516 Analyst, June, 1979, VoL. 104, pp. 516-524 Sequential Multi-element Analysis of Small Fragments of Glass by Atomic-emission Spectrometry Using an Inductively Coupled Radiofrequency Argon Plasma Source T. Catterick and D. A. Hickman The Metropolitan Police Forensic Science Laboratory, 109 Lambeth Road, London, SE 1 7LP A method is described for the quantitative multi-element analysis of small fragments (200-500 pg) of glass using an inductively coupled radiofrequency argon plasma source. The glass samples are digested with a mixture of hydrofluoric and hydrochloric acids and chromium is added as an internal standard. An ultrasonic nebuliser is used in order to reduce to a minimum the volume of solution required for each analysis. A single monochromator and detection system is employed, and the wavelength regions of interest are examined sequentially by means of a specially constructed control unit.The results for aluminium, barium, iron, magnesium and manganese show that the analysis of glass fragments in the range 200-500 pg can be achieved with coefficients of variation of approximately 10%. Standard glasses were analysed to assess the accuracy of the method. Keywords ; Glass analysis ; acid digestion ; control unit for automatic sequential selection of wavelength regions ; inductively coupled radiofrequency argon filasma ; forensic analysis Glass is a commonly encountered material in forensic science, but until recently much of the evidential value has been based on the measurement of physical properties such as refractive index and density.Several worker~l-~ have shown that chemical analysis of a glass sample will increase the evidential value and may well enable the sample to be classified as sheet, container, tableware, headlamp, etc., glass. Although the measurement of a large number of variables (such as trace-element concentrations) should give rise to the best classification,* this must be related to such factlors as the analytical technique employed and the operator time involved. The preferred situation would be to classify an unknown glass sample by measuring the concentrations of a relatively small number of trace elements. Previous work in this lab0ratory~9~9~ has shown that aluminium, barium, iron, manganese and magnesium are useful elements for classifying glass samples.Alternative analytical techniquesls2 have indicated that other elements, such as antimony, arsenic and potassium, are also useful for classification purposes, but again the determination of these elements must be viewed in relation to the capabilities of the analytical instrumentation available. A number of techniques have in fact been employed for the chemical analysis of glass: neutron- activation analysis,lv7 d.c. arc - atomic-emission spectr~graphy,~~~ atomic-absorption spectro- metry,596 spark-source mass spectrometry2 and X-ray fluorescence spectrometry.9 This paper describes an inductively coupled argon plasma (ICP) - atomic-emission spectro- metric procedure, which is a logical progression from the d.c. arc - atomic-emission3 and atomic-absorption6s6 spectroscopic methods previously reported from this laboratory.The emission spectrographic method, which involved grinding the glass fragments in a graphite matrix, was very demanding on the operator and was also prone to airborne contamination. It was also a fairly lengthy procedure, incorpclrating densitometry of the photographically recorded spectra. Atomic-absorption spectromletry is inherently a single-element technique, and the limited linear range for each element would mean that the measurement of several elements in a dissolved glass sample would be time consuming and might require dilutions of the sample. Electrothermal atomisation might be necessary for the determination of low concentrations of some elements, and this would increase further the analysis time.The inductively coupled argon plasma - atomic-emission method takes advantage of the large linear working ranges characteristic of plasma sources to reduce sample preparation to aCATTERICK AND HICKMAN 617 minimum. The sequential wavelength selection control unit employed enables a single monochromator and photomultiplier tube to be used for signal detection, and thus reduces instrumental costs. It also provides a more flexible system than a direct-read spectrometer. A dissolved glass sample can be analysed for five elements in under 4 min. Control unit Experimental Apparatus components and operating conditions are listed in Table I. A schematic diagram of the instrumental system is shown in Fig. 1, and details of the Argon 1 nebuliser U'trasonic I Fig.1. Schematic diagram of experimental system. Sequential analysis control unit The control unit was designed especially for this work, and operates by controlling auto- matically a stepping motor coupled to the wavelength drive of the monochromator, the synchronous motor of the monochromator and the detection and recording system. Up to TABLE I INSTRUMENTATION AND OPERATING CONDITIONS Plasma power supply . . Plasma torch . . .. Argon flow-rates . . Nebuliser . . .. Optics .. .. Spectrometer . . .. Wavelength selection . . Read-out . . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. Radyne, Model R50P, 27 MHz. Output power variable, 0-5 kW, normally operated a t 2.5 kW. Work coil, 3 turns 4 x 4 mm cross-section copper tubing. De-mountable torch; coolant gas tube, fused silica, 28 f 0.6 mm 0.d.; plasma gas tube, fused silica, 23 f 0.5 mm 0.d.; sample gas tube, borosilicate glass, 7 mm 0.d.; jet, 1.5mm i.d.A PTFE tube, 3.5 mm o.d., 3.0 mm i.d., is positioned inside the sample gas tube and is connected to the nebuliser. Coolant, 20.0; plasma, 0.0; sample, 0.55 1 min-l. Ultrasonic nebuliser, based on a published design,1° using a Siemens Sonostat 633 ultrasonic supply, of maximum output 12 W. Plasma is imaged on to the entrance slit of the spectrometer (a distance of 550 mm away) with an optical arrangement of two fused silica lenses. Entrance slit: width 0.02 mm, height 4-18 mm. Rank Hilger Monospek 1000; grating of 1200 lines mm-l blazed at 300 nm, reciprocal linear dispersion 0.82 nm mm-l. The control unit for sequentially selecting and scanning across wavelength regions is described in the text.Signal from photomultiplier tube [RCA, Type IP28 (selected)] , a t a potential of 600V (Brandenberg power supply, Model 475R), is amplified by the internal operational amplifier of the control unit and monitored on a potentiometric chart recorder (Servoscribe, Model RE 54 1.20). Sample uptake rate, 0.2 ml min-l.518 CATTERICK AND HICKMAN : SEQUENTIAL MULTI-ELEMENT ANALYSIS OF Analyst, VOl. 104 eight wavelength regions can be examined using the control unit, and for each wavelength region one of four possible amplifier gain settings can be selected. The stepping motor performs a rapid slewing (278 nm min-l) between each wavelength region. The response of the photomultiplier tube at the exit slit of the monochromator, suitably amplified by the operational amplifier in the control unit, is only monitored during the scan of the synchronous motor and the signal is fed to a chart recorder. The advantage of this system of scanning across a wavelength region is that the background levels on either side of the atomic-emission line are automatically recorded, and thus the need to carry out discrete measurements of background and background plus signal is avoided.Entering data for sequential analysis into the control unit The selected analytical atomic-emission lines are arranged in increasing order of wave- length, together with a reference “Start” position chosen at a convenient value just below the shortest wavelength. The stepping motor for altering the monochromator wavelength drive requires 96 pulses to complete one revolution, corresponding to a 2.5-nm wavelength region. It is thus straightforward to calculate the number of pulses required to advance the monochromator wavelength position from the chosen “Start” position to the selected wavelength regions.In practice, these numbers are then converted into equivalent binary numbers. The binary numbers associated with each of the selected wavelengths are entered sequentially in order of increasing wavelength into the control unit, using a set of 16 toggle switches. The 16-bit number chosen for each wavelength region is stored in a random access memory device (RAM, Sygnetics 82509). The position of each entry into the memory is determined by incrementing the setting of a multi-pole PROGRAM SWITCH.Each 16-bit binary number contains two items of information : (a) The first 14 bits represent the number of pulses required by the stepping motor for slewing the monochromator wavelength drive from the selected “Start” position to a position adjacent to the chosen analytical wavelength. In practice the maximum range from the “Start” position is about 425 nm. (b) The remaining 2 bits represent the amplifier gain selected from one of four possible settings. Logic of the control unit Automatic wavelength drive control. After the necessary combination of up to nine 16-bit binary numbers has been entered into the memory, the following sequence of events is initiated by pressing the START button. The COMPARATOR and COUNTER are cleared and the MEMORY ADDRESS CONTROL enters the second binary number from the MEMORY STORE into the MEMORY BUFFER (the Jirst binary number being the “Start” value).The first 14 bits of this 16-bit binary number are com- pared with the value in the COUNTER. While the COMPARATOR registers a “not equal to” state between the MEMORY BUFFER and the COUNTER, pulses are sent simultaneously from the INTERNAL CLOCK to the STEPPING MOTOR CONTROL and to the COUNTER. When the value in the COUNTER equals the 14-bit binary number in the MEMORY BUFFER the COM- PARATOR sends out an “equal to” signal. This “equal to” signal is sent to the MEMORY ADDRESS CONTROL, the STEPPING MOTOR CONTROL and the SYNCHRONOUS MOTOR CONTROL. This has the effect of stopping the pulses, which were both incrementing the value in the COUNTER, and also activating the stepping motor; the “equal to” signal results in the initiation of the time-controlled scans of the !synchronous motor.The switching of both motors is achieved via their relevant control units operating the appropriate relays (see Fig. 2). The SYNCHRONOUS MOTOR CONTROL uses pulses from the INTERNAL CLOCK to give a synchronous motor scan of fixed duration (about 9 s). At the end of the first scan the scan direction is reversed and a second scan is made across the same wavelength region. A counter in the SYNCHRONOUS MOTOR CONTROL registers when both scans have been completed. A signal is then generated, which is sent to the MEMORY ADDRESS CONTROL. This unit then updates the MEMORY BUFFER with the next 16-bit binary number and the above sequence of events is repeated for each binary number in the sequential program.The COUNTER This is shown schematically in Fig. 2.June, 1979 SMALL FRAGMENTS OF GLASS BY ATOMIC-EMISSION SPECTROMETRY 2 + Bits e 14 Bits -b Memory store (9 x 16-bit Program address control I I I I B e-. 1 Comparator w 4 UP 1 I Down Start L r t I Internal clock - - - Pu Ise generator v I L r Synch. motor Synch. : Relay - +- control motor Drive shaft to grating l-r-7 Step ping 519 r Stepping motor motor : Relay : control Fig. 2. Automatic wavelength drive control. keeps a running tally of how far the stepping motor has advanced from the original “Start” position. Thus, with each subsequent larger binary number (associated with the analytical regions of interest programmed in increasing order of wavelength), the number of pulses necessary to advance from the “old” to the “new” wavelength position is always available within the control unit.At the completion of each synchronous scan sequence the MEMORY ADDRESS CONTROL compares the position allocated in the MEMORY STORE of the current binary number in the MEMORY BUFFER with the position of the multi-pole PROGRAM SWITCH, This switch is left set at the position of the final entry. If the positions of the current binary number in the memory and the multi-pole switch are equal, the MEMORY ADDRESS CONTROL recognises that the sequential analysis program has been completed. A series of signals is then generated from the MEMORY ADDRESS CONTROL and the following actions result : (a) the COMPARATOR is cleared; ( b ) the Jirst 16-bit number in the program (i.e., the “Start” value) is entered into the MEMORY BUFFER; (c) the STEPPING MOTOR CONTROL reverses the direction of the stepping motor; ( d ) the COUNTER is set to count down as it receives pulses via the MEMORY ADDRESS The stepping motor then returns the monochromator to the selected “Start” position, and the control unit is now ready to repeat the programmed sequence of events. Automatic detection and recording system.In addition to the wavelength drive control system, the control unit contains a complementary system for controlling the detection and recording of the analytical signal. This is shown schematically in Fig. 3. CONTROL.520 CATTERICK AND HICKMAN : SEQUENTIAL MULTI-ELEMENT ANALYSIS OF Analyst, Vol.104 2-Bit binary Output from value from comparator memory buffer i output from PM T su p p I y to chart drive Chart recorder Fig. 3. Automatic detection and recording system. The output from the photomultiplier tube (PMT) detector is only switched through to the operational amplifier during the synchronous motor scans. This is achieved by activating a switch (SWITCH 1) with the “equal to” signal from the COMPARATOR. The same signal activates the chart paper drive of the recorder (SWITCH 2). The third part of the detection system that is automatically controlled is the gain selected by the AMPLIFIER CONTROL. This employs the remaining 2 bits of the 16-bit word. in the MEMORY BUFFER that is not involved with the wavelength drive.According to the value of these 2-bit binary numbers one of four pairs of resistors (Rl-R4 in Fig. 3) is selected. The linkage of the resistors to the operational amplifier gives the gain selected as appropriate for each particular wavelength region. The combination of the functions described above provides the necessary controls for the programmed sequential examination of wavelength regions and for selectively recording the signal from the photomultiplier tube during each of the synchronous motor scans. Ultrasonic fiebuliser As the method was designed to accommodate small sample sizes (200-500pg), it was necessary to keep to a minimum the corresponlding volume of dissolved sample, in order to retain reasonable concentrations of the trace elements.The requirement of effecting sequential analysis on a small sample volume led to the selection of an ultrasonic nebuliser, with the relatively low uptake rate of 0.2 ml min-1. The construction of the nebuliser was based on a design by Hoare et aZ.,lO with the modification of adding a PTFE sample transfer tube; this tube was extended to form an inner lining to the borosilicate glass sample injection tube of the plasma torch. The PTFE tube ended just short of the jet on the borosilicate tube. Although this meant that the jet was uinprotected from attack by hydrofluoric acid, in practice no visible signs of etching have been observed. No problems have been encountered with the trace-element concentrations in blank acid solutions run through the system.The problem of the presence of hydrofluoric acid could have been overcome by complexing the excess of hydrofluoric acid prior to analysis; the use of boric acid for this purpose has been described in the literature.1l This procedure was not adopted as it was felt that any additional steps in the sample digestion procedure would increase the risk of contamination and would also increase the sarnlple preparation time. Materials and Reagents Polystyrene tubes (5 ml, Sterilin) were emp1o:yed for the acid digestion of the samples and as containers for the working standard solutions. The tubes were pre-washed with an aqueous solution of hydrofluoric and hydrochloric acids (2HF + HC1 + 9H,O), were rinsedJune, 1979 SMALL FRAGMENTS OF GLASS BY ATOMIC-EMISSION SPECTROMETRY 521 with doubly distilled water and ethanol and dried in an oven at 60 “C.All acids used were of Aristar quality (BDH Chemicals), and elemental standard solutions were commercially available 1 000 pg ml-l solutions (BDH Chemicals or Hopkin & Williams). Procedure Sample preparation The glass fragments (200-500 pg) were cleaned by soaking in concentrated nitric acid for 30min, followed by a triple rinse with doubly distilled water. A final rinse with ethanol preceded drying at 60 “C. Glass samples were weighed on a microbalance (Perkin-Elmer, Model AD-2) before being transferred into the pre-washed 5-ml tubes. A 0.5-ml volume of a mixed acid solution (HF + 2HC1) was added to each sample,6 and these were then agitated for 30 min in an ultrasonic bath (Megason, Model 60-1, Schuco Scientific Ltd.).To the resulting glass solutions were added 1.5 ml of doubly distilled water and 0.5 ml of a 2.0 pg ml-l solution of chromium (as an internal standard) to give a total volume of 2.5 ml. Blanks were prepared in a similar manner. Standards tion ranges are listed in Table 11. samples, using equivalent amounts of the mixed acids and the chromium internal standard. Four multi-element standard solutions were used to calibrate each run, and their concentra- The standards were prepared in a similar manner to the TABLE I1 ANALYTICAL WAVELENGTHS, CONCENTRATION RANGES AND DETECTION LIMITS Element “Start” position Manganese . . Iron . . .. Chromium . . Magnesium A Magnesium B Aluminium . . Barium . . Ionisation Wavelength/ state nm ..249.80 . . I1 257.61 . . 11 259.94 . . I1 267.71 .. 277.98 . . I 285.21 .. I 396.15 . . I1 455.40 Concentration range of standards/ pg ml-1 - 0.007 5-0.045 0.05-0.30 (0.40) 1.0-6.0 1.0-2.0 0.25-1.50 0.007 5-0.045 Concentration Detection limit 0.5 mg of glass glass range for for 0.5 mg of - - 0-225 p.p.m. 10 p.p.m. Internal standard 0-0.15% o.oo5~0 0-3.0y0 0.2% o-1.0yo 0.05% 0-0.75% 0.05% 0-225 p.p.m. 5 p.p.m. Analysis The samples, still in the original 5-ml tubes used for digestion, are connected in turn to the ultrasonic nebulising system. The resulting aerosol is transferred into the plasma by a carrier flow of “sample” argon. After a delay of 25 s to allow equilibrium conditions to become established, and with the monochromator set to the “Start” wavelength, the pro- grammed sequence of scans is initiated at the control unit.The seven analytical wavelengths monitored are listed in Table 11, together with the “Start” wavelength position. Two magnesium emission lines are studied in order to cover the large range of magnesium concentra- tions found in glass samples. The control unit activates the stepping motor, which stops at the first programmed value, about 0.4 nm in front of the emission line. The stepping motor is electrically disengaged by the control unit and the synchronous motor of the mono- chromator is activated. This scans forward slowly (5.0 nm min-1) across a 0.8-nm spectral region. The direction of the synchronous motor is then reversed, and a second traverse across the spectral region is made, resulting in a duplicate record of each analytical feature.When the scan starting position is reached the stepping motor is re-engaged and moves the wavelength drive forward to the second programmed value. The procedure is repeated to cover all seven analytical wavelengths ; the stepping motor then returns the wavelength drive to the original “Start” position. During the scans of the synchronous motor the out- put of the photomultiplier tube, suitably amplified at one of the four gain settings selected in the initial programming, is fed to a chart recorder. The control unit holds the signal to the operational amplifier at earth potential except during the synchronous motor scans.522 CATTERICK AND HICKMAN : SEQUENTIAL MULTI-ELEMENT ANALYSIS OF Analyst, ‘cl‘ol.104 Similarly, the chart-paper drive of the recorder is only activated during the synchronous motor scan periods. Fig. 4 shows a typical chart recorder output from the analysis of a fragment of sheet glass. F C 1‘ il Fig. 4. Chart-recorder trace from the analysis of a 340-pg fragment of sheet glass (refractive index, 1.517 1). Emission lines: A, manganese 257.6nm; B, iron 259.9nm; C, chromium 267.7 nm; D, magnesium 278.0 nm; E, magnesium 285.2 nm; F, aluminium 396.1 nm; and G, barium 455.4 nm. The total analysis time for the programmed sequence of seven wavelength regions is 3 min 30 s; the control unit provides the option of scanning four times (instead of twice) over the emission lines, and using this option the analysis takes 5 min 40 s. A sample of distilled water is nebulised into the plasma in between each sample analysis, during the time when the wavelength drive returns from the highest wavelength to the “Start” position.The peak heights of the blanks, standards and samples are fed into a simple computer program together with the elemental concentrations of the standards and the masses of the glass samples. For each analysis the ratios of the peak heights to the chromium peak height are calculated. Using the data for the blanks and standards a line of best fit is constructed for each element and the ratioed sample peaks are compared with this. A final listing is generated of the samples and their corresportding concentrations of the trace elements measured. Listed in Table I1 are the working concentration ranges in glass (for 500-pg samples) for the elements determined by this method.Also listed are the corresponding detection limits, but these are not necessarily absolute values a s compromise experimental conditions were employed. The values represent the limits attainable under the conditions needed to cover the concentration ranges expected in glass samples. Results and Discussion Analysis of Standard Glasses (Accuracy Check) Six glasses were prepared3 by grinding together mixtures of Specpure (Johnson Matthey) oxides and carbonates and fusing them in platinum crucibles at 1400 “C. The results of three separate analyses of each glass are given in Table 111. The values in parentheses are the levels predicted from the masses of the compounds used in preparing the standard glass.June, 1979 SMALL FRAGMENTS OF GLASS BY ATOMIC-EMISSION SPECTROMETRY 523 TABLE I11 ANALYSIS OF STANDARD GLASSES Sample All % Ba, p.p.m.Fe, p.p.m. Mg, % Mn, p.p.m. A 0.16 (0.13) 23 (17) 1380 (1 890) 0.043 (0.052) 272 (303) B 0.20 (0.18) 50 (49) 1000 (1 220) 0.073 (0.096) 141 (170) D 0.40 (0.50) 95 (100) 380 (450) 1.49 (1.6) 169 (206) E 0.82 (0.81) 126 (134) 260 (290) 1.96 (2.2) 85 (82) F 0.64 (0.85) 576 (657) 90 (80) 3.99 (4.1) 28 (33) C 0.41 (0.51) 85 (80) 740 (870) 1.04 (1.1) 112 (119) Analysis of Sheet Glass (Precision Check) A typical commercially produced sheet glass (refractive index 1.517 1) was analysed a number of times in order to assess the over-all precision of the method. Table IV gives the results of a series of nine determinations made on one day, and a series of 24 determinations made over several weeks, with different operators and standard solutions. Also listed in Table IV are the results of analyses of the same sheet-glass sample by d.c.arc - atomic- emission ~pectrography~ and atomic-absorption spectrometry.6 This method gives better precision for replicate analyses of the sheet glass than it does for an equivalent number of analyses of one of the laboratory-produced standard glasses (coefficients of variation 4-10 yo compared with 8-25%). This is probably a consequence of inhomogeneity in the standard glass. TABLE IV ANALYSIS OF SHEET GLASS Method ICP - AES (within-day) ICP - AES (long-term) D.c. arc - AESB . . AASs . . .. .. Mean Standard deviation No. of determinations C.V., yo Mean Standard deviation No.of determinations C.V., yo Mean Standard deviation No. of determinations C.V., yo . . Mean Standard deviation No. of determinations C.V., yo A], % 0.51 0.049 9.6 9 0.51 0.063 12.2 24 0.45 0.05 11.1 11 Ba, p.p.m. 116 5.1 4.4 9 108 14 13.0 24 114 18 15.8 11 Fe, p.p.m. 560 38 6.7 9 611 86 14.1 24 576 75 13.2 11 5 70 20 10 3.5 Mg. % 1.79 0.069 3.9 9 1.73 0.15 8.5 24 1.70 0.17 10.0 11 2.03 0.041 2.0 10 Mn. p.p.m. 71 5.9 8.3 9 77 11 14.0 24 102 17 16.6 11 84 2.6 3.1 10 Conclusions A method has been described for the sequential quantitative determination of five elements in small samples of glass. The samples are dissolved in a mixture of hydrofluoric acid and hydrochloric acid at ambient temperature.6 Although the standard solutions are matched to the samples with respect to acid concentration, it was found that matching to the sodium, calcium and silica matrix present in a digested glass sample was unnecessary.This is in agreement with the conclusions of other workers12 with inductively coupled plasmas that chemical interferences are insignificant. An ultrasonic nebuliser with an acid-resistant transfer system enables the digested samples to be nebulised without undergoing neutralisa- tion or massive dilution. The use of an internal standard improves the precision of the method. The coefficient of variation for magnesium on replicate analyses of the sheet glass was 8.6% without the internal standard and 3.9% (see Table IV) with the internal standard. This finding is in agreement with that of Hoare and Mostyn,13 who used a very similar ultrasonic nebulising system.524 CATTERICK AND HICKMAN The method works well for samples in the range 200-500 pg, giving relative standard deviations of 4-10%.The variation in relative standard deviation reflects the demands of the analysis; magnesium occurs in glass at levels of several per cent. and is thus easier to determine than manganese, which is present at the 20-200 p.p.m. level. Iron determina- tions will lie somewhere between these two extremes, but may be subject to contamination. An advantage of the method is that it gives absolute levels of the trace elements, compared with methods such as spark-source mass spectrometry and X-ray fluorescence spectrometry, in which the results are normally expressed as ratios. For the purposes of data interpretation it is more satisfactory to use absolute trace-element values. A novel feature of the method is the wavelength-drive control unit, which enables a sample to be analysed for several elements sequentially, at a lower cost than a direct-read spectrometer. The unit also provides a greater degree of flexibility than a polychromator system as it can be re- programmed for a new set of elements in a few minutes. The authors thank Dr. A. G. Knapp and M:r. J. Russell for their help in the design and construction of the sequential analysis control unit. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. References Goode, G. C., Wood, G., Brooke, N., and Coleman, R. F., A.W.R.E. Re$., No. 024/71, 1971. German, B., and Scaplehorn, A. W., J. Forens. Sci. SOC., 1972, 12, 67. Blacklock, E. C., Rogers, A., Wall, C., and Wheals, B. B., Forens. Sci., 1976, 7, 121. Sneath, P. H. A., and Sokal, R. R., “Numerical Taxonomy,” Freeman, San Francisco, 1973, p. 5. Hughes, J. C., Catterick, T., and Southeard, G., Forens. Sci., 1976, 8, 217. Catterick, T., and Wall, C. D., Talanta, 1978, 25, 573. Sayre, E. V., and Smith, R. W., in Bishay, A., Editor, “Recent Advances in the Science and Tech- Harvey, C. E., J . Forens. Sci., 1968, 13, 269. Reeve, V., Mathiesen, J., and Fong, W., J. Forms. Sci., 1976, 21, 291. Hoare, H. C., Mostyn, R. A., and Newland, B. ‘r. N., Analytica Chim. Ada, 1968, 40, 181. Price, W. J., and Whiteside, P. J., Analyst, 197’7, 102, 664. Fassel, V. A., and Kniseley, R. N., Analyt. Chem., 1974, 46, lllOA and 1155A. Hoare, H. C., and Mostyn, R. A., Analyt. Chem., 1967, 39, 1153. nology of Materials,’’ Volume 3, Plenum Press, New York, 1974, pp. 47-70. Received December 18th, 1978 Accepted January 15th, 1979