JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1986, VOL. 1 9 Volatilisation Studies of Cadmium Compounds by the Combined Quartz Furnace and Flame Atomic Absorption Method: Effects of Magnesium Chloride and Ascorbic Acid Additives Tibor Kantor and Laszlo Bezur Technical University of Budapest, Institute for General and Analytical Chemistry, H- 152 1 Budapest, Hungary A flame atomic absorption spectrometer was coupled to a modified thermoanalytical quartz furnace for the element-specific detection of evolved decomposition products. Heating programmes with a constant heating rate (0.8 “C s-I) and with an exponentially decreasing heating rate (typically 50-5 “C s-1) were applied in the range 25-950 “C. Sample holders made of platinum, electrographite and solid pyrolytic graphite were used in furnace atmospheres of variable oxygen concentration. Decomposition and vaporisation of Cd(N03)2, CdCI2, MgCI2, ascorbic acid and mixtures of these compounds were studied.The radiation scatter of the aggregates formed from the ascorbic acid by decomposition was detected in a flow cuvette. Because of the extremely long life of the atomic cadmium vapour, cold vapour detection of this species was also feasible in the combined system. This was used to supplement the flame atomic absorption measurements in elucidating vaporisation mechanisms. Samples in the form of aqueous solutions and occasionally as solids were applied to the furnace. It was found that the hydrolysis of CdCI2 and MgCI2 takes place to a lesser extent on increasing the sample mass and the heating rate and by using pyrolytic graphite and platinum sample holders. The enhanced rate of hydrolysis on the surface of porous graphite is explained by an adsorption effect (adsorption hydrolysis).The fraction of a hydrolysing chloride salt evolved in chloride form is also dependent on the pre-heating conditions. In the presence of ascorbic acid, Cd(l), CdO(s) and MgO(s) are formed from both chloride and nitrate salts of these elements in the first period of heating. On the surface of the pyrolytic sample holder, the formation of Cd(l) is enhanced in the presence of ascorbic acid and thus complete evaporation of cadmium can be achieved at low temperature (32tL460 “C). The vaporisation of CdO(s) takes place by thermal dissociation from both platinum and graphite surfaces at a temperature (ranging from 580 to 800 “C) that depends on the effective oxygen concentration close to the sample.The decomposition products of the nitrate anion activate the graphite surface, which enhances the vaporisation rate of CdO(s) from such a surface. Keywords: Combined atomisation source; atomic absorption detection for thermal analysis; CdCI, and MgCI, hydrolysis; Cd, CdC12, CdO and MgC12 vaporisation; ascorbic acid decomposition The atomic spectrometric methods developed for metal speciation involve high-temperature volatilisation as one of the separation methods. 1-3 When using accurate temperature control and measurement, these systems provide thermo- analytical information, i.e., an “evolved gas analysis” can actually be performed.2.3 The other aspects of the use of atomic spectrometric observation for thermal analysis are represented by the studies of high-temperature reactions of technological4.5 and analyticab7 interest.Some of the com- bined spectroscopic sources have been found to be particularly useful for these purposes (see references cited in reviews 1 and 8). This paper attempts to clarify further the mechanism of chloride interference effects encountered when cadmium is determined by electrothermal atomisation (ETA) procedures. ETA methods are widely used for the determination of cadmium in clinical and environmental analysis. Slavin et uL.~ have recently reviewed the literature on the subject (71 references) and have studied in detail the effects of instrumen- tal parameters and different sample matrices on the atomic absorption sensitivity of this element.The presence of alkali and alkaline earth metal chlorides ( e . g . , magnesium chloride) have especially severe suppressive effects on the cadmium ~ignal.9~10 Various organic matrices ( e . g . , organic acids) also have significant effects in this respect. Under the influence of ascorbic acid additive (“matrix modifier”), the atomisation temperatures for lead11 and cadmium decrease. 12 The vapori- sation characteristics of cadmium halides have been studied with respect to sample loss during pre-heating (“ashing” or “charring”) and dissociation in the gas phase, and various additives have been recommended for increasing atomic absorption sensitivity.9J3-15 The dissociation of metal halide molecules in the gas phase could be increased by application of a platform sample holder15 and by means of a constant- temperature furnace.16 Other workers have investigated the atomisation mechanism of cadmium oxyacid salts17-” and determined the effect of the graphite tube structure on atomisation processes.20 Several workers have applied combined atomisation - radiation sources to study the vaporisation of cadmium compounds. Gegus et a1.21 used a combination of a ceramic furnace (“micro reactor”) and a stabilised arc discharge for this purpose, with either platinum or graphite sample holders. Significant differences were found in the vaporisation charac- teristics of cadmium depending on whether new or aged (multiply heated) graphite sample holders were employed.Robinson and Weiss22 used a dual stage atomiser for the speciation of cadmium compounds on the basis of their different vaporisation characteristics. Aziz et al.23 applied a combination of a graphite furnace and an inductively coupled plasma (ICP) for the determination of cadmium and studied the vaporisation of cadmium nitrate in the presence of various matrices. A similar combination was used by Crabi et al.,z4 who found that free cadmium atoms had an extremely long lifetime near room temperature, in agreement with our findings.25 The detectability of cadmium vapour at near ambient temperature was also utilised in this work. Experimental Instrumentation The combined quartz furnace and flame system is shown schematically in Fig. 1. Details of the furnace design and that of the temperature programmer have been described else-10 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1986, VOL.1 Burner Gas injector Mixing - chamber - Oxidant 1 - G + & T z T ” a l v e t Furnace gas Cooling gas Auxiliary cooling gas Fig. 1. atomic absorption system Schematic diagram of the combined quartz furnace - flame Pump I- + J’L I, - t Aerosol Fig. 2. radiation scatter method Windowless flow cuvette for detection of aerosols by a where.4 Argon was supplied as “furnace gas” (0.5 dm3 min-l) and “cooling gas” (0.8 dm3 min-I), unless indicated other- wise. A T-junction between the gas injector and the furnace outlet is incorporated for introducing “auxiliary cooling gas” (2 dm3 min-1) with the “valve” open.In these experiments, air was used as an auxiliary cooling gas, which contributed the “oxidant” gas (air) applied through the gas injector. A single-slot 10-cm burner was used to support a slightly reducing acetylene - air flame (observation flame height 10 mm). Because of the long lifetime of the cadmium atomic vapour, it was also possible to detect this species by switching off the acetylene flow and leaving all other conditions unchanged. This observation method will be referred to as the “cold vapour detection method” in the following discussion. For certain studies the sample components evolved in the furnace were transported into a flow cuvette, shown in Fig. 2. In this instance the valve shown in Fig. 1 was closed (no auxiliary cooling gas was applied) and a more heat-resistant silicone rubber tube (40 cm long) was used to connect the T-junction and the flow cuvette.With this system a higher sensitivity could be attained for the detection of cadmium vapour, but this was not desired in these present studies. The windowless cuvette shown in Fig. 2 was used for the detection of aggregates (aerosol particles) formed from ascorbic acid by heating, as this radiation scatter detection method has been applied for studying the thermal decomposition of several organic substances.4 The windowless version of flow cuvettes is useful for preventing the base-line drift experienced in earlier studies,4 which was caused by partial deposition of the aerosol particles on the cuvette windows. A Pye Unicam Model SP190A atomic absorption spec- trometer (equipped with a laboratory-made integrator unit) was operated under optical conditions suitable for the isolation of the Cd 228.8-nm line (Cd hollow-cathode lamp).The apparent absorbance signal caused by the radiation scatter of aerosol particles was detected at the same wavelength using a deuterium lamp. The temperature of the sample in the furnace was measured with a Pt/Pt - Rh thermocouple (which was in direct contact with the sample holder) and most of the studies were performed using a constant heating rate (0.8 “C s-I), referred to as the “slow heating” below. To attain “fast heating” for certain studies, the furnace was pre-heated to 950 “C and the sample holder (sitting on the measuring head of the thermo- couple and fixed in an isolating rod) was pushed rapidly into the furnace.Before this insertion, the sample holder was placed at the inlet part of the furnace tube to dry the sample at 130 “C. From the temperature versus time graphs recorded under these conditions an empirical function was obtained for the heating rate (RT): dT dt RT = - = 265 exp where T (“C) is temperature and t (s) is time. Equation (1) gives a good approximation of the heating rate in the range 350-850 “C under the conditions described. Accordingly, the values of RT were 52 “C s-1 at 350 “C and 8 “C s-1 at 850 “C. Heating rates in the ranges 130-350 and 850-950 “C were lower than those calculated from equation (1). The absorbance versus temperature graphs ( A - T curves) were directly recorded and were accepted only when two replicate recordings were in good visual agreement.Supplementary measurements were made with a DuPont Model 915 Thermal Evolution Analyser to investigate the decomposition of ascorbic acid. With this instrument a flame-ionisation detector is used for monitoring the evolution of organic species. For studying the reactions of ascorbic acid and certain metal salts in the solution phase, a Beckman Model 5260 recording spectrophotometer was used. Procedure Ringsdorff Type RWI spectrally pure graphite (“electrograph- ite”) and platinum foil were used to prepare sample holders in a “cup.” shape. Platform sample holders made of solid pyrolytic graphite (Perkin-Elmer) were also used in this work, as this material is becoming increasingly important in ETA methods.The electrographite sample holders were pre- conditioned at 950 “C. In studying the volatilisation of MgC12.6H20 at temperatures up to 950 “C, new sample holders were used for each measurement, because with repeated use MgO(s) accumulated on the surface of the sample holder. Aqueous solutions (1000 mg 1-1 of metal) were prepared from analytical-reagent grade CdC12 .2.5H20 and Cd(N03)2.4H20, from which 5-pl aliquots were injected on to the sample holders. The MgC12 (an aqueous solution of MgC12.6H20) was added to the CdC12 solution prior to injection when applied together, whereas ascorbic acid (5% aqueous solution) was added to the sample holder ( 5 pl) as a separate injection. Argon gas of the highest available purity was used, which, according to the manufacturer, contained 0.01 v01.-% of 02.In one experiment (see Table l ) , argon was passed through a purifying catalyst (BTS-Katalysator , BASF, Ludwigshafen am Rhein, FRG) to reduce the O2 content to 0.0001 vo1.-Yo. Results and Discussion Initial Vaporisation Temperature of Cadmium Metal, Cadmium Chloride and Cadmium Oxide Using a Platinum Sample Holder The characteristics of the initial observation temperature of vaporisation (Ti) have been discussed, and a method was suggested for the calculation of the Ti from vapour pressure and atomic absorption sensitivity data.7 The Ti value for magnesium oxide was determined with the use of a combined graphite furnace - flame atomic absorption method employing silver as a reference element.7 For the volatility studies carried out with the present quartz furnace - flame method, aJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1986, VOL.1 11 Table 1. Initial temperatures of vaporisation obtained using a platinum sample holder. Conditions used in this work: solid sample (s), 2 mg; solution sample, 5 p1 (1 mg ml-* Cd); heating rate, 0.8 "C s-1 Ti(exp.)l"C Ti Atmos- (calc.)/"C Sample phere Ref. 21 This work This work . . . . . . . . Cd(s) Ar* - 260 250 CdCI2.2.5H2O(s) Ar* 330 400 . . . . - CdCl, (aqueous solution) Ar* 240 320 - CdC12(1MHCI) Ar* - CdO(s) . . . . . . Air - - . . . . 280 290 CdO(s) Ar* 930 780 CdO(s) Ari- - 710 790 860 . . . . . . - . . . . . . * Oxygen concentration 0.01 v01.-Yo. -t Oxygen concentration 0.0001 v01.-Yo. reference material with a higher volatility than silver is desired, and NaCl was suitable for this purpose.Details of the measurements and data used for the calculation of Ti values of cadmium compounds will be published in a subsequent paper. A comparison of experimental and calculated Ti values found in this laboratory is given in Table 1, together with the experimental data of Gegus et af.,2J who used a combined atomic emission source. The Ti (calc.) values were derived assuming the following vaporisation reactions: . . . . Cd(s/l)eCd(g) * * (2) CdCl2(s)=CdCl,(g) . . . . . . (3) CdO(s) Cd(g) +?h.O,(g) . . . . (4) The experimental and calculated Ti values for cadmium metal were found to be in close agreement (Table 1). The experimental value for CdCI2 was closer to the calculated value when HC1 was added to the CdC12 solution.For the calculation of Ti(CdO), a purified argon environment was assumed with an oxygen partial pressure of 10-6 bar. The experimental value for Ti( CdO) under these conditions was found to be 80 "C lower than the calculated value. Table 1 indicates that the experimental value for T,(CdO) increased with increasing oxygen concentration in the furnace gas and it reached a value of 860 "C when argon was replaced with an air atmosphere. Comparative measurements were made in a nitrogen furnace atmosphere and it was observed that the A - Tcurves were consistent with the reported influence of oxygen partial pressure on Ti(CdO). The experimental values of Ti in Table 1 are in general agreement with the calculated values. The deviations obser- ved when two different combined sources are used to determine Ti(exp.) are not unexpected, considering the differences in the heating conditions and in the sensitivity of the detection methods.It can be concluded, therefore, that the data in Table 1 confirm the validity of reactions (2)-(4). Ascorbic Acid Fig. 3 represents the graphs characteristic of the thermal decomposition of ascorbic acid (ASC). Curve 1 was obtained by flame-ionisation detection (DuPont instrument) and curve 2 with radiation scatter detection under slow heating condi- tions. According to these curves (which are in close agree- ment), the decomposition of solid ASC takes place in the range 200-300 "C, the maximum rate of the process being at T , = 250-260 "C. When heated to 600 "C, a finely dispersed powder residue was obtained, the amount of which was 25.5% of the loaded sample and 62.3% of the theoretical carbon content of ASC.In the above experiments, solid ASC was decomposed under slow heating. When solutions of ASC were subjected to 100 200 300 400 500 600 Temperaturei'C Fig. 3. Thermal evolution analytical curves of ascorbic acid (ASC) detected by the flame-ionisation method (1) and by the radiation scatter method (2-4). 1, 5 mg of ASC, 2 dm3 h-1 of N2, aluminium holder, RT = 0.5 "C s-1; 2, 10 mg of ASC, 80 dm3 h-I of Ar, platinum holder, R, = 0.8 "C s-1; 3, aqueous solution of 1.5 mg of ASC, electrographite holder, fast heating; 4, as 3 but 3 mg of ASC thermal decomposition under fast heating conditions (curves 3 and 4 in Fig. 3), the maximum rate of decomposition was found at T,,, = 400-410 "C, about 150 "C higher than for solid ASC under slow heating conditions.In the experiments relevant to Figs. 4-9, a smaller amount of ASC (250 pg) was used and its radiation scatter signal could not be detected in the fast heating mode either. Magnesium Chloride Previously a combined graphite furnace - flame AAS method was used to investigate the volatilisation of hydrated magne- sium chloride (0.1 M HC1 solution) in the range 560-2400 "C applied on a polycrystalline electrographite sample holder .7 It was concluded that with the hydrated salt, partial hydrolysis takes place during heating, and a fraction vaporises according to the reaction MgC12(s) + MgCl,(g). The extent of hydrolysis depends greatly on the conditions of heating and the mass of the sample.Similar conclusions can be drawn from the A - T curves presented in Fig. 4. When the hydrated salt is heated slowly the evolution of MgC12(g) can be detected at Ti = 630 "C (curve 1). The absorbance maxima seen at higher temperatures (absorbance > 1) suggest that vaporisation takes place from a mixed condensed phase. In fact, several magnesium hydroxychloride compounds have been identified in such a melt.26 When an aqueous solution of 2% mlV Mg (MgC12), i . e . , 0.1 mg of Mg (MgCI2), was subjected to slow heating (RT = 0.8 "C s-I) a peak was not observed up to 950 "C and the measurements were therefore continued in a fast heating mode (curves 2-6 in Fig. 4). Using a pyrolytic graphite sample holder no signal was detected with 10 pg of Mg (MgC12), whereas application of 20 and 40 pg of Mg (MgCI2) resulted in curves 2 and 3, respectively [according to equation ( 1), RT = 20 "C s- 1 applied at 600 "C using the fast heating mode].The first peak in the range 170-320 "C (curve 3) can be explained by a blowing effect (removal of solid or liquid particles of sample) of the12 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1986, VOL. 1 HCl(g) evolved rapidly under fast heating. The hydrochloric acid was produced by the hydrolysis of the salt, and during a short drying period (up to 130 "C) it was not removed completely. It is expected that only the peak at Ti = 580 "C belongs to the MgC12(g) formed by volatilisation. With ASC added under otherwise identical conditions, no MgCl,(g) evolution is indicated (curve 4).Using an electrographite sample holder a vaporisation peak could not be detected with 20 pg of Mg (MgClZ), and on applying 40 pg of Mg (MgC12) curve 5 in Fig. 4 was recorded with a much smaller peak than that seen in curve 3 (pyrolytic graphite). These results suggest that MgC12.6H20 has a greater tendency to hydrolyse on the surface of the porous electrographite than on the surface of non-porous pyrolytic graphite. In the presence of ASC no MgClz(g) was produced from the electrographite surface either (curve 6). In summary, the evolution of MgC12(g) from hydrated MgC12 is hindered by the following factors: (a) a decrease in the mass or concentration of MgC12, (b) the use of a porous graphite sample holder, (c) a decrease in the heating rate and (d) the addition of ASC.The role of factors (a), (b) and (c) can be explained by their effect on the hydrolysis of magnesium chloride, as follows. With dilute aqueous solutions of MgC12, hydrolysis takes place at room temperature even in a sample holder with a non-reactive surface, and the degree of hydrolysis increases with sample dilution. It is known that adsorption hydrolysis takes place from neutral electrolyte solutions on the surface of activated carbon.27.28 It is likely that the sample holder prepared from polycrystalline electrographite and pre- conditioned at 950 "C has adsorption characteristics similar to those of activated carbon. The surface of carbon ignited in an inert gas of low oxygen content becomes basic and can bind acids.27 The binding of acids increases the hydrolysis of salts, and the hydroxides produced are retained in the pores of the activated carbon.* With more concentrated solutions hydroly- sis is limited at room temperature, and hydrated magnesium chloride remains on the surface when the temperature is increased to 130 "C. Water of crystallisation is released on further heating which can induce hydrolysis if the heating and the evaporation of water are not too rapid.7726 It may be assumed that the ASC additive blocks, at least partially, the pores of the graphite surface, resulting in a decrease in the adsorption hydrolysis of MgC12. From this, an increase in MgC12(g) evolution is expected if the adsorption hydrolysis remains the dominant process in the presence of ASC also.Because experience contradicted this expectation (curves 4 and 6 in Fig. 4), another hypothesis had to be found to explain the effect of ASC. Molecular absorption spectrophotometric measurements of pure ASC (10-4 M) and a mixture of MgC12 and ASC (both M) indicated that in the presence of MgC12 the absorption maximum of pure ASC was shifted from 258 to 248 nm, while the molar absorbance decreased by a factor of 0.67. Hence it can be concluded that the ASC reacts with MgC12 in aqueous solution at room temperature. During the thermal investig- tions (Fig. 4), the ASC was present in large excess, which would help to promote this reaction. The hydrochloric acid liberated in the solution phase is evolved in the first period of heating and the compound formed is decomposed to MgO(s) at higher temperature.* It is noted that adsorption hydrolysis can also take place on oxidised metal surfaces.27 In previous studies in this laboratory,2' certain anomalies (memory effect, non-linearity of analytical curves, etc.) were observed on the pneumatic nebulisation of neutral solutions of several salts (magnesium and cadmium salts, among others) and were explained by taking into account the adsorption hydrolysis that can occur on the surface of the metal injector tip of concentric nebulisers. 240 170 I 320 4- E r----1 i r" &20 5J 3 590 r-- I, 1- 4 760 630 r I I I I I I I I I 100 200 300 400 500 600 700 800 900 Tern perature/"C Fig. 4. Absorbance versus temperature curves for magnesium chloride using flame atomic absorption detection.1, 5 mg of MgC12.6H20, platinum holder, slow heating; 2,20 pg of Mg, pyrolytic graphite holder, fast heating; 3, as 2 but 40 pg of Mg; 4, as 2 but 40 pg of Mg + 250 pg of ASC; 5 , as 3 but electrographite holder; 6, as 4 but electrographite holder Under the spectrophotometric conditions described, the molecular absorption of a CdC12 - ASC solution was also studied (the vaporisation patterns of the dry residue of this mixture are depicted below). It was observed that the molecular absorption peak occurred at 261 nm, i.e., at a higher wavelength than that of pure ASC, and the molar absorptivity of the two-component system was 0.71 times lower. These results suggest that ASC also reacts with CdC12 in aqueous solution at room temperature. Recently, Slavin et aZ.30 summarised the importance of chloride interference effects in ETA methods.It has been found (see citations in reference 30) that the interference of MgC12 on the determination of lead is smaller if uncoated graphite tubes are used instead of pyrolytically coated tubes. This is explained by assuming that the "ordinary graphite" binds MgC12 to a greater extent by an intercalation process and thus the MgC12(g) is evolved at higher temperature than lead without exerting a gas-phase interference on this analyte. It seems to us that the intercalation of MgC12 was based on the analogy of the characteristics of other known examples of chlorine-containing compounds and there was no direct proof for this particular case. In the light of our results, it is clear that using an ordinary graphite tube in contact with the aqueous solution of MgC12 the hydrolysis of this salt takes place to a greater extent than on the surface of the pyrolytically coated tube during the pre-heating stage.Therefore, the evolution of MgC12(g) in the atomisation stage is less or it does not occur at all, depending on the other factors discussed above. Consider- ing the future prospects of solid pyrolytic graphite tubes and glassy carbon tubes,3O it can be predicted from our results that lowering the heating rate in the pre-heating stage will be more important with these materials for decreasing chloride inter- ferences. It must be noted, however, that the explanation and the prediction suggested do not apply to non-hydrolysing halide salts such as NaCl.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1986, VOL.1 13 Cadmium Nitrate Thermoanalytical (TG - DTG - DTA) studies31 of Cd(N0&.4H20 (75 mg, Pt sample holder, air atmosphere, RT = 0.05 "C s-1) showed that water of crystallisation was evolved in two steps in the range 50-250 "C. The anhydrous nitrate salt melted at 390 "C (DTA peak). However, a slow decomposition was observable from 330 "C and this process was completed at 430 "C. With the present method, applying 3 mg of Cd(N03)2.4H20 in a platinum sample holder (Ar atmosphere, RT = 0.8 "C s-1) and using flame atomic absorption detection, a sharp peak in the range 430-480 "C ( T , = 470 "C) and a larger second peak with Ti = 800 "C were observed. The first peak could not be detected with the cold vapour method, in contrast to the second peak. From radiation scatter measurements in a cuvette, it became evident that the low-temperature peak was caused by the blowing effect of the nitrogen oxide gases evolved rapidly from the melt of Cd(N03)2.It is not unexpected that with a heating rate 16-fold higher than that used in TG - DTG - DTA measurements the evolution of nitrogen oxide gases proceeds at a temperature higher by about 50 "C. With graphite furnace ETA methods, the atomisation of CdO(s) produced from Cd(N03)2 is explained by some workers18.20 as a solid-phase reduction -and vaporisation process, whereas others12J7J9,32 regard it as a thermal dissociation reaction. With the latter process some workers have proposed the interim formation of Cd(l) ,17J9 whereas others assume the validity of equation (4) under certain conditions.12732 Salmon and Holcombe20 found a higher appearance temperature of cadmium when using an oxygen content greater than 0.01% in the argon. This was explained by the chemisorption of oxygen at 500-900 "C in the graphite furnace, which decreased the rate of solid-phase reduction of CdO(s). L'vov and Ryabchuk32 questioned the possibility of solid-phase reduction on a graphite surface below 900 "C. They concluded also that graphite pre-heated to 200-600 "C in argon containing 0.5-1% of O2 binds oxygen at a higher rate than the graphite pre-heated in pure argon. More recently, Sturgeon et aZ.33 described a method for the determination of 0 2 in a graphite tube atomiser. It was shown unambiguously that by low-temperature heating in an argon atmosphere at an increased oxygen concentration, the graphite surface was activated for scavenging O2 at a much higher rate at elevated temperature.It has been proposed that oxidising gases formed by the thermal decomposition of oxyacid salts can also activate the graphite surface.2O It is also known that electrographite has a higher activity than pyrolytic graphite towards oxygen, and repeated heating at <2000 "C generally increases the affinity of graphite for oxygen.20,32,33 In Fig. 5, the A - T curves obtained with the cold vapour detection method are shown for Cd(N03)2 and Cd(N03)2 + ASC aqueous solutions when electrographite and pyrolytic graphite substrates are used (slow heating). The correspond- ing curves obtained with the flame atomic absorption detec- tion were almost identical, so they are not presented here.Curves 1-4 in Fig. 5 illustrate that atomic cadmium vapour is evolved inside the furnace and although the atoms are mixed with cold argon (first cooling step) and then with air (second cooling step, see Fig. l), free cadmium atoms still remain in a sufficient amount for atomic absorption detection. * Under the experimental conditions relevant to curve 1 (cold vapour detection), the time integrated signal was only 0.97 times ~ * It is noted that in earlier studies in this laboratory a similar characteristic of zinc atomic vapour was observed.6 Including the known case of mercury vapour, it follows that all the elements in Group IIB of the Periodic Table can be detected by the cold vapour atomic absorption method under certain conditions.600 1 710 I a NI 360 I I I I 1 I I 200 300 400 500 600 700 800 900 Tern peraturePC Fig. 5. Absorbance versus temperature curves for cadmium nitrate using cold vapour atomic absorption detection. Aqueous solutions, 5 pg of Cd, 250 pg of ASC, slow heating. 1, Cd(N03)*, electrographite; 2, Cd(N03), + ASC, electrographite; 3, Cd(N03),, pyrolytic graphite; 4, Cd(N0J2 + A X , pyrolytic graphite smaller than that using the flame atomic absorption detection. It is calculated that in the latter instance the rate of gas flow in the observation zone is about nine times higher than in the former. This and the close agreement of the integrated absorbance signals suggest that about one ninth of the material reaching the observation zone is atomic cadmium vapour.It should be noted that if the hot cadmium vapour evolved inside the furnace (Ar) is mixed with air in the first cooling step (see Fig. l), oxidation occurs and no cadmium signal is obtained without the use of the flame. No molecular absorption or non-specific scattering of the hollow-cathode lamp radiation was observed with or without a flame, as indicated by the zero absorbance signals measured when a deuterium lamp was used. This observation applies to the other spectral studies discussed in the following sections. As shown by curve 1 in Fig. 5, the evolution of Cd(g), from CdO(s) produced on decomposition of Cd(N03)2 on a porous graphite surface, is observed at 580 "C, which is 200 "C lower than the Ti(CdO) found for vaporisation from a platinum surface (Table 1).This can be explained by considering the high activity of pre-heated electrographite for oxygen,20J2J4 which results in a lower temperature of the reaction given by equation (4). The evolution of Cd(g) takes place at a much lower rate from a pyrolytic graphite surface (curve 3) than from an electrographite surface (curve l ) , which should be related to the lower activity of the pyrolytic sample holder towards oxygen. When ASC is added to Cd(N03)2 in an electrographite sample holder (Fig. 5, curve 2), an additional peak appears at Ti = 310 "C. The intense decomposition of ASC is completed at 300 "C (Fig. 3, curves 1 and 2), which just precedes the evolution of Cd(g). It may be assumed that the ASC or its decomposition products undergo reaction with Cd(N03)2 in the condensed phase and the compound formed decomposes to Cd(1) with further heating (the melting-point of cadmium metal is 321 "C).The vaporisation of cadmium metal was observable at Ti = 260 "C (Table l ) , i.e., at a lower temperature than under the conditions relevant to curve 2. It follows that in the latter instance the rate-determining step of14 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1986, VOL. 1 Cd(g) evolution is the decomposition of the intermediate compound. It can also be concluded from curve 2 that CdO(s) is formed from the major part of cadmium, which vaporises similarly to the situation when no ASC is present (curve 1). It is interesting that the effect of ASC additive becomes more pronounced with the use of a pyrolytic graphite sample holder (Fig.5 , curve 4). The vaporisation process taking place according to equation (2) is of greater significance and the second peak is shifted to a lower temperature (Ti = 450, T, = 600 "C). In curve 2 (electrographite) a slow evolution of Cd(g) between the two major peaks is also observable, which becomes a dominant peak in curve 4 (pyrolytic graphite). The significant differences between curves 2 and 4 can be correlated with the alteration of the form of the sample deposit. In the former instance the cadmium compounds are in contact with both the graphite crystals and the decomposition product of ASC in the pores of the electrographite. With the pyrolytic graphite the cadmium compounds are distributed in the decomposition product of ASC and this mixture forms an outer layer on the sample holder.The shift of the second vaporisation peak to lower tempera- tures (Fig. 5 , curve 4) may be explained by assuming that the finely dispersed carbon powder produced from ASC has even higher activity towards oxygen than the porous electrograph- ite. However, it is also possible that the decomposition product of ASC at 450 "C is not a pure carbon, but contains certain organic compounds that cause the reduction of CdO(s). According to the thermal evolution curve of ASC recorded with the use of flame-ionisation detection (Fig. 3, curve l), a small amount of decomposition product is also evolved at 330-530 "C. This must be some carbon and hydrogen containing gas that is not detectable with the radiation scatter method (curve 2).The curves presented in Fig. 6 were obtained when fast heating was applied for studying Cd(N03)2 and a Cd(N03)2 + ASC mixture. Recordings with flame atomic absorption detection (curves b) are also included in addition to the recordings relevant to the cold vapour detection method (curves a). Comparing these curves with those in Fig. 5, it is clear that the peak heights are greatly increased (the scale of absorbance is reduced to one quarter), but Ti and T, data are only slightly changed owing to the higher heating rate. It should be noticed, however, with this "fast heating mode" the heating rate decreases exponentially with increasing tempera- ture [see equation (l)].A comparison between curves 4a and 4b shows that cold vapour detection resulted in smaller peaks than flame atomic absorption detection in the presence of ASC using a pyrolytic sample holder. The time-integrated absorbance of curve 4a was a factor of 0.55 smaller than that of curve 4b. It can be seen from curves 3 and 4 in Fig. 3 that intense aerosol formation takes place from ASC in the range 330-510 "C when fast heating is applied. It is likely that the aggregates formed from ASC will adsorb a significant fraction of the cadmium atomic vapour, which otherwise tends to be deposited on the walls of the transport system.8 Hence the aggregates concerned serve as effective carriers of the cadmium vapour, resulting in an improved transport effi- ciency.8 As the aggregates are atomised in the flame the atomic absorption signal increases.Without flame atomisation (curve 4a), the fraction of cadmium adsorbed on the aggreg- ates cannot be detected by the atomic absorption method, and consequently a decrease in the signal is observed. Supplementary studies of the vaporisation of CdO(s) were performed using trace amounts of hexane vapour and trace amounts of isopropyl alcohol in the furnace atmosphere (Ar). Evolution of Cd(g) could be detected with the cold vapour method at Ti = 440 "C and Ti = 280 "C, respectively, owing to the heterogeneous reactions between CdO(s) and the gaseous decomposition products of these organic substances. Low- temperature evolution of cadmium (<450 "C) also takes place from mixtures of Cd(N03)2 and carbohydrates (e.g., glucose). It is known that the major part of the trace amounts of cadmium in tobacco is transferred into the cigarette smoke.34 This must be the consequence of the low-temperature reduction of cadmium compounds by the organic compounds and/or their decomposition products present. The trace amounts of cadmium in cigarette smoke could easily be detected with the present combined system using flame atomic absorption detection. To introduce the cigarette smoke into the transport system the cigarette was attached to the inlet of the "auxiliary cooling gas" (see Fig. 1) and was sucked in by the gas injector. Cadmium Chloride According to the thermoanalytical (TG - DTG - DTA) studies31 of crystalline CdC12.2.5H20 (48 mg, Pt sample holder, air atmosphere, RT = 0.05 "C s-I), water is released in two steps in the range 3&120 "C.The residue, which is CdC12, melts at 540 "C (DTA peak) and evaporates completely on heating to 690 "C ( T , = 660 "C). With the present method using flame atomisation, we observed the evolution of CdC12(g) at Ti = 400 "C when solid hydrate salt was applied (2 mg, Ar atmosphere, RT = 0.8 "C s-I), irrespective of the material of the sample holder (Pt, Si02 or graphite). The A - T curves of aqueous solutions of CdCl2 using electrographite sample holders and slow heating are shown in Fig. 7. Curves a and b represent absorbance measurements obtained with cold vapour detection and flame atomic absorption detection, respectively. According to curves l a and lb, obtained without additives, the first vaporisation peak (Ti = 370 "C) cannot be detected without flame atomisation, i.e., this peak is indicative of CdC12(g) evolution [equation 61 0 41,O 400 600 4b ' 740 I 730 I 4a 600 3b 3a : I 2 hl 3 2b 1 I I I I I 1 I 200 300 400 500 600 700 800 900 Tern peratu rePC Fig.6 . Absorbance versus temperature curves €or cadmium nitrate using (a) cold vapour detection and (b) flame atomic absorption detection. Aqueous solutions, 5 pg of Cd, 250 pg of ASC, fast heating. 1, Cd(N03)2, electrographite; 2, Cd(N03)2 + ASC, electrographite; 3, Cd(N03)*, pyrolytic graphite; 4, Cd(N03)2 + ASC, pyrolytic graphiteJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1986, VOL. 1 15 370 I 7 20 Ah ._ 370 790 4a E L 700 650 I 790 2 450 3 3b N 67O7Fo 800 3a 2b 2a I 440 370 I 500 600 l b - 1 600 I 1- l a I I I I 1 I 1 L 200 300 400 500 600 700 800 90 TemperaturePC Fig.7. Absorbance versus temperature curves for cadmium chloride using (a) cold vapour detection and (b) flame atomic absorption detection. Aqueous solutions, 5 pg of Cd, 5 pg of Mg (MgCl,), 250 pg of ASC, slow heating, electrographite sample holder. 1, CdCI,; 2, CdC1, + ASC; 3, CdCI, + MgCl,; 4, CdC12 + MgCI2 + ASC 400 I 4b 4a 3a A U I \ n l b l a 200 300 400 500 600 700 800 900 Temperature/'C Fig. 8. Absorbance versus temperature curves for cadmium chloride using (a) cold vapour detection and (b) flame atomic absorption detection. Conditions as in Fig. 7, but pyrolytic sample holder. 1, CdCl2; 2, CdCI2 + ASC; 3, CdC12 + MgCI2; 4, CdCl2 + MgCl2 +ASC (3)]. Cadmium chloride dissolved in water undergoes signifi- cant hydrolysis on the surface of electrographite in a similar manner to magnesium chloride (see above).This is proved by the vaporisation peaks with Ti = 600 and T, = 740 "C (curves l a and lb), which correspond to the evolution of Cd(g) from CdO(s) formed by hydrolysis of the chloride salt and decomposition of the hydroxide. It is interesting that these peaks are broader and appear at slightly higher temperature than the peak relevant to the same process [equation (4)], but the CdO(s) is produced by the decomposition of Cd(N03)2 (Fig. 5, curve 1). This difference may be related to the activation of the graphite surface by the oxidising gases evolved from the nitrate salt, which is not the case with the chloride salt.The low-temperature peaks (Ti = 320 "C) of curves 2a and 2b in Fig. 7 were obtained with ASC additive in a similar temperature range to that with Cd(NO& (Fig. 5 , curve 2). These correspond to the vaporisation model according to equation (2), as discussed above. The similarity of the shapes of curves 2a and 2b (without and with flame atomisation, respectively) suggests that no CdC12(g) is evolved from the residue of the aqueous solution of CdC12 + ASC mixture. This is due primarily to the reaction of these compounds in the solution phase, concluded from the molecular spectropho- tometric studies (see above). The second vaporisation peaks at T, = 630-640 "C (curves 2a and 2b) may be explained by the activity of the decomposition product of ASC for oxygen [vaporisation model according to equation (4)], discussed above.When MgC12 additive was employed with CdC12 (Fig. 7, curves 3a and 3b), similar A - Tcurves were obtained to those without additive (curves l a and lb). The curves recorded for the mixture of Cd(N0?)2 + MgCI2 (5 pg of Cd + 12 yg of Mg) were also almost identical with these. The first peak (Ti = 370 "C, curve 3b) is indicative of the fraction of cadmium liberated as CdC12(g), whereas the peaks in the range 650-800 "C are due to the evolution of Cd(g) from CdO(s). The fact that the MgC12 (5 pg of Mg) added does not influence these processes significantly is the consequence of its complete hydrolysis under the'conditions applied. It appears from consideration of curves 4a and 4b in Fig. 7 that the addition of ASC to the mixture of CdC12 + MgC12 results in the evolution of Cd(g) similarly to the situation when no MgC12 is present.However, in the latter instance (curves 2a and 2b) an interim peak is obtained at T, = 630-640 "C, which is not observed on addition of MgC12 (curves 4a and 4b). It is probable that the MgO(s) formed by hydrolysis decreases the affinity of the decomposition product of ASC for oxygen and this is reflected in the difference concerned. The A - T curves shown in Fig. 8 were obtained using a pyrolytic graphite sample holder, all other conditions being the same as those for Fig. 7. The significant changes in the curves relative to those in Fig. 7 are due primarily to the fact that virtually no hydrolysis of cadmium chloride takes place on the pyrolytic graphite surface, A much lower degree of magnesium chloride hydrolysis was also found on this surface (Fig.4). Hence the evolution of CdC12(g) (curves 1 and 3) and that of the Cd(g) (curves 2 and 4) takes place in a single step for the total (or dominant) mass of the cadmium sample, resulting in an increase in the magnitude of the vaporisation peaks. The A - T curves shown in Fig. 9 were obtained using fast heating. All the other conditions were identical with those in Fig. 8, except the absorbance scale, which was reduced to one fifth. The blowing effect of HCl(g) released at a high rate under fast heating conditions can be observed on several curves in the range 190-320 "C, in an analogous manner to that obtained with MgC12 solutions (Fig. 4).The atomic absorption detection of the cadmium-containing particles removed from the sample could only be detected, of course, when flame atomisation was used (curves b). Apart from these non-JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1986, VOL. 1 840 4b E, 4a LOU a1 r '1 3a 250 I 2b 2a I 720 I I b la I I I I I I I I 200 300 400 500 600 700 800 900 Tern peratu re/"C Fig. 9. Absorbance versus temperature curves for cadmium chloride using (a) cold vapour detection and (b) flame atomic absor tion detection. Conditions as in Fig. 8, but fast heating. 1, CdC12; 2, &C12 + ASC; 3, CdCI2 + MgC12; 4, CdC12 + MgC12 + ASC Table 2. Initial temperature of vaporisation obtained using electrographite and pyrolytic graphite sample holders. Conditions for measurements (except where indicated in footnotes): aqueous solutions; argon atmosphere with 0.01 vol.-O/~ of 02; heating rate 0.8 "C s-1; flame atomic absorption detection TJ"C Electro- Pyrolytic Sample* graphite graphite Ascorbicacid(ASC) .. . . . . 3201 220$ MgC12 (40) Cd(N03)2 + ASC (5 + 250) . . . . 310; 580 300; 450 . . . . . . . . 620 580 600 Cd(N03)2 (5) . . . . . . . . 580 CdCI2 (5) . . . . . . . . . . 370; 600 340 CdC12 + A X (5 + 250) . . . . . . 320;550;700 330 CdC12 + MgC12 (5 + 5) . . . . . . 370;650 350 Cd(N03)2 + MgC12 (5 + 12) . . . . 380;650 340 CdC12 + MgC12 + ASC (5 + 5 + 250) . . . . . . . . . . . . 320;600 340;480 * Values in parentheses are amounts of Mg, Cd or ASC, t Aqueous solution of 3 mg of ASC, fast heating, flow cuvette. $ Solid 10 mg of ASC, slow heating (0.8 "C s-l), flow cuvette.§ Solid 2-mg sample. CdO(s)$ . . . . . . . . . . . . 630 720 respectively, in pg. volatilisation peaks, the evolution of CdCl,(g) (curves 1 and 2) and that of Cd(g) (curves 2 and 4) is observed over broader temperature ranges than with the use of slow heating conditions (Fig. 8). It can also be seen that when ASC is added the areas of the peaks are much smaller with the cold vapour detection method (curves 2 and 4). This again can be explained by the adsorption of atomic cadmium vapour on the aggregate particles formed from ASC under fast heating conditions. This effect is more significant with the use of a pyrolytic (Fig. 9) than with an electrographite sample holder (Fig. 6 ) . Some of the initial temperature of vaporisation data obtained with the use of graphite sample holders are sum- marised in Table 2.Conclusions Conventional electrothermal atomisers use heating rates that are several orders of magnitude higher than those used with the quartz furnace applied in this work. An additional difference is that the mass of analyte measured in this study is several orders of magnitude higher than that in ordinary ETA measurement. The additives (possible sample matrices) applied in this work were, however, in the same mass range (5-250 pg) that is often applied in electrothermal atomisers. Obviously, these factors should be considered if the results obtained with the quartz furnace are used to interpret the processes taking place in electrothermal atomisers. Owing to the partial hydrolysis of cadmium chloride a double vaporisation peak was obtained (Fig.7, curve lb) that was similar in this respect to that reported previously when zinc chloride6 and magnesium chloride7 were investigated under slow heating conditions. The results obtained for zinc chloride can be compared with those of Yanagisawa et aZ.,35 who applied an extremely high heating rate (about 7 X 104 "C s-1) using a combination of a tungsten filament vaporiser and a low-pressure microwave-induced plasma. The emission signals were monitored with a storage oscilloscope.3~ By increasing the mass (0.0154.15 ng of Zn) of zinc chloride in a 0.001 M acidic solution, a single peak was observed. However, with a further increase in the sample mass (0.5 ng or higher), an additional, lower temperature peak also appeared, the height of which increased much more rapidly with increasing analyte mass than a linear relationship would give.This result is identical in essence with that obtained by us for zinc chloride6 and magnesium chloride ,7 with the difference that Yanagisawa et aZ.35 observed this phenomenon with a sample mass several orders of magnitude lower. Under the conditions described,35 double vaporisation peaks were also observed with zinc nitrate (5 ng of Zn), which cannot be explained by the partial hydrolysis of this salt. We also observed the double curves when using a relatively high mass of solid Cd(N03)2.4H20 with slow heating (see above). It turned out, however, that the low-temperature peak was due to the blowing effect of the nitrogen oxide gases evolved rapidly from cadmium nitrate.This can also happen with other nitrate salts, provided that the sample is not trapped in the pores of the sample holder. Based on the examples given above, it is clear that when using a higher heating rate similar evolution patterns may be observed with a small sample mass to those obtained with a high sample mass using a low heating rate. Campbell and Ottawaylo reported a four-fold decrease in the cadmium signal in the presence of 5 pg of Mg (MgC12) when applying a commercial graphite tube atomiser. Using slow heating of the dried (100 "C) residue of a diluted (1 + 1) sea-water sample, a single atomisation peak was obtained. In contrast, we obtained a double vaporisation peak in the presence of MgC12 (5 pg of Mg) when using flame atomisation and an electrographite sample holder (Fig.7, curve 3b). (In the work cited,lO the quality of the graphite tube was not mentioned, which probably means that a graphite tube without a pyrolytic coating was used.) It must be concluded that the fraction of cadmium evolved as CdC12(g) at a lower temperature was not dissociated in the graphite tube atomiser and could not be detected. Using an extremely high heating rate,l9 it is likely that the temperature of the gas within the graphite tube reaches a value at which the dissociation of metal chloride molecules occurs to a significant extent before they leave the tube. However, under such conditions a larger amount of MgC12(g) is also formed, as a shorter time isJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1986, VOL.1 17 available for hydrolysis to occur in the condensed phase. As a result, the chlorine species formed from the partial dissocia- tion of MgClz(g) decreases the dissociation of CdC12(g). By applying a platform sample holder9J5 or constant- temperature furnace,l6 the gas temperature inside the furnace is increased above the vaporisation temperature, so the gas-phase dissociation can be enhanced, resulting in a decrease in halide interference effects. According to our observations, “drying” at temperatures up to 130 “C using a step mode of heating is often not sufficient to drive out the hydrochloric acid formed by hydrolysis froml0- 20 pg salt matrices. However, if slow heating is applied, as much as 100 pg of Mg (MgC12) can hydrolyse completely in the aqueous solution when heated up to 300 “C.The large mass of non-hydrolysing NaCl present together with MgC12 in sea water may hinder the hydrolysis of MgC12. With such samples, NH4N03 or HN03 matrix modifiers have proved to be effective for the decomposition of the halide salts during the pre-heating stage.9 The vaporisation temperature of cadmium could be increased by the addition of (NH4)2HP04 or NH4H2P04 matrix modifiers.9330 However, the vaporisation temperature of cadmium (i.e., the permissible char temperat- ure) changed with the sea water to additive ratio.9.30 It follows that the relative concentrations of the matrix components and additives and the rate and time of pre-heating are critical factors in determining the magnitude of interference effects.The question9 of why increasing amounts of alkali and alkaline earth metals reduce the temperature at which char losses begin to appear can probably be answered by studying the thermal behaviour of the sample matrices and additives. For this purpose, combined spectroscopic sources seem to be more suitable than electrothermal atomisers, as demonstrated in this and previous investigations.7 It has been shown here that in the presence of ASC, Cd(1) is formed from CdC12 during heating on a pyrolytic graphite surface, which results in low-temperature evolution of Cd(g). MgC12 also reacts with ASC and less volatile MgO(s) is produced by thermal decomposition. These processes have been utilised to separate the atomic absorption signal of cadmium with time from the background signal of the matrix .I2 The aggregates (aerosol particles) formed from the ASC adsorb a fraction of the cadmium vapour.However, by applying a heating rate of 1000-2000 “C s-1 in the atomisation stage, which is readily available with commercial graphite atomisers, the aggregates will probably be decomposed before leaving the tube and therefore no decrease of the cadmium signal is expected. This was actually found in practice.12 Our experimental results support the assumption of L’vov and Ryabchuk32 that the vaporisation of CdO(s) formed from an aqueous solution of cadmium salts takes place by thermal dissociation in an atmosphere with a low oxygen content. We found that an increase in the oxygen concentration in the furnace gas increases the vaporisation temperature with both platinum and graphite sample holders, which is to be expected according to this vaporisation mechanism.However, when using graphite sample holders the initial observation tempera- ture of vaporisation was always lower than that with a platinum sample holder owing to the reaction between graphite and oxygen (the “purification effect of graphite”).32 The mechanism of oxygen scavenging and its consequences for the thermal dissociation of several metal oxides (including cadmium oxide) have been explained in detai1.32.33 The authors express their gratitude to their colleagues Jeno Paulik and Miklos Arnold for their most valuable contribu- tions in designing the gas purifier and to Vladislav Izvekov for the molecular spectrophotometric measurements.They are also grateful to Mr. Bruno Hutsch (Ringsdorff-Werke, Bonn-Bad Godesberg, FRG) who was kind enough to provide the solid pyrolytic graphite platforms. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. References Van Loon, J. C., Spectrochim. Acta, Part B, 1983, 38, 1509. Prack, E. R., and Batiaans, G. J., Anal. Chem., 1983,55,1654. Hanamura, S., Smith, B. W., and Winefordner, J. D., Anal. Chem., 1983,55, 2026. Kintor, T., Bezur, L., Sztatisz, J., and Pungor, E.,J. Thermal Anal., 1981, 22, 179. Simon, J., Kintor, T., Kozma, T., and Pungor, E., J. Thermal Anal., 1982, 25, 57. Kintor, T., Bezur, L., Sztatisz, J., and Pungor, E., Talunta, 1979, 26, 357. Kintor, T., Bezur, L., Pungor, E., and Winefordner, J. D., Spectrochim. Acta, Part B, 1983, 38, 581. Kintor, T., Spectrochim. Acta, Part B , 1983, 38, 1483. Slavin, W., Manning, D. C., Carnrick, G., and Pruszkowska, E., Spectrochim. Acta, Part B , 1983, 38, 1157. Campbell, W. C., and Ottaway, J. M., Analyst, 1977,102,495. Tominaga, M., and Umezaki, Y., Anal. Chim. Acta, 1982,139, 279. Hoenig, M., and Wollast, R., Spectrochim. Acta, Part B , 1982, 37, 399. Yasuda, S., and Kakiyama, H., Anal. Chim. Acta, 1977, 89, 369. Matsusaki, K., Anal. Chim. Acta, 1982, 141, 233. Kaiser, M. L., Koirtyohann, S. R., and Hinderberger, E. J., Spectrochim. Acta, Part B , 1981, 36, 773. Lawson, S. R., Nichols, J. A., Viswanadham, P . , and Woodriff, R., Appl. Spectrosc., 1982, 36, 357. Sturgeon, R. E., Chakrabarti, C. L., and Langford, C. H., Anal. Chem., 1976, 48, 1792. Czobik, E. J., and Matousek, J. P., Talanta, 1977, 24, 573. Chakrabarti, C. L., Wan, C. C., Teskey, R. J., Chang, S. B., Hamed, H. A., and Bertels, P. C., Spectrochim. Acta, Part B, 1981,36, 427. Salmon, S. G., and Holcombe, J. A., Anal. Chem., 1982, 54, 630. Gegus, E., Kreiter, J., Meray, L., and InczCdy, J., Acta Chim. Acad. Sci. Hung., 1979, 101, 347. Robinson, J. W., and Weiss, S . , Spectrosc. Lett., 1980, 13,685. Aziz, A., Broekaert, J. A. C., and Leis, F., Spectrochim. Acta, Part B, 1982, 37, 369. Crabi, G., Cavalli, P., Achilli, M., Rossi, G., and Omenetto, N., At. Spectrosc., 1982, 3, 81. Kintor, T., Bezur, L., and Pungor, E., Mikrochim. Acta, 1981, I, 289. Petzold, D., and Naumann, R., J. Thermal Anal., 1980,19,25. Lipatov, S. M., “Physical Chemistry of Colloids” (in Hun- garian), AkadCmiai Kiad6, Budapest, 1951. Hassler, J. W., “Purification with Activated Carbon,” Chem- ical Publishing Co., New York, 1974. Kintor, T., Grif, E. H., Nagy, B. H., and Pungor, E., Hung. Sci. Instrum., 1980, 50, 9. Slavin, W., Carnrick, G. R., and Manning, D. C., Anal. Chem., 1984, 56, 163. Liptay, G., Editor, “Atlas of Thermoanalytical Curves, Cumu- lative Index,” AkadCmiai Kiado, Budapest, 1977. L’vov, B. V., and Ryabchuk, G. N., Spectrochim. Acta, Part B , 1982, 37, 673. Sturgeon, R. E., Siu, K. W. M., and Berman, S . S., Spectrochim. Acta, Part B, 1984, 39, 213. Menden, E. E., Elia, V. J., Michel, L. W., and Petering, H. G., Environ. Sci. Technol., 1972, 6 , 830. Yanagisawa, M., Kawaguchi, H., and Vallee, B. L., AnaL Biochem., 1979,95, 8. Paper J5l4 Received May 28th, 1985 Accepted August 23rd, 1985