Analyst, May, 1973, Vol. 98, pp. 329-334 329 Analysis of High-purity Water by Flameless Atomic-absorption Spectroscopy Part II.* Signal Integration with a Non-resonance Line Correction System for Spurious Absorption Phenomena BY C. J. PICKFORD? AND G. ROSSIS (Chemistry Division, Euratom-CCR. 21020-Ispra ( Varese), Italy) A polychromator has been used in conjunction with a multi-channel integration system and an automatic sample injection unit in graphite-tube flameless atomic-absorption spectroscopy. The precision of the system has been evaluated a t high and low absorbances, and its ability to compensate for spurious absorption and variable volatility effects examined. IN Part I,l we described an automatic sampling unit used in conjunction with flameless atomic-absorption spectroscopy as part of a project to provide simultaneous determinations of a number of elements present at parts per lo9 (p.p.b.) levels in water.This paper describes the spectrophotometer and the detection system used in conjunction with this sampling unit for the above purpose. The system has been tested by using one main channel and one reference channel sequentially ; the system used for multi-element determinations will contain up to five channels and one or more reference channels. SPECTROPHOTOMETER SYSTEM- In the past, several systems have been described for multi-element determinations by atomic-emission,2 fluorescence3 or absorption4 spectroscopy with the use of polychromators or interference filter systems to separate the light beam into its individual components.For atomic-absorption spectroscopy, band-pass requirements usually preclude the use of interference filters, so that a polychromator is the obvious choice. SPURIOUS ABSORPTION PHENOMENA- Interferences in flameless atomic-absorption spectroscopy from spurious absorption effects, whether caused by condensed vapours of inorganic salts or by carbon dust, have often been For this reason, in a fully automatic system with low absorbance signals from samples that might be expected to exhibit such phenomena, some way of correct- ing for this effect is obviously desirable. There are several methods that can possibly be used for this purpose. A nearby non-resonance line of the same or another element can be used to measure the absorption due to non-specific effects alone, and this value can then be subtracted from the signal obtained by measurement at the element resonance line.The correction can be effected simultaneously, by using a dual-channel system,* as is achieved with a number of modern atomic-absorption spectrophotometers, or can be applied sequen- ti all^.^ The latter method is, of course, much easier to apply instrumentally, and can be carried out with even the simplest atomic-absorption instrument but, clearly, for the purpose of a study such as that being considered, the choice of a simultaneous correction system is imperative. Perhaps the most elegant compensation system is the deuterium-arc corrector,lO which involves the non-specific absorption of light from a deuterium arc at the same wavelength as the resonance line in order to compensate for background absorption.A similar system has been describedll in which a hydrogen-arc lamp, and polarisers to separate the two light beams, are used. In choosing the method of correction that is most suitable for multi-element determinations, the advantages and disadvantages of the two-line and the deuterium-arc methods should be considered. The principal disadvantages of the two-line method are that Gebouw voor Analytische Scheikunde, Technische Hogeschool Delft, Jaff alaan 9, * For details of Part I of this series, see reference list, p. 334. t Present address : $ To whom requests for reprints should be addressed. @ SAC and the authors. Delft, The Netherlands.330 PICKFORD AND ROSSI: ANALYSIS OF HIGH-PURITY WATER BY [Analyst, Vol.98 selection of the non-resonance line (which must be very close to the resonance line) may be difficult, except when a source rich in intense emission lines is used, and that isolation of the two lines with exit slits may be rather tedious unless lines from overlapping orders, as with grating instruments, can be selected. The principal disadvantages of the deuterium-arc system are that the continuum emitted may have a low intensity at certain wavelengths, and is essentially zero above 360 nm. Also, it is necessary, for instrumental reasons, for intensities of the two signal sources to be exactly balanced before measurements can be made. When multi-element determinations are to be carried out, this latter requirement indicates the use of the two-line method, because other- wise the intensity of each resonance line must be attenuated to that of the deuterium arc at that wavelength, which attenuation would be difficult to achieve and wasteful of energy (thus giving a greatly increased noise level).TREATMENT OF THE ABSORPTION SIGNAL- Although the amplitude of an absorption signal is usually used for the purpose of deriving calibration results, the integral of the signal is more often recorded in multi-channel instru- ments, mainly because of the greater ease of signal handling and subsequent display. It has also been ~ l a i m e d ~ ~ ~ ~ ~ ~ 3 that integration may, in addition, give greater precision when used in conjunction with flame or flameless atomic-absorption spectroscopy. For the above reasons, signal integration was preferred to amplitude measurement involving peak-sensing circuitry with memory capabilities. EXPERIMENTAL POLYCHROMATOR AND SLIT ASSEMBLY- A modified Jarrell-Ash, Model 75-000, fl6.3, plane grating 0-75-m spectrograph was used, with the photographic plate assembly removed and a light-tight box containing a series of exit slits and associated photomultipliers firmly attached.Commercially available aluminised quartz plates with 100-pm slits (Jarrell-Ash Co.) were carefully positioned at the focal plane of the spectrograph, a refractory quartz plate being used in front of each slit for the correct alignment of the slits with the spectral lines. The slit widths used corresponded with the opening of the entrance slit of the spectrophotometer.The minimum degree of separation of two consecutive slits that could be achieved was about 5 mm, which corresponded to a wavelength difference of 5 nm in the first order. When a smaller degree of wavelength separation was required two plane mirrors were used in order to bend the light path of the reference line sufficiently to bring it into line with the photo- multiplier. With this procedure, lines as close as 0.6 nm could be used. Sufficient space was available to enable up to ten lateral window photomultipliers to be installed. E.M.I. 9783 B in addition to RCA 1P 28 photomultipliers fed by a Keithley 245 high-voltage supply were used in this study. The above system (with individual slits) will later be replaced by a single aluminised quartz plate marked with various slits at the selected positions that correspond to the spectral lines to be used.Preliminary experiments with a reference beam system were performed by using a quartz plate positioned immediately after the graphite oven and tilted in such a way as partly to reflect the light from the multi-element hollow-cathode lamp, The reflected light was filtered with an interference filter so that a narrow wavelength band in the proximity of the resonance line to be studied could be passed and monitored by a photomultiplier. However, the dif- ference in band pass between the interference filter and the spectrophotometer led to results that were not very satisfactory. INTEGRATION ASSEMBLY- This assembly is shown in a simplified form in Fig. 1 for one channel (A) with a reference channel (B).This scheme is also applicable to other channels with which the same or different reference channels are used. The sequence of operations is as follows: the incoming photomultiplier signals are first converted into logarithmic form and then subtracted. The auto-balance circuit furnishes a d.c. level to the amplifier, which performs the subtraction step, so that during intervalsMay, 19731 FLAMELESS ATOMIC-ABSORPTION SPECTROSCOPY. PART I1 331 Fig.. 1. Elock diagram of dual-beam integrator circuit. k l energised; Re-set: k l and k3 energised; and Integrate: k l and k2 energised Balance : no relays energised ; Hold : between analyses (when there is no absorption) the subtractor output will be adjusted auto- matically to zero. This procedure is essentially equivalent to adjusting the intensities of the two light beams so that they become equal, but is achieved automatically.The output of the subtractor (now in absorbance units) is then integrated. The Balance, Hold, Re-set and Integrate cycles are controlled by three magnetic reed relays. The sequence can be achieved automatically, by being operated from the previously described central programming unit, or manually. The circuit can also be operated without the integration circuit, and also with direct amplification instead of log conversion. The log conversion, subtractor and integrator circuits are conventional, and have been described elsewhere.14 The auto-balance circuit is based on an auto-balance circuit described for use with gas-chromatographic integrators.15 The operational amplifiers used were the Burr- Brown 1556115 amplifiers.The output from the integrator is connected across a ten-turn helical potentiometer (Beckmann) so that for the preparation of linear calibration graphs, the integrator output can be scaled down directly to give a value in concentration units, which is displayed either by the digital voltmeter or the recorder. In normal operation the relays were controlled so that the integrator was in the Balance position continuously except when on Hold, Re-set or Integrate. Re-set was operated for 5 s before the atomisation cycle, Integrate was operated during the atomisation cycle and Hold was used as an intermediate position whenever a change was made, and for 30 s after Integrate so as to allow sufficient time for the integral to be recorded.A direct-current system was used in order to reduce the cost and complexity of the system in the event that the maximum number of channels is required. The principal disadvantage of this system in atomic-absorption spectroscopy, i.e., susceptibility to source emission and noise, is not a severe problem when the graphite tube is used, and in fact, when the light beam was correctly centred and focused, there was almost no interference from light emission originating within the crucible. The signal output was displayed either on a recorder (as previously described) or with a digital voltmeter (Keithley, Model 160). In fully automatic operation, a printer will be connected across the digital voltmeter for output display.OPERATING CONDITIONS- The operating conditions for the hollow-cathode lamps and the methods used for the preparation of solutions have been described previous1y.l Calibration graphs were obtained by using manual sample injection with a 100-pl Eppendorf pipette and manual operation of the integrator unit. Repetitive sampling was carried out by using the automatic sampling unit, with the integrator output displayed on a recorder. For measurements made for comparison purposes with a single-channel system, the Beckmann 1301/DBG spectrophoto- meter described previously was used.332 PICKFORD AND ROSSI : ANALYSIS OF HIGH-PURITY WATER BY [Analyst, Vol. 98 Selection of reference lines was made easier by initially preparing a photographic plate (Kodak Spectrum Analysis No.1, developed under standard conditions) of the hollow-cathode lamp concerned, by using the normal plate attachment for the spectrograph. After comparison of the plate with reference tables, suitable reference lines were selected according to the criteria given by billing^.^ RE s ULTS CALIBRATION GRAPHS- Calibration graphs were prepared for the elements cobalt, chromium, copper, iron, manganese, nickel and lead by using the reference line that had been established previously for each element. The chosen line pairs are listed in Table I. As an example, calibration graphs for lead are shown in Fig. 2; the intensity axis is expressed in arbitrary units, depending on whether the graph for 0 to 20 or 0 to 100 p.p.b. is considered. As can be seen, the region of linearity (ie., for which the integrator output can be calibrated directly in concentration units) extends up to about 60 p.p.b.TABLE I LINE PAIRS AND REPRODUCIBILITY RESULTS Element line/nm CO 240.7 Cr 357.9 CU 324.7 Fe 248.3 Mn 279-5 Ni 232.0 Pb 283.3 Reference line/nm CO 241.4 Ne 352.0 CU 323.1 CU 249.2 CU 282.4 Ni 231.4 Pb 280.1 Relative standard deviation with dual-channel integration system & Per cent. p.p.b. 0.5 20 1.5 2 0.9 50 1.8 5 1.0 20 2.3 2 0.8 10 1.5 2 1-1 10 1.6 2 1.3 50 2.2 5 1.8 50 2.3 6 Relative standard deviation with single-channel a.c. system +--7 Per cent. p.p.b. 0.6 50 1.0 100 1.2 50 1.1 30 1.7 10 2.4 50 Other reported values5 & Per cent. p.p.b. 4.9 2 4.7 2 4.7 10 3.5 2 2.6 20 3.9 5 EFFECTIVENESS OF THE COMPENSATION SYSTEM- This was tested by observing the effect produced on the zero base-line when a substance with a non-specific absorption at the wavelength used was introduced into the light path. With a metal-wire mesh screen, the compensation was complete (up to 96 per cent.absorption) when reference lines within about 15 nm of the main resonance line were used. However, when concentrated solutions of sodium chloride were vaporised in the crucible, thus producing a white “fog” in the light path, it was necessary for the reference line to be closer to the main line, usually within 4 nm, in order to give complete compensation. This requirement is to be expected and results from the small particle size and the greater light-scattering effect of the sodium chloride; it does not constitute a serious limitation as it does not affect the choice of a suitable reference line, when a multi-element hollow-cathode lamp or two independent lamps can be used.The effect of added sodium chloride on the absorption signal due to the elements con- cerned (at a concentration of 1Opg1-l) was examined both for the dual-beam integration system and for the normal single-beam a.c. instrument. Fig. 3 shows the results obtained for lead. As can be seen, a reduction in the integrated intensity occurs initially, possibly because of a chemical interference effect, and then the signal is virtually constant up to an 80 000-fold excess (corresponding to 95 per cent. absorption in channel A) ; beyond thisMay, 19731 FLAMELESS ATOMIC-ABSORPTIOX SPECTROSCOPY.PART I1 333 limit, the signal increases rapidly. Without compensation, the effect on the absorption signa is very much greater. Similar behaviour has been observed for the other elements considered in this study. COMPENSATION FOR VOLATILITY CHANGES- Because of changes in the voltage applied to the graphite tube, or to changes in the wall thickness, the absorption signal obtained for a standard solution may not always be constant on a day-to-day basis. In addition, various matrices may cause a depression in the amplitude of the absorption signal, compared with purely aqueous solutions. The rate of volatilisation of 10 pg 1-1 of lead was therefore varied artificially (by changing the atomisation voltage) in order to determine whether this problem could be solved more readily by signal integration than by amplitude measurement.Fig. 4 shows that, up to a certain limiting voltage value, there is no significant variation of the integrated signal, although above this value there is a rapid decrease. At lower concentration levels, this decrease at higher voltages was not observed. The variation of the signal amplitude with a normal single-beam system can be seen to be critically dependent on the response time of the amplifier and recorder. 20 40 60 80 100 Concentration, p.p.b. Fig. 2. Calibration mraphs for lead by using dual-channel Tntegration system; 283.3 and 280.1-nm lines. A, 0 to 20 p.p.b. of lead; and B, 0 to 100 p.p.b. of lend 1.0 w .- 8 3 0 m 8 L u 2 u .- L- F .- + 0.5 +? L 0 QJ 8 m - 9 0, Q Ll 0 200 400 600 800 1000 Concentration of added NaCI, p.p.m.Fig. 3. Effect of added sodium chloride on the absorption signal of 10 p.p.b. of lead: 1000 p.p.m. of NaCl= 100 000-fold excess. A, signal obtained with single-beam ax. instrument; and B, signal obtained with dual-beam inte- gration system REPRODUCIBILITY- The reproducibility was tested, as previously described, for the same elements as before, and in addition for lead. For each element two concentration levels were examined and the standard deviation of the integrator signal was evaluated from repetitive evaporations (at least sixteen for each concentration) from samples injected automatically. In Table I, the line pairs used are given and the precision achieved is compared with that of results obtained with a single-channel ax.system and with other values reported in the literature. The improvement in precision compared with that of previous results is slight, but the precision is maintained down to a lower level. At the lower concentrations chosen, the precision is affected adversely by the integrator base-line errors, caused by both drift and electrical pick-up of the relay switching signals of the oven unit and the central programmer, although these effects had been minimised by careful shielding of the input leads and by connecting capacitors in parallel with the switch contacts. CONCLUSIONS The use of a non-resonance line correction system and integration of the absorption signal improves the precision of graphite-tube flameless atomic-absorption spectroscopy,334 PICKFORD AND ROSS1 particularly at low concentration levels.Some variations in signal intensity caused by changes in the volatility of the sample can be eliminated, and the compensation for spurious absorption phenomena is satisfactory up to 95 per cent. absorption, provided that the reference line is close to the resonance line of the main element. 4 5 6 7 8 9 10 Atomisation voltage Fig. 4. Effect of variation of atomisation voltag; on the absorption signal of 10 p.p.b. of lead. .4, signal obtained with dual-channel inte- gration system; B, signal obtained with single-beam a.c. instrument with 0.5 s recorder response time; and C, signal obtained as for R but with 1 . 5 s recorder response time With the automatic sampler previously described, the above system does represent a first approach to a fully automatic scheme for multi-element determinations by flameless atomic-absorption spectroscopy, The results achieved indicate the feasibility and the applicability of such a system, particularly when a limited number of elements have to be determined on a routine basis or when the limited size and the nature of the sample (toxicity and activity) could make cumbersome or impossible the determination of more elements.A greater degree of flexibility could be expected by the use of a more appropriate spectro- photometer and of a more sophisticated exit slit assembly. MTe are grateful to Mr. M. Mol for constructing the electronic system, and to Euratom for the award of a postdoctoral fellowship to one of us (C.J.P.). 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. REFERENCES Pickford, C. J., and Rossi, G., Analyst, 1972, 97, 647. Haagen-Smit, J. W., and Ramirez-Munoz, J., Analytica Chiwt. Acta, 1966, 36, 469. Mitchell, D. G., and Johansson, A., Spectyochinz. Acta, 1970, 25B, 175. Mavrodineanu, R., and Hughes, R. C., Appl. Opt., 1968, 7, 1281. Fernandez, F. J., and Manning, D. C., Atom. Absorption Newsl., 1971, 10, 65. Alder, J. F., and West, T. S., Analytica Chim. A d a , 1970, 51, 365. Amos, M. I)., Bennett, P. A., Brodie, K. G., Lung, P. W. Y., and MatovBek, J. P., Analyt. Chern., Massmann, H., “Mdthodes Physiques d’Analyse,” Volume 4, G.A.M.S., Paris, 1968, p. 193. Billings, G. K., Atom. Absorption Newsl., 1965, 4, 357. Kahn, H. L., lbid., 1968, 40, 7. Woodriff, R., Culver, B. R., and Olson, K. W., A p p l . Spectrosc., 1970, 24, 530. L’vov, B. V., “Atomic Absorption Analysis,” Hilger and Watts, London, 1970. Nishita, H., Farmer, R., and Peterson, S., Analytica Chim. Acta, 1972, 58, 1. “Applications Manual for Computing Amplifiers,” Philbrick Research Tnc., Nimrod Press, Dedhani, Riggs, W. A., Analyt. Chevn., 1971, 43, 976. NOTE-Reference 1 is to Part I of this series. 1971, 43, 211. Mass., 1966. Received November 30th, 1972 Accepted January 5th, 1973