Diffractive Glossmeter For Measurement of Dynamic

See
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/230673876
Diffractive glossmeter for measurement of

dynamic gloss of prints
Article · January 2008
CITATIONS READS
0 529
5 authors, including:
Hanna Maarit Koivula Martti Toivakka

University of Helsinki Åbo Akademi University
24 PUBLICATIONS 118 CITATIONS 254 PUBLICATIONS 1,574 CITATIONS
SEE PROFILE SEE PROFILE
Kai Peiponen
University of Eastern Finland
268 PUBLICATIONS 2,075 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Atmospheric Pressure Plasma Surface Treatment (Plastek) View project
Nanosturctured Large-area Antibacterial Surfaces View project
All content following this page was uploaded by Martti Toivakka on 22 May 2014.
The user has requested enhancement of the downloaded file.

TAGA Journal
Technical Association of the Graphic Arts
Volume 4, 2008
Copyright 2008
SWANSEA PRINTING TECHNOLOGY LIMITED
All Rights Reserved
ISSN 1748-0337 (Print)

ISSN 1748-0345 (Online)
The TAGA JOURNAL is published by SWANSEA PRINTING TECHNOLOGY LTD on behalf

of the TECHNICAL ASSOCIATION OF THE GRAPHIC ARTS
Reproduction in any form by any means without specific written permission is

prohibited.
The technical content of the TAGA JOURNAL is the sole responsibility of the authors
and the articles do not necessarily reflect the views of the editor or the publisher.
Any product names mentioned herein appear as a matter of information only and do not
imply endorsement by the TECHNICAL ASSOCIATION OF THE GRAPHIC ARTS, the editor or
the publisher.
For complete instructions on how to submit articles for review and possible inclusion in
this publication, please see http://www.tagajournal .com
SWANSEA PRINTING TECHNOLOGY LTD

7 Glanmor Park Road
Sketty
Swansea
SA2 0QG
United Kingdom
Email: tagajournal@swanseaprinttech.com
TAGA JOURNAL VOL. 4 © 2008 SWANSEA PRINTING TECHNOLOGY LTD

Editor’s Note
Tim Claypole
Welcome to the fourth volume of the TAGA Journal, the premier peer reviewed Journal
for all professionals who use printing and related processes to create products. This
represents a milestone for the TAGA Journal. Swansea Printing Technology Ltd took
over the publication in 2005. This was a dark time for TAGA with worries about its
future and much restructuring. One of the possible victims of this need to focus on core
activities was the TAGA Journal. Hence, Swansea Printing Technology stepped in to
ensure the survival of the TAGA Journal, as the only peer reviewed publication for
printing science and technology. Since then TAGA has recovered its strength,
delivering one of the premier annual International conferences for those in graphic
applications. Thus, TAGA, now successful under GATF stewardship, is in a position to
run the Journal.
Volume 5 will be published by GATF on behalf of TAGA. The editorial office, which
handles the peer review process, will move from WCPC (Welsh Centre for Printing and
Coating, Swansea University) to GATF. Mark Bohan will take over as the editor. Mark
as well as being VP Research at GATF has had a long involvement with the Journal. He
was one of the group of TAGA members who first formulated the plan for a peer
reviewed Journal in San Diego hotel in the early 00’s. There is something of a Swansea
continuity as Mark at that time was a member of the academic staff in WCPC.
The successful publication of the Journal to date would not have been possible without
the work of others. In this regard, I would like to particularly thank my colleagues,
Eifion Jewell and David Bould. Many thanks goes to Christine Hammett, the
secretary/administrator of WCPC, for the hard work administering the peer review
© 2008 SWANSEA PRINTING TECHNOLOGY LTD TAGA JOURNAL VOL. 4

process, particularly chasing reviewers and authors, monitoring progress and formatted
the papers.
Finally, my thanks is to the authors who have submitted their work and the reviewers
who have ensure the quality of the Journal.
We at the WCPC pass the TAGA Journal back to TAGA and wish it every success in
the future.
Tim C. Claypole
Editor
TAGA Journal

TABLE OF CONTENTS
1 - 10 The use of non-destructive methods in the study of IJ ink
penetration
Tadeja Muck, Branka Lozo, Lenka Otáhalová, Markéta Držková,
Marie Kaplanová
12 - 24 Water-Borne and Solvent-Based Flexographic Inks –

Influence on Uncovered Area and Ink Leveling on PE-
coated Paperboard
Behudin Mesic, Gunnar Engström, Magnus Lestelius
25 - 34 Characterisation of the Lithographic Printing of

Conducting Films
B. Ramsey, D. Harrison, D. Southee
35 - 58 Grey Balance Control with a Re-Purposed GATF Colour

Circle
Manfred H. Breede
59 - 71 Diffractive glossmeter for measurement of dynamic gloss
of prints
Raimo Silvennoinen, Mikko Juuti1, Hanna Koivula, Martti Toivakka and Kai-
Erik Peiponen
72 - 83 The Effect of Paper Properties on the Color

Reproduction for Digital Proofing of Gravure
Publication Printing
Yu Ju Wu, Alexandra Pekarovicova and Paul D. Fleming
84 - 100 Factors Affecting the Printing Strength of Kaolin-Based

Paper Coatings
J.C.Husband, J.S.Preston, L.F.Gate, D.Blair and P.Creaton
101 - 115 A Comparison of Densitometric and Planimetric

Measurement Techniques for Newspaper Printing
Maria S. Wroldsen, Peter Nussbaum, Jon Y. Hardeberg

117 - 125 The Effects of Paper Coating on Gravure Ink Mileage
Curve
Renmei Xu, Yu Ju Wu, Alexandra Pekarovicova, Paul D. Fleming, and
Michelle X. Wang
126 - 148 From Multi Channel Publishing towards a Ubiquitous

Media Environment
Maria Åkesson and Carina Ihlström Eriksson, PhD
149 - 164 The evaluation of colour difference equations and

optimization of DE2000
Martin Habekost, Dr. rer. nat., Dr. Katrin Rohlf
165 - 177 Effect of Toner Fixing Temperature on Print Properties

in the Electrophotographic Process
Timo Hartus
179 - 191 Significance of Print Quality in Variable Data Printing
Åsa Gidlund, Thomas Mejtoft & Sofia Demnert
192 - 205 Effect of drying temperature profile and paper on

mechanical print quality in heatset offset printing
Timo Hartus
206 - 218 Factors Impacting the Evaluation of Printer Profile

Accuracy
Robert Chung
219 - 231 Characterization of Conductive Polymer Inks based on

PEDOT:PSS
Erika Hrehorova, Marian Rebros, Alexandra Pekarovicova, Paul D. Fleming,
and Valery N. Bliznyuk

T. MUCK, B. LOZO, L. OTAHALOVA, M. DRZKOVA, M. KAPLANOVA 1
The use of non-destructive methods in the study of IJ ink

penetration.
Tadeja Muck, Branka Lozo*, Lenka Otáhalová**, Markéta Držková**, Marie Kaplanová**
University of Ljubljana *University of Zagreb

Faculty of Natural Sciences and Engineering Faculty of Graphic Arts
Chair of Information and Graphic Technology Getaldićeva 2
Snežniška ul. 5 10 000 Zagreb
SI-1000 Ljubljana Hrvatska
Slovenia branka.lozo@grf.hr
tadeja.muck@ntf.uni-lj.si
**University of Pardubice
Faculty of Chemical Technology
Department of Graphic Arts and Photophysics
Studentská 95
532 10 Pardubice
Czech Republic
lenka.otahalova@upce.cz
marketa.drzkova@upce.cz
marie.kaplanova@upce.cz
Abstract
The final phase in printing process is drying of printing inks on the printing substrate. The
mechanism of drying and ink distribution on the substrate surface depends on the ink
characteristics as well as on many other factors, like surface energy, roughness, sizing
and porosity. The scope of the study is to evaluate the applicability of several non-
destructive methods in the analysis of vertical and radial ink distribution on diverse Ink-
Jet prints. The dye-based Ink-Jet inks printed on different types of coated and uncoated
papers were analyzed. The non-destructive measurement methods were used as follows:
Slit-scanning densitometry, photoacoustics and confocal laser scanning microscopy –
CLSM. The achieved results were verified by the use of the destructive measurement
method consisting of microtome cross-sections in z-direction of the prints and optical
microscopy. The destructive method gives a real insight into the depth of penetrated ink
and enables the correlation with the studied non-destructive measurement techniques.
Keywords: Ink-Jet ink, Confocal laser scanning microscopy, Photoacoustics, Cross-

section analysis, Slit-scanning densitometry
1 Introduction
The quality of ink-jet prints depends to a large degree on the interaction between the printing
substrate and Ink-Jet ink. The penetration of printing ink should be low in order to retain high
optical colour density [1,2]. The surface of the paper also plays an important role: the absorption
capacity for printing inks should be controlled. It should not dust and should have a suitable
level of smoothness as well as other properties necessary for good operability and printability
[3,4].
The efforts have been done in explanation of the ink-substrate interaction by non-destructive
methods. A few studies have included the application of different spectroscopic and
microscopic methods (Raman Spectroscopy, Fourier Transform infrared photoacoustic
© 2008 SWANSEA PRINTING TECHNOLOGY LTD TAGA JOURNAL VOL.4

2 THE USE OF NON-DESTRUCTIVE METHODS IN THE STUDY OF IJ INK PENETRATION
spectroscopy and UV resonance Raman spectroscopy) in the study of ink distribution inside the
print [5].
This study extends recent results in investigation of radial and vertical distribution of Ink-Jet
printing ink, both on and beneath the paper surface, on different types of papers, and determines
the impact of the paper surface on print quality. Consideration of advantages and drawbacks of
each method is discussed with the aim to find method that can provide satisfactory insight into
Ink-Jet ink penetration in the substrate structure.
2 Materials and Methods
2.1 Types of paper

In the study five completely different types of printing substrates were used (Table 1). The
reason for the use of such a wide range of samples was to check precisely the limits of the
applied non-destructive and destructive measuring techniques.
Table 1: The base characteristics of printing samples.
Sample Sample description Grammage /

mark Thickness
Z Zweckform Photo Paper, recommended for Ink-Jet 130 g/m2, 170 μm
prints, one-side coated, high glossy
K Carton board, one-side coated multilayered white 300 g/m2, 430 μm
board
I ICP Permanent Paper, quality corresponds to standard 70 g/m2, 110 μm
ISO 9706, no surface treatment
T Tissue paper, Three-ply with a low percentage of 60 g/m2, 185 μm
secondary fibers
A Scheufelen’s offset standard paper, coated, wood-free, 150 g/m2, 115 μm
without fluorescent whitening agents
For better understanding and gain of more complex insight in the results of photoacoustic
measurements, standard coated paper (mark A) was included in the study.
The prints were done on the Canon BJC 8500 ink jet printer with magenta water-soluble dye-
based ink. Beside magenta ink, only for the photoacoustic measurements also black ink was
used. Printing quality was 1200 dpi with colour intensity 100%.
2.2 Non-destructive methods
2.2.1 Slit-scanning densitometer is usually used in analytical chemistry for evaluation of thin
layer chromatograms. The separated compounds could be qualitatively and quantitatively
evaluated by measuring the optical density in reflection or transmission mode. Therefore the
samples with printed magenta dots were prepared. The diameter of spots and spacing between
spots was 3 mm. They were arranged as different lines and tracks. By slit-scanning densitometer
both the radial and vertical distribution of printing ink were investigated. Also, the method was
used for evaluating the influences of paper on the non-homogeneity of the signals from different
tracks and different lines of the models. The testing conditions for measurements were
remission mode and a wavelength 580 nm.
2.2.2 Confocal laser scanning microscopy (CLSM) was at first used in medicine, especially in
analyzing of pathophysiological samples. In the field of graphic arts the CLSM represents a
TAGA JOURNAL VOL.4 © 2008 SWANSEA PRINTING TECHNOLOGY LTD

non-conventional method for the non-destructive study of the interaction between Ink-Jet inks
and the printing substrate. It was used in the investigation of three-dimensional paper structures,
the determination of pore volume [6,7] and the evaluation of ink penetration in the z-direction
[8].
The type of the analyzed test form was the same as for the Slit-scanning microscopy. The spots
of the printing ink were observed at 460 µm × 460 µm scan area. The air objective with
numerical aperture 0.6 was used. The laser beam wavelength was 458 nm and resolution in the
z-direction 2.7 µm. The sample preparation, i.e. cutting the samples in z-direction was done
virtually, by computer software.
This software calculated the number of slices necessary to pass from the most fluorescent to the
non-visible part of the ink. The measurements were done on printed samples covered by
immersion oil, which enabled the proper transparency of the samples.
2.2.3 Photoacoustics is the non-destructive method for the study of optical and thermal
properties of a printed area at different depths beneath the sample surface at given wavelength
of excitation radiation. This implies the examination of ink penetration into the paper structure.
The physical principles of indirect photoacoustic effect and its possible applications in the field
of graphic arts are already well established [9,10,11,12].
Magenta and black Ink-Jet prints were measured together with the appropriate blank paper.
Samples were cut 11 mm in diameter and mounted on plexi-glass backing. As a reference, the
“carbon black” was used. A diode pumping solid-state 20 mW laser (Suwtech, 532 nm) was
used as the light source. The generated acoustic signal was detected by a capacitor microphone
Brüel and Kjaer type 4166 and preamplifier. The output signal was fed to a DSP lock-in
amplifier Stanford Research Systems model SR830. The laser light was modulated using the
internal oscillator of the lock-in amplifier. The measurements were controlled and processed by
computer.
To acquire basic parameters of studied samples, the following methods were used. Optical
spectra for optical density and whiteness determination were measured according ISO standard
13655 by a reflection spectrophotometer Eye-One. Paper roughness was evaluated by Parker
Print Surface Roughness tester (model 58-04-00-0001). The mass-per-area of all substrates was
determined through analytical weighing of 100 square centimetres paper samples (Karl
Schröder KG sample cutter) and thickness of the paper sheets was measured with a thickness
gauge. The mean density of individual substrates was determined from thickness and mass-per-
area values.
2.3 Destructive method
2.3.1 Cross-section analysis is the destructive method which enables a detailed access to the
undersurface migration of printing ink through the inside layers of the printing substrate.
The samples were immersed into the polyethylene glycol, frozen at −30 °C and covered with
Keyser glycerol Jelly. The cross cutting was done by microtome Cryostat Leica CM 1850. The
final microtome slice thickness was 30 µm. The slices were then examined by optical
microscope Zeiss Option Axioscope and captured by a Canon EOS Digital Rebel camera.
3 Results and discussion
3.1 Non-destructive methods

3.1.1 Slit-Scanning Densitometer
The magenta prints of all printing samples were measured line by line with the Slit-scanning
densitometer. The measurements were done in remission mode. The peak area value indirectly

presents the optical density of printed samples. The evaluated standard deviation values show
the homogeneity of ink distribution on the printed surface.
Table 2: The average peak area and RSD measured by densitometer in remission mode,
for magenta ink printed on all four samples.
Z1 Z2 K1 K2 I1 I2 T1 T2
average 47802.4 48468.8 41904.5 41602.4 40114.8 38897.3 25010.0 24899.8
st. dev 411.6 546.5 1637.7 1501.5 1133.0 577.1 2730.0 2242.4
RSD 0.9 1.1 3.9 3.6 2.8 1.5 10.9 9.0
Figure 1: The peak area of the magenta prints measured by densitometer in remission mode on
all four samples.
From the Table 2 and Figure 1 it can be seen that the highest intensity (optical density) of
magenta ink is achieved on the Z sample and the lowest value on the T sample. On the Z sample
the ink remains on the paper surface while on the T sample it penetrates into the paper. The
relative standard deviation (RSD) values show that on the Z sample the printed Ink Jet inks are
homogeneously distributed (Table 2). Such characteristics were expected from the paper
producer’s specification which guarantees a brilliant colour reproduction. The T sample has the
highest RSD value – around 10, because of its rough and open surface structure. The RSD value
for the K sample is higher than for the uncoated I sample. This can be explained by the type of
coating of the K sample, which is excessively hydrophobic. After the drying process, dye-based
ink remains on such a surface in a non-homogeneous layer. Therefore, the K sample is not
suitable for Ink Jet printing.

Figure 2: Surfaces of Z (left) and K (right) magenta prints observed by CLSM.
To confirm last statement, an image was taken of the Z and K surfaces of magenta prints by
Confocal Laser Scanning Microscopy. Figure 2 shows the surfaces of both prints observed by
CLSM. The size of the captured slice of the Z sample was 146.2 × 146.2 micrometers in x and y
directions. The z direction of the sample presents the thickness, i.e. virtual thickness expressed
as the CLSM resolution in z direction, which was 24.7 micrometers for the Z sample. The size
of the captured slice of the K sample was 206.8 × 206.8 in x and y directions and the CLSM
resolution in z direction of the sample was 12.4 micrometers. The picture proved the
conclusions from Slit-scanning densitometry. On the surface of K sample the non-uniformity of
ink layer as well as the holes in it are clearly visible. The ink on the Z sample is homogeneous
distributed. The micro cracks visible on the surface are used only for ink’s faster fixation.
3.1.2 Confocal Laser Scanning Microscopy (CLSM)

The CLSM measurements of vertical ink distribution were done on the magenta printed area for
each of the samples. By the computer software the slices in z-direction were virtually cut. The
maximum depth of penetrated ink was calculated using the number of slices from the top of the
printed area (the most fluorescent part of the magenta dot) to the lowest visible ink trace in the
sample (Table 3).
Table 3: Results of CLSM vertical distribution of magenta ink.
Sample Aver. Nu. of slices Slice thickness Max. ink thickness (µm)
Z 32.23 0.73 23.53
K 32.64 0.74 24.15
I 32.12 0.80 25.96
T 89.51 0.81 72.50

Figure 3: The printed surface of I sample evaluated by CLSM software (virtual slice).
Figure 3 shows the example of printed surface evaluated by CLSM software (virtual slice) for
the I sample. The stack size is 206.8, 206.8 and 31.3 µm for x, y and z-directions, respectively.
The deepest ink penetration is present on the T sample and the shallowest on the Z sample
(Table 3). These results correspond to those of the Slit-scanning densitometer. On the Z sample,
ink forms a layer on the paper surface; the optical density of such a layer is high. The measured
ink thickness for both T and I samples is high, because ink penetrates deeper into the uncoated
substrates (Table 3).
3.1.3 Photoacoustics
The preliminary measurements of spectra density were done before the photoacoustic
measurements. Figure 4 shows the influence of the base sample characteristics on the spectra
density changes.
1.4
Z
1.2
A
Optical density of magenta prints
K
1.0 I
T
0.8
0.6
0.4
0.2
0.0
350 400 450 500 550 600 650 700 750
λ (nm)
Figure 4: Optical density spectra changes for different substrate types.

The optical density spectrum differs for each type of sample, meaning that different colours will
be perceived on diverse media. The change in the appearance of printing colour is mainly due to
the different ink distribution in z-direction; in addition, in the case of the Z sample the dissimilar
low wavelength spectrum is probably due to presence of FWA (fluorescent whitening agents).
The highest optical density is achieved on the Z sample and the lowest on the T sample. These
results are completely in accordance with both previous applied non-destructive methods.
Photoacoustic measurements were done in the modulation frequency range from 10 Hz to 1 kHz
and modulation frequency dependencies for both intensity and phase of photoacoustic signal
were evaluated for the individual samples.
In modulation frequency range, where the absorbing layer is thermally thin, i.e. the signal arises
from the whole thickness of the layer, the photoacoustic signal is influenced by structural and
thermal properties of the substrate, the so-called sample backing. The theory predicts the
decrease of the photoacoustic signal intensity with increasing density of the backing. This is
supported by the measured dependence of the normalized intensity of photoacoustic signal on
mean density of substrates (Figure 5). Only for the K sample is the value of normalized
photoacoustic signal lower than the values for substrates with similar mean density. The value
corresponds to the value for the A sample, whose mean density is considerably higher.
This reflects the real structure of the K sample, as the main effect on its upper side density
comes from coating layer, as is also the case with the A sample. With increasing modulation
frequency the influence of the substrate should decrease and the photoacoustic signal should be
driven by the absorbing layer (i.e. the impact of ink density becomes more significant). In
accordance with this assumption, the dependence on substrate mean density observed for low
frequencies diminished when the highest frequencies were used.
0.10
Normalized photoacoustic intensity at 50 Hz
0.09
0.08
0.07 Z
0.06 A
K
0.05 I
0.04 T
0.03
0.02
0.01
0.00
0 200 400 600 800 1000 1200 1400
Mean density (kg/m³)
Figure 5: Dependence of normalized photoacoustic intensity at 50 Hz on mean density of

substrates.
For depth-profiling, the phase information of photoacoustic signal is also important. Differences
between individual samples are more pronounced for the phase lag frequency dependence than
in the case of intensity. As the reference carbon black sample used for the experiment has
shown some irregularities particularly for lower modulation frequencies, it was not possible to

gain absolute values of the phase lag; therefore, only phase lag differences between each sample
and carbon black reference were calculated (see Figure 6). However, ink penetration depth still
can be roughly estimated. According to the measured photoacoustic phase profiles, the ink layer
thickness of the studied samples is greater than several microns; penetrating into the substrate
structure to depth on the order of tens of microns.
From the relative differences of phase values for individual samples and the overall character of
phase curves further conclusions can be drawn. Generally, the phase lag of the photoacoustic
signal is minimal for strongly absorbing samples with high thermal diffusivity and increases
with decreasing values of sample optical absorption coefficient and thermal diffusivity.
Comparing the phase lag of photoacoustic signals originating at the same depth beneath the
sample surface, the weaker photoacoustic signal, the greater phase lag. The phase lag also
increases with increasing depth into the sample, as the photoacoustic signal from deeper layers
is delayed due to thermal transport to the sample surface.
Assuming thermal properties in the range typical for respective kinds of materials, at the highest
modulation frequency used, the penetration depth of photoacoustic signal is approximately 10
microns and at the lowest it is one order higher. This means that at high frequencies, the
photoacoustic signal is influenced only by the uppermost layer of the samples, while at low
frequencies the maximum penetration depth also plays a role. For samples studied, the phase lag
increased in the higher modulation frequency region in following order: A, K, Z, I and T (Figure
6); toward the lower frequencies, the phase curves get near each other except for the T sample,
whose phase lag remained apparently greater than for other samples. In the studied range of
modulation frequencies it is also possible to distinguish uncoated and coated papers based on
the shape of curves.
15
Photoacoustic phase (sample − ref.) (degrees)
Z
-15 A
K
I
-30 T
-45
-60
0 5 10 15 20 25 30 35
Square root of frequency (Hz)
Figure 6: Dependence of photoacoustic phase reduced by carbon black values on square root of
frequency.
Further experiments with the T sample have shown significant influence of compactness of its
three layers on modulation frequency dependence of both intensity and phase of photoacoustic
signal. In low modulation frequency region, the maximum difference measured for compact and
split sample was 0.042 for normalized photoacoustic intensity and −20.8 degrees for
photoacoustic phase. On the contrary, for high frequencies it was −0.027 for normalized
photoacoustic intensity and 10.3 degrees for photoacoustic phase. Thus, the generally known

ability of photoacoustic method for investigation of the subsurface structure was verified to be
suitable for the type of samples studied.
3.2 Destructive method

3.2.1 Cross section analysis and Optical Microscopy (OM)
The cross section analysis was used as a destructive method. The samples were cut by
microtome, observed by optical microscopy and captured by camera in order to measure the real
depth of penetrated ink. All the analyzed samples caused various difficulties in cross section
analysis (as the magenta dye ink is water soluble, the microscopy evaluation of the prints should
be done as quickly as possible after the sample preparation). The K sample (carton board) was
found to be the most problematic. Due to its thickness neither microtome cross-section nor
optical microscopy was done.
Table 4: Results of vertical distribution of magenta ink
Sample Ink thickness

Sample
thickness (µm) (µm)
Z 170 20–34
K 430 —
I 110 39–65
T 185 66–79
Table 4 shows that the results obtained by the microtome cross section measurements are in
correlation with all presented non-destructive measurements: slit-scanning densitometry,
photoacoustics and CLSM (Table 3). Results of the cross section analysis and CSLM are in very
good agreement particularly for Z and T sample. In case of the I sample, a possible reason of
deviation between ink thickness values determined by these methods is inhomogeneity common
for uncoated papers, influencing both the ink thickness in X-Y plane and uneven deformation of
sample thickness during preparation for cross section analysis.
Figure 7: The microtome cross section analysis of Z sample with added micrometer measuring
tool.
Figure 7 presents the measurement of the ink distribution in the z-direction on the microtome
cross-section. The thickness of the penetrated ink is measured by inserting the micrometer
measuring tool after the OM image has been taken.
3.3 Comparison
The results obtained by the microtome cross-sections present the real insight into the vertical ink
distribution in z-direction of the print as well as into the surface thickness of the ink layer.

Those facts make this destructive method a reliable control measurement for all the applied non-
destructive methods.
3.3.1 Cross section analysis vs. Slit-Scanning Densitometry

The real thickness of the Ink-Jet ink printed on the Z paper is in the range of 20–34
micrometers. This result corresponds to that obtained by slit-scanning densitometry showing the
highest optical density for the Z sample. This was explained by the remaining of Ink-Jet ink in a
compact layer on the paper surface. The Z paper is the high quality photo paper. Its coating
layer is thin and has a very good absorption characteristics. The correlation is also obtained for
other measured samples: The real ink layer thickness in the z-direction of the I sample is in the
range of 39–65 micrometers and of the T sample it is 66–79 micrometers. The results obtained
by slit-scanning densitometry show that the T sample has the lowest optical density and that the
value for the I sample is in between the extremes (as well as the K sample for which the cross
section was not possible).
3.3.2 Cross section analysis vs. Confocal Laser Scanning Microscopy (CLSM)
The results obtained by CLSM are in the close correlation with those of the microtome cross-
section analysis. The real thickness of the Ink-Jet ink layers are precisely in the same range with
those measured by CLSM for the Z (20–34 vs. 23.53 micrometers) and the T (66–79 vs. 72.50
micrometers) samples. Both methods agree for the I sample in between of the Z and T extremes.
3.3.3 Photoacoustics
Phase of the photoacoustic signal in higher modulation frequency region decreased in order A,
K, Z, I and T (see Figure 6). As the photoacoustic phase depends both on optical and thermal
properties and the optical density measured by conventional spectrophotometer (see Figure 4)
decreased in the same sequence except for the Z sample, it reflects dissimilarity in Z sample
structure in comparison with other samples. The Z sample is printed on photo paper, whose
coating is expected to have special properties intended to ensure optimal distribution of the ink,
so the ink penetration profile is probable to be specific.
At the other end of the modulation frequency range, only the phase lag of T sample is
apparently higher in comparison with remaining samples. This is in accordance with maximum
penetration depths determined namely by CSLM (see Table 3).
Unambiguous determination of the exact ink-penetration depth could not be so far addressed by
photoacoustic measurements due to lack of parameters needed to be known for assessment of
such a complex multilayered system. At this stage, by the other methods used, the penetration
depths for Ink-Jet ink printed on respective substrates were determined, but neither destructive
cross section analysis nor other non-destructive methods provided detailed depth-profiles of the
samples. Satisfactory discrimination of the phase curves measured for the individual samples
proves sufficient sensitivity of photoacoustics, to observe differences in both optical and
thermal properties of subsurface structure. In further research, correlation with results of all
discussed methods applied to extended and better defined sets of samples should enable the
actual depth-profile of Ink-Jet prints to be established.
4 Conclusions
The relevance of information obtained by different non-destructive and destructive

measurement techniques is achieved. Each of the studied methods, despite their limitations,
gives an explanation of the ink – paper interaction as well as the impact of the substrate on the
print quality.
A detailed study of the obtained results shows a good correlation among the applied non-
destructive methods: slit-scanning densitometry, photoacoustics and CLSM. Not only the
correlation among the non-destructive methods themselves, but the correlation with the results

of the destructive method is achieved. The cross section analysis by microtome and optical
microscopy give the real insight into the ink distribution in the z-direction. Slit-scanning
densitometry, photoacoustics and CLSM are introduced as the methods complementary to each
other and relevant for future applications.
Acknowledgements
This work, was partly supported by the Ministry of Education, Youth and Sports of the
Czech Republic, project No. MSM 0021627501.
References
1 Glittenberg D., Voigt A. and Donigian D., Novel pigment starch combination for
the online and offline coating of high quality inkjet papers, Pap. Technol, Vol. 44,
No. 7, pp.36-42, 2003
2 Muck T. and Hladnik A., Evaluation of radial and vertical distribution of ink jet
inks in paper, Professional papermaking, Vol. 2, No. 2, pp.62-64, 66-68, 2004
3 Lyne M. B. and Aspier J. S., Paper for Ink-Jet printing, Tappi Journal, pp.106-110,
1985
4 Baudin G. and Rousett E., Effect of paper properties on print quality, Imaging
Science and Technology, pp.120-124, 2001
5 Lozo B., Vyörykkä J., Vuirinen T. and Muck T., Nondestructive microscopic and
spectroscopic methods for depth profiling of ink jet prints, J. imaging sci. technol.,
Vol. 50, No. 4, pp.333-340, 2006
6 Goel A., Tzanakakis E.S., Huang S., Ramaswamy S., Hu W-S., Choi D. and
Ramarao B.V., Confocal laser scanning microscopy to visualize and characterize
the structure of paper, AICHE Symposium Series, Vol. 96, No. 324, pp.75-79, 1999
7 Auran P.G. and Bjorkoy A., Measuring the pore volume distribution of papers by
CLMS for printability, STFI, Proceedings, p.220, 1999
8 Hoang V., Huy H.L., Wei S. and Parker L.H., The interactions of ink-jet inks and
uncoated papers, 55th Appita annual conference, pp.285-292, 2001
9 Kaplanova M. and Katuscakova G., Photoacustic Study of the Thermal Effusivity
of Cellulose and Paper, Springer series in Optical Science Vol. 69., p.180, 1992
10 Kaplanova M. and Cerny J., Photoacoustic and Photocalorimetric Study of UV
Curable Inks and Varnishes, Advances in Printing Science and Technology, p.107,
1994
11 Kaplanova M. and Cerny J., Photoacoustic Study of the Ink and Paper
Interactions, Advances in Printing Science and Technology, p.301, 1997
12 Drzkova M., Otahalova L. and Kaplanova M., Photoacoustic Study of Printed
Samples, The First European Conference on Color in Graphics, Imaging and Vision
(CGIV), p.307, 2004

12 WATER-BORNE AND SOLVENT-BASED FLEXOGRAPHIC INKS – INFLUENCE ON …
Water-Borne and Solvent-Based Flexographic Inks –

Influence on Uncovered Area and Ink Leveling on PE-coated
Paperboard
Behudin Mesic*, Gunnar Engström** and Magnus Lestelius**
* Affiliated at Samrt Print

BrobyGrafiska Education 37
SE-686 80 Sunne
Sweden
behudin.mesic@brobygrafiska.se
** Affiliated at Karlstad University

Dept. of Chemical Engineering
Faculty of Technology and Science
SE-65188 Karlstad
Sweden
magnus.lestelius@kau.se; gunnar.engstrom@kau.se .
Abstract
A smooth and a rough PE-extrusion-coated paperboard, with and without corona
treatment, were printed flexographically using a solvent-based and a water-borne ink
with the objective to examine the influence of surface roughness and wetting of the
paperboard on ink transfer, uncovered area (UCA), ink-film thickness distribution, ink
leveling, and print unevenness. The results showed that ink transfer was significantly
lower for the solvent-based ink, but because this ink has a stronger pigment, less ink was
needed for a given ink density. Both inks were equally prone to yield UCAs, and both
inks also formed dry ink layers with the same thickness distribution in the 2–4 mm range.
Both inks also leveled out to same extent. The leveling was controlled by the corona
treatment and reduced the UCAs. The smooth substrate exhibited the least UCAs. On a
submillimeter level the print unevenness for the water-borne ink exhibited patterns that
resemble crawling worms or curled thread ends aligned in the direction of the printing.
The pattern for the solvent-based ink was grainy and without orientation. These patterns
are suggested to reflect the splitting pattern of the ink during the transfer to the
paperboard.
Keywords: printability, solvent-based ink, water-borne ink, surface roughness, PE-

extrusion-coated paperboard, ink transfer, uncovered area, ink leveling, print unevenness.
1 Introduction
Flexography is a progressive and widely applicable printing method for a wide range of
substrates and finished products, such as foil, plastic film, corrugated board, paper, paperboard,
or even fabric [1,2]. It is a printing method that has greatly expanded and is often preferred for
package printing [3,4]. Flexography generally uses low-viscosity inks, either solvent-based or
water-borne, which dry very quickly between the printing units in the printing press [5].
Solvent-based inks are formulated to contain a blend of alcohol that dissolves the resin and
produces a liquid ink, while water-borne inks are generally dispersions, using water and amines
for the liquefaction [6].
Both solvent-based and water-borne inks solidify by loss of solvent/water (drainage and
evaporation). Solvent-based inks give good adhesion, gloss, and flexibility on polymer-coated
substrates and aluminum foil, and they provide good resistance to water, acid, and alkali [7].

BEHUDIN MESIC, GUNNAR ENGSTRÖM AND MAGNUS LESTELIUS 13
Some disadvantages with using solvent-based inks are that they require venting, because odor is
given off, and that environmental hazards increase, because the inks contain highly inflammable
solvents. Water-borne inks share many common features with solvent-based inks and are said to
be cost effective, when material and compliance costs are factored in [7]. Water-borne inks have
advantages compared to solvent-based inks: they wash up with water, they do not produce
vapors hazardous to health, they reduce fire hazards and insurances costs, and their cost is less
affected by fluctuations in the price of crude oil. However, water-borne inks also have
disadvantages: they require more energy for drying, and they do not properly wet printing
substrates with low surface energy, such as polymeric films (for example, polyethylene (PE)
film and PE-coated board). Another disadvantage with water-borne inks is the handling of the
waste and recycling of packaging material [7,8].
A common and serious problem in flexographic printing, as well as with solvent-based and
water-borne inks, is uncovered area (UCAs). The UCAs consists of a number of small dots,
typically 0.04-0.9 mm2 in size, that are supposed to be covered with ink but are not. When
printing with water-borne inks, both local areas of low surface energy and depressions [2,9] in
the surface of the printing substrate have been reported to be able to cause UCAs. Experience
within the printing industry is that solvent-based inks cause less UCAs than water-borne inks.
Because solvent-based inks have significantly lower surface tension than water-borne inks, this
suggests that solvent-based inks wet the printing substrate better and also level out better.
Therefore, solvent-based inks are believed to be forgiving for areas of low surface energy of the
printing substrate.
This paper is focused on UCAs and the issues of how much of the UCAs stem from local areas
with low surface energy and how much stems from depressions in the surface of the printing
substrate. To shed light on this issue, a printing trial was performed in which a smooth and a
rough PE-extrusion-coated paperboard, with (44 mJ/m2) and without (<32 mJ/m2) corona
treatment, were printed with a solvent-based ink and a water-borne ink respectively. Besides
UCAs, the prints were tested with respect to ink transfer, dry ink-film thickness distribution, ink
leveling, and print unevenness. The printing was performed at two line pressures.
2.1 Inks
Two different inks supplied by Sun Chemical were used: 1) a water-borne ink (Cyan C
SCANFULLTITE RASTER 713-45686 B 60750), and 2) a solvent-based ink (Cyan 473-
QUATTROFLEX B 61659). Viscosity of in was measured using a Zhan 2 cup, while surface
tension was measured using the FIBRO DAT 500/1100 Dynamic Contact Angle and Absorption
Tester. The ink data are given in Table 1.
Table 1: Ink data
Type of ink ⇒ Water-borne ink Solvent-based ink

Viscosity, according Zhan 2 cup, (s) 29.5 29.5
Surface tension, (mJ/m2) 35.8 24.7
Dry solid content, (%) 32 22
Density, (g/mL) 1.0 0.93

2.2 Printing Substrates

A smooth pigment-coated (273 g/m2) and a rough uncoated (252 g/m2) liquid packaging
paperboard—supplied by Stora Enso, Skoghall, Sweden—were PE-extrusion coated on a pilot
scale by Tetra Pak, Lund, Sweden. The grammage of the PE layer was 12g/m2 for both
paperboards. After the PE-extrusion coating, the liquid packaging boards were cut into 5.5 cm ×
70.0 cm sheets, and half of the sheets were corona treated using Corona-Plus lab treater
equipment (Vetaphone, Kolding, Denmark), designed for the corona treatment of polymer films
or paperboard sheets. The corona power output was 2.0 kW using ceramic electrodes and an
aluminum roll with a perimeter speed of 1.65 m/s. All sheets were stored at a constant
temperature (23.5°C) and protected from light, air, moisture, and dust in sealed, partially
transparent PE bags (protected sheets were first stored in small sheaves, each wrapped in PE-
coated wrapping paper, and then stored in sealed, partially transparent PE bags). The untreated
(0 kW) paperboard displayed a surface tension less than 32 dynes/cm and treated (2.0 kW) 44
dynes/cm, as measured by Vetaphone’s PRO-DYN test pens (Vetaphone, Denmark).
The surface roughness of the PE-extrusion-coated sheets was measured using a profilometer
(Perthometer C5D, Perthen, Germany) equipped with a narrow diamond needle with a radius of
radius 2 μm [10]. The Ra value for the smooth pigment-coated paperboard was 0.68 µm and for
the rough paperboard 1.29 µm.
2.3 Experimental Design

The design of the printing experiments was performed using MODDE 5.0 software (Umetrics
AB, Umeå, Sweden, www.umetrics.com). With the intention to reduce the number of
experimental points, the experiments were chosen according to a central composite face-
centered (CCF) design. Four variables nip-line load, surface energy level, type of ink, and
printing substrate were included, which resulted in 22 experimental points.
2.4 Printing
All printing was performed using an IGT F1 equipped for flexographic printing (IGT Testing
Systems, Amsterdam, Netherlands, www.igt.nl). Three different printing plates (DuPont ACE,
produced by FlexoPartner AB, Sunne, Sweden, using straight light illumination) with 100%,
70% and 30% tone values were used. The screen ruling of the printing plate was 24 lines/cm,
the thickness was 1.70 mm, and its hardness was 71° shore A. The anilox roller had a screen
ruling of 180 lines/cm, with an ink volume of 2.7 g/m2, and a nip-line load of 125 N. The
conditions during the printing were kept constant (50% RH and 23.5°C). Two levels of the line
load between the printing plate and impression cylinder were used, 50 N and 150 N
respectively. The printing speed was 0.4 m/s. For each experiment, three strips of 4 cm × 40 cm
(long side of the strip printed in the MD direction) were printed (se Figure 1).
Figure 1: Image of a printed strip.
The amount of copper originating from the ink pigment on a selection of a sheet from the
printed material, as well as on the inks themselves, was measured using SCAN-CM 54:97 and
SCAN-38:96 with exception of a liquefaction/decomposition time of 3 h (adapted to measure
Cu) instead of 16 h (adapted to measure Cd).

2.5 Print Density, Print Unevenness, and UCAs

Print density was measured with an optical densitometer (Gretag Macbeth D19/47B/P,
Regensdorf, Switzerland) connected to a computer for data acquisition. Print density is given as
the mean value of 10 measurements on each strip. The optical print density of a printed area,
Dp, is defined as the logarithm of the ratio of the reflectivity of the unprinted area, R∞, of the
substrate to the reflectance factor of the printed area, Rp, (Equation 1).
R∞
Dp = log (1)
Rp
Print unevenness and percentage of UCAs in fulltone print were measured with the help of
image analysis using a flatbed scanner (Canon CanoScan D1250 U2F), with a resolution of 600
dpi for the imaging and software (STFI mottling v 2.6 software, developed by STFI, Stockholm,
Sweden) for the image processing. The processing consists of a frequency analysis based on a
fast Fourier transform (FFT). The result of print unevenness measurement is reported as the
standard deviation or the coefficient of variation within these wavelength bands (expressed in
mm): 0.125–0.25, 0.25–0.50, 0.50–1.0, 1.0–2.0, 2.0–4.0, and 4.0–8.0. The result of UCA
measurement is reported as a percentage of uncovered areas. The sample size was 21.7 mm ×
21.7 mm. The mean value is reported as mean value of the measurements of six such areas.
2.6 Visual Assessment of UCAs

The visual assessment (rating) of the amount of UCAs was made by eight observers (four male
and four female). The observers were asked to estimate the degree of UCAs in fulltone print.
The rating of UCAs required the observers to give a numerical assessment of what degree of
total printed area they considered to not be covered by ink. A scale of 1 to 8 (1 for the lowest
degree of UCAs and 8 for the highest) was used for the rating. The result is expressed as the
average value of eight statements. The size of the printed samples used for visual assessment
was 40 mm × 90 mm. Due to the large number of printed samples, the assessment was restricted
to the material printed at the lower printing pressure (50 N).
2.7 Light Microscopy

For microscopic evaluation of the print, a microscope of the brand Leica MZ 12 (Meyer
Instruments, Houston, U.S.A.) was used for the examination in side light. Side lighting was
provided by the Leica CLS 150X (150W cold light source with optical fiber light guides and
ringlights), with a light incident angle of 30 and a 10 cm distance from the investigated
sample. The microscope was equipped with 2.5 x1 magnifications and a Nikon DXM 1200
digital camera for capturing images (3.5 mm × 5.5 mm). Two images per strip were captured
and evaluated for each experiment.
3 Results and Discussion
3.1 Ink Transfer

To be able to fairly evaluate how surface roughness and surface energy of both PE-extrusion -
coated paperboards and the type of ink influence the printing performance, it is important that
the viscosity of the inks and the amount of ink (volume) transferred to the paperboards are
constant. This demand is fulfilled for the viscosity, because the inks were prepared to the same
viscosity. It is tempting to assume that this demand also is fulfilled for the amount of ink,
because it is reported that the transfer in applicator units similar to the anilox roller is volumetric
and governed by the open area between the roller and the paper [11]. However, these
researchers studied the transfer of water solutions. No solvent-based solutions were included in
their studies.
To estimate the amount of ink transferred to the paper, we measured the copper content in a
selection of representative prints and in the inks themselves by using atomic absorption. Copper

originates from the pigment in the ink, and according to the supplier, both inks contained the
same pigment but in different amounts. The result of the measurements is shown in Table 2,
where the corresponding print density is also given for sake of comparison.
Table 2: Copper content (mg/kg dry sample) in print and ink, and print density on smooth
corona-treated PE-extrusion-coated paperboard
Printing Copper
Sample identification Print density
pressure, N g/kg
Solvent-based ink on paperboard 50 0.023 0.593
Solvent-based ink on paperboard 150 0.023 0.643
Water-borne ink on paperboard 50 0.026 0.711
Water-borne ink on paperboard 150 0.026 0.724
Unprinted paperboard 0.0014

Solvent-based ink 57
Water-borne ink 36
From the data in Table 2 and the data given in “Materials and Methods,” the volume of ink on
paper was estimated:
Water-borne: 0.61 mL/m2

Solvent-based: 0.52 mL/m2
As is evident from this estimation, the volume of ink on paper was higher for the water-borne
ink (15%). Table 2 also shows the pigment strength for the water-borne ink was lower than for
the solvent-based ink. This suggests that to meet a given print density, more ink (mass) is
needed when printing with the water-borne ink.
The print density for all the printed material is shown in Figure 2. The highest print density was
obtained with the water-borne ink. This is in agreement with the copper content in the print (cf.
Table 2). The print density became higher on the smooth paperboard [12,13] as well as at the
higher printing pressure [14]. The corona treatment did not influence the print density. In a
previous study on the same material in which we investigated the effect of corona treatment on
the printability, we observed similar results [15].
0,9
50N 150N Smooth <32mJ/m2
0,8
Print density in full-tone
Smooth 44 mJ/m2
0,7
Rough <32 mJ/m2
0,6
Rough 44 mJ/m2
0,5
0,4
0,3
0,2
0,1
0,0
Water-borne Solvent- Water-borne Solvent-
ink based ink ink based ink
Figure 2: Print density for the printed PE-extrusion-coated paperboards. Error bars indicate
standard deviation.

3.2 UCAs
The prints were visually inspected in a light microscope with the view to trace the cause of the
UCAs—depressions or areas of low surface energy that reject the ink. Use of chosen
microscopic method failed in this respect. All UCAs appeared the same to us. However, that
does not mean that the cause for all UCAs was the same; it only means that the chosen method
did not manage to trace the cause.
To test the relevance of the instrumental measurement of the UCAs, a visual assessment (rating)
of them was performed. Due to the large number of printed samples, the assessment was
restricted to the material printed at the lower printing pressure (50 N). Figure 3 shows that the
agreement between measured and visually assessed UCAs was good.
5,0
4,0
UCA measured
3,0
2,0
1,0
0,0
0 2 4 6 8 10
UCA visually assesed
Figure 3: Relationship between measured and visually assessed UCAs.
The UCAs (instrumentally measured) of the printed material is shown in Figure 4. The material
printed with the solvent-based ink exhibited the highest values of UCAs. This is true at both
printing pressures, but lower values were, of course, obtained at the higher pressure. However,
to draw the conclusion that the solvent-based ink is more prone to impart UCAs than the water-
borne ink from this observation is incorrect, because the amount of ink on the paper was not the
same. As already mentioned in the previous paragraph, the amount of ink was 15% less for the
solvent-based ink, and to be able to correctly assess UCAs, the amount of ink must be constant.
Most likely the higher UCAs values for the material printed with the solvent-based ink were
caused by the lower amount of ink. Figure 4 also shows that surface roughness and corona
treatment affected the UCAs in the same way for both inks. Surface roughness increased UCAs
and corona treatment reduced UCAs.

4,5
50 N 150 N Smooth < 32 kJ/m2
4,0
Smooth 44 kJ/m2
3,5
3,0 Rough < 32 kJ/m2
2,5 Rough 44 kJ/m2

UCA, %
2,0
1,5
1,0
0,5
0,0
Water-born Solvent- Water-born Solvent-
Figure 4: UCAs for the printed material. Error bars indicate 5 % margins.
The conclusion stated above that UCAs for both inks was affected in the same way by the
paperboard properties finds support in the nice relationship between the UCAs for the materials
printed with solvent-based and water-borne inks respectively (see Figure 5).
3.3 Ink-Film Thickness Distribution

Mottling is the visual perception of unevenness in print density. This unevenness, in turn, is
governed by ink-film thickness distribution and print density. Mottling can be measured both
subjectively using a pairwise comparison method (or similar) or instrumentally using image
analysis. [16] reports good agreement between subjectively assessed mottling and mottling
measured using image analysis. The mottling measured using image analysis was expressed as
coefficient of variation, which is identical with the standard deviation in print reflectance over
the mean reflectance. This ratio is proportional to the standard deviation in print density [16].
In this work, the objective was to study the effect of paperboard surface characteristics on the
ink-film thickness distribution, rather than on mottling. For the printed material examined in this
work, the print density was low at most 0.78. Up to this print density, it can be assumed that
ink-film thickness is a linear function of print density. That means that the standard deviation in
print reflectance can be used as a measure of the ink-film thickness distribution.

5,0
4,0
UCA water-born ink

3,0
2,0
1,0
0,0
0,0 1,0 2,0 3,0 4,0 5,0
UCA solvent-based ink
Figure 5: Correlation between UCAs for material printed with water-borne and solvent-based
inks.
The standard deviation in print reflectance within the 2–4 mm wavelength range is shown in
Figure 6. The ink-film thickness was more uneven on the rough paperboard and that high
printing pressure reduced the unevenness. It is also evident that the corona treatment reduced
the unevenness slightly, which suggests that the leveling of the ink was improved [17]. This is
remarkable because the length scale was 2–4 mm Figure 6 also shows that the variations in ink-
film thickness developed the same way for both inks. This impression is strengthened by the
nice correlation, shown in Figure 7, between the standard deviations in print reflectance for the
materials printed with the water-borne and solvent-based inks respectively. In Figure 7, it is also
evident the standard deviation was somewhat lower for the material printed with water-borne
ink. This is very likely caused by the higher amount of ink on the paper and a more efficient
leveling.
1,6
Std dev. in print reflectance (2-4 mm)
Smooth < 32 mJ/m2

50 N 150 N
1,4 Smooth 44 mJ/m2
1,2 Rough < 32 mJ/m2

Rough 44 mJ/m2
1,0
0,8
0,6
0,4
0,2
0,0
Water-born Solvent- Water-born Solvent-
Figure 6: Standard deviation in print reflectance within the wavelength range of 2–4 mm.

1,6
Std. dev. in print reflectance

1,4
(2-4 mm) water-born ink

1,2
1,0
0,8
0,6
0,4
0,2
0,0
0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6
Std. dev. in print reflectance

(2-4 mm) solvent-based ink
Figure 7: Correlation between standard deviations in print reflectance for material printed with
water-borne and solvent-based inks.
It should be pointed out that UCAs does not interfere with the standard deviation in print
reflectance within the wavelength range of 2–4 mm reported here, because the length scale of
UCAs was significantly shorter (0.5–1 mm).
3.4 Print Unevenness on a Submillimeter Level

Figure 8 shows images of fulltone areas printed on the smooth paperboard without corona
treatment. The large unprinted spots or areas in the images with an extension of 0.3–0.6 mm are
UCAs. The reflectance of the printed area in between the UCAs is—as is evident in Figure 8
uneven, and the appearance of the unevenness is not the same for both the inks. For the water-
borne ink, the pattern resembles small, crawling worms or curled thread ends aligned in the
direction of the printing. [18] have observed a similar pattern for water-borne ink. For the
solvent-based ink, the pattern is grainy and without orientation. The pattern in the printed area
in between the screen dots (70%) on the same paperboard exhibits the same pattern (see Figure
9).
The pattern in the print on the rougher paperboard, as well as on the corona-treated paperboard,
exhibited the same appearance as that for the smoother and untreated paperboard, as shown in
Figures 8 and 9.
The variation in print reflectance can be analyzed, not only with respect to level, but also with
respect to orientation with help of software used here for image analysis. Figure 10, which
shows the standard deviation in print reflectance within the 0.125–0.25 mm range, summarizes
this analysis for the smooth paperboard without corona treatment printed with both inks. The
analysis showed that the standard deviation in print reflectance was higher in the print direction
than in the cross direction for the water-borne ink and that the standard deviation in print
reflectance for the solvent-based ink was the same for both directions. This is in agreement with
what we can see in Figures 8and 9.

Figure 8: Images of fulltone areas on the smooth paperboard without corona treatment
acquired using light microscopy.
Figure 9: Images of screen dots (70%) on the smooth paperboard without corona treatment
acquired using light microscopy.

3,0
Std dev. in print reflectance

Printing direction
2,5
Cross direction
(0.125-0.25 mm)
2,0
1,5
1,0
0,5
0,0
Water-born ink Solvent based ink
Figure 10: Standard deviation in print reflectance within the wavelength range of 0.125–0.25
mm for the smooth paperboard without corona treatment.
The orientated pattern for the water-borne ink probably reflects the ink-splitting pattern. This
pattern is caused by filament formation when the ink splits. The filaments are stretched and
broken during the ink transfer. If the broken ends of the filaments on the ink film on the printing
substrate do not level out, the printed surface will be patterned [17,19,20] report from work with
metered size press coating, in which the transfer of the coating colour is identical with the ink
transfer with an anilox roller, that inelastic filaments reduce the splitting pattern. These
researchers also report that high coat weights are prone to form filaments and a patterned
surface. This observation is consistent with ours (higher ink-film thickness for the water-borne
ink).
The elongation viscosity of the coating colour has been proposed to control the splitting pattern
in metered size press coating [21]. It is also known from studies of aqueous polymer solutions
that different polymers influence the elongation viscosity in different ways [22]. The solvent-
based ink and the water-borne ink contain different binder systems, which may influence their
elongation viscosity. A low elongation viscosity is desired [23].
4 Conclusions
• The ink transfer was significantly higher for the water-borne ink.
• The pigment strength was lower for water-borne ink, so for a given print density, more
ink was needed.
• Smoothness and corona treatment reduced UCAs in the same way for solvent-based ink
and water-borne ink. High printing pressure also reduced UCAs.
• The variations in ink-film thickness within the wavelength range of 2–4 mm were
influenced by the roughness and the surface energy of the paperboard, as well as by the
printing pressure, in the same way for both inks. Corona treatment reduced these
variations, which show that the inks leveled out on a fairly long length.
• On a submillimeter level, the print with the water-borne ink exhibited a pattern
resembling small, crawling worms or curled thread ends, aligned in the direction of the
printing. The print with the solvent-based ink was grainy without any orientation.

Acknowledgments
The authors are grateful to Tetra Pak R & D AB for producing the PE-extrusion-coated
paperboard and to Sun Chemical for supplying the printing inks. T2F is acknowledged for
financial support.
References
1 Castelli L., End-use Requirements and Ink Formulation, In Casatelli L. (ed.),

Flexographic Ink: A Process Approach, Foundation of Flexographic Technical
Association, NY, pp. 12–50, 1998
2 Zang Y.H. and Aspler J.S., Factors that affect the flexogrphic printability of
linerboards. Tappi J. 78(10):240-23-240-33, 1995
3 Chalmers I.R., Flexographic Printability of Packaging Grade Papers, Proc. 51st
Appita Ann. Gen. Conf., Melbourne, p.335, 1997
4 Dowdell W.C., The Flexographic Technical Association Issues, Its Annual Market
Perspective and Prospectus, Industry Forecast, Flexo, (1), 2001
5 Cusdin G., What is Flexography?, In Cusdin, G. (ed.) Flexography: Principle &
Practice, Foundation of Flexographic Technical Association, NY, pp 3 – 18, 1999
6 Crouch J.P., Inks for Flexography, In Crouch, J. Page. (ed.) Flexography Primer,
Second edition, GATF Press, U.S.A., p 85, 1998
7 Nelson R.E., The Place of Packaging in Our Society; Process Color Printing; and
Flexographic Inks, In Nelson, R.E. (ed.) Package Printing, Jelmar Publishing Co.,
Inc., New York, U.S.A., pp xvii, 3 – 18, 182 – 189, 346 – 349, 1993
8 Maust M.J., Correlation of Dispersion and Polar Surface Energies with Printing on
Plastic Films for Low VOC Inks, TAPPI J. 76 (5):95, 1993
9 Barros G.G., Fahlcrantz C.M. and Johansson P.Å., Topographic Distribution of
Uncovered Areas (UCA) in Full tone Flexographic Prints, TAGA J. 2(1):43-57, 2005
10 Wågberg P. and Johansson P-Å., Surface Profilometry – A Comparison Between
Optical and Mechanical Sensing on Printing Papers, Tappi J., 76 (12):115, 1993
11 Salt T., Paczkovski M. and Ohman S., Enhanceed MSP Performance with Gravure
Roll Premetering, MSP Forum, Orlando, FL, USA, 1-4 May 2002, pp 171-191,
TAPPI Press, 246pp, 2002
12 Lagerstedt P. and Kohlseth P., Influence of Surface Energetics on Ink Transfer in
Flexo Printing, In proc. 23rd IARIGAI Research Conference: Advances in Printing
Science and Technology, pp269-299, 1995
13 Aspler J., Cormier L. and Manfred T., Linerboard Surface Chemistry and
Structure Affect Flexographic Print Quality, Proceedings, International Printing and
Graphic Arts Conference, Vancouver, BC, Canada, pp. 167-177, 2004
14 Aspler J.S., Jordan B., Zang Y.H. and Nguyen N., Print quality of liner board in
commercial waterbased flexography, In proc. 50th Annual TAGA Conference, pp749-
781, 1998
15 Mesic B., Lestelius M. and Engström G., Influence of Corona Treatment Decay on
Print Quality in Water-borne Flexographic Printing of LDPE-coated Paperboard,
Journal of Packaging Technology and Science, 19:61-70, 2005
16 Johansson P-Å., Print Mottle Evaluation by Band-Pass Image Analysis, IARIGAI
22nd International Research Conference, Munich, Advances in Printing Science and
Technology 22, Pentech Press, pp 403, 1993
17 Desjumaux D.M. and Bousfield D.W., Levelling of Ink Film Defects to Obtain
High Gloss, 10th International Coating Science and Technology Symposium,
Scottsdale, AZ, USA, pp 276-279, International Society of Coating Science and
Technology, 288pp, 2000
18 Barros G.G. and Johansson P-Å., Mottle and Ink Distribution in Full-tone
Flexography Prints on Coated Linerboard, In Barros, G.G. Doctoral Thessis.

Influence of Substrate Topography on Ink Distribution in Flexography, Karlstad

University Studies 2006:17, Karlstad, Sweden, Paper V, 2006
19 Roper J.A., Selection of Thickeners for Optimatization of Metered Film Coater
Performance, Coating Conference, Toronto, Canada, 2-5 May 1999, pp 259-260,
TAPPI Press, 536pp, 1999
20 Gane P.A.C., Burri P., Drechsel J. and Reimers O., Formulation Optimatisation
for Improved Runnability of High Speed Pigmented Coatings on the Metered Size
Press, Coating Conference, Philadelphia, PA,USA, pp 15-22, TAPPI Press, 307pp,
1997
21 Issaksson P., Rigdahl M., Flink P. and Forsberg S., Aspects of the Elongational
Flow Behaviour of Coating Colours, Journal of Pulp and Paper Science, Vol. 24,
No. 7, pp 204-209, 1998
22 Jäder J., Willenbacher N., Engström G. and Arnström L., The Influence of
Extensional Properties on the Dewatering of Coating Colours, Journal of Pulp and
Paper Science, Vol. 31, No. 4, pp 181-187, 2005
23 Schoelkopf J., Gane P.A.C. and Fischer S., Determination of Elongational Strain
During Ink-on-Paper Tack Measurement: A Novel Technique to Measure Work
Cohesion and Adhesion of the Thin Films, 11th International Printing Graphic Arts
Conference, pp 11, Association Technique De I’Industrie Papetiere, 486pp, 2002

B. RAMSEY, D. HARRISON, D. SOUTHEE 25
Characterisation of the Lithographic Printing of Conducting

Films
B. Ramsey, D. Harrison, D. Southee
School of Engineering and Design

Brunel University
United Kingdom
Abstract
This paper reports the electrical characteristics of conductive lithographic films (CLFs),
formed by lithographically printing silver-loaded inks onto a variety of substrates. Sheet-
to-sheet variability in electrical resistance, and electrical resistance variability due to
location of the feature on the sheet, and orientation of the structure on the sheet are
reported. Achievable production tolerances for the lithographic printing of conducting
films are presented; ±2.5 % variation in resistance over short print runs; ±2.5 % variation
in resistance from the leading to trailing edge of a sheet. The inks do not present any
anisotropic behaviour at a macroscopic scale and at microscopic (sub mm) scale
variations in resistance dependent on orientation are due to front and side lay errors.
Keywords: conductive lithographic film, low-cost, production tolerances, thick film

circuit fabrication
1 Introduction
A growing interest in printed electronics has led to the investigation of a number of graphical
printing processes for the manufacture of electronic devices. Work in this area has included the
use of ink jet printing for the production of thin film transistors [1] and the use of micro-contact
printing to pattern organic transistors [2]. It is known from the previous work of the authors that
conducting films printed lithographically may be used as interconnect in a wide range of circuits
and systems. Previous work has described Conductive Lithographic Film technology [3], and its
application to passive filter networks [4], microwave frequency structures [5], capacitor
structures [6], [7], and use with light emitting polymers as electrode structures [8], and as
capacitive type humidity sensors [9].
This paper describes a number of experiments performed to electrically characterise

lithographically printed films and aspects of the lithographic printing process itself. The
repeatability and variability in sheet resistance were measured in a number of experiments and
are discussed. The work has been carried out to enable production tolerances to be determined
for this new approach to circuit fabrication.
2 Background
This paper reports on the use of lithographic printing as an additive circuit fabrication process.
Conductive Lithographic Films (CLFs) consist of a thin layer (3-5μm) of conductive silver
particles suspended in a resin binder, which form conductive tracks suitable for the manufacture
of a variety of electronic circuits and systems.
The process employs standard lithographic printing technology and purpose-developed ink
pastes to fabricate conductive film patterns on a range of flexible substrates. The technology
offers very high production speeds and low material use.

26 CHARACTERISATION OF THE LITHOGRAPHIC PRINTING OF CONDUCTING FILMS
2.1 The Lithographic Printing Process:-

Lithography relies on the action of two wetting functions on the surface of a smooth and un-
embossed printing plate. The plate chemistry repels water where the printed image is dark,
allowing an oil-based ink to adhere. A water film repels the ink in light regions of the image.
Contact with an ink and a moistening roller allows the printing plate to attract both water and
ink as required, and to form the image to be printed. The image is not printed directly on the
substrate material (e.g. paper), but is instead transferred to an intermediate or blanket cylinder
that has a yielding surface. The blanket cylinder then presses the ink film onto the surface of the
substrate, which is supported on a separate impression cylinder. The printed substrates rely on
evaporation and/or oxidation of the ink film for the image to become fixed.
Standard lithographic printing machines have the following characteristics:
• High Speed (Typically 3000 - 10000 Impressions/hour).

• Good Dimensional Control & Excellent Registration of Images.
• Low Cost per Sheet (Low Ink Volume/Substrate).
• Widespread Availability of Necessary Production Facilities.
2.2 Ink:-
Analysis of ink rheology and conductivity during progressive print trials has resulted in the
formulation of conductive inks (by the authors), which have suitable mechanical and electrical
properties. The ink constitutes 80% silver particulate by weight (1μm mean size). The vehicle
portion of the ink comprises a large proportion of organic alkyd resin with traces of solvent and
anti oxidant agents. Ink layers deposited by standard printing machines have a thickness of the
order of 1 - 5 microns. When compared with 25-micron thick conductors resulting from screen-
printing processes, and with the 35 or 70 microns of copper laminated onto conventional circuit
boards, it is apparent that the developed ink must exhibit high electrical conductivity, whilst
conforming to the various mechanical and chemical constraints imposed by the printing process.
2.3 Printed Film Substrates:-

The type of substrate used must conform to several basic parameters if a printed circuit is to
have appropriate electrical characteristics and service life. These include environmental
stability, dimensional stability, appropriate surface roughness and energy and suitable dielectric
and resistive properties. Synthetic and coated papers have yielded the best results, the former
having superior environmental/dimensional stability. Of the substrate materials evaluated to
date FyneWhite Art (a coated art paper), “Teslin” and “PolyArt” (synthetic papers) offer the
least sheet-to-sheet variability in resistance. Teslin (constituted of silica, polyethylene and 80%
air by volume) offers the lowest resistance printed films due to its high porosity, which causes
the mobile components (solvents) in the wet ink to be drawn in to the substrate. This allows
greater ink transfer (including particulate) to the substrate, and better packing of the particulate
once deposited. ‘Fine White Art’ has a smoother surface than the other papers trialled resulting
in a lower variability of the measured electrical sheet resistivity of printed films.
2.4 Film Analysis:-

To develop a circuit model of the films suitable for circuit simulation purposes, a series of tests,
including analysis from D.C. to microwave frequencies has been undertaken. The constructs
behave as a typical metallic conductor and can be described using any standard conductor
model. Exact values of track resistance depend upon the particular ink formulation and the
substrate printed on. This paper considers sheet-to-sheet variability in resistance, and resistance
variability due to location and orientation of the feature on sheet.

3 Methodology and Results
All results and observations made here regarding conductive lithographically printed inks refer
to work under taken by the authors or other members of the research group.
3.1 General Press Set up:-

The samples reported here were printed by the lithographic print machine operators on
commercial Offset Lithographic machines (Heidelberg and Man Roland presses), The machine
set up and process was the same as would be expected for graphical work and the presses were
not mechanically modified. Initially the plates were run using a commercial graphic ink to set
sheet alignment and ink duct key settings. Key settings were adjusted to give a consistent
opacity across the image (i.e. to accommodate varying areas of printed image across the width
of the plate) and then opened globally to deliver as much ink as could be deposited without
causing strike off. The graphic ink was then replaced with the conductive ink and the ink
delivery again globally adjusted. Unless otherwise stated, all the measured samples were printed
once the press had ‘stabilised’ i.e. not at the beginning of the print run.
3.2 Sheet Resistance Measurement Techniques:-

The sheet resistance of a planar conductor can be calculated by simply measuring the resistance
presented at the short edges of a rectangle of the printed film, and dividing the result by the ratio
of the length of the short side of the rectangle to the long side.
The measurement jig utilised brass electrodes with specimen contact lengths of 55 mm. The
distance between the electrodes was variable from 5 to 150 mm. A toggle clamp fastener
enabled the rapid and accurate placement of the specimens and electrodes. Measurements were
taken using a calibrated RS M210 Low Ohm meter. The low ohmmeter employs a four-probe
measurement system, negating the resistance of the leads. A control sample of copper was used
to determine the internal and contact resistance of the electrodes, which was found to be ≤ 0.001
ohms.
The four point probe test structure has the following dimensions:
Length = 14000 microns

Width = 100 microns
Mean ohms/sq. (total) for CLF = 0.865
This gives a mean nominal track resistance of 140 x 0.865 = 121.1 ohms
A full description of four point probe measurements of the printed films can be found in
reference [10].
3.3 Overprinting:-
Overprinting of the samples as is typical with these conductive films, acts to reduce variability
over a run by ensuring deposition of as complete and homogenous a film as is possible. This
results in a film resistance that is primarily a function of the bulk resistivity of the dried ink, not
as is otherwise the case, a function of the bulk resistivity of the dried ink and the holes in the
inhomogeneous layer.

Table 1: Tolerances of sheet resistance for R3 ink printed on “GlossArt” substrate on a

commercial Hidelberg GTO46 by a professional machine operator.
Certainty Print passes:

3 2 1
100 % ± 6.5 ± 15 ± 80
95% ± 6.1 ± 13 ± 48
The figures in Table 1 are typical of the reduction in variation seen as a result of overprinting.
These results are taken from 180 sequential sheets, 120 were overprinted, and then 60 of those
overprinted a third time, the sheets to be printed each time were marked and place within the
stack to ensure the press was stabilised prior to the printing of the samples. The 95 % certainty
figures are the tolerances within which 95 % of the samples fall, and act as a mechanism to
exclude erroneous outliers in the sample group. The tolerance difference between the single pass
figures is resultant of one (of sixty) rogue sample. The overprinting conducted in these studies
has been wet on wet; the pre-printed film has not been dried before overprinting.
This study shows that overprinting has an enormous effect on the sheet-to-sheet variability of
the process. A film of three print passes can show an eight-fold reduction in variability of
resistance over a film of one pass.
3.4 Resistance Repeatability and Variability:

Variation of resistance of a particular printed ink and substrate film falls into three categories:
• Sheet to sheet variability;
• Location on sheet;
• Orientation on sheet.
3.4.1 Sheet-to-Sheet Variability:-

As a press is set-up and run it is necessary to perform fine adjustments to achieve an optimum
image. These fine adjustments match the uptake of ink by the roller train from the ink-duct to
that of the substrate. Inaccurate settings result in a gradual change of the image, over large
numbers of sheets. The volume of ink on the roller train acts as a reservoir, slowing
manifestation of changes made to the ink duct. In graphic printing this is apparent as a gradual
change in colour density of the printed image. At the same time as any adjustments are made,
the ink/water balance on the plate is equalising. Alteration of the quantities of either, through
manual adjustment of the delivery systems or through a change in substrate or environment will
result in differing depositions.

Figure 1: Sequential variation of resistance of consecutively printed samples (% deviation from

mean).
Figure 1 illustrates typical sheet-to-sheet variation of electrical resistance results using a manual
GTO 46 press. The vertical scale is the percentage deviation from the population mean. Chart
(a) represents a 35-sample sequence from the middle of a run, a point at which the press is
considered to be stable, and shows a tolerance of little over ±2%. Chart (b) illustrates a
downward trend over a sample size of 100, of which every fourth sample was measured and
plotted. This characteristic is typical of a press stabilising during the first print pass.
The sheet-to-sheet repeatability of the process is affected by ink type, substrate, press condition,
operator control, temperature and humidity. Over short runs (of a couple of hundred sheets) it
has been proved that these factors can be optimised and controlled to result in resistances within
tolerances of ± 2.5% over the length of the run. These results have been achieved on an old
press in a commercial printing environment. Substrate material has a significant effect upon
sheet-to-sheet variability. Of the materials used in this study FyneWhite Art (a filled glossart
paper) and “Teslin” offer the best repeatability.
3.4.2 Position Dependent Variation in Resistance:-

Variation in the quantities of graphic ink deposited over an area of image will result in visual
irregularities. In the case of resistive/conductive inks the result will affect the sheet resistivity of
the deposited layer. To accommodate artwork consisting of areas of varying image density, ink
delivery to the plate is controlled via a row of ink duct keys across the width of the image plate.
Each key is adjustable between fully closed and fully open. This facilitates deposition of a
variable profile of ink onto the first delivery roller. The ink is subsequently transferred via an
oscillating roller to the delivery train, which distributes it to the plate inking rollers. Alteration
of the time in which the oscillating roller is in contact with the first ink roller varies the quantity
of ink of a given profile deposited across the whole width of the plate.
Thus the two primary adjustments are:
• ink duct keys which control variation in resistance across the image (X axis),
• and the contact time of the oscillating roller, which determines the quantity of ink
available down the length of the image (Y-axis).
Variation in resistivity across the X-axis (width) of printed samples:-

On modern automated presses ink duct settings are adjusted automatically. A proof of the image
is scanned and split into discrete strips corresponding to the ducts across the width of the image.

The quantity of ink necessary for each strip is calculated and the duct adjusted. On smaller,
older machines, such as the GTO 46 used in these trials, this is an entirely manual operation.
Figure 2: Plot of variation in resistance across a 40 cm width of printed sheet. This example is
printed on “Polyprime” polyethylene paper.
Trials have achieved overall tolerances of about ±15 %. Figure 2 illustrates a plot of varying
resistance for a 1 mm-width track measured in 20 mm sections over a 400 mm length. The
dipped “bathtub” shape of the plot is characteristic of a conventionally adjusted press, with ink
ducts set to reduce the quantity of ink delivered to the edges of an image. It has been beyond the
scope of this work to utilise an automatic duct setting system, or electrically test samples and
readjust the duct settings accordingly. It is envisaged that both these strategies would reduce the
variation currently experienced. Consequently the limits of achievable tolerance have not yet
been established.
Variation in resistance over the Y axis (length) of printed samples:-

Variation in resistivity along the Y-axis of printed samples is a result of the physical design of
offset presses. As the plate cylinder rotates, the leading edge of the image will always be the
first part of the plate to receive ink from the inking rollers. As a consequence there will be
slightly less ink available at the trailing edge of an image. This problem has been largely
overcome by the design of the ink delivery roller train, which serves to completely saturate the
ink receptive regions of the plate. However in situations where too little ink is delivered and the
plate is not fully saturated, the deposited film will exhibit a resistive gradient, increasing toward
the trailing edge on the image.
Figure 3: Plot of variation in resistance along a 120 mm track printed along the Y-axis. This
example is printed on “GlossArt”.
The variation across the Y-axis of the samples is lower and distribution more uniform than
across the X-axis. Figure 3 illustrates results of 12 x 10 mm measurements along a 1 x 120 mm
track printed parallel to the Y-axis. The tolerance of ±2.5% demonstrates the good control
achievable over this parameter.
3.4.3 Orientation Dependent Variation:

It was considered possible that deposited films might exhibit anisotropic characteristics. Such
parameters could be associated with the orientation of particles within the ink as it is transferred
to and from the plate, and onto the substrate. In order to establish the presence or otherwise of

such a phenomenon, analysis of 100µm scale structures using four point probe forced current
measurements was conducted.
3.5 Four point measurements:-
Figure 4: Principle of four-point resistance measurement.
The principle of four-point measurement is illustrated in figure 4; the grey structure represents
the printed conductive film. A known current is passed from pads a to d and the voltage across
pads b and c is measured. As the current flow through the voltmeter is (virtually) zero the
voltage read is a result of the forced current and resistance of the printed film over length l. If
length l and the track width are known the sheet resistance of the specimen can be calculated.
Electrical measurements were taken using a HP4062B Semiconductor Parametric Test System
coupled to a Solartron 7065 Microprocessor Voltmeter. The system was programmed in HP
basic and samples were positioned under the test probes on a manually controlled X Y table. Six
test probes contact pads on the test structure and multiple measurements of sheet resistivity at a
point and resistance of a track were conducted.
Figure 5: Four-point probe test structure artwork including the cross and perpendicular line
structures (not to scale).
Two test structures resembling crucifixes were designed to enable point and line measurements
of resistance, perpendicular and parallel to the direction of print (figure 5). The structures are of
nominal 100µm line width and were formed of inks printed on “GlossArt” paper. Samples were
air dried at 100°C. The cross structure resistivity measurements are averaged measurements as
detailed in reference 8. Thirty samples of each ink were measured and the mean for each
structure recorded.
Point resistivity measurements were made at the 100 x 100 µm centre of the crosses. To
evaluate differences in resistance dependent upon direction of measurement, a process of
rotating the measurement was implemented. At each orientation eight measurements were made

and the mean value recorded (this is standard practice to reduce errors due to equipment and
contacts). The eight mean measurements (0° + bias, 0° - bias, 90°+ bias…) for each sample
form two groups, of the 0°+, 0°-, 180°+ and 180°- readings, and the remainder. That the
measurement in each of these two groups correlate is expected, as they are the same
measurement conducted from different polarities. The difference between the mean
measurements of both groups is the difference between the X and Y-axis resistance of the
sample.
Table 2: Mean difference in X-axis and Y-axis resistance of 100µm R3 Structures on

“GlossArt”(30 samples, 60 structures).
R3 on “GlossArt” Structure
Mean (Ω/sq.)(total) 0.865
Mean (Ω/sq.) 0.863 0.868
(X-axis) (Y-axis)
(A)The difference between the mean of all the X-
axis measurements and the mean of all the Y-axis 0.5 %
measurements, as a percentage of the total mean
(b)The mean difference between individual X and 48 %
Y-axis measurements, as a percentage of the total
mean
Results for the structures are presented in table 2. The correlation between parallel and
perpendicular resistances for the printed ink is excellent producing a 0.5 % deviation from the
mean (a). The mean of the differences (b) 48% (which does not consider the polarity of the
difference and cancel it out) are larger figures. The real life appliance of (b) is in reliability of
predicting the difference between the parallel and perpendicular resistance of a particular
100µm x 100µm point, where as (a) is indicative of likely differences of a large series of points
forming a conductive track.
A general observation that the mean resistances are higher than has been achieved in previous
print trials indicates less material may have been deposited than ideal, resulting in samples
which have a wider variation of resistance. A further factor that may have caused the larger than
expected point measurement variations is the small scale of the structure (100µm2) and the
accuracy of the front and side lay mechanisms of the press employed (about 30 µm).

Direction of print
Figure 6: Micrograph of lithographically printed conductive ink on “GlossArt” paper, Note the
50µm alignment error.
As is visible from figure 6, the effect of alignment errors and track edge roughness becomes a
significant portion of the width of the track. Measurements of the structure indicate a mean
difference in parallel and perpendicular track width of ≈ 20 - 30 µm (on the image above only
horizontal distances will correlate to the scale as the sample was tilted away from the electron
receptor).
These series of tests have illustrated that it is not likely that there is a significant difference in
the X and Y-axis conductivity of the printed films. The mean correlation between the two
measurements can be very good (0.5%), indicating that any anisotropic characteristics of the
printed ink, if present, do not contribute significantly to macro resistance measurements.
4 Conclusions
This work has determined the following tolerances of the printing process:
• ±2.5 % variation in resistance over short print runs;

• ±2.5 % variation in resistance from the leading to trailing edge of a sheet;
• ±15 % variation in resistance across the width of a sheet.
These figures represent the best achieved for each parameter using a collection of old
commercial GTO presses. As such they may be considered achievable manufacturing
tolerances. On old manual presses all three tolerances are partially dependent on the machine
operator, as a consequence more modern automated machines offer better results, particularly
the variation in resistance experienced across the width of a sheet due to ink key/duct settings.
Other factors leading to wider than expected variations may include:
• Poor deposition (resultant of low quantities of ink during printing) indicated by

higher than expected resistances, resulting in the resistances measured being a factor
of not just the bulk resistivity of the deposited films, but of the reduced cross-
sectional area of the track, resultant of the gaps in it.
• Registration errors during the overprinting of the samples. Printed feature sizes
similar in magnitude to alignment error of the press will result in larger resistances
than exactly aligned overprinted structures. This results in the irregularities in
readings for small structures (100µm) that become negligible when larger features are

used. Differences in accuracy of the front lay (X-axis) and side lay (Y-axis)
alignments are considered to be the primary source of difference in the perpendicular
and parallel measurements of microstructures.
It should be noted that the ink and substrate combinations that give very low tolerances do not
necessarily offer the lowest resistance. A case in point is “FyneWhite Art”, which offers low
sheet-to-sheet variation but results in higher resistance deposits than “Teslin”. Consideration
needs to be given to which is most suitable for a particular task, a ±2.5% tolerance on 0.5 Ω/sq.,
or ±4% tolerance on 0.15 Ω/sq. The latter will consistently give a lower resistance.
References
1 Sirringhaus H. et al., Science 290, 2123, 2000

2 Rogers, J. et al., Adv. Mat. 11, 741, 1999
3 Ramsey , B., Evans P. and Harrison D., A `Novel Circuit Fabrication
Technique Using Offset Lithography, Journal of Electronics Manufacturing, 7(1),
pp63-67, 1997
4 Evans, P., Ramsey, B. and Harrison , D.,.(1999) Frequency Selective Filter
Structures. IEE Electronics Letters, Vol. 35 No.4, p.306, 1999
5 Evans P., Ramsey B., Harrison, D. .& .and Shepherd, P.R., (1997) A Novel
Lithographic Technology for Microwave Integrated Circuits, I.E.E. Electronics
Letters. Vol.33, No.6, p483, 1997
6 Harrey , P., Ramsey B., Evans P. and Harrison D., (1999) A novel manufacturing
process for capacitors, , Eco Design 1999,Proceedings of 1st International
Symposium on Environmentally Conscious Design and Inverse Manufacturing,
Tokyo, Japan, p842-848, 1999
7 Harrey P., Ramsey B., Evans P. and Harrison D., Interdigitated Capacitors by
Offset Lithography, Journal of Electronics Manufacturing, 10(1), pp69-77, 2000
8 Lochun D., Samuel I., Ramsey B. and Harrison D., Post (1999)– Processing of
Conductive Lithographic Films for Electronic Interconnect, Twenty–Fourth
International Electronics Manufacturing Technology Symposium, Austin, Texas,
USA, 1999
9 Harrey P.M., Ramsey B.J., Evans P.S.A. and Harrison D.J., Capacitive-type
humidity sensors fabricated using the offset lithographic printing process, Sensors
and Actuators B 87 2002, pp226-232, 2002
10 Walton A.J., Stevenson J.T.M., Fallon M., Ramsey B.J., Evans P.S.A. and
Harrison D.J., Characterisation of Offset Lithographic Films Using Microelectronic
Test Structures, IEICE Transactions on Electronics. (Japan) E-82-C, No. 4 pp. 576-
581, 1999

MANFRED BREEDE 35
Grey Balance Control with a Re-Purposed GATF Colour

Circle
Manfred H. Breede
Ryerson University
Toronto
Canada
Abstract
The purpose of this study is to compare L*a*b* Greyness measurement accuracy with
that achieved by a proposed new system which uses a re-purposed GATF Colour Circle.
It is generally accepted that the visual or instrumental evaluation of grey balance targets
is an effective method to control colour balance for the four-colour printing process.
While the various grey balance measurement systems in existence use different
instrumentation and quantify greyness in a variety of numeric and graphic formats their
ultimate goal is identical. All grey balance measurement systems seek to measure and
display greyness variations between a standard and a process greyness.
The results of this study and similar research by other researchers, show that both the
L*a*b* system and the GATF Colour Circle system are equally effective in quantifying
and displaying differences between an actual and a desired greyness.
The proposed grey balance control system, if suitable integrated in a closed-loop ink feed
control system of a printing press, will minimize press operator intervention.
Keywords: Grey Balance, Colour Balance, GATF Colour Circle, Ink Feed Control
1 Introduction
All printing processes rely on the four-colour processes to reproduce chromatic originals and as
such have to attain a seemingly elusive balance between cyan, magenta, yellow, and black in
order to create a facsimile of the original.
A study published by Printing Industries of America Inc. in 2001, based on 960 four-colour
process jobs compiled from 1998 to 2000, gives us a glimpse of the extent of the problem [1].
The study distinguishes between planned and unexpected spoilage, where planned spoilage is
the anticipated spoilage inherent in the make-ready of a four-colour process job and where
unexpected spoilage is caused by not meeting a customer’s expectations.
According to the study planned spoilage for four-colour jobs was 185.0%%, 41%, 17.8%,
10.6%, and 4.7% for very short runs (<500), short runs (501-2,000), moderate short runs (2,001-
5,000), moderate runs (5,0001-10,000) and long runs (>10,000) respectively. The figures for
unexpected spoilage listed in the same order as above were 45%, 10.9%, 5.4%, 3.2%, 1.7%.
While the study does not implicitly state that spoilage was caused by unacceptable colour
balance, other studies consistently corroborate that colour imbalance is a leading cause for
spoilage in four-colour process printing. Further indication that the problem is related to colour
imbalance is the clear correlation between run lengths and spoilage. Colour balance is
established in the early stages of a press run and can only be achieved by repeated adjustments
to the ink feed system of a press while it consumes paper. Once colour balance is established it
will usually stabilize and therefore, as a percentage of the whole, short runs will be subject to
more spoilage.

36 GREY BALANCE CONTROL WITH A RE-PURPOSED GATF COLOUR CIRCLE
The balance between process colours has in the distant past (and still often practiced today)
been controlled by press operators adjusting the ink feed system of a press in accordance with
their visual assessment of the output as compared to an approved standard (OK sheet). The
shortcomings of this approach are that the colour balance assessment depends on the vagaries of
individual press operators’ colour perception and the slow response time of an operator to
required ink feed changes. Given that a reproduction could consist of thousands of colours and
shades of colours, press operators are faced with the insuperable task to establish the correct
balance between the process colours, and consequently the risk of make-ready spoilage and
spoilage due to unacceptable output is high.
The introduction of control strips that could be measured with instrumentations such as
densitometers and spectrophotometers greatly improved the efficiency and control of four-
colour process printing, because print characteristics such as density, dot gain, ink trap, print
contrast, greyness, and colorimetric values could be objectively measured for each individual
process colour, resulting in faster make-readies, better colour balance and reduced waste.
Additionally, in advanced press systems, the measured data can be compared to pre-set aim
points in order to control the ink feed system of a press automatically, thus in a best case
scenario, the requirement for press operators to make colour balance decisions and the physical
action of making the adjustments is eliminated.
2 Origins of the Grey Balance Concept
One of the first scientists to investigate the mathematical properties of autotypical colour
synthesis was Hans E.J. Neugebauer. His research culminated with the publication of his
seminal paper in which he proposed a series of equations, quantifying the amounts of ink
needed to reproduce colours in the four colour-printing process [2]. While Neugebauer did not
specifically investigate grey balance, he laid the foundation for a mathematical approach to the
four-colour printing processes. Other colour scientists continued this research to specifically
compute the cyan, magenta and yellow ink amounts that produce neutral greys, because it was
realized, that a primary requirement of four-colour reproduction is the ability to create a neutral
grey scale throughout a picture’s tonal range. If, for example a neutral scale had a greenish
colourcast, yellowish colourcast, reddish colourcast etc., likewise would all other hues in the
reproduction be affected by these colourcasts. Archer [3], Clapper [4], Preucil [5], Pobboravsky
[6] and other researchers contributed to this grey balance research, but their primary area of
research was in the realm of internal picture structure for the purpose of improving colour
separation technology as opposed to the control of colour balance during the printing phase.
3 The Application of Grey Balance Measurements to Controlling Printing

Press Ink Feed Systems
One of the first practical implementations of grey balance patches for the purpose of colour
balance control on printing presses was described in 1968, and was subsequently reprinted in
GATF Research Project Report No. 76 [7].
Another early commercially available print control strip incorporating grey balance elements,
which at that time was still assessed visually, was Felix Brunner’s 1973 Print Control Strip
(Figure 1) [8].
Figure 1: The 2nd (1973) Print Control strip for checking dot gain, colour balance, solid ink
density, trapping, high light dots. Source: System Brunner web site (www.systembrunner.com).

MANFRED BREEDE 37
In 1979, a subsequent Brunner control strip generation became the central part of E.I. du Pont
de Nemours’s well known Eurostandard Cromalin proofing system [9]. Further developments
led to Brunner’s Picture Contrast Profile [10] and Print Expert [11] grey balance control
systems, which place a high emphasis on dot gain differentials to evaluate grey balance, and
furthermore categorize reproductions according to their grey balance sensitivity. Brunner’s
system has gained acceptance in a number of print control systems foremost of which are
QuadTech and MAN Roland. Brunner continues to be a staunch proponent of densitometers as
being the most suitable instrument for grey balance measurements [12].
Other printing press manufacturers, notably Heidelberg Druckmaschinen AG have chosen

spectrophotometry as their preferred grey balance measuring system. Similar to the Brunner
system, Heidelberg also uses print control strips that incorporate grey patches (Figure 2.) to
measure deviations in the grey fields and report them in ΔE*a b values.
Figure 2: CPC System 4GS digital Version 1.30 SM52, Heidelberg Druckmaschinen AG.
A further development in printing press colour balance control is the spectophotometric

measurement of the image itself, rather than control strips. Heidelberg’s Prinect Image Control
is such system. Measuring the image itself represents a radically different approach to colour
balance control and is as such not related to the principles of grey balance discussed in this
paper.
4 Grey Balance Patch Composition
In theory, equal amounts of pure cyan, magenta and yellow inks superimposed over each other
will produce black, because pure cyan, magenta and yellow inks each absorb a different third of
the spectrum. We know this not to be occurring with real process inks. Equal amounts of actual
cyan, magenta and yellow inks will in fact produce a brownish hue, because each process colour
has unwanted absorptions. In halftone printing a given tone is produced by adjusting the dot
areas, which makes it possible to compensate for the impurity of the process inks. A study by
Preucil [13] investigated printed Greyness values produced by various process inks and found
that yellow dots require a greater dot size than magenta dots if more bluish magentas are being
used, while redder magentas require somewhat smaller yellow dots than magenta. As well,
greyer or less saturated cyan inks require larger dot sizes to create neutral greys. As a general
rule, to produce a neutral grey, the dot size for cyan must be larger than the dot sizes for
magenta and yellow and the dot sizes for yellow and magenta must be similar.
Various standard organizations and producers of commercially available print control elements
have defined the screen percentages required to produce neutral greys slightly different.
Depending on the printing process they cater to, as well as the paper and inks being used, dot
gain and colorimetric values will vary. For example the SWOP specifications calls for 50%
cyan, 39% magenta and 39% yellow to produce mid-tone neutrals, whereas GRACoL specifies
values of 50% cyan, 40% magenta and 40% yellow. SWOP is a standard organization that
represents heat-set web offset printers, while GRACoL serves the sheet-fed offset market. Since
web offset tends to produce more dot gain than sheet-fed offset, SWOP specifies grey
percentages that are 1% smaller for magenta and yellow.
Table 1 shows the grey balance values used by major standard organizations. The black screen
patches are typically adjacent to the grey balance patches as a visual reference grey. If both the
grey balance and the black patch are indistinguishable from each other, perfect grey balance,
and by inference, perfect colour balance in the reproduction is achieved.

Table 1: Grey Balance Patch Screen Percentages. *ISO 12642
SWOP [14] SNAP Grey Target recommendation [15]

Grey Balance Percentages Grey Balance Percentages
Black Cyan Magenta Yellow Black Cyan Magenta Yellow
75 75 63 63 40 34 32
50 50 39 39
25 25 16 16
GRACoL [16] Brunner [17]
75 75 63 63 50 50 41 41
50 50 40 40
25 25 16 16
GATF Digital Compact Colour Control Ugra/FOGRA Media Wedge CMYK, V2.0
Strip version 1.0, [18] for digital proofs, [19]
40 40 30 30 10 10 6 6
20 20 12 12
40 40 27 27
60 45 45 45
80 80 65 65
100 100 85 85
FOGRA ECI/bvd Grey Control Strip DMACS Canadian Specifications, issue 4.9.2,
for offset printing [20] [21]
70 65 57 50 50 40 40
50 45 36 33
30 27 20 18
5 Spectrophotometry vs. Densitometry for Grey Balance Measurements
Early investigation of grey balance was driven by a desire to improve colour separation
techniques, and as such has dominated this field for many years. Although the general principles
of grey balance in both the pre-press and press areas are the same, the applications are different.
In the former, grey balance is used to balance colours within the picture by adjusting the cyan,
magenta and yellow dot sizes, while in the latter grey balance finds application to balance cyan,
magenta, and yellow ink flow in order to maintain the colour balance that was established in the
pre-press stages. Spectrophotometers together with device independent colour spaces, such as
L*a*b*, are unsurpassed in their ability to quantify colour the way the human eye perceives it,
because they incorporate the three stimuli of human vision, spectral energy distribution of a
light source, spectral reflectance of an object, and spectral response of a detector (standard
normal observer). Because modern colour management requires the measurement of colour
patches (IT8 targets) to characterize the output devices in a workflow with device independent
colour spaces such as L*, a*, b*, spectrophotometers are the only technical option.
Unlike grey balance control in the pre-press area where the possibility for adjustments are
extremely flexible with regard to localized colour manipulation, the printing process is
relatively inflexible in that it can only apply overall amounts of cyan magenta and yellow. For
example it is not possible to decide at the printing stages to cause the highlight to be bluer and
the mid-tones to be redder. It is only possible to cause both the highlights and the mid-tones to
be bluer or redder, or put another way, both tones cannot be adjusted independent from each
other. This restriction does not apply to prepress procedures and therefore the printing stage

MANFRED BREEDE 39
requires a different and highly standardized approach to colour balance. This is not to say that
spectrophotometers and device independent colour spaces cannot be used effectively to control
colour balance during printing, rather it means that densitometers can be used just as effectively.
A rational goal for colour acceptability should be closely related to the limits of the human
visual system to detect colour differences. Because the ability of the human visual system is
exceptionally adept at detecting colour differences when comparing a standard and a sample
side by side, it is standard practice in the printing industry to approve press runs upon visually
comparing the printing press output with some type of standard such as a digital proof or a
previously printed sample. This approved printing press output is usually called an OK sheet
and becomes the standard for the press run. If the Greyness values of the approved printing
press output have been measured and recorded, the aim will henceforth be to maintain the
Greyness values of the printing press output as closely as possible to the Greyness values of an
OK sheet, which virtually assures colour balance in the printed image.
Greyness deviations can be measured by spectrophotometers and quantified in ΔE*ab units, with
adequate accuracy [22]. However densitometers are also effective because of the sufficiently
good correlation between optical density units and ink amounts [23]. Given that optical densities
are a good measure of ink amounts, it stands to reason that the proportions of cyan, magenta and
yellow inks, that produce Greyness, can be monitored and controlled with densitometers.
An analogy from the world of cooking will be pertinent to this line of reasoning. In cooking the
aim is to satisfy the sense of taste, which is not unlike the world of printing where the objective
is the sense of sight. If, for instance a particularly tasteful dish was prepared from a recipe that
calls for three ingredients of known weights, one could reconstitute the dish by making a
complete chemical analysis of the dish. Although this approach would probably work, it would
be an excessively complex method to accomplish a relatively simple task. The obvious and
more sensible approach would be to use the exact same weights as specified in the recipe for
each of the three ingredients. The former approach is akin to the technologically complex
spectophotometric analysis of grey balance, while the latter approach is akin to using the much
simpler densitometric technology for the same task.
6 The GATF Colour Circle
The GATF Colour Circle was first introduced in 1957 in GATF Research Report No. 38 [24]
and is as such the oldest of the GATF colour diagrams. All of the GATF diagrams, including the
GATF Colour Hexagon and the GATF Colour Triangle are adaptations from other colour
notational systems using cyan, magenta, yellow solid ink density units as their basic input
values. They were primarily developed to show hue and chroma variations induced by printing
process related phenomena, such as ink trapping, colour sequence, ink strength, and substrate
gloss or absorbency.
The GATF Colour Circle (Figure 3) displays two colour dimensions called Hue Error and
Greyness, where Hue Error is plotted in the circumferential direction and Greyness is plotted in
the axial direction of the circle. The input (densities) and output (Hue Error and Greyness)
values are shown in Table 2. Similar to most colour spaces, colours found near the periphery of
the circle are more saturated than colours found toward the center of the circle. The circle is
transected by three axis, which represent the perfect primary and secondary colour coordinates
with regard to their hues. Each axis is shared by a primary and its secondary colour
complement. Thus cyan and red, magenta and green, and yellow and blue share the same axis.
Densitometers are made specifically for process colours and will measure the light absorption of
a process colour through a filter colour that is complementary to the colour measured. Thus
cyan, magenta, and yellow are measured through red, green, and blue filters respectively. The
complementary filter colour of the process colour measured will always yield a higher density

value than the other two filters, but the other two filter values are almost never zero. This means
process colours have unwanted absorptions or are, in other words, colorimetrically impure.
The formulas to calculate Hue Error and Greyness are as follows:
Hue Error = ⎛High Density - Low Density ⎞ x 100

Mid Density - Low Density
(1)
⎝ ⎠
Low Density
Greyness = ⎛High Density⎞ x 100 (2)
⎝ ⎠
Table 2: Density, Hue Error and Greyness
Filters R G B Hue Grey

Error
Cyan 1.40 0.53 0.27 23.01 19.29
Magnta 0.21 1.45 0.58 29.84 14.48
Yellow 0.03 0.05 1.10 1.87 2.73
Red 0.22 1.35 1.55 84.96 14.19
Green 1.44 0.14 1.18 80 9.72
Blue 1.51 1.62 0.61 89.11 37.65
Figure 3: GATF Colour Diagram
A colour will fall on its axis only if the two unwanted absorptions are equal. If a colour has
unequal absorptions, as is invariably the case, it will have a Hue Error.
Consider the example in Figure 3 and Table 2 where a cyan has density values of 1.40, 0.53 and
0.27 for the red, green, and blue filters respectively. The Hue Error is determined by the middle
value, which in this case is 0.53.
In the diagram the colour would then be plotted in the radial direction that is closest to the
process colour with which the measured colour is contaminated. In the foregoing example this
would mean that the cyan, having a middle value (0.53) for the green filter value, has a magenta
Hue Error and will therefore be plotted in the radial direction that is closest to magenta.
Secondary colours are plotted similarly, but since in the four-colour process, they are produced
by two primary colours their Hue Errors are a measure of the equality of both primary colours
that produce it. For example, if a red has density values of 0.22, 1.35, 1.55 in the red green and
blue filter positions respectively, then the red will be more yellowish by virtue of its higher blue
filter value (1.55) and will therefore be plotted toward the radial direction of yellow.
Greyness values are a function of the lowest density reading, because equal absorptions of red
green and blue light, which is indicative of a neutral grey, are possible only when measured
relative to the lowest filter value. This grey component in the colour will tend to de-saturate the
colour, causing the plot to move in the axial direction toward the center of the circle.

MANFRED BREEDE 41
7 Adaptation of the GATF Colour Circle to Intrinsic Greyness Measurements
Using the GATF Colour Circle for the purpose of controlling grey balance during the printing
process is a rather novel application for this colour space [25], because it was originally
developed for the measurement and display of the saturated primary process colours cyan,
magenta, and yellow, as well as their secondary colour derivates red, green and blue. Greyness
in the context of the original colour circle’s application means the dulling or de-saturation of a
measured colour, and not the measurement of Greyness itself. Unlike colormetrically expressed
Greyness, Preucils’s term Greyness does not take the illuminant and Standard Normal Observer
into consideration, but relies strictly on the lowest density value, which combined with equal
amounts of the other two density values constitutes the grey component of a colour. To make
the GATF Colour Circle useful for intrinsic Greyness measurements the red, green and blue axis
have to be reassigned a modified function. Unlike the original purpose of the colour circle,
which is to plot the three primary and secondary colours, intrinsic Greyness measurements
require only the measurement of the three primary colours. Therefore, the red, green and blue
axis are no longer input values but specifically indicate the output or hue shifts of a given
Greyness. Otherwise, the Hue Error and Greyness values are calculated as per equations (1) and
(2) and likewise the circle’s interpretation remains unchanged. Figure 4 shows the position of a
Greyness measurement that yielded 0.66, 0.56 and 0.46 for the red, green and blue filters
respectively, while Table 3 shows the input (density) and output (Hue Error and Greyness)
values. Using equations (1) and (2) we will get a Hue Error of 50% and a Greyness of 69.7%.
Since the highest value was found for the red filter density the dominant process ink in the
grey patch measured is cyan.
Table 3: Density, Hue Error, and Greyness
Hue
Filters R G B Grey
Error
Cyan 0.66 0.56 0.46 50 69.7
Figure 4: GATF Colour Circle

Greyness plot having a cyan colour cast
8 Using the GATF Colour Circle for Press Run Colour Balance Control
Suppose an approved (OK sheet) four-colour process job was measured for Greyness, and
yielded 0.62, 061, and 0.60 densities for the red, green and blue filter values. The resulting Hue
Error and Greyness values are therefore 50% and 96.77% respectively. At this point it must be
remembered that the job was not approved because of its Greyness, although it is reasonably
good, but because of the visual acceptability of the reproduction. Henceforth, the aim will be to
control the ink flow in such a way that the above red, green and blue filter densities are
maintained as close as possible, which in turn infers that the initially accepted quality is
maintained because of the Greyness/colour balance correlation discussed earlier. In essence,
the Greyness reading of the initially approved press sheet serves as a benchmark for the entire
press run. In Figure 5, the initial Greyness of the OK sheet is shown as a black plot, while three
press sheets’ Greyness readings taken during the press run are shown by plots that are coloured

according to their colourcasts. Table 4 lists the corresponding density, Hue Error, Greyness and
Delta values.
Greyness values express the magnitude of the deviation from the approved standard, both
numerically and graphically, while the ΔR, ΔG and ΔB values are used to control the ink flow
of the press. In concrete terms, this means press sheet No. 1 requires more magenta and yellow,
press sheet No. 2 requires the ink flow to be reduced to all colours but mostly to magenta and
least to yellow, and press sheet No. 3 requires also less ink flow in all colours but especially
less yellow and magenta, to reestablish the approved colour balance.
Figure 5: GATF Colour Circle Greyness plots clustered around a standard Greyness.
Table 4: Density, Hue Error, Greyness and Δ values
Filters R G B HE Grey ΔGrey ΔR ΔG ΔB
Grey (standard) 0.62 0.61 0.60 50 96.77
1. Grey (cyan hue) 0.62 0.54 0.50 33.33 80.65 -16.12 0 -0.07 -0.1
2. Grey (mag. hue) 0.65 0.70 0.61 44.44 87.14 -9.63 0.03 0.09 0.01
3. Grey (yell. hue) 0.66 0.73 0.75 77.78 88 -8.77 0.04 0.12 0.15
9 Experimental Method
The aim of the study was to measure Greyness values sampled from an experimental press run
and to determine whether a correlation with ΔE*a b exists. The experimental press run was
conducted under normal print conditions for sheet-fed offset lithographic printing of four-colour
process reproductions, while simultaneously aiming for nominally perfect grey balance. In the
context of this study normal print conditions means the measurable print characteristics as
defined by GRACoL (General Requirements for Applications in Commercial Offset
Lithography) [26], and nominally perfect grey balance means equal reflection densities through
the red, green and blue filters of a densitometer. The overriding emphasis however, was to
achieve nominally perfect grey balance, even if print conditions deviated from those stipulated
in GRACoL.
A test form (Figure 6) incorporates the following critical components:

MANFRED BREEDE 43
1. A grey balance strip composed of 50% cyan, 41% magenta and 41% yellow dot sizes
extending across all ink zones of the press.
2. An image containing memory colours, in particular skin tones, an image that is less
sensitive to colour shifts (i.e. a profusion of many colours) and an image that is highly
sensitive to colour shifts (i.e. very neutral colours).
3. Solid and 50% tint screen strips for all four process colours extending across all ink
zones of the press, in order to measure and control solid ink density and dot gain.
The test form was printed on a Heidelberg Quickmaster DI, waterless offset press, using 148
g/m2, M-Real, Euro, Art Gloss, D.I., coated offset paper. The inks used, were waterless inks by
Rycoline and the lay down sequence was K,C,M,Y.
Nominally perfect grey balance was achieved in one ink zone, against which all other readings
were compared. The same press sheet on which nominally perfect grey balance was found had
sufficient grey balance variation to produce 40 readings of continuously diminishing Greyness
values and will for the purpose of this study be referred to as Uncontrolled Greyness Variation
(UGV). Subsequent to finding a nominally perfect Greyness and its L*a*b equivalent, the
Uncontrolled Grey Balance Variation was measured as close as possible to 0.20 ΔE*ab
intervals.
The same press run also produced grey balance differences that were caused by gradually
increasing the ink flow to the magenta plate only. This resulted in 15 readings from 15 different
press sheets and will henceforth be called Controlled Greyness Variation (CGV).
All measurements were made with an X-Rite 530 spectrodensitometer. Every location
measured was simultaneously recorded in both L*a*b*, D50/10, and the three filter densities
(Status T), which were subsequently converted to Hue Error and Greyness values. The raw data
can be seen in Appendices A and B.
10 Evaluation Method
In view of the fact that colourmetric values derived from spectrophotometric measurements are
generally recognized to be an accurate method of quantifying colour differences, the proposed
Greyness measurement system will be evaluated by its nearness to it. Therefore, Greyness
values will be statistically evaluated using linear regression analysis, where ΔE*a b, is the
independent and ΔGreyness is the dependent variable. Additionally, an ANOVA analysis will
determine whether the association between ΔE*a b and Greyness is statistically significant.
Greyness and Hue Errors will also be compared to a* and b* to determine if similarities exist in
the way the L*a b and GATF Colour Circle colour spaces register colour shifts. The equations
used are defined as follows:
ΔL* = L* Sample - L* Standard (3)
Δa* = a* Sample - a* Standard (4)
Δb* = b* Sample - b* Standard (5)
2 2 2
ΔE* = ΔL* + Δa* + Δb* (6)
ΔGreyness = (Greyness Sample – Greyness Standard) x -1* (7)

*
ΔGreyness is converted to a positive value in order to create the quantitative equivalent of
ΔE*a b.

-------Trailing Edge------
---------Leading Edge--------
Figure 6: Test Form

MANFRED BREEDE 45
11 Print Quality Analysis
For this experimental press run, the densities and the 50% mid tone dot gain values producing
nominally perfect grey balance were as follows:
Solid Ink Density Dot Gain

Cyan 1.31 32%
Magenta 1.26 31%
Yellow 0.91 39%
whereas the grey bar produced the following densitometer readings:
Visual filter reading 0.84

Red filter reading 0.78
Green filter reading 0.78
Blue filter reading 0.78
thus using equation (2), a nominally perfect Greyness of 100% was achieved.
The L* a* b* equivalents are: L* = 45.61

a* = -1.75
b* = -3.61
The unusually high dot gain value for yellow can be explained by the fact that Heidelberg’s
direct imaging system uses a yellow halftone line screen ruling that is significantly finer than
the 175 lines/inch used in cyan and magenta. This is done, to prevent banding, which can
otherwise occur at the relatively low image resolution of 1,270 dots per inch. Since finer line
screen rulings are known to cause more dot gain, yellow’s dot gain must necessarily be high.
For this and other reasons, that are beyond the scope of this study, densities and dot gain values
are deviating from those recommended in GRACoL. Nonetheless, given the aforementioned
definitions, nominally perfect Greyness was achieved by virtue of equal red, green, and blue
filter readings. The equivalent L*a*b* values are however showing that the grey is not
perfectly neutral. Here again, an explanation is in order.
In three-colour process printing it is nearly impossibility to produce grey patches having three
equal densities, while in photographic imaging, this is easily achievable. For example a 24-step
Kodak Reflection Density Guide (continuous tone), if measured at the approximate density step
of the experimental press run’s grey density, produces the following densitometer readings:
Visual filter reading 0.83

Red filter reading 0.83
Green filter reading 0.83
Blue filter reading 0.83
and the L* a* b* equivalents are: L* = 42.42

a* = 0.08
b* = 0.66
The Kodak Reflection Guide L* a* b* values indicate clearly a more neutral grey than the grey
produced by the experimental press run, because its a* and b* values are closer to zero, yet this
difference is not recorded using Greyness equation (2), which calculates 100% Greyness for
both. Spectrophotometric measurements converted to the L*a*b* colour space are therefore
more accurate than densitometric measurements converted to Greyness values. The reason is
however, not necessarily an indication of inherently less accurate densitometric measurements,
but that this densitometer displays densities only to two decimal places of accuracy. Had the

densitometer a three decimal place accuracy, it probably would have registered the difference
between the two greys. This discussion should be born in mind for a better understanding of
some forthcoming test data.
12 Results and Discussion of Uncontrolled Greyness Variation (UGV) Data
To describe the relationship between the ΔE*a b and ΔGreyness both sets of data were plotted on
a scatter diagram fitted with a regression line. There are 40 pairs of data points ranging from 0.1
ΔE*a b vs. 0 ΔGreyness, to 7.87ΔE*a b vs. 15.56 ΔGreyness. An important consideration was the
selection of an interval for the independent variable ΔE*a b, because it would be pointless to use
ΔE*a b intervals that are beyond the human visual system’s ability to perceive colour.
A NPIRI Task Force on Colour Measurement [27] testing human subjects found that colour
difference perception is highly dependent on the colour viewed. According to the NIPIRI study,
70% of subjects perceived just noticeable colour differences from a low ΔE* a b value of 0.3 for
greys to a high ΔE* a b value of 5.0 for yellows. For this reason, aiming for intervals of 0.20
ΔE* a b seems reasonable. The actual achieved intervals for this study ranged from a low of 0.05
to a high of 0.49 ΔE*ab resulting in an overall average of 0.211.
Examining the data in Table 5, and the scatter diagram in Figure 7, it can be seen that a clear
relationship between ΔE*a b and ΔGreyness exists. The regression line fits the data reasonably
well and therefore in general, as ΔE*a b increases, ΔGreyness increases too. The strength of this
relationship can be determined by calculating the coefficient of determination, shown in Table 6
as the value of r2 = 0.979106. This value, ranges always from 0 to 1, and tells us what proportion
of the change in ΔGreyness can be explained by a change in ΔE*ab. Expressed as a percentage,
we can therefore state that 97.91% of a change in ΔGreyness is attributable to a change in
ΔE*a b.
The statistical significance of a test statistics can be determined by performing an ANOVA

analysis of the data, seen in Table 6. Accordingly, the result of the F-test produces a
Significance F value of 1.56E-33. (Table 6). Since Significance F is <0.01 the null hypothesis,
or the assertion that no statistical association between ΔE*a b and ΔGreyness exists, can
therefore be rejected at a 99% confidence level.
The overall results show a strong ΔE*a b and ΔGreyness statistical association, but seen also on
the scatter diagram are periodic data points that are not randomly distributed above and below
the regression line. These clusters of horizontal bands are an indication that the densitometer
does not register Greyness variations at the same level of sensitivity as the spectrophotometer,
because of the aforementioned lack of accuracy beyond a 2nd decimal place.

MANFRED BREEDE 47
Table 5: ΔE*ab vs. ΔGreyness (“UGV”)
No. ΔE ΔGreynes No. ΔE ΔGreynes

s s Delta E vs. Delta Grayness
1 0.1 0 21 4.06 8.33
2 0.43 1.27 22 4.29 8.33 18
3 0.66 2.53 23 4.52 9.41 16
4 0.82 2.53 24 4.75 9.41 14
5 1.1 2.53 25 5.04 10.59
Delta Grayness
12
6 1.27 2.53 26 5.31 10.34 10
7 1.47 2.53 27 5.5 9.3 8
8 1.52 3.75 28 5.68 10.59 6
9 1.65 2.53 29 5.91 10.59 4
10 1.82 3.8 30 6.21 10.59 2
11 2.07 5 31 6.39 11.63 0
12 2.22 5 32 6.55 11.76 0 2 4 6 8
13 2.42 5 33 6.69 12.5 Delta E

14 2.69 6.25 34 6.87 12.5
15 2.84 6.25 35 7.11 12.5
16 3.07 6.17 36 7.26 13.5 Figure 7: Scatter diagram and
17 3.3 7.41 37 7.4 14.61 regression line of ΔE*ab vs. ΔGreyness
18 3.52 7.41 38 7.64 14.61 (“UGV”).
19 3.67 6.1 39 8.13 15.56
20 3.93 8.33 40 8.33 16.48
Table 6: Regression Statistics and ANOVA Table (UGV).
SUMMARY OUTPUT
Regression Statistics
Multiple R 0.989498
R Square 0.979106
Adjusted R
Square 0.978556
Standard
Error 0.637164
Observations 40
ANOVA
Significance
df SS MS F F
Regression 1 722.9373 722.9373 1780.732 1.56E-33
Residual 38 15.42715 0.405978
Total 39 738.3644
Standard Upper
Coefficients Error t Stat P-value Lower 95% 95%
Intercept 0.683859 0.200544 3.41001 0.001552 0.277878 1.08984
X Variable 1 1.782447 0.042239 42.19872 1.56E-33 1.696938 1.867956
13 Results and Discussion of Controlled Greyness Variation (CGV) Data
A further objective of this study was to evaluate the relationship between a known cause of
colour shifts and the corresponding effects on the L*a*b* and GATF Colour Circle colour
spaces. To this end, the experimental press run included a period in which the magenta ink feed
was increased, thus causing an intentional colour imbalance. Whereas the ΔE intervals for the
previously described tests were selected on the basis of just noticeable colour differences, in this
series of tests the intervals were selected on the basis of colour acceptability. In 1990,

Heidelberg Druckmaschinen AG demonstrated the use of spectrophotometry for grey balance

measurements of press sheets at DRUPA [28] and later categorized ΔE units according to their
acceptability (Table 7) [29].
Table 7: ΔE Categories
ΔE between 0 and 1 In general, this deviation cannot be perceived

Very small deviation; only perceivable
ΔE between 1 and 2
by an experienced eye.
Medium deviation; perceivable even by an
ΔE between 2 and 3.5
inexperienced eye.
ΔE between 3.5 and 5 Large deviation.
ΔE exceeding 5 Massive deviation.
An extensive study of colour variation involving 9 web offset and 9 gravure press runs ranging
from 50,000 to 450,000 copies proposed ΔE tolerances shown in Table 8. [30].
Table 8: Proposed ΔE Tolerances for Gravure and Web Offset.
Gravure Web Offset

Within production run deviation 3.0 ΔE 5.5 ΔE
Average deviation from the OK sheet 1.8 ΔE 3.3 ΔE
To echo these acceptability tolerances, this part of the study uses 1.0 ΔE*ab intervals (Table 9-
10 and Figure 8). The actual achieved intervals for this study ranged from a low of 0.56 to a
high of 1.66 ΔE*ab resulting in an overall ΔE*ab average of 1.078.
Table 9: ΔE*ab vs.

ΔGreyness (CVC)
No. ΔE ΔGreyness
Delta E vs. Delta Grayness
1 2.23 3.75
2 3.28 4.94
25
3 4.51 6.02
Delta Grayness
4 5.21 7.06 20
5 6.87 7.95 15
6 7.85 8.89 10
7 9.13 10 5
8 10.53 11.71
0
9 11.57 12.5
0 5 10 15 20
10 12.55 15.46
11 13.76 15.84 Delta E
12 14.61 16.83
13 15.17 17.48
14 16.16 18.27 Figure 8: Scatter diagram and regression line of ΔE*a b
15 17.33 19.44 vs. ΔGreyness (CGV)

MANFRED BREEDE 49
Table 10: Regression Statistics and ANOVA Table (CGV).
SUMMARY OUTPUT
Regression Statistics
Multiple R 0.994963
R Square 0.989952
Adjusted R
Square 0.989179
Standard Error 0.543531
Observations 15
ANOVA
Significance
df SS MS F F
Regression 1 378.388 378.388 1280.823 2.25E-14
Residual 13 3.840535 0.295426
Total 14 382.2285
Standard
Coefficients Error t Stat P-value Lower 95%
Intercept 1.089446 0.329094 3.310439 0.005633 0.378482
X Variable 1 1.059952 0.029617 35.78858 2.25E-14 0.995968
High ΔE*a b values are typical early in the four-colour process make-ready phase, because
colour balance has not yet been established. Consequently, an accurate method to measure grey
balance is particularly critical at this stage, when the risk for spoilage is highest.
This set of data has a somewhat higher r2 value than the “Uncontrolled Grey Balance Variation”
data. Also the clusters of horizontal bands are no longer apparent. Overall, these results show
that an interval increase from 0.30 to 1.0 ΔE*ab units, causes the correlation between ΔE*ab and
ΔGreyness to become even stronger.
The marginally higher Significance F value is for all intents and purpose still zero, and can be
explained by the smaller sample size, but the statistical significance of the test statistics is still
maintained at a 99% confidence level, because the Significance F value 2.25E-14 < 0.001.
When a*b* and Greyness and Hue Errors are plotted in their respective colour spaces it can
clearly be seen that both colour spaces indicate colour shifts similarly (Figures 9-10). The a*b*
colour space indicates mainly a shift toward red and somewhat toward the blue direction, which
is indicative of a magenta shift, while the GATF Colour Circle indicates a strong magenta shift
increasingly drifting toward red. Furthermore, both colour spaces accurately register the process
related occurrence which led to the colour imbalance in the first place. It will be recalled that
during the experimental press run, ink flow to the magenta plate was intentionally increased. It
follows that grey balance corrections in reverse, i.e. from a grey with a colour cast toward a
neutral grey standard, can also be controlled with the GATF Colour Circle.

a*b* b
Yellow
10
2
-a a
0
Green Red
-15 -10 -5 0 5 10 15
-2
-4
-6
-8
-10
-b
Blue
Figure 9: a*b* (CGV) Figure 10: Greyness and Hue Error (CGV)
14 Related Grey Balance Research
Although the principles of grey balance have been known for many years, research in the area
of instrumental grey balance control and in particular the use of densitometers for grey balance
measurements has been accorded relatively little attention. In part, this deficiency is attributable
to the printing industry’s long standing reliance on visual grey balance evaluations, as well as
the lack of a suitable density based measurement space. When spectrophotometers came to be
used more widely for printing process control, colour spaces such as CIELAB became the grey
balance measurement method of choice, causing densitometers undeservedly to be regarded as a
somewhat secondary technology.
Nonetheless some researchers, foremost of whom the aforementioned Brunner dealt with the
use of densitometers for grey balance control extensively. Brunner created a grey balance space
in the shape of a hexagon (Figure 11), where each of the six corners of the hexagon signifies
one of the three primary and secondary colours. Furthermore, the hexagon encompasses three
zones of grey balance sensitivities, reflecting the fact that various types of original pictures have
different sensitivities to colour shifts. Brunner classifies pictures with a profusion of colours,
such as a still life of fruit for example as relatively insensitive to colour shifts, an average
picture as having moderate sensitivity to colour shifts and relatively colourless pictures such as
a snowy scene as having a very high sensitivity to colour shifts. The measurement technology
used in the Brunner systems is a densitometer, but densities are converted to dot gain values.
Brunner allows a dot gain divergence in the primary colours cyan, magenta and yellow of 6%.
4%, and 2% for pictures of low, medium and high colour shift sensitivities respectively. To date
the Brunner systems of grey balance control is considered to be the most sophisticated grey
balance control system using densitometers.

MANFRED BREEDE 51
Magenta 6% Dot Gain

Divergence
4% Dot Gain Divergence
Blue
Red
2% Dot Gain
Divergence
Cyan Yellow
Green
Figure 11: System Brunner Grey Balance Control Space
Another researcher with a theoretical interest in grey balance measurements by means of

densitometers is Tritton. While Tritton considers spectrophotometers more effective than
densitometers, he allows the use of densitometers as a practical alternative. Two comparisons
between dot area divergences and ΔE in neutral mid-tones and neutral shadows are shown in
Figure 12 [31].
Mid tone neutral, Cyan = 50% Shadow neutral, Cyan = 70%

Dot area divergence
Dot area divergence
20 20
from neutral (%)
from neutral (%)
16 16
12 12
8 8
4 4
0 0
0 4 8 12 16 20 24 28 0 4 8 12 16 20 24 28
Color difference (Delta E) Color difference (Delta E)
Figure 12: Dot area divergences from a true neutral grey in mid-tones and shadows versus
ΔE (printed with 60 lines/cm screen ruling on coated paper).
Noticeable in the graphs is the range of dot area divergences that produce the same ΔE values.
Tritton offers the following three explanations for this phenomenon: 1. grey balance requires
different amounts of ink; 2. variations in tone transfer between colours; 3. differences in
lightness between the th ree colours. He concludes however, that for small colour differences
below ΔE = 6, the correlation between dot area divergence and ΔE is approximately linear.
As a direct result of this author’s presentation on grey balance during the 58th Annual Technical
Conference in Vancouver, interesting and relevant research data emanated from discussions
with other researchers. Subsequently, John MacPhee of Baldwin Technology Company, Inc.
generously shared his grey balance research data, which he conducted on behalf of his
employer. The tests were conducted on the premises of the Graphic Arts Technical Foundations
(GATF) involving a press run under controlled conditions. The aim of the test was to print a
GATF grey balance chart consisting of a six by six grey balance matrix, where the amount of
cyan was a constant 50% (film dot area) throughout and the amounts of yellow and magenta
varied in the horizontal and vertical directions by 2% (film dot area) increments. The most
neutral field was selected on the basis of spectrophotometric measurements, and was quantified

in the CIELAB colour space. Six additional spectrophotometric measurements were made in
grey balance fields in which the magenta dot areas diverged from the “perfect” neutral field, and
seven more fields were measured in grey balance fields in which the yellow dot areas diverged
from the “perfect” neutral field. ΔE calculations were made (i.e. the difference between the
“perfect” neutral grey field and the less than perfect grey balance fields). Additionally, all fields
were measured through the red, green, and blue filters of a densitometer and the lowest value
was subtracted from the highest value. To analyze the results, the max-min densities were
plotted against ΔE as seen in Figure 13 [32].
Figure 13: Relationship of ΔE and corresponding differences in maximum and minimum

density readings of CMY neutral areas.
These results are consistent with those that have been presented previously when ΔE was
plotted against ΔGreyness, which is, as will be recalled, calculated from the maximum and
minimum density values. The MacPhee study also shows that the rate of change is not equal in
all process colours, but that magenta experiences a significantly higher rate of change than
yellow. MacPhee reasons that because typical process magentas are highly contaminated by
yellow and typical process yellows are comparatively pure, greater colour changes are
necessarily produced by magenta when all other factors are equal. Tritton [33] made a similar
observation, when he states: “There is measured evidence and practical experience to indicate
that the magenta inking level has a greater influence in achieving a neutral grey balance.” This
exacerbates colour balance because magenta’s greatest absorption is in the green spectral
regions, which are known to be the wavelengths where human colour vision sensitivity peaks.
In order to further validate the results of the ΔE vs. ΔGreyness study that is the subject of
investigation in this paper the maximum – minimum density values were extracted from the
Controlled Greyness Variation (CGV) data set and were compared to the magenta results of the
MacPhee study (Table 11).

MANFRED BREEDE 53
Table 11: Maximum – minimum densities and corresponding ΔE values of the MacPhee and
Breede studies
MacPhee Study Breede Study

Max – Min Max – Min
ΔE ΔE
Density Density
0.016 1.69 0.03 2.23
0.027 2.24 0.04 3.28
0.029 3.51 0.05 4.51
0.043 5.12 0.06 5.21
0.045 4.11 0.07 6.87
0.058 4.98 0.08 7.85
0.09 9.13
0.11 10.53
0.12 11.57
0.15 12.55
0.16 13.76
0.17 14.61
0.18 15.17
0.19 16.16
0.21 17.33
Each study is shown separately in Figures 14 and 15, while the results of both studies when
combined in the same graph is shown in Figure 16.
MacPhee Study
6
5 McPhee
4
Delta E
3
Least
2
Squares
1 (McPhee)
0
0 0.02 0.04 0.06 0.08
Difference betw een maximum and minimum density
readings
Figure 14: Maximum – minimum density readings vs. ΔE of the MacPhee study.

Breede Study
20
15 Breede
Delta E
10
Least Squares
5 (Breede)
0
0 0.05 0.1 0.15 0.2 0.25
Difference betw een maximum and minimum density
readings
Figure 15: Maximum – minimum density readings vs. ΔE of the Breede study.
MacPhee and Breede Studies
20 Breede
15
Delta E
McPhee
10
5 Least Squares
0 (Breede)
0 0.05 0.1 0.15 0.2 0.25 Least Squares
Difference betw een maximum and minimum (McPhee)
density
readings
Figure 16: Maximum – minimum density readings vs. ΔE of the MacPhee and Breede studies.
The results of both studies suggest that CMY grey fields which vary in magenta content from a
nominally perfect CMY grey field, change approximately linearly with respect to ΔE vs.
ΔMax./Min. density values.
15 Conclusion
Linear regression analysis of ΔE*a b vs. ΔGreyness shows a strong correlation, but at an
approximate sampling interval of 0.20 ΔE*a b, ΔE*a b has a greater sensitivity to colour variation
than ΔGreyness. Increasing the sampling intervals from 0.20 to 1.0 ΔE*a b, results in both ΔE*a
b and ΔGreyness having about the same sensitivity to colour variation. The colour variation
sensitivity of Greyness units could probably be improved if densitometers with 3rd decimal
place accuracy are used, but from a colour acceptability point of view, a 1.0 ΔE*ab accuracy is
adequate for the great majority of four-colour process work.
The fundamental purpose of ΔE*a b and ΔGreyness, in the context of this study, is to quantify
colour variation rather than influence the process. When a process related event such as an
increased ink flow to one particular plate occurs, both measurement systems record the
resulting colour shifts similarly in their respective colour spaces, but both systems can influence
or change the process only by reverting to the original data from which ΔE*a b and ΔGreyness
are derived.

MANFRED BREEDE 55
An independent study in which densitometric raw data was used, rather than converted
Greyness and Hue Error values, supports the results presented in this paper.
Densitometric values ΔR, ΔG, and ΔB could be used effectively in closed loop “on press”
measurement systems to correct undesirable colour casts.
The reliance of the GATF Colour Circle on optical density as its basic input value is
advantageous for controlling the ink flow of a press because, unlike L*a*b*, optical density
bears a close relationship to ink film thickness.
Acknowledgements
Test results of similar research by John MacPhee came to light when this author first
presented his findings to the 58th TAGA conference in Vacouver. The data so generously
shared by John MacPhee, gave more credence to the arguments presented here and
expanded the scope of the study.
Much gratitude is owed to Peter Roehrig, Technician-Printing Processes, Binding and

Finishing, School of Graphic Communications Management at Ryerson University, for
producing the test form and for making valuable suggestions for its design, as well as for
conducting the experimental press run.
Note: Figure 3 is a screen capture of a software published by PIA/GATFPress under the

name Abacus version 2. Figures 4, 5, and 10 are screen captures of a computer program
under development. Both applications were developed by the author.
References
1 Chadwick S., Business Management Advisory, Sewickley, PA, Printing Industries

of America Ltd., 2001
2 Neugebauer H.E.J., Die theoretischen Grundlagen des Mehrfarbendrucks,
Zeitschrift für wissenschaftliche Photographie, Photophysik und Photochemie, 36,
Heft 4, 1937
3 Archer H.B., Reproduction of grey with halftones, TAGA Proceedings, pp. 180-
193, 1954
4 Clapper F.R., Balanced halftone separations for process colour, TAGA Proceedings,
pp. 177-189, 1959
5 Preucil F., Some additive colour principles and the control of neutral greys in
process reproduction, Advances in Printing Sciences and Technology, vol. 3 Halftone
Printing, London, Pergamon Press, pp. 75-84, 1964
6 Pobboravsky I., Methods of computing ink amounts to produce a scale of neutrals
for photomechanical reproduction, TAGA Proceedings, pp. 10-34, 1966
7 Elyjiw Z., GATF Standard Offset Colour Control Bars, Research Project Report No.
76, Graphic Arts Technical Foundation, 1985
8 System Brunner web site, http://www.systembrunner.com, 2005
9 Field G.G. and Jorgensen G.W., Test Images for Printing, Second Edition
Pittsburgh, PA, Graphic Arts Technical Foundation, p.45, 1984
10 Brunner F., System Brunner PCP Picture Contrast Profile, TAGA Proceedings, pp.
256-263, 1987
11 Muirhead A.R., Print Expert(tm), TAGA Proceedings, pp 633-646, 1988
12 Sonntag P., Wuergler D. and Brunner F., rebuttal to an article by Juergen
Mittman of Heidelberg A.G., Deutscher Drucker, No. 13, 2003, promoting
colourimetry for ink regulation, System Brunner web site
http://www.systembrunner.com, 2005

13 Preucil F., Some additive colour principles and the control of neutral tones in
process reproduction, Advances in Printing Sciences and Technology, vol. 3 Halftone
Printing, London, Pergamon Press, p. 84, 1964
14 SWOP, Specifications Web Offset Publications, New York, NY, p.10, 1993
15 SNAP, Specifications For Non-Heat Advertising Printing, Second Printing,
Arlington, VA, Printing Industries of America, 1989
16 GRACoL, 6th Edition, General Requirements for Applications in Commercial Offset
17 Lithography, Alexandria, VA, IDEAliance, 2002
18 Field G.G. and Jorgensen, G.W., Test Images for Printing, Second Edition,
Pittsburgh, PA, Graphic Arts Technical Foundation, p.45, 1984
19 Field G.G., Colour and its Reproduction, Second Edition, GATFPress, p.179, 1999
20 Schmitt U. and Dolezalek F., Instructions for use Ugra/FOGRA Media Wedge
CMYK, V2.0, Munich, www.fogra.org, 2004
21 ECI, European Colour Initiative, Graubalance-Controlle im Offsetdruck mit dem
ECI/ bvdm Grey Control Strip, Berlin, www.eci.org, p.4, 2005
22 DMACS, Digital Magazine Advertizing, Canadian Specifications, issue 4.9.2,
Magazines Canada, 2005
23 Stokes M., Fairchild M.D. and Berns S., Colorimetrically Quantified Visual
Tolerances For Pictorial Images”, TAGA Proceedings, Volume 2, p. 775, 1992
24 Field G.G., Colour and its Reproduction, Second Edition, GATFPress, pp.93-94,
1999
25 Cox F.L., The GATF Colour Diagrams, Pittsburgh, PA, Research Progress No. 81
(Revised Edition), 1970
26 Breede M., Controlling Grey Balance with the GATF Colour Circle, Sewickley, PA,
GATFWorld, pp.22-24, 2004
27 GRACoL 6th Edition, General Requirements for Applications in Commercial Offset
Lithography, Alexandria, VA, IDEAliance, 2002
28 Basimir R., Costello G., Dibernardo A., DiPiazza J., Kuna D., Paulius K.,
Rybny C. and Zawackni W., A Comparison of Visual and Spectrophotometric
Evaluations of Paired Colour Prints, NPIRI Task Force on Colour Measurement,
TAGA Proceedings, pp. 558-578, 1995
29 Field G.G., Colour and its Reproduction, Second Edition, GATFPress, p.183, 1999
30 Heidelberg Druckmaschinen AG, Colour & Quality, Heidelberg Druckmaschinen
AG, Heidelberg, p.81, 1995
31 Schläpfer K., Widmer, E., Colour Deviation And Variations In Gravure And Web
Offset Printing, TAGA Proceedings, pp. 607-619, 1995
32 Tritton K., Colour control in lithography, Pira International, p.210, 1993
33 MacPhee J., Private Communication, Shelton, CT, Baldwin Technology Company,
Inc. June 5, 2006
34 Tritton, K, Colour control in lithography, Pira International, p. 166, 1993

MANFRED BREEDE 57
Appendix A
Raw Data for Uncontrolled Greyness Variation (UGV):

Greyness Greyness
ΔE*ab L*a*b* Densities ΔE*ab L*a*b* Densities
Hue Error Hue Error
V 0.84 V 0.85
L 45.61 Greyness 100 L 45.10 Greyness 93.83
C 0.78 No.16 C 0.81
Standard a -1.75 ΔGreyness 0 a -2.21 ΔGreyness 6.17
M 0.78 3.07 M 0.79
b -3.61 HE 100 b -6.60 HE 60.00
Y 0.78 Y 0.76
V 0.84 V 0.86
No.1 C 0.78 No.17 C 0.81
a -1.83 ΔGreyness 0 a -2.33 ΔGreyness 7.41
0.10 M 0.78 3.30 M 0.79
b -3.59 HE 100 b -6.80 HE 66.67
Y 0.78 Y 0.75
V 0.84 V 0.86
L 45.35 Greyness 98.73 L 45.01 Greyness 92.59
No.2 C 0.79 No.18 C 0.81
a -1.88 ΔGreyness 1.27 a -2.29 ΔGreyness 7.41
0.43 M 0.79 3.52 M 0.79
b -3.92 HE 100 b -7.03 HE 66.67
Y 0.78 Y 0.75
V 0.84 V 0.87
No.3 C 0.79 No.19 C 0.82
0.66 M 0.78 3.67 M 0.82
b -4.24 HE 50 b -6.82 HE 100
Y 0.77 Y 0.77
V 0.84 V 0.88
No.4 C 0.79 No.20 C 0.84
0.82 M 0.78 3.93 M 0.81
b -4.42 HE 50 b -6.91 HE 57.14
Y 0.77 Y 0.77
V 0.84 V 0.88
No.5 C 0.79 No.21 C 0.84
1.10 M 0.78 4.06 M 0.81
b -4.70 HE 50 b -7.01 HE 57.14
Y 0.77 Y 0.77
V 0.85 V 0.89
No.6 C 0.79 No.22 C 0.85
1.27 M 0.79 4.29 M 0.81
b -4.85 HE 100 b -7.02 HE 42.86
Y 0.77 Y 0.78
V 0.85 V 0.88
No.7 C 0.79 No.23 C 0.85
1.47 M 0.79 4.52 M 0.81
b -5.05 HE 100 b -7.31 HE 50.00
Y 0.77 Y 0.77
V 0.85 V 0.88
No.8 C 0.80 No.24 C 0.85
1.52 M 0.79 4.75 M 0.81
b -5.06 HE 66.67 b -7.64 HE 50.00
Y 0.77 Y 0.77
V 0.85 V 0.89
No.9 C 0.79 No.25 C 0.85
1.65 M 0.79 5.04 M 0.81
b -5.22 HE 100 b -7.84 HE 55.56
Y 0.77 Y 0.76
V 0.85 V 0.90
No.10 C 0.79 No.26 C 0.87
1.82 M 0.78 5.31 M 0.82
b -5.39 HE 66.67 b -7.51 HE 44.44
Y 0.76 Y 0.78
V 0.85 V 0.91
No.11 C 0.80 No.27 C 0.86
2.07 M 0.79 5.50 M 0.83
b -5.60 HE 75.00 b -8.30 HE 62.50
Y 0.76 Y 0.78
V 0.85 V 0.89
No.12 C 0.80 No.28 C 0.85
2.22 M 0.79 5.68 M 0.82
b -5.74 HE 75.00 b -8.82 HE 66.67
Y 0.76 Y 0.76
V 0.85 V 0.89
No.13 C 0.80 No.29 C 0.85
2.42 M 0.78 5.91 M 0.82
b -5.89 HE 50.00 b -8.98 HE 66.67
Y 0.76 Y 0.76
V 0.85 V 0.89
No.14 C 0.80 No.30 C 0.85
2.69 M 0.78 6.21 M 0.82
b -6.13 HE 60.00 b -9.37 HE 66.67
Y 0.75 Y 0.76
V 0.85 V 0.90
No.15 C 0.80 No.31 C 0.86
2.84 M 0.79 6.39 M 0.82
b -6.37 HE 80.00 b -9.51 HE 60.00
Y 0.75 Y 0.76
V 0.90 V 0.91
No.32 C 0.85 No.37 C 0.89
6.55 M 0.82 7.40 M 0.83
b -9.71 HE 70.00 b -9.92 HE 53.85
Y 0.75 Y 0.76
V 0.91 V 0.92
No.33 C 0.88 No.38 C 0.89
6.69 M 0.84 7.64 M 0.83
b -9.39 HE 63.64 b -10.14 HE 53.85
Y 0.77 Y 0.76
V 0.92 V 0.92
No.34 C 0.88 No.39 C 0.90
6.87 M 0.84 8.13 M 0.83
b -9.40 HE 63.64 b -10.53 HE 50.00
Y 0.77 Y 0.76
V 0.91 V 0.92
No.35 C 0.88 No.40 C 0.91
7.11 M 0.83 8.33 M 0.83
b -9.59 HE 54.55 b -10.14 HE 46.67
Y 0.77 Y 0.76
V 0.92
L 42.57 Greyness 86.52
No.36 C 0.89
a -4.57 ΔGreyness 13.50
7.26 M 0.84
b -9.57 HE 58.33
Y 0.77

Appendix B
Raw Data for Controlled Greyness Variation (CGV):
Greyness Greyness
ΔE*ab L*a*b* Densities ΔE*ab L*a*b* Densities
Hue Error Hue Error
V 0.84 V 0.95
C 0.78 No.8 C 0.83
Standard a -1.75 ΔGreyness 0 a 6.53 ΔGreyness 11.71
M 0.78 10.53 M 0.94
b -3.61 HE 100 b -7.27 HE 18.18
Y 0.78 Y 0.85
V 0.85 V 0.96
No.1 C 0.79 No.9 C 0.84
a -0.28 ΔGreyness 3.75 a 7.13 ΔGreyness 12.50
2.23 M 0.80 11.57 M 0.96
b -5.18 HE 66.67 b -7.85 HE 16.67
Y 0.77 Y 0.86
V 0.86 V 0.96
No.2 C 0.80 No.10 C 0.82
a -0.93 ΔGreyness 4.94 a 8.76 ΔGreyness 15.46
3.28 M 0.81 12.55 M 0.97
b -6.61 HE 75.00 b -6.86 HE 33.33
Y 0.77 Y 0.87
V 0.87 V 0.98
No.3 C 0.79 No.11 C 0.85
a 1.26 ΔGreyness 6.02 a 9.36 ΔGreyness 15.84
4.51 M 0.83 13.76 M 1.01
b -6.57 HE 20.00 b -7.15 HE 31.25
Y 0.78 Y 0.90
V 0.90 V 0.98
No.4 C 0.83 No.12 C 0.84
5.21 M 0.85 14.61 M 1.01
b -7.50 HE 66.67 b -7.98 HE 29.41
Y 0.79 Y 0.89
V 0.92 V 1.00
No.5 C 0.83 No.13 C 0.85
6.87 M 0.88 15.17 M 1.03
b -7.90 HE 28.57 b -8.12 HE 27.78
Y 0.81 Y 0.90
V 0.93 V 1.00
No.6 C 0.85 No.14 C 0.85
7.85 M 0.90 16.16 M 1.04
b -8.38 HE 37.50 b -8.20 HE 26.32
Y 0.82 Y 0.90
V 0.92 V 1.03
No.7 C 0.81 No.15 C 0.87
9.13 M 0.90 17.33 M 1.08
b -8.11 HE 0 b -8.19 HE 33.33
Y 0.81 Y 0.94

R. SILVENNOINEN, M.JUUTI, H. KOIVULA, M. TOIVAKKA AND K.-.E. PEIPONEN 59
Diffractive glossmeter for measurement of dynamic gloss of

prints
Raimo Silvennoinen1, Mikko Juuti1, Hanna Koivula2, Martti Toivakka2 and Kai-Erik Peiponen1
1
Department of Physics and Mathematics, University of Joensuu
P.O. Box 111, FI-80101 Joensuu, Finland
raimo.silvennoinen@joensuu.fi
mikko.juuti@joensuu.fi
kai.peiponen@joensuu.fi
2
Laboratory of Paper Coating and Converting, Åbo Akademi University
FI-20500 Turku, Finland
hanna.koivula@abo.fi
martti.toivakka@abo.fi
Abstract
Print gloss is an important quality measure for graphic communication papers. In

addition, the time evolution of the print gloss immediately after ink has been transferred
onto a paper is related to ink-paper interactions that control both paper printability and
run-ability of a printing process. However, currently the selection of glossmeters to
measure dynamic print gloss reliably is rather limited. Thus in this paper we propose a
glossmeter what is believed to be a novel system for dynamic print gloss detection from
papers. The intelligent part of the glossmeter is a diffractive optical element, which also
makes it easily possible to construct an online measurement device. Experiments proved
that the proposed sensor measures print gloss accurately at high sampling rates, and it
reveals the behaviour of the ink setting as well.
Keywords: Gloss, print gloss, ink gloss, dynamic print gloss, diffractive optical element.
1 Introduction
The concept of dynamic print gloss has become important factor to characterize print quality in
multi-colour technology worldwide in different type of print industry, which covers a wide
range of applications such as printing on paper, packing materials, plastic, metals, textiles to
name a few.
For instance in printing houses or for press machine manufacturers the quality of multi-colour
printing, especially on paper, is controlled by different methods utilizing standard measurement
and control technologies. This sets also feedback requirements for the paper makers, including
the filler and pigment makers, how to control the standardised quality in physical and optical
properties such as CIE whiteness, CIE colour/shade, ISO brightness, moisture, opacity, light
scattering coefficient, surface energy, surface roughness, abrasion resistance, dimensional
stability, mottle and water absorptivity. Altogether to control and supervise the paper surface
quality, and develop novel methods for paper surface research leads to a great variety of issues
in paper surface science. Here we refer only to few references, which deal with the recent
studies in relation to paper coating [1,2,3,4,5,6,7,8,9,10,11]. In a printing process the properties
of ink, such as colour, grind hue, percentage portion of solids, set time, strength, tack, viscosity,
water pick-up etc., play an important role. We refer only to basic hypothesis on the mechanism
of ink-splitting during printing [12] and few recent references in rheology of ink [13], in tack of
ink in substrate [5,7,14,15,16] without neglecting effects of liquid sorption into substrate, or
porous structures formed by pigment materials [17,18,19,20,21]. The liquid sorption causes also
different type of swelling processes [22].

60 DIFFRACTIVE GLOSSMETER FOR MEASUREMENT OF DYNAMIC GLOSS OF PRINTS
After printing process the appearance of the print has a great importance, and thus methods to
sense the contrast between the paper surface and the printed areas on it have been developed.
This relation can be characterized by concept of gloss. In general, the gloss of an object is
typically divided into six different indicators, namely specular gloss, sheen, contrast gloss,
absence of bloom gloss, distinctness of reflected image gloss and absence of surface texture
gloss [23,24]. Typical gloss standards describe the test methods for specular gloss of non-
metallic specimens at fixed incident angles of 20°, 60°, and 85°, which are the most popular
standard geometries for measuring specular gloss. To sense gloss from paper extra 75°
geometry is added in the standard specular gloss measurements to obey the ISO 1223 and
TAPPI T480 specifications. Recently NCR has developed a reference goniospectro-photometer
to provide high-accuracy traceable measurements of specular gloss at several standard
geometries, including 75° for paper samples, haze and absence-of-bloom gloss, and colour
appearance of gonio-apparent materials [25].
For pigment coated paper, the print quality and print gloss depend on many factors such as
pigment particle size and shape distribution, which control the porosity and surface roughness of
the coating layer [26,27,28]. Also binders and especially ink and ink levelling [29,30,31,32,33]
have influences to print gloss [34]. In printed matters the evaluation of temporal print gloss may
be a useful tool in investigation of ink-paper interaction, since the wet or non-dry ink film layer
correlates to final print gloss or paper-ink interface [35]. When the print gloss is monitored as a
function of time it is expected that the responses will show different responses as a consequence
of complicated ink setting. However, literature lists only few methods and tools to investigate
this important mechanism. The print gloss measurements reported by [36] utilize use of
goniometric arrangements. Whereas in the print gloss measurements reported by [30] use a laser
light incidence in oblique angle of 75o and in the print gloss measurements reported by [10]
utilize polarized light reflectrometry at oblique angles (75o) of laser light incidence. In addition,
the following authors as [37,38] have reported tentative results concerning the print gloss of
paper but not temporal evolution of print gloss. In those experiments a diffractive optical
element based geometric arrangement was used.
In this paper we report what is believed to be novel system of dynamic print gloss measurement
based on the use of a diffractive optical element based glossmeter (DOG) for the assessment of
the dynamic gloss of printed papers. This is a new method for this application and we believe
that the measurement of dynamic gloss with DOG will provide new information on ink-paper
interactions and ink setting mechanisms. The approach is also easily applicable to on-line
measurement of print gloss. Development of gloss is studied in detail with a large sample
matrix. Papers with a differing pore structure and a non-absorbing film are used as substrates.
Using black as well as coloured inks, and varying the ink amount will provide information about
the contribution of ink to development of gloss. Additionally, the measurement parameters of
the DOG have been optimised.
2 Materials and methods
A matrix of four coated paper samples having a matte (gloss target 30 during calendering) or
glossy (gloss target 70 during calendering) surface finish, and fast or slow ink setting speed
were selected for the dynamic gloss experiments. A smooth non-absorbing plastic film was used
as the fifth substrate. The four papers, labeled SG (slow glossy), SM (slow matte), FM (fast
matte) and FG (fast glossy), were double coated in pilot scale using coating formulations. The
differences in the printability and optical properties of the papers are a result from use of
different pigment sizes and shapes, which create different microscopic pore and surface
structures for the coatings. All the samples were calendered to adjust their gloss and
smoothness. The paper samples were printed in laboratory environment by using a black,
commercial heatset offset ink (Premoterm 2000, Flint Schmidt). The dynamic print gloss was
measured immediately after printing with the DOG sensor. Here we point out that because of

the fact that the paper matrix was made in pilot scale the machine and cross direction of paper
was not controlled and documented in detail, and thus we were not able to utilize all the
potential of the DOG sensor, such as ability to sense directional isotropy or anisotropy of the ink
setting as already reported by [38]. In this work only the average dynamic print gloss is
reported.
Schematic layout of the DOG system attached to UTP (Universal Test Printer) is shown in
Figure 1. Polarized light wavefront from a semiconductor laser (λ = 635 nm) is expanded with
c
the aid of a high refractive power lens system by such a way that the wavefront distribution is
uniform after passing through a circular aperture stop. Thereafter, the uniform wavefront is
focused by a lens system in collimating and focusing unit (C) on the printed ink-paper interface
of the sample (S). The ink-paper interface is produced on the paper located on the roller (R) by
using the printing unit (P). From the ink-paper interface scattered wavefront is captured
specularly on the aperture of the diffractive optical element (DOE). After passing the DOE
aperture the wavefront produces a diffraction pattern on the surface of the 2D photo-array of the
charge coupled device (CCD).
Figure 1: Schematic layout of the DOG system attached to UTP (Universal Test Printer): Diode
Laser (L), Collimating and focusing unit (C), Printing Unit (P), Printed paper sample (S), Print
roller (R), Diffractive Optical Element (DOE), Charge-coupled device (CCD).
The DOG presented in Figure1 utilizes advances of a diffractive optical element, which are the
ability to sense non-coherent and coherent properties of the laser wavefront that scatters from
the ink-paper interface in selected solid angle. Moreover, in the DOE artificially 16 separate
lenses are integrated. Each lens forms separately spot image in the focal plane of the element. It
is also worth to denote that the separate 16 wavefronts interact with each other making this
principle active to sense very small phase shifts in the wavefront. Typically the phase
differences in the coherent wavefront appear as speckle formation in the far-field regime as
shown in the inset of Figure 2a. The difference between the DOG and the conventional laser
glossmeter is that the latter cannot sense phase information of the scattered wavefront.
Next we consider the crucial part and the geometry of DOG sensor shown in Figure 2 in more
details. The DOE obeys the law of diffraction and principle of hologram imagery [39,40]. To
consider and analyse geometric properties of diffracted wavefront we have to determine mean
distance.

r WFR1
L se
u La WFR2
z
CCD
Paper DOE
(a)
(b) (c)
Figure 2: a) DOG sensor, L = lens, WFR1 is the ideal wavefront from a point source and WFR2
distorted wavefront, which is typically caused by appearance of dynamic speckle formation
when the ink-paper interface moves with the velocity u perpendicular to direction of the surface
normal z. A still image grabbed from a dynamic speckle at DOE aperture is shown in the upper
right inset. DOE image (in it’s image plane) is captured by a computer controlled CCD device
for further analysis, b) Binary DOE element and c) The ideal 4×4 light spot matrix in the image
plane of DOE.
In variables Rr, Ro, Rc, and Ri of point Q and respective angle variables αr, βr, αo, βo, αc, βc, αi, βi
with respect to the origo (Figure 3).
y
Q
RQ
z αQ
βQ
Figure 3: The mean distance variable RQ with respective angular variables αQ and βQ of point
source Q.
With the three variables Rr, Ro and Rc it is possible to determine the location of point source Q.
The subscripts r, o, c, i indicate reference, object, reconstruction and image -terms of mean
distance variables, respectively. According to Gaussian image properties the mean distance
variables and respective angle variables of point Q relate to each other as follows
1 1 μ ⎛ 1 1 ⎞
= ± 2 ⎜⎜ − ⎟, (1)
Ri Rc m ⎝ Ro Rr ⎟⎠

μ
sin α i = sin α c ± (sin α o − sin α r ) , (2)
m
μ
cos α i sin β i = cos α c sin β c ± (cos α o sin β o − cos α r sin β r ) , (3)
m
and
λc
μ= , (4)
λo
where m is a scaling factor (here m = 1). In the Equations (1-4) the factor μ denotes the ratio of
the reconstruction and recording wavelengths. The (±)-sign is + for a virtual image and - for a
real image. In the case of the DOE, the reconstruction source is infinity and thus the respective
angle variables are αr = 0 and βr = 0.
The geometry obeying hologram imagery including geometric and chromatic aberrations of the
DOE used in DOG is considered in details by [41]. The DOE element, used in the realisation of
the DOG-meter was calculated by using the Rayleigh-Sommerfield diffraction integrals [42], is
an on-axis binary amplitude diffractive element (Figure 2b). The element is focusing type and
its focal length is planned to be 100 mm at λo = 632.8 nm. The planar DOE produces a 4 × 4
light spot matrix in its focal plane (Figure 2c). Because of the large number of light spots (16), it
is possible to use statistical analysis in order to increase the accuracy and reliability of the gloss
measurement. The size of the aperture of the DOE is 4mm × 4mm, and the respective image
area of 0.5mm × 0.5mm including the 16 light spots with the distance of 125 micrometers
between nearest adjacent light spots in the focal plane when the reconstruction source locates in
infinity. The DOE’s wave optical principle was analysed in details by using diffraction based
wave-optical treatment [43]. After geometric and wave optical analysis the DOE was fabricated
by sputtering a 120 nanometer thick layer of chrome on a fused silica substrate. Next, a positive
electron-beam resist was spin-coated onto the chrome layer. This resist was exposed using an
electron beam writer. After the resist was developed, the chrome layer was wet-etched. More
details of DOE in materials inspection are described in the references [44,45].
After passing the DOE aperture the distorted wavefront performs a diffraction pattern on the
surface of the 2D photo-array of the charge coupled device (CCD). Thus also the ideal 4×4 light
spot matrix, as shown in Figure 2c, is distorted and its brightness is reduced as a consequence of
roughness as shown in Figure 4a. In the numerical analysis of the DOG sensor image to get
measure for the dynamic print gloss, the total irradiance over pixels locating in the signal
window (denoted by dashed area in Figure 4b) is calculated by using the following formula
N SW , M SW
1
I TOT =
N SW M SW
∑I iSW , j SW
iSW =1, j SW =1
, (5)
where N SW and M SW are the dimensions of the DOE image in x- and y-direction and I iSW , jSW is
the image irradiance observed by the (iSW , j SW ) th element of the CCD camera array. The ink-
paper interface is not smooth in optical sense and thus the irradiance of the peaks is reduced.
However to be absolutely sure that the irradiance of peaks (denotes coherent response) does not
make any uncertainties, the irradiance portion of peaks should be subtracted from the total
irradiance of that DOE image as follows
N SW , M SW N pks , M pks
1 1
I NC =
N SW M SW
∑ I iSW , jSW −
iSW =1, j SW =1 N pks M pks
∑I i pks , j pks
i pks =1, j pks =1
, (6)

where N pks and M pks are the total numbers of pixels in the DOE spot image in x- and y-
direction and I i pks , j pks is the image irradiance observed by the (i pks , j pks ) th element of the DOE
spot image (region of spots: dark-grey areas in Figure 4b). The light-grey areas enable the
calculation of visibility factor V = (Vmax −V min) (Vmax + Vmin ) for x- and y-directions to
estimate the anisotropy of the ink-paper interface as already reported for print gloss of paper
with clear separation of machine (MD) and cross-direction (CD) without temporal evolution of
print gloss [38]. However in this paper we present only the average dynamic print gloss arisen
from the reasons as already stated in the first section of this chapter. Finally, the gloss can be
calculated from the relation
I ink − substrate
G= × 100 . (7)
I reference
where Iink-substrate and Ireference are measured from an ink-substrate interface and from the gloss
reference.
Figure. 4: (a) Image pattern of 4 × 4 light spot matrix produced by DOE from print, (b)
minimum (light grey) and maximum (dark grey) intensity areas of the DOE image pattern. The
area used for calculation of the total irradiance is marked by dashed line.
In calibration of the DOG equipment for print gloss measurements, the gloss base line is
measured by using a gloss reference (black gloss standard). The non-uniformities of ink-paper
interface affecting different types of distortions in the scattered wavefront are thereafter
captured specularly on the aperture of the diffractive optical element (DOE) according to the
arrangement of Figure 1.
Table 1: Additional parameters of the DOG-meter used in dynamic print gloss measurements.
Subject Notes
Incidence angle of laser beam (αc, βc) (4.8, 0) deg.
μ = λc λo 1.0035
Gloss reference Black gloss standard
Measurement area ca. 5 mm x 5 mm
Measurement frequency ca. 100 ms
Measurement period 300 s
When the incidence angle of the laser beam (αc, βc) = (4.8 deg, 0 deg) is used in the DOG
equipment, the deviation of the DOG spot image of 1mm, on the 2D photo-array of the CCD
device, can be accepted caused by the wobbling of the paper web by 11.9 mm in the normal z-
direction (Figure 2a). Theoretically this limitation can be calculated using the Rayleigh criteria
( z R = πwo2 λc , where wo denotes the minimum radius of the focused laser beam waist). The

other additional parameters of the DOG-meter used in the dynamic print gloss measurements
contain the following specifications as shown in Table 1. The correlation between the industry
standard gloss and the DOG gloss can be found, e.g., from following references [46,47]. The
receptor angle of the present DOE sensor geometry is rather small, moreover, the DOE element
itself acts as a spatial filter for coherent light (coherent components diffract towards the peak
area, where the gloss is measured), and thus it is no need to use an extra bandpass filter in the
DOG sensor to cancel the harmful effects of background irradiance away.
In the dynamic print gloss experiments the responses, printed with constant slow/black ink
volume of 0.3cm3 ink on rolls, showed distinctive separation from sample to sample when
different type of papers were used in the test (Figure 5). In this experiment slow/black ink on
slow/glossy and fast/glossy papers showed (monotonic increasing) gloss evolution as a function
of time, the gloss of slow/glossy
1.2
SG
1
0.8
FG
G
0.6
0.4
SM
0.2 FM
0
0 50 100 150 200 250 300
t
Figure. 5: Dynamic gloss curves in gloss unit as a function of time in seconds from four
different paper samples print with 0.3cm3 ink amount. SG=slow/glossy, FG=fast/glossy,
SM=slow/matte and FM=fast/matte paper.
in the above paper being two times as high as that of the fast/glossy paper. This behaviour
indicates long time gloss formation with a high time constant τ (at 0.9GSAT –level, where GSAT is
gloss saturation), whereas in the slow/matte and in the fast/matte papers the gloss formation is
faster with smaller time constants than in the previous cases. The time constants for the gloss
dynamics shown in Figure 5 are listed in the Table 2. The results of the second experiment,
where the slow/black ink volume on rolls varied on the above paper surfaces, showed
interesting phenomena in ink-setting (Figure 6). In the ink amount experiments we confined
only to three different ink amonts (0.15, 0.30 and 0.50 cm3) caused by the huge number of paper
samples. In slow/ matte (Figure 6a) and fast/matte (Figure 6b) papers the ink- setting contributes
low gloss, and reaches maximum gloss magnitude with 0.3 cm3 volume of ink on rollers,
whereas the at the higher volumes the gloss will decrease significantly. The behaviour implies
that the ink may unify the non-isotropic structure of ink - matte paper interface as already
tentatively stated by [38]. This dynamic gloss evolution confirms the previous findings.

Table 2: Time constants of gloss evolution for four papers shown in Figure 3, when the
slow/black ink of 0.3cm3, was used in the printing.
Paper Time constant τ in second

at 0.9GSAT –level
Slow/Glossy 170
Fast/Glossy 140
Fast/Matte 60
Slow/Matte 35
(a) (b)
0.8 0.4
0.30cm 3
0.6 0.30cm 3 0.3
0.50cm 3
G
G
0.4 0.2
0.50cm 3
0.2 0.1
0.15cm 3 0.15cm 3
0 0
0 100 200 300 0 100 200 300
t t
(c) (d)
0.8
0.50cm 3
1 0.30cm 3
3
0.6
0.30cm 0.50cm 3
G
0.4
0.5
3
0.15cm
0.2
0.15cm 3
0 0
0 100 200 300 0 100 200 300
t t
Figure 6: Dynamic gloss curves in gloss units as a function of time in seconds from four
different papers printed using 0.15, 0.30 and 0.50 cm3 ink amounts (a) slow matte, (b) fast
matte, (c) slow glossy and (d) fast glossy paper.
The decrease of the gloss at high ink amounts can be related to formation of large ink filaments,
which are not levelled out by the surface tension prior to the final setting of the ink film. This is
supported by the observation that the final print gloss values are higher on papers designed to
have slow ink-setting speed. Also the changes in effective refractive index may also have some
insignificant contribution to decrease of gloss when ink volume increases. It may also be
possible that the gloss will decrease as a consequence of evolution of surface roughness, which
may appear at the formation of high ink coverage. This phenomenon is observed for different
type of papers [48] and for imprints [49]. On the contrary in slow glossy (Figure 6c) and fast
glossy (Figure 6d) papers the increase of ink from 0.3 cm3 to 0.5 cm3 do not cause significant
changes in uniformity of ink-paper interface, and the dynamic gloss responses coincide as
shown in Figure 5. This may indicate that the ink has achieved saturation, and the addition of
extra ink does not change the gloss. Although the dynamic colour measurements were not
performed simultaneously the observation by naked eye revealed that adding of more ink
increased the perceived colour. This subjective observation may cause by the coupling of
illuminating light through the air-ink interface into paper. The same type phenomenon appears
on the Nordic lakes in the spring when the scattering of daylight from ice surface reduces as a
consequence of coupling of daylight through ducts of water in ice into underlying lake water,
and the ice surface become dark.

As the general note all of the DOE measurements showed rather good repeatability from paper
to paper used in the test. Here we point out that the repeatability of DOG mesurement has been
observed to be 0.26% measured from high gloss metal plate [50]. However, there should be a bit
higher deviation in the present measurement, since the printing properties of labprinter also
vary. Also amount of ink on roller is rather difficult to control. As an example, we report here
the responses (Figure 7) performed with aid of the slow/black ink and fast/ matte paper as it is
the case in Figure 6b.
0.35
3
0.50cm
0.3
0.25
0.2 3
0.30cm
G
0.15
0.1
0.05 3
0.15cm
0
0 50 100 150 200 250 300
t
Figure 7: Repeatability of dynamic print gloss measurement.
7
3
0.30cm
6
3
0.50cm
5
4
G
2
3
0.15cm
1
0
0 50 100 150 200 250 300
t
Figure 8: Dynamic gloss development of slow/black ink on non-absorbing plastic film.
On the plastic film the dynamic gloss values from slow ink – plastic film interface (Figure 8) are
roughly five times as high as from the slow ink – slow glossy paper, and roughly ten times
higher as from ink- matte papers and fast glossy paper (Figure 6). The increase of dynamic gloss
is believed to be caused by the higher smoothness, lower porosity and better isotropy of the air-
ink-film interface than is the case with the pigment coated papers.

As an additional discussion concerning the gloss, which may have public interest, we point out
that the gloss depends on the reflectivity, which relates to the complex refractive index, on the
angle of incidence of light and on the texture of surface including the surface roughness. In the
case of DOG the normal incidence of probe light is usually used in the experiments to cancel
out the effects caused by the polarization but it is also possible to use oblique angle of light
incidence. The use of probe light close in normal direction of paper surface, which is possible
by using the DOG sensor without loosing the specular signal as it the case with the reflectance
measurements with laser light, allows also deeper penetration of the light compared with the
oblique angle geometry, which in turn may be an influencing factor to reveal a fall in gloss with
higher ink levels. Since the colour will change during the print run also the gloss will change as
a consequence of refractive index change and possible change of roughness [49]. The same
principle of gloss is also valid for the transparent inks and coatings as plastic, varnish and glass
etc. [45,51]. The DOG gloss values correlate with the gloss values drawn out by commercial
devices, and the studies with human perception has been carried out but not reported yet. In this
paper we have introduced the dynamic gloss properties of the DOG sensor, the anisotropy
parameters related to DOG gloss we have already introduced in several publications
[6,37,38,49,52]. The wider and deeper insight to the basis of the DOG method and to the merits
of the DOG sensor can be find, for instance, from the following references [41,50].
4 Conclusion
The DOG sensor, working in neighbourhood of normal incidence, was observed to be capable
of sensing dynamic print gloss. The sensor reveals the behaviour of the ink setting as a function
of time giving information from the remodelling of ink-paper interface also at low inking level.
In contrast to the conventional online paper glossmeters working at oblique angles of incidence
as already mentioned in introductory part of this article, the DOG sensor does not require a flat
sample surface due to its controlled waist cross section of the focused laser beam. Moreover,
from the geometrical reasons, the gloss meters working at the oblique angles of incidence (20o,
60o, 75o and 85o) are sensitive for paper-web flutter. At low inking levels, the paper properties,
surface roughness and absorption are important for the gloss evolution, and in that purpose the
DOE element can acts as a crucial multifunction tool in measurement of dynamic print gloss
development at various level of inking.
Acknowledgement
This study was performed in the Molecular Understanding of Printability Project

(MOLPRINT), which is part of Network of Competence in Formation of Surface
Properties (NETCOAT), funded by the Finnish Funding Agency for Technology and
Innovation (Tekes).
Nomenclature
λ = wavelength
Rr, Ro, Rc, and Ri = mean distances from point Q, and
αr, βr, αo, βo, αc, βc, αi and βi = respective angles, where subscripts r, o, c, i indicate respective
reference, object, reconstruction and image –terms
m = scaling factor (here m = 1)
μ = ratio of the reconstruction and recording wavelengths
I = irradiance
V = visibility
G = gloss
zr = Rayleigh regime
wo = minimum radius of the focused laser beam waist

References
1 Aspler J.S. and LePoutre P., “Transfer and setting of ink on coated paper,” Prog.
Org. Coat., vol. 19, no. 4, pp. 333-357. 1991.
2 Zang Y.H. and Aspler J.S., “The influence of coating structure on the ink
receptivity and print gloss of model clay coatings”, Tappi J., vol. 78, no. 1, pp. 147-
154. 1995.
3 Iyer R.R. and Bousfield D.W., “The Levelling of Coating Defects with Shear
Thinning Rheology,” Chem. Eng. Sci., vol. 51, no. 20, pp. 4611-4617. 1996.
4 Bousfield D. W. and Co. A., “Paper coating rheology,” in Advances in the flow and
rheology of non-Newtonian fluids Part B., Edited by Siginer, Kee and Chhabra, pp.
827-842. 1999.
5 Barbesta F., Bousfield D. W. and Rigdahl M., “Modelling of rheological
properties of coating colors,” J. of Rheology, vol. 45, no. 1 pp.139-160. 2001.
6 Silvennoinen R., Peiponen K.-E., Sorjonen M., Tornberg J. and Sumén J.,
“Diffractive optical sensing of the surface quality of coated paper,” Paper and
Timber, vol. 83, no. 5, pp. 395-399. 2001.
7 Preston J. S., Elton N. J., Legrix A., Nutbeem C. and Husband C., “The role of
pore density in setting offset printing ink on coated paper,” Tappi J., vol. 1, no. 3, pp.
3-5. 2002.
8 Ström G., Englund A. and Karathanasis M., “Effect of coating structure on print
gloss after sheet-fed offset printing,” Nord. Pulp Pap. Res. J., vol. 18, no. 1, pp. 108-
115. 2003.
9 Xu R., Fleming P. D., Pekarovicova A. and Bliznyuk V., “The Effect of Ink Jet
Paper Roughness on Print Gloss,” J. Imag. Sci. Technol., vol. 49, no. 6, pp. 660–666.
2005.
10 Elton N. J. and Preston J. S., “Polarized light reflectrometry for studies of paper
coating structure - Part I. Method and instrumentation,” Tappi J., vol. 5, no. 7, pp. 8
–16. 2006.
11 Stål M., ”The influence of starch addition strategy on the surface strength and
printability of SC offset paper”, ISBN 952-12-1761-8, Doctoral thesis in Paper
Coating and Converting, Faculty of Technology – Laboratory of Paper Coating and
Coverting, Abo Akademi University, Finland. 2006.
12 Taylor J .H. and Zettlemoyer A.C., “Hypothesis on the mechanism of ink-splitting
during printing,” Tappi J., vol. 12, pp. 749-757. 1958.
13 Bousfield D. W., “Particle Motion During Shear: The Influence of Particle Shape
and Roughness on Rheology,” Nordic Pulp and Paper Res. J., vol. 8, pp. 176-183.
1993.
14 van Gilder R L. and Purfeest R. D., “Commercial six-color press runnability and
the rate of ink tack build as related to the latex polymer solubility parameter,” Tappi
J., vol. 77 , no. 5, pp. 230-239. 1994.
15 Xiang Y., Bousfield D. W., Hassler J., Coleman P. and Osgood A., “Measurement
of local variation of ink tack dynamics,” J. Pulp and Paper Sci., vol. 25, no. 9, pp.
326-330. 1999.
16 Xiang Y. and Bousfield D. W., “Influence of coating structure on ink tack
dynamics,” J. Pulp and Paper Sci., vol. 26, no. 6, pp. 221-227. 2000.
17 Sapieha S., Inouem M. and Lepoutre P., “Conductivity and water sorption in
paper,” J. Appl. Pol. Sci., vol. 30, no. 3, pp. 1257-1266. 1985.
18 Ramarao B. V., Massoquete A., Lavrykov S. and Ramaswamy S., “Moisture
Diffusion Inside Paper Materials in the Hygroscopic Range and Characteristics of
Diffusivity Parameters,” Dry. Technol., vol. 21, no. 10, pp. 2007-2056. 2003.
19 Sørensen G. and Hoffmann J., “Moisture sorption in moulded fibre trays and effect
on static compression strength,” Pack. Technol. Sci., vol. 16, no. 4, pp. 159-169.
2003.

20 Xu G. G., Yang C. Q. and Den Y., “Mechanism of paper wet strength development
by polycarboxylic acids with different molecular weight and
glutaraldehyde/poly(vinyl alcohol),” J. Appl. Pol. Sci., vol. 101, no. 1, pp. 277-284.
2006.
21 Fabritius T. and Myllylä R., “Liquid sorption investigation of porous media by
optical coherence tomography,” J. Phys. D: Appl. Phys., vol. 39, pp. 4668–4672
2006.
22 Xiang Y., Bousfield D. W., Hayes P. C. and Kettle J., “Effect of latex swelling on
ink setting on coated paper,” J. of Graphic Tech., vol. 1, pp. 13-25. 2003.
23 Hunter R. S. The Measurement of Appearance, Wiley. 1975.
24 Hunter R. S. and Harold R .W. , The measurement of appearance, John Wiley
Sons, New York. 1987.
25 Liu J., Noël M. and Zwinkels J., “Design and characterization of a versatile
reference instrument for rapid, reproducible specular gloss measurements,” Appl.
Opt., vol. 44, pp. 4631- 4638. 2005
26 Glatter T. and Bousfield D. W., “Print gloss development on a model substrate,”
Tappi J., vol. 80, no. 7, pp. 125-132. 1997.
27 Donigan D. W., Ishley J. N. and Wise K. J., “Coating pore structure and offset
printed gloss,” Tappi J., vol. 80, no. 5, pp. 163-172. 1997.
28 Jeon S. J. and Bousfield D. W., “Print gloss development with controlled coating
structures,” J. Pulp and Pap. Sci., vol. 30, no. 4, pp. 99-104. 2004.
29 Lavelle J. S., “Gloss: theory and its application to printed ink films, National
Association of Printing Ink Manufacturers”. 1982.
30 Desjumaux D. M., Bousfield D. W. Glatter, T. P. and van Gilder R. L., “The
influence of latex type and concentration on ink gloss dynamics,” Prog. Org. Coat.,
vol. 38, pp. 89-95. 2000.
31 Xiang Y. and Bousfield D. W., “Effect of ink emulsification on ink gloss
dynamics,” Nord. Pulp and Pap. Res. J., vol. 17, no. 1, pp. 61- 66. 2002.
32 Preston J. S., Parsons D. J., Jones M., Gate L. F., Husband J. C. and Legrix A.,
“Ink gloss development mechanisms after printing- Part 1 - The influence of ink film
thickness,” Journal of Graphic Technology 1.2, pp. 29 - 37. 2003.
33 Karathanasis M., Fogden A. and Dahlvik P., “The concept of critical ink setting
time and its relation to print gloss: influence of latex binder,” Nord. Pulp Pap. Res.
J., vol. 19, no. 2, pp. 145-149. 2003.
34 Fetsko J. M. and Zettlemoyer A. C., “Factors affecting print gloss and uniformity,”
Tappi J., vol. 8, pp. 667. 1962.
35 Desjumaux D., Bousfield D. W., Glatter T. P., Donigian D. W., Ishley J.N. and
Wise K. J., “Influence of Pigment Size on Wet Ink Gloss Development,” J. Pulp
and Paper Sci., vol. 24, no. 5, pp. 150-155. 1997.
36 Arney J. S., Heo H. and Anderson P. G., “A Micro-Goniophotometer and the
Measurement of Print Gloss,” J. Imag. Sci. and Technol., vol. 48, pp. 458 – 463.
2004.
37 Sorjonen M., Jääskeläinen A., Peiponen K.-E. and Silvennoinen R., “On the
assessment of the surface quality of black print paper by use of a diffractive optical-
element-based sensor,” Meas. Sci. Technol., vol. 11, pp. N85-N88. 2000.
38 Palviainen J., Sorjonen M., Silvennoinen R. and Peiponen K.-E., “Optical
sensing of colour print on paper by a diffractive optical element,” Meas. Sci.
Technol., vol. 13, pp. N31-N37. 2002.
39 Latta J. N., “Computer-based analysis of hologram imagery and aberration I:
Hologram types and their monochromatic aberrations,” Appl. Opt, vol. 10, pp. 599 –
608. 1971.
40 Latta J. N., “Computer-based analysis of hologram imagery and aberration II:
Aberrations induced by a wavelength shift,” Appl. Opt, vol. 10, pp. 609 - 618. 1971.

41 Silvennoinen R., Räsänen J., Savolainen M., Peiponen K.-E., Uozumi J. and
Asakura T., ”On simultaneous optical sensing of local curvature and roughness of
metal surface,” Sensors and Actuators A, vol. 51, pp. 117-123. 1996.
42 Gaskill J. D., Linear systems, Fourier transforms, and optics, John Willey Sons, New
York. 1978.
43 Räsänen J., Savolainen M., Silvennoinen R. and Peiponen K.-E., “Optical
sensing of surface roughness and waviness by computer-generated hologram,” Opt.
Eng., vol. 34, no. 9, pp. 2574 - 2580. 1995.
44 Silvennoinen R., Peiponen K.-E. and Asakura T., “Diffractive optical elements in
materials inspection,” in International Trends in Optics and Photonics ICO IV,
Asakura T (ed.) pp. 281- 293. Berlin, Springer. 1999.
45 Myller K., “A glossmeter based on a diffractive optical element” (Doctoral
dissertation ISBN 952-458-510-3, University of Joensuu, Department of Physics).
2004
46 Myller K., Juuti M., Peiponen K.-E., Silvennoinen R. and Heikkinen E.,
”Quality inspection of metal surfaces by diffractive optical element-based
glossmeter,” Precis. Eng., vol. 30, pp. 443-447. 2006.
47 Juuti M., Koivula H., Toivakka M. and Peiponen K.-E., ”A diffractive glossmeter
for local gloss measurement of papers and prints,” TAPPI Coating and Graphics
Arts Conference. 2007.
48 Peiponen K.-E., Alarousu E., Juuti M., Silvennoinen R., Oksman A., Myllylä R.
and Prykäri T., “Diffractive optical element based glossmeter and low coherence
interferometer in assessment of local surface quality of paper,” Opt. Eng., vol. 45,
pp. 043601-043607. 2006.
49 Juuti M., Prykäri T., Alarousu E., Koivula H., Myllys M., Lähteelä A.,
Toivakka M., Timonen J., Myllylä R. and Peiponen K.-E., "Detection of local
specular gloss and surface roughness from black prints," Colloids and Surfaces A:
Physicochem. Eng. Aspects, vol. 299, pp. 101–108. 2007.
50 Silvennoinen R., Peiponen K.-E. and Myller K., Specular gloss, Elsevier,
Amsterdam, The Netherlands. 2008.
51 Myller K., Peiponen K.-E., Silvennoinen R., Tarvainen J.-P., Rainio J. and
Soinila-Oksanen S., “Glossmeter for detection of gloss and wear of concave glazed
ceramic products,” cfi/Ber. DKG (Ceramic Forum/ Deutschen Keramischen
Geselschaft), vol. 81, pp. E39-E42. 2004.
52 Juuti M., Kalima V., Pakkanen T.T. and Peiponen K.-E., "A novel method to
measure and analyse delta gloss by diffractive glossmeter," Meas. Sci. Technol., vol.
18, pp. L5–L8. 2007.

72 THE EFFECT OF PAPER PROPERTIES ON THE COLOR REPRODUCTION FOR DIGITAL…..
The Effect of Paper Properties on the Color Reproduction for

Digital Proofing of Gravure Publication Printing
Yu Ju Wu, Alexandra Pekarovicova and Paul D. Fleming
A-217 Parkview Campus

Western Michigan University
Kalamazoo, MI
y5wu@wmich.edu
Abstract
Gravure is the major printing process used for publication such as long-run magazine,
catalogs, newspaper, and miscellaneous commercial printing jobs. In gravure printing
processes, color proofing is done for the purpose of checking the color to ensure that it is
as desired, before proceeding to cylinder engraving. Recently, the use of digital proofing
gained popularity due to its benefits of high speed, wider color gamut, and affordable
prices for a device. Proofing on the actual production stock more closely predicts print
outcome. However, paper properties have significant effect on the image reproduction for
proofing processes. Working with different inks and devices, the actual production
printing substrate can have a very different color gamut and behave differently in the
digital proofing process. The color capability of a digital proofing system influences color
gamut and performance, which in turn affects accuracy of color for the press to match.
This study discusses how paper properties affect the color reproduction of digital
proofing. Publication printing substrates and manufacturer recommend proofing papers
were tested on the Epson Stylus Pro 4000 printer and Xerox Phaser 8550 printer. Paper
surface properties (roughness, porosity, pore size, formation) and optical properties
(brightness, whiteness, opacity, and gloss) were tested. The correlations between paper
properties and color gamut were investigated. It was found that paper with smooth and
uniform surface, low porosity or air permeability, high brightness or whiteness, and high
paper gloss properties exhibited a high gamut volume.
Keywords: Digital Proofing, Paper Properties, Color Gamut, Digital Printers, Publication
Gravure
1 Introduction
In gravure printing processes, similar to all other printing processes, color proofing for the
purpose of checking the color to ensure that it is as desired, is becoming a must, especially in
the field of catalog publishing. In the past, proofing was done by using a gravure proofing press,
which represented higher cost and longer production time. Digital proofers have the advantages
of high speed, wide color gamut, and affordable prices for a device; therefore, providing
significant time and cost savings compared to conventional gravure proofs.
Generally, proofing is done using the substrate that will be used for actual product printing.
When the digital printer is to be used for proofing, the substrate characteristics have to be taken
into the account [1,2,3]. These characteristics may differ significantly from the
ideal/manufacturer recommended substrate. Digital proofing systems work with spectral
characteristics of the colorants and the printed results could be different from a printing press.
The color gamut of a particular combination of printer, ink, print media, and RIPs is determined
by the uniformity of ink absorbency and color density. Color gamut in turn represents the color
fidelity and color matching results. A digital proofing system with wider color gamut is
desirable to provide better color reproduction capability for proof-press color matching.

Y.J. WU, A. PEKAROVICOVA, P.D. FLEMING 73
The color gamut of a digital proofing system is affected by the printer, software and substrate
involved in digital proofing processes [4,5,6,7]. Paper properties are probably the most
important factors affecting the completeness of the image transfer and image appearance in all
color reproduction processes. As a crucial variable in predicting and reproducing color, paper
has significant influence on the print quality, such as print contrast, density, color, tonal range,
and surface uniformity. The interaction between paper and ink, its porosity, roughness, together
with optical properties such as whiteness, opacity, light scattering, and gloss [8,9] must be
considered in all printing processes.
The printability of a paper surface is influenced by surface properties such as smoothness,

uniformity and the ink absorption into paper [10]. In general, high print quality is associated
with good formation and with smooth, compressible paper [11]. The most important surface
properties of paper include roughness, formation, porosity and air permeability. Roughness and
formation are considered as external surface properties, while pore size, porosity and
permeability are categorized as internal surface properties. Roughness affects ink gloss and
color, as well as print contrast. Formation is an indicator of how uniformly the fibers and fillers
are distributed in the sheet. Formation plays an important role as most of the paper properties
depend on it. A poorly formed sheet will have more weak and thin or thick spots. Paper
formation will affect its printing characteristics [10,12,13]. The porosity of a paper sheet greatly
affects properties such as apparent density, compressibility, resiliency, and the ability to absorb
inks. Most uncoated printing and writing papers are considered moderately porous, whereas
coated printing papers are relatively nonporous. The air permeability is an indicator that shows
how printing inks will penetrate and spread [10,14].
The optical properties of a paper influence the visual quality of the printed image and contribute
to its appearance and appeal. The best color reproduction will occur with papers that are bright
with uniform spectral reflection, smooth, glossy, and neutral shade [1]. The optical properties
include gloss, opacity, whiteness, brightness, and paper color. Brightness and whiteness are
important for print contrast development. The higher the brightness or whiteness, the higher the
contrast between the paper and printed image is [15]. Opacity, or paper’s ability to obstruct the
passage of light, is very important for publication grades [12,16]. Too much show-through of
printed images from the back side of the sheet will reduce print contrast and interfere with the
visual appearance of the image. The gloss of a paper affects the color of a print because it
affects the way light is reflected from the ink. In order to maintain uniform color printing
throughout a job, the variations in gloss must be monitored and kept to a minimum level. Glossy
papers are usually associated with high surface smoothness and good printing quality. High
quality printed images also achieve high gloss by increasing the gloss of printed ink films,
which enhances the brilliance and color intensity of printing [10.11,16].
The proof-press color matching highly depends on the range of colors and tonal ranges that
digital printers can reproduce and also how accurately these color numbers can be manipulated
in comparison to printing characteristics. The former is determined by ink-paper interaction of a
digital printer and the latter needs the aid of color management. Before performing proof-press
color matching, the color gamut of print combination of digital proofing system must be well
defined. This study discusses the color reproduction of digital proofing from the point of view
of paper properties. The correlations between paper properties and color gamut (gamut volume)
were also investigated.
2 Experimental
2.1 Equipment and Materials

The ECI2002R CMYK test target designed for DTP 70 spectrophotometer was employed for
this study. The patches of the target chart are specially designed to produce a good distribution
of colors in L*a*b* space. Three gravure publication printing substrates- free sheet coated
paper, lightweight coated paper, and newsprint- were selected as the actual production printing

substrates. Furthermore, three types of manufacturer recommended proofing papers- semimatte

photo paper, pearl proof paper, and selected proof paper were tested and compared. Table 1
provides basic information of substrates used in this study.
Table 1: Selected properties of tested substrates

Basis Weight [g/m2] Caliper [μm]
Substrates Mean Std. Dev. Mean Std. Dev.
Gravure Free Sheet 67.7 0.6 59 3
publication Light Weight Coated 50.4 0.8 47 2
printing substrates Newsprint 49.4 0.3 73 3
Manufacturer Semimatte Photo Paper 255.4 0.9 258 2
recommended Pearl Proof Paper 237.8 0.8 235 3
proofing papers Selected Proof Paper 192.1 0.7 184 2
Two commercially available digital printers were selected as digital proofing devices: an Epson
Stylus Pro 4000 printer with UltraChrome pigmented inks and a Xerox Phaser 8550 printer with
hot melt inks. The Epson Stylus Pro 4000 printer applies drop-on-demand technology to form
droplets, which land on the paper surface. The digital signal changes by heating or using a
piezoelectric effect [17,18]. The Xerox Phaser 8550 printer utilizes hot melt inks, rather than the
liquid or dry inks used in other processes, but also is actually a piezoelectric drop on demand
ink jet printer. Gravure printing was done on four color Cerutti rotogravure web press, located at
the Western Michigan University (WMU) Printing Pilot Plant. Commercial toluene-based
coated yellow, magenta, cyan and black inks for rotogravure from Flint Group were used. The
ink efflux time with a Shell cup #2 was kept at 22 ± 0.5 seconds for yellow and magenta, 25 ±
0.5 seconds for cyan, and 20 ± 0.5 seconds for black ink, respectively. Printing was done at 600
ft/min with electrostatic assist (ESA) on. The target print densities were 1.00 for yellow, 1.35
for magenta, 1.35 for cyan, and 1.50 for black, which was achieved for the first printed
substrate, and then the same printing condition was maintained during the printing process of
the rest of the substrates.
2.2 Research Procedure

In order to analyze the color gamut for a device, an ICC profile needs to be created. ICC profiles
were generated for the Epson Stylus Pro 4000 and Xerox Phaser 8550 printers, using selected
gravure publication substrates and proofing papers. The devices were profiled as CMYK. For
the Epson Stylus Pro 4000 printer, the ECI2002R CMYK chart was printed on the tested
substrates via CGS ORIS RIP software. For the Xerox Phaser 8550 printer, the ECI2002R
CMYK chart was printed on the tested substrates via its printer driver, which still is CMYK
addressable as a PostScript device. Those printed charts were measured with an X-Rite DTP70
spectrophotometer, operated by GretagMacbeth Measure Tool 5.0.7 software. The measurement
files were used to generate profiles using GretagMacbeth ProfileMaker Pro 5.0.7. ICC profiles
were then loaded into CHROMiX ColorThink Pro 3 software and the gamut volumes of the ICC
profiles were determined.
2.3 Data Collection and Analysis

Ten printed ECI2002R CMYK charts were collected for each substrate. The data of surface
properties (roughness, porosity, permeability, pore size, and formation) and optical properties
(whiteness, brightness, opacity, and paper gloss), together with gamut volumes for selected six
tested substrates were collected and the relationship between paper properties and color gamut
were studied. Paper properties and gamut volume for each paper were measured ten times to
reduce the measuring error. The color gamuts of proofing devices versus gravure press were
compared using ColorThink Pro 3.0 software. The instruments used for paper properties
measurement were summarized in Table 2.

Table 2: Instruments used for paper properties measurement

Paper Properties Measuring Instrument
Roughness [μm] EMVECO stylus profilometer
Formation M/K system formation tester
Pore size [nm] Mercury intrusion porosimetry
Porosity [%] Mercury intrusion porosimetry
Parker Print-Surf (PPS) tester
Air Permeability
Using equation of K = 0.048838 * Q * X (Q=PPS Porosity;
Coefficient [μm2]
X=thickness)
Brightness [%] Brightimeter MICRO S-5 (C/2°)
Whiteness Brightimeter MICRO S-5 (ASTM standard, C/2°)
Opacity [%] BNL-2 Opacimeter (TAPPI Standard)
Paper Gloss [%] Novo-Gloss™ Glossmeter (at 75°)
An Electronic Microgage Model 210-R (Emveco, Inc) with the spherical steel stylus having a
radius of 0.001 inch was used for profilometer measurement. The test conditions were 500
readings per group, 3 groups, 0.1 mm reading space, and 0.5 mm/s scanning speed. The
roughness R was then calculated using [19]:
R = | Xi+1 – Xi | /499, i = 1,2,…,499
An M/K system formation tester was employed to obtain formation index values, which
measure variations in light transmitted through the tested sheet as the light source scans across
the sheet rapidly. Each measurement is amplified and stored in one of 64 optically measured
“basis weight” classes or memory bins. The greater the deviation in optical density from the
instantaneous average, the further away a given data point is stored from the central bin or
average weight class of the histogram. At the end of each scan, the number of contiguous bins
and the amplitude or peak height of the histogram were recorded. The more uniform a sheet, the
greater its peak height is, and the fewer the number of bins into which the data fall. In other
word, the larger the formation index, the more uniform the sheet. The formation index was
calculated based on the equation of:
Formation Index = Peak Height / No. of Bins *1/100
An Autopore IV 9500 mercury porosimeter, measuring the incremental increase of volume
penetrated as the pressure rises, was employed for the porosity-related characteristic
measurements, such as of pore size, volume, and distribution of a paper. The pressure required
to intrude mercury into the sample’s pores is inversely proportional to the size of the pores.
Paper samples were placed in a penetrometer and evacuated at 50 μm Hg [20,21,22,23]. The
permeability of each substrate was calculated from the Parker Print Surf porosity value and its
thickness using the following equation [24].
K = 0.048838 * Q * X
where K is the permeability in μm2, Q is the flow rate of PPS 500 kPa in ml/min and X is the
sheet thickness in m.
A Brightimeter MICRO S-5 was employed to measure brightness % and whiteness complied
with TAPPI standard T 562 (C illuminant with 45°/0° geometry). Since the CIE whiteness [25]
can only be used in a limited region (3 > CIE tint > -3) and the tint value of newsprint was out
of the range, ASTM whiteness [26] was employed for data analysis in this study.
The ASTM Whiteness is given by
WASTM = 3.388Z – 3Y
where Y and Z are tristimuli measured with an illuminant C reference and a 2° observer. By
construction, a perfect C/2 white point (Y = 100, Z = 118.06) has a WASTM of 100.

A BNL-2 opacimeter was employed to test opacity, which is using TAPPI standard T 425
(opacity = R0 / R0.89). Paper gloss at 75˚ was measured using a Novo-Gloss™ Glossmeter based
on TAPPI standard T 480.
3.1 Color Gamut Comparison

The color gamut is the range of colors that a particular combination of printer, ink, print media,
and RIP can achieve. The proof-press color gamut comparison for the free sheet (FS), light
weight coated (LWC) and newsprint is discussed below.
Free Sheet (FS)

The free sheet was printed on the Epson Stylus 4000, Xerox Phaser 8550 printers and Cerutti
rotogravure press. When proofing on the free sheet coated paper with the Epson Stylus Pro 4000
printer, it was found that UltraChrome inks have severe ink smearing problems and cannot lay
on the paper surface properly. Therefore, free sheet coated paper with Epson Stylus Pro 4000
printer combination was excluded in further discussion. Figure 1 illustrates the proof-press color
gamut comparison for the free sheet coated paper. As seen from Figure 1, proofing papers,
printed via Epson Stylus Pro 4000 printer, yield wider color gamuts than press gamut for the
free sheet coated paper. The color gamuts of Epson Stylus Pro 4000 printer are wider than those
of the Xerox Phaser 8550 printer. For the Xerox Phaser 8550 printer, whether on the proofing
papers or free sheet coated paper, the digital printer tends to have a greater color gamut in
yellow and magenta areas. The Xerox Phaser 8550 printer gamut is larger in the higher L*
values area, while the press gamut is larger in the lower L* values area. This behavior in
highlights (large L*) is more a function of the whiteness/brightness of the proofing papers used,
than of the printer itself. There are some saturated colors that the press can achieve that the
digital printer cannot, and vice versa.
Proof gamut:
Proof gamut:
Pearl proof paper
Semimatte photo paper
with Epson 4000
with Epson 4000
Press gamut
Proof gamut:
Proof gamut:
Selected proof paper
Pearl proof paper
with Epson 4000
with Xerox Phaser 8550
Proof gamut:
Semimatte photo paper Proof gamut:
with Xerox Phaser 8550 Selected proof paper
(a) proof (FS with Xerox (b) proof (proofing papers )

Phaser 8500, black v.s. press (FS, true color)
wireframe) v.s. press
(FS, true color)
Figure 1: Proof-press color gamut comparison for the free sheet coated paper
Light Weight Coated Paper (LWC)

Figure 2 illustrates the proof-press color gamut comparison for the lightweight coated paper.
When proofing on the light weight coated paper, the Epson Stylus Pro 4000 printer yielded
greater color gamut in the yellow area, but smaller color gamut in magenta and red regions. The
press gamut is larger in the lower L* values area where Epson Stylus Pro 4000 cannot achieve
the dark shadow details. As expected, the color gamuts of selected proofing papers were wider
than that of lightweight coated paper that was printed via the Cerutti rotogravure web press. The
color gamut of Xerox Phaser 8550 printer is similar to that of press except for yellow and
magenta regions. The Xerox Phaser 8550 printer gamut is larger in the higher L* values area,
while the press gamut is larger in the lower L* values area. Again, there are some saturated
colors that the press can achieve that the digital printer cannot, and vice versa.

Proof gamut:
Proof gamut:
Pearl proof paper
with Epson 4000
with Epson 4000
Press gamut
Proof gamut:
Pearl proof paper
Proof gamut: with Xerox Phaser 8550
Selected proof
with Epson 4000
Proof gamut:
(a) proof (LWC with Epson (b) proof (proofing papers )

4000, black wireframe; v.s. press (LWC, true color)
LWC with Xerox Phaser
8500, red) v.s. press
(LWC, true color)
Figure 2: Proof-press color gamut comparison for the light weight coated paper
Newsprint
The proof-press color gamut comparison for the newsprint printed via the Epson Stylus Pro
4000 printer is shown in Figure 3. The color gamuts of proofing papers were all wider than that
of newsprint that was printed via the Cerutti rotogravure web press. When proofing on the
newsprint, the color gamut of the Epson Stylus Pro 4000 printer is similar to that of press, while
the color gamut of Xerox Phaser 8550 printer is relatively larger than that of press. The Epson
Stylus Pro 4000 printer gamut is larger in the higher L* values area, while the press gamut is
larger in the lower L* values area. As shown in Figure 3, proofing whether on the proofing
papers or printing substrate, the color gamut of Xerox Phaser 8550 printer is larger than that of
press. This was actually expected, because hot melt ink solidifies fast, prior to being able to
absorb into porous newsprint sheet.
Proof gamut:
Proof gamut:
Pearl proof paper
with Epson 4000
with Epson 4000
Proof gamut:
Pearl proof paper
Proof gamut: Press gamut

Selected proof
with Epson 4000
Proof gamut:
(a) proof (newsprint with (b) proof (proofing papers )

Epson 4000, black wireframe; v.s. press (newsprint, true color)
newsprint with Xerox Phaser
8500, red) v.s. press
(newsprint, true color)
Figure 3: Proof-press color gamut comparison for the newsprint (Epson Stylus Pro 4000
printer)
Table 3 shows the proof-press color gamut comparisons in terms of gamut volume. Compared
to the press gamut volume, the Epson Stylus Pro 4000 printer has larger gamut volume for the
selected proofing papers but smaller gamut volume for the actual production printing substrates.
The Xerox Phaser 8550 printer, however, produces larger gamut volume for proofing and
printing substrates, with the exception of lightweight coated paper. Working with different ink
technology, the gamut volume of the Epson Stylus Pro 4000 printer is larger than that of the
Xerox Phaser 8550 printer for the selected proofing papers.
Table 3: Gamut volume comparisons for recommended and printing substrates
Cerutti Rotogravure Epson Stylus Pro Xerox Phaser
Substrates Web Press 4000 Printer 8550 Printer
Free Sheet 332,000 ---------- 337,000
Light Weight Coated 318,000 219,000 309,000
Newsprint 149,000 123,000 265,000
Semimatte Photo Paper ---------- 700,000 421,000
Pearl Proof Paper ---------- 684,000 400,000
Selected Proof Paper ---------- 699,000 409,000

3.2 Paper Properties

Surface Properties
Roughness and formation measurements of tested substrates are shown in Table 4. Generally,
coated papers tend to have a smoother and uniform surface. The average roughness readings for
the proofing papers are in the range of 0.7 to 1.2 micron. Compared to other papers, newsprint
tends to have rougher surface and a larger standard deviation value. It is interesting to note that
lightweight coated paper has poor formation. The uniformity of newsprint used in this study is
better than that of lightweight coated paper.
Table 4: Roughness measurement

Substrates Emveco roughness [μm] Formation Index
Mean Std. Dev. Mean Std. Dev.
Free Sheet 0.96 0.12 105 4
Light Weight Coated 1.38 0.14 11.8 0.9
Newsprint 3.03 0.13 120 3
Semimatte Photo Paper 1.06 0.05 318 0.5
Pearl Proof Paper 0.78 0.06 212 4
Selected Proof Paper 1.13 0.08 59 3
Table 5 shows that the highest percent porosity was found in newsprint with 61.4%, followed by
lightweight coated (59.8%), free sheet (41.3%), pearl proof paper (37.3%), and semimatte photo
paper (36.7%). Selected proof paper has the lowest percent porosity at 26.4%. Newsprint has the
highest air permeability value of 9.27 x10-4 μm2 and largest standard deviation value, compared
to the other substrates. The average pore size readings for the proofing papers are in the range of
25 to 50 nm, whereas the average pore size readings for gravure publication printing papers are
larger than 200 nm. The average pore size of newsprint is up to 546 nm.
Table 5: Porosity-related characteristics
Total pore Ave. pore Porosity Air permeability
area [m2/g] diameter [nm] [%] [μm2]
Free Sheet 16.0 204 41.3 11.1 x 10-6
Light Weight Coated 16.3 203 59.8 10.6 x 10-6
Newsprint 6.49 546 61.4 927 x 10-6
Semimatte Photo Paper 36.9 47.4 36.7 8.9 x 10-6
Pearl Proof Paper 43.9 36.1 37.3 0.36 x 10-6
Selected Proof Paper 41.2 28.6 26.4 1.24 x 10-6
The pore size distribution curves are presented in Figure 4. The proofing papers show two peak
groups of pore sizes between 1500 to 1650 nm and 12 to 20nm, respectively. Lightweight
coated paper and free sheet have one peak group of pore size between 800 to 1000 nm, whereas
newsprint shows one maximum centered at 7000 nm. Coating is designed to make surfaces
smoother and less porous; therefore these results are not very surprising. Large maximum for
newsprint is quite understandable, because newsprint is not coated. The pore sizes over 10,000
nm were not considered, since they originate from raw stock porosity and possible artifacts.

Log Differential Intrusion vs. Pore Size

Log Differential Intrusion vs Pore size
Semimatte Photo Paper
C:\...\OMNOVA\GRAVUR~3.SMP Pearl Proof Paper
C:\...\OMNOVA\122706-1.SMP Selected Proof Paper
C:\...\OMNOVA\122706-2.SMP
Light Weight Coated Paper
C:\...\OMNOVA\517.SMP Free Sheet
C:\...\OMNOVA\FS#45.SMP Newsprint
C:\...\OMNOVA\122706.SMP
0.6
0.5
Log Differential Intrusion (mL/g)
0.4
0.3
0.2
0.1
0.0
10,000 1,000 100 10
Pore size Diameter (nm)
Figure 4: Pore size distribution from mercury porosimetry curves
Optical Properties
The optical properties measurements of tested papers are summarized in Table 6. Whiteness
provides a better idea about papers’ color reproduction capability because it measures
reflectance across all wavelengths in the visible light spectrum [27]. Semimatte photo paper and
selected proof paper containing optical whitener were measured automatically by build-in ultra
violet cut-off filter. It is not surprise that proofing papers have higher percent brightness and
whiteness. The opacity readings for all tested papers are in the range of 90% to 97%, with the
exception of lightweight coated paper (with relatively low opacity of 82%). The highest paper
gloss % was found in Pearl Proof paper with 63%. The free sheet coated paper has high paper
gloss value of 62%, but has the largest standard deviation value (2.5). Newsprint has the lowest
paper gloss %, comparing to other substrates.
Table 6: Optical properties measurement
TAPPI ASTM TAPPI Opacity Paper Gloss at
Substrates Brightness [%] Whiteness [%] 75° [%]
Std. Std.
Mean Std. Dev. Mean Mean Mean Std. Dev.
Dev. Dev.
Free Sheet 79.3 0.2 77.0 0.4 90.8 0.6 61.5 2.5
Light Weight Coated 69.9 0.1 42.4 0.3 82.0 0.7 49.7 1.9
Newsprint 56.4 0.5 34.6 0.5 90.3 0.9 27.2 1.2
Semimatte Photo 95.7 0.2 109.8 0.9 96.7 0.3 47.5 0.4
Paper
Pearl Proof Paper 89.3 0.1 89.0 0.2 94.6 0.5 63.1 1.5
Selected Proof Paper 93.0 0.3 99.0 0.8 90.1 0.5 57.0 1.0
Figure 5 represents color spectra for all tested papers. The color spectra data were measured by
using GretagMacbeth SpectroScanT spectrophotometer. As shown in Figure 5, proofing papers
tend to have higher reflectance values in the blue region (420-470 nm), resulting in blue white
appearance, most likely due to the presence of optical brighteners [28,29]. Conversely, light

weight coated paper and newsprint has higher light reflectance in the red region of the spectrum
and thus looks warm white.
100
90
80
Reflectance [%]
70
Semimatte Photo
60
Pearl Proof
50
Selected proof
40
Free Sheet
30
Light Weight Coated
20
Newsprint
10
0
380 430 480 530 580 630 680 730
Wavelength [nm]
Figure 5: Color spectra of tested papers
3.3 Correlation Analysis

Pearson's correlation coefficient was used to measure the association between gamut volume
and paper properties. The correlation analysis between gamut volume and paper properties for
the Epson Stylus Pro 4000 printer is displayed in Table 7. According to Table 7, the gamut
volume of the tested papers significantly correlates with all paper properties at the 0.05 level. It
shows that there is strong positive relationship between gamut volume and formation index,
whereas the relationship between gamut volume and roughness is negative, meaning smoother
papers have larger color gamuts. Gamut volume also has a negative relationship with porosity-
related characteristics. It also found that the relationships between gamut volume and optical
properties are positive. The brighter and whiter the paper is, the larger the gamut volume. This
is consistent with most observations, but there are some exceptions [29].
Table 7: Pearson correlation between gamut volume and paper properties for Epson Stylus Pro
4000 printer
Color Gamut Volume
Paper Properties
Sample Size Sig. (2-tailed) Pearson Correlation
EMVECO Roughness 5 .000 -.807*
Formation Index 5 .000 .882*
Porosity 5 .000 -.707*
Pore Size 5 .041 -.894*
Air Permeability 5 .000 -.701*
Brightness 5 .000 .978*
Whiteness 5 .000 .979*
Opacity 5 .000 .689*
Paper Gloss 5 .000 .753*
* Correlation is significant at the 0.05 level (2-tailed).
The correlation analysis results for the relationship between the gamut volume and paper
properties for the Xerox Phaser 8550 printer are exhibited in Table 8. Significant relationships
also were found between gamut volume and all paper properties (at the 0.05 level). It shows that
the rougher the paper, the smaller the gamut volume. With increasing brightness and whiteness,
gamut volume increases.

Table 8: Pearson correlation between gamut volume and paper properties for Xerox Phaser
8550 printer
Color Gamut Volume
Paper Properties
Sample Size Sig. (2-tailed) Pearson Correlation
EMVECO Roughness 6 .000 -.776*
Formation Index 6 .000 .842*
Porosity 6 .000 -.722*
Pore Size 6 .008 -.928*
Air Permeability 6 .000 -.715*
Brightness 6 .000 .990*
Whiteness 6 .000 .996*
Opacity 6 .000 .620*
Paper Gloss 6 .000 .631*
* Correlation is significant at the 0.05 level (2-tailed).
4 Conclusions
The color gamut of the digital printers is affected by the substrate properties. The digital printer
may be designed for a variety of substrates; however, the largest gamut and highest print quality
of the device is obtained when the manufacturer recommended substrate being used. The device
that is capable of reproducing a large gamut of colors on ideal substrates, may loose this
advantage when using a different substrate.
The results show that the color gamut of proofing papers was wider than that of actual
production printing papers printed via a Cerutti rotogravure web press. Proofing on the actual
production printing papers with Epson Stylus Pro 4000 printer tends to yield wider color gamuts
in yellow area but smaller color gamuts in magenta and red regions. The press gamut is larger in
the lower L* value area which the digital printer cannot reach because of low black density. For
the Xerox Phaser 8550 printer, proofing whether on the proofing papers or printing substrates,
digital printer tends to have wider color gamut in yellow and magenta areas. Xerox Phaser 8550
printer gamut is larger in the higher L* value area, while the press gamut is larger in the lower
L* value area. The press gamut is larger in the lower L* value area where digital printer cannot
reach. There are some saturated colors that the press can achieve that the digital printer cannot,
and vice versa. The Xerox Phaser 8550 printer color gamut for newsprint is larger than the press
gamut. Overall, proofing papers yielded larger color gamuts than actual production printing
substrates in terms of gamut volume. Working with different ink technology, the gamut volume
of Epson Stylus Pro 4000 printer is larger than that of Xerox Phaser 8550 printer for the
proofing papers. The Epson Stylus Pro 4000 printer tends to have smaller gamut volumes for
the printing substrates. The Xerox Phaser 8550 printer, however, produces larger gamut
volumes for the printing substrates, except for lightweight coated paper.
The gamut volume of tested papers significantly correlates with all paper properties. There is a
strong positive relationship between gamut volume and formation index, whereas the strong
negative relationship was found between gamut volume and roughness. Gamut volume also has
negative relationship with porosity-related characteristics such as pore size, porosity and air
permeability. The strong positive relationships were found between gamut volume and
brightness/whiteness properties, whereas the relationships strengths between gamut volume and
opacity/paper gloss are moderate. That is, paper with low roughness, high formation index, low
porosity or air permeability, high brightness or whiteness, and high paper gloss properties
exhibits a high gamut volume.
Acknowledgements
The authors thank National Science Foundation grant MRI- 0215356 and Omnova
Solutions for partial support for this work, and CGS Publishing Technologies for software
and proofing paper donations.

References
1 Wales T., “Paper - A Forgotten Variable in the Color Management Process,”

GATFWorld, vol. 16, no. 6, pp. 16-17. 2004.
2 Norberg O. and Andersson M., “The influence of paper properties on color
reproduction and color management,” Proceedings of the IS&T NIP19: International
Conference on Digital Printing Technologies, New Orleans, pp. 836-840. 2003.
3 Bandyopadhyay S., “Effect of paper properties on print quality of ink jet printer,”
Proceedings of the IS&T NIP17: International Conference on Digital Printing
Technologies, Fort Lauderdale, pp. 141-143. 2001.
4 Suchy M., Fleming P. D. and Sharma A., “Spot Color Reproduction with Digital
Printing”, Proceedings of the IS&T NIP21: International Conference on Digital
Printing Technologies, Baltimore, September 18-23,93-97. 2005.
5 Suchy M., Wu Y. J., Fleming P. D. and Pekarovicova A., “Spot Color
Reproduction with Digital Printing”, submitted to JIST. 2007.
6 Wu, Y. J., Suchy, M., Fleming, P. D., and Pekarovicova, A., “Quality Analysis of
Spot Color Reproduction with an Inkjet Printer”, Proceedings of the IS&T NIP22:
International Conference on Digital Printing Technologies, Denver, September 17-
22. 2006.
7 Wu Y. J., Fleming P. D. and Pekarovicova A., “Quality Analysis of Spot Color
Reproduction with Inkjet Printers”, submitted to JIST. 2007.
8 Lee H. K., Joyce M. K. and Fleming P. D., “Influence of Pigment Particles on
Gloss and Printability for Inkjet Paper Coatings”, Proceedings of the IS&T NIP20:
International Conference on Digital Printing Technologies, Salt Lake City, pp. 934-
939. 2004.
9 Lee H. K., Joyce M. K. and Fleming P. D., “Influence of Pigment Particle Size and
Pigment Ratio on Printability of Glossy Inkjet Paper Coatings”, Journal of Imaging
Science and Technology, Vol. 49, No. 1, pp. 54-61. 2005.
10 Wilson L. A., What the Printer Should Know about Paper, 3rd Ed, Pittsburgh, PA:
GATFPress, pp. 79-89; pp.140-187. 1997.
11 Thompson B., Printing Materials: Science and Technology, Pira International,
Leatherhead, pp. 260-284. 1998.
12 Biermann C. J., Handbook of Pulp and Papermaking, San Diego: Academic Press,
pp. 487-508. 1996.
13 Peel, J. D., Paper Science and Paper Manufacture, Bellingham, WA: Angus Wilde
Publications Inc. 1999.
14 Scott W. E. and Abbott, J. C., “Properties of Paper: An Introduction,” Atlanta,
Georgia: TAPPI Press. 1995.
15 Lee H. K., Joyce M. K. and Fleming P. D., “Influence of pigment particles size
and packing volume on printability of glossy inkjet paper coatings,” Proceedings of
the IS&T NIP19: International Conference on Digital Printing Technologies, New
Orleans, pp. 613-618. 2003.
16 Levlin J. E. and Soderhjelm L., Pulp and Paper Testing, Helsinki, Finland : Fapet
Oy, pp.138-159. 1999.
17 Brett G., “Short-run Digital Color Printing,” 2nd Ed, Pira International Ltd, pp. 34-
38. 2001.
18 Smyth S., “Digital Commercial Printing,” Pira International Ltd. 2001.
19 Xu R., Fleming P. D. and Pekarovicova A. and Bliznyuk V., “The Effect of Ink
Jet Paper Roughness on Print Gloss,” The Journal of Imaging Science and
Technology, vol. 49, no. 6, pp. 660-666. 2005.
20 Hrehorova E., Wood L. K., Pekarovic J., Pekarovicova A., Fleming P. D. and
Bliznyuk V., “The Properties of Conducting Polymers and Substrates for Printed
Electronics,” Proceedings of IS&T Digital Fabrication 2005, Baltimore, pp. 197-
202. 2005.

21 Malla P. B., Starr R. E., Werkin T. J. and Devisetti S., “The Effects of Pigment
Type and Pore Structure on Optical Properties and Rotogravure Printability of Light
Weight Coated (LWC) Paper – A Statistically Designed CLC Coating Study with
Mixtures,” 2000 TAPPI Coating Conference and Trade Fair, pp. 407-413. 2000.
22 Johnson R. W., Amick T. J., Abrams L. and Maynard R. B., “Use of HG
Porosimetry to Characterize Pore Structure and Model End-Use Properties of Coated
Papers.” TAPPI Proceedings - 1998 Coating / Papermaking Conference, pp. 337-
354. 1998.
23 Velho J., Santos N., Conceicao S., Ferreira P., Carvalho G. and Ferreira J. M.,
“Technical aspects relatively to surface, structure and print characterization of coated
papers.” Proceedings - PulPaper 2004 Conferences, Coating, pp. 7-14. 2004.
24 Pal L., Joyce M. K. and Fleming P. D., “A Simple Method for Calculation of the
Permeability Coefficient of Porous Media,” TAPPI Journal, Vol. 6, No. 9, pp. 10-16.
2006.
25 CIE.. 15: Colorimetry, Third Edition. 2004.
26 ASTM. “Standard Practice for Calculating Yellowness and Whiteness Indices from
Instrumentally Measured Color Coordinates”, E313-05. 2005.
27 Greenbaum P. J., “Brighter, whiter freesheet trend offers opportunity with
tradeoffs,” Pulp & Paper, Vol. 80, No. 7, pp. 30-33. 2006.
28 Chovancova-Lovell V. and Fleming P. D., “Effect of Optical Brightening Agents
and UV Protective Coating on Print Stability of Fine Art Substrates for Ink Jet”,
Proceedings of the IS&T NIP22: International Conference on Digital Printing
Technologies, Denver, September 17-22, 227-230. 2006.
29 Chovancova-Lovell V. and Fleming P. D., “Color Gamut – New Tool in the
Pressroom?” TAPPI J, in press. 2007.

84 FACTORS AFFECTING THE PRINTING STRENGTH OF KAOLIN-BASED PAPER COATINGS.
Factors Affecting the Printing Strength of Kaolin-Based Paper

Coatings
J.C.Husband, J.S.Preston, L.F.Gate, D.Blair† and P.Creaton‡.
†
Pigments for Paper Group Dept of Chemistry
Imerys Minerals Ltd University of Edinburgh
Par Moor Centre King’s Building
Par West Mains Road
PL24 2SQ Edinburgh
United Kingdom, EH9 3JJ,
john.husband@imerys.com
janet.preston@imerys.com
‡
School of Chemistry
University of Bristol
Cantock’s Close
Bristol
BS8 1TS.
United Kingdom
Abstract
The mechanical properties of paper coating layers are very important in printing and
converting operations. These include stiffness, resistance to fold cracking, dusting and
pick resistance. In this work we have studied the pick strength of kaolin coated papers as
a function of the coating pigment size, morphology and binder level.
We found that dry pick strength using oil depended on the pigment particle size. Coarse
clays gave much higher pick resistance than finer clays. A similar trend was observed
using oil-based inks. In the presence of water as a pre-damping fluid, the pick strength
depended on the interval between pre-damping and printing. The coarse clay coatings had
the greatest sensitivity to water, giving the lowest strength at short intervals, but the
highest at long time intervals. Finer clay coatings were less sensitive to damping. The
results indicate that the permeability of the coatings to water is the critical factor affecting
wet pick resistance.
Ink oils also reduced the strength of the coating layer. The extent of strength loss
depended on the polarity of the fluid, being greatest with water and least with mineral oil.
This effect of fluid may be a result of interruption of van der Waals or ion pair bonding
between latex and the kaolin surface. Hence the pick resistance of a coating layer is not a
measure of its intrinsic strength, but is a result of the interactions between the ink and
coating. Penetration of oil from the ink into the coating increases the applied stress
through tack build, and simultaneously reduces the strength of the coating. Improving
adhesion between mineral and binder may be the key to obtaining higher coating strength
with offset inks.
Keywords: Offset printing, coated papers, dry pick strength, wet pick strength, kaolin
aspect ratio

J.C.HUSBAND, J.S. PRESTON, L.F. GATE, D. BLAIR, P. CREATON. 85
1 Introduction
During printing the surface strength of a paper coating is a critical parameter. Papers can exhibit
picking when subjected to the high stresses, which develop during ink application. These
stresses result from the tack build associated with the loss of ink solvent into the pores of the
coating, and the consequent increase in cohesivity of the ink film [1,2,3]. Picking as a result of
the oil-based ink alone is normally referred to as “dry pick”. Additionally, the offset process
uses water, which is known to weaken the coating, leading to the phenomenon of “wet pick”
[1,4,5].
In order to avoid picking, binder must be present in a sufficient amount to give adequate
strength for the combination of pigment and basepaper in use. This is referred to as the “binder
demand” of the pigment, but there is still some uncertainty about which pigment properties are
most important in its determination. It is widely accepted that surface area is most important,
although Cobb [6] was unable to show this. Her work clearly established a correlation of binder
(casein) demand with the pore volume of the mineral-based coating layer. This is not
unexpected, since porosity in a material reduces its strength because the voids have no load-
bearing ability. Using this argument, pigments which increase the porosity of the coating should
require more binder. However, such pigment designs may lose binder more readily into the
basepaper, which might weaken the coating but paradoxically may strengthen the coating –
basesheet interface.
Since Cobb’s paper, there have been a surprisingly small number of published studies on binder
demand. Lepoutre and Hiraharu [7] investigated the differences in z-direction strength of
coating layers based on calcium carbonate (gcc) and kaolin. Using the IGT pick test with oil,
they concluded that the mechanical strength of the gcc coatings was higher than the clay. Using
a rolling cylinder technique to measure z-directional strength, they found that the rupture energy
of the gcc layer was higher than the clay, and therefore a higher level of binder was required for
the clay coating to obtain similar strength to the gcc in the z-direction. Petterson et al [8] also
investigated the wet pick strength of coated papers and concluded that gcc coatings were
stronger than clay coatings under their experimental conditions. They suggested that the role of
water was to interfere with the adhesion between latex and pigment. Skeppstedt et al [9] showed
that surface treating calcium carbonate by in-situ polymerisation of styrene led to improved dry
pick strength with SB latex binder. They attributed this to improved adhesion between the
grafted mineral surface and the latex.
Further studies of the z-directional strength of kaolin and GCC layers were reported by Inoue
and Lepoutre [10]. They found that the peel energy of gcc layers was higher than kaolin layers
at the same binder level, which in this study, somewhat unusually, was carboxymethyl cellulose.
Coatings of monosized polystyrene pigment had the lowest peel energy. The critical binder
content was determined from the peel test and correlated with pigment surface area. They
concluded that coating layers based on plate-shaped clay particles were more resistant to in-
plane stresses than isometric gcc.
Recently, some work has been published on z-directional coating strength of paper using a
micro-indentor [11]. Although applied to coated paper the technique is in an early stage of
development; the z-direction elastic stiffness estimated from the indentor showed order of
magnitude agreement with that calculated from in-plane tensile measurements for coating layers
containing calcium carbonate. For clay, the z-direction result was lower than the in-plane, which
the authors suggested might arise as a result of the anisotropic nature of kaolin particles. An
alternative view was provided by Granier and Sartre [12] who interpreted the strength of
coatings in terms of acid – base character of the mineral and latex. In their model, the kaolin
edges are basic and the faces acidic, so the acidic latex adheres much more strongly to the

edges. Clearly, the effect of surface chemistry on binder – mineral adhesion should form part of
any comprehensive theory of coating strength.
Reinius et al [5] studied the effect of fount solution and ink oil on the in-plane tensile strength of
unsupported coatings and showed that water reduced the strength to 12% of the original “dry”
strength. Oil also reduced the strength, but had less effect than water. These authors also
labelled fount solution on the press and were able to show that the fount solution penetrated the
coating to different extents depending on the coating pigment. Platy clay coatings gave the least
fount penetration and the greatest surface spreading. Narrow size distribution ground calcium
carbonate gave greatest fount penetration and least spreading.
Our own recent studies of coating strength explored the relationship between in-plane and z-
direction tensile strength of unsupported coating layers based on kaolins of different particle
sizes and aspect ratios as a function of latex level [13,15]. This work showed that anisotropy in
the coating layer as a result of the plate-shaped nature of kaolin particles leads to weaker tensile
strength in the z-direction compared to in-plane direction. For a given set of fine clays, there
was an inverse relationship between in-plane and z-directional tensile strength. Isotropic coating
layers based on ground calcium carbonate particles had similar tensile strength in the in-plane
and z-direction. In the present paper, our aim has been to establish whether the above trends
determined for unsupported coating layers also apply to papers coated with the same pigments
during printing. Therefore we used model coatings based on 100 parts kaolin, and plan to extend
the study the carbonate containing coatings in future work. We believe that the results will be
relevant to both web and sheet fed offset printing processes.
Two pairs of coating kaolins were used having approximately similar particle size distributions
as determined by sedimentation. These were chosen to cover particle shape factor (SF) values
from blocky (SF 10-16) to platy (SF 35-42). The mean (D50) particle size of the first pair was
0.40 µm equivalent spherical diameter (esd). The second pair was finer, with a mean size of
0.20µm. These four clays (FB, FP, UFB and UFP) were used in our previous studies [13,15].
An additional coarse high shape factor clay (CP) of mean size 2 µm was included in this study.
The properties of the kaolins are summarised in Table 1. Shape factor was measured using a
proprietary technique [14]. The latex binder used throughout was a carboxylated styrene
butadiene acrylonitrile copolymer of Tg = 10°C (DL920, Dow Chemical). The kaolins were
slurried at the optimum solids using 0.3wt% of a sodium polyacrylate dispersant - (CED3546,
Ondeo Nalco). For purposes of measuring the tensile strength of unsupported coatings, latex
was added at levels between 6 and 17 pph based on clay. 0.3 pph of sodium carboxymethyl
cellulose (Finnfix 10™, CP Kelco) was also added as a thickener. After pH adjustment to 8.0,
the colours were screened through a mesh of 53 µm aperture.

Table 1: Physical properties of kaolins
Kaolin Description Particle size distribution by Sedigraph™, Shape BET

wt% below factor surface
2µm 1µm 0.5µm 0.25µm D50 µm area,
m2g-1
CP Coarse 51 29 14 6 2.0 42 6
platy
FB Fine 93 81 63 34 0.37 16 14
blocky
FP Fine platy 90 74 55 31 0.44 34 16
UFB Ultrafine 100 99 94 68 0.19 10 27
blocky
UFP Ultrafine 97 90 79 53 0.23 40 24
platy
Unsupported coating films for the measurement of tensile strength were prepared by drawdown
onto polyester film, drying and subsequent removal from the substrate. In order to be self-
supporting, a coating thickness of between 50 and 100 µm was required. Tensile strength
measurements were made using the methods described in our earlier papers (xy direction, [13],
z-direction [15]).
Coated papers for print testing were prepared using these clays. The latex level for these
experiments varied from 8 – 14 pph with sodium carboxymethyl cellulose added at 0.3 pph. The
coatings were applied to a 46 gm-2 LWC mechanical basestock at 400 m min-1 using a
laboratory cylindrical coater (Heli-Coater™) equipped with infrared and hot air drying. The
coated strips were supercalendered for 10 nips at a roll surface temperature of 65°C and a
pressure of 89 kg cm-1 using a laboratory calender (Perkins).
The mean pore size and volume of unsupported coatings and coated papers were measured by
mercury intrusion using a Pascal 240 porosimeter (CE Instruments). Corrections for glassware
expansion were applied. When using coated papers, curve fitting was applied to separate the
pores associated with the basepaper from the smaller coating pores.
The ink setting rates of the coated papers were measured using the Ink Surface Interaction
Tester (ISIT, SeGan,) [2] using the Huber tack #1 and #3 inks used in the pick test.
The dry pick strength of the coated papers was measured using an AIC2-5 unit (IGT). The unit
was run in accelerating mode up to a maximum velocity of 2 ms-1. 1 cm3 of a standard pick oil
of viscosity 17 Pa.s was applied and the print was assessed visually using low angle
illumination to assess the speed and mode of failure, coating or basestock. Measurements were
made in triplicate. See Appendix for more discussion of this test.
The wet pick strength was measured using a Multipurpose Printability Tester (Prufbau). 0.5 cm3
of ink (Huber tack #1 or #3) was applied to the distribution rollers followed by distilled water to
the damping unit. The instrument allows the delay time between damping and printing to be
varied. The velocity at which picking first occurred was assessed visually and also the mode of
failure was recorded. Again the measurements were performed in triplicate. The delay time was
varied between 5 and 11 s and volume of water between 0 and 15 µl.
In all results given in this paper, picking was coating pick, not basestock or interface failure.

The absorption of fount solution was studied using the IGT printing unit by applying a
controlled volume of a 50 : 50 water : 2-propanol mixture containing a dye. This was applied to
the paper in the printing nip and the speed of absorption estimated by measuring the stain area
using image analysis.
Print density values were assessed after printing using an IGT laboratory printing unit at a speed
of 0.5 m s-1 at 500 N pressure, 0.3 cm3 of magenta sheetfed ink was applied to give a 1 µm thick
ink film. A portion of the print area was predamped with 1 gm-2 of water, using a very short
delay time of the order of 1 s. A SpectroEye™ densitometer (Gretag) was used to measure
optical density.
3 Results
3.1 Tensile strength of unsupported coatings

Table 2 summarises the tensile strength results for these clay coatings in the in-plane and z-
direction at 11 pph latex. These are taken from our previous paper [15]. Moving to finer or more
platy clays increases the in-plane strength but lowers the z-direction strength. The coarse platy
clay (CP) is an exception to this trend, exhibiting the strongest in-plane and z-direction strength.
Table 2: Summary of tensile strength values of unsupported coatings, 11 pph latex
Clay In-plane tensile strength, MPa z-direction tensile strength, MPa

CP 10.0 Too strong to measure
UFP 7.0 0.30
UFB 4.9 0.53
FP 4.6 0.63
FB 2.8 0.69
3.2 Tack development

The tack development curves (ISIT) for papers coated with kaolins FB, FP and CP at latex
levels of 9.5 and 12.5 pph are plotted in Fig. 1. The coarse kaolin CP gives coatings with the
slowest tack rise and the slowest tack decay. Next slowest is fine platy kaolin FP, followed by
fine blocky kaolin FB, showing the fastest tack decay. In all three systems increasing the
amount of latex slows down ink setting, although the effect is secondary to the effect of
changing kaolin.
10 10
9 9
8 8
Separation force, N
Separation force, N
7 7
Clay FB 9.5 pph
6 6
Clay FP 9.5 pph
5 5
Clay CP 9.5 pph
4 4
3 3
2 2 Clay FB 12.5 pph
Clay FP 12.5 pph
1 1
Clay CP 12.5 pph
0 0
0 50 100 150 0 50 100 150
Time, s Time, s
Figure 1: Tack development curves (ISIT) for coated papers containing kaolins FB, FP and
CP at low and high latex levels using #3 tack ink.

3.3 Pick strength of coated papers

The results for the IGT dry pick evaluation are shown in Fig. 2. The results show that pick
strength increases as the particle size of the pigment increases. There is no discernible influence
of particle shape, since clay pairs FB and FP and UFB and UFP give very similar responses.
The coarse platy clay CP gives a much higher dry pick strength than the finer pigments, a
similar trend to that shown by the tensile strength in Table 2. In broad terms, the trend in dry
pick strength follows that of the z-direction tensile, without the sensitivity to particle shape.
170
160 Clay FB
150 Clay FP
140 Clay UFB
-1
Clay UFP
Picking speed, cm s
130
Clay CP
120
110
100
90
80
70
60
50
40
7 8 9 10 11 12 13 14 15
pph latex
Figure 2: Dry pick strength of coated papers.
The pick speed of some of the papers was also measured on the Prubau tester using the Tack #3
ink with no pre-damping. The results (Fig. 3) broadly follow the dry pick trends, showing that
the coarse platy clay CP has the highest resistance to pick with this ink also.
120
Clay UFB
110 Clay UFP
Clay FB
100 Clay FP
Picking speed, cm s-1
Clay CP
90
80
finer
70
60
50
40
30
20
9 10 11 12 13
pph latex
Figure 3: Comparison of pick resistance measured using #3 tack ink but no pre-damping.

As expected, the effect of pre-damping with water was to lower the picking velocity of the
coating. However the delay time between damping and printing strongly influences this
weakening effect (Fig. 4). (The shortest delay that can be obtained with the Prufbau apparatus is
5 s). The coarse platy clay CP has the greatest sensitivity to pre-damping, especially at very
short delay times, when this pigment gives the lowest picking velocity of all the samples. At
longer delay times, the situation is reversed and this pigment gives coatings with the highest
picking velocity. It is interesting to note the reversibility of the detrimental effect water has on
strength – given sufficient delay after damping the pick strength values in Fig. 4 return to close
to the un-damped value. A similar trend towards lower pick velocity was found when the delay
time was kept constant at 11 s but the volume of water increased (Fig. 5). When a large volume
of water was added, the effect on strength of increasing the amount of latex binder was small
(Fig. 5). This implies that adding more binder is not an effective way to overcome severe wet
pick issues, unless pore sealing effects can reduce the ingress of water, but this may induce
water interference instead.
60.0
55.0
50.0
45.0
40.0
35.0
30.0
25.0 Clay FB
20.0
Clay FP
15.0
10.0 Clay CP
5.0 Clay UFP
0.0
4 5 6 7 8 9 10 11 12
time delay between damping and printing, s
Figure 4: Effect of delay time between damping and printing, 11 pph latex and 10 µL water
for pre-damping.

75.0
70.0 5 µL
65.0 10 µL
60.0 15 µL

55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
7 8 9 10 11 12
pph latex
Figure 5: Effect of amount of water time used for pre-damping, Clay CP, constant delay
time of 11 s.
3.4 Print density of printed papers

An offset printing evaluation using the IGT was conducted on the papers coated with clays FB,
FP and CP. Table 3 summarises the print density values obtained in the litho and dry regions of
the prints at 3 latex levels:
Table 3: offset litho print densities of coated papers, 10 – 12 gm-2 coat weight
Clay Latex level, pph Print density, Print density, Pre-damped : dry
Pre-damped dry region ratio
region
CP 9.5 Wet pick 1.27 -
11.0 1.26 -
12.5 1.28 -
FB 9.5 1.34 1.33 1.0
11.0 1.32 1.37 0.96
12.5 1.31 1.36 0.96
FP 9.5 1.31 1.34 0.98
11.0 1.30 1.36 0.96
12.5 1.33 1.38 0.96
The coatings containing the coarse platy clay CP gave severe picking in the pre-damped region
and this made measurement of print density impossible. In the dry region, coatings with this
clay had lower print density values than the finer clays. Both the finer clays gave pre-damped:
dry print density ratios between 0.96 and 1.0, indicating very little ink refusal. Hence there was
no evidence for water interference in these coatings, even at the highest binder levels. It is
interesting that the coarse platy clay coatings showed severe wet pick in the pre-damped area,
which corroborates the wet pick trends plotted in Fig. 4. The delay time between damping and
printing was about 1 s, the same order of magnitude as that observed to severely reduce the pick
resistance in Fig. 4.

3.5 Mercury porosimetry

Mercury intrusion results for coating films containing these clays are plotted in Figure 6. The
coarse platy clay CP gives coatings with much larger pores (by a factor of 2) than the finer
clays. The pore volumes are similar for all the clays. Also plotted are the results for coated
papers, showing that both the pore size and volume are lower than in unsupported films. This is
because the coated papers were calendered. We were unable to obtain good pore volume data
for the coated papers with clay CP, as the coating pore size overlapped with the base-sheet
pores. This confirms that the mean pore size is much larger than for clays FB and FP. The larger
coating pore size given by clay CP means that, certainly at longer timescales, the permeability
to free liquids will be high [16]. On the other hand the capillarity will be low, so there will be a
smaller driving force for liquid imbibition. This will give a slow rate of ink setting since the
driving force for oil removal from the ink layer is capillary pressure. It is likely that in coatings
based on clay CP, the larger pores will increase the rate of permeability to the pressure driven
application of water, leading to the trends shown in Fig. 4.
400 0.35
Clay FB
350 Clay FP 0.30
Clay CP
Pore volume, cm 3 g-1

mean pore diameter, nm
300 0.25
Closed circles
Coating films
250 0.20
Open circles
Coated paper
200 0.15
150 0.10
0.05
100
0.00
50
0 2 4 6 8 10 12 14 16 2 4 6 8 10 12 14 16
pph latex pph latex
Figure 6a: Coating pore size by mercury Figure 6: Coating pore volume by mercury
intrusion porosimetry for clays FB, FP and intrusion porosimetry for clays FB, FP and
CP. CP.
This conclusion was supported by measurement of the absorption of a stained water / 2-

propanol mixture in the IGT printing nip. The area of the stain showed a strong correlation with
the wet pick velocity (Fig. 7). A lower stain area indicates faster water absorption, showing that
the coarse platy clay gives the fastest absorption of water in the printing nip.

40
FB
30
Wet pick velocity, cm s-1
FP
20
CP
10
0
250 275 300 325 350 375
Stain area, arbitrary units
Figure 7: IGT stain area plotted against wet pick velocity for papers coated with clays CP,
FB and FP and 11 pph latex.
3.6 Effect of liquids on tensile strength of coating layers

The effects of various fluids on the tensile strength of coating layers were also investigated. We
approached this in a similar way to Reinius et al [5] by soaking pre-cut tensile pieces of
unsupported coatings based on ultrafine clay UFP in liquids of varying polarity ranging from
water to vegetable, aliphatic mineral (obtained from an ink manufacturer) and silicone oils. The
excess liquid was removed with filter paper and the in-plane tensile strength measured as
described in our previous paper [13]. The time between immersing the strips in liquid and
measuring the tensile strength was maintained at about 60 s. Measuring the z-direction strength
of saturated coatings was not possible since the method relies on adhesion to a water- soluble
tape.
Results are plotted in Fig. 8 for an immersion time of 60 s. The results for water reproduce the
wet pick trends very well. The tensile strength falls from around 5 MPa at 12 pph latex to ~ 0.4
MPa when wetted. Note that when saturated with water, increasing the latex level has little
effect on strength (compare with Fig. 5). Re-drying the strips in air at room temperature restored
the strength to close to the original level. A 50 / 50 mixture of 2-propanol and water behaved in
an identical way to water.
Experiments showed that oils also reduced the tensile strength of coatings. After 1 minute, the
more polar vegetable oils reduced the strength by about two-thirds. Mineral oil, being less polar,
had a smaller effect on strength. An experiment using silicone oil (BDH, 100 cs) showed this oil
to be the least interactive, so that the strength was less affected (Fig. 9).
To check the possibility that the strips were not fully saturated with oil, a further set were
prepared with an immersion time of 30 minutes. These results are plotted in Fig. 9(b). Clearly,
strength continues to fall with time, but the effect of the polarity of the fluid is still evident. The
caliper of the strips did not change after immersion for 30 minutes, indicating that no swelling
of the coating had taken place.

7.0
6.0
Original
Tensile strength, MPa 5.0 Water
50% propanol
4.0 Water redried
Linseed oil
3.0
Wood oil
2.0 Mineral oil
Silicone oil
1.0
0.0
4 6 8 10 12 14 16 18
pph latex
Fig. 8: Tensile strength of unsupported coatings at a range of latex levels based on Clay
UFP after immersing in fluids of varying polarity (contact time of 1 minute).
5 5
No solvent No solvent
Water / alcohol
Tensile strength, MPa
4 Water / alcohol
Water / redried 4
Vegetable oils
Vegetable oils
Non-polar oils Non-polar oils
3 3
2
2
1
1
Too weak to
0 measure
0
t
l
er
l
d
l
A
en
oi
oi
oi
oi
ie
:IP
at
er
l
l
lv
l
d
t
d
e
dr
al
oi
oi
oi
oi
en
W
oo
on
ee
er
so
at
er
re
al
d
e
d
lv
at
ns
W
W
l ic
in
on
ee
oo
er
+
o
so
w
M
N
Li
Si
er
ns
lic
in
W
0
o
at
M
:5
Li
Si
N
W
50
(a) (b)
Fig. 9: Tensile strength of unsupported coatings based on Clay UFP saturated with fluids of
varying polarity interpolated to 8.5 wt% latex. (a) after 1 minute, (b) after 30 minutes
contact with fluid.
We also made some coatings based on talc to compare with kaolin. Talc differs from kaolin in
having basal surfaces with siloxane instead of hydroxyl groups [17] and therefore has a much
lower affinity for polar fluids. We used a commercial talc slurry (C10, Mondo Minerals Oy,
Finland), which also contains a wetting agent. The talc had a surface area of 5.5 m2g-1 and an
average particle size of 2.0 µm esd.

9
No solvent
8
Water

7 Vegetable oils
6 Non-polar oils
5
4
3
2
1
0
er
l
nt
l
oi
oi
at
ve
al
W
ee
l
er
so
ns
in
o
M
Li
N
Fig. 10: Tensile strength of unsupported coatings of talc with 8.5 wt% latex after contact
with fluid for 1 minute.
The results (Fig 10) showed that talc coatings were significantly less sensitive than kaolin to
water. Talc retained 23% of its dry strength when wetted for 1 minute compared to only 7% for
kaolin. The effect of linseed oil was similar for both minerals, ~ 30%. The presence of a wetting
agent, essential for slurry preparation, may complicate the interpretation of these results.
Rousu [18] studied the interaction between latex films and ink oils in some depth. She showed
that certain types of oil, namely aromatic mineral oil and tall oil, were absorbed by latex films to
a much greater extent than other vegetable or mineral oils. These oils also tackified the latex
film. She also showed that acrylonitrile latexes, such as we have used here, had reduced oil
interactivity.
4 Discussion
This study has shown that many factors influence the surface strength of coated papers during
the offset printing process. Only some of these relate to the intrinsic strength of the coatings
themselves, as measured by tensile testing on unsupported layers. Other important factors relate
to the penetration of fluid, water or ink vehicle, into the pores and binder phase of the coating.
When applying ink alone, penetration of ink vehicle from the ink into the coating controls the
rate of tack build, and hence the force experienced by the coating during printing.
When we consider the strength of coatings in the presence of water, the situation becomes more
complex. Our results have shown that coatings, which are inherently very strong when
measured using the dry pick test, can be extremely weak when pre-damped. However, after the
pre-damping water has evaporated from the coating layer, the strength is regained. Hence to
successfully predict wet pick problems on the press, the delay time between damping and
printing is a crucial factor.
A coating consisting of a very coarse kaolin had intrinsically high tensile strength, both in-plane
and in the z-direction, when measured on unsupported coating layers. When coated onto paper,
coatings based on this clay also had the highest dry pick strength. They also exhibited the
highest strength when printed dry with a high tack ink. However, when pre-damped using a
short delay time of the order of 5 s, these pigment coatings had the lowest wet pick strength.
The strength was regained with longer delay times between damping and printing. Coatings

with this pigment exhibited the slowest rate of ink setting. This is related to the large average
pore size, hence low capillarity, which leads to a slow rate of removal of vehicle from the ink
layer.
The mechanism by which water causes such a catastrophic strength loss is not understood.
Latex binders are not strongly hydrophilic and are insoluble in water. Petterson et al [8]
concluded that water must attack the interface between the latex and pigment. Given the
hydrophilic nature of kaolin, we think that the most likely explanation is that water temporarily
forms a weak boundary layer, which disrupts the bonding between the latex and mineral surface.
The implication of this is that the latex – kaolin bonds are quite weak, probably originating from
van der Waals forces, or maybe ion pair bonding. Some support for this hypothesis comes from
the observation that coatings based on talc, which has a hydrophobic surface, were less sensitive
than kaolin to weakening by water.
When oil is the penetrating fluid, strength loss also occurs, but not to the same extent as we
observed with water. The extent of strength loss was related to the polarity of the oil. With the
more polar vegetable oils, after 30 minutes the effect on strength is almost as great as with
water. A weak boundary layer may again be formed since the slightly polar vegetable oils also
have an affinity for the kaolin surface [18]. It is well known that oily substrates are very difficult
to glue. Silicone oil exhibited the least weakening effect. Hence the high pick strength of coarse
clay coatings may in part be related to the slow rate of ink oil penetration, since the tack build
and weakening effects will be low. It is also likely that, after printing, papers in which non-
drying ink oil remains in the pores will have very weak coatings and this may contribute to
various problems encountered after printing, such as scuffing.
In the Appendix we have attempted to calculate the stress applied to the coating at failure during
the dry pick test. The results suggest that coating layers are at least 2 orders of magnitude
stronger than the stress applied during pick testing, and should not fail when they clearly do. It
is possible that the timescale of stress application in the tensile measurement is too slow.
Another possibility is that of cavitation occurring in the nip exit leading to higher splitting
forces than we calculated here. However, the established relationship between the picking
velocity and oil viscosity for any given paper (see Appendix) suggests that the forces involved
in picking are related to the Newtonian behaviour of the oil film, which would rule out
cavitation. A third reason, as detailed above, is that during printing, the presence of the oil
weakens the adhesion between binder and mineral surface within the coating. It is also possible
that picking may originate where there are flaws such as air bubbles in the coating.
5 Conclusions
This work has shown that:-
1. Coating pick resistance measured with oil-based fluids (dry pick) increases with the
average particle size of the clay. This confirms industry experience. This trend is also
seen for the in-plane and z-direction tensile strength of coating layers. There was no
clear effect of kaolin shape factor on dry pick strength unlike we observed in the tensile
strength measurements.
2. When a water film was applied before the ink, the pick strength was reduced, especially
for short delay times between damping and printing. The permeability of the coating to
water was the key factor here. Coatings based on coarse clays that had the highest dry
pick strength showed the lowest strength when pre-damped. This is probably related to
their large pore size, which is thought to lead to a high permeability of the coating to
water. The large pore size gives a low capillarity, which reduced the rate of tack build
with ink, and this lower tack force may contribute to an apparently higher dry strength.

3. Calculations of the forces associated with the dry pick test suggested that the maximum
stress applied to the coating at failure by the film split at the nip exit was at least 2
orders of magnitude lower than the z-direction strength of the coating. The much shorter
timescale of the pick test or the presence of flaws in the coating may explain why
picking is observed.
4. The imbibition of ink oils also reduces the strength of the coating. The polarity of the
oil was a key factor. The mechanism has not been established, but it is likely that the oil
can disrupt the latex – kaolin bonding to form a weak boundary layer between the latex
and mineral. It is also likely that some oils are absorbed by the latex. One implication is
that the process of ink setting and tack development is accompanied by a weakening
effect on the coating as ink oil is imbibed. Since ink oils remain within the coating
pores for some time after printing, the coating strength of freshly printed papers may be
considerably lower than anticipated. This may contribute to strength-related problems
during subsequent handling and converting operations.
In conclusion, measuring the tensile strength of dry coating layers is of value in providing a
basic understanding of how mineral particle morphology, or binder properties, influences the
material strength of the coating layer in isolation. However, in printing with offset inks, the
coating strength is only one part of the story. Other factors include the rate of tack rise of the
ink, since this determines the stress applied to the coating, and the extent of weakening of the
binder – mineral adhesion by the oil phase of the ink
The fundamental science behind the weakening effect of fluids such as oil and water is not well
understood. Our results suggest that the polarity of the fluid is an important factor. This points
to the nature of the bonding between latex and kaolin being relatively weak, and easily disrupted
by polar fluids. Establishing the mechanism would represent a significant step towards
engineering stronger coatings. More research is needed on this important subject, and colloid
probe AFM techniques, for example, may offer a promising way to study this phenomenon
fundamentally.
Acknowledgements
The authors are grateful to the board of Imerys Minerals Ltd for permission to publish
this work and to Dr A. G. Hiorns for helpful discussions.
References
1 Purfeest R.D. and Van Gilder R. F. “Tail edge picking, back trap mottle and
fountain solution interference of model latex coatings on a six-color press predicted
by laboratory tests”, Proc. TAPPI Coating Conf., (TAPPI Press, Atlanta) pp.461 –
472. 1991.
2 Gane P.A.C. and Seyler E.N. “Tack development: an analysis of ink paper
interaction in offset printing”, Proc. TAPPI Coating Conf., (TAPPI Press, Atlanta)
pp.243-260, 1994.
3 Gane P.A.C., Schoelkopf J. and Matthews G.P. “Coating imbibition rate studies
of offset inks : a novel determination of ink-on-paper viscosity and solids
concentration using the ink force-time integral”, Proc. TAPPI International Printing
and Graphic Arts Conf., pp. 71-88, TAPPI Press, Atlanta. 2000.
4 Smith D.A., Settlemeyer L.A. and McCoy J.W. “An investigation of coated paper
wet pick using scanning electron microscopy and the IGT Printability Tester”, Proc.
TAGA Conf., pp 320-345, TAGA, Rochester, New York. 1992.
5 Reinius H., Pajari H., Tahkola K., Mikkilä J., Pőhler T., Nieminen S.,
Hermansson E. and Schulze U. “Knowledge of interaction between fountain

solution and coating provides ways to improve printed paper quality”, Proc. PTS
Streicherei Symposium, 2003.
6 Cobb R.M.K. “Coating adhesive demand – what pigment function governs it?”,
TAPPI J., 41 (10), 581-600. 1958.
7 Lepoutre P. and Hiraharu T. “On the cohesion of clay and carbonate coatings”,
J.Appl.Polym. Sci., 37, pp. 2077-2084. 1989.
8 Petterson I., Rigdahl M., Fineman I. and Engström G. “On the wet and dry
strength of coated paper”, Trans. 8th Fundamental Symposium, Cambridge, (Mech.
Eng. Publications Ltd., London) vol 2, pp. 655-671. 1985.
9 Skeppstedt A., Borg J., Mälhammar G., Engström G. and Rigdahl M. “Surface
treatment of CaCO3 with polymers in order to improve the surface strength of coated
papers”, Proc. TAPPI Coating Conf., (TAPPI Press, Atlanta) pp. 191-197. 1991.
10 Inoue M. and Lepoutre P. “Influence of structure and surface chemistry on the
cohesion of paper coatings”, J.Adhesion Sci. Technol., 6 (7) pp. 851-857. 1992.
11 Barbier C., Larsson P-L., Östlund S., Hallback N. and Karathanasis M. “On
material characterisation of paper coating materials by microindentation testing”,
J.Coating Technol. Res., 2 (6), pp. 463-471, 2005.
12 Granier V. and Sartre A. “Ordering and adhesion of latex particles on model
inorganic surfaces”, Langmuir, 11, pp. 2179-2186. 1995.
13 Husband J.C., Preston J.S., Gate L.F., Storer A. and Creaton P. “The influence
of pigment particle shape on the in-plane tensile strength properties of kaolin-based
coating layers”, TAPPI J., 5 (12), pp. 3-8. 2006.
14 Gate L.F. and Webb T.W. US Patent 5,576,617. 1996
15 Husband J.C., Preston J.S., Gate L.F., Storer A. and Creaton P. “A study of in-
plane and z-direction strength of coating layers with varying latex content”, TAPPI
J.,6 (12), pp. 10-16. 2007.
16 Ridgway C.J. and Gane P.A.C., “Correlating pore size and surface chemistry
during absorption into a dispersed calcium carbonate network structure”, Nordic
Pulp Paper Res. J., 21 (5), pp.563 – 568, 2006
17 Trivedi N.C. “Talc”, in “Pigments for Paper”, ed. Hagemeyer, R.W., TAPPI Press,
Atlanta, p. 200. 1997.
18 Rousu S. “Differential absorption of offset ink constituents on coated paper”, PhD
dissertation, Åbo Academi University. 2002.
Appendix
Calculation of the stress applied in the dry pick test
The dry pick test used in our laboratory involves the application of a film of viscous
polybutyne-based oil of thickness 8µm to the coated paper substrate. The velocity of application
increases through the action of an accelerating applicator disc [1]. The critical velocity at which
the first signs of coating removal occurs is determined visually, in our case using a low power
microscope and side illumination.
An important question is what happens at the critical velocity. We note [2] that the product of
the pick velocity and the viscosity of the oil has a constant value for a particular substrate. This
is possible because the oil displays Newtonian viscosity behaviour [2]. The existence of the
relationship between pick velocity and viscosity implies that there must be a critical
(extensional) stress which acts on the paper surface to cause pick.
The splitting of a liquid film at the exit of a rolling nip leads to a negative pressure being
generated. As speed increases, the negative pressure also increases until it is large enough to
cause failure at the coating surface. The assumption is made that the stress is acting vertically to

the sample surface and therefore only the z-direction strength of the sample is important. There
may be an in-plane force if the negative pressure is generated a significant distance from the
centre of the nip, but we believe this is negligible due to the small size of the roll involved. The
other potential in-plane force may be generated by the deformation from the roll surface, which
will lead to an increase in the contact area in the nip and consequent stretching of the paper.
Examination of the impression of the inked disc left on the paper strip after pick measurement
allows us to calculate the longitudinal stretching of the coating. Results show that this is no
more than 0.05 % in strain, and does not normally lead to coating failure.
An experimental investigation by Chopra and Tawashi [3] provides a useful set of data on the
splitting force of films of viscous polymer solutions (lecithin) at short time scales. We have used
these data to calculate the approximate forces involved in the wet pick test at failure, as follows:
The classical equation governing the force needed to separate two discs of radius a separated by
a distance h when immersed in a liquid of viscosity η is as follows (equation 1):
(
F .t = 3 ηa 2 1 / h13 − 1 / h23
4
) (1)
where F is the force needed to separate the plates by the distance h1 – h2 in time t.
Note that the force increases for diminishing film thickness and that this parameter has a third
power dependence. This indicates the importance of film thickness and accurate metering of the
volume of oil applied during the dry pick test. Equation (1) assumes that the liquid is Newtonian
and that the separation rate is below a critical velocity where cavitation may begin. The
Newtonian relationship between oil viscosity and picking speed suggests that this latter
condition is adhered to.
Chopra and Tawashi measured the force at increasing rates of separation between two probes
containing liquid of known viscosity values. They plot their data in the form of stress curves as
a function of time for increasing separation rates. We used the data for their highest viscosity
solution (6.2 Pa.s) to calculate the stress generated by the IGT pick oil as follows:
Wheel (64mm diam)
C
D
Oil film
D C
A B Substrate
0.1 cm
Figure A1: Schematic diagram of IGT tacky oil disc.

The rate of stretching of the oil film at a pick velocity of 100 cm s-1 was first calculated
assuming the nip dimensions shown in Figure A1. In the time taken for the oil film to move
from the nip zone at point A to the point B, the oil film will be stretched by a vertical distance
CB. Suppose it splits at this point. Using trigonometry it can be calculated that the oil film is
stretched by 0.00156 cm. At a pick velocity of 100 cm s-1, the time taken to travel from A to B
is 0.001 s. Hence the rate of stretching is 1.56 cm s-1.
The fastest separation rate obtained by Chopra and Tawashi is 0.166 cm s-1. Hence the IGT
exceeds their experimental results by a factor of 10. However, we can extrapolate the peak
stress values reported in [2] to the required velocity, which gives a peak stress of 1700 g cm-2
with reasonable accuracy (Figure A2). Next we corrected for the faster build up of stress with
the IGT, which at 100 cm s-1 takes about 0.001 s to travel from A to B. This is 2 orders of
magnitude less than Chopra and Tawaski’s time scale of 0.1 s. We note that the stress rise is
approximately linear with time, so we can estimate that in 0.001 s the stress will have risen to
1700 (0.001/0.1) = 17.0 g cm-2.
20.0
18.0 1700 g cm-2

Separation force, g cm2 x 102
16.0
14.0
12.0
10.0
8.0
6.0
IGT @ 100 cm s-1
4.0
0.001 0.01 0.1 1 10
Separation rate, cm s-1
Figure A2. Separation force plotted as a function of separation rate for a lecithin solution of 6.2 Pa.s.
Replotted from Chopra and Tawaski, ref 3.
There now remains one final correction to be made. The viscosity of the oil used was 17.5 Pa.s,
higher than Chopra and Tawaski’s polymer solution of 6.2 Pa.s. From equation (1) we see that
the force F is linearly proportional to the viscosity so we can scale up our stress value by 17.5 /
6.2 giving 50 g cm-2, or 0.005 MPa.
This result for the maximum stress applied during the dry pick test at 100 cm s-1 is about a
factor of 100 less than the z-direction tensile strength of the coating layers reported in Figure 2
(main text). However, these were not calendered and the coated strips used in the pick test were.
Consideration of the effect of calendering which compresses the coating and reduces the
porosity, leads to the conclusion that the z-direction strength after calendering would increase
further, perhaps by a factor of 2 or 3.
Appendix References
1 IGT pick and blister test description, Applied Paper Technology Inc.,
www.appliedpapertech.com,
2 The pick test according to ISO 3783, IGT Newsletter 1, August 1997,
3 Chopra, S.K., and Tawashi, R., “Tack behaviour of coating solutions II”,
J.Pharmaceutical Sci., 73, 4, 477-481 (1984).

M. S. WROLDSEN, P. NUSSBAUM, J.Y.HARDEBERG 101
A Comparison of Densitometric and Planimetric

Measurement Techniques for Newspaper Printing
Maria S. Wroldsen, Peter Nussbaum, Jon Y. Hardeberg
The Norwegian Color Research Laboratory

Faculty of Computer Science and Media Technology
Gjøvik University College
P.O.Box 191
N-2802 Gjøvik
Norway
E-mail: Maria.Wroldsen@eidsivaenergi.no
peter.nussbaum@hig.no
jon.hardeberg@hig.no
Abstract
Two types of measurement technologies are used for process control in newspaper
printing, namely densitometric and planimetric technologies. Densitometric
measurements are done with densitometers or spectrophotometers, while planimetric
measurements are typically done with CCD image sensor-based instruments called dot
meters. Although these two technologies are fundamentally different, they are often used
interchangeably in print calibration and process control. In this paper we investigate the
statistical relationship between densitometric and planimetric measurements on
newspaper print.
The aim of our project was to investigate whether it was possible to estimate halftone
values measured by a densitometer, from the halftone values measured by different dot
meters. The applied model is based on regression analysis using second order
polynomials. The results are given as estimates of the polynomial parameters, i.e. the
polynomials give the relation between halftone measurements with one of the dot meters
and halftone measurements with the densitometer.
Our statistical analysis showed that due to the large uncertainty of the estimated
parameters, the model does not accurately describe the relationship between the two
measurement technologies. This can be explained in part by the poor repeatability
performance for dot meters applied to newspaper print. Moreover the measurement
results also have shown significant variations within the three dot meters used in this
experiment. Factors affecting the repeatability and determining the performance of the
model are considered and discussed in this work.
Keywords: densitometer, dot meter, newsprint, process control.
1 Introduction
In newspaper printing essentially two types of measurement technologies are used for process
control, namely densitometric and planimetric measurements. In densitometric measurements,
the optical density is measured, and if needed converted to halftone values, typically using the
Murray-Davies equation. In planimetric measurements, it is attempted to directly measure the
halftone values, that is, the dot area coverage, typically using devices containing a CCD
imaging sensor; such devices are often called dot meters. Although these two technologies are
fundamentally different, they are often used interchangeably in print calibration and process
control, in particular in the Norwegian newspaper industry [1,2].

102 A COMPARISON OF DENSITOMETRIC AND PLANIMETRIC ....
This motivated us to investigate whether there is a statistically significant relationship between

halftone measurements on newspaper print done with densitometers (converted into halftone
value with the Murray-Davies-equation) and halftone measurements done with dot meters, The
objective of this study is thus to find out whether it is possible to convert planimetric halftone
measurements into densitometric halftone measurements and vice versa. Since these
technologies are used interchangeably, it is important to know how to convert from planimetric
measurements into densitometric measurements, to keep the printing process under control and
to achieve high print quality.
This study is limited to newspaper printed in coldset offset lithography using AM-screening.
The test-target is printed in three different printing devices with different process parameters.
Factors that possibly could have affected the final print (like printing parameters) are considered
as “noise” and will not be discussed. Our focus is to find whether there is a statistical
relationship between densitometric and planimetric measurements on newspaper print
independent of factors that possibly could have affected the printing process.
This study also includes repeatability analysis for the measuring devices; three dot-meters and
one densitometer. It is necessary to know the repeatability of the measuring devices, to indicate
the validity of a possibly relationship.
After giving a brief overview of the different measurement technologies in Section 2 and the
state of the art of research discussing their properties and relations in Section 3, we present our
experimental setup and preliminary tests in Section 4. In Section 5 we present and discuss our
experimental results, before concluding and proposing ideas for further research in Section 6.
2 Measurement technologies
The size of the halftone dots increase during the printing process. This is called dot gain. It is
important to know the dot gain characteristics to achieve high print quality. The dot gain is
divided into mechanical and optical dot gain. Mechanical dot gain is the result of growth during
the printing process [3]. Optical dot gain appears due to absorption and light scattering in the
ink and the paper. This makes the dots seem larger and darker than they really are. The sum of
mechanical dot gain and optical dot gain is called total dot gain.
Density is the light absorption ability of a material. The measurement of density is done with an
instrument called a densitometer and is used to control colours in the printing process. Density
is given by:
Ii
D ink = log 10 , (1)
Im
where Im is the reflected light intensity and Ii is the intensity of the incident light [4]. High
density corresponds to high absorption.
A reflection densitometer measures the amount of reflected light from a surface. It consists of a
light source to illuminate the sample, optics to focus the light, filters to define the spectral
response of the sample and a detector to monitor the reflected light. The sample is viewed at 45
degrees from the surface. The reflected light is then converted to density with a logarithmic
amplifier and displayed digitally. The densitometer sees the dot almost like the human eye and
provides an optical density value [5].
Murray [6] expressed the relationship between the reflection density of halftone prints and the
dot area R, known as the Murray-Davies equation:

1 − 10 − D R
R = × 100 %, (2)
1 − 10 − D H
where DR is the density for a sample and DH is the solid ink density.
Traditionally halftone dot area measurements have been done in laboratories with an instrument
called a planimeter [7]. This is the same as an image analyzer. In planimetric measurements, the
dot area coverage is measured by using devices containing a CCD imaging sensor. Such
devices, designed for measuring printing plates, are often called dot meters. A dot meter
combines a microscope and a CCD imaging sensor. According to [7] the major variables in a
system like this are image capture, aperture selection and thresholding. The dot meter analyzes
the digital image and decides what is a part of the dot and what is not based on a threshold. The
camera takes a snap-shot of the area being measured and literally counts the number of black
and white pixels in the image. The dot meter is actually measuring the dot area and provides an
absolute value of dot coverage [8].
The focus of the camera is an important factor. The depth of focus is typically less than 0.2 mm
for any such system [8]. The dot meter uses the image histogram and a threshold to calculate dot
area. The threshold defines how dark a pixel should be to be taken into account.
3 Literature review
In the past, several studies regarding dot meter and densitometer measurements have been done;
considering their reliability for different materials [8,9] comparisons of the two measurement
technologies [9,10] and the use of an image analysis system to measure density [11,12]. Most of
these studies deal with measurement of printing plates. However, lately densitometers and dot
meters are used interchangeably and not only for printing plates, but also on newspaper print
[1,2].
Yule and Nielsen [13] studied whether halftone values from density measurements
corresponded to real dot areas. They found that halftone value calculated from density values
with the Murray-Davies equation did not correspond to the real dot area coverage, because the
effect of the penetration of light into the paper is usually neglected. Especially for uncoated
papers the density of middle tones increases and multiple internal reflections from the paper
surface increase it still further, so that the usual simple equation relating dot area to density is
not accurate. Their general conclusion is that the relationship between dot area and halftone
density is not nearly as simple as it appears.
Arnaud [14] compared three methods of image analysis to determine the physical area of dots
on five different substrates, among these uncoated paper and printing plates. The three devices
were all based on an optical microscope. His conclusion was that the dot gain (mechanical and
optical dot gain) is a parameter at least as important as the solid ink density in process control.
Arnaud states that an image analysis software needs to be created specifically for the printing
industry and that this software should be able to accurately measure dot area on any substrate
(including papers).
Romano [7] states that measuring halftone dot areas on printing plates with a video image
analyzer is a simple procedure, but tends to be rather subjective. It is very important to obtain a
high quality image. According to Romano, the image quality is dependent of two criteria’s; the
distance between the histogram’s peaks (contrast) and the depth of the histogram valley
(sharpness) of the 50% tint. Illumination is also an important factor. When it comes to size of
aperture (field of view), Romano states that this is a critical factor to make accurate

measurements. With small aperture size (enclosing only a few dots), errors can occur when the
aperture is randomly placed.
4 Experimental approach
As mentioned previously the aim of this study is to find out whether it is possible to convert
planimetric halftone measurements into densitometric halftone measurements. Hence, a specific
test target consisting of 16 patches (patch size 8x8mm) in different halftone values for each
process colour (CMYK), was designed (Table 1 and Figure 1). The target was printed in coldset
web-offset lithography using AM-screening, in three different printing plants (three different
Norwegian newspapers), namely Bladet Sunnhordland, NR1 Trykk, and Orkla Trykk. In the
following, the test targets are referred to with the name of the printing plant and a number
indicating the sequence number of the copy. For instance, the test-target used as training-set is
referred to as “NR1 Trykk 24000”.
Table 1: Digital halftone values (as specified in the image file) for the test target used in this
study (see Figure 1)
C100% C97% C89% C81% C73% C65% C58% C55%
C50% C45% C40% C32% C24% C16% C8% C3%
M100% M97% M89% M81% M73% M65% M58% M55%
M50% M45% M40% M32% M24% M16% M8% M3%
Y100% Y97% Y89% Y81% Y73% Y65% Y58% Y55%
Y50% Y45% Y40% Y32% Y24% Y16% Y8% Y3%
K100% K97% K89% K81% K73% K65% K58% K55%
K50% K45% K40% K32% K24% K16% K8% K3%
Figure 1: The test target
As densitometer, a GretagMacbeth Spectrolino spectrophotometer was used, under the

following setup: Physical filter: Pol, White base: Paper, Illuminant: D65, Observer angle: 2º,
Density standard: DIN NB. Three commercially available dot meters were used in this study
(brand names withheld for anonymity). Considering the aperture size and the treshold method
used to segment the digital image into ink and substrate areas the manufacturer’s default settings
have been used in this project.
Given the halftone values measured by one of the dot meters, the aim was to predict halftone
values of the densitometer. The applied prediction model is based on regression analysis using
second order polynomials. The results are given as estimates of the polynomial parameters, i.e.
the polynomials give the relation between halftone measurements with one of the dot meters and
halftone measurements with the densitometer, as follows:

2
y densitomet er = axdotmeter + bxdotmeter + c (3)
In polynomial regression, it is important to avoid over-fitting. Graphs with measurement data

indicated that the relation could be described with second order polynomials; the scatter plots
showed slowly decreasing graphs. Third order polynomials were also investigated, but the third
order terms were extremely small. Hence, the polynomials used are second order to avoid over-
fitting, for details, refer to [15].
Two limitations of the model were introduced; if the predicted densitometer value exceeds
100% or is below 0%, the value is clipped to 100% and 0%, respectively.
Empirical correlation coefficients were calculated to indicate whether a statistical relationship

exists between the measurement datasets, as follows:
∑ ( x − x )( y − y )
r = Correl ( X ,Y ) = (4)
2 2
∑ (x − x) ∑ (y − y)
Because of significant measurement differences between the process colours, it was necessary
to study each of them individually. Furthermore the measurement data were divided into two
sets; a training set to establish the model and a test set to evaluate its performance. The residuals
between the predicted and measured halftone values with the densitometer (with test set =
training-set and test set ≠ training set, respectively) were used to judge the performance of the
model.
Because of significant measurement differences between the dot meters and also between the
process colours for each densitometer-dotmeter-combination, it was necessary to study both the
instrument combinations and the process colours individually [15]. The modelling and data
analysis were therefore conducted separately for each dot meter.
The following method was used in this study to build and test the model (describing a possible
relationship between densitometric and planimetric measurements) for each combination of
instruments: First, three series of measurement data from one test target were used to establish
the model (one model for each process colour). This measurement data constitutes the training
set. Then, the residuals between predicted and measured halftone values with the densitometer
(with the test set being part of the training set and with the test set totally independent of the
training set, respectively) were used to judge the performance of the model. Some of the test
targets in the test set were printed in another printing plant than the test target of the training set.
4.1 Preliminary repeatability tests

To justify that the densitometer could be used as a reliable representative for all densitometers,
we did a preliminary test with two different densitometers. This was done to verify whether
different densitometers give the same result (in contrast with the dot meters which are based on
thresholds). For this test we used the test-target named “Bladet Sunnhordland 1500”. The
following patches were measured for each process colour: 100%, 81%, 50% and 24% (white
base: paper).
The density values were converted to halftone values using the Murray-Davies-equation, and
the densitometer pretest showed the largest deviations for halftone values below 24%. This is
probably caused by the conversion from logarithmic density values into halftone values. Low
density values converted to halftone values using the Murray-Davies equation result in larger
variations than high density values. This effect is getting even more obvious with low solid ink
densities, like in newspaper printing. Another critical factor is the number of decimals used for

density measurement. Even though the densitometer pre-test showed some deviations between
the two densitometers, only one of them is used in the analysis. This was necessary to limit the
analysis.
The repeatability analysis of the densitometer was satisfying. We measured the 50% patch 10
times. The variation was less than 0.01 density for all the process colours. The tolerance density
deviation for densitometer measurements is ±0.01 according to DIN 16536-2 [16].
A repeatability analysis was also conducted for the dot meters. On newspaper print the 50%-
patch for each process colour (CMYK) was measured 10 times with each dot meter. Based on
these measurements, we calculated the average, range (absolute value of maximum halftone
value minus minimum halftone value) and standard deviation were for the three dot meters on
newspaper print (note: not printing plates). The repeatability analysis showed low repeatability
for all three dot meters, as shown in Table 2, 3 and 4.
Table 2: Repeatability analysis – dot meter 1

Dot meter 1 Test-target: “NR1 Trykk 24000”
C50% M50% Y50% K50%
Average 49.70% 46.35% 47.50% 41.05%
Range 3.00% 3.00% 3.50% 1.00%
Standard deviation 0.92% 1.11% 1.00% 0.28%
Table 3: Repeatability analysis – dot meter 2

C50% M50% Y50% K50%
Average 48.86% 48.25% 52.53% 39.69%
Range 5.80% 6.10% 7.00% 2.80%
Table 4: Repeatability analysis – dot meter 3.

C50% M50% Y50% K50%
Average 49.15% 42.25% 47.80% 40.40%
Range 4.00% 3.50% 2.50% 2.00%
DIN 16536-2 [16] states the tolerance variation of density measurements to be ±0.01. However,
there is no standard dealing with acceptable variations for dot meter measurements. In
accordance with ISO 12647-3 [17] the optical density for CMY should be 0.9 and for black 1.1.
Outside U.S the 26% tonal value curve is used. This means that the tone value increase at 40%
or 50% should be 26%. Optical solid density 0.9 and 26% tone value curve make ±0.01
correspond to approximately 2% tone value for the middle tones (see Table 5). None of the
three dot meters fulfilled this requirement.
Table 5: Tolerance tone value for the middle tones

Density Murray-Davies
DH DR Halftone value
0.90 0.90 100.0%
0.48 76.52%
0.47 75.64%
0.46 74.73%

According to the presented results, it is not possible to decide whether this low repeatability is
caused by the measuring devices and/or inhomogeneous halftone values within one patch. Print
irregularities cause noticeable differences in measured halftone values and reduce the
repeatability when the aperture is small. Large screen dots used in newspapers in combination
with small aperture is therefore unfavourable. It is not unambiguous to decide what is
substratum and what is part of a screen dot, especially for middle halftone values, due to high
optical dot gain (see Figure 2).
Figure 2: Image of a 50% halftone value printed on newspaper with magenta ink taken with a
Zeiss Axioplan 2 imaging microscope (© Maria S. Wroldsen)
As mentioned previously the dot meter analyzes the digital image and according to a certain
defined threshold operation the colour image will be converted into a high-contrast, black-and-
white image. All pixels lighter than the defined threshold are converted to white; all pixels
darker are converted to black. Figure 3 illustrates variations in terms of defining an appropriate
threshold to determine the size of the dot area. Taking a porous and scattering substrate such as
newsprint into consideration the main difficulty is to determine an ideal threshold.
Figure 3: Threshold variations and the corresponding % dot area. Image A with threshold 100
results in 25% dot area, Image B with threshold 128 results in 42% dot area and Image C with
threshold 150 results in 52% dot area
5 Experimental Results and Discussion
The calculation of empirical correlation between the densitometer and dot meter values
indicated relationship (see Table 6). Due to the fact that the correlation coefficients were close
to 1, it can be assumed that there must be a correlation between the halftone values measured by
the various instruments.

Table 6: Empirical correlation coefficients

Dot meter 1 Dot meter 2 Dot meter 3
Densitometer, C 0.984 0.986 0.979
Densitometer, M 0.991 0.984 0.982
Densitometer, Y 0.988 0.997 0.991
Densitometer, K 0.986 0.983 0.985
Based on three measurement series of one test target, second order polynomials estimating the
relationship between halftone measurements with the dot meters and the corresponding halftone
measurement with the densitometer were established.
Figures 4-7 show measurement data with belonging trend lines for the three measurement series
with dot meter 1, of the test target “NR1 Trykk 24000”; this constitutes the training set for our
model. These graphs show that the trend lines highly fit the measurement data. To test this
model, we calculated residuals between predicted and measured halftone value.
Cyan
100,00
90,00
Halftone value in %, (densitometer)
y = -0,006x 2 + 1,6953x - 4,7409

80,00
R2 = 0,9937
70,00
60,00
50,00
40,00
30,00
20,00
10,00
0,00
0,00 10,00 20,00 30,00 40,00 50,00 60,00 70,00 80,00 90,00 100,00
Halftone value in % (dot meter 1)
Figure 4: Relation for cyan (“NR1 Trykk 24000”, dot meter 1)

Magenta
100,00
90,00
y = -0,0042x 2 + 1,4406x + 4,4598

80,00
R2 = 0,9953
70,00
60,00
50,00
40,00
30,00
20,00
10,00
0,00
0,00 10,00 20,00 30,00 40,00 50,00 60,00 70,00 80,00 90,00 100,00
Figure 5: Relation for magenta (“NR1 Trykk 24000”, dot meter 1)
Yellow
100,00
90,00
y = -0,0049x 2 + 1,468x + 3,606

80,00
R2 = 0,9965
70,00
60,00
50,00
40,00
30,00
20,00
10,00
0,00
0,00 10,00 20,00 30,00 40,00 50,00 60,00 70,00 80,00 90,00 100,00
Figure 6: Relation for yellow (“NR1 Trykk 24000”, dot meter 1)

Black
100,00
90,00
y = -0,0052x 2 + 1,5301x + 1,5519

80,00
R2 = 0,9967
70,00
60,00
50,00
40,00
30,00
20,00
10,00
0,00
0,00 10,00 20,00 30,00 40,00 50,00 60,00 70,00 80,00 90,00 100,00
Figure 7: Relation for black (“NR1 Trykk 24000”, dot meter 1)
The obtained regression polynomials for dot meter 1 are given in Equations 5-8.
2
Cyan : ydensitometer = −0.006 xdotmeter1 + 1.6953 xdotmeter1 − 4.7409 (5)
2
Magenta : ydensitometer = −0.0042 xdotmeter 1 + 1.4406 xdotmeter1 + 4.4598 (6)
2
Yellow : ydensitometer = −0.0049 xdotmeter1 + 1.468 xdotmeter1 + 3.606 (7)
2
Black : ydensitometer = −0.0052 xdotmeter1 + 1.5301xdotmeter1 + 1.5519 (8)
The residual between predicted and measured halftone value with the densitometer was
calculated as in the following example with cyan 65%:
• Measured halftone value with dot meter 1: 59.50%

• Measured halftone value with the densitometer: 78.12%
• Predicted halftone value with the densitometer: 74.89% (Equation 5)
• The residual between measured and predicted halftone value with the densitometer: 3.23%
Table 7 shows the residuals when the test-set is part of the training-set. Because the test set is
part of the training set, it is expected that the differences between the predicted and measured
halftone values are rather small. As it can be seen in Table 7 the variations are colour and
halftone value independent and does not follow a certain trend although cyan shows the largest
variations. However, there is no significant trend for the obtained variations. Although the
model performs well it is important to test the model with another test-set.
Table 8 shows the residuals when the test-set is not part of the training-set. It can be seen that
the model does not perform that well applying another test-set. Consequently the differences
between the predicted and measured halftone values are larger. Although the residuals have
increased, the variation still does not follow a certain trend.

Table 7: Dot meter 1, densitometer: Residuals in % when the test set is part of the training set.
(Test set: “NR1 Trykk 24000”, 1. measurement series)
Residuals
Patch Cyan Magenta Yellow Black
100% 0.00 0.00 0.10 0.00
97% 0.13 0.14 0.14 1.54
89% 1.20 1.13 0.85 0.58
81% 3.63 1.61 1.50 0.58
73% 3.23 0.92 0.97 0.04
65% 3.23 2.62 0.87 0.68
58% 0.79 1.33 1.53 0.99
55% 1.60 1.68 3.59 3.73
50% 1.47 0.15 1.49 0.35
45% 0.95 1.86 0.84 1.30
40% 1.19 2.01 1.18 0.37
32% 2.09 0.11 4.46 1.20
24% 0.78 1.00 0.97 2.66
16% 0.98 1.36 1.51 0.83
8% 0.55 3.77 0.53 0.82
3% 4.25 4.87 1.52 1.09
Table 8: Dot meter 1, densitometer: Residuals in % when the test set is different from the
training set. (Test set: “Orkla Trykk 5000”)
Residuals
Patch Cyan Magenta Yellow Black
100% 0.00 0.00 0.00 0.00
97% 2.15 1.59 1.66 0.94
89% 2.12 3.24 1.86 3.16
81% 3.25 5.83 2.15 1.46
73% 4.38 5.57 3.78 3.71
65% 5.04 0.40 3.82 4.99
58% 4.82 1.46 4.17 6.04
55% 3.65 0.78 4.07 4.85
50% 1.06 2.66 2.64 1.0
45% 0.16 1.48 4.51 1.75
40% 1.45 1.64 3.99 2.43
32% 2.61 5.86 1.69 1.54
24% 4.55 5.52 3.43 3.07
16% 2.60 2.19 0.59 1.01
8% 0.17 4.09 0.37 0.72
3% 4.91 1.80 4.66 0.48
Residuals between predicted and measured densitometer values were calculated for several test
sets different from the training set. Some of these test-sets where printed in different printing
processes than the training-set.
As stated earlier, 2% can be considered the tolerance variation for dot meters. Some residuals
are larger than 2% and this indicates that the model is not good enough to describe a relation.
The low repeatability is an unfavourable factor.

We have so far presented the detailed results from the first dot meter only. In the following we
summarize briefly the results for the two other dot meters, for more detailed information refer to
[15].
For dot meter 2 the obtained regression polynomials are shown in Equations 9-12. The residuals
are larger when test-set is different from the training-set (see Table 10) than for test-set part of
training-set (see Table 9) and indicates that our second order polynomials do not satisfactory
describe a possibly relationship between halftone measurements with dot meter 2 and the
densitometer. This is partly caused by the low repeatability for dot meter 2.
2
Cyan : y densitometer = −0.0061x dotmeter 2 + 1.6114 x dotmeter 2 − 2.0895 (9)
2
Magenta : y densitometer = −0.0051x dotmeter 2 + 1.5363 x dotmeter 2 + 1.092 (10)
2
Yellow : y densitometer = −0.0018 x dotmeter 2 + 1.0991x dotmeter 2 + 7.1581 (11)
2
Black : y densitometer = −0.0063 x dotmeter 2 + 1.6296 x dotmeter 2 + 0.3169 (12)
The polynomials for dot meter 3 is are given in Equations 13-16, and the residuals are given in
Table 9 and 10. We see the same trend as for the two other combinations of instruments. The
average residuals are larger when test-set is not part of training-set. As for the two other
combinations, the low repeatability of the dot meter is one factor that makes our model
unsatisfactory.
2
Cyan : y densitometer = −0.0049 x dotmeter 3 + 1.3198 x dotmeter 3 + 13.486 (13)
2
Magenta : y densitometer = −0.0044 x dotmeter 3 + 1.3316 x dotmeter 3 + 10.239 (14)
2
Yellow : y densitometer = −0.0043 x dotmeter 3 + 1.2947 x dotmeter 3 + 11.319 (15)
2
Black : y densitometer = −0.0047 x dotmeter 3 + 1.4000 x dotmeter 3 + 5.5886 (16)
Table 9: Residuals in % between dot meter and densitometer when the test set is part of the
training set. (Test set: “NR1 Trykk 24000”, 1. measurement series)
Residuals
Cyan Magenta Yellow Black
Dot Max 4.25 4.87 4.46 3.73
meter 1 Average 1.63 1.54 1.38 1.05
Dot Max 5.88 4.11 3.73 4.80
meter 2 Average 2.27 1.29 1.54 1.32
Dot Max 8.50 4.56 3.74 4.96
meter 3 Average 1.89 2.19 1.88 1.79
Table 10: Residuals in % between dot meter and densitometer when the test set is different from
the training set. (Test-set: “Orkla Trykk 5000”)
Residuals
Cyan Magenta Yellow Black
Dot Max 5.04 5.86 4.66 6.04
meter 1 Average 2.68 2.76 2.71 2.32
Dot Max 4.37 8.39 6.87 6.87
meter 2 Average 2.09 3.65 2.63 2.35
Dot Max 8.94 6.31 10.02 6.17
meter 3 Average 2.93 3.39 4.87 2.44

6 Conclusions and Perspectives
Our statistical analysis showed that due to large uncertainty of the estimated parameters, the
model does not accurately describe the relation between the two measurement technologies.
This can be explained by the poor repeatability performance for dot meters applied in
newspaper print. The repeatability analysis provided low confidence using dot meters in
newspaper print. None of the three dot meters fulfilled the requirement of 2% tolerance
deviation (note: these are requirements which are not defined in an official standard). Dot
meters are originally developed for measuring printing plates only.
The residuals between predicted and measured half tone values with the densitometer increased
when the test set was different from the training set, as would be expected. Moreover, the
measurement results have shown significant variations within the three dot meters. Some factors
affecting the repeatability and determining the performance of the model are listed in this
section.
Important factors that impair the use of dot meters on newspaper print:
• Print irregularities cause noticeable differences in measured halftone values and reduce the
repeatability when the aperture is small.
• Small aperture in combination with large halftone dots used in newspapers are
unfavourable.
• Due to high optical dot gain (especially for the middle tones and in newspaper print) it is
ambiguous to decide what is substratum and what is part of a screen dot.
• Large residuals between predicted and measured halftone values for the middle tones could
partly be explained by the high optical dot gain and problems due to determination of
threshold (what is substratum and what is not) in the image analysis.
Important factors that impair the use of densitometers on newspaper print (when using the
Murray-Davies equation to calculate halftone values):
• The conversion from logarithmic density values into halftone values with the Murray-
Davies equation causes a slowly decreasing graph that makes low density values converted
to halftone values result in larger variations than high density values.
• The effect of this conversion is even more obvious when used with low solid ink densities
like in newspapers.
Based on these results, dot meters are not recommended for halftone measurements on paper
substrates in newspaper printing.
Throughout this project some ideas of further research to investigate a possibly relation between
densitometric and planimetric measurement emerged.
The test target was printed in different printing processes with different solid ink densities, even
though the instructions for printing said K 1.10 and CMY 0.9 in accordance with [17]. It is
difficult to control this in newspaper printing. It would have been interesting to do the same
experiment with a print medium where accurate solid ink densities are possible. The uncertainty
of dot meters and densitometers for use in newspaper printing is too high. Another type of paper
(with lower optical gain) would also be preferable. To increase the repeatability it is
advantageous with coated paper, accurate solid ink density and finer screen ruling (the aperture
size would not be so critical). Moreover, FM-screening could be used. The reason why we did
this experiment with newspaper in the first place is the increasing use of dot meters in the
newspaper industry.

More than one copy of each instrument could have been included in the repeatability analysis to
make any variations between copies become visible.
Image analysis of halftone images would be interesting to investigate the decision of threshold
(what is part of a screen dot and what is not) and to illustrate the percent of optical dot gain for
different halftone values. Different thresholds could be set and the result (dot area coverage)
could be compared to measured values for the different dot meters. Different size of aperture
could also be simulated to observe the influence of calculated dot area coverage. This
experiment could perhaps lead to a recommendation of optimal size of aperture for different
screen rulings; what size of aperture is necessary to avoid systematical errors?
Acknowledgments
First, we would like to thank senior lecturer Sven Erik Skarsbø for providing supervision
to this master thesis project (in collaboration with Professor Jon Y. Hardeberg and
lecturer Peter Nussbaum). Thanks for your interest, motivation and help throughout this
project! Secondly, we would like to thank associate professor Are Strandlie for valuable
help with the statistical analysis. Finally, we would like to thank Mediebedriftenes
Landsforening for financial support and the newspapers (Bladet Sunnhordland,
Romerikes Blad and Drammens Tidende) for offering to print the test targets.
References
1 Aasen E., Danielsen Ø. and Bovolden A.J., Halftone measurements in newspaper
print, Bachelor thesis, Gjøvik University College, Gjøvik, Norway, In Norwegian,
2002
2 NADA (Visited 2007), Support – Lineær avisproduksjon – kort og godt, Online:
http://www.nada.no/support/linprod/kortoggodt.html. In Norwegian. (NADA is a
company working with digital ad delivery for Norwegian newspapers, see also
http://www.nada.no/omnada/english.html)
3 Malmqvist K., Verikas A. and Bergman L., Consistency of mechanical dot gain -
a hidden quality parameter, TAGA 51st Annual Technical Conference, pp 409-417,
1999
4 Bergman L., Using Multicoloured Halftone Screens for Offset Print Quality
Monitoring, Licentiate Thesis No. 1147, Linköping University, 2005
5 Tobias Associates, Inc. (Visited 2007), Reflection Densitometry, Online:
http://ourworld.compuserve.com/homepages/tobiasinc/TAPages/bulletin3.htm
6 Murray A., Monochrome Reproduction in Photoengraving, J. Franklin Institute,
vol. 221, pp.721-744, 1936
7 Romano D., The Image Analyzer – A True Dot Area Meter? In TAGA Proceedings,
pp.318-334, 1996
8 Colthorpe S. and Imhoff G., CTP – Why densitometers do not work, White paper,
Centurfax Ltd and Grip Digital Inc, 1999
9 Hsieh Y., Wu Y. and Lin W., An Experimental Research to Compare Devices for
Measuring Aluminium Lithographic Printing Plates, In TAGA Proceedings, pp.250-
273, 2003
10 Spotts T. and Desai N., A comparison: The measurement of dot area by
densitometry and physical measurement methods, Visual Communications Journal,
pp.19-24, 2005
11 Malmqvist K., Bergman L., Buck H. and Malmqvist L., The 3-colour CCD
camera as a densitometer for measuring density of cyan, magenta and yellow in
printed solid areas and in screen areas, In Advances in Printing Science and
Technology vol. 22, 342-351. IARIGAI, 1993

12 Brydges D., Deppner F., Kuenzli H., Heuberger K. and Hersch R., Application
of a 3-CCD color camera for colorimetric and densitometric measurements, In Color
Imaging: Device-Independent Color, Color Hardcopy, and Graphic Arts III,
Proceedings of SPIE, vol. 3300, 292-301, 1998
13 Yule J.A.C. and Nielsen W.J., The penetration of light into paper and its effect on
halftone reproduction, TAGA Proceedings, pp.65-76, 1951
14 Arnaud S., Measurement of dot area, In TAGA Proceedings, pp.685-702, 2001
15 Wroldsen M.S., Densitometriske og planimetriske målinger av rasterstrukturer,
Master’s thesis, Gjøvik University College, Gjøvik, Norway. In Norwegian, 2006
16 DIN 16536-2., Prüfung von Drucken und Druckfarben der Drucktechnik –
Farbdichtemessung an Drucken. Teil 2: Anforderungen an die Messanordnung von
Farbdichtemessgeräten und ihre Prüfung, Deutsches Institut für Normung, 1995
17 ISO 12647-3., Graphic technology – Process control for the manufacture of half-
tone colour separations, proofs and production prints. Part 3: Coldset offset
lithography on newsprint, International Organization for Standardization, 2005

R. XU, Y.J. WU, A. PEKAROVICOVA, P.D. FLEMING, M. X. WANG 117
The Effects of Paper Coating on Gravure Ink Mileage Curve

Renmei Xu1, Yu Ju Wu2, Alexandra Pekarovicova2, Paul D. Fleming2, and Michelle X. Wang3
1 2
Ball State University Western Michigan University
Department of Technology
Muncie
IN 47304 USA
3
Armstrong World Industries
Abstract
An ink mileage curve is a plot of the printed optical density of an ink on a substrate as a
function of ink film thickness. It is helpful to predict how much ink is needed to achieve
target density and has been studied for many years. It was found that it was affected by
both paper properties and ink properties. However, most of the previous research was
done for offset lithographic printing and little was done for gravure printing, due to the
difficulty of measuring gravure ink mileage. In our previous studies, ink mileage was
measured using two methods based on the same analytical principle: the tracer method by
adding a tracer into the ink, and the direct method utilizing metal ions that already exist in
the ink. In this study, both of the two methods were used. The substrates used were pilot
coated papers using different coating formulations. The coated papers were printed on a
pilot rotogravure web press. The concentrations of copper ions in both cyan liquid ink and
ink film were analyzed by Atomic Absorption Spectrometry (AAS) and used to calculate
the amount of ink transferred to the paper. The ink was also doped with a metal ion tracer,
and the amount of ink transferred was obtained in the same way. It was found that the
transferred amounts of cyan pigments were higher than those of the tracer, thus the direct
method is slightly more reliable. The analysis of the obtained ink film coat weight and
reflection density data showed that both the Oittinen and Calabro-Savagnone model fitted
the experimental data well, with the latter a little better. The relationships between
regression coefficients and tested paper properties were studied. Saturation density Ds
correlated with all of the tested paper properties, but not strongly. Higher roughness, air
permeability, pore size, or porosity resulted in higher saturation density. Density
smoothness m was found to correlate very well with roughness and air permeability.
Higher roughness and air permeability resulted in lower m, thus lower density. No
significant correlations were found between the ink film coat weight exponent n and
tested paper properties.
Keywords: Ink Mileage Curve, Gravure, Ink Film Coat Weight, Coating.
1 Introduction
An ink mileage curve is a plot of the printed optical density of an ink on a substrate as a
function of ink film thickness. The optical density of a print increases from zero to a saturation
value with increasing ink layer thickness on the paper. The saturation density comes from first
surface reflection [1]. In practice, ink film thickness is difficult to measure, so ink film coat
weight, the amount of ink in gram per square meter of substrate or gsm, has been used to
represent the ink film thickness by assuming the ink film density is the same for the same ink.
An ink mileage curve is helpful to predict how much ink is needed to achieve target density and
has been studied for many years [2-9]. In order to study the ink mileage curve quantitatively,
several models for curve fitting have been reported by different researchers. It was found that
the ink mileage curve was affected by both paper properties and ink properties.

118 THE EFFECTS OF PAPER COATING ON GRAVURE INK MILEAGE CURVE
However, most of the previous research was done for offset lithographic printing and little was
done for gravure printing, due to the difficulty of measuring gravure ink mileage. The previous
studies were based on the experimental data of prints made on IGT and/or Prüfbau printability
testers using offset lithographic inks. The quantity of transferred ink, and hence the amount of
ink on paper, was determined by the weight difference of the printing disc before and after
printing. This method can not be applied to gravure inks, because the solids content in solvent-
based gravure ink was about 30%, thus the weight of the ink film compared to variation of
substrate grammage was too small to achieve reliable results. An analytical method was used in
our previous studies. In an earlier study, the inks were doped with selected tracers that did not
affect the ink color and viscosity. The concentrations of the tracer element in both the liquid ink
and the ink film were analyzed using Atomic Absorption Spectrometry (AAS) to calculate the
mass of the ink transfer, and hence the ink film coat weight [10, 11]. This tracer method can be
used in all kinds of ink types including solvent-based gravure and flexographic inks. It can also
be applied to commercial printing presses. However, adding a tracer into an ink is not
convenient and the effect of the tracer on ink performance was minimal, but still existed in
reality. Therefore, a later study analyzed metal ions in ink pigments, more precisely, copper ions
in cyan ink, for ink transfer calculation, instead of a tracer added into the ink [12]. Copper is
part of the cyan pigment, so it does not interfere with ink viscosity, or ink color. This direct
method seems more convenient and accurate.
Our previous studies found that equation 1 and 2, which were modified from two of previous
models [4,6], fitted the data well, with equation 2 having a little smaller sum of the squares of
the residuals.
D = Ds (1 − e − mw )
n
Oittinen (1)
1 1 1 n
Calabro and Savagnone = +( ) (2)
D Ds mw
Where w: ink film coat weight
D: relative reflection density
Ds, m, n: regression coefficients
Ds is called “saturation density”, the reflection density of an ink film at infinite ink film
thickness. Parameter m determines the steepness of density curve in the region of very thin film
and is called “density smoothness” [2]. The higher the density smoothness is, the higher the
reflection density is. It was found that saturation density Ds had higher correlations with air
permeability and pore size than surface roughness. Parameter m correlated well with paper
roughness measured by the profilometer, and at a less significant level, with air permeability.
Parameter n correlated well with both roughness measured by the profilometer and air
permeability [11, 12].
In this work, papers coated using formulations with different latex binders were printed on a
rotogravure press. Both the tracer and direct method were applied and compared. The effects of
paper coating properties on gravure ink mileage curve were studied.
2 Experimental
2.1 Paper Coating

A series of coating trials were run with a 32# base paper. Coating formulations are shown in
Table 1. The ratio between delaminated clay and fine clay was 80:20, which was the same for
all the samples. The latex binder type used was different but the same concentration of 6 parts
for sample 1 to 5. The coating formulation for sample 6 contained the same latex as sample 2
but at a lower level of 4.5 parts.

Table 1: Coating formulations
Sample ID 1 2 3 4 5 6
Delaminated clay, Astraplate 80 80 80 80 80 80
No. 1 clay, KCS 20 20 20 20 20 20
Latex, XL 3005 6
Latex, XL 2457 6 4.5
Latex, XL 2373 6
Latex, XL 411-16 6
Latex, XL 411-09 6
Lubricant, Dispex N40 0.1 0.1 0.1 0.1 0.1 0.1
2.2 Paper Testing

A Parker Print-Surf (PPS) Model 90 (Messmer Instrument) was used to measure roughness at
490 kPa with soft backing, according to TAPPI standards. Air permeability coefficient as
determined by PPS porosity [13] was measured at 980 kPa clamping pressure. An Electronic
Microgage Model 210 from EMVECO Inc. (now Lorentzen & Wettre USA, Inc.) with a
spherical steel stylus having a radius of 0.001 inch was also used for roughness measurement.
Average pore sizes and porosity were determined by mercury intrusion porosimetry.
Measurements were carried out using an Autopore IV 9500 (Micromeritics Instrument).
2.3 Printing
The coated papers were printed on a pilot rotogravure web press (Cerutti Model 118, Italy)
located at Western Michigan University (WMU) Printing Pilot Plant. Commercial toluene-based
coated cyan ink for rotogravure (Flint Group) was used. It was doped with a selected tracer. The
ink efflux time with Shell cup #2 was kept at 25 ± 0.5 seconds. Printing was done at 600 ft/min
with electrostatic assist (ESA) on. The cyan cylinder has compressed cells and the screen ruling
was 175 lpi. The print layout contains different tonal values from 10% to 100 %. The printing
condition was kept the same for all the paper substrates so it didn’t have an effect on the ink
mileage behaviors.
2.4 Ink Film Testing

A known amount of liquid ink samples and certain areas of unprinted paper samples and printed
paper samples with different tonal values were digested in a mixture of nitric acid and hydrogen
peroxide (1:1 volume) by slow boiling for at least 3 hours to ensure complete digestion. The
copper (Cu) and tracer concentrations were analyzed quantitatively using a Varian Atomic
Absorption (AA) Spectrometer Model AA240, which is a flame AA [14]. The ink film coat
weights at different tonal values were calculated using the equation:
c1 - c 2
w= (3)
c3
where w: ink film coat weight (gram per square meter, or gsm)
c1: element concentrations in printed paper (gsm)
c2: element concentrations in unprinted paper (gsm)
c3: element concentrations in liquid ink (wt %)
The reflection densities at different tonal values were measured with reference to the reflection
density of unprinted paper using an X-Rite 530 SpectroDensitometer.
The characteristics of the papers are reported in Table 2. Less latex content in sample 6 than that
in sample 2 resulted rougher and more porous coating surface.

Table 2: Characteristics of papers
PPS Profilometer Permeability

Grammage Pore Size Porosity
ID Roughness Roughness Coefficient
(g/m2) (nm) (%)
(micron) (micron) (micron2)
1 48.7 1.73 2.03 2.02E-5 239.2 64.98
2 48.8 1.74 2.02 2.22E-5 204.2 63.41
3 48.7 1.62 1.94 1.90E-5 208.8 62.10
4 48.0 1.69 2.04 2.27E-5 188.6 56.50
5 48.2 1.62 1.99 2.10E-5 231.6 60.87
6 48.5 1.78 2.09 2.53E-5 250.1 65.99
The measured results of relative reflection density and ink film coat weight obtained from both
direct copper and tracer method are shown in Table 3. The ink film coat weight values obtained
from direct copper method are higher than those from tracer method. In order to find out if the
high speed printing condition or ESA system affected different transfer amounts of pigment and
tracer, two samples were printed on a K-Proofer with the same cyan ink collected during the
printing trial. The printed samples were then analyzed using the same procedure. The ink film
coat weight values obtained from direct copper method are also higher than those from tracer
method. The ratios between these two values are 1.234 and 1.231 respectively for the two K-
Proofer samples, which are very close to the ratio of 1.258 for the press sample. Therefore, the
differences between the results from the direct copper and tracer method were not caused in the
printing process. The possible reason can be the analytical error. The detection limit of flame
AA instrument is 3 ppb for copper, while 5 ppb for the tracer element [14]. Therefore, the
results obtained from the direct copper method are slightly more reliable. Pigments, as part of
ink film, play important role on reflection density, and copper is part of pigments. Based on
these two reasons, the results obtained from the direct copper method were selected for further
studies.
The ink film coat weight data obtained from direct copper method and the relative reflection
density data were analyzed using appropriate OriginPro 7.5 nonlinear fitting routines. Equations
1 and 2 were examined. Ds, m, and n were treated as regression variables. One example of the
ink mileage curves is shown in Fig. 1. The solid dots are experimental data of cyan ink on paper
sample 5. The solid line is the non-linear fitting result using the Oittinen model, while the dotted
line using the Calabro-Savagnone model. These two curves were indistinguishable in our data
range and began to separate with increasing ink film coat weight. The ink mileage curve
obtained using the Calabro-Savagnone model is above that using the Oittinen model; therefore,
the saturation density obtained from the Calabro-Savagnone model is higher than that from the
Oittinen model. The hollow dots are residual values. The residual is equal to the experimental
value minus the value calculated from the fitting model. Residual values can be used to tell how
good the fit is. The smaller the absolute values of the residuals are, the better the curve fit is.

Table 3: Relative reflection density and ink film coat weight results
Relative Ink Film Ink Film

ID Tone (%) Reflection Coat Weight, gsm Coat Weight, gsm
Density (Direct Method) (Tracer Method)
25 0.28 1.34 1.01
50 0.56 2.70 1.96
70 0.84 4.56 3.46
1
80 1.01 5.84 4.64
90 1.19 7.29 6.03
100 1.30 8.28 7.02
25 0.27 1.40 1.01
50 0.55 2.76 2.02
70 0.83 4.67 3.61
2
80 1.01 5.96 4.78
90 1.19 7.45 6.05
100 1.30 8.45 7.06
25 0.28 1.35 0.95
50 0.56 2.71 1.89
70 0.83 4.36 3.26
3
80 1.01 5.69 3.79
90 1.18 6.88 5.43
100 1.30 8.06 6.20
25 0.27 1.43 1.12
50 0.55 2.85 2.30
70 0.83 4.66 3.89
4
80 1.00 5.98 5.16
90 1.18 7.57 6.53
100 1.30 8.56 7.67
25 0.27 1.40 1.05
50 0.56 2.77 2.18
70 0.83 4.55 3.71
5
80 1.01 5.85 4.86
90 1.18 7.27 6.23
100 1.30 8.32 7.24
25 0.26 1.36 1.14
50 0.54 2.71 2.16
70 0.81 4.50 3.96
6
80 0.99 5.78 5.24
90 1.16 7.20 6.70
100 1.28 8.03 7.81

2.0
1.8
1.6
Relative Reflection Density

1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Ink Film Coat Weight, gsm
Figure 1: Ink mileage curve of sample 5 (the solid line is the curve fitting using the Oittinen
model and the dotted line using the Calabro-Savagnone model)
The residual values obtained from curve fitting of all samples were plotted against ink film coat
weight. The degree of fit of an equation to the experimental data can be determined by the sum
of the squares of residuals and the distribution of residuals around zero. A small sum of the
squares of residuals and an even distribution indicate a good fit. The residual plots of the two
models are compared in Fig. 2. The results indicate that both models fitted the experimental data
well, with Calabro-Savagnone model a little better. This agrees with the finding in our previous
studies [11, 12].
0.20 0.20
0.15 Oittinen Model 0.15 Calabro-Savagnone Model
0.10 0.10
0.05
Residual (Density)
0.05
Residual (Density)
0.00 0.00
-0.05 -0.05
-0.10 -0.10
2 2
SUM (Residual ) = 0.0116 SUM (Residual ) = 0.0104
-0.15 -0.15
-0.20 -0.20
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9
Ink Film Coat Weight, gsm Ink Film Coat Weight, gsm
Figure 2: Sum of the squares of reflection density residuals and their distributions
The regression coefficients, Ds, m, n, derived from curve fitting for both models are listed in
Table 4. The values for different coated papers are close because of the same clay contents in
coating formulations, but there are still significant differences due to the effects of different
latex binder types and contents. Saturation density Ds values derived from the Calabro-
Savagnone model are higher than those from the Oittinen model, which has been shown in Fig.
1.
It is difficult to conclude which values are more reliable based on present data. It can only be
found out by applying thicker ink film which will fall in the region where the reflection density
approaches the saturation density. However, achieving reflection density close to 2 requires
relatively thick ink film and makes the gravure printing trials difficult to run.

Table 4: Regression coefficients of Oittinen and Calabro-Savagnone model
Oittinen Model Calabro-Savagnone Model

ID
Ds m n Ds m n
1 2.250 0.101 1.013 3.370 0.245 1.053
2 2.161 0.098 1.046 3.213 0.241 1.089
3 2.181 0.103 1.040 3.276 0.252 1.079
4 2.152 0.097 1.047 3.174 0.239 1.093
5 2.077 0.102 1.065 3.032 0.249 1.116
6 2.374 0.089 1.031 3.608 0.233 1.067
The correlations between paper characteristics and regression coefficients for both models are
shown in Table 5. Saturation density Ds correlated with all of the tested paper properties, but not
strongly. Saturation density results from first-surface reflection, which is affected by the
smoothness of the ink film surface. Surface roughness of an ink film is related to the ink’s
leveling, which is mainly determined by the ink’s properties, but also is affected by the paper
properties. Therefore, the correlations are not very strong, but still exist. Higher roughness, air
permeability, pore size, or porosity resulted in higher saturation density. Picollet et.al. [15]
studied gravure ink penetration and spreading on LWC coated papers. They found that rough
surface and pores in coating layer facilitated ink spreading and penetration, respectively, and
thus improved ink leveling.
Table 5: Correlation matrix of regression coefficients of Oittinen and Calabro-Savagnone

model
Oittinen Model Calabro-Savagnone Model

Ds m n Ds m n
PPS
0.77 -0.81 -0.59 0.74 -0.86 -0.61
Roughness
Profilometer
0.64 -0.87 -0.36 0.58 -0.94 -0.35
Roughness
Permeability
0.55 -0.96 0.01 0.50 -0.95 -0.04
Coefficient
Pore Size 0.59 -0.34 -0.38 0.57 -0.17 -0.40
Porosity 0.69 -0.32 -0.58 0.70 -0.18 -0.63
Parameter m, also called density smoothness, determines the steepness of density curve in the
region of very thin film. It was found to have strong negative relationships with both roughness
and air permeability. Higher roughness resulted in lower m because the ink film was too thin to
cover the rough surface, thus the corresponding reflection density was low. Higher air
permeability also resulted in lower m because the solvent was absorbed so fast that the ink did
not have time to level, thus the reflection density was also low. Tollenaar and Ernst [2] reported
that m correlates with the degree of contact between the ink film and the paper. Calabro and
Savagnone [6] found that m was correlated mainly with the smoothness, gloss and absorption
values. The results obtained in this study confirmed previous findings.
The ink film coat weight exponent n determines how fast the density approaches the saturation
density Ds. No significant correlations were found between parameter n and tested paper
properties. It was more likely affected by the ink properties. Calabro and Savagnone [6] found

that parameter n had a better correlation with ink rheological variables, such as viscosity, yield
value and tack, and its optical properties, such as the absorption coefficient and the fineness of
grind of the pigment.
In summary, for a paper with higher roughness and higher air permeability, the reflective
density will increase slower with increasing ink film coat weight in the beginning, but will
approach a higher saturation density. Therefore, a paper with low roughness and air
permeability will get high ink mileage in the low density range. However, if the target density is
high, ink mileage curve is needed to find a paper with optimal roughness and air permeability.
4 Conclusions
Two methods based on the same analytical principle were tested: the tracer method by adding a
tracer into the ink, and the direct method utilizing metal ions that already exist in the ink. It was
found that the transferred amounts of cyan pigment were higher than those of the tracer, thus the
direct method is slightly more reliable. The analysis of the obtained ink film coat weight and
reflection density data showed that both the Oittinen and Calabro-Savagnone model fitted the
experimental data well, with the latter a little better, as evidenced by smaller sum of the squares
of residuals. The saturation densities obtained from the Calabro-Savagnone model were higher
than those from the Oittinen model, which needs further investigation by applying very thick
ink film. The relationships between regression coefficients and tested paper properties were
studied. Saturation density Ds correlated with all of the tested paper properties, but not strongly.
Higher roughness, air permeability, pore size, or porosity resulted in higher saturation density.
Density smoothness m was found to correlate very well with roughness and air permeability.
Higher roughness or air permeability resulted in lower m, thus lower density. No significant
correlations were found between the ink film coat weight exponent n and tested paper
properties. In conclusion, a paper with low roughness and air permeability will get high ink
mileage in the low density range. However, if the target density is high, ink mileage curve is
needed to find a paper with optimal roughness and air permeability.
Acknowledgements
The authors want to thank Mr. Mike Glenn of WMU Printing Pilot Plant for help with the
printing trial. Financial support from OMNOVA Solutions, Inc. and International Paper is
gratefully acknowledged.
References
1. Yule J.A.C., Principles of Color Production, John Wiley & Sons, New York, pp. 151-
158, 1967
2. Tollenaar D. and Ernst P.A.H., Optical Density and Ink Layer Thickness,
Proceedings of International Conference of Printing Research Institute (IARIGAI),
pp. 214-234, 1962
3. Kornerup A., Fink-Jensen P. and Rosted C.O., Tristimulus Values of Prints and
Mileage of Printing Inks, Die Farbe, Vol. 18, pp. 29-64, 1969
4. Oittinen P., Graphic Arts in Finland, Vol. 1, No. 2, pp. 11-22, 1972
5. Calabro G. and Mercatucci F., Method for Evaluating Newsprint Printability,
Proceedings of 12th International Conference of Printing Research Institute
(IARIGAI), pp. 155-159, 1974
6. Calabro G. and Savagnone F., Method for Evaluating Printability, Adv. Printing
Sci. Technol. 17, Chap. 23, pp. 358-380, 1983
7. Blom B.E. and Conner T.J., Optical Density and Ink Film Thickness; A
Comparison of Models, TAGA Proceedings, pp. 213-225, 1990
8. Chou S.M. and Harbin N., Relationship between Ink Mileage and Ink Transfer,
TAGA Proceedings, pp. 405-432, 1991

9. MacPhee J. and Lind J.T., Insight into the Relationship between Print Density and
Ink Film Thickness, TAGA Proceedings, pp. 479-496, 2002
10. Xu R., Pekarovicova A., Fleming P.D. and Bliznyuk V., Physical Properties of
LWC Papers and Gravure Ink Mileage, TAGA Proceedings, pp. 485-497, 2005
11. Xu R., Pekarovicova A., Fleming P.D. and Wang X., Study of Ink Mileage Curve
of Gravure Printing, TAGA Proceedings, pp. 443-452, 2006
12. Xu R., Pekarovicova A., Fleming P.D., Wu Y.J. and Wang M. X., A Method for
Evaluating Ink Mileage in Gravure Printing, TAPPI Journal Vol. 6, No. 12, pp 27-
32. 2007
13. Pal L., Joyce M.K. and Fleming P.D., A Simple Method for Calculation of the
Permeability Coefficient of Porous Media, TAPPI J., Vol. 5, No. 9, pp. 10-16, 2006
14. Tyler G., AA or ICP - Which Do You Choose?, published on the website of
Varian, Inc., https://www.varianinc.com/media/sci/apps/icpes003.pdf, 1991
15. Picollet M., Morin V., Piette P. and Le Nest J.F., Competition Between Gravure
Ink Penetration and Spreading on LWC Coated Papers, Proceedings of the 1998
Coating/Papermakers Conference, TAPPI Press, Atlanta, pp. 383-392, 1998

126 THE EFFECTS OF PAPER COATING ON GRAVURE INK MILEAGE
From Multi Channel Publishing towards a Ubiquitous Media

Environment
Maria Åkesson* and Carina Ihlström Eriksson, PhD**
* Halmstad University ** Halmstad University

Sweden Sweden
maria.akesson@hh.se carina.ihlstrom_eriksson@hh.se
Abstract:
As most of today’s media houses publish in multiple channels, they also exist in a
multitude of infrastructures. Given the ongoing diffusion of personal computers, handheld
devices, and mobile telephones as well as the advent of new technologies such as the e-
paper, a new type of ubiquitous media environment is emerging. Drawing on this
background we address the research question what are the challenges when aligning a
new promising technology (such as the e-paper) with existing value networks enabling
value creation in ubiquitous media environments? In this paper we report from a study
with newspaper publishers in Sweden, Belgium, France and the Netherlands consisting of
18 interviews and 9 workshops. The findings show that value networks in a multiple
channel publishing environment are closely related to infrastructure environment. The
nature of ubiquitous media environment will have implications on value networks
openness and flexibility. Value networks will need a fluid structure to cope with changes
in very short cycles.
Keywords: E-paper technology, Newspaper, Multiple channels, Value networks,

Ubiquitous media environment
1 Introduction
News publishing organizations of today are often organized as media houses publishing in
multiple channels, distributed in a multitude of infrastructural environments, such as print,
online, telecom and broadcast. Given the frequent introduction of new devices and improved
wireless communication technologies as well as the advent of new technologies such as the e-
paper, new business opportunities emerge for these organizations. This development enables
reaching a mass-scale audience anytime, anywhere through a multitude of devices resulting in a
ubiquitous information environment [1]. A ubiquitous information environment is a
heterogeneous assemblage of interconnected infrastructure and service solutions to support
mobility, mass scale and digital convergence [1]. We use the term ubiquitous media
environments (UME) to discuss the integrated infrastructures and services applicable to the
news publishing industry.
Future UME differs from multi-channel publishing in that services would be channel-
independent. Services would be accessible anytime, anywhere in the most convenient way
depending on users needs as well as physical and social situation [2,3]. This requires that
services can migrate seamlessly between devices and that the environment encompass the
intelligence required to make the necessary context adaptations following users’ movements [3,
4]. As recognized by Jonsson, Westergren and Holstrom [5], these digital advancements lead to
transformation of existing value networks, or even to disruption of value networks and business
models [6,7].
This disruptive effect of digitization has been experienced in the news publishing industry [8].
Since the introduction of the internet in the mid 80´s circulation of print newspapers has

M. AKESSON, C. IHLSTROM ERIKSSON 127
declined in Australia, New Zeeland, North and Latin America, and Western Europe and the
advertisers are following the readers to digital media [9]. It is against this background that news
publishing organizations have expanded their business to publishing in multiple channels.
However, this development has been troublesome. For example, establishing business models
and value network enabling profitable business online has been a struggle. This situation is now
experienced in publishing mobile services. There is still a lot of uncertainty related to value
networks and business models in digital environments for media houses [10,11].
Thus, the development towards UME challenges how media houses organize their business and
technological environments in value networks. In this paper we analyze how these
considerations are treated by management in media houses preparing for a new innovation, the
e-newspaper published on e-paper technology. E-paper is reflecting, giving the same reader
experience as paper (such as high contrast and the possibility to read in sunlight) and is thin,
flexible and non-sensitive. These properties give hope that the e-newspaper can replace the
printed edition in time. Given that the e-newspaper combines the readability and overview from
the printed newspaper with the possibilities of digital media such as constant updates,
interactivity and video it offers a good news reading experience. In UME, it will also be
possible to offer personalization, context awareness and location based services.
However, the challenge of realizing these inherent values of UME and an e-newspaper lies in
the ability to change strategies rather than technology [12]. When the realization of inherent
value requires new or radically changed value networks, the innovation is complex even if the
technology is not. The innovative character of UME, composed by an assemblage of
heterogeneous integrated technologies [1] together with the disruptive nature of digital media to
newspaper organizations [8] indeed makes it interesting to study the complexity of forming
value networks in this domain.
In this paper we address the research question; what are the challenges when aligning a new
promising technology (such as the e-paper) with existing value networks enabling value
creation in UME? We do so by discussing the considerations that arise with the introduction of
a new channel in UME. The aim is to contribute to the understanding of value networks in UME
by further exploring new issues and putting the different issues from the literature into a more
detailed perspective. Moreover, the ambition is to contribute to management with insights of
how to align and organize business and technological environments in UME.
The paper proceeds as follows. In section 2 we describe the selected research method followed
by an update on eReader devices in section 3. The theoretical background is presented in section
4. Section 5 presents the empirical results and in section 6 these are related to literature. The
findings are discussed in section 7 and section 8 concludes the paper.
2 Method
This research took place within the DigiNews project (ITEA 03015), a two-year collaborative
research project, including several major technology firms, media houses and universities across
Europe, which ended mid-year 2006. The partners were from Belgium, Spain, Netherlands,
France and Sweden. The project was initiated by Philips Applied Technology in Belgium
together with the Swedish Newspaper Publishers’ Association. The Swedish media house
partners were Aftonbladet (AB), Göteborgs-Posten (GP), Nerikes Allehanda (NA), Norrköpings
Tidningar (NT), Sundsvalls Tidning (ST), Sydsvenskan (SS), Östgöta Correspondenten (ÖC).
European partners from the media sector were Concentra Media (CM), De Telegraaf (DT) and
Le Monde (LM). The overall goal of the project was to explore research and development issues
for an electronic newspaper of the future. The project aimed at combining the accessibility,
simplicity and mobility of printed newspapers, with the advantages of digital media,
communication technologies and portable consumer electronics.

Two different types of data collection methods were used, i.e. interviews with 18 newspaper
managers and designers, and 9 workshops with media house representatives from management,
IT, marketing and design (Table 1).
Table 1. Data collection activities
Interviews Workshops
NT1 (Aug 25th 2004) Editor-in- GPw1 (Jun 2nd 2004) – 6
chief new media participants
ÖC1 (Aug 25th 2004) Business SSw1 (Sep17th 2004) – 8
developer participants
SS1 (Sep 16th 2004) Marketing STw1 (Oct 6th 2004) – 5
manager participants
SS2 (Sep16th 2004) Layout director ABw1 (Oct20th 2004) – 3
participants
ST1 (Oct 6th 2004) Quality STw2 (Oct 7th 2004) – Design
Assurance Manager focus group
th
ST2 (Oct7 2004) Editor STw3 (Nov 24th 2004) – Design
focus group
ST3 (Oct 7th 2004) Web publisher CMw1 (Oct 26th 2005) – 4
participants
AB1 (Oct 20th 2004) Editor-in-chief DTw1 (Nov 22nd 2005) – 10
new media participants
AB2 (Oct 20th 2004) Layout LMw1 (Feb3rd 2006) – 3
director participants
GP1 (Oct 27th 2004) Development
director
GP2 (Oct 27th 2004) Managing
Development Editor
ST4 (Nov 24th 2004) CEO
CM (Mar 23rd 2005) Head of
research
NT2 (Apr 27th 2005) Head of
Editorial Department
DT1 (Nov 22nd 2005) Director of
new media
DT2 (Nov 22nd 2005) Development
officer
DT3(Nov 22nd 2005) Editor
LM (Feb 3rd 2006) Chief Operations
Officer and Managing Director
The interview study included a total of 18 respondents. Each of the semi-structured interviews
was about 60-90 minutes in length. The semi-structured interview guide used at all sites
facilitated the consistency of data collected between sites and interviewees. While allowing
individual perspectives and experiences to emerge, the interview guide provided a systematic
way of delimiting issues discussed in the interview [13]. It covered a variety of topics such as
organization, technology, business models, new services and design. All respondents had key
functions within each newspaper, e.g. editor-in-chiefs, managers, or designers (see Table 1). At
each interview at least two members of the research team was present, one leading the interview
and one taking notes and making sure no key issues were missed. All interviews were recorded
and later transcribed.

The newspaper staff at the four first workshops and the three last was selected to represent
managers, designers, marketing and IT people. In the beginning of the project we formed a
design focus group consisting of representatives from Aftonbladet, Göteborgs-Posten,
Norrköpings Tidningar and Sundsvalls Tidning that were specifically interested in design issues
of the future e-newspaper. Two of these full day workshops concerned strategic issues as well
and are therefore included in this paper (see Table 1). The results from the workshops are
mainly illustrated in the empirical part concerning the pros and cons of today’s media channels.
All workshops were recorded, and notes were taken by one member of the research team.
Relevant parts of the workshops were later transcribed.
In order to categorize the data collected, patterns were identified in the transcribed material [14].
Excerpts in the transcribed material were marked with assigned colors, facilitating data
categorization according to corresponding themes. Within these themes, we scanned the data
material for similarities and differences, providing input for re-examining the initial themes.
3 eReader update
In this paper we use the term eReader for digital reading devices using electronic paper
technology. Electronic paper (e-paper) is the common term for several different technologies
that can be used to produce screens with a number of specific characteristics. The e-paper is
reflecting, giving the same reader experience as paper (such as high contrast, good color
representation and the possibility to read in sunlight). The resolution is around 160 dpi which is
the same as printed newspapers. The e-paper is thin, flexible and non-sensitive. In addition, it
does not require high battery performance, ultimately, the screen image is stable and fix even
when there is no electrical voltage applied. The three different technologies for e-paper,
electrophoreses, dipolar rotation and electrowetting have earlier been described by [15]. In this
paper we present examples of products that at present on the market and that are expected to be
launched during 2008.
Sony introduced the first eReader called Sony LIBRIé on the Japanese market 2004[a]. In 2006
they launched their next generation eReader on the U.S. market, i.e. the Sony Reader (Figure 1).
It has a 6-inch screen, weight is less than 9 ounces and one can do 7.500 page views for each
charge by an AC adapter. It can hold up to 80 eBooks at the same time, and allows PDFs,
personal documents, newsfeeds, blogs and JPEGs [b].
iRex Technologies BV, a spin-off from Royal Philips Electronics, introduced the iLiad (Figure
2) in 2006 and the iLiad book edition in 2008 includes an 8.1 inch screen with 16 levels of grey
and 160 dpi resolution, Wi-Fi, USB ports and MP3 capabilities [c]. Using a special marker,
readers can comment on articles and scribble their notes on the screen. The French financial
newspaper Les Echos, the Chinese newspaper Yantai Daily and the Dutch NRC Handelsblad are
currently publishing on the iLiad [d].
Bookeen Cybook is another product on the market (Figure 3). It has a 6" E Ink display with 4
grayscale and a resolution of 166 dpi. It can be used in both portrait and landscape mode [e].
Another product on the market is STAReBOOK (Figure 4) with a 6" E Ink display with 4
grayscale [f]. The French financial newspaper Les Echos also publish in the STAReBOOK
device.

Figure 1: Sony Reader Figure 2: iRex iLiad
Figure 3: Bookeen Cybook Figure 4: STAReBOOK
Amazon has recently introduced the Kindle (Figure 5) a branded eReader to which over 100 000
books, U.S. and international newspapers, for example The New York Times, The Wall Street
Journal and Le Monde, as well as numerous blogs can be purchased through their own wireless
delivery system, Amazon Whispernet [g].
Plastic Logic is building the first commercial manufacturing facility targeted at flexible active-
matrix display modules (Figure 6) and will start the production during 2008. They estimate an
initial production of one million units per year [h].
Polymer Vision announced in January 2008 the Readius® product for commercial launch by
mid 2008 [i]. It is a hand held mobile phone device with a five-inch monochrome rollable
screen with 16-levels of grey.
Bridgestone is working on a color e-paper. In October 2007, they announced a 0.29-millimeters

thick e-paper that is capable of displaying 4,096 colors on an eight-inch display [j].

Figure 5: Amazon Kindle Figure 6: Flexible display
Figure 7: Readius Figure 8: Bridgestone e-paper
As this overview shows there are several eReader devices with newspaper content. There are
also new devices announced to be launched on the market with new features such as bendable
and color displays.
4 Service Distribution in Ubiquitous Media Environments
The literature review is informed from two theoretical traditions. First, we give an overview of
concepts of ubiquitous computing as it is manifested in information systems research and,
second, of theories of value networks from business administration.
4.1 Service distribution in UME

The emergence of UME indeed renders new challenges for news publishing organizations.
Offering ubiquitous media services requires distribution of services whenever and wherever
users need them, and also for services to be accessible through multiple devices at different
locations. As a result, distributing services in UME, dynamically supporting users anytime
anywhere, is very challenging.
The multi-contextual nature of ubiquitous information environments requires seamless support

for individual’s context switches, to be a well working computing environment [4]. This
requirement means that the ubiquitous environment must be capable of handling the dynamics
of users’ social contexts, i.e. capable of context awareness. Context-awareness is typically
related to physical location and user identity, but can also include knowledge about time,
history, social context, other people, work or private etc [2]. In addition, to meet the personal
needs of mobile users, ubiquitous services will require personalization, support for dynamic
mobility, and support associated channel adaptations [1].
Moreover, the distribution of services needs to be adapted to the dynamics of users´

environments in terms of changing technological capabilities in the environment, and to
changing hardware capabilities [16]. Services may be accessed through multiple devices by the
same user, and seamlessly migrate to another during service delivery [1,4]. As a result, use
patterns can be described as fluid [4] and consequently services cannot be designed on
assumptions of devices or infrastructure capabilities.
The infrastructure resources of UME must accordingly enable distribution in diverse ways, and
in large scale. A global information infrastructure will be geographically spread and
institutionally disperse delivering telecommunication solutions, wired and wireless

communication solutions [1]. Not only need an infrastructure in UME support geographically
distributed and mass scale distribution of services but must also support a wide variety of digital
data such as text, numbers, audio, video and images [17]. The diversity of different digital data
formats creates challenges for effective handling as there are no conceptual basis for
representing the underlying semantics of for example a video [17].
The rapid advances in technology continuously changes how services can be produced,
distributed and consumed. The infrastructure needs to be scalable for millions of users with a
large number of devices. Further, the technical and software solutions must be generalizable
across multiple application domains to be profitable for solution providers. All of these issues
also have an impact on performance that will have to be on an acceptable level for consumers
[17]. Accordingly, IT infrastructure concerns global reach and range, scalability, flexibility, and
openness to emerging technologies [18]. For a service provider, global infrastructures include a
wide network of external actors and potential partners with steady or loose arrangements, even
on a need basis.
Not only do the challenges of distributing services in UME cause changes in media houses so
does the ongoing convergence across the computing, communications, and content industry
[18]. This convergence redefines the roles and relationships within and between the industry
firms and has resulted in a development of new IT architectures. In turn, this has influenced how
IT activities and relationships are organized. Driven by IT development, firms are developing
market responsiveness and flexible ways of creating customer value. Digital convergence,
globalization and the competition constantly challenges the IT organization. Therefore, new
strategic networks and alliances are being built with key technology partners [18].
As a result, the emergence of UMEs has impact beyond a single organization. The rapid
development of technology supporting social and physical mobility accelerate the span and the
complexity if interorganizational coordination. In addition to the governance and control of
infrastructure issues within an organization, a key challenge is the control and ownership across
various media in multiple contexts [1]. Thus, there is a need to integrate these issues in
decisions and strategies as the rapid convergence media, service and product companies, causes
profound changes in the organizational and industry structures and the associated value
networks [1]. How to manage infrastructure and strategy decisions in this very complex
environment, having impact on external and internal organizational relations as well as on
business and value chains and networks is a major challenge for IS researchers [18].
To address the challenge of organizing business and technological environments in digital

economy [18] suggest a platform (see Figure 9). The platform consists of the components
capabilities, relational structures, and integration structures.
Figure 9: Overview of the platform logic (cf. Sambamurthy & Zmud, 2000)
IT Capabilities refer to IT-based assets and routines that are the most important and vital for
value creation and successful business. Relational structures are the internal and external
relationships and networks that enable the IT capabilities. Finally, integration structures refer to
the integration between IT capabilities and relational architectures. This includes coordination
of IT and business capabilities to create value as well as to give identity to the role of IT
capabilities in the value creating internal and external networks. Indeed, value creating
processes and value networks are central components of organizing business and technological
environments in digital economy.

4.2 Value creation and value networks

The inherent value of a new innovative technology is realized through a value creating process
[3,19]. In business management, value creation refers to the activities a firm performs to create a
value proposal. Porter’s (1985)[20] chain framework analyzes value creation as a chain of
activities at firm level that all contribute to the value of a product or service offered to a market.
The main managerial objective of this analysis is to maximize value at a minimum cost. This
approach is challenged in the digital economy [21].
The digital economy is dynamic and characterized by rapid development and high
competitiveness [22]. Already in 1993 Normann and Ramirez argued to shift from a single firms
value chain thinking to value network thinking where all stakeholder co-produce value. Value
creation can also be co-created with and between customers [23]. Value network is a more fluid
structure than the value chain construct and therefore regarded to fit better to the complex
dynamics of value creation in the fast moving e-commerce [21]. Value networks can be
understood as the network of relationships to a firms stakeholders and the internal
organizational structure of a firm, i.e. “the context within which a firm identifies and responds
to customers’ needs, solves problems, procures input, reacts to competitors, and strives for
profit” (p 36, [6]).
As demonstrated by Vanhaverbeke and Cloodt [7] interorganizational networks need to be set

up and managed to commercialize innovation successfully. These networks are important to
jointly create value to targeted customer groups. Value networks are linked through business
models unfolding the potential value of an innovative technology [24, 25]. Orchestrated by
firms architecting business models network members are aligned to realize that value [24]. A
business model encompasses articulating customer value, identifying customer segments and
the structures for value creation including financial structures, competitive strategy as well as
participants in value networks [24].
All participants should profit in a value constellation. The strength of a value network is
determined by the added value compared to alternative networks, and by the commitment from
the participants [7]. A value network creates value through complex and dynamic exchanges of
three types of value currencies; goods, services, and revenue; knowledge; and intangible
benefits [21]. According to Allee [21], knowledge and intangible values are of equal
importance as revenue exchanges, not the least in the information and knowledge economy.
Value networks and value creation has been addressed in ubiquitous computing research.
Lyytinen and Yoo [1] highlight the transforming effects of ubiquitous information environments
on a business level as an important path for research. Responding to this call, Jonsson,
Westergren and Holstrom [5] conducted an explorative study of how ubiquitous computing can
influence business. This study showed that value creation process in ubiquitous computing
environments should be designed cross organizational boundaries.
Adding to this line of research we aim to contribute to the understanding of value networks in
ubiquitous information environments by further exploring challenging issues. In this paper we
address the research question what are the challenges when aligning a new promising
technology (such as the e-paper) with existing value networks for enabling value creation in
UME? We do so by analyzing the newspaper organizations considerations concerning the
introduction of a new channel, the e-newspaper. The data presentation will highlight newspaper
organizations considerations regarding the capabilities in UME, existing channels with their
advantages and disadvantages, the opportunities and barriers with including with the e-
newspaper, and finally the challenges of aligning the e-newspaper with existing value networks.
The data presentation will be followed by an analysis in relation to the referenced literature. In
order to give a comprehensive view, we organize this analysis according to the platform logic
presented by Sambamurthy and Zmud[18] discussing the concepts from literature regarding IT
capabilities and design visions of UMEs [1,2,4] as well as concepts of value creation actors

[6,7], currencies of value exchange [21], and realization of business value [12,24] in value
networks. Drawing on this background we will discuss challenging issues related to aligning a
new promising technology with existing structures.
5 Adding the e-newspaper channel
At present, many news publishing organizations are organized as media houses with a multiple
channel publishing strategy. We start by discussing this environment and critical issues related
to the alignment of a new promising technology and the ongoing development towards UME.
This part mainly derives from the nine workshops. Thereafter, we discuss the visions and
challenges related to the introduction of a new channel, the e-newspaper, and discuss the major
issues related to the alignment with existing structures identified in the interviews illustrated
with quotations. Finally, we discuss the empirical findings in relation the literature on
ubiquitous information environments.
5.1 Existing structures

During the nine workshops we have worked in an iterative process collecting participants’
views on the existing channels of today. Since only a few of the newspaper organizations have a
broadcasting channel (e.g. TV and radio) we here focus on the three main environments, i.e.
print, online and mobile.
Print
In the print channel the distribution is physical and the majority of media houses own their
distribution, thereby controlling the whole value-chain. The value networks are to a large extent
organized within the newspaper industry. Within the industry it is common to cooperate on
distribution. One example is that two newspapers located nearby each other have an agreement
that one of the newspapers buy the distribution service from the other. Since the print channel
has a very long history, typically more than 100 years, these networks are stabile and well
established. However, the business in the print channel has been experiencing declined revenues
from advertising and decrease in subscriptions. In most of these organizations, this has lead to
re-organizations and having to let employees go. The threat discussed to be the most pressing is
that the printed newspaper does not attract young adults. Historically, the respondents mean that
young adults typically starts to read and subscribe to newspapers when they start a home and a
family. This pattern is now changing and these newspaper organizations are worried about the
future for the printed newspaper. The respondents discuss the introduction of digital publishing
as the most important reason for this development. However, the printed newspaper is still the
dominant business.
Online
Since the mid 90´s. these media houses all have online channels. The online newspaper on the
web is the dominant service but they also offer pdf newspapers, mail services etc. In the initial
years the media houses had a god economy and could afford experimenting online with little
return on investment. But after a few years many of these organizations have had a long struggle
to make the online publishing profitable. It has taken a long time to attract advertisers and it is
very difficult to charge readers for online content. One of the reasons mentioned for the later is
that readers pay for their computer and to the internet operator for the internet access and traffic
and that they therefore are not prepared to pay for content. Another reason mentioned is that
building the technical environment and competence they have today has been very costly. Some
of the Swedish newspapers have invested in a joint company, i.e. Citygate, for mutual
development of technical solutions for online services. They share the development costs and
then market the services under their own brand. Examples of such services are: stock
information, dating services, weather services, games etc. Some organizations have moved the
online business to separate companies or business units to be in control of the business and
separate it from the print channel. These are still strongly interrelated with the printed
newspaper on management level. In most organizations, the online newspapers have found their

place and are considered an important channel for the media houses. The banner market has
grown considerably during the last years, as well as revenues from services for which the
readers register and pay. Webb TV and other multimedia content have been intensively
developed the last few years and many of these media houses have employed new staff working
with the web.
Mobile
Even though most media houses publish some simple services to mobile phones, the most
common strategy is to wait and see. Investments in new technology and competence are limited
in most cases. The media houses recognize there is a big future potential in mobile services but
they do not think either the users or the advertisers are mature enough. But the most important
barrier for investments in the mobile channels is the revenue sharing between the content
providers and the telecom operators. The media houses regard that this today is in favor of the
telecom operators. There are ongoing negotiations between the actors in the networks of mobile
infrastructures. Another problem is the diversity of devices and the lack of standards. Some of
the media houses have found a way around that by developing their own downloadable
programs that runs on any mobile phone. The media houses are active in developing strategies
for their mobile channels and are currently trying to find suitable business models as well as to
developing new and value adding services.
5.2 Relation between channels

In the workshops, one of the tasks was to compare the pros and cons related to publishing news
services in print, online and mobile channels. In table 2, we present a summary of the most
common views on the pros and cons with focus on the value creation of products/services.
These views are from a media house perspective as well as the respondents considerations from
a consumer point of view.
Table 2: Pros and cons of today’s media channels
Channels Pros Cons

Print No power consumption needed Old news
Shareable Waste of paper
Durable No interactivity
Portable Geographically limited
Dispensable Expensive production and
Scannable distribution
Good overview
Long term and loyal customer relations
Online Up to date Uncertain business model
Always on (digital content is regarded as free)
Interactive Not mobile
Searchable Poor overview
Archiving Difficult to know who the reader is
No space limitations Lost control over distribution
No geographical borders for distribution
Mobile Anytime, anywhere Expensive
Push and pull Small screen
Immediacy Very poor overview
Personalization Slow (most people still have slow
High penetration phones)
Many diverse devices
Lack of standards
Bad revenue share
Unsatisfying business model
Lost control over distribution

As can be seen in table 2, the view of the pros and cons are distinguishing rather than
complementary. There are many new opportunities offered for value creation in digital
channels. However, digital channels are not without shortcomings. One of the most important
cons for the newspaper organizations is the lost control over distribution in digital channels
which strongly affect their business models. As stated in the introduction the media houses are
now facing an introduction of yet another channel, i.e. the e-newspaper published on e-paper
technology. Since the e-newspaper is regarded to hold the potential of replacing the printed
edition in the long run, and thereby reducing the distribution and printing costs dramatically, the
media houses follows the development with interest. In table 3 we present the expected pros and
cons of the e-newspaper.
Table 3: Expected pros and cons of tomorrow’s e-newspaper channel
Channel Pros Cons

e-newspaper High readability Poor overview
Broad content base Another device
No computer feeling Small screen size
Up to date - more editions No color in the beginning
Durable Expensive
Mobile and portable Cannibalizing on existing
Anytime, anywhere channels
Environmental friendly
Interactive
Personalization
Searchable
No geographical borders
The pros and cons related to distribution were not included in these discussions. As shown in
table 3, there are many hopes for extended opportunities comparable to visions of ubiquitous
services. The cons are mostly related to the participant’s opinions of e-paper devices, but also
give expression for anxiety of how an e-newspaper would affect existing channels.
In the following we will focus on insights from the 18 interviews grouped into the following
themes; general visions of the e-newspaper, infrastructure for distribution, relationship to other
channels, internal organizational issues, and finally external and interorganizational issues.
Thereafter, we summarize the expected pros and cons of the e-newspaper channel.
5.3 General visions of the e-newspaper

The respondents agree on that the greatest gain for the e-newspaper channel is the possibility of
reduced distribution costs. Even though many of the respondents believe that the printed
newspaper will be a parallel channel for a long time still, some think that the e-newspaper
channel can replace the printed newspaper in the long run. The dream is to be able to shut down
the printing press, eliminating the paper cost and physical distribution. In addition, the
geographical limitations of physical distribution would be overcome as illustrated by this quote.
The foremost opportunity with the e-newspaper is to replace the printed

newspaper, to get rid of the printing press and save the woods, and to get away
from the physical distribution…But the e-newspaper has to be very simple and
mobile. That is the vision but step one will probably be to complement the other
channels and in step two it might replace the printed newspaper.
E-paper devices are expected to have some level of computing capability, memory and storage
capacity. Mobility was considered an important property of the e-newspaper by the respondents.
They assume that for the e-newspaper to be successful and meet user expectations it needs to be

mobile and pocketable and light weighted. The vision of being able to deliver news services and
updates, even very local news, anywhere and anytime is regarded as an attractive opportunity
for media houses. As demonstrated by the quote below, this opportunity is regarded as even
more attractive if it can be accomplished with wireless technology.
Wireless updates increases mobility. All reader types are more or less mobile.
Mobile workers, travel to and from work, young readers are all mobile. Even
senior citizens are very mobile today. That is why I believe we will be successful;
we are much more mobile today.
This is regarded as an important development towards UME. Further, added value such as
location based services, personalization and context aware services are expected to be important
for a successful e-newspaper. The respondents predict changed reader habits in the sense that
you will have your own newspaper, that it is personal, and that you will bring it wherever you
go and that you will read it several times during the day. One respondent expressed it like this.
An e-newspaper will be read from the morning until the evening, it is mobile so you
will bring it along all day. It will be MY newspaper. You can take it out of your
pocket whenever you like.
However, there are contradicting views on the value of personalized content. Some believe that
a careful approach to personalizing content is wise with the argument that part of the news
experience is “not knowing” what you will read beforehand, as demonstrated by this
respondent.
I can imagine that we will customize content for different target groups,
geographically and interest wise. I think personalization of content is overrated. I
have never wanted personalized services for myself. I don’t think there is a mass
scale reader demand for that.
The respondents all agree that attracting advertisers to the e-newspaper is a great challenge.
First of all, there must be a mass scale audience. Second, the e-newspaper must have capability
such as color for presenting ads. Lastly, it is regarded as very important to offer interactivity in
advertisements to attract advertisers. Some of the respondents also stress that there are very
interesting possibilities with this bottom-less media for new types of advertising space. As
stated by this respondent searchable ads may be interesting in an e-newspaper.
To be successful in attracting advertisers color is a must. The ads must also be as

effective for the advertisers as today’s ads, I mean that customers react on them. It
is also important that the ads are interactive and maybe it should be possible to
search for ads.
5.4 Infrastructure for distribution

There are no clear opinions of which infrastructure environment that would be preferred for the
e-newspaper. The thoughts on what influences these decisions are focused on capacity, cost and
value-chain. The importance of stabile performance and secure delivery are stressed, and are
regarded as prerequisites Capacity for multimedia content and interactivity are considered as
important, as illustrated by this quote.
For us, it does not matter what the infrastructure for distribution is, the cheapest
and most effective perhaps. Satellite or broadcast, broad band internet or mobile,
wireless – I don’t know. Everything is ok as long as it can handle multimedia, we
must have support for that. The future is multimedia, it is only the printed
newspaper that is not.

Some respondents believe that converging technology and the development towards UME will
bring new and interesting possibilities. For example, this respondent thinks that GPS technology
would add value to the e-newspaper.
I think GPS technology is interesting. Depending on where you are the news could
change. That would be truly mobile!
5.5 Relationship to other channels

One critical issue discussed is the e-newspaper´s relation to other channels. The respondents are
anxious that adding a channel might reduce consumers’ interest in another. Some regard the e-
newspaper to be a competitor to the printed newspaper. Others think it will compete with the
online channels. Some believe that the development towards UME will mean that all digital
channels will be integrated and together replace the printed newspaper, as illustrated below.
I think that all channels together will replace the printed newspaper. My wish is to
have one editorial department that produces multimedia material that is packaged
and spread in the different channels as it suits best.
Some respondents’ hope of replacing the printed newspaper is more restrained. There is also the
opinion that the printed newspaper will survive as shown here.
I regard the e-newspaper to be a complement to other channels. I do not think it

will kill the printed newspaper. That is only a wet dream. Historically new
channels have started as complementing channels and then maybe replaced others.
There is no channel that has been completely replaced though, except for maybe
smoke signals.
There are also strategic and organizational considerations that are related to multiple channels.
Some of the respondents regard the multiple channel environment of the media house to be
necessary for future news publishing enterprises. In this line of thinking, one channel would not
be regarded as more valuable than another. They would be regarded as an assemblage of
channels that together would reach the audience in the best possible way given the resources the
reader has in the particular situation. As this respondent witness, the newspaper brand is
regarded as a very important asset in UME.
That we are a media house today, with many different channels, is not a
coincidence. The foundation for choosing this strategy is that it is getting harder
and harder to charge for content…We believe that our competitive advantage in
the future is the newspaper brand and credibility, in new media and in multiple
channels.
Others have a strategy of prioritizing amongst the different channels, regarding some to be more
important than others.
We prioritize our efforts and investments for different channels. The order is 1) the
printed newspaper, 2) the web, 3) the PDF newspaper, 4) mobile services, and if
we start publishing on e-paper it will be number 5.
5.6 Internal organizational issues

As illustrated by the quotations from the interviews with newspaper managers, adding a new
channel will have an impact on the organizational structures. The organizing and competence
profile of many newspapers is concentrated to the production of the printed newspaper. As new
channels are introduced these structures are disturbed.

Some professional roles will radically change. We are tremendously many editors
for fairly little material. E-paper will bring a totally new way of editing. It will be
more and more automated. The workload will be loading the databases and
packaging the content.
After experiencing the difficulties of getting online services profitable, many organizations cut
down on technical staff. Now, this has influence on the level of technical competence among the
editorial staff, resulting in educational efforts regarding digital technology and publishing, as
shown here.
We used to have more technical people, but when the dot com balloon busted we let
almost all of them go, they were 80 people. They had no editorial skills. Now we
are retraining editorial staff to handle the technology.
However, the technical and competence issues are not regarded to be the most challenging.
There are strong conservative traditions related to the printed newspaper that influence the
attitudes towards introduction of new media. These attitudes build barriers that according to the
respondents are difficult to challenge, as illustrated with this quote.
There will not be any press stop as today, but some do not realize that. There is
nothing as reactionary as a newspaper editorial staff when it comes to change. The
willingness to change is zero. It is in people’s heads, you cannot even talk about
shutting the printing press down. For them the printed newspaper will be there
forever. To make them leave all of their traditions behind – I do not think so.
Some respondents are even that drastic that they think they would have to re-new the entire
staff, as quoted below. This is because the heritance from the long tradition with the printed
newspaper is very deeply rooted in newspaper organizations.
For us to succeed with the e-newspaper we probably would have to exchange the
entire staff. The e-newspaper will require a totally different competence profile
than we have today. E-paper is still hypothetical to us, but we can compare with
the introduction of the web. The web is still not taken seriously enough by the
editorial staff. Paper is more prestigious, whereas the web is second-class.
But there are exceptions to this picture. In addition to those organizations that are companies or
business units separated from the printed newspaper, the mindset in some traditional newspaper
organizations is more positive. One example from a respondent in a newspaper organization
positive to change and have a climate of welcoming new technology is given with this quote.
Our new editorial system is a multiple channel system. As soon as we have

implemented this system we will create a channel for e-paper and test. We want to
be first with e-paper. We want to hit the big dragons on their fingers. We can do
that because we are not as sluggish and slow, idea to decision is a lot faster in our
organization.
5.7 External and interorganizational issues

The most important external relationships are obviously readers and advertisers representing the
two customers of a newspaper. Further, the relationships between industry actors considered to
be very important. These relationships are perceived as critical for business models and for
strategic alliances. The general understanding is that media houses would gain from acting on
an interorganizational and international level when forming networks for the future. The
disruptive effect of digital channels has lead to the awareness that every media house cannot
build their own value networks and their own business models. To be competitive in UME, they
recognize that interorganizational relationships are very important. To take the position that

owns the customer relation is regarded to be the most important in network building as
illustrated below.
The business models have to be thought through and what role we fill in it. The one
who owns the relationship to the customer or reader mediates, we can stimulate
each other if we own that mediator role together in an industry network.
External relationships are closely related to strategy and distribution issues. As presented above,
there are contradicting views on technical and distribution issues that influence. Some
organizations have an outsourcing strategy. This quote shows that concentrating competencies
on content rather than on distribution is one of the reasons behind this strategy.
We have outsourced the whole IT department, we do not house any online services
our selves. Our core competence is not distribution, it is quality content. That is
why we have outsourced distribution, and technology, even the payment systems.
We will have the same strategy with the e-newspaper – if it requires technical
competence – outsource. If it is cheaper – outsource.
There are other organizations who regard ownership of technical solutions to be important. In
some cases that ownership is a joint venture between the newspaper publishing companies, in
Sweden typically between companies with the same owner. As told by this respondent, the
argument for this strategy is ownership and control over distribution, reducing the complexity of
external relationships needed to distribute the e-newspaper.
We as newspapers must own the distribution channel. It does not work well for us
as it is on the mobile phone side. I can imagine that we can give the e-paper device
to our subscribers for free and for a payment they can get access to additional
content within a network of cooperating actors owning the infrastructure together.
This strategy is highly influenced by the experiences from the mobile channels. This quote
shows that the worst scenario for newspaper organizations is to become a middle hand in the
distribution chain.
I think a shared infrastructure among content providers is to prefer. It is important

not to become dependent on one or two operating companies that build their
business model on holding the content provider as their hostage.
Summing up, the empirical material presented gives a comprehensive view of considerations
related to the alignment of a new publishing channel. In the following we will present an
analysis in relation to the literature.
6 Aligning the e-newspaper in a ubiquitous media environment
We start the analysis by presenting a graphical overview of considerations related to the

alignment of the e-newspaper within the existing multi channel environment drawn from the
interviews and workshops (Figure 10). As can be seen in figure 10, the multi channel
environment is very heterogeneous. The number and character of relationships differ from
channel to channel. Since the efforts and investments required from customers differ between
infrastructure environments, the value offer needs to be differentiated between channels.
Clearly, the decisions on infrastructure for distribution have consequences for value networks
and business models. Drawing on the visions of a future UME, coping with the heterogeneity
between value networks in different infrastructure environments will be a core challenge.

Figure 10: Considerations related to the alignment of the e-newspaper within the existing multi
channel environment
Hereafter, we discuss the most critical considerations by discussing the empirical findings in
relation to the literature on distribution of services in ubiquitous information environments and
on value networks. In order to get a comprehensive view of challenging issues, we have
organized the discussion according to the platform logic by [18], i.e. IT capabilities, relational
structures, and integration structures.
6.1 IT capability
Anytime and anywhere delivery of services for the e-newspaper channel is considered to be the
most promising possibility by the media houses in our study. Some of the pros mentioned as the
most important for the e-newspaper channel are the possibilities of personalization and local
awareness. As illustrated by the quotations, there are however contradicting views on the
importance of personalization. Still it is anticipated that the IT capability of the infrastructures
and e-paper devices (i.e. an eReader device) must have support for location awareness and
personalization functionality. Further, supporting users dynamic movements between physical
and social context [1] and seamlessly support fluid use patterns [4] are capabilities anticipated to
be of importance for an effective UME. As a result, devices in UME need to be functional in
multiple infrastructural environments. Moreover, the eReaders for consuming the services
offered through this channel need enough computing capacity, memory and storage to support
these capabilities as well as to support multimedia content presentation. These capabilities are
also important to leverage value to advertisers. There is a contradiction in the product
development between IT capacity of the eReaders and the demand for a pocketable and light
weighted device. Consequently, one important issue to address when aligning the e-newspaper
with existing structures is: What are the requirements on devices to leverage customer value in
UME?

Some of the capabilities of UME are related to a well functioning return channel as shown by
the empirical evidence. In some cases automatic identification of device, location and other
information related to context [2] need a return channel. As suggested by one respondent, GPS
technology might be used to locate readers. In other cases users might actively be using a return
channel, to interact with content such as advertising or to make a comment or send a picture.
The requirements on the return channel are related to the publisher push and/or pull strategies of
service offers. The capabilities of managing and storing the information flow, as well as
package and deliver information and knowledge back to for example advertisers are important
parts of value creation in UME. Added to this, the return channel is a prerequisite for systems to
be aware of users’ context and for users to be aware of the environment. This is ideally as
unproblematic for users as possible and preferably wireless. The second critical issue to address
is therefore: What are the requirements of return channels to enable value creation in UME?
The capacity of the integrated infrastructures for delivery in UME is of course fundamental for
value creation. In the study it was shown that all media houses are convinced that the news
services of tomorrow are multimedia content based. Thus, the assemblage of technologies used
for distributing the e-newspaper content need to have capacity for heavy traffic as well as for a
large variety of data formats [17]. Furthermore, the infrastructure must be flexible and open to
new types of technologies to meet the expectations by the media houses. Delivery anytime and
anywhere puts requirements on well functioning integration structures cross infrastructure
environments. Therefore, an important question for distribution is: What is the minimum
capacity level for integrated distribution infrastructures in UME?
In UME the value offer to consumers is built on the features of mobile and ubiquitous services.
In the interviews it became clear that media houses do not expect consumers to pay for content
or services, the value currency between the content provider and consumer is ubiquitous
services in exchange for market communication exposure, i.e. an intangible value currency [21].
The respondents believe that the monetary revenue will come from the value exchange from the
advertiser who will pay for communicating with their markets in UME. The added values to the
advertiser besides exposure are increased targeting and information and knowledge about
customers reaction, i.e. knowledge currency [21]. These values are created through the features
of ubiquitous services. Given the characteristics of UME, these values cannot be built on single
devices such as eReaders [1], these values need to be co-created cross channels, independently
of infrastructure environment. As a result, the following issue is important to address: How can
value be created independently of device and infrastructures in UME?
6.2 Relational structures

As our empirical findings show, the development towards UME will have large impact on the
media houses. The structures for organizing competence, IT management and value networks
are all challenged. The logics for governing an environment of this character need to be
rethought [18]. As testified, there are conservative attitudes and traditions in these organizations
that are difficult to overcome. In addition to this, the interorganizational and customer
relationships of the media house are challenged. One important aspect is the relationship to
customers, in this case readers as well as advertisers. In the case of the printed newspaper the
organizations own the distribution, or they have an alliance with another newspaper
organization that does. In these cases they also own the relationship with the newspaper reader.
For online and mobile channels the ownership of distribution is outside the media house, and
sometimes the customer relation goes via an operator. These conditions have impact on the
value chains and business models [24], thus it is important to understand business needs in
relation to ubiquitous information environments [5,25]. As the empirical material show, this is
one of the major issues for the media houses when aligning a new technology with existing
structures. Given the characteristics of value creation in UME it can be expected that co-
creation of value and flexible relationships are required to leverage customer value.

Consequently, an important question is: What are the desired structures for ownership of
distribution and customer relations to enable value creation in UME?
The study gave evidence that the media house is part of a complex world with different alliances
and structures for relationships in different channels. In addition to this heterogeneity, each
channel infrastructure has its network regulations, industrial standards and actors. In a global
infrastructure, these relationships are increasing in numbers [1]. As described by the empirical
data, some of the organizations have formed joint efforts to handle these relations. According to
literature [7,19] this approach is essential to be able to realize the inherent values of UME. On
top of this, there are two different customer groups, consumers and advertisers, demanding for
different values, enabled by different types of technical solutions, but still interdependent. When
aligning the e-newspaper it is therefore important to ask: What are the desired relations to
providers of technical solutions to enable value creation in UME.
6.3 Integration architectures

The media houses discuss new structures for interrelationships regarding the integration of IT
capabilities with business networks, relations to distribution owners and technical solution
providers. The empirical data disclose the uncertainty and tension in the relationships with for
example telecom operators. Accordingly this is a challenging avenue of large scale with many
contradicting interests involved. Still this is a key area for strategic decisions [1,18]. Further,
these integration structures must have the flexibility, the preparedness and the alertness that is
required to meet the constant development and convergence of technology, the numerous and
variety of devices as well as changes in customer needs and expectations. This means that roles
of the participants in the value network might change for each delivery depending on users’
actions. As discussed by Vanhaverbeke and Cloodt [7] this also means that the incentives and
value exchanges change. This is an escalation of the characteristics of the digital economy as
discussed by Amit and Zott [22] and requires open and flexible value network structures
[19,21]. Drawing on Samabamurthy and Zmud’s [18] discussion on integration architectures,
the overall question is: How should the network structures for value creation be integrated with
the IT capabilities of UME for a successful business agenda?
As discussed in literature [1,18] and shown by the interviews the organizational structures and
competence profile is challenged by the development towards UME. The disruptive effect [6] of
new publishing channels causes re-configuring internal value creating networks. This includes
re-defining the core competencies and resources required to create value. The empirical material
also suggests that this includes changing attitudes and mindsets within the existing work force.
As a result, when aligning a new technology to UME the following question needs attention.
What is the desired internal network structure of a content provider in UME?
In table 4 on the next page, we summarize the challenging issues related to the alignment of a
new promising technology in UME.

Table 4: Alignment issues and considerations in relation to the platform logic components
Platform Logic Challenging issues Corresponding

Components considerations
(Sambamurthy
& Zmud, 2000)
IT Capabilities Device capabilities • What are the requirements
on devices to leverage
customer value in UME?
Return channel • What are the requirements
capabilities of return channels to enable
value creation in UME?
• What is the minimum
capacity level for integrated
Capacity and distribution infrastructures
integration of in UME?
infrastructures • How can value be created
independently of device and
IT independent infrastructures in UME?
value creation
Relational Role of distribution • What are the desired
structures structures for ownership of
distribution and customer
relations to enable value
Role of technology creation in UME?
solution provider • What are the desired
relations to providers of
technical solutions to
enable value creation in
UME?
Integration Integration of • How should the network
architectures network structures structures for value
creation be integrated with
the IT capabilities of UME
Internal network for a successful business
structure agenda?
• What is the desired internal
network structure of a
content provider in UME?
In the following section we will discuss the practical and theoretical implications of these
findings.
7 Discussion
This study aimed at exploring challenges of aligning new technology with existing value
networks enabling value creation in UME and to put the different issues discussed in literature
into a more detailed perspective. We did so by studying media houses in the newspaper industry
as focal actors enacting in the DigiNews project with the incitement of finding new business
opportunities by unfolding potential value of the e-paper innovation in newspaper publishing
market. This resulted in a comprehensive description of considerations related to the alignment
of a new promising technology, the e-newspaper, with existing value networks. Indeed this
study demonstrates that there are implications for all components of organizing business and
technological environments [18] when aligning a new technology with existing value networks

in UME. As shown in the previous section there are many promising opportunities of an e-
newspaper and a development towards UME but there are many challenges to address (see table
4). Hereafter we will discuss the implications of these challenges for management and the
contribution to literature.
7.1 Practical implications

This study indeed portrays the complexity of value networks in UME and the complex and
challenging issues management need to address. There are risks of re-experiencing the problems
and disruptive effects of digital media if media houses do not take these challenges seriously.
Even today, the parallel networks in different infrastructures cause uncertainty of value creation
and the associated value networks and business models. Given the assumptions and visions of
UME presented in this paper, it can be anticipated that value networks in UME will be required
to be very open, cross existing boundaries, organizational, interorganisational, technical, as well
as cross existing value networks. To realize the inherent values, all of these boundaries are
challenged since services and content delivery will be driven by customers’ needs and
behaviors.
Management in media houses would gain from developing a more open and flexible approach to
the relationships and exchanges forming value networks. Today, some of these organizations are
“stuck” in traditional mindsets of the printed newspaper. The consequence is an internal
entrance barrier opening up for other actors to create gain the market in the new environment.
As shown by the character of the challenges when aligning a new technology in an emerging
UME, strategies cannot be built on tradition model of publishing newspapers. For those media
houses that are not capable to unleash of the historical straightjacket, the doomsday prophets
foreseeing the death of traditional newspapers might be right.
Co-producing value in dynamic networks, recognizing that readers and advertisers,

infrastructure and technical solutions providers are all part of the networks, together with an
open approach spanning the boundaries of traditional thinking (selling information to readers
and exposure to advertisers, total control over value chain, building all solutions in-house, only
cooperating within their own domain etc) is recommended. In addressing the challenges
described with the case of introducing the e-newspaper this shift in strategic view has surfaced.
Some media houses have realized that exchanges of knowledge and other intangible values are
as important as traditional value exchanges between participants in external value networks in
the networked digital economy. These organizations have expanded their tradition of
cooperation of technology (competing with content) competence boundaries (internally) and
cross industry boundaries.
Our ambition is that the list of challenging issues can serve as useful guidance for management
when preparing for the alignment of a new technology and the development towards UME.
Even though the development towards UME involves numerous and difficult challenges, there
are many new opportunities for the news publishing industry. These opportunities may lead
traditional newspaper organizations in to a new era, aligning new digital publishing with their
core business and competences of journalism, publishing technology and market
communication. We believe that this illustration of considerations for media houses, could apply
to any organization providing services for a wide audience in a ubiquitous information
environment.
7.2 Contribution to literature

In line with previous research (see e.g. [1,5,25]) this study disclose the heterogeneity of
ubiquitous information environment. This study contributes to previous research with increased
understanding of implications of the heterogeneity to value networks in UME. To realize the
inherent values of UME, the value networks must be open [7], flexible [21] and even fluid to
cope with e.g. fluid use patterns, converging media, infrastructures, technology and emerging
standards.

Traditional basis for value creation and business models such as media channel, device or
infrastructure for delivery will not be utilizable. In newspaper industry, value creation is
dependent on type of customer (in general reader or advertiser), on device (such as paper,
computer or mobile), and on the infrastructure for distribution. At present these are parallel
structures and conditions for value creation, i.e. value differs between customer groups and
channels. However, the development towards UME is changing this, for example the mobile
internet is one example of the effect of convergence. In UME the value creation will depend on
user needs and actions to a much larger extent and on user demand. Anytime, anywhere,
seamless, location based, and personalized service delivery in any convenient way requires just
in time service delivery on new and fluid terms. The value creation process and the value
networks cannot be designed on assumptions of use situation or technical capabilities of devices
and infrastructures. Neither will traditional groupings of customer segments be utilizable.
Moreover, the relationships in value networks will change. Depending on user actions and
individual preferences different stakeholders might have the primary customer relation or be in
a mediating role. As this study shows the stakeholder in control of distribution is often the
owner of the end customer relation and thereby holds a power position in the value network. In
UME, the stakeholders in control of distribution might be interrelated during service delivery.
That means that the roles in the value network are fluid depending on what, when, where and
why an action is taken by a user.
This fluid nature of value networks in UME will also apply to the value exchanges between
stakeholders. Given the different context for user actions the same service and content might
provide different value. For example, a user might be willing to pay for a service in one
situation but not in another situation for the same service. In some cases advertisers may have
an interest in exposure and in other cases not, even though the value to the customer might be
equivalent in both cases. In summary, these changing conditions result in interrelated and fluid
value networks which in turn mean that the value exchanges, currencies, relations and roles in
the value crating process will change in very short cycles.
8 Conclusions
In this paper we addressed the research question: What are the challenges when aligning a new
promising technology (such as the e-paper) with existing value networks enabling value
creation in UME? To address this question we have conducted a literature review of
infrastructure issues in ubiquitous information environments and literature on value creation and
value networks. Drawing on this literature together with empirical results from 18 interviews
and 9 workshops with media house representatives in Sweden, Belgium, France and the
Netherlands, we suggest eight challenging issues, presented in section 6.
This study shows how the heterogeneity of ubiquitous information environments influences
value networks. In this paper we have discussed that enabling value creation in UME requires
fluid value networks. The value networks are formed and reshaped on the basis of user needs
and actions, i.e. a pull oriented rather than push oriented architecture. The fluid nature means
that the value creating processes, value constellations and relationships, and value exchanges
and currencies might change for each service delivery, i.e. in very short cycles. This requires an
enormous openness and flexibility and challenges existing ideas of boundaries in the digital
economy. As a consequence it can be suggested that organizing business and technological
environments cannot be separated from value creation processes. To enable the envisioned
values of ubiquitous environments all of the components of a business platform might need to
be co-produced in integrated networks and not separated by the boundaries of traditional
practices.

There are several limitations to this study. First, the broad approach to challenges should be
followed by deeper investigation with concerned stakeholders to verify the significance from
several viewpoints. Second, the notion of fluid value networks should be validated in other
industries where ubiquitous information environments apply.
References
1 Lyytinen K. and Yoo Y., Research Commentary: The Next Wave of Nomadic
Computing. Information Systems Research, Vol. 13, No. 4, pp. 377-388. 2002
2 Abowd G. D. and Mynatt E. D., Charting Past, Present, and Future Research in
Ubiquitous Computing. ACM Transactions on Computer-Human Interaction, Vol. 7,
No. 1, pp. 29-58. 2000
3 Yoo Y. and Lyytinen K., Social impacts of ubiquitous computing: Exploring
critical interactions between mobility, context and technology – a special issue for
information and organization. Information and organization, (15:2), pp 91-94. 2005
4 Henfridsson O. and Lindgren R., Multi-contextuality in ubiquitous computing:
Investigating the car case through action research. Information and Organization,
Volume 15, Number 2, pp. 95-124. 2005
5 Jonsson K., Westergren, U. H. and Holström J., Technologies for value creation:
an exploration of remote diagnostic systems in the manufacturing industry.
Information Systems Journal. 2007
6 Christensen C. M., The Innovator’s Dilemma. When New Technologies Cause
Great Firms to Fail. Harvard Business School Press, Boston, MA. 1997
7 Vanhaverbeke W. and Clood, M., Open Innovation in Value Networks, in
Chesbrough, H., Vanhaverbeke, W. and West J., eds., Open Innovation: Researching
a New Paradigm. Oxford: Oxford University Press, pp. 258-281. 2006
8 Christensen C. M. and Davis A. B., “The Future of Newspapers”, Forbes.com,
October 2006.
9 The Economist. Who killed the newspaper. August 26, 2006.
10 Picard R. G., Cash Cows or Entrecôte: Publishing Companies and Disruptive
Technologies. Trends in communication, 11(2), 127–136. 2003
11 Ziv N. D., Towards a New Paradigm if Innovation on the Mobile Platform:
Redefining the Roles of Content Providers, Technology Companies, and
Users. In Proceedings of the international conference on Mobile Business,
IEEE. 2005
12 Christensen C.M. and Rosenbloom R., Explaining the attacker's advantage:
technological paradigms, organisational dynamics and the value network.
Research Policy 24, 233-257. 1995
13 Patton M. Q., Qualitative Research and Evaluation Methods, (3rd ed., Sage
Publications, Inc, California. 2002
14 Miles M. B. and Huberman A. M., Qualitative Data Analysis. Sage Publications,
Inc. California. 1994
15 Ihlström C., Sabelström-Möller K. and Åkesson M., The Challenge of Production
in e-paper Publishing - from new consumption to new workflows. Proceedings of
TAGA 2005, Toronto, Canada. 2005
16 Banavar G. and Bernstein A., Software Infrastructure and Design Challenges for
Ubiquitous Computing Applications. Communications of the ACM. Vol. 45, No. 12,
pp. 92-96. 2002
17 March S., Hevner A. and Ram S., Research Commentary: An Agenda for
Information Technology Research in Heterogeneous and Distributed Environments.
Information Systems Research, Vol. 11, No. 4, pp. 327-341. 2000

18 Sambamurthy V. and Zmud R. W., Research Commentary: The Organizing Logic

for an Enterprise’s IT Activities in the Digital Era – A Prognosis of Practice and a
Call for Research. Information Systems Research, Vol. 11, No. 2, pp. 105-114. 2000
19 Chesbrough H. and Rosenbloom R. S., The role of the business model in capturing
value from innovation: Evidence from Xerox Corporation's technology spin-off
companies. Industrial and Corporate Change, 77(3), 529-555. 2002
20 Porter M., Competitive advantage: Creating and sustaining superior performance.
New York: The Free Press. 1985
21 Allee V., Reconfiguring the Value Network. Journal of Business Strategy, Vol 21,
No4. 2000
22 Amit R. and Zott C., Value creation in e-business. Strategic Management Journal,
Vol 22, pp 493-520. 2001
23 Stabell C. B. and Fjellstad Ö. D., Configuring value for competitive advantage: on
chains, shops, and networks. Strategic Management Journal, Volume 19, Issue 5 , pp
413 – 437. 1998
24 Chesbrough H., Open Innovation: The New Imperative for Creating and Profiting
from Technology. Harvard Business School Press. 2003
25 West J., Vanhaverbeke W. and Chesbrough H., Open Innovation: A Research
Agenda, in Chesbrough, H., Vanhaverbeke, W. & West, J. eds., Open Innovation:
Researching a New Paradigm. Oxford: Oxford University Press, pp.285-307. 2006
26 Andersson M. and Lindgren R., The Mobile-Stationary Divide in Ubiquitous
Computing Environments: Lessons from the transport industry. Information Systems
Management, Volume 22, Number 4, pp. 65-79. 2005
a Sony LIBRIé. Available at: http://www.sony.jp/products/Consumer/LIBRIE/ (April,

16, 2008)
b Sony Reader. Available at: http://products.sel.sony.com/pa/prs/reader_features.html
(April, 16, 2008)
c iRex iLiad. Available at: http://www.irextechnologies.com/ (April, 16, 2008)
d Les Echos e-paper edition. Available at:
http://www.lesechos.fr/epaper/inscription.htm (April, 16, 2008)
e Bookeen Cybook. Available at: http://www.bookeen.com/ebook/ebook-reading-
device.aspx (April, 16, 2008)
f STAReBOOK. Available at: http://www.stareread.com/en/reader.html (April, 16,
2008)
g Amazon Kindle. Available at: http://www.amazon.com/Kindle-Amazons-Wireless-
Reading-Device/dp/B000FI73MA (April, 16, 2006)
h Plastic Logic. Available at: http://www.plasticlogic.com/hi-res.php (April, 16, 2008)
i Polymer Vision. Available at:
http://www.polymervision.com/frameset.php?id=&page= (April, 16, 2008).
j Bridgestone e-paper. Available at:
http://www.engadget.com/2007/10/22/bridgestone-shows-off-ultrathin-full-color-e-
paper/ (April, 16, 2008)

M. HABEKOST, K. ROHLF 149
The evaluation of colour difference equations and

optimization of DE2000
Martin Habekost, Dr. rer. nat., Dr. Katrin Rohlf
School of Graphic Communications Management and Department of Mathematics

Ryerson University
350 Victoria Street
Toronto, Ontario, M5B 2K3
Canada
Email: mhabekos@ryerson.ca
Email: krohlf@ryerson.ca
Abstract
The original colour difference equation DEab published by the CIE in 1976 built the
foundation for modern colour differencing methods. This equation had its imperfections
by giving colour difference numbers that do not correlate with the visual impression of
colour difference. Through the release of the DE94 and later on the DE2000 colour
difference equation the CIE addressed these imperfections. The Colour Measurement
Committee of the Society of Dyers and Colorists of Great Britain released the DECMC-
colour difference equation. This equation has not been modified since it initial inception
in 1984.
With four available colour differencing equations one has to ask oneself which one of
these equations correlates better with the way human observers perceive colour
differences. In this report 35 colours with 4 variations for each colour were evaluated and
ranked by 24 observers. The initial differences were determined under DEab. The rankings
given by the observers were weighted and correlated with the DE values calculated with
the four colour differencing equations. The correlation values were plotted individually
for each equation in respect to Lightness, Hue and Chroma to determine which equation
corresponds better with the perceived differences. The DECMC and DE2000 equations
corresponded well with the perceived differences with an advantage for DE2000.
The DE2000 equation, like the DECMC and DE94 equations, contains weighting factors.
An attempt was made to determine these factors to see if the correlation between the
numerical differences and the perceived differences could be improved. A numerical
optimization method was employed but only minor improvements were observed.
Therefore, it can be said that the DE2000 formula in its current form correlates quite well
with the way human observers perceive colour differences.
Keywords: Colour, CIE, Colorimetry, Differencing, Delta E, CIE DE2000
1 Introduction
Ever since the introduction of colorimetry researchers have attempted to express visual
differences numerically. Various equations were developed and used, but colour scientists could
not exchange the generated colour difference numbers since they were derived using different
equations, which were probably not based on the same colour space [1]. This is akin to people
talking to each other in different languages.
In 1976 the CIE (Commission International D’Eclairage) published the DEab-equation for the
numerical expression of colour differences which is based on the L*a*b* colour space. A DE or
colour difference number of 1.0 was set as the just noticeable difference. For the first time

150 THE EVALUATION OF COLOUR DIFFERENCE EQUATIONS AND OPTIMIZATION….
colour scientist around the world were able to exchange colour difference numbers since the
numerical difference was based on one colour space and one formula [2]. The DEab or ΔEab-
equation can be seen in equation 1:
ΔEab = ΔL*2 + Δa*2 + Δb*2

ΔL* = L* Sample − L*Standard
(1)
Δa* = a* Sample − a*Standard
Δb* = b* Sample − b*Standard
A colour is notated in 3 dimensions. The L*-value represents the Lightness of a colour and has a
value from 0 to 100. The red-greenness of a colour is notated with the a*-value, where +a
indicates the red direction while –a indicates the green direction. The b*-value denotes the blue-
yellowness of a colour with +b indicates the yellow direction and –b indicates the blue
direction. This equation was soon established as the standard equation to express colour
differences and for the first time colour scientists were able to exchange colour difference data,
since they were all speaking the same language. Criteria were published that classified the
severity of numerical colour differences [3].
Shortly after the introduction of the DEab-equation it was realized that the numerical difference
of 1.0 had a different visual meaning depending on the location of the colour in colour space. A
difference of 1.0 for a dark blue could mean no visible difference, while a difference of 1.0 for a
pastel type colour could mean a large visible difference. This is caused by the perceptual non-
uniformity of the CIE L*a*b*-model.
Since the CIE L*a*b* colour space is not perfectly perceptually uniform two possible
approaches can be taken: a.) modify the colour space model, so the colours are distributed
unformingly or b.) modify the colour difference equations to compensate for the lack of
uniformity.
The first attempt to modify the DEab-colour difference equation was made by the Colour
Measurement Committee (CMC) of the Society of Dyers and Colourists of Great Britain in
1984 [4]. The CMC-equation is based on the L*C*h*-notation of colours which can be derived
through simple mathematical transformations from the L*a*b*-notation. It is a transformation
from a Cartesian to a polar coordinate system. The CMC-equation was originally developed for
the textile industry, but has found its way into the graphic arts industry. This equation takes into
account the varying sensitivity of the human eye to colour differences depending on the location
of the colour in colour space. Weighting factors in the CMC-equation modify the size of the
allowable tolerance depending on the location of the colour in colour space. A DECMC of 1.0
under this equation gives a similar visual difference in all regions of colour space.
The DECMC-formula can be seen in equation 2:
2 2 2
⎛ dL* ⎞ ⎛ dC * ⎞ ⎛ dh* ⎞
dECMC = ⎜ ⎟ +⎜ ⎟ +⎜ ⎟ (2)
⎝ lS L ⎠ ⎝ cSC ⎠ ⎝ S h ⎠
The SL, SC and Sh are the main weighting factors for lightness (dL*), chroma (dC*) and hue
(dH*). The two other factors l and c are constant and are defined by the user and weight the

importance of lightness and chroma relative to the hue of the measured colour. The ratio of the
factors l and c are normally set at 2:1 for the Graphic Arts industry.
In 1995 the CIE introduced a colour differencing formula that is based on the RIT/Du Pont data.
This data stems from experiments with automotive paints. The data was compiled by Berns et al
(Berns 1991 [5]) and published as CIE94 or DE94 in 1995 [6].
Like the DECMC-formula this DE94 formula is also based on the L*C*h*-notation and reads as
can be seen in equation 3:
2 2 2
⎛ dL* ⎞ ⎛ dC * ⎞ ⎛ dh* ⎞
dE *
= ⎜ ⎟ +⎜ ⎟ +⎜ ⎟ (3)
⎝ k L S L ⎠ ⎝ kC S C ⎠ ⎝ k h S h ⎠
94
This equation has two sets of coefficients. The factors SL=1, SC=1+0.045C*ab and
Sh=1+0.015C*ab
The k-coefficients are also known as parametric factors and refer to effects influencing colour-
difference judgment. The s-coefficients account for CIELab’s lack of visual uniformity [7].
The DE94 formula corrects a uniformity error for colour pairs of high chroma. Originally, they
gave larger numerical difference than their visual appearance would suggest.
Although the DE94 formula showed an improvement over the DEab-formula, certain colours
still produced larger numerical differences than other hues. The correction of this imperfection
was done by the release of the DE2000 colour difference formula [8][9]. This formula contains
also a correction factor for the blue-violet region. The DE2000 formula can be seen in equation
4:
2 2 2
⎛ ΔL' ⎞ ⎛ ΔC ' ⎞ ⎛ ΔH ' ⎞ ⎛ ΔC ' ⎞ ⎛ ΔH ' ⎞
DE2000 = ⎜ +
⎟ ⎜ +
⎟ ⎜ ⎟ + RT ⎜ ⎟⎜ ⎟ (4)
⎝ k L S L ⎠ ⎝ kC S C ⎠ ⎝ k H S H ⎠ ⎝ kC S C ⎠ ⎝ k H S H ⎠
An observant reader notices that the equation does not use L*C*h* anymore but L’, C’ and H’.
This is a transformation of the L*C*h*-colour space. The mathematical transformation from
one to the other color space goes beyond the scope of this paper. This has been done extensively
at the Rochester Institute of Technology (Sharma, 2005 [10]). The DE2000 formula has five
corrections to CIE L*a*b* as Luo [11] writes: “A lightness weighting function (kLSL), a chroma
weighting function (kCSC), a hue weighting function (kHSH) and an interactive term between
chroma and hue differences for improving the performance for blue colours and a factor (RT) for
re-scaling the CIELAB a*-scale for improving performance for grey colours” [11].
All the newer colour difference equations should be used for numerical differences DEab smaller
than 5 [12]. The corrective terms used in all newer colour difference equations compensate for
the non-uniformity of the L*a*b*-colour space. Attempts have been made in Germany to adapt
the L*a*b*-colour space by modifying each axis, so that the colour difference formula from
1976 can be used and represents the true Euclidian difference between two points in colour
space [13].

2 Experimental section
The experimental part consists of four parts:

a.) Generating colour patches with known numerical differences and print them on an
offset lithographic press
b.) Human test subject rate the perceived colour differences
c.) Rank the perceived differences and correlate them with the numerical colour difference
numbers given by DEab, DECMC, DE94 and DE2000 and determine which one gives a
better correlation
d.) Attempt to improve the weighting factors from DE2000 for obtaining a better
correlation with the observed colour differences.
Before the human test subjects could rank the colour samples it was necessary to generate the
test patches. This was done by generating 35 colours with 4 variations per colour. These
variations were made with a colour difference DEab of approximately 2, 5, 5.5 and 7. These DE-
values were selected based on research previously done [14]. All colours samples were drawn in
Adobe Photoshop CS2 using the L*a*b*-mode to enter the colour information. The variations
were obtained by modifying the a* and b* values, while keeping the L*-value constant. The
same variations of a* and b* were applied to all colours used in this project.
The colour patches were approximately 2 x 2 cm in size. The saved Photoshop file contained no
ICC-colour profile information. This procedure was chosen to ensure that colours would be
chosen that fell within the gamut of the printing device. This was more economical than running
a printing press in this stage of the study.
2.1 Profiling the Devices

An ICC-profile of an HP 5550 colour laser printed was generated using an IT8.7/4 target,
Monaco Profiler software and an X-Rite DTP70 measurement device. The profile was generated
on the same paper stock that was used during the press run of the colour patches.
After generating the ICC-profile for the proofing device this profile was applied to the test
colours and proofs created. All L*a*b* values of the test patches were measured in five
different spots with an X-Rite 530 instrument for each patch to minimize the influence of colour
variations that might occur within each patch. Chromix® ColorThink software was used to
verify that no colours were out of gamut. This can be seen in figure 1.
Figure 1: Colour space of the proofing device with test colours

The test colours are equally distributed within the colour space and cover light and dark colours,
saturated and less saturated colours. A table with the L*a*b*-values of the test colours is listed
in the appendix of this paper.
After the test colours were established on the proofing device it was necessary to generate an
ICC-profile for the four colour offset press. An IT8.7/4 target was printed at the target densities
for coated paper and an ICC-profile generated using an X-Rite DTP70 device and Monaco
profiler software. A comparison of the two profiles is shown in figure 2.
Figure 2: ICC-profile of the proofing device (wire frame) and press (solid colour)
Figure 2 clearly shows that ICC-profile of the proofing device and the press are quite similar.
The largest deviation exists in the yellow region; otherwise, the proofing device somewhat has a
gamut greater than the press. The difference between both devices is not very large; meaning the
colours used in this study did not need any modification since the colours are within the colour
space of each colour reproduction device.
2.2 Perceptual Rating of the colour samples

The viewing test took place in a GTI Colour Viewing Station viewing booth. The colour
temperature and CRI index of the viewing booth were confirmed with an Eye-One instrument
and i1Share-software. The 24 participants could arrange the colour standard and the four
variations for each colour in any way or form and apply the rating scale to them. Before the
participant could continue they had to undergo a colour blindness test using Ishihara test plates.
From the many volunteers a few could not continue the test due to a colour deficiency that was
discovered with the Ishihara test. Each participant used the same rating scale:
• Match
• Slightly different
• Different
• More different
• Very different
The numbers of responses were counted for each colour and a weighting scheme was applied.
The weight allocation took place in the following matter: The number of responses for “match”
given by the participants were multiplied by a factor of five, the number of responses for
“slightly different” were multiplied by a factor of four, the number of responses for “different”
were multiplied by a factor or three, the number of responses for “more different” were
multiplied by a factor of two and the number of responses for “very different” were multiplied
by a factor of one. The scores for each weighting were than added up and a total score for the

corresponding DE-number resulted. Table 1 shows a typical result for a colour sample The DE-
values refers to DEab-values. Table 1 shows a typical ranking.
Table 1: Ratings and rankings of a colour, DEab-values
DE: 2.26 DE: 5.87 DE: 5.26 DE: 8.84

# of # of # of # of
Factor responses Score responses Score responses Score responses Score
Match: 5 6 30 0 0 3 15 0 0
Slighly different 4 10 40 1 4 10 40 0 0
Different 3 1 3 4 12 3 9 2 6
More Different 2 0 0 5 10 1 2 3 6
Very Different 1 0 0 6 6 0 0 12 12
Total Score: 73 32 66 24
These DEab-values were plotted against the total number and the r2-values were obtained. This
was done for the DE-values from all four equations and all colour samples used in this study. A
typical plot of this is shown in figure 3.
Figure 3: Example of a correlation between colour difference DEab and rating score
The r2-values for this sample for the other colour difference equations were 0.68 for DE94, 0.70
for DE2000 and 0.70 for DECMC respectively. The r2-value gives an indication on how well the
colour difference values calculated with the four colour difference equations correspond with
the observed visual differences. A value closer to 1 will mean a better correlation between
calculated and observed difference.
The r2-values from all colour samples and the various DE-equations were then plotted against
the L*, C* and h* values of all the colours evaluated in this study to see how each colour
difference equation performed. The evaluation of these graphs made it possible to conclude
which colour difference equation performs best within this study.

3 3. Results
3.1 Performance of the four difference equations against Lightness of the evaluated samples
The first evaluation of the colour difference equations is against the Lightness values of the
samples.
The r2-values obtained from the colour samples in relation to the four colour differencing
equations are plotted against the Lightness, Chroma and Hue values of the tested samples. The
spread of the r2-values makes it necessary to have a polynomial trend line present to give a
measure of performance for each of the colour differencing equations in respect to Lightness,
Chroma and Hue. The level of polynomial was set at 4 to reflect the varying correlations that
were obtained from the ranking of the human observers to each of the colour differening
equations. The results in regards to Lightness can be seen in figures 4 and 5.
1
0.9
r(ab)
0.8 r(cmc)
r(94)
0.7
r(2000)
Correlation
0.6
r(cmc)
0.5
r(2000)
0.4
0.3
0.2
0.1 r(94)
0
31.65
33.1
40.27
48.5
49.34
49.66
50.29
r(ab)
56.99
57.58
61.57
67.66
68.99
75.94
76.83
79.74
79.93
Lightness
80.61
89.37
Figure 4: Threedimensional plot of the r2-values in regards to Lightness
Figure 5 contains the same data points but only the fourth order polynomial curve is shown.
0.9 DE ab
DE94
DE2000
0.8
DE cmc
0.7
0.6
0.5
0.4
30 40 50 60 70 80 90 100
Lightness
Figure 5: Performance of all four colour differencing equations in relation to the Lightness
values of the tested colour swatches

This figure demonstrates that DE2000 and DECMC equations show similar behavior with regards
to lightness. Both equations improve their correlation with increasing Lightness of the samples.
This is important since small numerical differences in lighter colours are more noticeable than
the same numerical difference among darker colours. Interestingly the DEab equation shows a
decline in performance as the lightness values increase. This translates into the fact that at
lighter colours the DEab equation will generate colour difference numbers that do not correspond
well with the perceived visual difference. The DE94 equation shows also a similar behaviour in
relation to increasing lightness of the evaluated samples, but not as severe as the DEab equation.
3.2 Performance of the four difference equations against Chroma of the evaluated samples
The second evaluation of the colour difference equations is against the Chroma of the colour
samples. The graphs for this are shown in figures 6 and 7.
1
0.9
0.8
)r(ab
0.7
)r(94
Correlation
0.6
r(cmc) )r(200 0
0.5
)r(cmc
0.4
0.3
r( 2000)
0.2
0.1
0
r(94)
2.96
8.26
11.04
11.86
13.94
14.34
15.9
18.23
22.43
r(ab)
23.8
26.05
29.25
29.92
31.28
36.3
Chroma
43.63
45.32
56.31
Figure 6: Three dimensional plot of the r2-values in regards to Chroma
Figure 7 contains the same data points as figure 6 but only the fourth order polynomial curve is
shown.
0.9
DE ab
DE94
0.8
DE2000
DE cmc
0.7
0.6
0.5
0.4
0 10 20 30 40 50 60
Chroma
Figure 7: Performance of all four colour differencing equations in relation to the Chroma
values of tested colour swatches

From this figure it can be seen that DE2000 and DECMC equation show similar behavior
throughout the chroma range of the tested samples. The DECMC-equation shows a more coherent
correlation in the low chroma area. This is important since the human visual system is quite
sensitive with regards to small changes in this area.
In the high chroma area (>40) the DE2000 equation shows a more coherent relation to the
perceived differences then the DECMC-equation. Overall the DE2000 and the DECMC-equation
show a very similar behavior with a slight advantage for the DECMC-equation, since the r2-values
are in a more coherent band compared to the DE2000-equation.
The DEab equation shows the largest spread of data point throughout the entire chroma range of
the samples. With the increasing complexity of the colour difference equations the data points
are in a more coherent band.
3.3 Performance of the four difference equations against Hue of the evaluated samples
The third evaluation of the colour difference equations is against the hue angles of the colour
samples. The evaluation against the hue angles of the tested colour samples is important to
determine if which equation performs better at certain hue angles, especially between DE94 and
DE2000, since an extra term was introduced to the equation to weight the colour differences
according to the position of the tested colour in colour space. The graphs for this are shown in
figures 8 and 9.
1
0.9
0.8
0.7 )r(ab
Correlation
0.6 )r(94
r(c mc) )r(2000
0.5
0.4 )r(cmc
0.3
r(2000)
0.2
0.1
0
r(94)
1.77
11.83
31.41
71.72
77.99
98.5
105.29
130.5
149.43
r(ab)
173.29
181.1
195.92
234.48
243.38
276.15
317.86
Hue Angle
355.64
357.58
Figure 8: Threedimensional plot of the r2-values in regards to Hue Angle
Figure 9 contains the same data points as figure 8 but only the fourth order polynomial curve is
shown.

0.9
DE ab
0.8 DE94
DE2000
0.7
DE cmc
0.6
0.5
R Y G B
0.4
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360
Hue Angle
Figure 9: Performance of all four colour differencing equations in relation to the hue angle
values of tested colour swatches
From Figure 8 and 9 it can be seen that DEab, DE94 and the DECMC-equation show a negative
trend towards the blue-violet region of the hue angle chart. It needs to be noted that the
performance of the DE2000 equation does not show a large variation in relation to the hue
angle. It is clearly visible that the DE2000 curve (blue curve) shows a much better correlation
between the rating done by the human observers and the numerical expression of colour
differences using this equation. The improved performance of DE2000 vs. DE94 and DEab can
be attributed to the correctional term that was introduced in the DE2000 equation to improve its
performance in especially this region of the visible spectrum.
3.4 Overall performance indicators of the colour difference equations

In addition to the figures 4 – 9, which show how each of the colour difference equations
correlate to Lightness, Hue and Chroma of the evaluated samples, other means are necessary to
evaluate which of the four equations correlate the most with the perceived visual difference of
the 24 human observers used in this test. A measure for this is the average r2-value of the visual
rating scores and the standard deviation these r2-values for each one of the colour differencing
equations. The following table depicts this in more detail.
Table 2: Average r2-value, standard deviation and coefficient of variation for all four colour
difference equations
DEab DE94 DECMC DE2000

Average r2-value 0.818 0.851 0.877 0.885
Standard Deviation 0.161 0.124 0.108 0.104
Coefficient of 0.197 0.146 0.123 0.118
variation (Standard
deviation/mean)
The table 2 shows that the DE2000 equation has the lowest coefficient of variation of all four
colour difference equations and highest average r2-value. The coefficient of variation is defined
as the division of standard deviation by the mean value. It allows for a more valid comparison
between data sets with different mean values. The lower the number the lower is the dispersion

of the data. Since DE2000 has the lowest coefficient of variation it means that the numerical
colour difference values correlate quite well with the observed colour differences. From table 2
is can be seen that the data for the DE2000 equation has the lowest dispersion. In figure 10
these coefficient of variation values are shown in a bar graph.
0.25
0.2
0.15
0.1
0.05
0
DEab DE94 DEcmc DE2000
Colour difference equation
Figure 10: Colour difference equations vs. coefficient of variation values
3.5 Optimization of the DE2000 equation

The DE2000 equation contains three weighting factors kL for lightness, kC for chroma and kH for
hue angles. These values are set by default to a value of 1. This means that there is no emphasis
on the lightness (kL), chroma (kC) or hue (kH) of the measured colours. In the past attempts have
been made to see if the DE2000 equation can be optimized [15], [16]. Gilbert et al [15] found
that the DE2000 equation was only marginally better than the DECMC equation, while Melgosa
et al. [16] introduced another weighting function. This weighting function improved the
performance of the DE2000 equation, but the improvement was statistically not significant
enough. A recent study by Johnson & Green [17] done in this area found that an adjustment of
the parametric factors to the values of kL = 1.5, kC = 1 and kH = 0.5 in the DE2000 as well as the
DE94 equation improves the performance of these equations versus the DEab equation.
Luo et al [11] did a verification of DE2000 using industrial data and came to contradicting
results in regards to the weighting factors kL, kC and kH. In one case a ratio of 2:1:1 got better
results and in the other case a ratio of 1:1:1 corresponded better on how the human observers
perceived the colour differences of the samples. It needs to be said, that for the 2:1:1 ratio
untrained observers were used and in the case of the 1:1:1 ratio professional observers were
used.
These two studies were the starting point for the attempt to improve the r2-value between the
ranking done by the human observers and the numerical difference obtained using the DE2000
equation with the default values for weighting factors.
Based on the work done by Johnson & Green [17] and Luo [11] it seems that an emphasis on
the kL factor seemed to improve the correlation between observed colour differences and
numerically calculated differences.
The optimization of the weighting factors for DE2000 was done with the help of MAPLE
software. The maximum of the average r2-values was obtained using the Optimization Package

within MAPLE. The package uses built-in NAG (Numerical Algorithms Group) C library
routines that are well documented and widely used in the scientific community. In particular, the
SQP (Sequential quadratic programming) method was invoked using NLPSolve, which is a
function call within MAPLE to solve nonlinear programs. The numerical method, as well as the
NAG library itself is well documented [20].
The DE2000 equation was entered together with the rating scores obtained from the 24
observers. The criteria for the optimized weighting factors were to get an improved average r2-
value and a lower standard deviation. As a result the following weighting factors were found:
kL = 0.98
kC = 1.05
kH = 0.97
The average r2-value improved slightly from 0.8845 to 0.8861 and the standard deviation
improved slightly from 0.1045 to 0.1038. Therefore no real improvement was achieved. The
weighting factors are also quite similar to each other with no real emphasis on one or the other.
The MAPLE worksheet is listed in the appendix.
This attempt to improve the performance of DE200 equation resulted only in a very minor
improvement in regards to a reduced standard deviation of the r2-values. The average of the r2-
values could not be improved.
Melgosa et al [16] introduced a chroma-tolerance weighting function, which they have

borrowed from the LCD equation, to improve the performance of the DE2000 function [18].
Although performance was improved it was statistically not significant. Gilbert et al [15] did an
evaluation of colour difference formulae and found the correlation efficients for DE2000 were
only marginally higher than that for DECMC.
Based on the results and results found in literature only one instant gives concrete new
weighting factors for DE2000. Other results show only minor improvements that indicate the
use of the DE2000 equation with the default weighting factors of 1. Therefore the DE2000
formula in its current form is the best colour difference equation that is currently available.
4 Conclusions
As this study shows, there are vast differences in regards to the correlation between visually
observed colour differences and numerically calculated difference, depending on which colour
difference equation is used.
Consequently it is important to name the equation used to determine colour difference. The
evolution of colour difference equations demonstrates that with increasing equation complexity
the correlation between numerical and observed difference improves. The DE2000 equation
correlates slightly better with the observed colour differences than the DECMC equation. The r2-
value for the correlation between numerical and observed colour difference is 0.877 for the
DECMC equation and 0.883 for the DE2000 equation.
Both previously named equations correlate numerically better with the perceived colour
differences than the DEab-equation, which has an average r2-value of 0.818.
The attempt to improve the performance of the DE2000 equation by modifying the weighting
factors kL, kC and kH resulted in similar new weighting factors that gave only in a minor
improvement in regards to the average r2-value. This agrees with the majority of the work done
by other researchers in this area. Unless an internationally agreed set of new weighting factors
for DE2000 is available one should continue to use the default weighting factors of 1.

Acknowledgements:
This project was made possible through the generous support of the Faculty of
Communication and Design through a Project Grant.
We would like to acknowledge the support and help of Peter Roehrig, Print Technician at
the School of Graphic Communications Management, in preparing the printed colour
samples.
We would also like to thank Assistant Professor Chris Kular for the Ishihara Colour
blindness test plates.
We would like to thank Dr. Abhay Sharma, Chair of the School of Graphic
Communications Management at Ryerson University for his guidance and support.
We would also like to thank all the volunteers who helped us with this project.
References:
1 Luo R., 75th CIE Jubilee conference presentation, Ottawa, May 2006
2 CIE., Colourimetry, 2nd ed., CIE Publications No 15.2. Vienna: Central
Bureau of the CIE, 1986
3 Heidelberg Publication, Heidelberger Druckmaschinen AG, Colour and
Quality, p.81, 1995
4 Clarke F.J.J., McDonald R. and Rigg B., Modification to the JPC79 colour
differencing formula, J. Soc. Dyers and Col., 100: 128 – 132, 1984
5 Berns R.S., Altman D.H., Reniff L., Snyder G.D. and Balonen-Rosen M.R.,
Visual determination of suprathreshold colour-differences using probit
analysis, Col. Res. App., Vol .21, 459 – 472, 1991
6 CIE., Technical report: Industrial colour-difference evaluation. CIE
Publication 116, Vienna: CIE Central Bureau, 1995
7 Billmeyer Saltzman., Principles of Colour Technology, Berns, R.(Ed.), 3rd
edition, Wiley & Sons, New York, p. 114, 2000
8 CIE, Technical Report: Improvement to industrial colour-difference
evaluation, CIE Publication 142, Vienna: CIE Central Bureau, 2001
9 Luo M.R., Cui G. and Rigg B., The development of the CIE 2000 colour
difference formula, Col. Res. Appl. 26: 340 – 350, 2001
10 Sharma G., Wu W. and Dalal E.N., Col. Res. Appl., 2005, 30, 2005
11 Luo M.R, Minchew C., Kenyon P. and Cui G., Verification of CIEDE2000
using industrial data, AIC Colour and Paints, Interim Meeting of the
International Colour Association, Proceedings, 97 – 102, 2004
12 Johnson T., Green P., Issues of Colour Measurement and Assessment, TAGA
Proceedings, 518 – 534, 1991
13 Buering., Deutsche farbwissenschaftliche Gesellschaft e.V. , Presentation
2001, http://www.dfwg.de/doc/dfwg-homepage-419.htm , retrieved Feb 4, 2006
14 Basimir R., Costello G., Dibernardo A., DiPiazza J., Kuna D., Paulius K.,
Rybny C. and Zawackni W., A Comparison of Visual and
Spectrophotometric Evaluations of Paired Colour Prints, NPIRI Task Force on
Colour Measurement, Rochester, NY, TAGA Proceedings, pp. 558-578, 1995
15 Gilbert J.M., Daga J.M., Gilabert E.J., Valldeperas J. and the
Colourimetry Group of the Spanish Fastness, Evaluation of colour
difference formulae, Colour. Technol. 121, 147 – 152, 2005
16 Melgosa M., Huertas R., Yebra A. and Perez M.M., Are Chroma Tolerance
dependent on Hue-angle?, Col. Res. Appl., 29: 420 – 426, 2004

17 Johnson A. and Green P., The Colour Difference Formula CIEDE2000 and
its Performance with a Graphic Arts Data Set, TAGA Journal, vol. 2, pp. 59 –
71, 2006
18 Kim D.H., The influence of parametric effects on the appearance of small
colour differences, PhD thesis, University of Leeds, UK, 1997, Chapter 6
19 Kim D.H., Cho E.K. and Kim J.P., Evaluation of CIELAB-based colour
difference formualae using a new data set, Col. Res. Appl. 26, 369 – 375 2001
20 Numerical Algorithm Group,
http://www.nag.com/numeric/cl/cldocumentation.asp retrieved January 28,
2008
Appendix
The following table list the L*a*b*-values of the colour tested.
Table 4 L*a*b*-values of the tested samples
Sample # L* a* b*
1 49.66 36.17 -1.53
2 75.94 -4.40 29.44
3 50.43 20.20 23.56
4 61.68 -9.85 20.37
5 50.29 -17.20 -0.33
6 40.27 19.31 -17.47
7 76.81 8.24 28.76
8 39.65 6.51 -30.60
9 67.66 -21.34 17.77
10 50.03 12.36 2.59
11 68.99 -5.50 28.73
12 81.85 7.62 -7.99
13 79.74 2.58 13.19
14 79.93 0.96 -8.91
15 80.61 -7.32 8.57
16 71.12 9.08 0.28
17 89.37 -3.69 13.46
18 68.28 -15.29 -4.36
19 79.79 -3.98 -11.17
20 80.27 5.96 1.97
21 79.08 -14.09 2.64
22 76.83 13.42 40.63
23 57.54 -23.03 -5.99
24 57.58 -18.11 2.13
25 60.03 46.99 -8.19
26 48.50 -19.95 -39.01
27 49.65 38.68 23.62
28 49.34 -27.65 16.33
29 48.98 26.01 -1.53
30 41.47 -19.86 0.46
31 33.10 -2.60 -1.41
32 32.38 14.51 2.43
33 31.65 -10.32 -19.92

34 56.99 56.15 -4.28

35 61.57 -25.67 -35.96
Worksheet used in MAPLE software for the optimization of the weighting factors for DE2000
> restart
> # INPUT TEST COLOR DATA AND COMPUTE DE2000 value
> # M[i,j] will be an array where i is the number of data pts for a given test color (1 to 4), and j
is the number of the test colors (1 to 35)
> # DeltaLprime35 etc. are data files with the 35 values for the corresponding terms in the
DE2000 formula
>
> for j from 1 by 1 to 35 do
for i from 1 by 1 to 4 do
L[i, j] := op(1, fscanf("DeltaLprime35", "%f"));
C[i, j] := op(1, fscanf("DeltaCprime35", "%f"));
H[i, j] := op(1, fscanf("DeltaHprime35", "%f"));
RT[i, j] := op(1, fscanf("RT35", "%f"));
SL[i, j] := op(1, fscanf("SL35", "%f"));
SC[i, j] := op(1, fscanf("SC35", "%f"));
SH[i, j] := op(1, fscanf("SH35", "%f"))
end do
end do
>
DE2000[i, j] := sqrt(L[i, j]^2/(kL^2*SL[i, j]^2)+C[i, j]^2/(kC^2*SC[i, j]^2)+H[i,
j]^2/(kH^2*SH[i, j]^2)+RT[i, j]*C[i, j]*H[i, j]/(kC*SC[i, j]*kH*SH[i, j]))
end do
end do
>
> # INPUT RATER INTEREST SCORES FOR THE COLOR SAMPLES
RI[i, j] := op(1, fscanf("RaterI35", "%f"))
end do
end do
>
> # CALCULATE CORRELATION COEFFICIENT FOR EACH TEST COLOR RSQ[j],
j=1..35
> # n=number of samples (n= 4)
> n := 4
>
sdri[j] := (n*(sum(DE2000[k, j]*RI[k, j], k = 1 .. n))-(sum(DE2000[k, j], k = 1 ..
n))*(sum(RI[k, j], k = 1 .. n)))/(n*(n-1));
sdsquared[j] := (n*(sum(DE2000[k, j]^2, k = 1 .. n))-(sum(DE2000[k, j], k = 1 ..
n))^2)/(n*(n-1));
sRIsquared[j] := (n*(sum(RI[k, j]^2, k = 1 .. n))-(sum(RI[k, j], k = 1 ..
n))^2)/(n*(n-1));
RSQ[j] := sdri[j]*sdri[j]/(sdsquared[j]*sRIsquared[j])
end do
>
> # Compute average R^(2 )value, and variance

> RSQAve := 1/35*(sum(RSQ[k], k = 1 .. 35))

> varRSQ := 1/34*(sum(RSQ[k]^2, k = 1 .. 35))-1/1190*(sum(RSQ[k], k = 1 .. 35))^2
>
> with(Optimization)
> soln := NLPSolve(RSQAve, {.1 <= kH, .1 <= kC, .1 <= kL}, maximize, method = sqp)
soln := [0.886121262375657914, [kL = 0.980008961557212177, kC = 1.04917820804548900,
kH = 0.970995457163454034]]
> kCmax := op(2, op(1, op(2, soln)))
kCmax := 0.980008961557212177
> kLmax := op(2, op(2, op(2, soln)))
kLmax := 1.04917820804548900
> kHmax := op(2, op(3, op(2, soln)))
kHmax := 0.970995457163454034
> RSQAvemax := op(1, soln)
RSQAvemax := 0.886121262375657914
> stddevRSQmax := sqrt(subs(kC = kCmax, subs(kL = kLmax, subs(kH = kHmax, varRSQ))))
stddevRSQmax := 0.1037577101
> subs(kL = 1, subs(kH = 1, subs(kC = 1, RSQAve))); sqrt(subs(kL = 1, subs(kH = 1, subs(kC =
1, varRSQ))));
0.8845555344
0.1044760282

T. HARTUS 165
Effect of Toner Fixing Temperature on Print Properties in the

Electrophotographic Process
Timo Hartus
Department of Forest Products Technology

Paper and Printing Technology
Helsinki University of Technology
P.O. Box 6400
FI-02015 TKK
Finland
Email:Timo.Hartus@tkk.fi
Abstract
Fixing by fusing is the final stage in the electrophotographic printing process. It is an

important part of the printing process and greatly determines the final print quality.
Melting begins at the surfaces of the toner particles, which at first start to cohere and then
as melting progresses the filming or fusing toner starts to wet and adhere to the paper.
The print evenness and gloss are seen to develop simultaneously with the adhesion phase
as a function of spreading on the paper surface.
The aim of the work is to define the effect of fusing temperature on the development of
toner adhesion, print evenness and print gloss. To achieve this aim, a new test method has
been developed for the estimation of toner interaction in electrophotographic applications.
The results of this study show that the degree of toner adhesion greatly determines the
filming properties and hence the final evenness and gloss level of a printed surface.
Keywords: Adhesion, Electrophotography, Paper, Testing, Toner, Fusing
1 Introduction
The electrophotographic printing process generally includes six main steps, which are charging
of the paper surface, exposing, developing, transferring toner to paper, fusing on the paper and
cleaning of the photoreceptor [1]. Fusing is the final process in an electrophotographic printer or
copier, and it determines print quality properties, such as fixing strength, glossiness and density
of the image [1]. In the fusing step, toner is permanently fixed to paper by effectively melting
the toner onto the paper surface.
The toner particles become fixed on paper by heating them above their glass transition
temperature. In addition to heating, pressure can be used to enhance the fixing result [2]. During
heating, toner particles partially melt, predominantly on their surfaces, and thereby begin to bind
to each other and onto the paper.
Thermal changes in the toner during the fusing process can be divided into three stages [3]:
1. Warming (temperature increase of toner particles and paper)
2. Softening (melting of the toner begins from the particle surfaces and they start to cohere
and adhere to each other)
3. Melting (partly melted toner adheres to the paper)
Immediately after heating is removed, the melted toner starts to cool and the toner particles
become bonded to the paper. If the toner has not fused sufficiently, the adhesion to paper is
poor.

166 EFFECT OF TONER FIXING TEMPERATURE ON PRINT PROPERTIES …
Energy consumption during the fusing process can be split into several parts: the first is
consumed by the toner particles in initial surface melting leading to their cohering and adhering
to each other. In this stage, the toner changes from a powder to a highly viscous particulate
mass, reaching a maximum viscosity at a temperature ranging from 120 to 130 °C [3].
Subsequently, the remaining energy required is consumed to complete the melting, associated
with a lowering of the viscosity in turn allowing the toner to penetrate controllably into the
paper. The process, however, is subject to limitations. For example, a too high fusing
temperature has a negative effect on paper and print quality, even if the toner adhesion may
reach its maximum. The most significant negative effects are yellowing and curling of the paper
[3]. Following the fusing stage, the print temperature should decrease sufficiently so that the
toner hardens. If the fixing temperature is too high, however, the still warm and soft toner can
cause detrimental smearing. Furthermore, too high a drying temperature may cause too low
toner viscosity, leading to excessive spreading of toner in images, and an increased tendency for
slump, or flow, into the paper. Clearly, if the temperature applied is too low, i.e. below the glass
transition temperature of the toner, no fixing occurs [3]. Though the glass transition
temperatures of toners are typically more or less the same in practice, quite significant
differences in the melting temperature ranges of toners can be found [3].
Electrophotographic toners consist mainly of polymer resin and pigment [1]. In addition charge
control and flow agents and waxes may be used. Resin types used in toners include styrene-
acrylates, polyesters and epoxides. Toner pigments themselves do not have significant affinity
to a paper surface, thus the main factor determining the toner-paper interface is the toner
polymer phase, and indirectly, depending on polymer state and pigment concentration, the
viscosity of toner. The secondary factors are thermal conductivity, surface energy and toner pile
height [4,5].
Toner rheological properties must be optimized to allow for the correct balance of coalescence,
spreading and penetration into paper [1]. Toner thermal properties are a compromise between
their being soft enough for the fuser, but not too soft in order to prevent the toner particles from
being fused onto the photoreceptor at the development stage, or in the storage before use. In
addition, the paper must be optimally designed to have the necessary thermal response
characteristics to support the fusing and cooling cycle. Its structure must also be porous and
permeable enough for the toner to be able to partly penetrate into the paper. Thick paper grades
need a longer time to warm than thin papers with low surface porosity. This may cause
incomplete fixing of toner in the printing and copying process.
Both toners and papers combine to give a wide variety of melting rates according to the system
used and their individual properties. Although a toner with a small particle size and a more
spherical shape undoubtedly produces a higher quality print, it needs some additional heating
energy compared to the conventional toner in order to produce the same fixing level and print
gloss [6,7]. The real fusing temperature reached on the print can vary greatly in the printing
process. A large range of paper grades are commonly accommodated in a single process under
equal conditions. The printing quality, however, varies if the thermal properties of papers differ
too widely from each other.
The gloss of the printed image and the evenness of the gloss have a subjective effect on the print
quality. Although the legibility of extremely glossy prints decreases, high gloss is often
associated with high print quality [8].
The roughness of the paper greatly determines the roughness of the print [9]. The gloss of a
print is strongly depending on the underlying paper roughness [8]. Printing normally increases
the gloss level but it often increases the gloss variation [10]. Typically, the raising of fusing
temperature in electrophotography increases the gloss level of the final print [11].

T. HARTUS 167
The purpose of this study is to explain the effect of different paper grades on print quality and
fixing degree in the electrophotographic printing process. To achieve this, the effect of fixing
temperature on the quality properties of electrophotographic prints with various types of digital
printing papers is studied, and the results used to clarify the possibility of controlling the
electrophotographic print gloss by adjusting the fixing temperature. In this latter context, the
development of print gloss and roughness as a function of fixing temperature is also studied.
2 Experiments
All the prints, thermal fusing processes and measurements were carried out under standard
laboratory conditions (23 °C and 50 % RH).
2.1 Materials
Studied papers were: 80 g/m2 uncoated copy paper (1), 130 g/m2 coated art paper (2) and 200
g/m2 glossy coated art paper for digital printing (3). Papers were cut to a size of 40 mm x 200
mm for the experiments.
Two commercial black toners (A and B), which are commonly used in copy machines, were
examined in this study. The toners differed in shape and particle size. Toner A was a
conventional toner with the somewhat greater particle size than toner B, which in turn was
described as a chemical toner with a more spherical shape. According to the thermogravimetric
analyses, differential scanning calorimetric analyses and surface energy determination, however,
the thermal properties and surface chemical properties of both toner grades were virtually
identical, showing only some differences in the melting energy consumptions of the toners
[3,12,7]. They were thus assumed to be constituted of the same material, differing only in
particle size and shape, and hence having the same thermal and chemical properties. The impact
of particle size and shape could therefore be studied. The formulation and physical properties of
the studied toners are presented in Table 1.
Table 1: Formulation and property facts of the toners

Particulate Lubricational Glass Specific melt Surface
size wax transition energy energy (at 150
/ µm (on /% temperature / J/g °C), mJ/m2
average) / °C
Toner A 8.5 5 65.6 14.0 19.5
(platy)
Toner B 7.0 15 66.6 19.8 17.7
(spherical)
2.2 Printing
Toners were spread on the papers by a hand coater using a bar applicator, the application
procedure being the following:
A stencil board, thickness 0.5 mm, was set on the paper to be used as substrate. Toner was
spread over the paper by pulling a handcoater bar over the stencil board a number of times
so that an even toner layer was created. The print areas accepted for the analysis were
defined as having optically even full coated black surfaces. The target ink (toner) amount
applied after metering was approximately 2 g/m2, on a fully coated paper surface. Each
accepted black area was roughly 3 cm wide and at least 10 cm long.
The even thickness of the “printed” toner layer was visually controlled before fixing toner on
the substrate in the experiment. Only even, full colour, black area was accepted for testing.
These chosen samples were then further evaluated by measuring the optical density with a
Vipdens 2000 densitometer (Vipdens 2000, Viptronic GmbH) without a polarization filter. The

density of the prints was taken as a control of the amount of toner on the paper. Print density
was evaluated only on full printed areas.
2.3 Fixing
The toner was thermally fixed using IR-heaters. Two prints, one of each toner, were heated
together so that the different toners could be evaluated on the same paper substrate
simultaneously. The sheets were placed on a plywood carriage for transporting under the
heating unit. The heating unit consisted of two sections of IR heating elements (Elstein
Hochtemperaturstrahler HTS/1 250). The distance between the thermal elements, mounted
above the papers, and the print samples was 50 mm. Both the IR heating elements emitted 250
W each and generated a maximum surface temperature of 700 °C. The samples were kept for
various controlled times under the IR-heaters until the chosen target temperatures were reached.
Surface temperatures of prints were measured by a portable Dickson IR-temperature meter. The
measuring time of the meter was less than one second. The temperatures reached at the end of
the fixing stage were chosen to be 75, 100, 125, 150 and 200 °C, respectively.
2.4 Toner adhesion measurements

The toner fixing and film forming properties was characterized by a specially designed adhesion
measurement, using a print tack-meter, which is a special accessory mounted on a laboratory
printing device [13]. The instrument is illustrated in Figure 1. The thermally fixed prints were
taped on the printing sector. The measurement disc is a 20 mm wide rubber coated aluminium
roller. In this study, however, the measurement disc was pre-inked with pick test oil (IGT,
normal tack). The distribution of pick oil was made by a high speed inking device (speed 100
m/s, time 10 s). The disc was weighed before and after the tack oil distribution, so that the total
amount of pick oil on the measurement disc was determined, and found to be about 8 mg, which
corresponds to an amount of 2 g/m2 after application on the paper. The distance between the
disc (the upper position, when the disc is totally released from the print) and print on the
printing sector was constant; approximately 5 mm. The difference in the distance between the
disc and the printing sector has naturally a great effect on measured releasing time, because the
moving time of the measurement disc is included into the release time and that is why it has to
be constant to get comparable results.
The meter is commonly used to survey the change of print tack or setting and drying of offset
ink as a function of time after printing, and it records the release time of a measurement disc as
it is removed from a printed surface under constant force, the print substrate being taped onto
the printing sector. The release time of the measuring disc is termed the printack-value. More
commonly, tack measuring instruments record the release force needed to remove such a disc.
Comparison between time and force may allow greater resolution between the samples with
high evenness and high toner adhesion, or good surface strength, in general (the area of long
release times in printack) in further studies using the tack approach.
In general, several measurements are made from a print as a function of time to monitor drying
of the print. The number of printack-measurements from every print varied between 10 and 20.
In this study, the average printack-value of each measured test sheet was calculated from the
multiple individually measured values.

T. HARTUS 169
Figure 1: A Universal Testprinter laboratory printing device with Printack-meter.
2.5 Print gloss measurements

Print gloss was measured by a gloss meter (Vipgloss-1, Viptronic GmbH) after the thermal
fixing and cooling of the toner on paper or on microscopy glass slides. The device measures the
gloss value according to the standard DIN 16537, in which a 45 ° angle of incidence is in use.
Possibly this gloss measurement standard may have set samples some different relative order
than the other commonly used gloss standards would have done. The measured gloss values are
relative values between 0 (no gloss at all) and 10 (very high gloss). At first, the gloss meter was
calibrated onto the low gloss and onto the high gloss standard surfaces, after this the glosses of
the prints were measured. Five measurements were made from every sample, and from these
values the average values were calculated. The measurement area diameter is 3 mm, which is
the minor diameter of ellipsoidal illuminated area in the device. A relative small measurement
area allowed the gloss measurements from the samples on microscopy glass slides also.
2.6 Roughness of the print

Roughness (rms) values of toner films on microscopic glass slides were measured with an
atomic force microscope (AFM) using a tapping mode [14]. The basic mode of an AFM records

the forces between the sample surface and a fine tip mounted on a cantilever. The tapping mode
atomic force microscope (TM-AFM) is a development of the contact mode AFM. The
oscillation is driven by a constant driving force and the amplitude of its oscillation is monitored.
The tip touches the surface at the bottom of each oscillation, and this reduces the oscillation
amplitude of the cantilever. The feedback control loop of the system then maintains this new
amplitude constant as the oscillating, or tapping, tip scans the surface. This is accomplished by
the z component of the scanner, which changes the tip height to adjust exactly for the surface
topographic variations as the tip scans the sample surface. Measured areas were necessarily
fairly small, 5 μm x 5 μm.
For the practical AFM measurements, the toners were spread on microscopic glass slides by a
hand coater in a manner identical to that used for the paper sample preparation, and they were
likewise heated using the IR-heater. The measurements gave information only about the part of
roughness which is due to toner itself. The measurements were carried out in air atmosphere,
and only on those samples which were fixed well enough to avoid dusting in the sensitive
environment of the AFM (only samples, which were heated during fixing in the oven to over
100 °C were analyzed using this method). The sample surface tested had also to be fairly even
with no excessive variations in the surface height.
Printack-values measured from unfixed or partly fixed prints (low fusing temperatures) were
analogous with the friction meter generated values found in a previous study [3]. In the latter
approach an adhesive tape is applied to the fused image by controlled and repeated application
of pressure. The tape is then pulled from the paper under controlled force conditions, at a given
speed and angle according to ASTM D 1894 (Standard Test Method for Static and Kinetic
Coefficients of Friction of Plastic Film and Sheeting).
The printack-values decreased once the toner particles had started to adhere together. At the
second step, the printack-values started to increase once toner particles had begun to stick also
onto the paper surface. The adhesion to paper was seen to begin at a temperature range of 100 -
120 °C. This result correlates also with previous toner adhesion measurements [7]. Printack-
values increased until the surface has reached the final state of evenness (film roughness). The
final evenness of a surface was seen to depend mainly on the original roughness of the paper.
Once the fixing was completed, only minor changes occurred in the printack-value and these
only if the evenness of the print increased. The development of printack-values as a function of
fusing temperature of the prints is illustrated in Figure 2.

T. HARTUS 171
75
70 A1
Printack-value, ms
A2
65
A3
B1
60
B2
55 B3
50
60 80 100 150 200
Fusing temperature, °C
Figure 2: Printack value of prints using toners A (conventional toner) and B (chemical toner) as
a function of temperature on papers (1) – 80 g/m2 uncoated, (2) – 130 g/m2 coated and (3) –
200g/m2 glossy coated.
In the case of the non-fused toner samples (the samples which were heated below 60 °C); the
pick oil penetrated into the toner layer. The splitting occurred in the dry toner layer. The
penetration depth determined the splitting depth of the toner layer. A part of toner is removed
with the oil onto the disc and a part remains on the paper. The measured Printack-value is
determined mainly according the friction forces between the toner particles.
At somewhat higher temperatures, i.e. those samples which were heated to above the toner glass
transition temperature in the range from 80 to 100 °C, and hence some film forming had
occurred, the toner has been partly melted on its surface and begun also to fix onto the paper
surface. In this case, the pick oil can only partially penetrate into the toner layer. The splitting
occurs in the middle of the unmelted toner layer which is between the formed surface film of the
toner and the toner which is beginning to fix on the paper. This toner layer on the tack wheel
was thinner than in the case of totally non-fused toner surface, and this, together with the
smoother surface creating stronger oil interface forces due to the more even oil layer
distribution, may account for the slightly lower observed adhesion forces.
The samples which were heated over 100 °C were more completely melted, and a hard surface
film is seen to have formed, such that the pick oil cannot penetrate into the toner layer. The
splitting then occurs between oil and the toner surface. The toner film surface evenness
determines the measured printack-value in this case. No failures were observed in the toner
layer or paper surface itself. The more even surface means a closer, more uniform oil film
contact and hence the greater Printack-value.
Print gloss is, as already described, quite strongly dependent on the paper gloss [15]. The
roughness of paper greatly determined the maximum gloss level which could be reached. The
gloss of electrophotographic prints increased as the fixing temperature increased, especially in
the cases of thick ink layers, depending on the melting degree of the toner. The greater the
melting degree of toner, the higher the print gloss obtained. The print gloss was seen to develop
similarly to the printack-values as a function of temperature (Figure 3). First, we see that the
gloss-values are almost stable when the toner particles start to melt and adhere together. At the
second step, the surface of the toner layer being largely melted, the gloss-values start to increase
strongly and the print becomes more even, the toner particles also begin to stick onto the paper

surface. In this temperature range, approximately from 100 °C to 200 °C, the printack-values
show quite a clear correlation with the measured gloss values. The gloss increases with higher
fusing temperatures.
10
9
8 A1
7 A2
6
Gloss
A3
5
B1
4
3 B2
2 B3
1
0
60 80 100 150 200
Fusing Temperature, °C
Figure 3: Gloss (relative scale 0-10) of prints using toners. A (conventional toner) and B
(chemical toner) as a function of temperature on papers (1) – 80 g/m2 uncoated, ( 2) – 130 g/m2
coated and (3) – 200g/m2 glossy coated.
The difference in the gloss values between toner A and B is significant, as is their mechanism of
development. Toner B has greater roughness (Fig. 4), lower final gloss potential (Fig. 5), though
the gloss of toner B does rise earlier in the heating series, and lower printack-value (Fig. 2) at
every studied fusing temperature. Toner B, therefore, though having reduced gloss potential
compared to Toner A, does have the ability to generate equal gloss to Toner A over a range of
greater microroughness levels. These data support the hypothesis that Toner A, being of larger
particle size and a more flat shape, may be able to mask some of the underlying larger scale
paper roughness and so give a coverage effect. The limit in gloss, however, is apparently
determined by the fine-scale fusing structure and the degree of filming, which eventually is
inversely dependent on toner particle size.
The connection between print roughness and print gloss for the toner samples on the microscope
glass slides heated to the different fusing temperatures is seen in Figure 5. It can be clearly
observed that the toner shape and particle size have a significant effect on the development of
print roughness (film forming) and gloss under the various fusing conditions. When toner B is
used, higher fusing temperatures are needed to reach the same print evenness and gloss than
when using toner A, though, as observed in Figure 4, the gloss-roughness relationship
continually favours toner B.

T. HARTUS 173
45
40
35
Roughness rms, µm
30
25
20
15
10
0
100 150 200
Temperature, °C
Figure 4: Connection between roughness (filming) and fusing temperature of toner (◦ toner A
and ■ toner B) measured on microscopy glass slides.
Combining Figure 4 and Figure 5 we can see that print smoothness and print gloss develop with
increasing temperature for both the toners. Interestingly, toner A shows an initial range of gloss
following a trend with rms smoothness, however, the highest gloss is reached by an apparent
secondary effect, i.e. increasing gloss at equal roughness. It is once again suspected that the
gloss development for toner A is strongly related to the distribution of particles and polymer on
the paper surface, and that initial gloss may be related to physical coverage of the paper derived
from the larger toner particle size. Only if it is heated further, the gloss becomes secondarily
developed at the higher temperature rather than only being a function of rms roughness. On the
other hand, toner B shows a continuous relationship between smoothness and gloss and we can
interpret this function of heating as a direct melting, fusing and flowing of the toner.

10
9
8
7
Gloss value
6
5
4
3
2
1
0
0 10 20 30 40
Roughness rms, μm
Figure 5: Connection between roughness (filming) and gloss (◦ toner A, - conventional toner
and ■ toner B – chemical toner) of toner fused at different temperatures on microscopy glass
slides.
There may be further likely reasons for the reduced final gloss potential of Toner B, for example
it may be related to the greater amount of lubrication wax in the formulation in order to produce
good quality with fast performance in black and white printing, e.g. FAX application of
multifunction machines, and purposefully to remove the gloss from text to aid readability. Toner
B having smaller particle size and a more spherical shape and also some greater specific melting
energy consumption may have different thermal conductivity properties than Toner A. The
possible difference in the thermal conductivity of the unmelted toner powders may also be a
reason why toners A and B reached different gloss levels at the defined applied fusing
temperature. It is possible that Toner B did not reach as uniform a temperature distribution as
toner A during the limited heating time, and this could manifest itself as lower gloss level.
The connection between print gloss and printack-value is indicated in Figures 6 and 7 for the
different paper surfaces using each toner, respectively. The uncoated paper (1) resulted in the
lowest printack-value (the lowest adhesion) at given gloss for both the toners, reflecting the
impact of larger scale roughness on the pick oil layer distribution. The coated papers (2) and (3)
did not display any differences in the printack-value or print gloss.

T. HARTUS 175
10
9
8
7
6 A1
Gloss
5 A2
4 A3
3
2
1
0
50 60 70 80
Printack-value
Figure 6: Connection between print gloss and toner fixing degree at different fusing
temperatures with toner A ( conventional toner) on paper (1 – 80 g/m2 uncoated, 2 – 130 g/m2
coated and 3 – 200g/m2 glossy coated.
10
9
8
7
6 B1
Gloss
5 B2
4 B3
3
2
1
0
50 60 70 80
Printack-value
Figure 7: Connection between print gloss and toner fixing degree over a range of fusing
temperatures with toner B (chemical toner) on paper (1 – 80 g/m2 uncoated, 2 – 130 g/m2
coated and 3 – 200g/m2 glossy coated.
.

4 Conclusions
Adhesion measurement employing pick oil using a tack testing device provides a new way to
produce fairly easily and reproducibly a useful estimation of toner fixing degree and print
surface evenness. The Printack-device used in this work measures the release time of a disc
from the printed surface, instead of the release force needed to remove the disc as is measured in
most other such devices. Further study would be needed to determine if there are any relative
merits between the methods.
Incomplete fixing of toner on paper is generally caused by a too low fusing temperature,
provided the conditions for surface energy and adhesion are met. The final print evenness and
the print gloss clearly depend on the melting degree of toner at the fixing stage, in addition to
the original paper roughness. The gloss and evenness of the print are increased with increased
fusing temperature. High gloss of the print is obtained at high temperature after complete
melting of toner.
Though the glass transition softening temperatures of the studied toners were approximately
equal, there was quite a significant difference in the effective melting temperature ranges of
toners. It is obvious that the difference in the toner melting energy consumption determines
partially the final gloss and evenness of printed surface, and although the fusing temperatures as
measured were the same, the thermal energy distribution in the toners was likely to be different.
The differences in toner particle size and shape reflected these effective thermal fusing
properties, in that the coarser toner particle size acted to reduce print gloss though it achieved
greater smoothness by potentially improved paper coverage and hence greater gloss at lower
initial temperature. Higher temperatures were needed to bring the transition at constant
smoothness to the then higher gloss potential. Higher temperatures were needed for the finer
toner to achieve fixing and to manifest its glossing potential. Both the toner and the paper
properties, therefore, have an important effect on the apparent fixing temperature of the toner on
paper. In an electrophotographic printer there is usually no possibility to adjust fusing
temperature or time in the printer fusing unit, and so the right paper grades for predefined toner
types have to be selected to obtain optimized printing quality.
Acknowledgements
The author would like to thank Mr. Mikko Jokinen for AFM analyses of prints. The
author is also grateful to Professor P.A.C. Gane for his valuable comments during writing
this article.
References
1 Schein L.B., Electrophotography and Development Physics, 2nd Edition, Springer-

Verlag. New York, 1992
2 Oittinen P. and Saarelma H., Printing, Papermaking Science and Technology,
book 13, Fapet Oy, Helsinki, 1998
3 Hartus T., Adhesion of electrophotographic toner on paper, Graphic Art in Finland
31(1), 14-19, 2002
4 Britto I.L., An Evaluation of Factors that Control the Fixing of Toner to Paper in
Laserprinters, NIP 7 Vol 1. pp. 386-400, 2001
5 Sipi K., Formation of Quality and the Effect of Toner in Combined IR-fusing and
Hot Roller Fusing, M.Sc. thesis. Helsinki University of Technology, Espoo Finland
121 p, 1998
6 Heilmann J., Heikkilä I., Majava M. and Oittinen P., Print Quality in Hot Air
Fusing of Toners, IT&T´s Eleventh International Congress on Advances in Non-
Impact Printing Technologies. October 29 – November 3, Hilton Head, South
Carolina. 214-218, 1995

T. HARTUS 177
7 AL-Rubaiey H., Hartus T. and Oittinen P., The influence of Flash Fusing
Variables on Image Fixing Quality, Graphic Art in Finland 312, 7-10, 2002
8 Handbook of Print Media: Technologies and production methods Kipphan, H.
(ed.) Springer Berlin 2001 pp. 223-226, 2001
9 MacGregor M.A., Johansson P.-Å. and Béland M.-C., Measurement of small-
scale gloss variation in printed paper, Topology explains much of the variation for
one paper. Proceedings of the 1994 International Printing and Graphic Arts
Conference, 33-43, 1994
10 MacGregor M.A. and Johansson P.-Å., Gloss uniformity in coated paper;
Measurements of commercial papers, TAPPI Coating Conference Proceedings, 1991
11 Briggs J.C., Tse M.-K., Cavanaugh J. and Telep D.A., The Effect of Fusing on
Gloss in Electrophotography, IS&T NIP14 International Conference on Digital
Printing Technologies. October 18-23, Toronto, Canada, 1998
12 Hartus T. and AL-Rubaiey H., The Influence of Flash Fusing Variables on Image
Fixing Quality. Experimental Report, Helsinki University of Technology, Laboratory
of Media Technology. Otaniemi Finland 58 p, 2000
13 Testprint, Inc. Product Information, from http://www.testprint.com/ Retrieved
19.11.2005
14 Helsinki University of Technology, Laboratory of Paper and Printing
Technology, Research Equipments and services, from
http://www.tkk.fi/Yksikot/Paperi/laitteet.html/ Retrieved 15.1. 2006
15 Oittinen P., The limits of gloss in prints, Paperi ja Puu 65(11), 718–724, 1983

A. GIDLUND, T. MEJTOFT, S. DEMNERT. 179
Significance of Print Quality in Variable Data Printing

Åsa Gidlund*, Thomas Mejtoft** & Sofia Demnert***
**
*STFI-Packforsk STFI-Packforsk
Linköping University Royal Institute of Technology (KTH)
thomas.mejtoft@stfi.se
***
Royal Institute of Technology (KTH)
Abstract
Print quality and variable data printing are two important topics when discussing digital
printing and its ability to be a commercially successful printing technology. The aim of
this paper is to determine the significance of print quality when using variable data
printing to customize printed matters. In this study three levels of colour print quality and
three levels of personalization; entirely static, personalized with name, and personalized
with name, text and image, were established. Sample material was obtained by creating
fictitious advertisements in the personalization levels, which were then printed at the print
quality levels. A panel of respondents was asked to arrange the fictitious advertisements
in order of preference. At all print quality levels, personalizing with only name led to a
rather small increase in respondent preference compared to the static advertisements.
However, personalizing with name, text and image led to a significant increase in
preference compared to a static or name personalized advertisements. Hence, to attract
the respondents a higher level of personalization should be used. The respondents were
more forgiving towards low print quality when the advertising information was
personalized. Furthermore, personalization contributes more at lower print quality levels,
indicating that print quality is more significant in static printed matters or printed matters
with a low level of personalization.
Keywords: Digital printing, Print quality, Variable data printing
1 Introduction
Before the 20th century, printing was the only way of spreading marketing messages and
information to customers when referring to non-personal communication channels [1]. During
the later half of the 20th century many new, non-printed, channels for reaching a mass audience
have been introduced. Communication media like radio, television and the Internet have
changed the way that information and marketing messages are pushed and pulled to and from
customers. The new channels, with the global spreading of the Internet leading the way, have
however made print just an output channel among others. Until professional digital printing was
introduced in the beginning of the 1990’s, printing was only possible to use when trying to
reach a mass audience with static messages. Digital printing, in contrast to conventional printing
technologies does not require a static printing plate (e.g. [2,3]), and is therefore suitable for
producing mass customized [1] printed advertising material in e.g. one-to-one marketing [4].
Customized printing is based on the “combination of variable information with output devices
that do not require intermediate films or plates” [3] Using digital printing for customization to
make documents more effective as a business tool is a major economic driving force identified
by the non-profit digital printing industry consortium PODi (e.g. [5,6]).
Variable data printing can be used to gain business success since customized advertising
material can make a significant increase in response rate compared to static advertisements [7].
However, earlier studies indicate that the use of variable data printing is in general low among

180 SIGNIFICANCE OF PRINT QUALITY IN VARIABLE DATA PRINTING
printing houses [8] even though the technology has been available for over ten years. During
this period the print quality of digitally printed matters has steadily improved and is today
classified as good enough for many applications [2]. Print quality is a general measure of the
success of a colour printing system and is an important customer requirement, along with other
requirements such as cost, productivity, connectivity and reliability [9]. Furthermore, print
quality is one of customer’s top considerations when purchasing printed matters [10]. The
digital printing equipment available today with the capacity to produce variable data prints
range from inexpensive office copy-printers to high-speed production equipment, which also
means large variations in print quality. Furthermore, the importance of colour has been stressed
in earlier research [7]. The response rate from a mailing campaign increased by 46% when
colour was added to a static advertisement. For a customized advertisement the increase for
colour was 167%.
Print quality and variable data printing are two important topics when discussing digital printing
and its ability to be a commercially successful printing technology. The aim of this paper is to
determine the significance of print quality when using variable data printing to customize
printed matters. The following research questions have been formulated to investigate the
problem:
• What effect do different types of personalization have on the respondents’ preference

at different print quality levels?
• Is colour of any significance for the experience of personalized advertisements?
2 Material and Method
2.1 Research Approach

In spring 2006, 28 respondents participated in a significance evaluation of print quality and
variable data printing. The group of respondents was selected to be equally distributed over age
and gender. In this study, three levels of colour print quality and three levels of personalization
were defined (Figure 1). The print quality levels were established by technical measurements
and visual assessments. Fictitious advertisements were created at the three personalization
levels. To adapt the advertisements to the panel of respondents, each respondent completed a
questionnaire with information forming the basis for the personalization. The advertisements
were then printed in the three colour print quality levels. In all, nine colour samples were
obtained. In addition, three black and white samples were printed in the medium print quality
level, one for each personalization level. An evaluation was conducted to determine the
significance of print quality in personalized printed matter. The respondents were asked to
arrange the fictitious advertisements with different levels of print quality and personalization
levels in order of preference.

Quality level
verification
Technical
measurements
Visual
assessment
Sample
Quality production
levels
Printer 1
QBW
QL Printer 2
QM
QH Printer 3 Twelve samples
for each respondent
Personalization
levels Panel of
respondents
Significance
P1
evaluation
P2
P3
Order of
preference
1.
Questionnaire
Visual
assessment
Analysis
Figure 1: Schematic overview of the research process.
2.2 Personalization and Evaluation Material Design

The questionnaire, which formed the basis of the personalization of the test material, concerned
travel and specifically the respondents’ opinion of travel. The respondents were asked whether
he/she preferred to travel in Sweden or abroad, if he/she preferred metropolises, seaside resorts
or skiing resorts, with whom the respondent preferred to travel and how important high hotel
standard, reasonable prices and the destination itself were when choosing destination. The
respondents were also asked to grade how much they enjoyed different activities (e.g. golf, art,
tourist attractions etc.) using a five-point Likert Scale, spanning from very uninterested to very
interested [11]. The questionnaire also contained a section with background data e.g. age and
gender.

DREAMTRAVEL
Bästa Åsa Gidlund! 08-652 52 00
DREAMTRAVEL
Vi vet att människors drömmar ser olika ut. Därför har vi på DreamTravel
Bästa Åsa Nordin! 08-652
arrangerat resor, 52Sverige
både inom 00 och till andra spännande länderDREAMTRAVEL
över
hela världen, i snart 50 år. I vårt program finns många intressanta resor,
både för Dig Bästaen Thomas
semester Mejtoft!
08-652 52 00
Vi vet att människors drömmar ser olika ut. Därför harsom
vi pådrömmer
DreamTravel om utomlands och för Dig som
arrangerat resor, både inom Sverige och till hellre upptäcker
andra spännandevad Sverige har att erbjuda.
länder över
hela världen, i snart 50 år. I vårt program finns många intressanta resor, Vi vet att människors drömmar ser olika ut. Därför har vi på DreamTravel
både för Dig som drömmer om en semester utomlands arrangerat resor, både inom Sverige och till andra spännande länder över
Just nu har vi ettoch för Dig som
oemotståndligt helaerbjudande till Dig!
världen, i snart 50 år. I vårt program finns många intressanta resor,
hellre upptäcker vad Sverige har att erbjuda.
både för Dig som drömmer om en semester utomlands och för Dig som
I Nice kombinerar Du lata sol- och baddagar
hellre medvad
upptäcker storstadssemesterns shopping
Sverige har att erbjuda.
och aktiviteter.
Just nu har vi ett oemotståndligt erbjudande till Dig! Här i staden, som är franska rivierans största turistort, är man
mycket noga med att bevara sitt historiska arv. Gamla byggnader hålls efter,
Kultur, äventyr och några av världens allraoch de ockrafärgade
främsta sevärdheter väggarna
väntar Dig tvättas
Justi nu rena
har vioch
ettkonserveras.
oemotståndligt Härerbjudande
lyckas mantill Dig!
förenaavsvunnen
Peking, huvudstaden som ligger i norra utkanten skönhet med
den nordkinesiska en blommande turistindustri. Hotell Vendante är
slätten.
Mest imponerande är förstås Kinesiska Muren, ett trevligt
sommellanklasshotell
nästan 700 milinrymt Beläget
lång i eni Indiska
vacker byggnad
oceanenfrån
ligger1800-talet.
vackra Mauritius, som lockar med fantastiska
sträcker sig från Gula havet i öster till sin sista utpost i Gobiöknen. I Peking orörda stränder, korallrev och färggranna fiskar. På trestjärniga Hotell Bretin,
finns så mycket att upptäcka. Hotellet Gloria 3 nätter
är avfrån 3.510 :- vackert beläget mellan de smaragdgröna bergen och det turkosa havet, har
turistklass.
Priset gäller del i dubbelrum med kontinental frukost.
Du nära till både stranden och den vackra kolonialstaden. För Dig som gillar
7 dagar från 8.680 :- vandring har även de mäktiga bergen mycket att erbjuda.
Priset gäller del i dubbelrum med frukostbuffé.
Uppfyll Din dröm - ring oss redan idag!
1 vecka från 11.995 :-
RESEDRÖMMAR?
Uppfyll Din dröm - ring oss redan idag! Priset gäller del i dubbelrum med halvpension.
RESEDRÖMMAR?
RESEDRÖMMAR?
Uppfyll Din dröm - ring oss redan idag!
RESEDRÖMMAR?
RESEDRÖMMAR?
RESEDRÖMMAR?
Figure 2: Example of fictitious advertisements used as test material in the significance
evaluation.
All advertisements were designed as an offer for a trip to a specific destination and hotel. The
advertisements were created in 210 x 148 mm (A5) and consisted of a logotype, a catch phrase,
an image and a text area (Figure 2). All advertisements were in Swedish since all respondents
were native speaking Swedes. As in an ordinary advertisement, the variable images in this study
were mainly selected to give the respondent a feeling of the destination. Three levels of
personalization were determined and denoted level 1, level 2 and level 3. The level 1
advertisements were entirely static and hence all respondents were given the same
advertisement. In these advertisements the respondents were addressed as “Dear Traveller”. The
level 2 advertisements were the same as level 1 and hence had the same offer and the same
image. However, in these advertisements the respondents were addressed with their own name
(e.g. “Dear Thomas Mejtoft”). In the level 3 advertisements, the offer was adapted to each
respondent in terms of destination, hotel standard and hotel location. The choice of image was
based on the destination or the preferred activity while on vacation. The hotel standard and the
vicinity of the hotel were based on the respondent’s answer on the importance of the hotel
standard and the preferred activity while on vacation respectively. This resulted in the use of
nine different images. Further, the respondent was addressed by name in the same manner as the
level 2 personalization.
2.3 Defining and Verifying Print Quality Levels

To create and define different levels of print quality a test form was printed on a number of
different paper grades in a number of different printers. Based on a visual evaluation performed
by the authors, three paper/printer combinations were chosen to represent the three levels of
colour print quality. The levels were denoted low, medium and high. To verify the accuracy and
distribution of these three colour print quality levels a more comprehensive evaluation was
performed in terms of technical measurements and a visual assessment with a panel. A test form
consisting of a test chart for technical measurements and images for visual evaluation was
printed in all three printers. The test chart contained colour patches for measurements of print
gloss, print mottle, sharpness, and colour gamut. The test form was printed on the chosen paper
grade in the three printers.
The colour gamut volume was calculated on the basis of CIELAB-values from printed areas of
full tone cyan, magenta, yellow, red, green, blue, black and paper white. The CIELAB-values
were measured with a GretagMacbeth Spectroscan spectrophotometer. The settings used were
D50, two degrees and the No filter. The test areas for print mottle and sharpness were scanned at
300 ppi using an Epson Expression 10000 XL desktop scanner. The measures of these quality

factors were calculated using Matlab® image analysis routines, which are developed to
correlate with the visual appraisal. Print mottle was measured from areas of full tone cyan and
full tone green. Sharpness was measured from two regions; a black vertical and horizontal line
on white background and a black vertical and horizontal line on a yellow background. Hence,
the sharpness metrics black raggedness vertical, black raggedness horizontal, colour
raggedness vertical and colour raggedness horizontal could be calculated. Print gloss
measurements were performed with a Zehntner glossmeter ZLR 1050M. Measurements were
done on a full-tone black area and on a full-tone green area at an angle of 75º, according to the
standard [12].
As mentioned earlier, a visual assessment was conducted to establish the accuracy of the print
quality levels used in the study. The outcome of a visual assessment depends on the motif used
[13]. Further, an image evaluation will be more useful if several types of images are used [10].
To avoid any influence from the different images included in the study, this assessment
involved all nine images used in the advertisements at each of the three colour print quality
levels, in total 27 samples. The samples were presented to a panel of 14 experienced print
quality observers. The images were presented three at the time; one image printed in each print
quality level. The method used was category scaling [10]. Each observer was asked to express
their opinion of the general print quality of each sample on a 10-point numeric rating, where a
larger number implied better print quality. The assessment was performed in a standard daylight
viewing illumination, D50. The result was presented as the mean value for each print quality
level.
Earlier research [14] has shown that experienced print quality observers and inexperienced
respondents might judge print quality differently. Therefore, another visual assessment was
performed, this time using the respondents participating in the significance evaluation of the
advertisements. Hence, it could be investigated if this difference could affect the results of this
study. The samples used were the level 1 advertisements printed in the three colour print quality
levels, i.e. the most frequent advertisement in the significance evaluation. The assessment was
performed using the same method and under the same conditions as the assessment with the
experienced observers. The data regarding each respondents view on the print quality levels also
gave another dimension in the analysis of each respondent’s answer in the significance
evaluation. This data made it possible to adjust the results and base the three colour print quality
levels on how each individual respondent perceive the print quality at the different levels. That
is, if a respondent perceived the established medium level quality to be the best, the ranking of
this sample was adjusted to be used in calculating the mean value of the high print quality level.
Furthermore, this assessment enabled a comparison between the experienced print quality
observers and the inexperienced respondents regarding appraisal of print quality.
2.4 Sample Production

The three colour print quality levels were all produced by using the same paper grade in
different printers. To produce the samples, the three levels of personalization were then printed
at the three colours print quality levels. In addition, the three levels of personalization were
printed in black and white on the paper/printer combination producing the print quality level
denoted medium. Hence, a total of twelve types of samples were produced, nine colour and
three black and white. The encoding of the samples is presented in Table 1.

Table 1: Encodings of levels of personalization and print quality.

Personalization
Level 1 Level 2 Level 3
(Static) (Name) (Name, Text
& Image)
Black & White P1-QBW P2-QBW P3-QBW

Low Color P1-QL P2-QL P3-QL
Medium Color P1-QM P2-QM P3-QM
High Color P1-QH P2-QH P3-QH
For each type of sample a unique set was produced for each respondent and all samples of
personalization level 2 and level 3 were conformed to the specific respondent based on the
answers obtained from the questionnaire.
Even though the static sample could be produced using conventional printing technology, all
samples were printed in digital printing to ensure consistency of print quality between the three
levels of personalization. In general, conventional printing is regarded to give slightly higher
print quality than digital printing. This implies that the static advertisements could potentially
get a higher print quality if printed using conventional printing technology. However, this
quality difference is constantly changing as the quality of digital printing is improving. As the
purpose of this study is not to compare digital and conventional printing the potential quality
differences between the two printing technologies will be disregarded.
Since all respondents’ sets of samples were unique, there was a risk of copy to copy variation in
the production of the samples. Nevertheless, to ensure the accuracy between the different copies
and ensure a minimal variation between the different sets, all samples were visually investigated
after production. The samples were found to be visually indistinguishable and hence any
possible copy to copy variation should not affect the results of this study.
2.5 Significance Evaluation

Rank order [10] was the method used in the significance evaluation of the advertisements. This
method is suitable when having few samples that are easy to mix up and when the effort from
the respondents should not be too demanding. The evaluation was carried out by the 28
respondents. Each respondent was presented to a unique set of all 12 samples. The respondents
were asked to view the advertisements one at the time – look at the image and read the text –
and decide how appealing each sample was to them, i.e. how well each sample caught their
attention. The respondents were then asked to sort the samples in order of preference. The order
of precedence was translated to ranking points. The best ranked sample got twelve points; the
second best got eleven points and so on down to the lowest ranking sample that got one point.
The evaluation was performed in a standard daylight viewing illumination, D50. Each
respondent performed a colour vision deficiency test before the evaluation. This test contained
two samples from the Ishihara test plates, designed to separate the persons with colour defects
from those with normal colour appreciation [15]. No respondents showed any tendencies
towards colour defects.
3 Results
3.1 Print Quality Levels

The accuracy of the defined print quality levels was confirmed both by the technical
measurements (Table and Table ) and the visual assessment by the experienced observers. The
high quality samples had larger colour gamut volume and higher print gloss than the medium
quality samples. No significant differences between these two print quality levels were found in
print mottle or sharpness. The medium quality sample had larger gamut volume and lower
colour raggedness, i.e. better sharpness, than the low quality sample.

Table 2: Technical measurements of colour gamut volume, print mottle and print gloss.
Colour Gamut Print Mottle Print Gloss
Volume Cyan Green Black Green
QL 219050 0.06 0.25 67.08 68.85
QM 308330 0.03 0.14 32.57 42.73
QH 417910 0.02 0.25 97.80 99.10
Table 3: Technical measurements of sharpness.

Black raggedness Colour raggedness
Horizontal Vertical Horizontal Vertical
QL 5.51 18.75 44.83 38.07
QM 5.78 5.26 5.63 4.96
QH 7.66 6.18 7.66 5.73
The visual assessment of all the nine images in the three colour print quality levels, performed
by the experienced observers, showed that the print quality levels were accurately defined
(Figure 3), regarding both order and distribution.
Print Quality Levels

10
9
8
7
6
5 5.5
4
4.3
3
2 2.9
1
0
QL QM QH
Figure 3: The results from the visual assessment of all the nine images in the three colour print
quality levels (with 95% confidence intervals), performed by the experienced observers.
The visual assessment performed by the inexperienced respondents, indicated that they also
found the print quality levels as accurately separated. Hence, in general both the experienced
and the inexperienced respondents assessed the print quality levels similar. However, as
expected, the individual observers had different views on the range, order and distribution of the
print quality levels.
3.2 Outcome of Significance Evaluation

The results from the significance evaluation were cross-referenced to the background data
collected from each respondent. It could be concluded that there were no considerable
differences in opinion based on gender or age.
In the significance evaluation, the respondents put the samples in order of precedence. The
significance evaluation showed that the samples with level 3 personalization were, at all three
colour print quality levels, preferred over the level 1 and level 2 personalization samples (Figure
4 and Table 2).

12
10
8
6
4
P3
2
0 P2
QBW
QL
QM P1
QH
Figure 4: Mean value according to print quality
level and personalization level.
Table 2: Mean value according to print quality

Personalization
& Image)
Black & White 2.82 3.04 6.18

Low Color 3.46 3.68 8.79
Medium Color 7.39 7.86 10.11
High Color 6.86 7.14 10.68
Basically print quality is something that is perceived by the customer rather than objectively
determined by a printing house, printing press manufacturer or an expert panel. As mentioned
earlier, the visual assessment performed by the respondents indicated that there were differences
in how the different respondents appraised the print quality. Adjustments were made to the data
from the significance evaluation using the outcome of the visual assessment performed by the
respondents. This made small but noticeable corrections to the results (Figure 5).
12
10
8
6
4
P3
2
0 P2
QBW
QL
QM P1
QH
Figure 5: Adjusted mean value according to print quality

Table 3: Adjusted mean value according to print quality

Personalization
& Image)
Black & White 2.82 3.04 6.18

Low Color 3.43 3.64 8.93
Medium Color 7.00 7.54 10.00
High Color 7.29 7.50 10.64
It could be noted in Figure 5 that the dip in Figure 4 in grading between medium and high
quality for level 1 and level 2 personalization evened out when the results were adjusted to
conform with how each respondent perceived the print quality.
4 Discussion
Defining accurate and relevant print quality levels was a difficult task since print quality is a
subjective matter. It was also hard to analyze the exact print quality experience of each
respondent. However, the results from the general visual assessment with experienced print
quality observers indicated that the print quality levels were relevant. Adjusting the results from
the significance evaluation and basing the three colour print quality levels on how each
individual respondent perceived the print quality of each different level further limited the
influence of any differences in perceived print quality among the respondents.
The influence of the choice of personalization levels, type of advertisement, layout etc. can not
fully be determined from this investigation. However, the use of a general subject like travel, a
basic layout and three distinct personalization levels would give a good indication of the general
outcome of an analysis of the significance of print quality in variable data printing.
As mentioned earlier, digital printing and conventional printing differ in their ability to produce
customized printed matters. While digital printing can be used for both static and variable data
printing, conventional printing technologies, such as offset, can only be used for printing static
matters. If cost and speed is taken into account conventional printing technologies are often
used for longer run lengths and digital printing for shorter run lengths or variable data printing
since the production cost for digital printing is higher than conventional printing for longer runs
(e.g.[2,16]).
Personalization
& Image)
Conventional Printing
Technologies
Digital Printing
Technologies
Figure 6: Possible printing technologies to use at different levels of personalization.
Translating this study into a commercial perspective, means that when moving from a level 1
personalization (static) to a level 2 personalization (name), a change in printing technology has
to take place. Accordingly, a level 3 personalization (name, text & image) also has to be printed
using digital printing, since some information is personalized (Figure 6). However, it should be

noted that in this study all samples have been printed in digital printing to ensure consistency of
print quality between the three levels of personalization.
The mean value over all print quality levels of a level 2 personalization does not differ
substantially from using a level 1 personalization, leading to an increase of 5.8% in appraisal by
the respondents. Further, adding personalized text and images (level 3) improves the appraisal
by another 65%. As mentioned before, customizing some part of a print requires digital printing.
Taking into account the higher production cost of digitally printed advertisements (if a long total
run length is presumed), the motivation to personalize by using name only is in fact low.
Regarding print quality, it is possible to distinguish a mean increase of 33% in appraisal when
using colour at the low print quality level instead of black and white printing. However, a
further increase of 53% is noticed when increasing the print quality from the low level to the
medium level. When increasing the print quality even more there was only a 3.6% increase in
appraisal.
Looking at personalization level 3, the low quality samples differed only slightly from the
medium and high quality samples (8.9 vs. 10.0 and 10.6). However, at level 1 and level 2
personalization, the low quality colour samples clearly differed from the other two levels
(level 1: 3.4 vs. 7.0 and 7.3; level 2: 3.6 vs. 7.5 and 7.5). This indicated that the respondents
were more forgiving towards low print quality when the advertising information was
personalized.
12
10.6
10.0
10
8.9
8 7.5 7.3 7.5

7.0
6.2 P1
6 P2
P3
4 3.4 3.6
2.8 3.0
0
QBW QL QM QH
Figure 7: Adjusted mean value, sorted according to print quality level.

12
10.6
10.0
10
8.9
8 7.5 7.5
7.3
7.0
QBW
6.2
QL
6
QM
QH
4 3.4 3.6
2.8 3.0
0
P1 P2 P3
Figure 8: Adjusted mean value, sorted according to personalization level
The black and white samples showed the same tendency as the colour samples, i.e. a slight
increase in preference between the level 1 and level 2 personalization, but a considerable
increase between level 2 and level 3 (Figure 7). At each personalization level the colour samples
were more appealing to the respondents than the black and white samples (Figure 8). Further, a
level 3 personalization black & white advertisement is less appealing than both level 1 and
level 2 personalization colour samples at high and medium print quality in colour. However, as
noted in Figure 7 and Figure 8 the respondents believed that the level 3 personalized black and
white sample were more appealing than the level 1 and level 2 personalization of the low and
medium print quality in colour. This means that if the cost of colour printing, in comparison
with black and white, is much higher than the cost of personalization, a black and white printed
matter with high personalization could be preferred.
This research study is just a brief introduction to investigating the interaction between variable
data printing and print quality. Nevertheless, in this investigation the importance of specific
images, subject and layout of the advertisements have become noticeable. Future research must
take in consideration these factors by extending the investigation even further. For example, the
choice of images could affect the results not only by the motif but also by the balance and
composition of the colours in the images. That is, an image with highlights would most likely
print differently than a highly saturated image. Furthermore, the selection of the image on the
static sample is important, since this image is the one that most samples had printed. In this
study this was the image that the correction was based on, which raises the validity of the
investigation.
The levels of personalization could be divided into more unique levels to separate the effect of
personalized text and images and investigate these two individually. This is especially important
since the results from this study show that personalization with name only (level 1) does not
make any significant change in appraisal from a static sample. Could appraisal be raised even
further by for example removing the name from level 3 personalization, and making the
personalization more “invisible” to the respondent? This could also be studied with for example
eye-tracking equipment to investigate how the respondents observe each sample before making
a decision of order or preference.

5 Conclusions
With the type of printed advertisements, print quality levels and personalization levels used
here, personalizing with only name led to a rather small increase in respondent preference
compared to the static advertisement. However, personalizing with name, text and image led to
a significant increase in preference compared to a static or name personalized advertisements.
Hence, to attract the respondents a higher level of personalization should be used.
The respondents were more forgiving towards low print quality when the advertising
information was personalized. Furthermore, personalization contributes more at lower print
quality levels, indicating that print quality is more significant in static printed matters or printed
matters with a low level of personalization.
In general, colour prints seemed to be preferred over black and white prints. No differences
were found between the three sets of colour advertisements and the black and white
advertisements regarding the significance of print quality and personalization. The highest
personalized black and white sample was more appealing than the lower personalization levels
of the low and medium print quality in colour. Consequently, a black and white printed matter
with high personalization could be preferred if the cost of black and white printing is much
lower than colour printing.
The conclusions in this study were consistent with previous research regarding both the
effectiveness of variable data printing (e.g. [5,7]) and the importance of print quality [10].
However, this study was inconclusive with earlier research that suggests a considerable increase
in the response rate when using name personalization instead of static advertisements [7].
Instead, the results from this study indicated an insignificant difference between static and only
name personalized advertisements. Future research should consequently include a fourth
personalisation level in which the text is personalised but without the name personalisation to
clarify the significance of text personalization.
In contrast to other studies (e.g. [7]) this study has focused on the respondents experience in a
qualitative study rather than actual response rates in a quantitative study. This illustrates that it
is possible to get knowledge on the importance of print quality in relation of variable data
printing in a time and cost effective study.
This study further advances the scientific knowledge by introducing the print quality concept in
the discussion of variable data printing. The two concepts are individually important in both
research and the industry and this study gives knowledge about where resources should be
focused to make a printed matter more effective in catching the attention of the customer.
Acknowledgements
The authors would like to dedicate a special thanks to all the respondents in the
significance evaluation and the observers in the visual assessment for their participation.
We would also like to thank our supervisors Prof. Nils Enlund, KTH, Prof. Björn Kruse,
Linköping University, Dr Marianne Klaman, STFI-Packforsk, and Dr. Per-Åke
Johansson, STFI-Packforsk for their valuable comments. The Kempe Foundations, the
EU Structural Fund, the Swedish printing research program T2F, and the industry
participants in the DigiPrint project are gratefully acknowledged for their financial
support.
The authors are grateful for the scholarship from Tryckeriföreningens stiftelse för
utbildning och forskning (TUF), that made possible to present this research at the 2007
Taga Conference in Pittsburgh, USA.

References
1 Kotler P., Wong V., Saunders J. and Armstrong G., Principles of Marketing,
Fourth European Edition, Pearson Education Limited, 2005
2 Kipphan H., Handbook of Print Media, Technologies and Production Methods,
Springer, 2001
3 Romano F.J., Lee B., Rodrigues A. and Sankarshanan., Professional prepress,
printing, and publishing, Prentice-Hall, 1999
4 Peppers D. and Rogers M., The One to One Future, Currency Doubleday,1993
5 PODi., Best Practices in Digital Printing, Third Edition, Caslon & Company, January
2003
6 PODi., Best Practices in Digital Print, Sixth Edition, Caslon & Company, 2006
7 Broudy D. and Romano F., An Investigation: Direct mail responses, Based of
color, personalization, database, and other factors. Digital Printing Council, White
Paper, Retrieved July 5, 2005, from
http://www.gain.net/PIA_GATF/PDF/romano1.pdf, 1999
8 Mejtoft T., Strategies for Successful Digital Printing, Journal of Media Business
Studies, Vol. 3, No. 1, pp. 53-74, 2006
9 Dalal E.N., Rasmussen D.R., Nakaya F., Crean P.A. and Sato M., Evaluating
Overall Image Quality of Hardcopy Output, Proceedings from IS&T's 1998 Image
Processing, Image Quality, Image Capture, Systems Conference, Portland, Oregon,
pp. 169-173, 1998
10 Engeldrum P., Psychometric Scaling: A Toolkit for Imaging Systems Development,
Imcotec Press, Winchester, 2000
11 Saunders M., Lewis P. and Thornhill A., Research Methods for Business Students,
Third Edition. Prentice Hall, 2003
12 ISO 8254-1:1999, Paper and board -- Measurement of specular gloss -- Part 1: 75
degree gloss with a converging beam, TAPPI method, 1999
13 Field G.G., Test Image Design Guidelines for Color Quality Evaluations,
Proceedings from IS&T/SID’s Seventh Color Imaging Conference: Color Science,
Systems and Applications, Scottsdale, Arizona, pp. 194-196, 1999
14 Cui L.C., Do experts and naive observers judge printing quality differently?
Proceedings from IS&T/SPIE’s 2004 Image Quality and System Performance
Conference, San Jose, California, pp. 132-145, 2004
15 Ishihara S., The Series of Plates Designed as a Test for Colour-Deficiency, 24
Plates Edition. Kanehara Trading Inc. Tokyo, 2004
16 Mejtoft T., The Cost of Digital Printing in Newspaper Production, STFI-Packforsk
Report 111, STFI-Packforsk, 2005

192 EFFECT OF DRYING TEMPERATURE PROFILE AND PAPER ON MECHANICAL ….
Effect of drying temperature profile and paper on mechanical

print quality in heatset offset printing
Timo Hartus
Department of Forest Products Technology

Paper and Printing Technology
Helsinki University of Technology
P.O. Box 6400
FI-02015 TKK
Finland
Timo.Hartus@tkk.fi
Abstract
In this paper, the mechanical surface strength properties of offset prints were studied from
six different papers and three different heatset drying temperature profiles. The most
common types of uncoated and coated papers used in the heatset offset printing process
were included in this research.
Print gloss, roughness and abrasion resistance were investigated. Roughness values were
measured both using the Bendtsen method, and the possibility investigated to characterize
paper and print surface evenness by a modified print tack measurement application.
The gloss of the dried prints compared to the gloss of the corresponding paper increased
predictably. However, as shown by Marttila [1] high drying temperature diminished the
gloss level of prints on uncoated papers, related to an observed roughening of the surface
made up of the fibre mat In the case of coated papers, the gloss level of prints increased
slightly as a function of drying temperature, and exhibited diminished post-print
roughness, indicating either the insulating nature of the coating, or its base paper isolating
property in respect to ink diluents.
Abrasion resistance of the prints increased on uncoated paper grades as the heatset drying
temperature rose. The greater surface roughness value predictably also slightly increased
the smearing tendency compared to coated print surface in the rub tests, despite the lower
overall values of ink transfer. The prints on coated papers had some poorer abrasion
resistance when they were dried at high drying temperature than when they had been
dried under lower drying temperature.
Keywords: Heatset printing, abrasion resistance, gloss, roughness, paper
1 Introduction
During offset printing, viscous ink is transferred onto the printing plate through a train of rollers
to form of a thin ink layer. Heatset inks set different than sheet fed inks and have additional
parameter on drying. Heatset offset printing provides flexibility and added value to print on
coated papers, as well as improved quality on newsprint [2]. In the heatset process, the setting
and drying of the ink is promoted by applying surface heat to the print. Both coated and
uncoated paper grades are used, and the paper plays a significant role in defining print quality.
However, less work on ink gloss and print rub has been done for heatset prints compared to
coldset.
Due to the high speeds increasingly being used in heatset offset printing, the degree of drying of
the print has become a most important question. To avoid problems such as smearing of prints,

T. HARTUS 193
evaporation has to be as fast as possible. Heatset inks contain 30-40 % mineral oil-based solvent
(boiling temperature range about 240-290°C), of which 70-90 % evaporates in the dryer [1].
The other main components are pigment and resin. Resins bind pigment particles together and
to the paper. In inks, resin can be in either liquid or solid state, termed soft and hard resins
respectively. Pigments are insoluble and crystalline. Simultaneously, therefore, with the
evaporation of oil, binders harden, and the ink film develops a solid and glossy appearance. The
amount of residual oil in the print is depending mostly upon the drying temperature encountered
in the heatset drying oven. Residual oil proportion is also dependent on interactions of the oils
and binders in ink and paper compounds [3].
The roughness of the paper greatly determines the roughness of the print [4]. Print roughness
largely determines the final print gloss level [5]. The gloss of a print is strongly depending on
the underlying paper roughness [6]. In principle, the gloss of paper determines the limits for
print gloss [7]. Roughness and porosity affect ink gloss after printing and it shows that ink film
splitting event is influenced by these parameters [8]. Recent work has shown that there is also
an expected contribution from the refractive index of the ink layer, the higher the refractive
index, the greater the Fresnel reflectance [9, 10, 11]. Additives like waxes in ink can contribute
to the forming of high print gloss level. In cases where increased drying temperature acts to
diminish print gloss, it is accepted as being due to surface fibre roughening, especially on
uncoated and lightweight coated papers [12, 13]. With conventional printing methods, evenness
or homogeneity of paper greatly determines the limits for micro scale gloss uniformity of the
print [7, 14]. Although applying ink normally increases the gloss level as part of the image, it
often increases the gloss variation also [15]. Typically, the raising of drying temperature in
heatset offset increases the gloss level of the final print with coated paper grades, whereas with
uncoated paper grades, gloss value can decrease because of fibre roughening. Fiber roughening
is the commonly suspected cause of gloss decrease in heatset printing in the drying stage. The
problem is significant when printing paper is uncoated or only lightly coated [16]. Lowering of
drying temperature diminishes fibre roughening, but in that case the residual solvent content in
the print increases and that may produce setoff smearing problems or weak abrasion fastness.
Ink setting and ink drying on paper play important roles in determining the tendency for
smearing of the print, either in terms of setoff or in print rub. That is the reason why there have
been developed numerous methods for following ink setting and drying after the printing
process, for instance, ink tack force development (splitting force) [17, 18, 19]. Also, a novel
method and device have been proposed to quantify the point-to-point variation in the rate of ink
tack development at millimeter scale [20]. Similarly, abrasion resistance tells how resistant a
print surface is against rubbing. There are numerous print abrasion resistance (rub off) testers
available in the market. They nearly all adopt a similar rubbing principle and they all cause
failure within the ink or between the ink and the paper surface by either rotation. More recently,
a new test was proposed by [21], following earlier work by [22], in which a strain was applied
to an ink layer without inducing actual rub. This is thought to be more easily related to practice
and example correlations were given. In this study, however, the aim is to determine the
intrinsic cohesion and adhesion of the ink on paper and so traditional abrasion testing is
considered relevant to the job in hand.
The purpose of this study is to explain the impact of different paper grades and dryer
temperature profile in the print drying stage on print quality and fixing degree (adhesion and
hardening) in the heatset printing process. To achieve this, the effect of drying temperature on
the quality properties of heatset offset prints with various types of printing papers was studied,
and the results used to clarify the possibility of controlling the heatset offset print gloss by
adjusting the drying temperature. In this latter context, the development of print gloss and
roughness as a function of print drying temperature in a heatset oven is also investigated. Also,
it is also observed if the printed surface parameters can predict the resistance against mechanical
failures in prints in the cases of different papers and heatset drying oven temperatures.

2 Experimental
The studied papers were commercial products: 45 g/m2 news (1), 45 g/m2 heatset news (2), 45
g/m2 LWC (3), 56 g/m2 SC paper (4), 75 g/m2 LWC (5) and 80 g/m2 uncoated copy paper (6).
Papers were cut to a size of 55 mm x 250 mm for the experiments and printed area was 50 mm
times 200 mm in every print. These definitions are summarized also in Table 1.
Table 1: Summary of commercial paper grades.

Sample Type Coated Uncoated Basis weight /
g/m2
1 Mechanical Newsprint 45
2 Mechanical Heatset news 45
3 Mechanical LWC 45
4 Mechanical Supercalendered 56
5 Mechanical LWC 75
6 Woodfree Copy paper 80
All the printing experiments were done using one commercial, commonly used, black ink for
heatset printing process1.
Papers were printing with a Universal Testprinter2 laboratory printing machine using a printing
speed of 1 m/s, at a nip pressure of 630 N and forming a printing length of 200 mm. The width
of printing disc was 50 mm. In the beginning, 0.3 cm3 of ink were distributed onto the inking
device (Universal Testprinter High Speed Inking Device). The distributed ink was spread until
ink forms an even ink film. The spreading time was about ten seconds at the rotation rate of 100
m/min. Then a rubber-coated offset disk was set in contact with the ink roller on the inking
device, duration five seconds. After each inking of the offset disk, an ink addition of 0.02 cm3
was made on the inking roller to keep ink level constant on the offset disk. Although the ink
amounts on the offset disk were nearly constant, the ink uptake in the prints varied greatly
among the paper grades. No fountain solution was used – this was omitted primarily for ease of
laboratory printing but also to eliminate the effects of moisture variation during the experiment,
although in practice this is expected to have a significant effect, especially on fibre roughening.
Ink setting onto the papers was estimated by splitting force measurements. The measurements
were carried out using a print tack meter, Printack (Figure 1), which is a special accessory
mounted on the Universal Testprinter -laboratory printing device [23], and records the time
taken to remove a print contact disk, a 20 mm wide rubber coated aluminium roller, via the
action of a spring loaded cantilever. This time parameter is related to the viscosity properties,
mass of ink available for splitting (roughness dependent) and disc-adhesion properties of the
ink, especially being sensitive to the elongation or shortness of the ink as a function of time and
extension. The spring force was adjusted to about 2 N before the tests. The print tack meter
makes one pull from one position and the print sector turns to the next measurement point. Up
to 30 measurements can be done from one print. Whilst it is difficult to represent a given tack
force by this cantilever method, as might be offered by other techniques, such as Deltack [24] or
ISIT [25], it is useful to compare the release time parameter for the properties of the same ink
used on different papers.
1
Black heatset production ink for LWC and SC papers, manufactured by Sun Chemical.
2
Universal Testprinter is a product of Testprint B.V. The Netherlands.

T. HARTUS 195
Printack
Measurement disc
Print sector
Figure 1: The print tack meter (Printack) connected to the laboratory printing machine
The time delay between the end of printing and the first splitting measurement was one second.
The ink levels used for the splitting force tests were kept the same as for the drier samples. The
release time of the measuring disc is termed the print tack-value. The distance between the disc
(the upper removed position, when the disc is totally released from the print) and print on the
printing sector was constant; approximately 5 mm. The difference in the distance between the
disc and the printing sector has naturally a great effect on measured releasing time, and that is
why it has to be constant to get comparable results.
Immediately after printing the papers by the Universal Testprinter, the prints were extracted
from the printing sector and manually transferred and taped at both ends onto a special, self-
made, print carriage. The print carriage has an even area for 50 mm wide print instead of the
more usual 40 mm wide even area commonly used in Prüfbau-type printability testers. All the
other parameters and properties of the print carriage were maintained the same as in the
carriages designed for Prüfbau-type printability testers.
The carriage-mounted prints were subsequently dried in a FograHot oven. The speed of the
carriage passage through the oven was 0.10 m/s. The oven consists of three hot air blowing
sections and an on-line temperature meter, which automatically measures surface temperature of
the print just after the drying session. No air forming filter was in use. Drying temperature
profiles used in this study are presented in Table 2. The purpose of the selected drying
temperature profiles was to create a regime of fast heating, and to maximize the web
temperature differences, whilst still being able to get an acceptable drying result after the drying
units in order to study the changes in print due to temperature rising.
Table 2: Temperature profiles of hot air dryer.

Temperature of dryer units / °C
Heating profile code 1st unit 2nd unit 3rd unit
A 300 300 300
B 300 200 150
C 150 150 100
Print gloss was measured by a gloss meter3 (L&W Gloss tester SE224) after the drying of prints.
The devise produces gloss values according the standard T480. There was a 75° angle of
incidence is in use. The gloss values are relative values from 0 to 100. Five measurements were
taken from every sample, and, from these, the average values were calculated.
Printed samples were then further evaluated by measuring the optical density with a Vipdens
2000 densitometer (Vipdens 2000, Viptronic GmbH) without a polarization filter. The density
3
L&W SE224 gloss meter is a product of Lorentzen & Wettre

of the prints was taken as a control of the amount of ink on the paper, and also for calculation of
relative abrasion fastness values of prints. Print density was evaluated only on full printed areas,
which in this case meant that the inking disc was loaded to a constant amount but the papers
accepted different weights.
Bendtsen roughness of studied paper and the dried prints was measured according to the
standard SCAN P21-67 with an L&W SE114 Bendtsen Tester4.
The evenness of the papers and dried prints after pick testing was estimated by a specially
designed measurement, using the print tack meter described above. The fixing degree of digital
toner was characterized with this equipment in a previous study [26]. In this study also, the tack
measurement disc was “pre-inked” with pick test oil (IGT, normal tack). The distribution of
pick oil was made by a high speed inking device (speed 100 m/s, time 10 s). The disc was
weighed before and after the tack oil distribution, so that the total amount of pick oil on the
measurement disc could be determined and found to be about 14 mg, which corresponds to an
amount of 2.8 g/m2 after application on the paper.
In general, when used as a tack device, several measurements are made from a print as a
function of time to monitor drying of the print: the number of print tack measurements from
every print is 20. In this study, the average print tack value, in the presence of the pick oil, of
each measured test sheet was calculated from the multiple individually measured values, as the
test is applied to set ink and the strength of the ink and its adhesion values are being studied
which are constant over the time of testing after drying.
A PATRA Rubproofness Tester was used for measuring the degree of ink rub off from prints.
The test area was round with a diameter of 3 cm. Prints are rotated against a non printed paper
surface (uncoated copy paper 80 g/m2, paper 6, was used throughout as the contact paper) under
a controlled pressure, being applied by two standard loads having weights of 1 144 g and 134 g,
respectively. The number of rotations was 50 in each test.
The loosened/transferred ink amount was estimated by measuring the ISO brightness value of
the contact besmeared rubbing papers. The ISO brightness values were measured by an L&W
Elrepho 2000 spectrophotometer.
The prints after the abrasion resistance tests were photographed by a Highwood HW-10L digital
microscope connected to a computer. The magnification factor was 200. The images were
viewed by a Video viewer program. The images were grey level inverted to optimize the
resolution power in the printed image.
3.1 Printing
The various different types of papers took significantly different ink amounts from the offset
printing disc. The papers which were produced especially for the heatset printing process,
however, gave almost equal print density levels, and reflect the uniformity of papers devised for
the market sector. The uncoated news and copy paper took the greatest ink amounts and
produced the lowest print density levels. Printed ink amounts and respective measured density
values are collected in Table 3.
4
L&W SE114 Bendtsen Tester is a product of Lorentzen & Wettre

T. HARTUS 197
Table 3: Ink amounts on the prints and metered density values.

Paper Code Ink amount / g/m2 Density
1 1.86 1.54
2 2.47 1.18
3 0.94 1.53
4 1.40 1.66
5 1.17 1.71
6 2.07 1.17
3.2 Setting of the prints

The setting of ink on the papers was studied as described above, by metering splitting time with
the print tack meter immediately after printing, before the drying stage. The disc release times
from the prints are visualized in Figure 2.
The coated papers have the longest overall setting times. The slowness of tack rise of the coated
papers illustrates that the LWC grades studied are probably clay-rich and, therefore, not fast
setting, because ink can not be absorbed directly into the substrate, but rather ink vehicle needs
to be separated from the viscosifying ink layer. Furthermore, the surface is more even and
ink/paper contact area is smaller than it is with the uncoated papers, with their porous and rough
surface.
The uncoated papers (newspaper and uncoated copy paper), which are not commonly used in
the heatset printing process, gave the shortest setting times, because of the relative ease of ink
absorption into the substrate with a quite rough surface structure and also relatively greater
surface area. The uncoated papers which are developed especially for the heatset printing
process (paper code 1 and 4) have a more tightly packed surface structure and so behaved more
like the coated papers.
5
Release time, ms
3
4
4
3
2
5
1 2
1
6
0
0 20 40 60 80 100
Time, s
Figure 2: Setting of ink on studied papers / Release times in the print tack meter
(- paper 1, - paper 2, - paper 3, - paper 4, - paper 5 and - paper 6)
The step-wise behaviour of some of the print tack curves reflects the measurement method, in
which separation time is recorded and not actual tack force. The nature of the extensional
properties of the ink to filament breakage can remain constant over certain periods, despite the
expected progression of concentration.

3.3 Drying of the prints

A part of the prints, which was not used for the splitting measurement, was dried in the
FograHot-oven using selected heating temperatures previously described in Table 2. The web
temperatures just after the hot air blowing unit were measured. The observed web temperatures
are collected in Table 4.
The effect of paper and temperature profile produced significant variations in web temperature
between the prints. Predictably, the highest web temperatures were measured from the prints
which were dried using the hardest drying, profile A, and with uncoated paper grades. The
lowest web temperatures were measured from the prints which were dried using the softest
drying, profile C, and with coated paper grades, respectively. All the prints were dry enough
after the drying sessions to avoid any setoff smearing.
Table 4: Print surface temperature (°C) in the dryer after the 3rd heating unit
Paper Code
Heating profile 1 2 3 4 5 6
A 210 205 184 216 186 220
B 140 135 125 113 130 108
C 122 126 117 119 124 88
3.4 Estimation and development of print gloss and print roughness

The results of Bendtsen roughness measurements are visualized in Figure 3. Evenness of pick
response from the papers and prints estimated also by a print tack devise, Printack, and by the
developed application, are shown in Figure 4. The samples, which produce greatest roughness
values, had lowest pick oil print tack-values, as might be expected from lower contact area, and,
therefore, lower thin film forces, due to surface roughness. The uncoated samples, paper and
print, 2 and 6, had the greatest roughness and also the poorest pick contact evenness values.
Respectively, the coated papers and the prints on coated paper had the lowest roughness and
also the highest print pick strength evenness values.
The roughness of the prints was, under certain conditions and print-paper combinations, greater
than the roughness of the corresponding unprinted papers. The uncoated papers had the greatest
difference in the roughness level before and after heatset printing. Print tack pick measurements
gave significantly higher values from the papers than from the prints.
300
Roughtness (Bendtsen), ml/min
250
200 P
A
150
B
100 C
50
0
1 2 3 4 5 6
Samples
Figure 3: Roughness–values (Bendtsen) measured from the papers and the dried prints with
different temperature profiles. Samples 1-6 mean the codes of studied papers

T. HARTUS 199
300
250
Release time, ms
P
200
A
B
150
C
100
50
1 2 3 4 5 6
Samples
Figure 4: Print tack pick-oil release times measured from the papers and the dried prints with
different temperature profiles. Samples 1-6 mean the codes of studied papers
The connection between the roughness values and pick-oil release times in the print tack meter
values is evident (Figure 5). Print tack pick-oil measurements separate better the samples which
have low roughness than does the Bendtsen roughness measurement method. It can be observed
that at a range of high roughness, the Bendtsen-method separates samples better than the print
tack pick-oil measurement in both the cases of papers and prints.
300
250
Coated papers
200
Release time. ms
Uncoated heatset papers

150
100
Uncoated papers
50
0
0 0,001 0,002 0,003 0,004 0,005 0,006 0,007
1/(Roughness)2
Figure 5: Print tack pick-oil release times of papers and prints as a function of Bendtsen
roughness (■ means on print and Ο means on paper)
It can be observed that Figure 5 can be fitted to an exponential decay function where both
unprinted and printed papers tend to fall on the same line. The air leak method of the Bendtsen
roughness [27] depends on the square of the cross sectional lateral roughness area, i.e. obeys
Poisueille flow, whereas the print tack pick-oil value may be expected to follow the two
dimensional contact area as also surface gloss is proposed to correlate to the surface roughness
[28], which, for Gaussian surface profile statistics, will depend on the reciprocal of the
roughness area such that the graph should follow a 1/x2 behaviour. Surface roughness and
coating pore structure control the pick-oil or ink tack dynamics [29], i.e. smooth paper surface
with small scale pores is proposed in this study to have faster pick-oil or ink setting rate, and

also longer print tack release times, than paper surfaces having higher roughness or a more
rough-grained pore structure.
3.5 Abrasion resistance of prints

In Figure 6, the ISO brightness values of the PATRA rubbed papers are presented. High
brightness value means that only a small amount of rubbed ink has been transferred to the
rubbing paper, i.e. good abrasion resistance. In Figure 6, the rubbed prints, i.e. the remaining
printed surfaces after rubbing, are presented as inverted microscopic images, in which the white
areas correspond to the retained black printed ink and the dark areas correspond to the paper
surface where ink has been rubbed away.
ISO Brightness of contract rub
94
92
90
A
88
paper
B
86
C
84
82
80
1 2 3 4 5 6
Samples
Figure 6: Abrasion test results estimated from ISO brightness values of the contact rubbing
papers. (Studied papers are marked using the codes 1-6 and dryer heating profiles using the
codes A, B and C as in the previous Figures)
The application of drying temperature is making a glossy and even print surface and supports
the fixing of ink onto the paper surface by causing an intimate contact between paper and the
ink layer. The better fixing/adhesion degree is reached mainly by the greater contact and greater
adhesion forces affecting between the ink and paper layers. Also, it is assumed that few
chemical bonds can be formed between fibres and ink binders [30]. Too little residual solvent
content may cause a brittle ink layer which may reduce the mechanical strength between ink and
paper, and in this case, mechanical failures between the layers can occur easily under quite
small stress, especially in the case of coated papers, where high drying temperatures are
required to fix ink to the paper coating [3]. On the other hand, too much residual oil can lead to
smearing.
From Figure 7, it can be seen that the hard drying temperature, profile A, caused most fibre
roughening. From the raised fibre surfaces more ink was rubbed away than from the more even
or low lying print surfaces. The fibres, which have been raised above the print surface plane, are
easily seen in prints 1A, 2A, 4A and 6A.

T. HARTUS 201
Figure 7: Prints after the abrasion test (gray scale inverted images; more black colour in image
corresponds to more white in the print). Paper codes 1-6 and drying temperature profiles A, B
and C
3.6 Gloss of the prints

The measured gloss values (Figure 8) showed, as expected, print gloss to be higher than the
gloss of the unprinted papers. No significant gloss differences were found between the drier
heating profiles. Naturally, high drying temperature caused minor lowering of gloss especially
in uncoated paper grades, which is probably due to the observed fibre roughening. In coated
paper grades high heating temperature produced even higher gloss than lower drying
temperatures.
90
80
70
Gloss-value, %
60 P
50 A
40 B
30 C
20
10
0
1 2 3 4 5 6
Samples
Figure 8: Gloss of studied papers and prints (P means unprinted paper before printing, A, B, C
refer to the printer dryer heating profile)
The measured print tack pick-oil release times, i.e. strength evenness of the printed surface,
correlated well with the measured gloss values. We see that there are different correlations with
gloss depending on whether we are considering unprinted or printed samples, which are clearly
observed in Figure 9. This may be due to surface chemical differences between papers and
prints, though more likely it is the extra dependence of gloss on ink refractive index. This is also

supported by the analysis of print tack pick oil release times as a reciprocal function of
roughness.
90
80
70
60
Gloss, %
50
40
30
20
10
0
0 50 100 150 200 250 300
Release time, ms
Figure 9: Gloss of papers and prints as a function of print tack pick-oil release time (■ means a
print and Ο means a paper)
3.7 Effect of surface evenness on abrasion resistance of the prints

Only minor variations were found in the abrasion resistance between the heatset drying
temperature profiles of prints on any one paper (Figure 10). Roughness or print tack pick oil
measurement results alone do not explain abrasion resistance of the prints in this study. Selected
paper type, however, did produce major differences in the abrasion resistance. Uncoated, news
and copy paper (1, 2 and 6), had the best abrasion resistance, though they had high roughness
and low print tack pick-oil values. The coated papers (3 and 5) had the highest surface evenness
but they gave poorer abrasion resistance than news and copy paper. The SC paper (4), which
was produced specially for heatset, had evenness between coated and uncoated papers, but its
abrasion resistance was the poorest of all the prints.

T. HARTUS 203
100 6
ISO brightness of the contact rubbing

1
95
2 3
90
85
5
papers 80
75
4
70
65
60
55
50
0 50 100 150 200
Release time, ms
Figure 10: ISO brightness of the contact rub papers as a function of print tack pick-oil release
time of the prints
The lack of correlation between purely surface-related measurements and print rub resistance
has been discussed in depth by [31] in which it was shown that print rub resistance is a
combination of substrate compressibility, as well as contact smoothness and surface pore
structure. The differences between single coated and multicoated grades were demonstrated, and
the increased rub resistance of single coated papers compared with multicoated papers was
accounted for by the manifestation of the compressibility of the substrate through the single
coating layer. A similar argument can be used here now for the uncoated grades, i.e.
compressibility of the uncoated grades is greater than for the coated papers, and the protection
of the exposed ink layer within the uncoated paper surface roughness provides for improved
resistance to abrasion.
4 Conclusions
The roughness of the paper determines greatly the final print roughness or evenness level. The
surface roughness also determines greatly the limits for the surface gloss. High drying
temperature causes fibre roughening in the heatset printing process, especially in the case of
uncoated paper grades, and this is associated with diminishing gloss. On the other hand, higher
drying temperature produces a more even ink surface strength and a glossy ink film on paper,
manifest in raised print gloss-sheet gloss differential.
The method for predicting surface evenness by print tack measurements using tack-oil as an
inter-medium proved to be useful for defining the surface strength parameters in terms of
contact forces. The connection between contact evenness and print or paper gloss has been
demonstrated.
Abrasion resistance is depending on developed interactions between ink and paper or paper
coating. Paper surface structure is one important factor in adhesion of ink on paper. Rough
surfaces have greater surface area for good mechanical fixing. The bonding of paper coating
onto base paper has to be stronger than the adhesion of ink to the coating to prevent
delamination at the coating base paper interface. Neither paper nor print roughness/evenness
provided an unequivocal estimate of print abrasion resistance.

If heatset drying temperature is too high, it can cause too low residual solvent content, and ink
film becomes too hard and rigid. Too rigid an ink layer seems to weaken the mechanical
strength between ink and paper, or the paper coating layer, and, in this case, mechanical failures
between the layers can occur easily under small stress. In the case of coated papers this failure
can be between ink and paper coating. Also fibre roughening became a reason for poorer
abrasion resistance in the cases of some hard dried prints.
Acknowledgements
The author is grateful to Professor P.A.C. Gane for his valuable advice and comments
during writing this article.
References
1 Marttila J., Oittinen P., Grönlund A., Hartus T., Korhonen J. and Marchart
P.,Heat-set offsetpainatuksen kuivatusolosuhteiden hallinta, Teknillinen
korkeakoulu. Graafisen tekniikan laboratorio. Julkaisu 19. 40 p. (in Finnish), 1995
2 Wells N., Hybrid heatset and coldset printing, Newspapers & Technology, 1-5, 2001
3 Hartus T. and Oittinen P., Characterisation of the Drying Properties of Heatset
Inks by thermal Methods, Graphic Arts in Finland, Vol. 25, no 1, pp. 9-15, 1996
4 MacGregor M.A., Johansson P.-Å. and Béland M.-C., Measurement of small-
scale gloss variation in printed paper. Topology explains much of the variation for
one paper, Proceedings of the 1994 International Printing and Graphic Arts
Conference, pp. 33-43, 1994
5 Bousfield D.W., A model to predict the leveling of coating defects, Tappi J., Vol 74,
no 5, pp. 163–70, 1991
6 Handbook of Print Media: Technologies and production methods, Kipphan, H.
(ed.) Springer Berlin 2001 pp. 223-226, 2001
7 Oittinen P., The limits of gloss in prints, Paperi ja Puu Vol. 65, no 11, pp. 718–724,
1983
8 Jeon S.J. and Bousfield D.W., Print gloss development with controlled coating
structures, Journal of Pulp and Paper Science, Vol. 30, no 4, pp. 99-104, 2004
9 Preston J.S., Hiorms A.G., Elton, N.J. and Ström J., Application of imaging
reflectometry to studies of print mottle on commercially coated papers, Tappi
Coating & Graphic Art Conference, Atlanta, Proceedings, 2006
10 Sorjonen M., Jääskeläinen A., Peiponen K-E. and Silvennoinen R.,On the
assessment of the surface quality of black print paper by use of a diffractive optical-
element-based sensor, Measurement Science and Technology, Vol. 11 (2000), pp.
N85-N88, 2000
11 Granberg H., Spectral Light Absorption and Gloss of Cyan & Magenta Inks, Acreo,
T2F Report f59, 2002
12 Chinga G., Stoen T. and Gregersen Ø., On the roughening effect of laboratory
offset printing on SC and LWC surfaces, Journal of Pulp and Paper Science, Vol.
30, no 11, pp. 307-311, 2004
13 Sederholm R., The influence of Coating layer on heatset roughening, MSc. Thesis,
Helsinki University of Technology, 59 p. (in Finnish), 1998
14 Béland M.-C. and Bennett J.M., Effect of local microroughness on the gloss
uniformity of printed paper surfaces, Applied Optics Vol. 39, no 16, pp. 2719-2726,
2000
15 MacGregor M.A. and Johansson P.-Å., Gloss uniformity in coated paper;
Measurements of commercial papers, TAPPI Coating Conference Proceedings,1991
16 Korhonen J., The Effect of Paper and Ink on Heatset Drying, Graphic Arts in
Finland, Vol. 22, no 2, pp. 16-22, 1993
17 Concannon P. W. and Wilson L.A., A Method for Measuring Tack Built of Offset
Printing Ink on Coated Paper, TAGA Proceedings 44 Vancouver, 1992

T. HARTUS 205
18 Plowman N., Ink tack – Its effects and results, Graphic Arts Monthly Vol. 61, pp. 6-
8, 1989
19 Gane P.A.C., Seyler E.N. and Swan A., Some novel aspects of ink/paper
interactions in offset printing, International Printing and Graphic Arts Conference,
Halifax, Nova Scotia, Tappi Press, Atlanta, pp. 209-228, 1994
20 Xiang Y., Bousfield D.W., Hassler J., Coleman P. and Osgood A., Measurement
of local variation of ink tack dynamics, Journal of Pulp and Paper Science, Vol. 25,
no 9, pp. 326-330, 1999
21 Gane P.A.C., Kozlik T, and Schoelkopf J., Print Rub Determination: A Novel
Laboratory Method to Simulate Practice, Pulp and paper Canada Vol. 106, no 10,
pp. T21 -T219, 2005
22 Gumbel R., Converting prints on matt coated paper, International Symposium on
Paper Coating Coverage, Helsinki. Training Centre of Finnish Forest Industries
AEL-Metsko, 32pp, 1989
23 Testprint, Inc. Product Information, from http://www.testprint.com/ Retrieved
19.11.2005
24 Prüfbau. Technical Information, Prüfbau Multipurpose Tack Measuring System
Deltack, Prüfbau, Peißenberg/München 33 p, 2003
25 Gane P.A.C. and Seyler E.N., Tack development: an analysis of ink/paper
interaction in offset printing, Tappi Coating Conference Proceedings. San Diego,
Tappi Press, Atlanta, pp. 243-260, 1994
26 Hartus T., Effect of Toner Fixing Temperature on Print Properties in the
Electrophotographic Process, TAGA Journal, 3.3, pp 165-177, 2008
27 Heinemann S., Estimation of Surface Roughness and Air Permeabilility of Paper
According to Bendtsen, Papier (Darmstadt), Vol. 50, no 5, pp. 233-241, 1996
28 Gate L., Windle W. and Hine M., The relationship between gloss and surface
microtexture of coatings, Tappi Journal, Vol. 56, pp. 61-65, 1973
29 Xiang Y. and Bousfield D.W., Influence of Coating Structure on Ink Tack
Dynamics, Journal of Pulp and Paper Science, Vol. 26, no 6, pp. 221-227, 2000
30 Hartus T. and Oittinen P., Päällysteen ja painovärin kuivumistulokseen vaikuttavat
vuorovaikutusilmiöt (Drying Interactions of Paper Coating and Heatset Ink),
Uudistuva Paperi -seminaari, Espoo, Poster presentation (in Finnish), 1995
31 Gane P.A.C., Ridgway C.J. and Gliese T., A re-evaluation of factors controlling
print rub on matt and silk coated papers, Tappi Coating and Graphic Arts
Conference, New Orleans, Tappi Press, Atlanta, 2006

206 FACTORS IMPACTING THE EVALUATION OF PRINTER PROFILE ACCURACY
Factors Impacting the Evaluation of Printer Profile Accuracy

Robert Chung
Rochester Institute of Technology

School of Print Media
69 Lomb Memorial Drive
Rochester, NY 14623, USA
rycppr@rit.edu
Abstract
The evaluation of printer profile accuracy is affected by many factors and the printing
device represents only one such factor. This paper examines the effect of profiling target
layout and patch size on printer profile accuracy. Two digital color printing devices were
evaluated using a single color measurement system. The evaluation avoids temporal
variation associated with the color printing device on purpose. It was found out that (1)
the spatial color uniformity of the device bears a larger impact on its colorimetric
accuracy than factors such as target layout and patch size; (2) the use of the random target
over the visual target helps minimize spatial non-uniformity in the shadow region of the
color gamut; and (3) there is no significant difference in colorimetric accuracy when
reducing the patch size of the random target from 6mm to 4mm using the color
measurement instrument in this experiment.
Keywords: ICC, profile, accuracy, colorimetry
1 Introduction
Many people in North America live in the world of eight-and-half-by-eleven. The notepad is 8.5
by 11; the magazine is 8.5 by 11; and the most popular paper size for desktop printers is “Letter
Size,” and that is 8.5 by 11 inches. Having the same dimension helps to keep documents in a
neat pile.
There is finite area in an 8.5 by 11 inch space. If one takes one inch out as margins, the printable
area of an 8.5 by 11 page is 6.5 x 9 inches or 58.5 in2. This translates into 37,741 mm2 (one inch
is equal to 25.4mm). You may question what’s all the fuss!
When the first CGATS-endorsed profiling target, IT8.7/3, came out in mid-1990s, it contained
928 6mm patches and the entire target would fit into an 8.5 by 11 page with 6mm patch size
nicely. As a matter of fact, the “Letter Size” page with one-inch margin can accommodate 1,048
6mm patches.
When CGATS introduced the IT8.7/4 profiling target with 1,617 6mm patches [1], the magic
world of 8.5 by 11 no longer could cope with the addition of patches that were deemed
necessary by the color management communities. The only recourse without changing the paper
size while keeping the target intact is to reduce the patch size. For example, if one reduces the
patch size from 6mm to 4mm, it can accommodate up to 2,358 patches. Incidentally, neither
IT8.7/3 nor IT8.7/4 specifies the patch size as the normative part of the specification. The
default patch size of 6mm came from the days of IT8.7/3 and constraints in color measurement
instrumentation. In other words, the patch size of 6 mm is not a part of the standard, but the user
will confront the issue of target size when implementing color management.

R. CHUNG 207
In addition to the increase in number of patches, CGATS also introduced a random layout of the
target along with the visual layout in the informative part of the standard. Figure 1a is a reduced
size of the IT8.7/4 random target and Figure 1b is a reduced size of the IT8.7/4 visual target.
Figure 1: Two default layouts of the IT8.7/4 target
1.1 Problem Statements

Printer profile accuracy depends on the following immediate factors: the profiling target, the
color printing device, the color measurement instrument, the profile-making software, the
CMM, etc. Depending how printer profile accuracy is tested, spatial uniformity of the printing
device becomes a factor if it involves comparison of colorimetric measurements from more than
one location within the sheet. If it involves comparison of colorimetric measurements between
the press run that generates the press profile and the press run that applies the press profile,
color repeatability of the printing device then becomes a factor. Thus, it is important to ask,
“Which testing condition is suitable for testing printer profile accuracy?”
There are two default target layouts, i.e., random vs. visual, mentioned in the IT8.7/4
specifications. The concept of randomizing patches is believed to minimize the effect of
inherent device noises associated with inking evenness, ink starvation, etc. The question of
interest becomes, “To what extent is the random target more effective in minimizing printer
noise than the visual target?”
The dimension of IT8.7/4 with a patch size of 6mm no longer fits within the letter size. A
question of equal interest becomes, “Is there an adverse effect in color measurement accuracy
when patch sizes are reduced?”
These questions regarding target orientation, patch size, and measurement noise were initially
explored by Chung [2] and are now more fully examined. The researcher recognizes that the
experimental findings will not be absolute, but relative to the variables compared, e.g., visual
layout vs. random layout, 6mm patch size vs. 4mm patch size. Furthermore, specific findings
depend on the testing conditions, e.g., printing devices tested and color measurement instrument
used.
2 Literature Review and Pilot Study
A printer profile contains Look-up Tables (LUTs) between CMYK and CIELAB. It assumes
that there is no variability within a device so that each combination of CMYK is mapped to a
unique CIELAB values, and vice versa. In fact, a printer can vary both spatially and temporally

[3]. By understanding the magnitude of the variations, one can begin to have a handle on the
estimation of printer profile accuracy.
Variability is the degree of repeatability or precision of the measured outcome. When a person
measures anything only once, he knows nothing about its variability. If one measures a printed
color chart more than once, he can assess the variability associated with a color measurement
system or a printing device.
2.1 Instrument Repeatability

Researchers at Rochester Institute of Technology devised a method to estimate color
measurement instrument repeatability [4]. The Committee for Graphic Arts Technology
Standards also recommends the method whereby cumulative probability plot of ∆E distribution
from a multi-patch color chart is used for evaluation [5]. The procedures are to (1) measure a
printed IT8.7/3 (basic) target twice using one color measurement instrument; (2) calculate
colorimetric difference between any one of the two colorimetric measurements and its average;
and (3) arrange the ∆E distribution of all 182 patches in the form of relative frequency (better
known as a pie chart) or as a cumulative relative frequency (CRF). The CRF of ∆E becomes a
graphic depiction of the instrument repeatability. A Spectrolino/Spectroscan was used in this
study. Measuring a printed color chart twice is the minimum number of times to assess
repeatability of a measurement system.
The repeatability of this instrument, as shown in Figure 2, shows that one-half of the time, the
uncertainty of a color measurement is 0.2 ∆E or less; and 90% of the time, the uncertainty of a
color measurement is 0.4 ∆E or less. For purpose of comparing variations induced by different
factors, the scaling of the CRF curve is normalized from 0 to 10 ∆E in the x-axis.
1.0
0.9 CRF Curve -- Instrument Repeatability

0.8
0.7
0.6
0.5
0.4
0.3
0.2 Spectrolino/Spectroscan
0.1
0.0
0 1 2 3 4 5 6 7 8 9 10
²E
Figure 2: Instrument repeatability of Spectrolino/Spectroscan
2.2 Spatial Uniformity of the Printing Device

Spatial uniformity of a CMYK printing device can be estimated similarly with the use of a
minimum of two identical IT8.7/3 (basic) color blocks printed on the same sheet (Figure 3).
Colorimetric difference between any one of the two colorimetric measurements and its average
are calculated. The ∆E distribution of all 182 patches in the form of cumulative relative
frequency (CRF) becomes a graphic depiction of the spatial uniformity of the output device [6].

R. CHUNG 209
Figure 3: Two IT8.7/3 (basic) color blocks with one rotated

Figure 4 shows an example of spatial uniformity of a digital press using the analysis method
outlined above. By comparing with the instrument repeatability, one can see that one-half of the
time, the uncertainty of spatial uniformity is 0.6 ∆E or less; and 90% of the time, the uncertainty
of a spatial uniformity is 1.5 ∆E or less.
1.0
0.9 CRF Curves -- Spatial Uniformity
0.8
0.7
0.6
0.5
0.4
0.3
iGen3
0.2
Spectrolino/Spectroscan
0.1
0.0
0 1 2 3 4 5 6 7 8 9 10
²E
Figure 4: Spatial uniformity of a digital press
2.3 Temporal Consistency of the Printing Device

Temporal consistency of a CMYK printing device can be estimated similarly with the use of
two IT8.7/3 (basic) color blocks printed at different time. Here, spatial uniformity (the average
of the sample measurements) is differentiated from temporal consistency (the difference
between the sample value and its target value).
Figure 5 shows the CRF curve between two digital press runs. One can see that one-half of the
time, the uncertainty of temporal consistency of the digital press is 2.6 ∆E or less; and 90% of
the time, the uncertainty of temporal consistency is 5.7 ∆E or less. Multiple CRF curves may be
generated between sampled press sheets and the target to reflect temporal consistency of the
press run.

CRF Curves -- Temporal Consistency

1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2 NexPress Run 1 vs. NexPress Run 2

Spectrolino/Spectroscan
0.1
0.0
0 1 2 3 4 5 6 7 8 9 10
²Eab
Figure 5: Temporal consistency between two press runs
Instrument repeatability, spatial uniformity, and temporal consistency are independent factors
that impact the repeatability of color measurement data. When evaluating printer profile
accuracy, one has to live with the uncertainty of a color measurement instrument and the spatial
uniformity of the printing device.
2.4 Profiling Target Layout

Different target layout and the profile-making software react to spatial non-uniformity
differently. There was a curiosity to learn if target layout influences the color measurement
accuracy of the profiling target that, in turn, impacts the printer profile accuracy. If the spatial
uniformity of an output device is good, patch layout of the profiling target should have little
impact on the colorimetric accuracy of the resulting profile. On the other hand, if the spatial
uniformity of an output device is poor, patch layout of the profiling target and the profile-
making software may have a significant impact on the colorimetric accuracy of the resulting
profile. Thus, to determine the effect of target layout on printer profile accuracy remains as an
objective of the study.
2.5 Measurement Error due to Patch Size

Spooner discusses the effect of color measurement error due to patch size [7]. When the light of
the measuring instrument that illuminates in all directions of a translucent substrate such as
paper, some of the light that diffuses laterally out of the lighted area diffuses back to the lighted
area. If the measured area is equal to the patch size, then some laterally diffused light will exit
through the sample edges and back, and thus the measurement is influenced by adjacent colors.
He called such an effect, lateral diffusion error (LDE). To avoid this type of measurement error,
ISO 5/4 [8] specifies that the patch size should be 2mm larger on all sides from the measured
area.
The default width of a color patch in the IT8.7/4 target is 6mm wide. The diameter of the
Spectrolino/Spectroscan’s hold-down aperture, used in the experiment, is 5mm. The measured
area is about 4mm (Figure 6). In this case, there was a concern regarding the measurement error
when reducing the patch size from 6mm to 4mm.

R. CHUNG 211
Patch size (6mm)
Measured area
(4mm)
Hold-down aperture
(5mm)
Figure 6: Schematics of patch dimension and measured area
3 Methodology
In this research, the IT8.7/4 (full or 1,617 patches) targets were used to build printer ICC
profiles. The IT8.7/3 (basic block with 182 patches) target was used as input to the evaluation
process. These targets were printed to a KPG Approval color proofer and a Xerox DocuColor
6060 digital printer. The GMB Spectrolino/Spectroscan was used to measure all color patches
(CIELAB, D50, 2 degree). In other words, one set of color measurement was made from one
KPG Approval print and one Xerox Docucolor 6060 print containing the targets as described. In
addition, the GMB ProfileMaker 5.0 was used for ICC profile construction and CHROMIX
ColorThink 3.0 Pro for data extraction from ICC profiles. ∆Eab was used to express the color
difference as opposed to using other ∆E formulas because it is recommended by CGATS and
ISO 12647.
3.1 Testing the Effect of Target Layout

To test if there is a significant difference in colorimetric accuracy between the visual target and
the random target, two devices, KPG Approval and Xerox 6060 and the standard 6mm target
were chosen for the experiment. It is hypothesized that the 2,540 spots/in (spi) with dye
diffusion thermo transfer based KPG Approval is a spatially uniform output device. Thus, the
difference in target layout will have less impact on colorimetric accuracy than that of the 600 spi
dry toner based Xerox 6060. Below is the experimental procedure for testing the effect of target
layout.
a) Determine spatial uniformity of the devices by printing two IT8.7/3 (basic)

color blocks within an A3 sheet. The ∆E distribution (CRF curve) between
individual measurements and their averages is an indication of the spatial
uniformity of the output device. More importantly, the average CIELAB
values between the two corresponding patches of the two IT8.7/3 (basic)
targets represent the reference values when assessing colorimetric accuracy of
ICC profiles made from different layouts and from different patch sizes.
b) Print the IT8.7/4 visual target and the IT8.7/4 random target to KPG Approval
and Xerox 6060. Construct ICC profiles, with the same CMYK constraints
using GMB ProfileMaker 5.0.
c) Test colorimetric accuracy of these ICC profiles by means of output
simulation. This is done using the Worksheet feature of ColorThink 3.0 Pro to
perform A-to-B or device-to-PCS color conversion. Briefly, a CIELAB list
can be generated from a CMYK list via a specific ICC profile and the absolute
colorimetric rendering intent. The CIELAB list is the simulated outcome of
printing the CMYK target. Because no physical printing device is used, there
is no process variation involved.

d) Compute colorimetric difference (CRF curve) between the simulated output

and the reference value established in Step (3.1a). The CRF curve is a measure
of the printer profile accuracy.
3.2 Testing the Effect of Patch Size

To test if there is a significant difference in colorimetric accuracy due to patch sizes, KPG
Approval with both the 6mm (visual and random) and the 4mm (visual and random) targets
were chosen as the testing conditions. It is hypothesized that if patch size is a significant factor,
larger colorimetric errors will be detected in the ICC profiles built from reduced patch sizes.
The testing procedure is similar to Testing the Effect of Target Layout.
4 Results and Analysis
If there are colorimetric differences due to target layout or patch size, the difference has to be
relative to the inherent spatial variation of the device. The researcher will use the results of the
spatial uniformity that includes color measurement system error as the starting point to discuss
the effect of target layout and patch size on printer profile accuracy.
4.1 Spatial Uniformity of Output Device

The spatial uniformity of KPG Approval is shown in Figure 7. Colorimetric differences were
between individual measurements and their averages. The ∆E statistics shows that the median
∆E is 0.3 and the 90-percentile ∆E is 0.5. The maximum ∆E of 4 was from the color patch ID 92
with %dot area value of 70C, 100M, 20Y, and 0K. This purplish color patch has the largest
spatial color difference. Color patch ID 92 with a coordinate of H1 is located at the bottom
center of the IT8.7/3 (basic) target. There was no physical flaw associated the patch and the
cause of the color difference was unknown.
1.0
0.9
CRF Curves CRF ²Eab
0.8 Median 0.3
0.7 90% 0.5
Max. 4.0
0.6
0.5
0.4
0.3
0.2 Approval_Average vs. Approval_90
0.1 Approval_Average vs. Approval_0
0.0
0 1 2 3 4 5 6 7 8 9 10
²Eab
Figure 7: Spatial uniformity of KPG Approval

The spatial uniformity of Xerox 6060 is shown in Figure 8. The ∆E statistics shows that the
median ∆E is 0.5 and the 90-percentile ∆E is 1.2 with a maximum ∆E of 2.4. Figure 7 and 8
help verify that KPG Approval is more spatially uniform than Xerox 6060 digital printer.

R. CHUNG 213
1.0
0.9
CRF ²Eab
CRF Curves
0.8 Median 0.5
90% 1.2
0.7 Max. 2.4
0.6
0.5
0.4
0.3
0.2 X6060_Average vs. X6060_90
0.1 X6060_Average vs. X6060_0
0.0
0 1 2 3 4 5 6 7 8 9 10
²Eab
Figure 8: Spatial uniformity of Xerox 6060
4.2 Colorimetric Differences Due to Target Layout

To test the colorimetric accuracy of ICC profiles by means of output simulation, a CMYK list of
the IT8.7/3 (basic) target was set up in ColorThink 3.0 Pro as a Worksheet. By specifying an
ICC profile and absolute colorimetric rendering intent, the software transforms the CMYK list
into a CIELAB list via the A-to-B look-up table (LUT). The researcher realizes that the
accuracy of the color conversion depends on the CMM and algorithms used in the transform.
This is how one would simulate the output device without temporal variability of the printing
device.
The CIELAB list derived from the above simulation is known as the sample. The sample
CIELAB list and the reference CIELAB list, derived from Step (3.1a) of the methodology, are
used to calculate ∆E between them.
The comparison of colorimetric accuracy between target layouts of KPG Approval is shown in
Figure 9. Curve A is the spatial uniformity of KPG Approval, as described in Figure 7. Curve B
is the colorimetric difference associated with the profile built from the IT8.7/4 6mm version of
the visual target. Curve C is the colorimetric difference associated with the profile built from the
IT8.7/4 6mm version of the random target.
A
1.0
0.9
CRF Curves
0.8
0.7
B
0.6
0.5
0.4
C
0.3
Approval_Average vs. Approval_V6
0.2
Approval_Average vs. Approval_R6
0.1 Approval Non-uniformity
0.0
0 1 2 3 4 5 6 7 8 9 10
²Eab
Figure 9: Printer profile performances between two target layouts of KPG Approval
Figure 9 shows that the colorimetric error, via output simulation, is around 0.7 ∆E at the 50
percentile. This value is almost three times greater than the spatial uniformity of the Approval or
0.25 ∆E. This means that KPG Approval can match a colorimetric specified color with any
possible combination of CMYK values with an average ∆E error of 0.7. In this instance, the
error comes from the color measurement device, profile-making software, and CMM used in the
ColorThink 3.0 Pro software.
Figure 9 also reveals printer profile accuracy between the two target layouts. While the median
∆E from the visual target is slightly smaller than that of the random target, it was uncertain if
such a difference is significant. As stated earlier, KPG Approval is a uniform color output

device. Profiles made with different target layouts may have little effect on their colorimetric
accuracy.
More importantly, Figure 9 shows that there are more occurrences of large ∆E (greater than 4)
values in the visual target than in the random target. Upon further analysis with the use of ∆E
sorting feature in the ColorThink 3.0 Pro (Figure 10), these color patches were found to be all 4-
color black tints with patch ID from 129-134 in the IT8.7/3 (basic) target. The significance of
larger errors in 4-color black tints is that the use of gray component replacement (GCR) may
help reduce printing variation.
Figure 10: Color patches with ∆E values greater than 4 found in KPG Approval visual target
There was only one color patch with ID 92, as shown in Figure 11, having a ∆E larger than 4 in
the random target and the patch is made up of 70C, 100M, 20Y, and 0K, the same color patch
that was found to have the largest color difference due to spatial non-uniformity. Given that
there are choices, the use of the random target over the visual target, couple with the profile-
making software, can help minimize color measurement errors in the shadow region of the color
gamut.
Figure 11: Color patches with ∆E values greater than 4 found in KPG Approval random target
The colorimetric accuracy comparison between target layouts of Xerox 6060 is shown in Figure
12. Curve A is the spatial uniformity of Xerox 6060, as described earlier. Curve B is the
colorimetric difference associated with the profile built from the IT8.7/4 6mm version of the
visual target. Curve C is the colorimetric difference associated with the profile built from the
IT8.7/4 6mm version of the random target.
A
1.0
0.9
CRF Curves
0.8 C
0.7
0.6
0.5
0.4 B
0.3
X6060_Average vs. X6060_V6
0.2
X6060_Average vs. X6060_R6
0.1 X6060 Non-uniformity
0.0
0 1 2 3 4 5 6 7 8 9 10
²Eab
Figure 12: Printer profile performances between two target layouts of Xerox 6060

R. CHUNG 215
Figure 12 shows that colorimetric errors, via the Xerox 6060 output simulation, are greater than
1.5 ∆E at the 50 percentile. This value is, again, three times greater than the spatial uniformity
of Xerox 6060 or 0.5 ∆E. Without further testing if the difference between spatial uniformity
and temporal consistency are significant, Table 1 reconfirms that colorimetric accuracy of a
printer profile begins with spatial uniformity that is inherent in the output device.
Table 1: Comparison of colorimetric errors by output simulation
²E at 50 percentile
Spatial non- Visual Randomized
uniformity target target
KPG
0.25 0.7 0.9
Approval
Xerox
0.5 1.7 1.5
6060
In term of the effect of target layout on colorimetric accuracy of Xerox 6060 printer profiles,
Figure 12 suggests that the median ∆E from the visual target is slightly larger than that of the
random target. It is uncertain if such a difference is significant.
Figure 12 also shows that there are more occurrences of large ∆E (greater than 4) values in the
visual target than in the random target. Using the ∆E sorting feature in the ColorThink 3.0 Pro,
it was found out that the same group of color patches with patch ID from 129-134 in the IT8.7/3
target yielded larger ∆E values from the visual target as discussed in the Approval case. Both
the magnitude and the occurrence of larger ∆E (greater than 4) were reduced in the random
target.
4.3 Colorimetric Differences Due to Patch Size

Colorimetric accuracy of two KPG Approval profiles, i.e., Visual_6 and Visual_4, is shown in
Figure 13. Curve A is the spatial uniformity of the Approval, as described in Figure 7. Curve B
is the colorimetric difference associated with the profile built from the IT8.7/4 6mm version of
the visual target and the curve C is that of the IT8.7/4 4mm version of the same target.
A
1.0
0.9 B
CRF Curves
0.8
C
0.7
0.6
0.5
0.4
0.3
0.2 Approval_Average vs. Approval_V6
Approval_Average vs. Approval_V4
0.0
0 1 2 3 4 5 6 7 8 9 10
²Eab
Figure 13: Device simulation of KPG Approval using visual target in 6mm and 4mm
Figure 13 shows that (1) the median ∆E between the visual 6mm and the visual 4mm KPG
Approval targets is the same; (2) there are larger ∆E values associated with the reduced patch
size. For example, ∆E at the 90 percentile increased from 1.9 ∆E to 2.6 ∆E as the patch size
reduced from 6mm to 4mm. Spooner’s lateral diffusion error (LDE) may very well explain why
this had occurred. But the effect of lateral diffusion error did not support the two random targets
(Figure 14). As shown in Figure 14, colorimetric differences are relatively small between the
random 6mm and the random 4mm targets.

A
1.0
0.9 B
CRF Curves
0.8
C
0.7
0.6
0.5
0.4
0.3
0.2 Approval_Average vs. Approval_R6
Approval_Average vs. Approval_R4
0.0
0 1 2 3 4 5 6 7 8 9 10
²Eab
Figure 14: Device simulation of KPG Approval using random target in 6mm and 4mm
By stepping back and examining both Figure 13 and 14 together, it became evident that (1) the
random target tends to yield better profile accuracy than the visual target by reducing the
occurrence of larger ∆E, and (2) the random target is less affected by the effect of reduced patch
size.
5 Discussion
This research work was initially motivated by curiosity of the causal relation between the
profiling target variations and the printer profile accuracy. Although measurements can also be
affected by ambient illumination entering the instrument, it was realized that printer profile
accuracy is the response of a system, and not a single element alone.
Figure 15 summarizes the factors that may influence printer profile accuracy. Major factors that
contribute to the evaluation of printer profile accuracy include printer repeatability, color
management, color measuring instrument, and the evaluation method used. In this research, the
researcher paid special attention to the effect of profiling target and the type of printer on printer
profile accuracy. He held a number of factors, e.g., color measuring instrument, profile-making
software, as constants in the experiment.
Figure 15: Factors contributing to the evaluation of printer profile accuracy
Based on our own research work, the magnitude of variation due to color measuring instrument
is the smallest of all variations in the system. The smaller patch size of the profiling target can
cause additional measurement error due to geometric precision of the positioning mechanism
and the halftone structure of the print. It is recommended that more instruments with different
measured areas be included as a follow-up study. Particularly, the alignment between the

R. CHUNG 217
instrument and the patch becomes critical as the measured area approaching to the size of the
color patch. Lateral diffusion errors will be more severe for profiling target with small patch
sizes in randomized layout than for profiling target in visual layout. This is because the
magnitude of these errors varied with the size of the area being measured and the location of the
adjacent colors [7].
Printing repeatability is a major factor in the evaluation of printer profile accuracy. Spatial
uniformity of printing is adopted as a baseline when evaluating colorimetric accuracy of color
printing devices. Temporal consistency and run-to-run variability associated with color printing
devices were purposely avoided by means of output simulation with the Worksheet feature of
the ColorThink 3.0 Pro.
Other than variations in profiling target layout and patch size, color management parameters,
e,g., profile-making software, CMYK constraints, CMM, were held as constants in this
experiment. As such, the researcher had to accept errors in profile predicted values via device
simulation that was three times in ∆E magnitude in comparison to the spatial uniformity of the
printing device. If the researcher were to experiment with the profile-making software settings,
including the computational ability of the CMM, he would have reduced colorimetric errors in
the simulation process.
Printer profile accuracy may be assessed by different evaluation methods. When a synthetic
color target is used, it involves the A-to-B color conversion where the ‘A’ space is a CMYK
color space. The use of a colorant-based synthetic target, such as the IT8.7/3 (basic), proved to
be effective in the evaluation of color repeatability of printing devices in color-managed
workflows. Other target, e.g., the ECI 2002 profiling target, with ECI published characterization
data set as the reference was also used [9]. In this case, the device uniformity is excluded in the
evaluation process.
There are two limitations with the use of a colorant-based synthetic target as experienced in this
research: (1) only the A-to-B LUT of the printer profile is evaluated and the B-to-A LUT is not;
(2) the findings do not necessarily predict visual responses of pictorial color image match. As
suggested in Figure 15, it is useful to include pictorial color images with synthetic colors that
are sampled from various pictorial scenes as the input target in the evaluation process. It is
recommended that pictorial color images and an image-based synthetic target are included in the
evaluation of printer profile accuracy in the context of digital proofing workflow as a follow-up
study.
6 Conclusion
There was little research on conducting printer uniformity and temporal consistency when film-
based proofs were the norm and pressmen were asked to print to visually match the supplied
color proofs. ICC-based color management changed all that. In particular, color proofs are to
match standard printing conditions. Color matching between two printing devices begins with
accurate ICC profiles. This paper is an attempt to better understand factors that impact printer
profile accuracy.
Colorimetric accuracy of a printer ICC profile can be affected by a number of factors. From the
variables tested, device uniformity is the most important factor, i.e., the more uniform the
device is, the more repeatable the color will be. Factors such as target layout and patch size are
of secondary importance.
The use of the random layout over the visual layout of the profile target helps minimize spatial
non-uniformity in the shadow region of the color gamut. This requires that the profile-making
software have the data fitting and color space modeling capabilities.

Measurement noise is proportional to the spatial non-uniformity of the printing device.

Measurement noise from the randomized target layout may be smoothed out by profile-making
software. So, printer profile accuracy is a combination of target layout and data smoothing
ability of the profile-making software.
Back to the world of 8.5 by 11. When reducing the IT8.7/4 profiling target from 6mm to 4mm,
there are colorimetric differences between the two visual targets, i.e., there are more larger ∆E
values associated with the reduced patch size. But this effect did not repeat in the random
targets with different patch sizes. In this case, there were no significant colorimetric errors
caused by the reduced patch size. For all practical purposes, there is no penalty to use the
random target at the 4mm reduced patch size.
Acknowledgments
The author wishes to thank Mr. Steve Upton of CHROMIX who shared many insights
and new features in the ColorThink 3.0 Pro software. He wants to recognize his graduate
student assistant, Dimitrios Ploumidis, for providing the laboratory support. He is also
grateful to his RIT colleagues, Franz Sigg and Edline Chun; and his industry colleagues,
David McDowell, Gary Field, and William Birkett, for reviewing the paper.
References
1 CGATS IT8.7/4, Graphic technology – Input Data for Characterization of 4-Color

Process Printing – Expanded Data Set, 2005
2 Chung R., The Effect of Profiling Target Variations on Colorimetric Accuracy of
Printer Profiles, Test Targets 6.0, pp. 19-24, 2006
3 Sigg F., Spatial Uniformity of Offset Printing, TAGA Proceedings, pp 649 – 658,
2007
4 Chung R. and Shimamura Y., Conducting a Press Run Analysis, Proceedings of
the 28th IARIGAI Research Conference, Advances in Color Reproduction, GATF,
pp. 333-345, 2001
5 CGATS, Recommended Industry Practice, Color characterization data set
development — Analysis and reporting, 2006
6 Chung R., Gravure Research Agenda: Achieving Repeatable Color in Packaging
Printing, Gravure, pp. 44-49, 2006
7 Spooner D.L., Evaluation of a Method for Correcting for Measurement Errors
Caused by Adjacent Colors, TAGA Proceedings, pp. 416-427, 2002
8 ISO 5/4, Photography - Density Measurements - Part 4: Geometric Conditions for
Reflectance Density, International Organization for Standardization, 1995
9 Sharma A., WMU Profiling Review, Western Michigan University, 2005

E. HREHOROVA, M. REBROS, A. PEKAROVICOVA, P.D. FLEMING, V.N. BLIZNYUK 219
Characterization of Conductive Polymer Inks based on

PEDOT:PSS
Erika Hrehorova, Marian Rebros, Alexandra Pekarovicova, Paul D. Fleming, and Valery N.
Bliznyuk
Department of Paper Engineering

Chemical Engineering and Imaging
Western Michigan University
4601 Campus Dr.
A-217 Parkview Campus
Kalamazoo, MI 49008-5462
Email: hrehorova@gmail.com
mrebros@gmail.com
a.pekarovicova@wmich.edu
dan.fleming@wmich.edu
valery.bliznyuk@wmich.edu
Abstract
The main driving force for implementation of printing in manufacture of flexible

electronics is the possibility to reduce the cost by high speed R2R processing at ambient
conditions. This work focuses on characterization of inks that can be used to print simple
components or layers for various applications in electronics. More specifically, inks
containing conductive polymer, poly(3,4-ethylenedioxy-thiophene)-poly(styrene
sulfonate), known as PEDOT: PSS, were tested. In order to deposit smooth and uniform
functional layers, it is important to optimize ink spreading and leveling on the substrate.
PEDOT:PSS is commercially available as an aqueous dispersion with high surface
tension (71 mN/m), which leads to poor ink spreading and substrate wetting. In this work,
addition of alcohols and surfactants was used to lower the surface tension of polymer ink
and the effect of concentration on dynamic and static surface tension was studied.
Dynamics of ink spreading were also tested using dynamic contact angle measurements.
It is shown that ink containing primary alcohol wets the surface more readily than
surfactant containing systems. Addition of alcohols and surfactant had also positive effect
leveling and bulk conductivity of final inks. More uniform layers were produced when
using inks with lower surface tension. Increased bulk conductivity by addition of ethylene
glycol is believed to be a result of conformational change and increased interaction
between polymer chains. Stronger interactions between polymer chains were also
confirmed by rheological measurements.
Keywords: conductive polymers, surface tension, wetting behavior, topography,

rheology
1 Introduction
Nowadays, more and more electronics manufacturers are embracing printing technologies as
high-potential manufacturing methods for mainstream electronic components. However, to fully
utilize the benefits of printed electronics, manufactures need advanced materials that are well
suited for specific electronic applications and also printing systems, and are available in
commercial quantities. Functional materials needed for printing of electronic components
include i) conductors, ii) semiconductors and iii) dielectrics.

220 CHARACTERIZATION OF CONDUCTIVE POLYMER INKS ….
This work considers the properties of conductive inks that can be used in various printed
electronics applications. Ink conductivity can be achieved by different mechanisms, such as
incorporating metallic or other conductive particles into a non-conducting polymer matrix, or by
using polymers that exhibit electrical conductivity in a suitable solvent. Among conductive
polymers, solvent based (xylene, toluene) polyaniline inks [1, 2] and water-based poly (3,4-
ethylenedioxy-thiophene)-poly(styrene sulfonate) (PEDOT:PSS) inks [3, 4] are widely studied
for applications in organic electronics. Several different printing methods are employed in these
studies, such as flexography, offset lithography and gravure printing, as well as ink-jet printing.
The present work focuses on characterization of PEDOT:PSS based polymer inks. PEDOT: PSS
is commercially available as a water-soluble polyelectrolyte system with good film-forming
properties, high visible light transmittance, and excellent stability [5]. Some applications of
PEDOT:PSS include antistatic coatings, conductive layers in organic light emitting diodes
(OLEDs), capacitors and thin film transistors [6]. A PEDOT:PSS complex is prepared by
oxidative polymerization of ethylenedioxythiophene (EDOT) in aqueous dispersion using
sodium peroxodisufate as the oxidant. A template polymer (usually polystyrene sulfonic acid –
PSS) is present during the polymerization [5]. The PSS in the resulting complex acts as a source
for the charge balancing counter ion. Moreover, it keeps the PEDOT chains dispersed in water,
forming stable, easy to process, deep blue microdispersions [6]. The chemical structure of
PEDOT: PSS is shown in Figure 1.
As already mentioned, PEDOT:PSS is commercially available as aqueous dispersions. The

water-based nature of such polymer systems gives rise to the issues of substrate wetting and ink
spreading. Water has a high surface tension and thus water based inks are very often formulated
with alcoholic co-solvents and/or surfactants in order to lower surface tension for printing. The
addition of alcohols lowers the surface tension monotonically with increasing concentration, due
to a preferential adsorption of the organic molecule at the liquid-air interface. Surfactants,
however, quickly reduce the surface tension at very low concentrations up to the critical micelle
concentration (CMC), due to a strong adsorption of the surfactant at the liquid-air surface. At
concentrations higher than the CMC, the surface tension is practically constant, because any
additional amount of surfactant will form micelles in bulk [7].
O O O O O O
S
+
S S
S S S
+
O O O O O O
_ _
SO3H SO3H SO3H SO3 SO3H SO3H SO3H SO3
Figure 1: Chemical structure of poly(3,4-ethylenedioxy-thiophene)-poly(styrene sulfonate)

complex (PEDOT:PSS)
Although the static surface tension is widely used in the printing industry to predict wetting
behavior of printing inks, especially water based printing inks, it is also important to
characterize interfacial surface tension under dynamic press conditions, where the ink is under
constant compositional change. A new liquid-air interface is created characteristically in the
order of milliseconds. Dynamic surface tension of inks containing surfactants and other
polymers is determined by diffusion, adsorption and desorption processes and it is typically
higher than the equilibrium (static) surface tension.

Analysis of wetting behavior of different functional fluids and their interaction with polymeric
substrates is an important subject in order to develop and optimize various materials for
electronics manufacture. The quality of the interface between functional layers in an electronic
device is crucial for its performance [8]. In general, the wetting process is reflected by the
contact angle, defined as the angle that a liquid makes with a solid surface. The equilibrium
relation of a three phase system can be described by the Young – Dupre equation [9, 10]:
γ lv cos Θ = γ sv − γ sl , (1)
where Θ is the contact angle and γ is the surface/interfacial tension at the liquid-vapor interface
(lv), solid-vapor interface (sv) and solid-liquid interface (sl).
As already discussed, there are several different ways to reduce surface tension of water based
systems. With PEDOT:PSS dispersions, addition of secondary alcohols has yet another positive
effect on the resulting films. It has been reported that electrical conductivity can be enhanced by
addition of different organic compounds. The conductivity improvement is strongly dependent
on the chemical structure of the compound. Among the alcohols, ethylene glycol and glycerol
were found to be the most efficient [11]. Enhancement of conductivity is believed to be a result
of an increased interchain interaction caused by conformational change of the PEDOT chains
from the coil structure into expanded-coil or linear structures [12].
When formulating conductive polymer inks, it is necessary to consider the method of printing
and desired application, because requirements for ink properties can greatly vary for different
printing processes. In addition, it is important to understand the effect of individual components
employed in ink formulation on electrical properties of printed layers. Additives typically used
for PEDOT-PSS inks or coatings include co-solvents and surfactants, bonding and cross linking
agents, adhesion promoters and additives for conductivity improvement [13].
This work studies the effects of ink additives such as co-solvents and surfactants on ink
properties in both fluid and solid form. More specifically, we studied the effect of addition of
ethylene glycol, ethanol and surfactant on dynamic and static surface tension of PEDOT: PSS
dispersions. The dynamic contact angle of the resulting ink was measured in order to examine
the wetting behavior on poly(ethylene terephthalate) (PET) substrate before and after corona
treatment. Surface topography and conductivity of resulting films on glass substrates as well as
on PET were also studied. The effect of solvents, and additives on rheological behavior of
PEDOT:PSS was examined.
2 Materials and Experimental Procedures
2.1 Materials
The conductive polymer (Baytron® P) dispersion was obtained from H.C.Starck GmbH & Co,
which contains 1.2-1.4% of PEDOT:PSS in water. Three different types of PEDOT: PSS based
inks were prepared. Other materials, such as ethylene glycol, ethanol and TWEEN80 (nonionic
surfactant) were purchased from Sigma Aldrich. Ethylene glycol was used in the formulation of
PEDOT:PSS based inks to enhance conductivity. Ethyl alcohol and TWEEN80 were used to
decrease surface tension of the PEDOT:PSS dispersion. Table 1 shows the tested ink
compositions and ID’s that will be used throughout this article.

Table 1: Composition of tested PEDOT: PSS based inks

Ink ID Ink Composition
PEDOT:PSS Pure PEDOT:PSS (Baytron® P)
EG-PEDOT:PSS Ethylene Glycol in PEDOT:PSS (50% v/v)
EtOH-EG-PEDOT:PSS Ethanol in EG-PEDOT:PSS (25% v/v)
TWEEN80-EG-PEDOT:PSS Surfactant Tween80 in EG-PEDOT:PSS (0.31 wt %)
2.2 Procedures
A SensaDyne Tensiometer was used to measure dynamic surface tension of the inks during
addition of ethylene glycol, ethanol and surfactant to the PEDOT:PSS dispersion. This test is
using the maximum differential bubble pressure method [14] based on creation of air bubbles in
the fluid, at the end of two orifices with different diameters. The differential pressure of the
formed bubbles is measured and the surface tension of the liquid is directly proportional to the
pressure difference.
Measurement of the static surface tension of inks was done using the contact angle analyzer
FTA200 from First Ten Angstroms. The values of ink surface tension were calculated from the
pendant drop[15] shape of the ink.
Wetting behavior of inks was tested by measuring dynamic contact angle using the FTA200.
The values of contact angle were calculated from the sessile drop [16]. PET (DuPont) was used
as a substrate for contact angle measurements. A Corona treatment (SOA, Inc.) was applied in
order to increase surface energy of the PET substrate [17].
The rheological behavior of PEDOT:PSS based inks was studied using a TA AR 2000 Dynamic
Stress Rheometer together with Rheology Advantage software. Concentric cylinder geometry
was employed to measure the ink samples. In order to eliminate possible shear history effects
from loading, each sample was allowed to equilibrate for 2 minutes. The geometry was
maintained at a constant temperature using a circulating water bath (25 °C). A steady state flow
test was performed in the range of shear rates from 0.0001 to 2000 s-1. During this test, a shear
rate is applied and viscosity measured when the material reaches steady state flow. After the
viscosity is measured, the shear rate is again increased and the process repeated yielding a
viscosity flow curve. Measured flow curves were then fitted with appropriate flow models, with
the help of Rheology Advantage Data Analysis software.
In order to observe the changes in surface topography of PEDOT: PSS based films; the tested
inks were solution casted onto glass slides and dried for 30 minutes at 120 °C. Solution casting
is a simple drop casting, where a certain volume of liquid is dropped onto the substrate and let
dry to produce solid polymer film.
White Light Interferometry in the Vertical Scanning Mode (WYKO RST-Plus microscope) was
used to study topography of the resulting films. White light interferometry is a non-contact
method for optical surface profilometry of various surfaces [18]. In vertical scanning mode
(VSI), the interferometric lens scans the surface at varying heights by vertical movement
through the focus and captures interference data at fixed intervals. The interference signal for
each point of the sample surface is recorded, providing information about the fringe modulation,
which is consequently used to calculate the surface height profile [19]. Another method for
surface topography characterization employed in this study was Atomic Force Microscopy
(AFM). An Autoprobe CP machine (Thermomicroscopes, USA) operated in a tapping mode and
with typical scan sizes from 30 x 30 μm2 down to 2 x 2 μm2 was used to study the
morphological features at different levels of structural organization and depending on the

preparation conditions of the samples. Root mean square (RMS) roughness was then measured
and used as a quanity characterizing perfectness of the polymer film on glass.
The conductivity of PEDOT:PSS based inks casted on glass was measured using a Keithley
2400 multimeter in four-probe mode. Measured resistance, cross-sectional area, and length of
tested film sample can be used to calculate resistivity, ρ (Ω-cm). The inverse of resistivity yields
conductivity (S/cm).
3.1 Static and Dynamic Surface Tension

Firstly, ethylene glycol was added to the PEDOT:PSS dispersion. This addition caused decrease
of static surface tension of PEDOT:PSS ink from 70.7±0.7 to 59.3±0.2 mN/m as measured by
the pendant drop method. However, the surface tension of EG-PEDOT:PSS measured under
dynamic conditions is higher than the static (equilibrium) surface tension and depends strongly
on time of interface existence - surface age. For shorter surface age, the alcohol molecules have
less time to migrate onto the newly created interface and thus the surface tension is higher than
that measured under equilibrium conditions (Figure 2).
90
80
75.5
70.6 70.0
70
Surface Tension [mN/m]
60
53.3
50
40
30
20
10
0
1
1.12 sec 1.52 1.86 sec Infinite Time = static ST
Surface Age [s]
Figure 2: Dynamic surface tension of EG-PEDOT:PSS ink
Figures 3 and 4 show the change in dynamic surface tension and bubble frequency of EG-
PEDOT: PSS ink during addition of ethanol and nonionic surfactant TWEEN80, respectively. It
can be seen, that addition of alcohol into the system caused a gradual decrease in surface tension
within the measured range of ethanol addition. It has been reported [20] that the surface tension
decreases relatively slowly or is almost constant when the ethanol content in ethanol/water
mixture exceeds 20 vol%. Therefore, and also due to low concentration of polymer, we have
used only up to 20 vol% of ethanol in this work. In the case of the surfactant TWEEN80 (Figure
4), the initial drop in surface tension is more dramatic and further addition of surfactant caused a
rather slow decrease in surface tension, indicating that the system is above the CMC of the
tested surfactant at the measured bubble frequency. This conclusion is also confirmed by a
steady bubble frequency observed.

80.00 1.2
75.00 1.0
BF [bubble/sec]
70.00 0.8
65.00 0.6
60.00 0.4
55.00 0.2
50.00 0.0
0 5 10 15 20 25
EtOH Addition [% v/v]
Figure 3: Surface tension and bubble frequency changes during addition of ethanol to EG-
PEDOT: PSS ink
80.0 1.2
75.0 1.0
70.0 0.8
BF [bubble/sec]
65.0 0.6
60.0 0.4
55.0 0.2
50.0 0.0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Surfactant Conc. [wt %]
Figure 4: Surface tension and bubble frequency changes during addition of TWEEN80 to
EG-PEDOT:PSS ink
Static surface tensions of the tested inks are shown in Table 2. The “rule of thumb” in the
printing industry is to have the surface tension of ink at least 10 mN/m lower than the surface
energy of the substrate to be printed on. Typical values of static surface tension for water-based
inks used in gravure or flexo printing are in the range of 28 - 45 mN/m [21]. The lowest surface
tension was found for EtOH-EG-PEDOT:PSS ink. However, printing of such ink might be still
problematic for some polymeric substrates with lower surface energy.
Table 2: Static surface tension of tested PEDOT:PSS based inks

Ink ID Static Surface Tension [mN/m]
PEDOT:PSS 70.7 ± 0.9
EG-PEDOT:PSS 59.3 ± 0.2
EtOH-EG-PEDOT:PSS 37.4 ± 0.2
TWEEN80-EG-PEDOT:PSS 41.8 ± 0.1

3.2 Dynamic Contact Angle

The dynamic contact angle was measured for all prepared inks on corona treated PET substrate.
As expected, the lowest contact angle was found for the ink with the lowest surface tension
(EtOH-EG-PEDOT:PSS). It can be seen from Figure 5, that the contact angle of pure
PEDOT:PSS ink and inks containing only alcohols stabilize after a short time (around 1.5 sec),
corresponding to initial spreading of the ink drops on the substrate. In the case of surfactant-
containing system, the contact angle did not reach a stable value, even after 30 seconds and
continued to decrease. This behavior has been seen previously for dynamic contact angle of
liquids containing surfactants [22]. Figure 6 shows the contact angle of TWEEN80-EG-
PEDOT:PSS ink on PET substrate after 2, 5 and 10 minutes of observation.
85
PEDOT:PSS
75
Contact Angle [deg]
EG-PEDOT:PSS
65
TWEEN80- EG-PEDOT:PSS
55
EtOH- EG-PEDOT:PSS
45
35
0 5 10 15 20 25 30
Time [s]
Figure 5: Dynamic contact angle of PEDOT:PSS inks on PET substrate
Figure 6: Contact angle of TWEEN80-EG-PEDOT:PSS ink on PET substrate
During printing, however, there is only a very shot time available for ink to spread on the
substrate before it goes into the drying station. Thus, the dynamic contact angle is more
important than contact angle at equilibrium conditions. Therefore, it is reasonable to look at
contact angle measurements only for short time periods (Figure 7). It can be seen, ethanol is
more efficient than the surfactant (TWEEN80) at short time scale. This is valid for both static
and dynamic conditions for the tested system.

85
75
Contact Angle [deg]

PEDOT:PSS
65
EG-PEDOT:PSS
EtOH-EG_PEDOT-PSS
55
TWEEN80-EGH-
PEDOT-PSS
45
35
0 0.2 0.4 0.6 0.8 1
Time [s]
Figure 7: Dynamic contact angle of PEDOT:PSS based ink at short time scale
3.3 Surface Topography

In order to avoid effects of the substrate and study only the effects of ink formulation of surface
topography, conductive polymer films were drop casted on glass slides and dried at 120 °C for
30 minutes to assure complete removal of ethylene glycol from polymer films. It can be seen
from Figure 8, that addition of alcohols into the PEDOT:PSS system significantly improves
uniformity of the film surface at the millimeter scale (2.5x1.9 mm2) as measured by vertical
scanning interferometry (VSI). This is due to lower surface tension of the alcohol-containing ink
and thus improved wetting of the glass substrate. However, AFM scans made at micrometer
scale (10x10 μm2) show smoother surface of PEDOT:PSS films. EtOH-EG-PEDOT:PSS film
show appearance of some larger domains, which can be a result of conformational change of
polymer chains and swelling of the PEDOT:PSS complex, indicating stronger interchain
interactions caused by alcohol addition [12]. It was found that the RMS roughness of
PEDOT:PSS film measured by VSI was reduced from 902 nm down to 67 nm by addition of
ethylene glycol and ethanol. Simultaneously, the RMS roughness measured by AFM shows
only 2.3 nm for PEDOT:PSS and 12.9 nm for EtOH-EG-PEDOT:PSS. A similar effect on RMS
roughness measured by AFM was found for addition of glycerol into the PEDOT:PSS system
[23].

a) PEDOT: PSS b) PEDOT: PSS

10x10 μm2
c) EtOH-EG-PEDOT:PSS d) EtOH-EG-PEDOT:PSS
10x10 μm2
Figure 8: Surface topography of PEDOT:PSS and EtOH-EG-PEDOT:PSS ink film on glass

substrate as studied by a), c) VSI and b), d) AFM
3.4 Rheological Behavior

Viscosity curves (viscosity vs. shear rate) or flow curves (shear stress vs. shear rate) can provide
information about ink processing and performance and thus, they are very important to monitor.
Low shear rates can be related to storage conditions of materials, such as sedimentation, phase
separation and structure retention. High shear rates give information about performance on the
press.
Factors influencing ink flow characteristics typically include pigment content, particle shape
and size distribution, binder system and other additives used in ink formulation. In the case of
polymer solutions, flow properties depend on polymer concentration, temperature, molecular
weight and polydispersity index. Polymer solutions of higher concentrations deviate from
Newtonian behavior more than diluted solutions, due to higher number of chain entanglements
per unit volume [24].
Figure 9 shows the viscosity curves of four different PEDOT:PSS based inks. The initial
polymer solution (PEDOT:PSS) first shows an increase in viscosity with increasing shear rate.
This can be due to orientation of polymer chains with applied shear and thus, increasing the
polymer-polymer interaction up to the point of maximum viscosity, after which it slowly shear-
thins. Addition of EG to PEDOT:PSS solution lowers the concentration of the polymer,
however, as already mentioned, it causes the polymer chains to expand, resulting in stronger
interchain interactions, which led to increased viscosity at lower shear rates.
Addition of surfactant caused even further viscosity increase at lower shear rates. The possible
explanation for such behavior is that because the concentration of surfactant in the ink
formulation was above the CMC (3 g/l), there is a possibility of formation of rod-like micelles
of surfactant molecules [24] and generation of additional entanglements within the polymer
system.

10.00
PEDOT:PSS
EG-PEDOT:PSS
EtOH-EG-PEDOT:PSS
TWEEN80-EG-PEDOT:PSS
1.000
viscosity (Pa.s)
0.1000
0.01000
1.000E-3 0.01000 0.1000 1.000 10.00 100.0 1000 10000
shear rate (1/s)
Figure 9: Viscosity curves for different PEDOT:PSS based inks
In order to quantitatively compare measured curves, mathematical model function fittings can
be used to describe flow curves (shear stress, σ, vs. shear rate, γ) using only a few parameters.
There are a lot of fitting models available. When selecting a flow model, it is important to
consider whether the samples is (i) idealviscous (Newtonian), (ii) shear-thinning, (iii) shear-
thickening, (iv) without a yield point or (v) showing a yield point. For inks tested in this work,
the Herschel-Bulkley model [25] was found to be the best fit (Equation 2). This model
incorporates the elements of Newtonian, Power Law and Bingham models, such as yield stress
(σy), consistency coefficient (K) and power law index (n).
σ = σ y + K ×γ n (2)
Figure 10 shows the flow curves of tested inks fitted with the Herschel-Bulkley model for low
shear rates (0.001 - 1 Pa). At shear rates higher than 1 s-1, results show linear dependence of
shear stress on shear rate with similar slope of around 0.8 and y intercept -0.9 for all inks.
Resulting model parameters are presented in the Table 3.
100.0
PEDOT-PSS
EG-PEDOT:PSS
10.00 EtOH-EG-PEDOT:PSS
TWEEN80-EG-PEDOT:PSS
1.000
shear stress (Pa)
0.1000
0.01000
1.000E-3
1.000E-4
1.000E-4 1.000E-3 0.01000 0.1000 1.000 10.00 100.0 1000 10000
shear rate (1/s)
Figure 10: Flow curves for different PEDOT:PSS based inks fitted with mathematical model
according to Herschel-Bulkley

Table 3: Solid content and parameters of Herschel-Bulkley model for tested PEDOT:PSS based
inks
Solid
K Standard
Sample ID Content σy [Pa] n
[Pa.sn] Error
[%]
PEDOT:PSS 1.29 1.39x10-3 0.122 0.982 5.30
EG-PEDOT:PSS 0.92 1.17x10-3 0.115 0.961 2.89
EtOH-EG-PEDOT:PSS 0.82 3.09x10-4 0.112 0.832 3.80
TWEEN80-EG-PEDOT:PSS 1.21 2.22x10-3 0.118 0.936 4.68
When using the Herschel-Bulkley model, the following applies [26]:

• Newtonian fluids >> σy = 0 and n = 1
• Power law fluid >> σy = 0 and n ≠ 1 (n > 1 for shear-thickening and n < 1 for shear-
thinning fluids)
• Bingham fluid >> σy ≠ 0 and n = 1
It can be seen from the Table 3 that all of the tested inks exhibit yield stress. However, low
values indicate that only a small stress is required to induce the flow. Addition of EG caused
only slight decrease in yield stress and consistency coefficient (also known as viscosity
coefficient), even though the solid content decreased by almost 30%. On the other side, further
decrease in % of solids by ethanol addition led to a decrease in yield stress of nearly one order
of magnitude. Considering the power law index, the lowest value was calculated for EtOH-EG-
PEDOT:PSS, while the rest of tested inks show an n value close to 1. The model fitting software
used in this work also calculates standard error of the fit as shown in the Table 3. A reasonable
fit gives a value of less than about 20 [27].
3.5 Conductivity
The electrical conductivity of the PEDOT:PSS based inks was calculated from resistance
measurements on films casted on a glass substrate. As it was previously reported, addition of
ethylene glycol to PEDOT:PSS dispersion enhances conductivity of the resulting films up to
200 S/cm [12]. In our case, addition of 25 vol% of ethylene glycol resulted in conductivity
increase from 5.3 to 92.1 S/cm. Addition of ethanol to EG-PEDOT:PSS caused a decrease in
conductivity, which can be attributed to lower concentration of polymer in dispersion. On the
other hand, the presence of surfactant TWEEN80 in EG-PEDOT:PSS has led to increased
conductivity (Table 4). Conductivity results also correlate well with roughness of casted films.
The highest roughness measured at millimeter scale was found for PEDOT:PSS and the lowest
for TWEEN80-EG-PEDOT:PSS inks indicating that the surface uniformity is very important to
the charge transport of conductive polymer layer.
Table 4: Conductivity of tested PEDOT:PSS based inks

Sample ID Conductivity [S/cm]
PEDOT:PSS 5.3 ± 0.1
EG-PEDOT:PSS 92.1 ± 0.1
EtOH-EG-PEDOT:PSS 62.4 ± 0.2
TWEEN80-EG-PEDOT:PSS 115.6 ± 2.1
Conductive polymers continue to gain importance in many electronics applications, due to the
possibility of processing from solution, which enables the use of printing technologies for their
deposition and patterning. Properties of polymer inks need to be often optimized for printing in
order to achieve desired properties of final prints. Additives used traditionally in graphic ink
formulation can work in conductive ink formulation as well, although their effect on electrical
performance is very often not known.

This work studied the effect of ink additives on processing and final properties of PEDOT:PSS
based polymer inks. PEDOT:PSS is a conductive polymer widely studied for printed electronics
applications. The surface tension of PEDOT:PSS aqueous solutions was reduced by addition of
alcohols (ethylene glycol and ethanol) and non-ionic surfactant. It was found that addition of
ethylene glycol to PEDOT:PSS solution lowers the static surface tension by 11 mN/m. Further
decrease can be achieved by addition of ethanol or surfactants, whereas ethanol showed higher
effectiveness in lowering both, dynamic and static ink surface tension. This was also confirmed
by dynamic contact angle measurements, where the surfactant-containing ink showed higher
contact angle and slower rate on ink spreading than the ink containing ethanol. Efficient ink
spreading and leveling is crucial for smooth and uniform deposition of PEDOT:PSS layers.
Uniformity and smoothness of functional layers for printed electronic devices is even more
important than for graphic inks. In printed electronics, non-uniform film can lead to lower
device performance or even short circuit. Surface topography studies showed that addition of
alcohols into PEDOT:PSS solution helped in improving the resulting film roughness on a
millimeter scale, however the opposite effect was detected at the micrometer range due to
swelling of PEDOT:PSS molecules. Conductivity of PEDOT:PSS films was improved by
addition of ethylene glycol. Rheological tests were performed by using rotational viscometry to
obtain viscosity curves for tested inks, which were further analyzed by using Herschel-Bulkley
flow model. Higher values of yield stress for ethylene glycol containing inks confirmed
increased interaction between PEDOT:PSS chains.
References
1 Makela T. et al., Utilizing Roll-to-Roll Techniques for Manufacturing Source-Drain

Electrodes for All-polymer Transistors, Synthetic Metals, vol. 153, pp. 285. – 288,
2005
2 Hrehorova E.L., Wood K., Pekarovic J., Pekarovicova A., Fleming P.D. and
Bliznyuk V., The Properties of Conducting Polymers and Substrates for Printed
Electronics, Proceedings of IS&T Digital Fabrication DF01, pp. 197 – 202, 2005
3 Fenoll M., Catusse R. and Rousse E., Gravure Printing: Material Characterization
for All-Organic Capacitor, IARIGAI’s 32nd International Research Conference on
Digitalization and Print Media, pp. 83 – 92, 2005
4 Zielke D, Hübler A.C., Hahn U., Brandt N., Bartzsch M., Fügmann U. and
Fischer T., Polymer –based organic field-effect transistor using offset printed source
and drain structures, Appl. Phys. Lett., vol. 87, 123508, 2005
5 Groenendaal L., Jonas F., Freitag D., Pielartzik H. and Reynolds J.R., Poly (3,4-
ethylenedioxythiophene) and Its Derivatives: Past, Present, and Future, Adv.
Mater.,vol. 12, no. 7., pp. 481-494, 2000
6 Kirchmeyer S. et.al., Scientific importance, properties and growing applications of
poly(3, 4-ethylenedioxythiophene), J. Mater. Chem., vol. 15, pp. 2077 – 2088, 2005
7 Huibers P.D.T., Lobanov V.S., Katritzky A.R., Shah D.O. and Karelson M.,
Prediction of Critical Micelle Concentration Using a Quantitative Structure-Property
Relationship Approach. 1. Nonionic Surfactants, Langmuir, vol. 12, pp. 1462 – 1470,
1996
8 Fahlman M. and Salaneck W.R., Surfaces and interfaces in polymer-based
electronics, Surface Science, vol. 500, pp. 904 – 922, 2002
9 Young T., Miscellaneous Works, G. Peacock, ed., (Murray, London), Vol. I, 418 pp,
1855
10 Dupre A., Theoris Mecanique de la Chaleur, (Paris), 368 pp, 1869
11 Ashizawa S., Horikawa R. and Okuzaki H., Effect of Solvent on Carrier Transport
in Poly(3,4-ethylenedioxithiophene)/poly(4-styrenesulfonate), Synthetic Metals, vol.
153, pp.5-8, 2005
12 Ouyang J. et al., On the mechanism of conductivity enhancement in poly (3,4-
ethylene-dioxythiophene):poly(styrene sulfonate) film through solvent treatment,
Polymer, vol. 45, pp. 8443 – 8450, 2004

13 Kirchmeyer S., et al., Intrinsically Conductive Polymers, Kunststoffe, 10, pp. 202–
208, 2005
14 Klaus J.P. et al., U.S. Patent 4,416,148, 1983
15 Andreas J.M., Hauser E.A. and Tucker W.B., Boundary Tension By Pendant
Drops, J. Phys. Chem., vol. 42, pp. 1001, 1938
16 Nutting G.C. and Long F.A., The Change with Time of the Surface Tension of
Sodium Laurate Solutions, Journal of the American Chemical Society, vol. 63, pp. 84
– 88, 1941
17 Thomas P.E. and Raley G.E., U.S. Patent 4,351,784, 1982
18 Caber P.J., Interferometric profiler for rough surfaces, Applied Optics, vol. 32, no.
19, pp. 3438 – 3441, 1993
19 Veeco Metrology Group, WYKO Profilers Technical Reference Manual, Version
2.2.1, 1999
20 Sharma M.K., Surface Phenomena in Coatings and Printing Technology, in Surface
Phenomena and Fine Particles in Water-Based Coatings and Printing Technology,
Plenum Press, NY and London, 1 pp, 1989
21 Podhajny R.M., Surface Phenomena in Coatings and Printing Technology, in
Surface Phenomena and Fine Particles in Water-Based Coatings and Printing
Technology, Plenum Press, NY and London, 41 pp, 1989
22 Rebros M., Fleming P.D. and Joyce M.K., UV-Inks, Substrates and Wetting,
Proceedings of the 2006 TAPPI Coating & Graphic Arts Conference, Atlanta, 2006
23 Snaith H.J., Kenrick H., Chiesa M. and Friend R.H., Morphological and
electronic consequences of modifications to the polymer anode ‘PEDOT:PSS’,
Polymer, vol. 46, pp. 2573 – 2578, 2005
24 Gupta R.K. Polymer and Composite Rheology, Marcel Dekker, Inc., New York, 390
pp, 2000
25 Mezger T.G., The Rheology Handbook, 2nd Ed., Vincentz Network, Hannover, pp.
299, 2006
26 Hemphill T, Campos W. and Pilehvari A., Yield-Power Law Model More
Accurately Predicts Mud Rheology, Oil & Gas Journal, vol. 91, no. 34, pp. 45-50,
1993
27 TA Instruments, Rheology Advantage Data Analysis: Rheometrics Series Getting
Started Guide, Revision E, pp. 60, 2004

View publication stats

Diffractive Glossmeter For Measurement of Dynamic

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Diffractive Glossmeter For Measurement of Dynamic

Uploaded by

Copyright:

Available Formats

See

Diffractive glossmeter for measurement of

Article · January 2008

Hanna Maarit Koivula Martti Toivakka

SEE PROFILE SEE PROFILE

Atmospheric Pressure Plasma Surface Treatment (Plastek) View project

Nanosturctured Large-area Antibacterial Surfaces View project

The user has requested enhancement of the downloaded file.

ISSN 1748-0337 (Print)

The TAGA JOURNAL is published by SWANSEA PRINTING TECHNOLOGY LTD on behalf

Reproduction in any form by any means without specific written permission is

SWANSEA PRINTING TECHNOLOGY LTD

TAGA JOURNAL VOL. 4 © 2008 SWANSEA PRINTING TECHNOLOGY LTD

© 2008 SWANSEA PRINTING TECHNOLOGY LTD TAGA JOURNAL VOL. 4

TAGA JOURNAL VOL. 4 © 2008 SWANSEA PRINTING TECHNOLOGY LTD

12 - 24 Water-Borne and Solvent-Based Flexographic Inks –

25 - 34 Characterisation of the Lithographic Printing of

35 - 58 Grey Balance Control with a Re-Purposed GATF Colour

72 - 83 The Effect of Paper Properties on the Color

84 - 100 Factors Affecting the Printing Strength of Kaolin-Based

101 - 115 A Comparison of Densitometric and Planimetric

© 2008 SWANSEA PRINTING TECHNOLOGY LTD TAGA JOURNAL VOL. 4

126 - 148 From Multi Channel Publishing towards a Ubiquitous

149 - 164 The evaluation of colour difference equations and

165 - 177 Effect of Toner Fixing Temperature on Print Properties

192 - 205 Effect of drying temperature profile and paper on

206 - 218 Factors Impacting the Evaluation of Printer Profile

219 - 231 Characterization of Conductive Polymer Inks based on

TAGA JOURNAL VOL. 4 © 2008 SWANSEA PRINTING TECHNOLOGY LTD

The use of non-destructive methods in the study of IJ ink

University of Ljubljana *University of Zagreb

Keywords: Ink-Jet ink, Confocal laser scanning microscopy, Photoacoustics, Cross-

© 2008 SWANSEA PRINTING TECHNOLOGY LTD TAGA JOURNAL VOL.4

2 Materials and Methods

2.1 Types of paper

Table 1: The base characteristics of printing samples.

Sample Sample description Grammage /

2.2 Non-destructive methods

TAGA JOURNAL VOL.4 © 2008 SWANSEA PRINTING TECHNOLOGY LTD

2.3 Destructive method

3 Results and discussion

3.1 Non-destructive methods

© 2008 SWANSEA PRINTING TECHNOLOGY LTD TAGA JOURNAL VOL.4

TAGA JOURNAL VOL.4 © 2008 SWANSEA PRINTING TECHNOLOGY LTD

Figure 2: Surfaces of Z (left) and K (right) magenta prints observed by CLSM.

3.1.2 Confocal Laser Scanning Microscopy (CLSM)

Table 3: Results of CLSM vertical distribution of magenta ink.

© 2008 SWANSEA PRINTING TECHNOLOGY LTD TAGA JOURNAL VOL.4

Figure 4: Optical density spectra changes for different substrate types.

TAGA JOURNAL VOL.4 © 2008 SWANSEA PRINTING TECHNOLOGY LTD

Figure 5: Dependence of normalized photoacoustic intensity at 50 Hz on mean density of

© 2008 SWANSEA PRINTING TECHNOLOGY LTD TAGA JOURNAL VOL.4

TAGA JOURNAL VOL.4 © 2008 SWANSEA PRINTING TECHNOLOGY LTD

3.2 Destructive method

Table 4: Results of vertical distribution of magenta ink

Sample Ink thickness

© 2008 SWANSEA PRINTING TECHNOLOGY LTD TAGA JOURNAL VOL.4

3.3.1 Cross section analysis vs. Slit-Scanning Densitometry

The relevance of information obtained by different non-destructive and destructive

TAGA JOURNAL VOL.4 © 2008 SWANSEA PRINTING TECHNOLOGY LTD

© 2008 SWANSEA PRINTING TECHNOLOGY LTD TAGA JOURNAL VOL.4

Water-Borne and Solvent-Based Flexographic Inks –

* Affiliated at Samrt Print