LC Detector

1
Chrom-Ed Book Series
Raymond P. W. Scott
LIQUID CHROMATOGRAPHY DETECTORS
This eBook is protected by Copyright law and International Treaties. All rights are reserved. This book is covered by an End User Licensee Agreement (EULA). The full EULA may be seen at http://www.library4science.com/eula.html.
COPYRIGHT @2003 by LIBRARY4SCIENCE, LLC ALL RIGHTS RESERVED Neither this book or any part may be reduced or transmitted in any form or by any means, electronic or mechanical , including photocopying, microfilming, and recording or by any information storage and retrieved system without permission in writing from the publisher except as permitted by the in-user license agreement. World Wide Web http://www.library4science.com/
Contents
Introduction .............................................................................................................1 Detector Specifications.....................................................................................2 Dispersion in Connecting Tubes................................................................4 Low Dispersion Tubing...............................................................................7 Dispersion in the Detector Sensor Volume.......................................... 10 Apparent Dispersion from Detector Sensor Volume....................... 13 Dispersion Resulting from the Detector Time Constant................... 16 LC Detectors Based on Refractive Index Measurement............................ 19 The Refractive Index Detector .................................................................... 20 The Angle of Deviation Method............................................................. 21 The Fresnel Method .................................................................................. 23 The Christiansen Effect Detector ........................................................... 26 The Interferometer Detector ................................................................... 29 The Thermal Lens Detector .................................................................... 33 The Dielectric Constant Detector........................................................... 35 The UV Detectors ............................................................................................... 40 The UV Absorption Detectors..................................................................... 40 The Fixed Wavelength UV Detector..................................................... 42 The MultiWavelength UV Detector ......................................................... 48 The MultiWavelength Dispersive UV Detector ................................ 49 The Diode Array Detector ........................................................................... 52 The Fluorescence Detector........................................................................... 57 The Single Wavelength Excitation Fluorescence Detector .............. 59 The Multi Wavelength Fluorescence Detector......................................... 62 Transport Detectors............................................................................................ 66 The Moving Wire Detector .......................................................................... 67 The Chain Detector........................................................................................ 69 The Modified Moving Wire Detector......................................................... 70 The Disc Detector........................................................................................... 76 The Evaporative Light Scattering Detector............................................. 78 Liquid Light Scattering Detectors.............................................................. 81 The Low Angle Laser Light Scattering Detector.............................. 83 The Multiple Angle Laser Light Scattering (MALLS) Detector.... 85 The Electrical Conductivity Detector......................................................... 88 The Electrochemical Detector ..................................................................... 93 Electrode Configurations.......................................................................... 94 Electrode Construction............................................................................. 96 The MultiElectrode Array Detector.......................................................100 References...........................................................................................................105
Introduction
Although chromatography was discovered late in the 1890s its development was almost negligible until the 1940s and this was largely due to the lack of an inline sensitive detector. The first, effective inline liquid chromatography (LC) detectors were the refractive index detector reported by Tiselius and Claesson (1) in 1942 and the conductivity detector described by Martin and Randall (2) in 1951. These two devices should have evoked a growth in LC development, but, in the early fifties, gas chromatography (GC) was invented which completely eclipsed the development of LC. It was not until the early 1960s that the renaissance of LC took place, initially based on the use of the refractive index of Tiselius and Claesson. Although a significant number of GC detectors were developed over two or three years, the development of LC detectors was much slower, largely due to the fact that low concentrations of solute in a liquid do not change the properties of a liquid nearly as much as they do a gas. In fact, the development of LC detectors was gradual and arduous. In a similar way to the development of GC there has been a continuous interaction between improved detector performance and improved column performance. Initially, separations monitored by detectors with improved sensitivity permitted a precise column theory to be developed and experimentally substantiated. This allowed new columns to be designed with reduced dispersion and higher efficiencies. The improved efficiencies, however, produced small volume peaks, small, that is, compared with the volume of the detector sensor and the dispersion that took place in the conduits of the detector system..
2 As a consequence, the ultimate efficiency obtainable from the column was determined by the geometry of the fluid conduits of the detector and not its sensitivity. This provoked detector redesign, with smaller sensor volumes, different geometry and shorter connecting tubes between the column and sensor. In turn, these modifications allowed much smaller particles to be used in the column resulting in even lower column dispersion and higher efficiencies. In this way, just as in GC, detector design and column design have interacted over the years to a point where the performance of LC columns are now commensurate with those of GC columns. Unfortunately, even today, there is no LC detector that has an equivalent performance to the flame ionization detector (FID) used in GC. In general, LC detectors have sensitivities of two to three orders of magnitude less than their GC counterparts and linear dynamic ranges one to two orders of magnitude lower. Only highly specific LC detectors have sensitivities that can approach those of GC detectors.
Detector Specifications
Detector specifications are like those for GC detectors and are listed as follows, 1. Dynamic Range 2. Response Index or Linearity 3. Linear Dynamic range 4. Detector Response 5. Detector Noise Level 6. Detector Sensitivity or Minimum Detectable Concentration 7. Total System Dispersion 8. Sensor Dimensions 9. Detector Time Constant 10. Pressure Sensitivity 11. Flow Sensitivity 12. Operating Temperature Range
3 In general the specifications are the same for both GC and LC detectors with the exception of detector dispersion. Although, detector dispersion has a minimal effect on the resolution in GC separations, detector dispersion can actually destroy a separation achieved in an LC column if the system is not designed correctly. Dealing with the other specifications, the dynamic range and linear dynamic range are the same as those defined in book 4. The response index, the measure of detector linearity, can also be determined in exactly the same way, either by the incremental method of calibration, or the logarithmic dilution method. In the logarithmic method of calibration, mobile phase, now a liquid, is passed continuously through an enclosed stirred vessel containing a known mass of solute, the eluent passing directly into the detector. The logarithm of the detector output is plotted against the logarithm of the calculated solute concentration and the magnitude of the response index determined from the slope of the curve in the manner described in book 4. The response, noise and sensitivity are measured in exactly the same way as for GC detectors. Pressure sensitivity and pressure tolerance have a more important significance in LC as in multidimensional LC, the detector may be situated between two or more columns and thus must tolerate pressures up to the input pressure (e.g., several thousand p.s.i). Pressure sensitivity and flow sensitivity are also more important in LC due to the relatively high pressures involved and the sensitivity of many sensors to pressure changes (e.g., the refractive index detector and the UV detector). However, LC columns have a high impedance to flow and so pressure pulses are often smoothed out in the column and do not reach the detector. Dispersion that takes place in a column is very important and will be dealt with in some detail. Dispersion in Detector Sensors There are three sources of dispersion in LC detector sensors, 1. Dispersion from Connecting Tubes(Newtonian) 2. Dispersion from Sensor Cell Volume (Newtonian)
4 3. Dispersion from Sensor Cell Volume ( Dilution) Each of these sources of dispersion are controllable by careful sensor design and employing appropriate cell geometry.
Dispersion in Connecting Tubes
The dispersion that takes place in an open tube results from the parabolic velocity profile that occurs under conditions of Newtonian flow, (i.e. when the velocity is significantly below that which produces turbulence). Under condition of Newtonian flow, the distribution of fluid velocity across the tube adopts a parabolic profile as shown in figure 1. The velocity at the walls is virtually zero and that at the center a maximum. This situation is depicted diagramatically in Figure 1. Newtonian Flow
Tube Walls
Mobile Phase Velocity
Parabolic Velocity Profile
5 Figure 1. The Parabolic Velocity Profile of a Solute Band Passing Through a Tube Due to the relatively high velocity at the center of the tube and the very low velocity at the walls, the center of the band of solute passing down the tube will move ahead of that situated at the walls. The resulting effect of band dispersion is depicted in figure 2.
Tube Walls Initial Band Width Dispersed Band
Figure 2. Band Dispersion Resulting from Newtonian Flow The dispersion in open tubes was examined by Golay (3) and Atwood and Golay (4) and experimentally by Scott and Kucera (5) and Lochmuller and Sumner (6). The variance per unit length of an open tube (H) according to Golay is given by 2D m r 2u H = + u 24 D m where (Dm) is the diffusivity of the solute in the mobile phase, (u) is the linear velocity of the mobile phase, and (r) is the radius of the tube. At relatively high velocities (i.e., at velocities much greater than the optimum velocity of the tube, which will usually be true for all connecting tubes) r 2u H = 24 D m
6 Furthermore, Q = p r 2u where (Q) is the flow rate through the tube. Q 24 p D m Now, (H) is the variance per unit length of the tube but a more useful parameter to the analyst is the volume variance (sv2). This can be derived using the relationship predicted by the Plate Theory (see book 6). Thus, H = s2 v =
( V0 )
n
p r2 l n
p 2r 4l 2 = n
where (Vo) is the volume of the tube and (l) is the length of the tube l Now H = , consequently s 2 v n different length is
p l Q 0.5 2 r sv = 24 Dm
p r4 l Q = p r lH = 24 D m
2 4
Thus, expression for the volume standard deviation ( s v(l) ) for tubes of (1)
Employing equation (1) it is possible to calculate the value of (s v(l) ) for a range of cylindrical connecting tubes of different radii and different lengths. Table 1 Standard Deviation of Connecting Tubes of Different Sizes Connecting Tubes for Liquid Chromatography
Standard Deviation of Tube Dispersion Tube Diameter 0.001 in, 0.00254 cm 0.002 in, 0.00508 cm 0.003 in, 0.00762 cm l=1 cm 22.3 nl 47.6 nl 107 nl l=2 cm 31.5 nl 67.3 nl 151.3 nl l=5 cm 49.9 nl 106.4 nl 239.2 nl l=10 cm 70.5 nl 150 nl 0.34 ml l=15 cm 86.4 nl 184.4 nl 0.41 ml
7
0.005 in, 0.01270 cm 0.010 in, 0.02540 cm 298 nl 1.19 ml 421 nl 1.68 ml 0.67 ml 2.66 ml 0.94 ml 3.76 ml 1.15 ml 4.61 ml
(Dm) is taken as 2 x 10-5 cm2sec-1 and the flow rate at 0.5 ml/min. All values are fairly typical for the normal operation of the chromatographic system near optimum conditions. It is seen from table 1 that the effect of dispersion in connecting tubes is large due to the very low diffusivity of solutes in liquids. It will be shown in book 8 that for the successful use of microbore columns (columns less than 2 mm I.D.) tube dispersion needs to reduced to about 80 nl. Again assuming that to minimize the chance of tube blocking, the limiting minimum I.D. for the connecting tube is made to be 0.003 in (and tubes of this diameter will still easily block) then the connecting tube must be less than 1 cm long. It is clear that the length of the connecting tube between column and detector must be reduced to an absolute minimum. If the tubing diameter is reduced further and the column diameter is increased, then longer tubing lengths may be possible. Alternatively, the resolution of the early peaks can be sacrificed in favor of later eluting peaks which will also allow longer connecting tubes to be used. These techniques to reduce the effect of connecting tube dispersion in LC are common with most manufacturers. The simple solution of designing the chromatographic system such that the detector sensor is situated very close to the end of the column does not appear to be considered a practical option.
Low Dispersion Tubing
In order to avoid dispersion in mobile phase conduits a number of attempts to design low dispersion tubing has been reported. The first attempt was by Halasz et al. (8), who crimped and bent the tube into different shapes to interrupt the Newtonian flow and introduce radial flow within the tube. His devices had limited success and the tubes had a tendency to block very easily.
8 In 1978 Tijssen (9), developed a theory to describe the radial flow that was induced into coiled tubes by the continual change in direction of the fluid as it flowed round the spirals (his theory will be considered in detail in Book 9). Tijssen found that by coiling the tubes significantly reduced dispersion, particularly at high flow rates However, the coils were a little clumsy to form as the radius of the coil was required to be less than 3 times the internal radius of the tube for optimum performance. A more practical system was introduced by Katz and Scott (10), who developed a serpentine form of connecting tube that met the requirement that the radius of the serpentine bends (a/2 in the diagram) was less that 3 times that of the internal radius of the tube. A diagram of a serpentine tube is shown in figure 3.
Figure 3 Low Dispersion Tubing During passage through the tube, the direction of mobile phase flow changed by 180o as it passed from one serpentine bend to another.
Variance per Unit Length ( ml 2 /cm)
0.5 Straight Tube 0.007 in I.D.
1.0
1.5 Serpentine Tube 0.10 in I.D.
1.0
2.0 3.0 Flow rate (ml/min)
4.0
Figure 4 Graphs of Peak Variance against Flow Rate for Straight and Serpentine Tubes This violent change in direction resulted in extensive radial flow which aided radial transfer and greatly reduced the dispersion. This effect is clearly shown by the curves relating the variance against flow rate for straight and serpentine tubes shown in figure 4. It is seen that at high flow rates, the dispersion is reduced by over an order of magnitude by the serpentine tubing relative to the dispersion that occurred in the straight tube. Despite the apparent advantages, low dispersion serpentine tubing appears to have been employed in only one commercial LC detector. It should be pointed out that any conduit system that has low dispersion will also provide very fast heat transfer rates. Serpentine tubing has been also used in commercial column ovens to heat the mobile phase rapidly to the column oven temperature before it enters the column. The serpentine tubing allows effective heat exchange with a minimum of heat exchanger volume to distort the concentration profile of the solvent
10 gradient. The different forms of dispersion profiles that are obtained from various types of connecting tubes used in LC are shown in figure 4.
Dispersion Peak from Serpentine Tube Dispersion Peak from Coiled Tube
Dispersion Peak from Straight Tube (0.25mm I.D.)
Dispersion Peak from Straight Tube (0.18mm I.D.)
Figure 4 Dispersion Profiles from Different Types of Tube These dispersion curves were obtained using a low dispersion UV detector (cell volume, 1.4 ml) and a sample valve with a 1 ml internal loop. All tubes were of the same length and carried the same mobile phase at a flow rate of 2 ml/min. employed. The peaks were recorded on a high speed recorder. The peak from the serpentine tubing is seen to be symmetrical and has the smallest width. The peak from the coiled tube, although still very symmetrical is the widest at the points of inflexion of all four peaks. The peak from the straight tube 0.25 mm I.D. is grossly asymmetrical and has an extremely wide base width. The width and asymmetry is reduced using a tube with an I.D. of 0.18 mm but serious asymmetry remains. Where the design of the chromatograph precludes a close proximity between the column and the detector, the use of low dispersion serpentine tubing may be a satisfactory alternative.
Dispersion in the Detector Sensor Volume
The finite nature of the detector sensor volume can cause peak dispersion and contribute to the peak variance by two processes. Firstly
11 there will be dispersion resulting from the Newtonian flow of fluid through the cell in much the same manner as the flow of a viscous fluid through an open tube. This will furnish a variance similar in form to that predicted by Golay but, as the tube length is small and the tube length to radius ratio much larger than that from a connecting tube, a different equation is necessary to describe the dispersion effect. Secondly, there will be a peak spreading which results from the finite volume of the sensor. If the sensor has a significant volume, the concentration measured will not be that entering the detector cell but the average concentration throughout the cell. Thus, the true profile of the peak can not be monitored. If the sensor volume is significantly smaller than the peak volume the effect will merely give the peak an apparent dispersion. However, if the sensor volume becomes of the same order of magnitude as the peak volume, then the peak profile will be distorted and resolution will be lost. In the extreme case two peaks could coexist in the sensor at one time and only a single peak will be represented. The effect of viscous flow on dispersion will first be considered. Dispersion in Detector Sensors Resulting from Newtonian Flow Most sensor volumes are cylindrical in shape, are relatively short in length, and have a relatively small length-to-diameter ratio. The small length-to-diameter ratio is in conflict with the premises assumed in the development of the Golay equation for dispersion in an open tube. Atwood and Golay (11) extended the theory of dispersion in open tubes to tubes having small length-to-diameter ratio. The theory is complex and not relevant here as, if appropriate cell design is employed, the dispersion from viscous sources will be negligible. Nevertheless, the effect on solute profiles is shown in figure 5.
12
n=30
Sample Concentration
1 0.3 0.1 0.03 0.01
10
1.0 2.0 3.0 Normalized Elution Volume V/V T
Figure 5 Elution Curves Presented as a Function of the Normalized Tube Length It is seen that serious peak distortion can occur but as the length of the sensor cell is increased the distortion is reduced, but the dispersion increases. However, if the conduits to the cell are appropriately designed to produce secondary flow in the cell, then the parabolic velocity profile is destroyed, and the dispersion and peak distortion eliminated. The manner of entry of the mobile phase from the connecting conduits are, consequently, designed to produce this secondary flow and the manner in which this is achieved is shown in figure 6.
13
Optical Window Inlet from Column Mobile Phase Exit
To Waste
Detector Cell Optical Window
Figure 6 The Design of a Modern Absorption Cell The Newtonian flow is distorted by the manner in which the inlet and outlet conduits are connected to and from the cell. Mobile phase enters the cell at an angle that is directed at the cell window. It follows, that the mobile phase flow has to virtually reverse its direction to pass through the cell producing a swirling action which introduces strong radial flow and disrupts the Newtonian flow. The effect also occurs at the exit end of the cell. The flow along the axis of the cell now must reverse its direction to pass out of the port which is accomplished by attaching the exit conduit at an angle to the axis of the cell. Employing this type of entry and exit connections eliminates dispersion resulting from viscous flow.
Apparent Dispersion from Detector Sensor Volume
The detector can only respond to the average value of the solute concentration throughout the sensor cell. At the extreme, if the sensor cell volume was large enough to contain two closely eluted peaks the
14 response would appear as a single peak, albeit very distorted in shape. This extreme condition rarely occurs, but serious peak distortion and loss of resolution can quite often happen. This will be evident when the sensor volume is of the same order of magnitude as the peak volume. The problem can be particularly severe when columns of small diameter are being used. The situation is depicted in figure 7.
1.25E+00
Concentration in Arbitrary Units
1.00E+00
Column Length 3 cm Column Diameter 3 mm Particle Diameter 3 m m Column Efficiency 5000 plates k' of Solute 2
7.50E-01
5.00E-01
2.50E-01
0.00E+00 -0.02
-0.01
0.01
Volume Flow in
ml
Cell Volume 2.5 ml
Figure 7 Effect of Sensor Volume on Detector Output Consider the elution profile of a peak eluted from a column 3 cm long, 3 mm I.D. packed with particles 3 m in diameter as shown in figure 7. If the peak is eluted at a (k') of 2, from figure 7 it is seen that the peak width at the base is about 14 ml wide. The sensor cell volume is 2.5 ml and the portion of the peak in the cell is depicted in the figure. The detector will obviously respond to the mean concentration of the slice
15 contained in the 2.5 ml sensor volume. It is also clear that, if the sensor volume is increased, a larger part of the peak will be contained in the cell. As a consequence, the output will be an average value of an even larger portion of the peak which will produce serious peak distortion. The effect of a finite sensor volume can be easily simulated with a relatively simple computer program and the output from such a program is shown in figure 8.
Column Length 15 cm Column Diameter 1 mm Particle Diameter 5 mm (k') of first eluted peak 1.0 0.1 m l 1.0 m l 2.0 m l 3.0 m l 5.0 m l
Solute Concentration (normalized)
Volume Flow of Mobile Phase
Figure 8 The Effect of Detector Sensor Volume on the Resolution of Two Solutes The example given, is, by far, not the worst case scenario, but is a condition where the detector sensing volume has a very serious effect on the peak profile and, consequently, the resolution. The small bore column produces peaks having relatively small volumes and which are commensurate with the volume of the sensing cell. From figure 8, it is seen that even a sensor volume as small as 1 ml will have a significant effect on the peak width and the maximum resolution will not be obtained from the column. It is clear that the sensor cell volume must be
16 no greater than 2 ml if the column performance is not to be denigrated to a significant extent. To emphasize the effect of cell volume, it should be noted that the results from a sensor cell having a volume of 5 ml are virtually useless. Unfortunately, many commercially available detectors have sensor volumes as great as, if not greater than 5 ml. If small bore columns are to be employed, such sensor volumes must be carefully avoided.
Dispersion Resulting from the Detector Time Constant.
In addition to the sources of dispersion so far discussed, the peak can appear to be further dispersed by the combined time constant of the sensor and its associated electronics. It must be emphasized that the time constant of the system can not actually disperse an eluted peak, but its effect of it on the sensor measurement can produce an apparent peak dispersion. Thus the term appear is used as the solvent profile itself is not changed, only the profile as presented on the recorder or printer. The effect of the detector time constant can be calculated and the results from such a calculation are shown in figure 9.
T' = 0 sec T' = 0.6 sec
Detector Response
T' = 1.5 sec
Time
Figure 9. Peak Profiles Demonstrating Distortion Resulting from Detector Time Constant
17 The undistorted peak, monitored by a detecting system with a zero time constant, is about 4 seconds wide. An LC column operating at a flow rate of 1 ml/min. and having a peak base-width of 4 seconds would represent a peak with a volume of about 67 ml. It follow, that the peaks depicted would represent those eluted fairly late in the chromatogram. However, despite the late elution, the distortion is still quite severe. To avoid distortion of the early peaks the time constant would need to be at least an order of magnitude less. Scott et al. (12) measured the time constants of two photocells and their results are shown in figure 10.
Cadmium Sulphide Photocell Log Decay Curve
Normalized Curve
Decay Curve
0 IP-28 Photomultiplier Log Log Decay Decay Curve Curve Normalized Curve
10
Decay Curve
0.25 Time (seconds)
0.5
18 Figure 10. The Response Curves of Two Photocells The output each photocell to fast transient changes in incident light intensity was monitored with a high speed recorder. The curves for the cadmium sulfide photocell, figure 10 (chosen as an old type, sensor with a very slow response) is shown at the top of the figure. From the slope of the log curve, the time constant was calculated to be about 2.5 seconds. Such an extremely slow response would be impractical for modern chromatographic systems (i.e., two or more peaks could elute within the period of the time constant). The result of the slow response of the cadmium sulfide sensor, would be to cause the peaks merge into a single distorted peak. The performance of the photomultiplier (representing a sensor with a fast response) is shown in the lower curves of figure 10. The time constant, determined from the slope of the log curve, was only 40 milliseconds. A response time of 40 milliseconds is acceptable for most LC separations. Nevertheless in fast LC separations, solutes can be eluted in less than 100 milliseconds in which case an even faster response might be necessary. Contemporary sensors and electronic systems use fast solid state sensors and solid state electronic components. Thus, most commercial detector systems are sufficiently fast for the vast majority of chromatography applications. As a general rule, the overall time constant of an LC detecting system should be less than 50 milliseconds. For specially very fast separations, a lower value of 15 milliseconds may be necessary. Fast sensors and electronics will respond to high frequency noise so the chromatographic system must be designed to reduce short term noise. This may involve magnetic screening to reduce the effect of stray, lowfrequency electromagnetic fields from nearby power supplies and any high energy consuming laboratory equipment. In general, as the peaks in LC separations can be extremely small all sources of dispersion must be taken into account. It follows that in the design of the chromatograph, careful steps must be taken to minimize
19 the effect from such dispersion sources and to ensure the integrity of the separation is maintained.
LC Detectors Based on Refractive Index Measurement
LC detectors range from those that are exclusively non specific (i.e., bulk property detectors, e.g., the refractive index detector) through those that are partially specific (i.e. partial solute property detectors, e.g., the UV detectors) to the totally specific detectors (i.e., solute property detectors, e.g., the fluorescence detector). In general, the sensitivity increases progressively as the detector becomes more specific, the highest sensitivities being obtained from the specific detectors. Refractive index is a bulk property of the column eluent and so detection depends on the solute modifying the overall refractive index of the mobile phase sufficiently to provide a signal twice that of the noise. Bulk property detectors have an inherently limited sensitivity irrespective of the instrumental technique that is used. Consider an hypothetical bulk property detector that monitors the density of the eluent leaving the column. Assume it is required to detect the concentration of a dense material, such as carbon tetrachloride (specific gravity 1.595), at a level of 1 mg/ml in n-heptane (specific gravity 0.684). Let the change in density resulting from the presence of the solute at a concentration of 10-6 g/ml be (D d). It follows, that to a first approximation, X s (d1 - d 2 ) Dd = d1 where (d1) is the density of the solute, carbon tetrachloride, (d2) is the density of the mobile phase, n-heptane and (Xs) is the concentration of the solute to be detected. Thus for the example given,
20 Dd =
(1.595 - 0.684 ) x 10 -6
1.59
= 5.71 x 10 -7 The coefficient of cubical expansion of n-heptane is about 1.6 x 10-3 per oC. The temperature (Dq) that would produce a change equivalent to the presence of carbon tetrachloride at a concentration of 10-6 g/ml can, therefore, be calculated. Thus, 5.71 x 10 -7 o Dq = C 3 1.6 x 10 = 3.6 x10 -4 o C Therefore, to detect a concentration of one part per million of carbon tetrachloride (at a signal to noise ratio of two), then the temperature variation must be maintained below 1.8 x 10-4 oC. Such temperature stability is extremely difficult to maintain and, thus, temperature control will limit the sensitivity obtainable from the detector. Even the heat of adsorption and desorption of the solute on the stationary phase can produce temperature changes of this order of magnitude. Similarly, the density of the contents of the cell will change with pressure and, if there is a significant pressure drop across the cell, also with flow rate. These stability problems apply to all bulk property detectors and, thus, bulk property detectors in general will all have a limited sensitivity (on average for most compounds, this will be about 10-6 g/ml). In addition, even to achieve this sensitivity, the sensor must always be operated under very carefully controlled conditions.
The Refractive Index Detector
One of the first on-line detectors to be developed was the refractive index detector originally described by Tiselius and Claesson (14) in 1942. Despite its limited sensitivity, this detector can be very useful for
21 detecting those compounds that are nonionic, do not adsorb in the UV, and do not fluoresce. Since 1942, there have been many types of refractive index detectors introduced and a number of different optical systems utilized. Only those in common use or having particular interest will be described here.
The Angle of Deviation Method
When a monochromatic ray of light passes from one isotropic medium, (A), to another, (B), it changes its wave velocity and direction. The change in direction is called the refraction and the relationship between the angle of incidence and the angle of refraction is given by Snell's law, namely, n sin (i ) n' B = B = nA sin (r) where (i) (r) (nA) (nB) and (n'B) is the angle of incident light in medium (A), is the angle of refractive light in medium (B), is the refractive index of medium (A), is the refractive index of medium (B), is the refractive index of medium (B) relative to that of medium (A).
Refractive index is a dimensionless constant that normally decreases with increasing temperature. The reported values are usually taken at 20o or 25oC and are mean values measured for the two sodium lines. If the mobile phase is allowed to flow through a hollow prism and a ray of light passes through the prism it will be diverged from its original path and can be focused onto a photocell. If the refractive index of the mobile phase changes, due to the presence of a solute, the angle of deviation of the transmitted light will also alter and the amount of light falling on the photocell will change. A number of manufacturers have employed the angle of deviation method in refractive index detector design.
22
Sample Mirror Lens Mask Light Source
Reference
Zero Adjust
Sensor Recorder Amplifier and Power Supply
Figure 11. The Refractive Index Detector Based on the Angle of Deviation Method of Measurement A diagram of a simple refractive index detector that is based on the angle of deviation method of measurement is shown in figure 11. The differential refractometer monitors the deflection of a light beam caused by the difference in refractive index between the contents of the sample cell and those of the reference cell. A light beam from an incandescent lamp is confined to the region of the cell by an optical mask. A lens collimates the light beam through both the sample and reference cells to a plane mirror. The mirror reflects the beam back through the sample and reference cells to a lens which focuses it onto a photocell. In fact, it is the location of the beam, rather than its intensity, that changes with the refractive index difference between the contents of the two cells. As the position of focus of the beam on the photoelectric cell changes, the electrical output changes which is electronically modified to provide a signal proportional to the concentration of solute in the cell.
23
sucrose glucose maltose xylose lactose maltotriose
Time
Figure 12. Chromatogram from an RI Detector Based on the Angle of Deviation Method of Measurement An example of a separation monitored by a refractive index (RI) detector for a typical sugar application is shown in figure 12.
The Fresnel Method
The relationship between the reflectance from an interface between two transparent media and their respective refractive indices is given by Fresnel's equation, 1 sin 2 (i - r) tan 2 (i - r) R = 2 + 2 sin (i + r) tan 2 (i + r) where (R) is the ratio of the intensity of the reflected light to that of the incident light and the other symbols have the meanings previously assigned to them. n sin(i) Now, = 1 sin(r) n 2 where (n1) is the refractive index of medium (1), and (n2) is the refractive index of medium (2). Consequently, if medium (2) represents the column eluent, any change in (n2) will change (R) (i.e., DR) and, thus, measurement of (DR) will detect
24 changes in solute concentration. The first to utilize this principal of detection was in the construction of a practical detector was Conlon (4). Conlon's device is now obsolete but it illustrates the principle of the Fresnel method of detection very simply. A diagram of Conlon's detector is shown in figure 13.
Light Source Photocell Glass Rod From Column
To Waste
Figure 13 A Simple Detector Based on the Fresnel Method of Refractive Index Measurement The sensing element consists of a rod prism sealed into a tube through which the solvent flows. The rod (6.8 mm in diameter and 10 cm long) is made from a glass rod, bent to the correct optical angle (just slightly less than the critical angle) and an optical flat is ground on the apex of the bend (see figure 13). The optical flat is then sealed into the window of a flow-through cell. The photocell is arranged to be one arm of a Wheatstone bridge and a reference photocell (not shown) which monitors light direct from the cell, is situated in another arm of the bridge. A commercial refractive index detector working on the Fresnel principle is shown diagramatically in figure 14. Light from a tungsten lamp passes through an IR filter (to minimize thermal effects) onto a magnifying assembly and prism that also splits the beam into two.
25
To User View Port Flow Cell Lens/Splitter Prism Flat Mirror IR Filter
Focusing Lenses Photocell
Flat Mirror Lamp
Figure 14 A Diagram of the Optical System of a Refractive Index Detector Operating on the Fresnel Method The two light beams are arranged to pass through the sample and reference cells respectively. Refracted light from the mobile phase/prism surface passes through the prism assembly and focused onto two photocells. The prism is also arranged to reflects some light to an aperture where the surface of the prism can be observed. The photocell outputs are electronically processed and passed to either a potentiometric recorder or a computer data acquisition system. The refractive index range monitored by the device for a given prism is limited and, consequently, there are usually three different prisms available to cover the RI ranges of 1.351.4, 1.311.44 and 1.401.55 respectively. In figure 15 is shown the separation of a series of polystyrene standards monitored by this type of detector.
26
Polystyrene 1. MW 1,650,000 2. Mw 480,000 3. MW 180,000 2 4 4. MW 76,000 5. MW 39,000 6. MW 11,600 3 7. MW 2,900 8. MW 580. 1 5
5 Time (minutes)
10
15
Figure 15 The Separation of Some Polystyrene Standards Using a RI Detector Operating on the Fresnel Method The separation was carried out by size exclusion on a column packed with 5 mm particles and operated at a flow rate 0.8 ml/min. As a result of limited sensitivity and restricted linear dynamic range, the RI detector is only used for those applications where, for one reason or another. all other detectors are inappropriate or impractical. This type of detector does, however, have one particular area of application for which its characteristics make it particularly suitable and that is for monitoring the separation of polymers. This is because for those polymers containing more than six monomer units, the refractive index is proportional to polymer concentration and independent of its molecular weight. Consequently, quantitative estimation of each polymer mixture can be obtained by simple normalization of peak areas and no individual response factors are required. RI detectors have sensitivities of about 1 x 10-6 g/ml, a linear dynamic range of about 200 and a response index (r) lying between 0.97 and 1.03.
The Christiansen Effect Detector
27 This procedure for measuring refractive index arose from the work of Christiansen on crystal filters (14,15). Consider a cell packed with particulate material having the same refractive index as the mobile phase passing through. If a beam of light passes through the cell there will be little of no refraction or scattering. However, if the refractive index of the mobile phase changes, there will now be a refractive index difference between the mobile phase and that of the packing. As a consequence some light will be refracted away from the incident beam and the intensity of the transmitted light will be attenuated. Thus, if the transmitted light is focused onto an appropriate photocell, then any change in refractive index caused by the elution of a solute will produce scattering and a consequent change in electrical output. In practice, he optical dispersions of the media are likely to differ, and consequently the refractive index will only match at one particular wavelength. As a result the fully transmitted light will be largely monochromatic. Light of different wavelengths will be proportionally dispersed depending on the wavelength at which the two media have the same optical dispersion. Thus, a change in mobile phase refractive will change both the intensity of the transmitted light and its wavelength. This device was made by GOW-MAC Inc., who claimed it had a sensitivity of 1 x 10-6 refractive index units (the maximum that cold be expected). This would be equivalent to a sensitivity of 9 x 10-6 g/ml of benzene (refractive index 1.501) eluted in n-heptane (refractive index 1.388). The cell volume was kept to 8 m l (a little large for modern sensors) which was small enough to work satisfactorily with 4.6 mm I.D. LC columns. Different cells packed with appropriate materials were necessary to cover the refractive index range of 1.31 to 1.60. A diagram of the Christiansen detector is shown in figure 5.
28
Solid Packing Eluent from Column To Waste
Achromat Lamp Photocells
Condensor Aperture
Prism To Waste
Reference solvent
Figure 16. The Christiansen Effect Detector In the optical unit there is a pre focused lamp having an adjustable voltage supply to allow low energy operation when the maximum sensitivity is not required. The condensing lens, aperture, achromat and beam splitting prisms are mounted in a single tube which permitted easy optical alignment prevented contamination from dust. The device contains two identical and interchangeable cells. The disadvantage of this detector is that the cells must be changed each time a different mobile phase is chosen in order to match the refractive index of the packing to that of the new mobile phase. The refractive indices of the cell packing can be closely matched to that of the mobile phase by using appropriate solvent mixtures. In most cases solvent mixing can be achieved without affecting the chromatographic resolution significantly (e.g. by replacing a small amount of n-heptane in a mixture with either n-hexane or n-octane depending on whether the refractive index needs to be increased or decreased. However a considerable knowledge of the effect of different solvents on solute retention is necessary to accomplish this procedure
29 successfully. As a result of limitations inherent in his type of detector combined with the general disadvantages of the RI detector per se has not made the Christiansen Effect Detector very popular.
The Interferometer Detector
The interferometer detector was first developed by Bakken and Stenberg (16) in 1971. The response of the detector depends on the change in the effective path length of a beam of light passing through a cell when the refractive index of its contents changes due to the presence of an eluted solute. Light that has passed through the cell is focused on a photocell. Coincidentally a reference beam of light from the same source is focused on the photocell, interference fringes are produced. The fringes change as the path length of one light beam changes with reference to the other, thus, as the concentration of solute increases in the sensor cell during an elution of a peak, a series of electrical pulses will be generated as each fringe passes the photocell. The optical path length (d) of light through the cell depends on the change in refractive index (Dn), and the path length (l), thus, d = Dn l In addition, the number of fringes (N) which move past a given point on the photo cell (or the number of cyclic changes of the central portion of the fringe pattern) is given by, 2Dnl l where (l) is the wavelength of the light employed. The larger the value of (N) for a given (Dn), the more sensitive the detector will be. It follows that, (l) should be made as large as possible. However, this procedure for increasing the sensitivity is limited by the dead volume of the column and the dispersion that can be tolerated before chromatographic resolution is impaired. A diagram of the simple optical system originally employed by the authors is shown in figure 17. N =
30
Plain Mirror From Column Plain Mirror Lamp
Sensor Cell Half Mirror To Waste Photocell
Figure 17 The Original Optical System Used by Bakken and Stenberg in Their Interferometer Detector Light from an appropriate source strikes a half silvered mirror and is divided into two paths. Part of the beam is reflected by a plane mirror back along the same path and onto a photocell.
Time
31 Figure 18 Chromatogram from the Bakken and Stenberg Interferometer Detector The other part of the beam passes through the sensor cell to a plane mirror, where it is reflected back again through the sensor cell to the half silvered mirror that reflects it onto the photocell. Interference takes place with the other half of the light beam on the surface of the photocell. The trace resulting from the elution of 8 m l of dioxane through the cell is shown in figure 18. Each peak shown in figure 18 represents the passage of a fringe across the surface of the photocell. The four interference peaks represents a single chromatographic peak. The number of fringes will be directly proportional to the total change in refractive index, which, in turn, will be proportional to the total amount of solute present. In this form the detector has limited use, but has been developed into a commercially viable instrument called the Optilab DSP by Wyatt Technology Inc. A diagram of the optical system of the Optilab interference detector is shown in figure 19. Light from the source is linearly polarized at -45o to the horizontal plane. Horizontal and vertical polarized light beams are produced and after passing through the Wollaston prism, one beam passes through the sample cell and the other beam through the reference cell. The sample cell beam is horizontally polarized and the reference cell beam is vertically polarized. After passing through the cells, the beams are focused onto a second Wollaston prism and then through a quarter-wave plate which has its fast axis set -45o to the horizontal plane. A beam that is linearly polarized in the fast axis plane after passing through the plate will lead another linearly polarized but orthogonal beam by a quarter of a wavelength. The phase difference results in a circularly polarized beam. Each of the beams focused on the Wollaston prism consists of two such perpendicular beams which, after the quarter wave plate, result in two circularly polarized beams of opposite rotation. These beams will interfere with each other to yield the original linearly polarized beam. A
32 second polarizer is placed at an angle (90 b) to the first, allowing about 35% of the signal to reach the photocell. A filter transmitting light at 546 nm precedes the photocell.
Wollaston Prism Sample Cell Wollaston Prism Interference Filter
Lamp
Lens Polarizer Mask Reference Cell Lens Quarterwave Plate
Photocell Analyzer
Courtesy of Wyatt Technology Figure 19 The Optilab Interference Refractometer Detector If the sample cell contains a higher concentration of solute than the reference cell the refractive index will be higher and the interfering beams will be out of phase. The refractive index difference (Dn) and the phase difference (Dp) are related by 2pLDn Dp = l where (L) is the length of the cell, and (l) is the wavelength of the light. The circularly polarized beams will, therefore, interfere to yield a linearly Dp polarized beam which is rotated radians, and the amplitude of the 2 light striking the photocell (Ap) will be given by D p D p = A o cos b A p = A o cos 90 - b 2 2
33 An extremely high sensitivity is claimed for this system but it is difficult to interpret the data in terms of minimum detectable concentration The smallest cell (1.4 ml) is reported to give a sensitivity of about 2 x 10-7 RI units at a signal-to-noise ratio of two. Consequently, for benzene (RI = 1.501) sensed as a solute in n-heptane (RI=1.388 ) this sensitivity would represent a minimum detectable concentration of 5.6 x 10-5 g/ml. The alternative 7 m l cell would decrease the minimum detectable concentration to about 1 x 10-6 g/ml, similar to that obtained for other refractive index detectors. A number of LC detectors have been developed that are either based on refractive index measurement or function on some physical property of the mobile phase system that is related to the refractive index. Although most are not commercially available, they demonstrate the range of sensing techniques that have been investigated as possible methods of detection.
The Thermal Lens Detector
If a laser is focused on an absorbing substance, the refractive index of the material can be modified in such a way that the medium behaves as a lens. The thermal lens effect was first reported by Gorden et al. (25,26) in 1964 but since that time the phenomenon has been investigated by a number of workers. Thermal lens formation results from extremely weak laser light adsorption The excited-state molecules subsequently decay back to ground state causing localized temperature increases to occur in the sample. Since the refractive index of the medium depends on the temperature, the ensuing spatial variation of refractive index produces an effect which appears equivalent to the formation of a lens within the medium. For most liquids, the temperature coefficient of refractive index is negative and consequently, the insertion of a liquid in the laser beam produces a concave lens that results in beam divergence. Buffet and Momis (27) used the thermal lens effect to develop a small volume
34 detector, a diagram of which is shown in figure 20. The device consists of a heating laser, the light from which is passed directly through the sample via two lens and a half mirror. Another laser, the probe laser, passes light in the opposite direction through one lens, through the sample to the half mirror where the light is reflected onto a photocell.
Half Mirror Heating Laser
Reference Photocell Probe Laser
Lens
Lens
Pin Hole Mask Heat Filter Photocell Filter Sample Cell
Figure 20 The Layout of a Thermal Lens Detector A filter and a pinhole screen is placed between the mirror and the photocell to remove the heating laser light. When an absorbing solute is eluted from the column through the cell, a thermal lens is produced causing the probe light to diverge, and the intensity of the light passing through the pinhole and on to the photocell is reduced. The cell volume can be as little as a few microliters and, thus, would be suitable for use with microbore columns. A sensitivity of 10-6 AU has been claimed for the detector and a linear dynamic range of about three orders of magnitude. The thermal lens detector is, in fact, a special form of the refractive index detector and might, therefore, be considered a universal detector. Nevertheless, like other bulk property detectors, it can
35 not be used with gradient elution or flow programming and has sensitivity that is no better, if as good as, other refractive index detectors.
The Dielectric Constant Detector
T he refractive index of a substance is a complementary property to the dielectric constant and in some circumstances is a direct function of it. For non-polar substances, the relationship between dielectric constant (e) and refractive index (n) is given by e = n2 For semi-polar substances or mixtures of semi-polar substances and nonpolar substances the Lorentz-Lorenz equation applies e -1 n2 - 1 = 2 e +2 n + 2 However, for polar substances and mixtures of polar and semi-polar substances the relationship breaks down and no simple functions describe dielectric constant in terms of refractive index. The more polar the substance, the larger is its dielectric constant. In n o r m a l chromatography (as opposed to reversed phase chromatography) the mobile phase is normally less polar than the solutes being eluted. Thus, the presence of a solute in the mobile phase will increase the dielectric constant of the mobile phase. Conversely, in reversed phase chromatography the solute is usually less polar than the solvent and the dielectric constant of the mobile phase is reduced by the presence of a solute. Thus. a device situated at the end of the column which responds to changes in dielectric constant would act as a chromatography detector. The sensor often takes the form of a cylindrical or parallel plate condenser. The volume of the sensor must be as small as possible to minimize dispersion. In addition, as the sensitivity
36 of the device is proportional to the electrical capacity of the sensor, the capacitor plates must be very close together. A suitable circuit for use in dielectric constant measurement is an electrical "bridge", the detector cell being situated in one arm of the bridge. If the sensor cell has a capacity greater than 100 pF, then a Wein bridge can be used; however such a cell may well have a fairly large volume. For smaller capacity cells, the Schering bridge is more appropriate and a diagram of a Schering bridge is shown in figure 21.
Cell C D
r R
Ro Co C'
Figure 21 The Schering Bridge for the Measurement of Small Capacities No capacitor is ideal, all will have some inductance and resistance in addition to its capacity. In fact, because the plates of the capacitor are situated in the mobile phase, if uninsulated, it is very likely to have a significant resistance component. The current though the resistive component of a conductor is in phase with the applied voltage and the capacity component lags the applied voltage by 90o. Thus, there are two components to be balanced before the output of the bridge (across (D)) can be used to monitor the elution of a solute.
37 The Schering bridge is balanced by the iterative adjustment of (Ro) and (C'). At balance the following relationships will hold: C Co = Ro R and r C' = R Co
The resistance-component of the cell reduces the bridge sensitivity to changes in capacity and thus the plates should be well insulated to prevent conductivity through the mobile phase. The capacity of the sensor can also be measured by making it one component of a resistance/capacity or an inductance/capacity oscillator. The frequency will depend, among other things, on the capacity of the sensor and, in turn, on the dielectric constant of the material between the plates. The frequency general can be heterodyned against a reference oscillator and the frequency difference will then be proportional to the change in capacity and hence the dielectric constant of the mobile phase. Poppe and Kunysten (28) described a dielectric constant detector which included a reference cell for temperature compensation. The cell consisted of two stainless steel plates 2 cm x 1 cm x 1 mm separated by a gasket 50 mm thick. The two cells were identical and clamped back to back, sharing a common electrode.
38
Figure 22 The Sensor of a Dielectric Constant Detector The device was reported to have a sensitivity of 10-6 g/ml for chloroform (e = 4.81) in n-octane. As might be expected, it was found to be very sensitive to pressure changes in the cell (thought due to plate deformation) even when constant flow pumps were employed. The first dielectric constant detector became commercially available in 1979 (29) and was described by Benningfield and Mowery (30). Several applications were reported by Bade et al. (31). A diagram of the sensor is shown in figure 22. Each cell consisted of a concentric cylinder (inner electrode) inside a larger cylinder (the outer electrode) which formed the outer wall of the cell. Both electrodes were made of stainless steel. The two cylinders were electrically isolated with a cylindrical flow path through the cell. The inner cylindrical electrodes were 1.26 cm in diameter and 0.625 cm long separated from the outer cylinder by about 0.009 cm.
39
Methylene Dichloride Trumyristin
Tripalmatin
5 10 Time (minutes)
15
Figure 23 The Separation of Some Triglycerides Monitored by a Dielectric Constant Detector The linear dynamic range of the detector was reported to be 3.5 x 104 . The sensitivity was quoted as about 1 x 10-7 g/ml, which would be close to the theoretical limit for bulk property detectors. An example of the use of the dielectric constant detector to monitor a separation of triglycerides is shown in figure 23. Bulk property detectors have neither the sensitivity nor the linear dynamic range of solute property detectors and are less frequently used in modern LC analyses. None can be used satisfactorily with gradient elution, flow programming or temperature programming and so they restrict the choice of development. They do have certain unique areas of
40 application, some of which have already been mentioned. Their use probably represents less than 5% of all LC analyses.
The UV Detectors
Although over the years a large number of LC detectors have been developed and described, the vast majority of all contemporary LC analyses are carried out using one of four detectors, the UV detector in one of its forms, the electrical conductivity detector, the fluorescence detector and the refractive index detector. In addition, some form of the UV detector probably accounts for 80% of those analyses.
The UV Absorption Detectors
UV absorption detectors respond to those substances that absorb light in the range 180 to 350 nm. Many (but not all) substances absorb light in this wavelength range, including those substances having one or more double bonds (p electrons) and substances having unshared (unbonded) electrons, e.g. all olefins, all aromatics and compounds, for example, containing > C = O , > C = S , N = N groups. The sensor of a UV detector consists of a short cylindrical cell having a capacity between 1 m l and 10 m l through which passes the column eluent. UV light is arranged to pass through the cell and fall on a photoelectric cell (or array). The output from the photocell passes to a modifying amplifier and then to a recorder or data acquisition system. The relationship between the intensity of UV light transmitted through a cell (IT) and the concentration of solute contained by it (c) is given by Beer's Law. I T = I o e -klc or ln (IT) = ln (Io) - kcl
where (Io) is the intensity of the light entering the cell, (l) is the path length of the cell,
41 and (k) is the molar extinction coefficient of the solute for the specific wavelength of the UV light. Differentiating,
d l og dc
IT Io
= - kl
The sensitivity of the detector, as measured by the transmitted light, will be directly proportional to the extinction coefficient (k) and the path length of the cell (l). To increase the sensitivity of the system, (l) must be extended but there is a limit to which (l) can be increased as the cell volume and, in particular, the length of the cell must be restricted. This is necessary to minimize peak dispersion in the sensor and to avoid more than a small fraction of a peak existing in the cell at any one time This problem has already been discussed. To restrict peak dispersion, the radius of the cell must also be reduced as (l) is increased. Thus, less light will fall on the photocell, the signaltonoise ratio will be reduced and thus the detector sensitivity or minimum detectable concentration denigrated. Thus, increasing the detector sensitivity by increasing the path length has limitations and a welldesigned cell involves a careful compromise between cell radius and length to provide the maximum sensitivity. Most modern UV detector sensors have path lengths that range between 1 and 10 mm and internal radii that range from about 0.5 to 2 mm I Now, Log T = k' l c = A Io where (A) is termed the absorbence Now (D A) is sometimes employed to define the detector sensitivity where the value of (D A) is the change in absorbence that provides a signal-to-noise ratio of two. Thus D A = k' l D c
42 where ( D c) is the detector concentration sensitivity or minimum detectable concentration. Thus Dc= DA k' l
Thus. two detectors, having the same sensitivity defined as the minimum detectable change in absorbence, will not necessarily have the same sensitivity with respect to solute concentration. Only if the path lengths of the two sensors are identical will they also exhibit the same concentration sensitivity. This can cause some confusion as it would be expected that two instruments having the same spectroscopic sensitivity would also have the same chromatographic sensitivity. To compare the sensitivity of two detectors given in units of absorbence the path lengths of the cells in each instrument must be taken into account. UV detectors can be used with gradient elution providing the solvents do not absorb significantly over the wavelength range that is being used for detection. In reversed phase chromatography, the solvents usually employed are water, methanol, acetonitrile and tetrahydrofuran (THF), all of which are transparent to UV light over the total wavelength range normally used by UV detectors. In normal phase operation more care is necessary in solvent selection as many solvents that might be appropriate as the chromatographic phase system are likely to absorb UV light very strongly. The n-paraffins, methylene dichloride, aliphatic alcohols and THF are useful solvents that are transparent in the UV and can be used with normal distribution systems (e.g. a polar stationary phase such as silica gel).
The Fixed Wavelength UV Detector
The fixed wavelength UV detector uses light of a single wavelength (or nearly so) which is produced by a specific type of discharge lamp.
43 100
253.7
Mercury Lamp 10
100 Relative Emission
302.2 307.6 213.9
313.2
Zinc Lamp
10
280.1 277.1
100 Cadmium Lamp

228.8
326.1
346.6 340.3 325.4 308.1 313.2
10
226.5 214.4
293.1 288.1 283.6
200
250
300
350
Wavelength in nm Figure 24 Emission Spectra for Three Discharge Lamps The most popular lamp is the low pressure mercury vapor lamp, which generates most of its light at a wavelength of 254 nm. Other lamps that could be used are the low-pressure cadmium lamp which generates the majority of its light at 225 nm and the low pressure zinc lamp that emits largely at 214 nm. None of the lamps are strictly monochromatic and
44 light of other wavelengths is always present but usually at a significantly lower intensity. The emission spectra of the mercury, cadmium and zinc lamps are shown in figure 24. It is seen that to obtain monochromatic light an appropriate filter would be needed. The low pressure mercury light source (wavelength 253.7 nm) provides the closest to true monochromatic light of all three lamps. However, there is light present of significant intensity below 200 nm, but light of such wavelengths is generally absorbed by the mobile phase. The zinc lamp has a major emission line at 213.9 but the emission line at 307.6 is of comparable intensity and a suitable filter would be needed if detection was required to be exclusively at the lower wavelength. The cadmium lamp has a major emission line at 228.8 but light is emitted at both lower wavelengths and at substantially higher wavelengths and so an appropriate filter would again be desirable. Suitable interference filters can be quite expensive to construct, which may account for the unpopularity of these two lamps. They do, however, emit light at wavelengths which would give an increased sensitivity to substances such as proteins and peptides, which might make their use worthwhile in the biotechnology field. A diagram of a typical optical system for a fixed wavelength UV detector is shown in figure 25. Light from the UV source is collimated by a suitable lens and passed through both the sample cell and the reference cell and then on to two photo cells The cells are cylindrical with quartz windows at either end. The reference cell compensates for any absorption that mobile phase might have at the sensing wavelength. The outputs from the two photo cells are passed to a signal modifying amplifier so that output is linearly related to the concentration of solute being detected. For reasons already discussed, modern sensor cells have angular conduits that form a 'Z' shape to reduce dispersion. The UV sensor can be sensitive to both flow rate and pressure changes but this instability can be greatly reduced if the sensor is well thermostatted.
45
To Waste From Column
Quartz Windows
Sample cell
Quartz Lens Low Pressure Mercury Lamp To Waste Quartz Windows Reference Flow
Photo Cells Reference Cell
Figure 25. The Fixed Wavelength UV Detector The fixed wavelength UV detector is one of the most commonly used LC detectors; it is sensitive, linear and relatively inexpensive. Sensitivity (minimum detectable concentration) can be expected to be about 5 x 108 g/ml with a linear dynamic range of about three orders of magnitude for 0.98 < r < 1.02. The separation of some aromatic hydrocarbons by exclusion chromatography on a very high efficiency column (efficiency ca 250,000 theoretical plates) monitored by a fixed wavelength detector is shown in figure 26. All the solutes are distinctly resolved despite their having molecular weight differences equivalent to only two methylene groups. The peaks from such columns are only a few microliters in width and so a specially reduced volume sensor cell was necessary to cope with the high efficiencies and allow the consequent improved resolution to be realized. The molecular weight of
46 decyl benzene is 218 and, thus, one methylene group would represent a differential of only 6.4% of the molecular weight.
Column, length, 10m, I.D., 1 mm, mobile phase tetrahydrofuran, flow-rate 30 ml/min., adsorbent Partisil 10. Column efficiency ca. 250,000 theoretical plates. Solutes benzene, ethyl benzene, butyl benzene, hexyl benzene, octyl benzene and decyl benzene. Such columns must be packed in short lengths (about 1 m) and subsequently joined and are thus somewhat tedious to construct.
Figure 26. The Separation of Some Alkyl Benzenes by Exclusion on a High Efficiency Column. A very simple fixed wavelength detector suitable for use in preparative chromatography is shown in figure 27. This detector was invented by Miller and Strusz (32) and originally manufactured by GOW-MAC Instruments. As opposed to detectors used for analytical purposes, detectors for preparative work need to have a very low sensitivity as sample sizes are large and consequently the solute concentrations are very high. If analytical detectors are used for preparative work a portion of the eluent is split from the main stream, diluted with more mobile phase and then passed through the detector. In practice, this is a rather awkward procedure. As seen in figure 27 the column eluent passes through a delivery tube and onto a supporting plate that is usually made
47 of fused quartz, so that adequate UV light can reach the photo cell placed on the other side of the plate.
From Column Glass Plate UV Lamp From Column Glass Plate
Flow of Mobile Phase Over Plate
Mask
Photo Cell
Figure 27. Fixed Wavelength Detector for Preparative Work The liquid flows over the plate and the effective path length of the sensor will be the film thickness which will be unique to the particular solvent used as the mobile phase. The UV lamp is situated above the upper side of the plate and the photo cell on the lower side. A reference photo cell (not shown) is situated close to the lamp and the output used to compensate for changes in light intensity from variations in lamp emission. The short path length results in a low sensitivity and the detector can operate satisfactorily at concentrations as high as 10-2 g/ml (1% w/w), which is ideal for preparative chromatography. A particular advantage of this type of sensor is its very low flow impedance and thus
48 can easily accommodate the high flow rates used in preparative LC. The film thickness does depend, among other factors, on the column flow rate, consequently, precise flow control is necessary for the detector to perform satisfactorily.
The MultiWavelength UV Detector
The multiwavelength detector employs a light source that emits light over a wide range of wavelengths. Employing an appropriate optical system (a prism or diffraction grating), light of a specific wavelength can be selected for detection purposes. Light of a specific wave length wavelength might be chosen where a solute has a absorption maximum to provide maximum sensitivity. Alternatively, the absorption spectra of an eluted substances could be obtained for identification purposes by scanning over a range of wavelengths. The latter procedure, however, differs with the type of multiwavelength detector being used. There are two basic types of multiwavelength detector, the dispersion detector and the diode array detector, the latter being the more popular. In fact, very few dispersion instruments are sold today but many are still used in the field and so their characteristics will be discussed. All multiwavelength detectors require a broad emission light source such as deuterium or the xenon lamp, the deuterium lamp being the most popular. The two types of multi-wavelength detectors have important differences. In the dispersive instrument, the light is dispersed before it enters the sensor cell and thus virtually monochromatic light passes through the cell. However, if the incident light is of a wavelength that can excite the solute and cause fluorescence at another wavelength, then the light falling on the photo cell will contain the incident light together with any fluorescent light that may have been generated. It follows, that the light monitored by the photo cell may not be monochromatic and light of another wavelength, if present, would impair the linear nature of the response. This effect would be negligible in most cases but with certain fluorescent materials the effect could be significant. The diode array detector operates quite a differently. Light of all wavelengths generated
49 by the deuterium lamp is passed through the cell and then dispersed over an array of diodes. Thus, the absorption at discrete groups of wavelengths is continuously monitored at each diode. However, light falling on a discrete diode may not be derived solely from the incident light but may contain light generated by fluorescence excited by light of a shorter wavelength. Unfortunately, this effect is exacerbated by the fact that the cell contents are exposed to light of all wavelengths emitted by the source and so fluorescence is more likely. Thus, under some circumstances, measurement of transmitted light may involve fluorescent light and the absorption spectrum obtained for a substance may be a degraded form of the true absorption curve. The ideal multiwavelength detector would be a combination of both the dispersion system and the diode array detector. This arrangement would allow a true monochromatic light beam to pass through the detector and then the transmitted beam would itself be dispersed again onto a diode array. Only that diode sensing the wavelength of the incident light would be used for monitoring the transmission. In this way any fluorescent light would strike other diodes, the true absorption would be measured and accurate monochromatic sensing could be obtained.
The MultiWavelength Dispersive UV Detector
A diagram of the multiwavelength dispersive UV detector is shown in figure 28. Light from the deuterium lamp is collimated by two curved mirrors onto a holographic diffraction grating. The dispersed light is then focused by means of a curved mirror, onto a plane mirror and light of a specific wavelength is selected by appropriately positioning the angle of the plane mirror. Light of the selected wavelength is then focused by means of a lens through the flow cell and, consequently, through the column eluent. The exit beam from the cell is then focused by another lens onto a photo cell which gives a response that is some function of the intensity of the transmitted light. The detector is usually fitted with a scanning facility that allows the spectrum of the solute contained in the cell to be obtained. There is an inherent similarity between UV spectra of
50 widely different types of compounds, and so UV spectra are not very reliable for the identification of most solutes.
Deuterium Lamp
Curved Mirror
Plane Mirror To Waste Photo Cell From Column
Curved Mirror Plane Mirror Diffraction Grating
Quartz Lenses
Sample Cell
Courtesy of the Perkin Elmer Corporation
Figure 28 The MultiWavelength Dispersive UV Detector The technique can be used, however, to determine the homogeneity of a peak (e.g., by comparing spectra taken from both sides of the peak. Both spectra are normalized and either one is subtract one from the other and the difference is shown to be zero, or the ratio of the two spectra is calculated and the result shown to be unity. A common use of multi-wavelength choice is to enhance the sensitivity of the detector by selecting a wavelength that is characteristically absorbed by the substance of interest. Conversely, a wavelength can be chosen that substances of little interest in the mixture do not adsorb and, thus, make the detector more specific to those substances that do. An example of the use of the variable wavelength UV in this way is afforded
51 by the separation of some carboxylic acids that is monitored by UV absorption at 210 nm. The separation is shown in figure 29. The separation of a series of common fatty acids was carried out on a reversed phase column using water buffered with phosphoric acid as the mobile phase.
1 2 3 5 4 6 7 8
Time (minutes)
Column: Spherisorb Octyl, 25 cm x 4.6 mm I.D., 5 mm particles. Mobile Phase: 0.2 M phosphoric acid. Flow rate 0.8 ml/min. monitored at 210 nm. 1. tartaric acid, 2. lactic acid, 3. malic acid, 4. formic acid, 5. acetic acid, 6. citric acid, 7. succinic acid, 8. fumaric acid. Courtesy of Supelco Inc.
Figure 29. The Separation of Some Carboxylic Acids Monitored by UV Absorption at 210 nm Multi-wavelength dispersive detectors has proved extremely useful, providing adequate sensitivity, versatility and a linear response. As a result of the need for an optical bench inside the instrument, however, it is somewhat bulky In addition, it has mechanically operated wavelength selection and requires a stop/flow procedure to obtain spectra "on-thefly". In contrast, the diode array detector has the same advantages but none of the disadvantages.
52
The Diode Array Detector
The diode array detector also utilizes a deuterium or xenon lamp that emits light over the UV spectrum range. Light from the lamp is focused by means of an achromatic lens through the sample cell and onto a holographic grating. The dispersed light from the grating is arranged to fall on a linear diode array. The resolution of the detector (Dl) will depend on the number of diodes (n) in the array, and also on the range of wavelengths covered (l2 - l1). Thus Dl = l 2 - l1 n
Consequently, the ultimate resolving power of the diode array detector will depend on the semiconductor manufacturer and on how narrow the individual photo cells can be commercially fabricated. A diagram of a diode array detector is shown in figure 30.
53
Achromatic Lens System Outlet
Deuterium Lamp
Photo Diode Array Shutter Flow Cell Eluent from Column
Holographic Grating
Figure 30. The Diode Array Detector Light from the broad emission source is collimated by an achromatic lens system so that the total light passes through the detector cell onto a holographic grating. In this way the sample is subjected to light of all wavelengths generated by the lamp. The dispersed light from the grating is allowed to fall onto a diode array. The array may contain many hundreds of diodes and the output from each diode is regularly sampled by a computer and stored on a hard disc. At the end of the run, the output from any diode can be selected and a chromatogram produced using the UV wavelength that was falling on that particular diode. During chromatographic development, the output of one diode is recorded in real time producing a real time chromatogram. It is seen that by noting the time of a particular peak, a spectrum can be obtained by recalling from memory the output of all the diodes at that particular time.
54
Ratio A 235 \A 245
Ratio = 2.5
Absorbance 254 nm
I max = 274 nm P.I. =0
The chlorthalidone was isolated from a sample of tablets and separated by a reverse phase (C18) on a column 4.6 mm I.D., 3.3 cm long, using a solvent mixture consisting of 35% methanol, 65% aqueous acetic acid solution (water containing 1% of acetic acid). The flow rate was 2 ml/min. Courtesy of the Perkin Elmer Corporation
Figure 31. Dual Channel Plot from a Diode Array Detector Confirming Peak Purity The diode array detector can be used in a number of unique ways and an example of the use of a diode array detector to verify the purity of a given solute is shown in figure 31. The chromatogram monitored at 274 nm is shown in the lower part of figure 31.As a diode array detector was employed, it was possible to ratio the output from the detector at different wavelengths and plot the ratio simultaneously with the chromatogram monitored at 274 nm. If the peak was pure, the ratio of the adsorption at the two wavelengths (those selected were 225 and 245 nm) would remain constant throughout the elution of the entire peak. The upper diagram in figure 31 shows this ratio plotted on the same time scale and it is seen that a clean rectangular peak is observed which unambiguously confirms the purity of the peak for chlorthalidone. The
55 wavelength chosen to provide the confirming ratio will depend on the UV adsorption characteristics of the substance concerned, relative to those of the most likely impurities to be present, consequently the wavelengths must be chosen with some circumspection.
Anthracene
The separation was carried out on a column 3 cm long, 4.6 mm in diameter and packed with a C18 reversed phase on particles 3 m in diameter. Courtesy of the Perkin Elmer Corporation
Figure 32. The Separation of Some Aromatic Hydrocarbons Another interesting example of the use of the diode array detector to confirm the integrity of an eluted peak is afforded by the separation of the series of hydrocarbons shown in figure 32.The separation appears to be satisfactory and all the peaks appear to represent individual solutes; without further evidence, it would be reasonable to assume that all the 250 nm peaks were pure. However, by plotting the adsorption ratio, , for 255 nm the anthracene peak it becomes apparent that the peak tail contains an impurity as the clean rectangular shape of the peak top is not shown. The absorption ratio peaks are shown in figure 33.
56 The ratio peaks depicted in figure 33 clearly indicate the presence of an impurity by the sloping top of the anthracene peak. This is further confirmed by the difference in the spectra for the leading and trailing edges of the peak.
250/255 nm Ratio
Irregular top to peak showing presence of impurity
Figure 33 Curves Relating the Adsorption Ratio, 250 nm , and 255 nm Time Spectra taken at the leading and trailing edge of the anthracene peak are shown superimposed in figure 34. Further work identified the impurity as t-butyl benzene at a level of about 5%.
Spectra of Trailing Edge of Peak Adsorption Units
Spectra of Leading Edge of Peak
57 Figure 34 Superimposed Spectra Taken at the Leading and Trailing Edges of the Anthracene Peak Another example of the use of the diode array detector to demonstrate peak purity is shown in figure 35.
Impure mAU mAU Pure
a
200 240 280 (nm) 320 190 230 270 310 (nm)
b d c e
Time (minutes)
Figure 35 Diode Array Spectra Demonstrating Peak Purity A chromatogram is shown containing five peaks and spectra have been taken of peak (a) and peak (b) halfway up the rising side of each peak, at the top of each peak and halfway down the trailing edge of each peak. The spectra are also included in the figure 35. The impure and pure peak are unambiguously identified illustrating the value of this type of detecting system for analytical purposes. The performance of both types of multi-wavelength detectors are very similar and typical sensitivities would be about 1 x 10-7g/ml (significantly less than the fixed wavelength detector) with a linear dynamic range of about 5000 and a response index lying between 0.97 and 1.03.
The Fluorescence Detector
58 When light is emitted by molecules that are excited by electromagnetic radiation, the phenomenon is termed photoluminescence. If the release of electro-magnetic energy is immediate, or stops on the removal of the excitation radiation, the substance is said to be fluorescent. If, however, the release of energy is delayed, or persists after the removal of the exciting radiation, then the substance is said to be phosphorescent. Fluorescence has been shown to be extremely useful as a detection process and detectors based on fluorescent measurement have provided some of the highest sensitivities available in LC. When a molecule adsorbs light, a transition to a higher electronic state takes place and this absorption is highly specific for the molecules concerned; radiation of a specific wavelength or energy is only absorbed by a particular molecular structure. If electrons are raised to an upper excited single state, due to absorption of light energy, and the excess energy is not immediately dissipated by collision with other molecules or by other means, light will be emitted at a lower frequency as the electron returns to its ground state and the substance is said to fluoresce. As some energy is always lost before emission occurs then, in contrast to Raman scattering, the wavelength of the fluorescent light is always greater than the incident light. Detection techniques based on fluorescence affords greater sensitivity to sample concentration, but less sensitivity to instrument instability, (e.g. sensor temperature and pressure). This is due to the fluorescent light being measured against a very low light background (i.e., against a very low noise level). This is opposite to light absorption measurements where the signal is superimposed on a strong background signal carrying a high noise level. Unfortunately, relatively few compounds fluoresce in a practical range of wavelengths. However, some compounds, including products from foods, drugs, dye intermediates etc., do exhibit fluorescence and can be monitored by fluorescent means. In addition, many substances can be made to fluoresce by forming appropriate derivatives.
59 The optical system of most fluorescent detectors is arranged such that the fluorescent light is viewed at an angle to the exciting incident light beam. This minimizes the amount of incident light that can interfere with the fluorescent signal. Under such circumstances, the fluorescent signal is viewed against a an almost black background and thus, furnishes the maximum signal to noise. A filter can be used to reduce the background light still further by the removal of any stray scattered incident light. The fluorescence signal (If) is given by I f = f I o 1 e -k c l
where (f) is the quantum yields (the ratio of the number of photons emitted and the number of photons absorbed), (Io) is the intensity of the incident light, (c) is the concentration of the solute, (k) is the molar absorbence, (l) is the path length of the cell. Fluorescence detectors can be simple or complex, the simplest consists of a single wavelength excitation source and a sensor that monitors fluorescent light of all wavelengths. For certain samples, this form of fluorescence detector can be very sensitive and relatively inexpensive. However, employing excitation light of a single wavelength and only a broad emission wavelength, it is not very versatile. Conversely, the fluorescence spectrometer fitted with a small sensor cell is far more complex but with both selectable excitation wavelengths and emission wavelengths is extremely versatile. In addition, excitation and emission spectra can be obtained as required.
The Single Wavelength Excitation Fluorescence Detector
60 The single wavelength excitation fluorescence detector is probably the most sensitive LC detector that is available, but is achieved by forfeiting versatility. A diagram of a simple form of the fluorescence detector is shown in figure 36. The excitation light is normally provided by a low pressure mercury lamp which is comparatively inexpensive and provides relatively high intensity UV light at 253.7 nm. Many substances that fluoresce will be excited by light of this wavelength.
Quartz Window Photo Cell Quartz Window Lens
Excitation Light Fluorescent Light From Column To Waste UV Light Source
Figure 36. The Single Wavelength Excitation Fluorescent Detector The excitation light is focused by a quartz lens through the cell. A second lens, set normal to the incident light, focuses the fluorescent light onto a photo cell. A fixed wavelength fluorescence detector will have a sensitivity (minimum detectable concentration at an excitation wavelength of 254 nm) of about 1 x 10-9 g/ml and a linear dynamic range of about 500 with a response index of 0.96 < r <1.04.
61
5
11 2
10 4 6 12 3 8 9 1 14 7 13 15
Column: 2 Pecosphere5C C18 (150 mm x 4.6 mm) in series. Mobile Phase: 90% acetonitrile/10% water. Flow rate: 2.0 ml/min. Detector Fluorescence (Excitation 254 nm total emission sensed). Sample: 20 ml of NBS Standard. 1. Naphthalene 4. Phenanthrene 7. Pyrene 10. Benzo(b)fluoranthene 13. Dibenz(a,h)anthracene 2. Fluorene 5. Anthracene 8. Benzo(a)anthracene 11. Benzo(k)fluoranthene 14.Indeno(1,2,3,cd)pyrene 3. Acenaphthene 6. Fluoranthracene 9. Chrysene 12. Benzo(k)fluoranthene 15. Benzo(ghi)perylene
Courtesy of the Perkin Elmer Corporation
Figure 37. Separation of the Priority Pollutants Monitored by the Simple Fluorescence Detector
62 An example of a separation monitored by a simple fluorescence detector is the separation of a mixture of priority pollutants shown in figure 37. The excitation light was approximately monochromatic at 254 nm and all wavelengths of the fluorescent light was sensed by the photo cell. There are some compromises between the expensive fluorescence spectrometer detector and the single wavelength excitation fluorescence detector. Some have a single monochromators that select the wavelength of the excitation light, others employ a single monochromator to select the emission wavelength or provide emission spectra.
The Multi Wavelength Fluorescence Detector
The multi wavelength fluorescence detector contains two monochromators, one to select the excitation wavelength and the second to select the fluorescence wavelength or produce a fluorescence spectrum A diagram of the multi wavelength fluorescence detector is shown in figure 38. The detector comprises a fluorescent spectrometer fitted with suitable absorption cell that is sufficiently small so as not to degrade the resolution of an LC column. There are two distinctly different light paths one for the excitation light and one for the emitted light. The different light paths are depicted separately, the excitation light in dark blue and the emitted light in light blue. The broad band excitation source (usually a deuterium lamp) is placed at the focal point of an ellipsoidal mirror (shown at the top left hand corner of the diagram). The resulting parallel light beam falls on a toroidal mirror that focuses it onto the grating on the left-hand side of the diagram. This grating selects the wavelength of the excitation light. Light of the selected wavelength passes to a spherical mirror and then to a ellipsoidal mirror (shown at the base of the diagram) which focuses it onto the sample. The excitation light path is in dark blue and is situated on the left-hand side of the diagram. A beam splitter is situated between the spherical mirror and the ellipsoidal mirror (in the center of the
63 diagram) which reflects a portion of the incident light onto another toroidal mirror which focuses it onto the reference photo cell.
Toroidal Excitation Light Source Mirror Photo Cell
Ellipsoidal Mirror
Spherical Mirror
Toroidal Mirror
Spherical Mirror Beam Splitter Reference Cell
Grating that Selects the Wavelength of the Excitation Light Ellipsoidal Mirrors Sample Cell
Grating that Selects the Wavelength of the Emitted Light
Figure 38. The Fluorescence Spectrometer Detector The path of the fluorescent light is in light blue and is largely on the right hand side of the diagram. Fluorescent light from the cell is focused by an ellipsoidal mirror on to a spherical mirror which then focuses the light onto a grating situated (seen at about center right of the figure). The grating selects the specific wavelength of the fluorescent light to be monitored. Light of the selected wavelength passes to a photoelectric cell which monitors its intensity.
64 The instrument is quite complex but is extremely versatile. The use of the detector to optimize both the excitation light and the fluorescence light to provide high selectivity for the Fluoropa derivative of neomycin is shown in figure 39. It is a very good example of the selection of a specific excitation light wavelength and the complementary emission light wavelength to provide maximum sensitivity.
Neomycin
4 6 2 Time (minutes)
Column: Supercosil LC8, 15 cm x 4.6 mm, 5 m m particles: Mobile Phase: tetrahydrofuran : 0.0056M sodium sulfate/0.007M acetic acid/0.01M pentane sulfonate, 3:97. Flow rate: 1.75 ml/min. Post Column reagent: 1L 0.4M boric acid/0.38M potassium hydroxide containing 6 ml 40% Brij35, 4 ml mercaptoethanol, 0.8g o-phthalaldehyde. Flow rate 0.4 ml/min. Mixer 5 cm x 4.6 mm column packed with glass beads. Reactor 10 ft x 0.5 mm knitted Teflon capillary tubing. Reaction Temperature 40oC. Sample 20 ml of a mobile phase extract of a commercial sample. Excitation wavelength 365 nm;emission wavelength 418 nm. Courtesy of Supelco Inc.
Figure 39. Detection of Neomycin OPA Derivative at an Excitation Wavelength of 365 nm and an Emission Wavelength of 418 nm Optimizing excitation and emission light wavelengths to obtain maximum sensitivity for a complex mixture can become quite involved as shown by the separation of some priority pollutants depicted in figure
65 40. The separation was carried out on a column 25 cm long, 4.6 mm in diameter and packed with a C18 reversed phase.
5 Fluorescence Intensity 2 1 4 3 8 9 10 11 12 6 7 15 14 13
Time in Seconds 1 Naphthalene 2 Acenaphthene 3 Fluorene 4 Phenanthrene 5 Anthracene 6 Fluoranthene 7 Pyrene 8 Benz(a)anthracene 9 Chrysene 10 Benzo(b)fluoranthene 11 Benzo(k)fluoranthene 12 Benzo(a)pyrene 13 Dibenz (a,h)anthracene 14 Benzo(ghi)perylene 15 Indeno(123-cd)pyrene
Fluorescence Detector Program

Time (seconds) 0 220 340 510 720 1050 1620 Wavelength of Excitation Light 280 nm 290 nm 250 nm 260 nm 265 nm 290 nm 300 nm Wavelength of Emitted Light 340 nm 320 nm 385 nm 420 nm 380 nm 430 nm 500 nm
66 Figure 40. Separation of a Series of Priority Pollutants with Programmed Fluorescence Detection The mobile phase was programmed from a 93% acetonitrile, 7% water to 99% acetonitrile, 1% water over a period of 30 minutes. The gradient was linear and the flow rate was 1.3 ml/min. The separation illustrates the clever use of wavelength programming to obtain the maximum sensitivity. During development both the wavelength of the excitation light and that of the emission light were changed to provide maximum sensitivity for the particular solute. The detector can provide fluorescence or excitation spectra by arresting the flow of mobile phase when the solute resides in the detecting cell and scanning either the excitation or fluorescent light. (This is the same technique as that used to provide UV spectra with the variable wavelength UV detector). As a consequence, it is possible to obtain excitation spectra at any chosen fluorescent wavelength or fluorescent spectra at any chosen excitation wavelength. Thus, even with relatively poor spectroscopic resolution many hundreds of spectra can be produced, any or all of which (despite many spectra being very similar) can be used to confirm the identify a compound.
Transport Detectors
A transport detector consists of a carrier such as a metal chain, wire or disc that passes continually through the column eluent extracting a sample of the mobile phase containing the solute as a thin film adhering to its surface. The mobile phase is eliminated by evaporation leaving the solute as a coating on the carrier. The carrier is then scanned by an appropriate sensing technique to monitor the residual solute. For example, an FID could be used to sense the pyrolysis products of the solute by heating the carrier and most of the pyrolysis products containing carbon would be detected. The system is obviously restricted to those solutes that are involatile and, in addition, the solvents used for the mobile phase must be volatile (and extremely pure). The former condition is usually met in LC, otherwise the analysis would probably be
67 carried out by GC. The latter condition is normally easy to achieve as there is a wide choice of solvents readily available for LC. The system appears to be ideal but the early models had some disadvantages. The instruments were bulky, expensive and some incorporated a 90strontium source (for the argon detector) all of which were unpopular. In addition, as a result of the basic design, the anticipated high sensitivity was not realized and the apparatus was clumsy and difficult to operate. However, as it was a universal detector and was unaffected by the solvents used, it was readily accepted by the soap and cosmetic industry.
The Moving Wire Detector
The original wire transport detector was developed by James et al. (33) and subsequently manufactured and marketed by Pye Unicam. A diagram of the Pye Unicam detector is shown in figure 41.
Evaporator Oven From Column Nitrogen Coating Block Oxygen Detector Oven Hydrogen Nitrogen Pyrolysis Oven Flow Restrictors
FID
Nitrogen
Feed Spool
Cleaner Oven
Collection Spool
Figure 41 The Pye Unicam Moving Wire detector The carrier wire from a spool passes through an oven at 750C and any residual lubricants remaining on the wire after drawing are burnt off.
68 After passing round a pulley the wire then goes through a coating block where it is coated with a thin film of mobile phase. The wire then passes through an evaporator at about 105C (or a temperature appropriate for the solvent(s) employed). In the evaporation chamber a nitrogen stream is allowed to flow counter-current to the wire movement to improve the rate of solvent evaporation. The wire carrying the solvent-free solute then enters (via a restriction) a pyrolysis tube that is maintained at about 500C. The pyrolysis tube has a nitrogen supply entering at either end which sweeps the tube contents, including the pyrolysis products that are formed, out through a center tube into the FID. The FID was a standard detector and the device functioned well and, as expected, was completely independent of the nature of the solvents used for the mobile phase. The detector sensitivity was disappointing, and found to be little better than the average refractive index detector viz. 5 x 10-6 g/ml. The poor sensitivity resulted from excessive noise (not a weak signal) which may have been caused by high boiling impurities in the solvents, fluctuations in the nitrogen flow and/or irregularities in the pyrolysis process. The linear dynamic range was also found to be less than two orders of magnitude. However, the device did establish the viability of the transport method for LC detection.
69
Synchronous Motor Gold Chain
Evaporation Tunnel
LC Column Gauze Jet Air Inlet FID
Coating Block
Figure 42. The Chain Detector

The Chain Detector
About the same time as James et al . developed the wire transport detector, Haahti and Nikkari (34) described a similar device, simpler in design, that employed a chain loop in place of the wire as a carrier. A diagram of their apparatus is shown in figure 42. The carrier, a gold chain driven by a synchronous motor, passes through a coating block where it is wetted with the column eluent. The chain then enters an evaporator tunnel, is heated, and the solvent volatilized leaving the solute deposited on the chain. The chain then exits the tunnel into the actual flame of an FID.
70
Surfactant
Mineral Oil
Ethyl Alcohol
n-Heptane
Sample: mineral oil and a surfactant, solvent: n-heptane/ethyl alcohol, column: 2 x 300 mm, column packing: silica gel, flow rate: 0.7 ml/min., chart speed: 24 cm/min., evaporator temperature: 150C, nitrogen flow: 30 ml/min., hydrogen flow rate: 25 ml/min., oxygen flow rate: 30 ml/min.
Figure 42. Chromatogram Obtained from the Chain Detector Combustion takes place, ions are produced in the expected manner, and the ion current processed in the usual way. Unfortunately, due to the occlusion of local, high concentrations of solute between the links of the chain, the detector output was extremely noisy and thus the system exhibited a relatively poor sensitivity. A chromatogram obtained with the chain detector is shown in figure 42. The noise spikes on each peak are obvious, which, besides affecting the overall sensitivity of the detector, makes quantitative analysis extremely difficult.
The Modified Moving Wire Detector
71 In the early 1960s, a number of attempts were made to improve the performance of the transport detector. In 1966 Karmen (35) introduced an aspirating system to draw the pyrolysis products into the hydrogen flame detector. Later in 1970, Scott and Lawrence (36) developed the Karman's system further and introduced a modified form of the original moving wire detector. The sensitivity of the original detector, in addition to being degraded by the high noise level, was also limited by the proportion of the pyrolysis products that entered the FID. Excepting synthetic polymers, (which often quantitatively produced monomers) many compounds yielded only a few percent of volatile compounds on pyrolysis. Thus the FID could only sense a very small fraction of the products from the solute deposited on the wire. If the solutes, were completely combusted in oxygen or air, however, then all the carbon in the solute would be converted to carbon dioxide. Furthermore, if the carbon dioxide was then reduced to methane by mixing it with hydrogen and passing it over a nickel catalyst, the carbon dioxide would be quantitatively converted to methane which could be detected by the FID. This procedure would increase the sensitivity of the detector to those substances that gave poor yields on pyrolysis and, in addition, increase the linear dynamic range and possibly provide a predictable response. A diagram of the original moving wire detector modified in this way is shown in figure 43. The hydrogen lines entering the detector body were enlarged to reduce the flow impedance and permit the satisfactory operation of the aspirator. The detector was connected to a 2 in. length of 1/2 in I.D. thin walled stainless steel tube filled with about 2 g of nickel catalyst and then to the aspirator. The wide tube was closed with a loose plug of quartz wool and .
72
Air
Evaporator Oven From Column Coating Block Oxygen
Detector Oven
FID
Nickel Catalyst Molecular Entrainer Hydrogen/Argon Oxygen
Oxygen
Flow Restrictors
Feed Spool
Oxidation Oven
Cleaner Oven
Collection Spool
Figure 43 The Modified Moving Wire Detector The nickel catalyst was prepared by absorbing a saturated solution of nickel nitrate onto 20/40 BS mesh brick dust, decomposing the nitrate at 500C for 3 hours followed by reduction of the nickel oxide so produced to metal in a stream of hydrogen at 250C. The jet/venturi aspirator (supplied commercially as a molecular entrainer) was placed in line with the hydrogen flow to the detector. To improve the aspirating efficiency, the gas used was a mixture of hydrogen and argon, and with this mixture, the jet/venturi pressure drop continuously sucked the combustion gasses into the hydrogen stream. In figure 43 the reduced pressure side of the aspirator is shown connected to the side limb of the oxidation tube and the two tube system used in the original moving wire detector, was replaced by a single tube. The oxygen or air is fed into the center of the tube and, thus, provided both the evaporator flow and the oxidation flow. All tubes were constructed of quartz. The linear dynamic range of the system was determined to be about four orders of magnitude a considerable improvement on the
73 pyrolysis instrument. The response index for a series of compounds of different chemical types ranged from 0.96 to 1.04. and the actual response was found to be proportional to the carbon content of the solutes.
6 Solvent 1 2 3 4 5 7 8 9 10 11 12
Injection
Figure 44 The Separation of Blood Liquids Employing Incremental Gradient Elution and Monitored by the Modified Moving Wire Detector However, due to the limited number of compounds that were tested this relationship should be assumed only with caution. A chromatogram of blood lipids obtained by incremental gradient elution and monitored by the modified detector is shown in figure 44. As incremental gradient elution involves a program of 12 solvents ranging from hydrocarbons,
74 chlorinated hydrocarbons, nitro-paraffins, esters, ketones and alcohols. This separation illustrates the versatility that is provided by this detector for solvent selection. Van Dijk (37), attempted to improve the sensitivity of the detector by using a spray procedure for coating the wire. The column eluent entered an atomizer and spray from the nozzle was directed onto the wire. The spray partly concentrated the solute by evaporation and also increase the load on the wire. A linear dynamic range of about 3 x 103 was claimed for the system and a sensitivity increase of 50 over the original wire transport detector (probably about 3 x 10-6 g/ml). Yang et al. (38) also used a thermal spray for coating the wire and also claimed an increased sensitivity. A heated chamber (through which the conduit from the column passed) was placed above a moving stainless steel belt. The solvent was rapidly brought to its boiling point resulting in spray leaving the exit of the conduit and coating the belt. Yang et al. also employed a photo-ionization detector and an electron capture detector as alternatives to the FID. Compton and Purdy (39) redesigned the Pye Unicam FID by inserting a rubidium silicate glass bead above the flame and thus made its response specific and changed it into a nitrogen phosphorus detector. Stolyhwo et al. [40] employed metal spirals wound on wire and stranded wire to increase the surface area of the carrier to increase the proportion of the column eluent taken into the detector. A detectable mass of 100 ng of triolei was claimed but the actual concentration sensitivity was not reported. Pretorious and Van Rensburg (41) tried to increase the carrier take-up by coating the wire with sodium silicate, kaolin and copper kaolin. some improvement to appears to have been realized. Slais and Krejei (42) replaced the normal FID with the NPD detector and used it to detect chlorine compounds. They mixed the combustion gases with hydrogen and passed the mixture directly into the NPD. At a column flow rate of 0.37 ml/min., the sensitivity of the detector was stated to be about 3 x 10-7 g/sec, which is equivalent, in concentration units, to about 1.6 x 10-6 g/ml.
75 The moving wire detector has also been modified by Dugger (43) for radioactivity detection (e.g., detection of 14 carbon labeled compounds). To detect 14 carbon compounds, the solute on the wire was oxidized to carbon dioxide and the radioactive gas passed to a Geiger-Muller tube. To detect tritium, the tritiated water produced on combustion was passed over heated iron to reduce it to hydrogen and tritium, which was then also passed to a Geiger-Muller tube. More recently De Pea et al. (44, 45) have developed the transport detector still further. A photograph of their detector, produced by Scientific Detectors Ltd., is shown in figure 45.
Figure 45. The Uni-Mass Transport Detector The major novel feature of this transport detector is the use of an oxidized titanium tape as the carrier which has a greatly improved loading capacity and readily wets with both dispersive and highly polar solvents (e.g., n-heptane and water). In addition, the coating, evaporating
76 and pyrolysis system has been made much more compact and the argon detector (which is used as the sensor) is made physically very close to the pyrolyser (ca 2-3 cm) A photograph of the internal layout of the detector is shown in figure 46.
Figure 46. The Internal Lay-Out of the Uni-Mass Detector The performance of this detector remains to be established. It has found great interest in biotechnology field as it responds to all substances and the relative responses to different compounds appear to be within a factor of 5 and thus all components are detected. This detector should have a theoretical limit of detection of about 1x10-9 g/ml (i.e. similar to the fluorescence detector) but this sensitivity has not been realized yet.
The Disc Detector
The disc detector was originally described by Dubsky (46) who used a rotating gauze disc as the carrier. A diagram of the device is shown in figure 47. The rotating disc has a perimeter made of wire gauze and the
77 column exit is situated just above the gauze so that the eluent flows through the gauze coating it with a film of mobile phase. The excess of mobile phase is collected in a container situated below the disc. An infrared lamp, or some other appropriate heater, is placed a little ahead of the point of coating, in the direction of rotation of the disc which evaporates the solvent leaving the solute coated on the gauze.
Gauze Disc
Column
Heater Electrodes
Synchronous Motor Flame FID
Figure 47. The Disc Detector The FID is placed diametrically opposite to the point of coating and the flame jet is situated beneath the gauze in such a manner that the flame itself is in contact with the gauze. The FID electrodes are placed just above the gauze, directly over the flame. The ion current is collected by the electrodes and amplified in the usual way and fed to a recorder or data acquisition system. This arrangement is simple compared with the conventional wire transport detector but although a tenfold increase in sensitivity is claimed, this is difficult to confirm from the published data given. Szakusito and Robinson (47) reported that the metal gauze disc carrier was a source of
78 excessive noise. Specifically, significant "spikes" were caused by local concentrations of solute accumulating at the intersections of the wire mesh during evaporation. These authors employed an alumina disc, 4.5 in diameter, with the edge tapered to 0.25 mm thick. The thin edge was used for coating and detection. It appeared that there a significant reduction in noise, but again, sensitivities were not given in terms that would allow comparison with other detectors. It appeared that in continual use, the pores of the alumina could become blocked with residue from incompletely combusted solutes or mobile phase components and so the life of the alumina disc may well be limited. The disc does appear to be a simpler transport system than the wire or chain but its reliability and sensitivity remain to be established.
The Evaporative Light Scattering Detector
Nebulizer Gas
Eluent from Column Nebulizer Photomultiplier Optical Fibers
Light Source Light Trap
Fan Exhaust
Courtesy of Polymer Laboratories Inc.
Figure 48. A Diagram of the Evaporative Light Scattering Detector
79 The evaporative light scattering detector, as its name implies, utilizes a spray that continuously atomizes the column eluent into small droplets. These droplets are allowed to evaporate, leaving the solutes as fine particulate matter suspended in the atomizing gas. The atomizing gas can be air or, if necessary, an inert gas. The suspended particles pass through a light beam and the scattered light viewed at 45o to the incident light beam by means of a pair of optical fibers. The scattered light transmitted through the fibers is sensed by a photomultiplier and the output electronically processed and passed either to a computer data acquisition system or to a potentiometric recorder. A diagram of the light scattering detector is shown in figure 48.
Eluent and Atomizing Gas
Atomized Stream of Particles
Collimated Light Beam
To Photocell
Courtesy of Polymer Laboratories Inc.

80 Figure 49 The Sensor of a Commercial Evaporative Light Scattering Detector Theoretically, the detector responds to all solutes that are not volatile and as the light dispersion is largely Raleigh scattering, the response should be proportional to the mass of solute present; as a consequence, it is sometimes referred to as the mass detector. For a linear response, the droplet size must be carefully controlled as it also determines the particle size of the dried solutes. Detector sensitivity is claimed to be between 10 and 20 ng of solute. However, in these terms it is difficult to compare with other detectors. The great advantage of the detector, however, is its catholic response and that its output is linearly related to the mass of solute present. However, the magnitude of the response does vary widely between different substances. A diagram of the sensor of the evaporative light scattering detector manufactured by Polymer Laboratories is shown in figure 49.
Courtesy of Polymer Laboratories Inc. Peak 1 2 3 4 5 6 Compound cholesterol ester triglyceride cholesterol unknown phosphatidyl choline phosphatidyl ethanolamine Mass (mg) 5 18 10 10 10 Retent.Time(min). 0.717 1.746 4.687 8.860 10.028 17.390
81 Figure 50. The Separation of Some Lipid Class Materials Monitored by a Evaporative Light Scattering Detector The eluent is atomized in a stream of nitrogen and the finely divided spray passes down a heated chamber during which time the solvent is evaporated. The removal of the solvent produces a stream of particles which then pass through a collimated beam of light. The scattered light at an angle to the incident light is focused onto a photomultiplier tube and the output is processed in an appropriate manner electronically. The device is fairly compact and relatively simple to operate. An example of the results obtained when used for monitoring a general lipid class analysis is shown in figure 50. The minimum detectable mass estimated from this chromatogram appeared to be about 10 ng of solute. To some extent, this detector provides a replacement for the transport detector as it detects all substances irrespective of their optical or electrical properties.
Liquid Light Scattering Detectors
Light scattering detectors differ from evaporative light scattering detectors in that they respond to the light scattered by a polymer or large molecular weight substance present in the column eluent itself. The scattering is measured as it passes through an appropriate sensor cell while illuminated by a high intensity beam of light. The high intensity light source is achieved by the use of a laser (light amplification by the stimulated emission of radiation) that also generates the light at the appropriate wavelength for measurement. There are two forms of the detector: the low angle laser light scattering (LALLS) detector and the multiple angle laser light scattering (MALLS) detector. Both devices are commonly used but the multiple angle laser light scattering detector is more versatile as it provides molecular dimensions as well as the molecular weight of the eluted solute. In the LALLS detector, the scattered light is measured at a very small angle to the incident light (virtually 0o), and consequently the signal can be affected by scattering from contaminating particulate matter present
82 as impurities in the eluent. This can cause significant noise, and thus, reduce the detector sensitivity. Discussions on the subject have been given by Wyatt (48) and some early experimental results reported by D. T. Phillips(49). The ratio of the intensity of the light scattered at an angle (f), (If) to the intensity of the incident light (Io), for Rayleigh light scattering is given by If = a w Rf Io where, (a) is the attenuation constant, (w) is a function of the refractive index, and (Rf) is Rayleigh's constant Thus, Rf = If a w Io
Now the molecular weight (Mw) of the solute is related to the Rayleigh factor by the following expression, Rf Mw = c(K - 2A 2 Rf) where (c) is the concentration of the solute, (A2) is a function of polymer-polymer interactions, and (K) is the polymer optical constant. Substituting for (Rf) If Mw = where K = a w Io If c K - 2B 2 a w Io = If c a w I o K - 2B 2 I f
2 p 2 h2 dh 2 l4 N dc
where (h) is the solvent refractive index,

83 (l) is the wavelength of the light in vacuum, and (N) is Avogadro's number. Thus, a basic relationship can be developed between the molecular weight of the scattering material, the intensity of the scattered light and the physical properties of the materials and equipment being involved. However, constants are in used, the magnitude of which are difficult to determine. In practice a simple graphical procedure is used to determine the molecular weight of the solute without the need to determine all the pertinent constants. Rearranging the equation c(K - 2A 2 Rf) cK 1 = = - 2cA 2 Mw Rf Rf or cK 1 = 2cA 2 + Rf Mw
Now (c), (K), and (Rf) are either known or can all be calculated from cK known data and light scattering measurements; thus, by plotting Rf against (c) a straight line will be produced with the intercept being 1 . Mw
The Low Angle Laser Light Scattering Detector
The optical system of the low angle laser light scattering detector produced by LDC Analytical of the Thermo Instruments Corporation is shown diagramatically in figure 51. To conserve space, a folding prism is used that allows the device to be contained to a reasonable size yet accommodate the length of the laser generator. Light from the laser passes through a diverging lens, through a chopper and then through the folding prism. On leaving the prism the beam passes through some measuring attenuators and a calibrating attenuator shutter and then through the cell. An annular mask is situated between the cell and the relay lens and only allows light scattered in the cell at a low angle to pass
84 to the relay lens. Between the annular shutter and the relay lens is a safety attenuator that ensures that none of the potentially damaging laser light can reach the photomultiplier. The scattered light is focused through a field stop onto the forward detector lens. Between the field stop and the forward detector lens is a prism that allows the scattered light to be viewed through a microscope.
Diverging Lens Chopper LASER Folding Prism Forward Detector Lens Photomultiplier Rear Detector Lens Sensor Aperture
Measuring Attenuators Apertures
Annulus Mask
Cell Windows Opal Condensing Diffuser Lens Relay Viewing Lens Calibrating Microscope Safety Attenuator/ Analyzer Red Attenuator Shutter Field Filter Filter Polarizer Stop Holder Courtesy of LDC Analytical, Thermo Instruments Corporation.
Figure 51 Optical Diagram of a Low Angle Laser Light Scattering Detector Between the forward detector lens and the rear detector lens is A filter holder and an analyzer/polarizer. Finally the light is focused through a sensor aperture to an opal diffuser that spreads the scattered light through a red filter and onto the photomultiplier.
85 The device is frequently used with a refractive index detector in series to coincidentally measure the refractive index of the eluent. This is necessary to calculate (K) from the refractive index. As discussed, the molecular weight of a solute is determined from the intercept of the cK graph relating ( ) to the solute concentration (c) as shown in figure Rf 52. The concentration calculated from the refractive index detector employing the data from prior calibration.
cK Rf 1 Mw
cK 2A 2 Rf
c Figure 52 Determination of Molecular Weight from Low Angle Light Scattering Measurements The detector sensitivity appears to be similar to that of the refractive index detector and with about the same linearity. However, the greatest advantage of this detector is that it can provide molecular weight data for extremely large molecules.
The Multiple Angle Laser Light Scattering (MALLS) Detector
The multiple angle laser light scattering detector differs from the low angle device, in that scattering measurements are made at a number of different angles, none of which are close to the incident light. This significantly reduces problems associated with light scattering from
86 particulate contaminants. Data taken at different angles to the incident light allows the root-mean-square (rms) of the molecular radius r 2
1/ 2
to be calculated in addition to the molecular weight of the substance. The relationship that is used is as follows:
1/ 2 cK = a r2 sin (q )2 + b M w Rf Theory can provide explicit functions for (a) and (b) but these constants are obtained in practice by the use of calibrating standards of known molecular weights and molecular radii. In addition, a photocell will not have precisely the same response to low light intensities and calibration procedures are also necessary to account for their different responses.
The number of different angles of measurement differs with different instruments, and some measure the scattered light intensity at 16 different angles. In general, the more data points taken at different angles, the more precise the results will be. A diagram of a (MALLS) detector system which measures the light scattered at three different angles is shown in figure 53.
Sensor 2 Sensor 1 Sensor 3
LASER Glass Cell
LASER Beam
87
Courtesy of Wyatt Technology Corporation
Figure 53 The Multiple Angle Laser Light Scattering Detector (miniDawn) This device (the miniDawn), manufactured by Wyatt Technology Corporation, contains no mirrors, prisms or moving parts and the light paths are direct and not "folded". Light passes from the laser (wavelength 690 nm) directly through the sensor cell. Light scattered from the center of the cell passes through three narrow channels to three different photocells, set at 45o and 90o and 135o to the incident light. Thus scattered light is continuously sampled at three different angles during the passage of the solute through the cell. A continuous analog output is provided from the 90o sensor and all the sensors are sampled every 2 sec. The molecular weight range extends from 10 3 to 10 6 Daltons and the rms radii from 10 to 50 nm. The total cell volume is about 3 ml and the scattering volume is 0.02 ml. The detector has a sensitivity, defined in terms of the minimum detectable excess Rayleigh ratio of 5 x 10-8 cm-1. This is difficult to translate into normal concentration units but appears to be equivalent to a minimum detectable concentration of about 10-6 g/ml.
Kc R(q )
Slope Gives a Value for the rms Molecular Radius Intercept gives a Value for the Molecular Weight 0 0.5 sin 2( q /2) 1.0
Figure 54. Calibration Curves

88 The relationship between the intensity of the scattered light, the scattering angle and the molecular properties are as follows: cK 1 = 2cA 2 + where P(q ) describes the Rf M w P(f) dependence of the scattered light on the angle of scatter and the other symbols have the meanings previously attributed to them. The relationship between the angle of scattering, (q ), the molecular weight and the rms molecular radius of the solute is obtained using the appropriate equation. Employing suitable reference materials, graphs of the form shown in figure 54 can be constructed to evaluate constants (a) and (b) and thus permit the measurement of the molecular weight and molecular radius of unknown substances.
The Electrical Conductivity Detector
The electrical conductivity detector measures the conductivity of the mobile phase. There is usually background conductivity which must be backed-off by suitable electronic adjustments. If the mobile phase contains buffers, the detector gives a base signal that completely overwhelms that from any solute usually making detection impossible. Thus, the electrical conductivity detector a bulk property detector. and senses all ions whether they are from a solute or from the mobile phase. In order to prevent polarization of the sensing electrodes, AC voltages must be used and so it is the impedance not the resistance of the electrode system that is actually measured. From a physical chemistry stand point the conductivity of a solution is more important than its resistance. However, it is the resistance (impedance) of the electrode system that determines the current across it. The resistance (R) of any conductor varies directly as its length (L) and inversely as its cross sectional area (a). rL Thus, R = a where (r) specific resistance of the conductor.
89 The specific resistance of a conductor is the resistance across two opposite faces of a 1 cm cube made of the conductor material. The specific conductance (k ) of a solute is the reciprocal of the specific resistance, i.e., 1 k = r Thus the conductance of a given solution (C) is given by, ka 1 C = = ohms -1 R L The conduction of all the ions produced by 1 g equivalent of an electrolyte at any particular concentration can be evaluated by imagining two large parallel electrodes, 1 cm apart, and the whole of the solution placed between them. The conductivity of the system is called the equivalent conductance (L). If 1 g equivalent of the electrolyte is dissolved in (v) ml of solution; then it follows that this will cover (v) cm2 of electrode area. It follows that in the above equation (a) becomes (v) and (l) is unity and 1000k L = ohms -1 cm 2 c where (c) is the electrolyte concentration in gram equivalents /liter. or Thus, C = k = Lc 1000
L ca 1 = ohms -1 R 1000 L
Thus, the cell resistance is inversely proportional to the electrolyte concentration (c), the equivalent conductance of the electrolyte and the cell geometry [(a) and (L)]. Thus the cell output (which is, in fact proportional to the resistance change of the cell) must be modified to provide an output linearly related to concentration. The basic sensor, being so simple, has changed little from that first described (1). The sensor volume has been reduced but still consists of
90 two electrodes situated in the column eluent, the resistance (or strictly the impedance) between which is measured by an appropriate electronic circuit. Figure 55 shows three simple forms of the conductivity sensor although many other geometric forms have been described in the literature. The basic sensor is, indeed, very simple, the top design is easily constructed and can have an effective sensing volume of only a few microliters.
From Column
Electrodes To Waste
PTFE Insulating Tube From Column
Sensor Volume
PTFE Waste Tube To Waste
Stainless Steel Tube. (Electrode Grounded)
Stainless Steel Tube. (Electrode to Amplifier) Sensor Volume PTFE Waste Tube
PTFE Insulating Tube From Column
Stainless Steel Tubes. (Isolated Electrodes)
To Waste
Figure 55. The Electrical Conductivity Detector The center sensors are made from short lengths of stainless steel tubing (about 0.020 in I.D. and 1/16 in O.D). The two tubes slide into a PTFE
91 tube and the space (inside the PTFE tube) between the two tubes is sensing volume. The lower sensor is very similar, but two lengths of stainless steel tube are employed, both of which are isolated from any other metal connection by PTFE tubing. The two electrodes are, thus, electrically isolated (this can be advanatgeous for certain types of electronic measuring circuits). These sensors have sensing volumes of only a few nanoliters and are, thus, often used in detectors for capillary electrophoresis. The electrical system consists of a frequency generator (1,000-5,000 Hz) that provides an AC potential across the cell. The sensor is placed in one arm of a Wheatstone bridge as shown in figure 56.
Sensing Cell
Out-of-Balance Signal
AC Supply If a single sensor is employed, it is situated in one arm of the bridge; if, however, a reference sensor is also used, then the two sensors are situated in opposite arms of the bridge. In either case the outofbalance signal can be passed to a precision rectifier and the output either handled by an analog circuit or sampled directly by the computer. Sometimes a variable resistance is situated in one of the other arms of the bridge and is used for zero adjustment to back-off any signal arising from ions contained by the mobile phase.
Figure 56 One Form of the Wheatstone Bridge Used with the Electrical Conductivity Detector
92 The out-of-balance signal is rectified and the DC signal either passed to a nonlinear amplifier or a computer data acquisition system. The nonlinear amplifier is necessary to render the output linearly related to ion concentration. If the output is passed directly to the computer, then the linearization can be carried out with appropriate software and the chromatogram is presented on the printer. An example of a separation monitored with a conductivity detector useing an ion supression technique (see book 15) is shown in figure 57.
Time 10 Minutes Courtesy of Dionex Inc. A proprietary ion exchange column, IonPacCS12, was used and the mobile phase consisted of a 20 nM methanesulphonic acid solution in water. A flow rate of 1 ml/min was employed and the sample volume was 25 ml. 1. Lithium 4. Potassium 2. Sodium 5. Magnesium 3. Ammonium 6. Calcium
Figure 57. Determination of Alkali and Alkaline Earth Cations The electrical conductivity detector has a sensitivity (minimum detectable concentration) of about 5 x 10-9 g/ml and a linear dynamic range of about 200 where 0.97 < r < 1.03. It is used in probably over 95% of all
93 analyses involving ion exchange procedures to separate inorganic and organic ions.
The Electrochemical Detector
The electrochemical detector responds to substances that are either oxidizable or reducible and the electrical output is an electron flow generated by a reaction that takes place at the surface of the electrodes. If the reaction proceeds to completion (exhausting all the reactant) the current becomes zero and the total charge generated will be proportional to the total mass of material that has been reacted. This process is called coulometric detection. If, however, the mobile phase is flowing past the electrodes, the reacting solute will be continuously replaced as the peak passes through the detector. All the time there is solute present between the electrodes, a current will be maintained, albeit varying in magnitude. Until relatively recently, this procedure was that most common employed in electrochemical detection and is called amperometric detection. The electrochemical detector requires three electrodes, the working electrode (where the oxidation or reduction takes place), the auxiliary electrode and the reference electrode (which compensates for any changes in the background conductivity of the mobile phase). The processes taking place at the electrode surface can be very complex; nevertheless, the dominant reaction can be broadly described as follows. At the actual electrode surface the reaction is extremely rapid and proceeds almost to completion. This results in the layer close to the electrode being virtually depleted of reactant. As a consequence, a concentration gradient is established between the electrode surface and the bulk of the solution. This concentration gradient causes solute to diffuse into the depleted zone at a rate proportional to the solute concentration in the bulk of the mobile phase. Thus, the current generated at the electrode surface will be determined by the rate at which the solute reaches the electrode and consequently, as the process is diffusion controlled, will depend on solute concentration and the
94 magnitude of solute diffusivity. The response of the detector or current (i) is described by the following equation, i = n F A K T c ua where (n) is the number of electrons per molecule involved in the reaction, (F) is the Faraday Constant, (A) is the area of the working electrode, (KT) is the limiting Mass Transfer Coefficient, (c) is the solute concentration, (u) is the linear velocity of the mobile phase over the surface of the electrode, and (a) is a constant usually taking a value between 1/3 and 1/2. It is seen that the current (i) (and, thus, the sensitivity), can be raised by either increasing the electrode area, increasing the transfer coefficient or increasing the velocity of the mobile phase past the electrodes. It would appear that increasing the electrode area would be the easiest, however, increasing the electrode surface area while maintaining an amperometric response also increases the noise, often to such an extent that there is an overall reduction in detector sensitivity. Weber and Purdy (50) and Hanekamp et al. (51) showed that under certain conditions a reduction in the sensor size produces a significant increase in signal-to-noise and, thus an increase in sensitivity. Higher flow rates will also increase the rate of solute transfer and the sensitivity which would be an added advantage to miniaturization. However, the sensor will be very flow sensitive and thus the flow rate must be kept very constant and the detector would not be amenable to flow programming development. Miniaturization would also reduce peak dispersion in the sensor.
Electrode Configurations
The electrodes can take a number of different geometric forms which have been described in some detail by Poppe (52). Some examples of different electrode configurations are shown in figure 57. Examples (A) and (B) are two common forms of thin-layer cells. (A) has the working
95 electrode sealed into the cell wall with the reference and auxiliary electrodes situated down-stream to the working electrode. (B) is similar to (A) but with the auxiliary electrode sealed in the wall of the tube opposite the working electrode and, again, the reference electrode downstream. (C) is a wall-jet electrode where the eluent is allowed to impinge directly onto the working electrode which is situated opposite the jet. This arrangement not only increases the value of (u) (the velocity of the liquid passing over the electrode and thus the transfer coefficient (KT)) but also scrubbs the surface of the working electrode helping to reduce the need for frequent cleaning. (D) and (E) are two examples of cylindrical electrodes; in (D) the working electrode is in the form of a rod stretching across the diameter of the sensor cell and in (E) the working electrode comprises an annular ring set in the cell wall.
96 A
Auxilary and Reference Electrodes Downstream
B
Auxilary Electrode Reference Electrodes Downstream
Working Electrode
Working Electrode
C
Flow from Column Auxilary and Reference Electrodes Downstream Working Electrode
D
Working Electrode Working Electrode
Figure 57. Different Electrode Configurations In both cases the auxiliary and the reference electrodes are situated down-stream to the working electrode.
Electrode Construction
The choice of electrode construction material is restricted due to the need for mechanically ruggeness and long term stability. The most common material is carbon paste made from a mixture of graphite and some suitable dielectric substance. This material has the disadvantage that it is soluble in some solvents, although, using special waxes or polymers as dielectric binders to contain the graphite, helps reduce the solubility problem. Vitreous or 'glassy' carbon is an excellent electrode material particularly if organic solvents are to be used and is probably the most
97 popular contemporary electrode material. Glassy carbon is produced by slowly baking a suitable resin at elevated temperatures until it is carbonized and then heating it to a very high temperature to cause vitriation. Vitreous carbon is relatively pure, mechanically strong, has good electrical properties and can be readily cleaned mechanically. It also performs particularly well when operated at a negative potential. Glassy carbon electrodes are preferable to carbon paste electrodes due to their inherent resistance to solvents. Electrochemical detection imposes certain restrictions on the type of chromatography that is employed and the mobile phase that is used. The detecting system requires a conducting mobile phase and thus must contain water. Thus, the majority of 'normal phase' systems are not usable. In addtion, very high solvent concentrations may render it insufficiently conducting. Reversed phase chromatography, however, is ideally suited to electrochemical detection. Nevertheless, certain precautions must be taken for its effective use. The mobile phase must be completely oxygen free which can be removed by bubbling helium through the solvent reservoir. It is also important to remove oxygen from the sample before making an injection. The solvents must also be free of metal ions or serious base line instability will result. Non-aqueous solvents such as pure acetonitrile can be employed but certain salts like tetrabutyl-ammoniumhexafluorophosphate must be added to render the solvent conducting. Basic Electrochemical Detector Electronics A simplified form of the circuit that is in general use is shown in figure 58. The amplifiers can best be considered as FET operational amplifiers such as the TLO81. The auxiliary electrode is held at a fixed potential by the first amplifier, the voltage being selected by the potentiometer (P) that is connected to a regulated power supply. The current flowing through the working electrode is processed by the second amplifier and the output fed to the recorder or data acquisition system. The electrochemical detector as described is extremely sensitive but has certain disadvantages.
98
Regulated Power Supply
Amp.1 P
Amp.2
Feed Back Resistanc e Output
Figure 58 Basic Circuit Used with Electrochemical Detectors The mobile phase must be extremely pure, oxygen free and devoid of metal ions. A more serious problem arises from the adsorption of the oxidation or reduction products on the working electrode surface. As a result, the electrode system must be frequently calibrated to ensure accurate quantitative analysis. In addition, the detector must be regularly dissembled and cleaned (usually by a mechanical abrasion). Many attempts have been made to avoid this contamination problem but, although it has been reduced, it has not been completely eliminated (particularly in the amperometric form of operation). Due to potentially low sensing volume the detector is very suitable for use with small bore columns. An example of the use of the electrochemical detector to monitor the separation of a series of catacholamines is shown in figure 59. It is seen that the detector is very sensitive but its value in g/ml (which would allow some comparison with other detecting methods) can not be evaluated from the data.
99
8 9 1 4 10 2 3 6 5 12 17 15 14 18 11 7 13 16 100 pA 6
15 Time (Minutes)
20
Courtesy of the Perkin Elmer Corporation Column: HC-3 C18 (100 mm x 4.6 mm); mobile phase: aqueous solution of 100 nM formic acid , 0.35 nM octane sulphonic acid, 1.0 nM citric acid 0.10 nM EDTA, 5% acetonitrile, 0.25% v/v diethylamine, pH to 3.10 with potassium hydroxide; flow rate 1 ml/min; detection: oxidative amperometric with glassy carbon electrode at 100 mV potential vs. Ag/AgCl electrode. 1. 3,4 dihydroxymendelic acid 200 pg 2. L-dopa 600 pg 3. vanillymendelic acid 400 pg 4. norepinephrin 200 pg 5. a-methyl dopa 600 pg 6. 3-methoxy,4-hydroxyphenylglycol 400 pg 7. epinephrine 200 pg 8. 3,4-dihydroxybenzylamine 200 pg 9. normetanephrin 400 pg 10. dopamine 200 pg 11. metanephrine 400 pg 12.3,4-dihydroxyphenylacetic acid 200 pg 13. N-methyl dopamine 400 pg 14. tyramine 1 ng 15. 5-hydroxyindole-3-acetic acid 200 pg 16. 3-methoxytyramine 400 pg 17. 5-hydroxytryptamine 200 pg 18. homovanillic acid 400 pg
100 Figure 59. The Separation of Some Catacholamines Monitored by an Electrochemical Detector. The column appears to be packed with 3 micron particles which should provide an efficiency of ca. 16,000 theoretical plates. This high efficiency means the peaks would have a very small volume, thus, even with the small peak masses, the peak concentrations could be relatively high. Quoting sensitivities in terms of the mass contained in each peak will always give an enhanced impression of detector sensitivity when a high efficiency column is used.
The MultiElectrode Array Detector
The success of the electrode array as an LC detector is probably due to the development of the porous carbon electrode. This electrode is made of porous graphitic carbon, which has a very high surface area, is mechanically robust and, more important, is permeable to the mobile phase. As a consequence, flow through electrodes can be constructed. The material ideal for electrochemical detection in a number of ways. As the surface area is greatly in excess of that required for efficient electrochemical reaction, it can be severely contaminated before it fails to function. In fact, as much as 95% of the surface can be contaminated before it requires cleaning. When the electrode becomes sufficiently contaminated to require cleaning (which, according to the manufacturers, may occur between one and three years of continual use), the contamination can be rapidly removed by flushing with nitric acid. The porous graphitic carbon electrode facilitates the construction of electrode arrays. In use, the large surface area of the porous electrode ensures that 100% of the eluted material is reacted. Thus, the electrochemical reaction is no longer amperometric, but now coulometric. This is an important difference and makes the array system practical. The electrode system is shown diagramatically in figure 60. Each electrode unit has a central porous carbon electrode, on either side of which is situated a reference electrode and a auxiliary electrode. As the pressure drop across the porous electrode is relatively small, these electrode units can be connected in series forming an array. Up to 16
101 units can be placed in series and these arrays are commercially available. However, a sensor system that contains as many as 80 electrodes in the space of a few millimeters has also been constructed (53).
Counter Electrodes Counter Electrodes
Reference Electrodes
Reference Electrodes Porous Working Electrodes Courtesy ESA Inc.
Figure 60. The Coulometric Electrode System Employing Porous Graphitic Carbon Electrodes A progressively greater potential is applied sequentially to the electrodes of each consecutive unit. This results in all the solutes migrating through the array until each reaches the unit that has the required potential to permit its oxidation or reduction. A comparison between amperometric and coulometric detection is shown in figure 61. In amperometric detection only a small part of the solute is reacted and so the remainder can proceed to the next electrode system and be detected again. In this way each of the units will detect all the solutes and the graph shown in the upper part of the figure is produced. It is seen that there is no discrimination by the different
102 electrode voltages. Coulometric detection results in the total solute being reacted and thus it will be detected by that unit that has the required potential and not be sensed by other units.
Retention Time
Electrode Voltage
Retention Time
Electrode Voltage Courtesy of the Analyst.
Figure 61 Three Dimensional Graphs Demonstrating the Difference between Amperometric and Coulometric Detection Employing an Electrode Array It follows that distinct peaks will be produced only at those units with the appropriate potentials. The result is a three dimensional graph similar to that shown in the lower part of the figure. It is seen that each unit produces a peak at a potential that is characteristic of the solute being detected. Although the voltage increases progressively from one unit to the next, reaction is not completed at one unit only. This is because reaction will not take place at a specific potential, but over a narrow
103 range of potentials. In practice the signal is usually detected over three contiguous electrode units. For example, the first will oxidize a very small amount of the solute, the second unit will be the dominant unit and oxidize the majority of the solute, while the third unit will oxidize the small remaining quantity of solute.
7 8 6 4 5 3 1 2 Electrode Units
7 8 6 5 3 4 2 1 Electrode Units Courtesy of the Analyst.
Figure 62. Use of the Electrochemical Array to Confirm Compound Identity This produces a characteristic pattern of peaks for a particular solute. The ratio of peak height for the three contiguous electrode units that sense the substance will be different for different substances although they may be reacted at the same three electrodes. This is obviously a
104 method of confirming the identity of the solute and is demonstrated in figure 62. The upper graph shows the reference chromatogram of two standards each showing specific retention times and a specific peak pattern as provided by the electrode array detector. The lower graph is for a similar sample and, although the retention times for the pair of solutes is very similar, the pattern given by the electrode array detector clearly shows that the second compound eluted is not the same as that of the second standard.
27 30 29 17 6 7 21 28 24 11 8
23
12 26
9 10
16 14 25
0 20 3
2 5 1
10 Retention Time (minutes) 4 22
20 19 15 18 13
Mobile Phase 1% Methanol to 40% Methanol in a Phosphate (0.1 mol l-1 buffer with ion pairing (pH 3.4) 1. Dihydroxyphenylacetic acid 2. Dihydroxyphenylethylene glycol 3. L-Dopa 4. Dopamine 5. Epinephrine 6. Guanine 7. Guanosine 8. Homovanallic acid 9. Hydroxybenzoic acid 10. Hydroxyindoleacetic acid 11. Hydroxyphenylacetic acid 12. Hydroxyphenyllactic acid 13. Hydroxytryptophan 16. Metenephrine 17. Methoxyhydroxyphenyl glycol 18. Methoxytyramine 19. N-methylserotonin 20. Norepinephrim 21. Normetenephrine 22. Salsolinol 23. Octopamine 24. Seratomin 25. Tryptophan 26. Tyrosine 27. Uric Acid 28. Vanillic acid
105
14. Kynurenine 15. Melatonin 29. Vanylmandelic acid 30. Xanthine Courtesy of the Analyst.
Figure 11 The Separation of 30 Neuroactive Substances Monitored by an Electrochemical Array The electrode array detector also gives improved apparent chromatographic resolution in a similar way to that of the diode array detector. Two peaks that have not been chromatographically resolved and are eluted together can still be shown as two peaks that are resolved electrochemically and can be quantitatively estimated. Another advantage is that high oxidation potentials can be used without the high background currents and noise that usually accompany such operating conditions. The electrodes that are operating at high voltages are "buffered" by the previous electrodes operating at lower voltages which results in reduced background currents and noise. Another example of the application of the detector to the separation of a number of neuroactive substances (54)is shown in figure 63. It is seen that for certain applications the electrochemical array detector can be extremely useful.Nevertheless, in order to use the detector, the solutes must be amenable to electrochemical reaction and capable of being separated using a mobile phase that will conduct an ion current
References
1. A. Tiselius and D. Claesson, Arkiv. Kemi Minearl. Geol., 15B(No. 18)(1942) 2. A, J. P. Martin and S. S. Randall, Biochem. J., 49(1951)293 4. M. J. E. Golay, "Gas Chromatography 1958", (Ed. D. H. Desty), Butterworths, London (1958)36. 5. J. G. Atwood and M. J. E. Golay, J. Chromatogr., 218(1981)97. 6. R. P. W. Scott and P. Kucera, J. Chromatogr. Sci.,18(1971)641. 7. C. H. Lochmuller and M. Sumner, J. Chromatogr. Sci.,18(1980)159. 8. I. Halasz, H. O. Gerlach, K. F. Gutlich and P. Walkling, US Patent
106 3,820,660, (1974). 9. R. Tijssen, Separ, Sci. Technol., 13(1978)681. 10. E. D. Katz and R. P. W. Scott, J. Chromatogr., 268(1983)169. 11. J. G. Atwood and M. J. E. Golay, J. Chromatogr., 218(1981)97. 12. R. P. W. Scott, P Kucera and M. Munroe, J. Chromatogr., 186(1979)475. 13. A. Tiselius and D. Claesson, Ark. Kemi. Mineral. Geol. 15B(No.18)(1942). 14. C. Christiansen, Ann. Phys. Chem., 3(1884)298. 15. C. Christiansen, Handbook of Chemical Microscopy (Ed. E. M. Chanot and C. W. Mason), John Wiley and Sons, New York, Vol. 1((1958)101 and 189. 16. M. Bakken and V. J. Stenberg, J. Chromatogr. Sci., 9(1971)603. 17. 7. J. P. Gorden, R. C. C. Leite, R. S. Moore, S. P. S. Posto, J. R. Whinnery, Bull. Am. Phys. Soc., (2) 9(1964)501. 18. J. P. Gorden, R. C. C. Leite, R. S. Moore, S. P. S. Posto, J. R. Whinnery, J. Appl. Phys., 36(1965)3. 19. C. E. Buffet and M. D. Momis, Anal. Chem. 54(1982)1824. 20. R. A. Grant, J. Appl. Chem. 8(1959)136. 21. M. Poppe and J. Kunysten, J. Chromatogr. Sci. 10(1972)16A. 22. L. V. Benningfield Jr., Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, March 5-9(1979)paper 123. 23. L. V. Benningfield Jr. and R. A. Mowery Jr., J. Chromatogr. Sci., 19(1981)115. 24. 14. R. K. Bade, L. V. Benningfield Jr., R. A. Mowery and E. N. Fuller, Am. Lab.13(10)(19810130. 25. J. P. Gorden, R. C. C. Leite, R. S. Moore, S. P. S. Posto, J. R. Whinnery, Bull. Am. Phys. Soc., (2) 9(1964)501. 26. J. P. Gorden, R. C. C. Leite, R. S. Moore, S. P. S. Posto, J. R. Whinnery, J. Appl. Phys., 36(1965)3.
107 27. C. E. Buffet and M. D. Momis, Anal. Chem. 54(1982)1824. 28. M. Poppe and J. Kunysten, J. Chromatogr. Sci. 10(1972)16A. 29. L. V. Benningfield Jr., Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy,March 5-9(1979)p 123. 30. L. V. Benningfield Jr. and R. A. Mowery Jr., J. Chromatogr. Sci., 19(1981)115. 31. R. K. Bade, L. V. Benningfield Jr., R. A. Mowery and E. N. Fuller, Am. Lab.13(10)(19810130. 32. J. M. Miller and R. Strusz, Am. Lab. Jan(1970)29 33. A. T. James, J. R. Ravenhill and R. P. W. Scott, Chem. Ind. (1964)746. 34. E. O. A. Haahti and T. Nikkari, Acta. Chem. Scand., 17(1963)2565. 35. A. Karmen, Anal. Chem. 38(1966)286. 36. R. P. W. Scott and J. F. Lawrence, J. Chromatogr. Sci.,8(1970)65. 37.12. 14. L. M. Van Dijk, J. Chromatogr. Sci. 10(1972)31. 38. L. Yang, G. J. Ferguson and M. L. Vertal, Anal. Chem.,56(1984)2632. 39. B. J. Compton and W. C. Purdy, J. Chromatogr., 169(1979)39. 40. A. Stolyhwo, O. S. Privet and W. L. Erdahl, J. Chromatogr. Sci. 8(1970)65. 41 V. Pretorius and J. F. van Rensburg, J Chromatogr. Sci., 11(1973)263. 42. K. Slais and M Krejci, J. Chromatogr., 91(1974)181. 43. N. A. Dugger, US Patent 528,343. 44. De Pea 45 De Pea 46. H. Dubsky, J. Chromatogr.,71(1972)395. 47. J. J. Szakusito and R. E. Robinson, Anal. Chem., 46(1974)1648. 48. P. J. Wyatt, Applied Optics 7(196801879.
108 49. D. T. Phillips, Nature 221(1969)1257. 50. S. G. Weber and W. C. Purdy, Ind. Eng. Chem. Prod. Res. Dev., 20(1981). 51. H. G. Hanekamp, P. Boss and R. W. Frei, Trends in Anal. Chem, 1(1982)135. 52. H. Poppe, Instrumentation for High-Performance Liquid Chromatography, (Ed. J. F. K. Huber) Elsevier, Amsterdam and New York (1978). 53. A. Atsushi, T. Matsue and I. Uchida, Anal Chem., 64 (1992)44. 54. C. N. Svendsen, Analyst, 118(Feb)(1993)123.

LC Detector

Uploaded by

Document Information

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

LC Detector

Uploaded by

Copyright:

Available Formats

1

Chrom-Ed Book Series

LIQUID CHROMATOGRAPHY DETECTORS

Mobile Phase Velocity

Parabolic Velocity Profile

Tube Walls Initial Band Width Dispersed Band

Variance per Unit Length ( ml 2 /cm)

0.5 Straight Tube 0.007 in I.D.

1.5 Serpentine Tube 0.10 in I.D.

2.0 3.0 Flow rate (ml/min)

Dispersion Peak from Straight Tube (0.25mm I.D.)

Dispersion Peak from Straight Tube (0.18mm I.D.)

1 0.3 0.1 0.03 0.01

1.0 2.0 3.0 Normalized Elution Volume V/V T

Optical Window Inlet from Column Mobile Phase Exit

Detector Cell Optical Window

Concentration in Arbitrary Units

Solute Concentration (normalized)

Volume Flow of Mobile Phase

T' = 1.5 sec

0.25 Time (seconds)

Sensor Recorder Amplifier and Power Supply

Focusing Lenses Photocell

Flat Mirror Lamp

Achromat Lamp Photocells

Sensor Cell Half Mirror To Waste Photocell

Lens Polarizer Mask Reference Cell Lens Quarterwave Plate

Half Mirror Heating Laser

Reference Photocell Probe Laser

Pin Hole Mask Heat Filter Photocell Filter Sample Cell

100 Relative Emission

302.2 307.6 213.9

100 Cadmium Lamp

346.6 340.3 325.4 308.1 313.2

293.1 288.1 283.6

Photo Cells Reference Cell

Flow of Mobile Phase Over Plate

Plane Mirror To Waste Photo Cell From Column

Curved Mirror Plane Mirror Diffraction Grating

Courtesy of the Perkin Elmer Corporation

Achromatic Lens System Outlet

Photo Diode Array Shutter Flow Cell Eluent from Column

I max = 274 nm P.I. =0

Irregular top to peak showing presence of impurity

Spectra of Leading Edge of Peak

Excitation Light Fluorescent Light From Column To Waste UV Light Source

Courtesy of the Perkin Elmer Corporation

Spherical Mirror Beam Splitter Reference Cell

Grating that Selects the Wavelength of the Emitted Light

Fluorescence Detector Program

LC Column Gauze Jet Air Inlet FID

Figure 42. The Chain Detector

Evaporator Oven From Column Coating Block Oxygen

Nickel Catalyst Molecular Entrainer Hydrogen/Argon Oxygen

Synchronous Motor Flame FID

Eluent from Column Nebulizer Photomultiplier Optical Fibers

Light Source Light Trap

Courtesy of Polymer Laboratories Inc.

Figure 48. A Diagram of the Evaporative Light Scattering Detector

Atomized Stream of Particles

Collimated Light Beam

Courtesy of Polymer Laboratories Inc.

where (h) is the solvent refractive index,

Measuring Attenuators Apertures