You are on page 1of 5

ARTICLE IN PRESS

Applied Radiation and Isotopes 68 (2010) 6670

Contents lists available at ScienceDirect

Applied Radiation and Isotopes


journal homepage: www.elsevier.com/locate/apradiso

Neutron back scattering for the search of the Battle of Anghiari


V.R. Bom a,, A. Cosentino b, M. Seracini b, R. Rosa c
a
Delft University of Technology, Department of Applied Physics, Mekelweg 15, 2629 JB Delft, The Netherlands
b
Center of Interdisciplinary Science for Art, Architecture and Archaeology, University of California San Diego, 9500 Gilman Dr. La Jolla, CA 92093, USA
c
ENEA, Casaccia Research Center, Via Anguillarese 301, 00123 Rome, Italy

a r t i c l e in f o a b s t r a c t

Article history: The Battle of Anghiari is a wall painting made by Leonardo Da Vinci around 1505. Its present day
Received 17 June 2009 location is unknown but some indications suggest that the mural might be concealed behind a brick
Received in revised form wall. Test measurements are presented demonstrating that neutron back scattering (NBS) can be used to
30 July 2009
search through the wall for the painting. NBS is a non-destructive technique to establish the presence of
Accepted 31 July 2009
the hydrogen contained in the painting materials that were probably used by Da Vinci.
& 2009 Elsevier Ltd. All rights reserved.
Keywords:
Neutron back scattering
Leonardo Da Vinci
Battle of Anghiari
Non-destructive testing

1. Introduction research was done by Seracini in 2000 using laser scanning,


thermography and ground penetrating radar (Pieraccini et al.,
In 1503 Leonardo Da Vinci accepted a commission from the 2005). The thermographic investigation showed that Vasari had
chief magistrate of the Republic of Florence to paint a large mural built brick walls to support six new murals and the radar
on a wall of the Hall of 500 in the Palazzo Vecchio. This painting measurements provided insight in the internal structure of east
was to commemorate the historic Battle of Anghiari (BoA) in and west walls. The west wall appeared to be homogeneous as
which the Florentine army defeated the Milanese one in 1440 and radar images showed no traces of an internal interface. However,
would be Leonardo Da Vincis largest and most substantial work, inside the east wall a discontinuity clearly appeared at 15 cm
about three times the size of the famous Last Supper mural he depth, which could correspond to an air gap between the masonry
painted in Milano in 1495. Work began presumably in June 1505 built by Vasari and the original wall, maybe to preserve the BoA
but it would never be completed. Contemporaries describe the mural.
painting as Da Vincis most magnicent work ever and it is of The presence of the BoA behind the masonry front wall might
great interest because the artist used a new experimental painting be conrmed by using non-destructive and non-invasive neutron
technique based on oil. The unnished mural remained in the Hall techniques such as neutron nanosecond analysis (NNA) and
of 500 until 1563 when the architect and painter Giorgio Vasari neutron back scattering (NBS). NNA is based on n; g reactions
undertook renovation of the space. The ceiling and the walls were and can be used to detect the elemental composition of materials
raised and Vasari himself covered the new walls with frescoes and inside the wall. Neutron back scattering is based on the slowing
all traces of the Battle of Anghiari were lost. Some scholars believe down of fast neutrons by hydrogen nuclei. NBS can only detect
that Vasari was a far too great admirer of Leonardo Da Vinci to hydrogenous materials but has a high speed of operation
have destroyed the artwork and that it would have been a simple compared to the NNA technique. The hydrogen in the case of
enough matter to have built a new wall over the work of Leonardo, the BoA search would be present in the materials used by
as was sometimes done in those days. Leonardo Da Vinci such as Greek pitch, linseed oil, walnut oil and
A rst investigation to determine if the BoA was behind one of gesso, which have been used according to original documents.
the walls was carried out in 1975 by Seracini, Newton and Asmus, Gesso is the Italian word for Board chalk (akin to the Greek
using a variety of introspection methods. The research was word gypsum), and is a powdered form of the mineral calcium
unsuccessful due to the lack of appropriate technologies. Further carbonate, CaSO4  2H2 O. Each calcium carbonate molecule nor-
mally contains two water molecules bound into the crystal. Gesso
was traditionally mixed with animal glue, usually rabbit-skin
 Corresponding author. glue, to be used as a primer coat preventing paint from soaking
E-mail address: V.R.Bom@TUDelft.nl (V.R. Bom). into the support layer. Leonardo Da Vinci plastered his wall with

0969-8043/$ - see front matter & 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apradiso.2009.07.025
ARTICLE IN PRESS
V.R. Bom et al. / Applied Radiation and Isotopes 68 (2010) 6670 67

gesso because he tried out an experimental technique using the impossible to build an exact reproduction of the east wall cross
wall as it were a big panel painting, and used oil based paints. section because the pictorial technique used by Leonardo Da Vinci
Gesso was not applied by Vasari. He used mortar instead, made and the composition of the wall on which the mural was painted
from sand and lime, to build the wall and as preparation for his are unknown. DUNBID was evaluated on two different test
frescoes. The Vasari paintings do not interfere with the NBS structures to cope with this limitation: the Delft structure and
method since dried mortar does not contain hydrogen. the ENEA one. They were designed to cover a range of likely
The feasibility of NBS for the search for the BoA has been tested compositions of the east wall. The real east wall, regarding to
using mock walls at the Delft University of Technology in the neutron back scattering, is expected to behave in a way that will
Netherlands and at the ENEA laboratories in Casaccia near Rome, be in between the two structures. Both structures are composed of
Italy. The results of both tests are presented in this paper, leading the three main parts.
to the conclusion that the NBS technique may be applied Front brick wall: Samples of the bricks used by Vasari for the
successfully in the search for the BoA mural and that the method front wall 15  4  30:5 cm are available and so the reproduction
does not constitute any risks neither to the existing Vasari of the front wall was not an issue.
paintings nor to the public in the Hall. Mural: The technique used by Da Vinci can be guessed from the
supplies he bought between August 1504 and April 1505, namely:
gesso, Greek pitch, and white soda. The back wall must rst have
2. Neutron back scattering been covered with wall gesso. A ground layer of Volterrano gesso
mixed with Greek pitch, most likely made to react with the white
Neutron back scattering is a well established method to show soda (sodium hydroxide), may have been applied next on the wall
the presence of hydrogen. It is used, among others, for land mine gesso. Leonardo must have painted a ground color with Alexan-
detection (Brooks et al., 1999; Datema et al., 2001, 2002; Bom drian white and linseed oil on this ground layer, nally followed
et al., 2005). by the pigments, mixed with linseed and walnut oils. It cannot be
A NBS detector operates by irradiating the surface under guessed how much gesso, Greek pitch and linseed oil were applied
investigation with high energy (MeV) neutrons. The neutrons lose on each square meter, therefore the mural of the Delft structure
energy by scattering from atomic nuclei beneath the surface and was made with a high concentration of linseed oil and the ENEA
become thermal after a number of collisions. The thermalization structure mural was made with three different smaller amounts.
process takes far fewer collisions when scattering from hydrogen Back wall: The east wall probably is made of mortar, stones and
as compared to other elements. The concentration of thermal likely clay elements and its average thickness is about 70 cm, but
neutrons in regions containing hydrogen-rich materials will the actual composition is not known. The Delft back wall consists
therefore be relatively high. A thermal neutron detector that of 1 m dry sand, which depth is enough to be considered innite,
monitors the neutron ux coming back from the irradiated surface and the ENEA test structure has a back wall made of bricks (less
will show an increased count rate above hydrogenous regions. An silicon and more aluminum) with a natural moisture level and a
important advantage of the NBS method is the high speed of nite depth. The actual back wall in the Palazzo Vecchio is
operation (Bom et al., 2006). The main limitation of the NBS supposed to be in between these two extreme cases.
method lies in the sensitivity to moisture. The masonry in case of
the BoA search, however, is very dry because the east wall in the 3.1. Delft structure
Hall of 500 is an inner wall and there is no observed leakage from
the roof. The hydrogen containing materials inside the east wall
The Delft structure was built at the Delft test lane facility,
can therefore only be the oils, the resin and the gesso used by
which was especially constructed to test NBS demining systems.
Leonardo Da Vinci.
The facility consists of a box of 3  8  1 m3 lled with sand. The
The NBS system used is the Delft University neutron back-
sand is heated from the bottom to keep it as dry as possible. This
scatter imaging detector (DUNBID). It consists of 16 position
box constitutes the back wall.
sensitive thermal neutron proportional counter tubes that are
The mural was simulated by a plasterboard 200  60 
placed in a tray covering an area of 50  50 cm2 . The position of a
0:9 cm3 laid down on the sand, building a horizontal wall
neutron hit along a tube is determined by charge division, the
structure. The plasterboard is made of a gypsum layer and covered
tube number provides the position in the direction perpendicular
with paper and is assumed to suitably simulate gesso for the
to the tubes. Both positions are fed into a process computer and a
neutrons. A section 60  60 cm2 of the plasterboard was painted
2D image of the back scattered thermal neutron intensity is
with linseed oil 1:38 l=m2 to simulate the mural on the gesso. An
formed, reecting the hydrogen distribution beneath the surface.
aluminum plate 60  60  0:5 cm3 was also used, to evaluate the
There is extensive blurring in the image due to the heavy
sand box background. Three different front walls, made with
scattering of the neutrons. The advantage of using a 2D approach
bricks 6:5  5  20 cm3 , were built on top of the plasterboard
is the enhanced sensitivity for the detection of discontinuities in
and the aluminum plate and tested (see Fig. 1):
comparison to only monitoring the overall count rate. The spatial
resolution is approximately 3  3 cm2. The detection efciency is
(A) Two layers of bricks 260  60  10 cm3 .
almost 1 for neutrons with thermal energies and decreases to
(B) Three layers of bricks 260  60  15 cm3 .
around 104 for 2.5 MeV neutrons. The neutrons are provided by a
(C) Three layers of bricks 260  60  15 cm3 plus a 5 cm
radioisotope 252Cf point source that is positioned in the center of
distance DUNBID-wall.
the detector. The sensitivity for hydrogen detection varies over the
detector area due to the dependence of the neutron ux on the
distance from the source.
3.2. ENEA structure

3. Test structures The ENEA structure was built at the ENEA laboratory in
Casaccia, Rome. The front wall 180  120  15 cm3 was built
The east wall can be regarded as composed of three main with one layer of bricks. The mural was replicated using one of
structural elements: a front wall, the mural and a back wall. It is four plasterboards 100  120  0:9 cm3 of which three were
ARTICLE IN PRESS
68 V.R. Bom et al. / Applied Radiation and Isotopes 68 (2010) 6670

Fig. 1. Layout of the experimental congurations of the Delft test structure.

Fig. 2. Layout of the experimental conguration of the ENEA test structure.

painted with different quantities of linseed oil: 0.41, 0.83 and


1:25 l=m2 .
The back wall was built with two layers of bricks and was 5 cm
away from the front wall. Fig. 2 shows the ENEA structure cross
section. Fig. 3. The count rate excesses above background as fraction of the background
count rate for the three front wall congurations A, B and C mentioned in the text
and the three congurations DELFT-4, DELFT-3 and DELFT-2 (from left to right).
4. Results The drawn and dashed lines reect the trend for the 15-cm bricks with standoff
and the 10-cm bricks front walls, respectively (walls A and C).

4.1. Delft structure

Measurements were done with each of the three front walls (A, a fraction of that background count rate, for the three front wall
B and C, see Section 3.1 and for the following four experimental congurations A, B and C.
congurations (see Fig. 1): The absolute count rate for the background measurements
DELFT-1. Background: DUNBID above the aluminum plate. varies for cases A, B and C as: 1.00, 0.93 and 0.82, respectively,
DELFT-2. Plasterboard: DUNBID above the plasterboard with- showing a decreasing rate as the distance of the detector to the
out linseed oil. back wall (the sand in this case) increases. The plasterboard
DELFT-3. Linseed oil plasterboard: DUNBID between the (conguration DELFT-2) could be detected in all the setups with
plasterboard without and with linseed oil. an excess count rate between 12% and 15%. The plasterboard
DELFT-4. Linseed oil 1:38 l=m2 : DUNBID above the plasterboard painted with linseed oil (conguration DELFT-4) could be detected
with linseed oil. with a 2528% excess count rate. DUNBID could also report a
Counts were collected with an acquisition time of 500 s using a lesser linseed oil amount (conguration DELFT-3) giving an excess
252
Cf source of 1:5  105 n=s. The background count rate was fraction between 19% and 22%. Fig. 3 shows two trend lines for the
270 c/s. Fig. 3 shows the count rate excess above the count rate two front wall congurations A and C. The slope of the line is an
measured at the background position (conguration DELFT-1), as indication for the sensitivity of the method to the presence of
ARTICLE IN PRESS
V.R. Bom et al. / Applied Radiation and Isotopes 68 (2010) 6670 69

linseed oil. The sensitivity is higher the closer the detector is to ENEA-5. Oil1:25 l=m2 : The plasterboard with 1:25 l=m2 lin-
the oil-containing layer as could be expected. seed oil is used.
ENEA-6. Empty=oil1:25 l=m2 : The plasterboard with linseed
4.2. ENEA structure oil 1:25 l=m2 covers half DUNBID while the other half is left empty.
ENEA-7. Plasterboard=oil1:25 l=m2 : DUNBID covers half the
The experimental congurations were: bare plasterboard and half the plasterboard with 1:25 l=m2 board.
ENEA-1. Background: The gap between the front and back wall Fig. 4 shows the relative excess count rates for measurements
is empty. made with the DUNBID detector placed at a standoff distance
ENEA-2. Plasterboard: The plasterboard without oil is inserted from the front wall of 4 and 5 cm. All the data were collected with
between the two walls. an acquisition time of 60 s and a 252Cf source of 2:1  106 n=s. The
ENEA-3. Oil0:41 l=m2 : The plasterboard with 0:41 l=m2 lin- background count rate was 3150 c/s.
seed oil is placed between the two walls. The gure shows that with the NBS method slight variations in
ENEA-4. Oil0:83 l=m2 : The plasterboard with 0:83 l=m2 is oil concentration can be detected. The count rate is strongly
linseed oil placed between the two walls. dependent on the distance between the detector and the wall or
oil. Measurements done without the back wall hardly showed any
effect of plasterboard or oil. A substantial back wall proved to be
necessary to observe the effect. The large difference of the
observed effect between the Delft and the ENEA tests (30% and
15%, respectively) may be explained by the difference between the
back walls used in both setups.

4.3. NBS images

The intensity of the back scattered neutrons can be determined


as a function of position using the position sensitivity of DUNBID.
The resulting images will be blurry because of the scattering of
the neutrons in the wall structures but may still give important
Fig. 4. Excess count rates relative to the background for distances DUNBID-front- information about the distribution of the hydrogenous materials.
wall of 4 and 5 cm. The sensitivity over the image varies strongly as the edges/corners

Fig. 5. Images of the backscattered neutron ux obtained with DUNBID. Corrections for background and sensitivity variations have been applied as mentioned in the text.
The intensity in the images is an indication for the oil content on the boards. Intensity 1.0 corresponds to a bare plasterboard ENEA-2. The two bottom images represent the
results for the congurations ENEA-6 (left) and ENEA-7 (right). All images have the same intensity scale.
ARTICLE IN PRESS
70 V.R. Bom et al. / Applied Radiation and Isotopes 68 (2010) 6670

get less primary neutrons compared to the center because of the 10 m high scaffolding for the experiments in the Palazzo Vecchio.
central position of the source. Corrections have been applied by The dose rate at the oor in that case will be only 0:05 mSv=h,
subtraction of the background image to remove the scattering which is far below the allowed rate for public.
from the bricks followed by a division by the background-
subtracted plasterboard image to correct for the sensitivity
variations. Fig. 5 shows DUNBID images obtained in this way 6. Conclusion
with the ENEA tests. The images have been compressed into 5  5
pixels to improve statistics, giving a resolution of 10  10 cm2 . It is certain that Leonardo Da Vinci made a painting on a wall in
A pixel value of 1.0 would correspond to plasterboard. The two top the Palazzo Vecchio. What materials he used and in what
images show results obtained with the plasterboards with 0:41 concentrations remains uncertain, but an educated guess can be
and 1:25 l=m2 oil concentrations. The pixel values for the made among others from still existing shopping lists. The setups
0:41 l=m2 are around 1.0, indicating that such small oil used in the tests described here were based on these guesses and
concentrations can hardly be distinguished from plasterboard variations in the setups were applied to cover a wide range of
without oil. paint materials and concentrations.
The pixel values for the 1:25 l=m2 board are clearly above 1 The wall with the BoA mural may contain sections with and
showing the effect of the oil. The two bottom images in Fig. 5 without preparation layers and the prepared sections may be
show the two mixed compositions: ENEA-6 (to the left) and ENEA- painted or unpainted. It has been demonstrated that the NBS
7 (to the right). The lack of counts is obvious in the left-hand part device DUNBID may be applied successfully to search of the
of the image of the ENEA-6 conguration where there was no prepared sections, painted or not, because the base layers give a
board. In the image of the ENEA-7 combination the left-right strong signal. The painted sections may also be found since the
difference is still noticeable. concentration of painting materials used by Leonardo Da Vinci
likely is within the range of the concentrations used in these tests.
4.4. Uncertainties DUNBIDs imaging capabilities may prove useful in distinguishing
the various sections. The DUNBID NBS system may thus be
The uncertainties in the results due to counting statistics are successfully applied in the search for the Battle of Anghiari
small: 0.003 for the ENEA tests and 0.004 for the Delft tests. The painting.
main source of errors is the variation in the distance of the NBS
detector to the wall since back scattering from the wall itself is the
main count rate source. A correction for distance variations can be Acknowledgements
applied by measuring the standoff and careful calibration. The
effects of variations in the wall structures, possibly due to uneven We thank Prof. DuVarney, Emory University Department of
brick size or to irregular joints, are small because of the Physics, for suggesting the use of neutron techniques to detect the
considerable scattering and diffusion of the neutrons through presence of pigments behind the masonry front wall in 2005. Prof.
the structures that will average out small scale irregularities. DuVarney has been a consultant to the project since then.
The University of California starting from 2007 has supported
Prof. Seracinis efforts to pursue the search for the lost mural with
5. Safety neutron techniques after appointing him Scientic Director of
CISA3: Center of interdisciplinary Science for Art, Architecture and
The time during which the wall is irradiated in one single Archaeology.
position is of the order of minutes thanks to the high speed of
operation of the NBS method. The radioactivity that is induced in References
the wall structure is therefore only short living and is far below
the legal limits. Bom, V.R., van Eijk, C.W., Ali, A.M., Osman, A., El-Monem, A.A., Kansouh, W.,
The radioactivity of a brick similar to the ones used for the Megahid, R.M., 2006. A feasibility test of land mine detection in a desert
ENEA structure was measured before and after irradiation with environment using neutron back scattering imaging. IEEE Trans. Nucl. Sci. 53
(4), 22472251.
the source that was used for the ENEA tests. Gamma ray Bom, V.R., van Eijk, C.W., Ali, M.A., 2005. DUNBID, the Delft University neutron
measurements were performed of the natural radioactivity for backscattering imaging detector. Appl. Radiat. Isot. 63 (56), 559563 Eighth
24 h. Then the brick was irradiated from a distance as used in the International Conference on Applications of Nuclear Techniques.
Brooks, F.D., Bufer, A., Allie, S.M., 1999. Detection of plastic land mines by neutron
tests for 7 days and measured again for 24 h. Two isotopes were
back scattering. In: Sixth International Conference on Applications of Neutron
formed by this irradiation: 24Na T1=2 14:9 h and 56Mn Science, Crete, Greece. Report IAEA/PS/RC-799, December 1999.
T1=2 2:6 h. These two isotopes decay quickly and were detect- Datema, C., Bom, V.R., van Eijk, C.W.E., 2001. Land mine detection with the neutron
back scattering method. IEEE Trans. Nucl. Sci. 48 (4), 10871091.
able only after an irradiation time of many hours, long enough to
Datema, C., Bom, V.R., van Eijk, C.W.E., 2002. DUNBLAD, the Delft University
reach the saturation activity. DUNBID requires a measurement neutron backscattering landmine detector. In: Proceedings of the Fifth
time that is more than two orders of magnitude smaller. International Symposium on Technology and the Mine Problem. /http://www.
The 2:1  106 n=s Cf neutron source gave a dose rate at a demine.org/SCOT/Papers/pdfpapers.htmlS.
Pieraccini, M., Mecatti, D., Luzi, G., Seracini, M., Pinelli, G., Atzeni, C., 2005. Non-
distance of 1 m of 5 mSv=h. This is below the allowed dose rate of contact intrawall penetrating radar for heritage survey: the search of the Battle
10 mSv=h for radiological workers. DUNBID will be mounted on a of Anghiari by Leonardo Da Vinci. NDT&E Int. 38, 151157.

You might also like