Mechanism of cobalt atomization from different atomizer surfaces in graphite-furnace atomic-absorption spectrometry

0039-9140/90$3.ocl+0.00 PcrgamonPressplc

Talanrcr, Vol. 37, No. 12, pp. 1111-1117,1990 Printedin GreatBritain

MECHANISM OF COBALT ATOMIZATION FROM DIFFERENT ATOMIZER SURFACES IN GRAPHITE-FURNACE ATOMIC-ABSORPTION SPECTROMETRY C. L. CHAKRABAR~* and S. J. CATHUM Centre for Analytical and Environmental Chemistry, Department of Chemistry, Carleton University, Ottawa, Ontario, Canada KlS 5B6 (Received 20 December 1989. Revised 26 June 1990. Accepted 4 July 1990) Summary-The mechanism of cobalt atomization from different atomizer surfaces in graphite-furnace atomic-absorption spectrometry has been investigated. The atomizer surfaces were pyrolytically coated graphite, uncoated electrographite, and glassy carbon. The activation energy of the rate-determining step in the atomization of cobalt (taken as the nitrate in aqueous solution) in a commercial graphite furnace has been determined from a plot of log /c. us. l/T (for T values greater than the appearance temperature), where k, is a first-order rate constant for atom release, and T is the absolute temperature. The activation energy E, , can be correlated either with the dissociation energy of Coo0 or with the heat of sublimation of Co,, , formed by carbon reduction of COO,, , the latter being the product of the thermal decomposition of Co(NO,),. The mechanism for Co atomization seems to be. the same for the pyrolytically coated graphite and the uncoated electrographite surfaces, but different for the glassy carbon surface. The suggested mechanisms are consistent with the chemical reactivity of the three atomizer surfaces, and the physical and thermodynamic properties of cobalt and its chemical compounds in the temperature range involved in the charring and atomization cycle of the graphite furnace.

The mechanisms of atomization in graphitespectrometry furnace atomic-absorption (GFAAS) have been studied by several authors, who have used different theoretical models based on kinetic,‘v2 thermodynamic*’ or combined kinetic-thermodynamic considerations.6 Investigations of the atomization mechanism requires both kinetic and thermodynamic factors to be taken into account.7 It is also important to investigate the dependence (if any) of the atomization mechanism on the nature of the atomizer surface. Mechanisms for atom formation and dissipation in GFAAS are of importance in designing more efficient atomizers and in optimizing experimental procedures. Valuable information about the mechanism of atomization in GFAAS can be obtained from the absorbance signal profile,

analysis volume at any instant t, as a convolution integral of the form:

THEORY

R(t) = k,n

Paveri-Fontana et a1.8and van den Broek and de Galan have provided an analytical expression for the number of atoms present in the

where k, is the temperature-dependent firstorder rate constant for the combined rate of atom removal, i.e., by diffusion, expansion and convection; re-adsorption onto the atomizer surface is not considered. If it is assumed that

*Author for correspondence.

I n=

S(t’)R(t - t’) dt’ I0

(1)

where S(t) is the rate of atom formation, R(t) is the rate of atom removal and t’ is a variable of integration. s(t) is given by’O S(t) = nOkooexp x,xp(-[:k,exp[-$]df’)

(2)

where n, is the number of analyte atoms initially present, k, the pre-exponential factor, E,, the activation energy, R the gas constant and T the absolute temperature at time t. The rate of atom removal, R(r), is given’s9 by:

1111

(3)

1112

C. L. CHAKRABARTI and S. J. CATHUM

the rate of atom removal is much faster than the rate of atom fo~ation, another form of the Smets equation’ can be obtained from the rate of atom formation [equation (2)] as follows. Assuming that the first-order rate constant k, for the rate of atom formation is approximate by the Arrhenius law, and that the amount of analyte jy S(t) dt left unatomized at time t is expressed” by

s co

S(t) dt

I

=~exp(-Sjk,exp[-~]dt)

(4)

then

J I

A dt

S(t) dt JI

and A and where k, = k, -exp(-EJRT) jy A dt are the absorbance and the integrated absorbance, respectively, at time t. Equation (5) is valid if the rate constant for atom removal, k,, is larger than the rate constant for atom formation, k, This condition can be fulfilled in a graphite furnace by using forced convection of inert gas during atomization.‘*” This point will be raised again in the Results and Discussion section. Using transition-state theory, Sturgeon et aL6 formulated the following expression for the rate constant of atom formation, k, and the change in free energy of activation, AGO”:

.

where AGo* = AHo* - TASO*; AHo* and ASo* are the enthalpy and entropy of activation, respectively, and a is a constant. Sturgeon et aL6 assumed that the term con~ning the linear function of temperature is essentially temperature-independent because of compensating effects from other temperature-dependent terms in the equation. Also, according to transitionstate theory, the rate constant of atom formation, k,, may be formulated by statistical mechanics to yield the expression k, = bT” exp( - E,/RT)

(7)

where b and m are constants; the value of m may be deduced from the partition function. E. is the molar energy change at absolute zero (assumed to be equal to the activation energy E* in the Arrhenius law).

Arranging terms and taking the logarithm of both sides of equation (7) yields logy = -

2.30E;RT+logb

(8)

where y = k,/T”. Equation (8) indicates that a plot of logy us. l/Z’ should be linear with slope - EJ2.303R and intercept log 6. Earlier investigations6 of mechanisms of atom formation in GFAAS showed the importance of the reduction of the metal oxide, MO, according to the equation MO, + C, = CO, + M,

(9)

This simple equation for atom formation was considered to explain the observed activation energy and the appearance temperature in the graphite furnace for many elements, including cobalt. Sturgeon et al.6 found an activation energy of 406 kJ/mole for the atomization of cobalt (as the nitrate) and an appearance temperature of 1430 K for cobalt atom formation, and concluded that Co(NO& was first converted into Cools), which was then reduced to Co,, followed by sub~mation of Co, to Cof,). Ham and McAllister” investigated the atomization of Co in a pyrolytically coated graphite tube furnace operating in a vacuum, by using a rapid-scan mass spectromet~c detection system, and by thermochemical calculation of the equilibrium in the furnace during atomization. Their calculation showed that the atomization reaction was Co,, + Co,, for both cobalt nitrate and cobalt chloride at the appearance temperature for cobalt. However, Grimley et ~1.‘~reported the dissociation energy of COO&)as 361 kJ/mole from mass spectrometry of Cow and CoOe, arising from CoO(,, in a Knudsen cell between 1578 and 1744 K. Thermochemical calculations by Ham and McAllister” demonstrated that if more than 5 x lo- ” mole of cobalt was used as the sample Coo, would replace Co(,) at equilibrium above 1880 K, which would result in a heat of atomization of 309 M/mole for cobalt, as estimated by using Trouton’s rule. These conflicting reports have left the mechanism of cobalt atomization still unresolved. Also, the mechanisms of cobalt atomization from different surfaces in GFAAS need to be investigated. One objective of this work was to investigate the mechanisms of cobalt atomization from different atomizer surfaces, namely pyrolytically coated graphite, uncoated electrographite, and glassy carbon. The other objective was to investigate how the linearity of Arrhenius plots is

Mechanism of cobalt atomization

affected by different values for the constant m in equation (7). EXPERIMENTAL

Apparatus A Perkin-Elmer model 503 atomic-absorption spectrometer, equipped with a deuterium arc background corrector and a model 76B heated graphite atomizer (HGA) was used. The atomic-absorption spectrometer was modified in our laboratory to allow signals to be registered with a time constant of 20 msec. The graphite furnace was heated with a laboratorymade power supply capable of supplying variable heating rates up to 1 K/msec. The signals were recorded with a model 4094 programmable Nicolet oscilloscope, and the integrated absorbance was obtained by using a software package provided by Nicolet Instrument Corporation. A Perkin-Elmer cobalt hollow-cathode lamp was used at a lamp current of 10 mA. The 240.7 nm line of cobalt was used as the analysis line, with a spectral band-pass of 0.7 nm. The temperature of the graphite tube inner surface just below the sample injection hole was measured with a model 1100 automatic optical pyrometer (Ircon Inc., Niles, IL., USA), the temperature being read off the calibration curve supplied by the manufacturer. Atomizer tubes Atomizer tubes fabricated from three different materials were used: pyrolytically coated graphite tubes (Perkin-Elmer, part No. BOO9 l504), uncoated electrographite tubes (PerkinElmer, part No. 0290-1820), and glassy carbon tubes (Sigri RingsdorfI Werke, GmbH, West Germany). All these tubes had the same dimensions: length 28 mm, outside diameter 8 mm, inside diameter 6 mm.

1113

sition and atomization curves. The cobalt sample was atomized in a pyrolytically coated graphite tube, with the furnace used in the gas-interrupt mode. For determination of the activation energy for cobalt atomization from different atomizer surfaces the furnace was used in the gas-flow mode, with an argon flow-rate of 625 ml/mm. Aliquots (10 ~1) of the cobalt solution (0.02 r g/ml) were deposited in the graphite furnace and atomized with a charring temperature of 1070 K and atomization temperature of 2700 K. RESULTS AND DISCUSSION

Decomposition and atomization curves Figure 1 shows the decomposition and atomization curves for cobalt (taken as the nitrate in aqueous solution) atomized from a pyrolytically coated graphite tube. The decomposition curve is defined as that obtained as a function of the charring temperature when the sample is atomized at a constant (optimum) temperature. The atomization curve is defined as that obtained as a function of the atomization temperature when a constant (optimum) charring temperature is used. Each datum point represents the arithmetic mean of three values, corrected for blanks. In Fig. 1, curve A is the decomposition curve and curve B the atomization curve. For the decomposition curve, the sample was dried at 420 K and charred at various temperatures from 670 to 2770 K. The atomization temperature was kept constant at the optimum temperature of 2700 K. For the atomization curve, the charring temperature was kept constant at the optimum charring temperature of 1070 K and the atomization temperature was varied from 1070 to 3 170 K. The graphite tube was cleaned after each run by firing at 3170 K. The data for

Reagents The stock solution of 1000 pg/ml cobalt was prepared by dissolving pure cobalt metal in pure nitric acid (Baker ULTREX) and diluting the solution with ultrapure water obtained direct from a Milli-Q2 water purification system (Millipore Corporation). The test solutions were prepared immediately prior to use, by serial dilution of the stock solution with ultrapure water. Procedure Aliquots (10 ~1) of the cobalt solution (0.02 pg/ml) were used for determining the decompo-

TEMPERATURE, K

Fig. 1. Decomposition and atomization curves for 0.2 ng of Co (taken as the nitrate in aqueous solution in 1% v/v nitric acid) by atomization from a pyrolytically coated graphite tube.. A, decomposition curve; B, atomization curve.

1114

C. L. CI~KRABAR-IY and S. J. CATHUM

the decomposition curve show that the maximum charring temperature that can be used without noticeable loss is - 1100 K. The appearance temperature (defined as the temperature at which the analyte signal is just distinguishable from the baseline) and the optimum atomization temperature (defined as the lowest temperature giving the maximum absorbance) are - 1400 and -2700 K, respectively. The decomposition curve (A) can be interpreted as follows. As the charring temperature is increased, the cobalt nitrate is decomposed at about 970 K (not shown in Fig. 1) to form [email protected] A noticeable loss of Co atoms occurs at a charring temperature of - 1100 K. In Fig. 1, since there is no significant difference between the temperature in the decomposition curve at which the absorbance starts to decrease, and the appearance temperature in the atomization curve (B), it is reasonable to conclude that the decrease in the absorbance is due to loss of Co@)when the Co(NO,), is charred at a temperature > 1100 K in the pyrolytically coated graphite tube. Any mechanism for the atomization of Co(NO,), must be consistent with these observations in order to be acceptable as satisfactory. Rate constants for function

the supply and removal

The time-dependence of the supply of analyte atoms into and their removal from the graphite furnace has been reported in the literature. Fulleti has proposed a model consisting of a mathematical expression with two exponential terms to describe the release and removal of copper in the Perkin-Elmer model HGA-70 graphite furnace. He concluded that the rate of removal exceeds the rate of atom formation by a factor of 3-20, depending on the atomization temperature. More rigorous studies of supply and removal of analyte atoms during the atomization were published later.6*8*9~14*‘5 To verify the results reported earlier by others2.9 that the value of the rate constant, k,, for atom formation is less than the value of the rate constant for atom removal, k,, we will calculate k, and k, at each datum point and present the results graphically to show the difference between the two rate constants. It is assumed that k, is approximated by the Arrhenius law [equation (511and that k, is given’ by k, = kd + k, + k,

(10)

where kd, k, and k, are first order rate constants for loss by diffusion, expansion and convection, respectively. This assumption is approximately true for the system that we are investigating if we exclude (as we have done) other loss processes. For kd we haveI k

(11)

where D,, is the diffusion coefficient at STP, T is the absolute temperature, I is the tube length, and the value of n (to be called the gas combination factor) varies from 1.5 to 2.0 for various combination of gases;” k, is given byi k, = al4T

(12)

where a is the heating rate of the atomizer in K/set. The value of k, is expressed by9 k, = FT/3OOMV

(13)

where F is the flow-rate of the purge gas in ml/mm, A4 is a proportionality constant which depends on the gas flow pattern through the furnace, and V is the analysis volume. Data used in the calculation are given in Table 1. The results are presented in Fig. 2, which shows that k, is greater than k,. Under the experimental conditions shown, our results agree with those of van den Broek and de Galan. Activation nitrate

energy for

atomization

of cobalt

Table 2 summarizes the results for determination of the activation energies for the atomization of cobalt nitrate from the three different surfaces. The standard free energy change at the appearance temperature of 1400 K was calculated by Sturgeon et al6 on the assumption that COO(,) is the precursor for Co,, and was found to be - 100 kJ/mole. We may therefore conclude that the reduction of COO,, by C,,J at the appearance temperature is thermodynamically favorable. Table 1. Data used in the calculation of the rate constants k, and k,+ Activation energy E,, kJ/mole Heating rate K, K/set Pre-exponential factor k, , set -’ Gas constant R, J.mole-‘. K-’ Gas combination factor, n Tube length I, cm Tube volume V, cm3 Diffusion coefficient D0 at 273 K, cm’lsec Proportionality factor, M Flow-rate F, mllmin

387.0 650 2.0 x 10’ 8.314 1.89 ::x 0.097 90.0 625.0

*The temperatures used are the same as in Fig. 3.

Mechanism of cobalt atomization 1.0.

:I/

1115

A B

-1.0.

C

-3.0 ’ 20. -5.0. B

O-00

-7.0 4 4.0

1400

5.0

6.0

T, 10’

TEMPERATURE, K

7.0

K

Fig. 2. Rate constants for the supply and loss of Co, as a function of the absolute temperature: A, rate constant k, for Co atom removal; B, rate constant k, for Co atom formation.

Fig. 3. Arrhenius plots for atomization of 0.2 ng of Co (taken as the nitrate in aqueous solution in 1% v/v nitric acid) in different atomizer tubes: A, pyrolytically coated graphite; B, uncoated electrographite; C, glassy carbon.

We have calculated E, values from the slopes of the Arrhenius plots, drawn by use of the data from the leading edge of the absorbance signal profile (Fig. 3). The slopes were calculated by the least-squares method. The initial linear portion of the slope covers a temperature range of about 340 K. Figure 4 shows that the value of m does not have any discernible effect on the slope of the Arrhenius plots. The effect of the heating rate on E, for cobalt atomization from a pyrolytically coated graphite tube is presented in Table 3. The heating rate a was determined graphically by using the equation T = To + at, where T is the absolute temperature at time t. No significant effect is seen, which agrees with the experimental observations reported by Sturgeon et aL6

In Table 2 there is no significant difference between the activation energies obtained by using the pyrolytically coated graphite surface and the uncoated electrographite surface, but these values are significantly different from the activation energy obtained with the glassy carbon tube If surface spreading and penetration by the aqueous solution of the sample had played a role in the atomization, then all three E, values should be appreciably different because significant spreading and penetration would be expected for the uncoated electrographite surface, but not for the pyrolytically coated graphite surface and the glassy carbon

-3.0.

MECHANISMS

OF ATOMIZATION *

Table 2 presents a comparison of the activation energies for Co atomization from the three different atomizer surfaces. Since the dimensions of the three atomizer tubes are identical, and the appearance temperatures and the heating rate at the appearance temperatures are not very dissimilar, the activation energies obtained can be compared and interpreted in terms of the mechanisms for atom formation of Co from the different atomizer surfaces.

B

-5.0.

-7.0 J 4.5

6.0 T,lO’K

Fig. 4. Arrhenius plots for the atomization of 0.2 ng of Co (taken as the nitrate in aqueous solution in 1% v/v nitric acid) in a pyrolytically coated graphite tube, as a function of m [equation @)I: A, m = 1.5; B, m = 1.2; C, m = 1.0.

Table 2. Activation energy ( f standard deviation of four determinations) for atomization of cobalt from different atomizer surfaces Tube* PG UG Gc

Appearance temperature, K

Atomization temperature, K

Heating rate, K/see

E,, kJ/mole

1420 1450 1500

2700 2700 2700

650 800 850

387 f 15 375 f 19 285&9

lPG, pyrolytically coated graphite tube, UG, uncoated electrographite tube; GC, glassy carbon tube.

7.5

1116

C. L. CIUKRAFMRIIand S. J. CATHUM

surface. The pyrolytically coated surface is expected to show much less spreading than the uncoated electrographite surface, but no penetration as long as the pyrolytic graphite coating remains intact. The near identity of the E, values for the pyrolytically coated graphite surface and the uncoated electrographite surface rules out surface spreading and penetration as responsible for the observed difference in the E, values in the present case. This means that surface spreading and penetration cannot explain the low E, value for the glassy carbon tube. Perhaps the cause of the observed low E, value for the glassy carbon tube lies in a different atomization mechanism. The E, values (Table 2) for the pyrolytically coated tube and the uncoated electrographite tube may be correlated with either the heat of sublimation of Co,,, 428.4 kJ/mole, or the heat of dissociation of Coo(,), 361 kJ/mole (Table 4). Since reduction of COO(,) by carbon is thermodynamically favorable at the appearance temperature of 1400 K, identification of E, by correlating it with the heat of sublimation of Co(,) is not unreasonable. These two mechanisms are shown below as equations (14) and (15).

Table 3. Activation energy ( f standard deviation of four determinations) for atomization of cobalt from a pyrolytically coated graphite tube at different heating rates Appearance temperature, K

Heating rate,* K/set

E., kJ/mole

1410 1440 1420

580 1050 2800

390+12 387 + 12 405 + 16

*At the appearance temperature.

adsorbed onto the condensed-phase CoOo . The expected low surface coverage by CO, (because of its very low partial pressure) will cause the reduction of COO,, to be kinetically unfavorable. Reactions (17) and (18) are coupled in the graphite furnace. The dissociation energy of CoOw is so much smaller than the heat of sublimation of Coo0 [reaction (17)] that the sublimation of COO(,)will occur with the attendant dissociation of CoOti) [equation (15)]. It is possible that the small mass of the sample taken allows the sublimation of COO,, to deplete all COO(,) to form COO,, for which the dissociation energy is so much lower than the sublimation energy of COO,, that the sublimation of Coo0 would result in the dissociation of CoOti), the dissociation energy being observed COO@)+ C@)-co ( (9 + co,,, (14) as the activation energy, Ea. On the basis of the above, it is reasonable to conclude that the more likely mechanism of COO@, + co,, + 0,) (1% atomization of Co(NO,), in the pyrolytically coated graphite tube or the uncoated electroThe mechanism shown as equation (15) requires graphite tube is the dissociation of COO(,) folthe prior formation of COO@,,from COO,,. lowing the sublimation of COO,,. Reaction (14) should be kinetically favorable because the COO,,, formed by the decomposition sublimation dissociation of Co(NO,), in the charring cycle is in intimate -coo COO(s) (B) CO(,) + 0,) contact with a very large excess of Co, (present as the heated graphite tube). Reaction (15) is The E, value of 380 kJ/mole (Table 2) correlates also kinetically favorable at the high temperabetter with the heat of dissociation of CoOu), ture of atomization. 361 kJ/mole (Table 4), than the heat of sublimaThe possibility that any of the following tion of Co,, 428.4 kJ/mole. However, both reactions may be an effective pathway for the mechanisms [equations (14) and (15)] are conformation of Co, should also be examined: sistent with the observations made earlier about the decomposition curve in Fig. 1. COO@)+ CO,, + CO@)+ CO,, (16) The E,, value (285 kJ/mole) of Co atomization COO@)+ COO@) (17) for the glassy carbon tube is more difficult to account for, as it does not correlate with any (18) value of the thermochemical properties given COO(,) + CW+ CO(,)+ CO,,, Although reaction (16) is thermodynamically in Table 1. By thermochemical calculations favorable, it is kinetically less favorable than the Ham and McAllister” have demonstrated that other reactions considered earlier, including if more than 5 x lo-” mole (2.945 ng) of Co is reactions (17) and (18). This is because of the taken, Co,,, should replace Co(,) for equilibria at very low partial pressure of CO@,,.” In order for above 1880 K. Since the E, values for Co CO,,, to reduce Coo(,), COO molecules must be atomization were obtained by using 0.2 ng of

Mechanism of cobalt atomization Table 4. Physical and ~e~~he~~ species Parameter Melting point of Co Boiling point of Co Melting point of Co0 Co,O, transformed into Co,04 Co,O, transformed into Co0 CO@)* co Co,, -+ 2 c?o(8) co,, -+ co< COO@,-+ c o& coo, -+ co, + or&

Co, the Co atoms should be in the solid state rather than the liquid state. However, for a monolayer distribution of the analyte atoms on the atomizer surface, the energy of interaction between the surface and the analyte atoms may affect the activation energy. It has been reported’**‘9 that glassy carbon is chemically less reactive than graphite, and perhaps pyrolytitally coated graphite. If lower chemical reactivity results in a weaker interaction between the glassy carbon surface and the Co atoms on it, and if the measured E, corresponds to this interaction, then E, would have a low value. The stronger interaction between Co, and the p~olyti~lly coated graphite surface or the uncoated electrographite surface should accordingly result in high E, values. The experimental E* values (Table 2) are consistent with this. The E, value (285 kJ/mole) for Co atomization from the glassy carbon surface can be correlated with the heat of vapo~zation, 309 kJ/mole, as estimated by Trouton’s rule. However, such a correlation would require that the cobalt sample be converted into Co,, near the appearance temperature of N 1500 K for atomization from the glassy carbon surface (Table 2). It is possible that cobalt exists as liquid on the tube surface even though its appearance temperature is lower than the melting point, since the latter can be lowered by traces of impurities in the sample. However, it is also possible that the small amount of Co, present can be totally sublimed at a temperature below the melting point of Co,, By thermochemical calculations, the mass of cobalt which is atomized at the appearance temperature of 1400 K to give a measurable signal (at the appearance temperature) is found to be approximately 3.7 x lo-$ ng. This small amount of cobalt can be totally sublimed at well below its melting point, which accounts for the appearance temperature for cobalt also being well below its melting point,

properties

1117 of probable cobalt

Value

Refereuce

1768 K 3373 K 2208 K at 1168K at f223K 428.4 Id/mole 167 M/mole 370 kJ/mole 510.5 kJ/mole 361 kJ/mole

20 13 13 13 ;: 20 z: 23

Acknowledgements-The authors are grateful to the Naturat Sciences and Engineering Research Council of Canada for financial support. One of the authors (Shamil J. Cathum) is grateful to the Government of Iraq for a postgraduate scholarship.

REFERBNC~ 1. B. Smets, Spectruchim. Actu, 1980, 3SB, 33. 2. C. W. Fuller, Analysi, 1974, 99, 739. 3. W. C. Campbell and J. M, Ottaway, Talanta, 19?4,2f, 837. 4. W. Frech, E. Lundberg and A. Cedeqren, Prog. Anal. Atom. Spectrom., 1985, 8, 257. 5. C. L. Chakrabarti, S. B. Chaug and S. E. Roy, Spectrochim. Acta, 1983, 38B, 447. 6. R. E. Sturgeon, C. L. Chakrabarti and C. H. Latqford, Anal. Gem., 1976, 48, 1792. 7. J. P. Matousek and H. K. J. Powell, Spectrochint. Acta, 1988,43B, 167. 8. S. L. Paveri-Fontana, G. Tessari and G. Torsi, An&. Chem., 1974, 46, 1032. 9. W. M. 0. T. van den Broek and L. de Galan, Anal. Chem., 1977,49, 2176, 10. W. Frech, N. G. Zhou and E. Lundberg, S~ctroch~. Acta, 1982, 37B, 691. Il. N. S. Ham and T. McAllister, &ti., 1988, 43B, 789, 12. R. T. Grimley, R. P. Burns and M. G. fnghram, J. Chem. Phys., 1966, #,4158. 13. C. Duval, Inorganic Thermogravimetric Analysis, 2nd Ed., Elsevier, New York, 1963. 14. G. Torsi and G. Tessari, Amzl. Chem., 1975, 47, 839. 15. G. Tessari and G. Torsi, ibid., 1975, 47, 842. 16. H. Falk, CRC Crit. Ra. Anal. Chem., 1988, 19, 29. 17. B. V. L’vov, Atomic Absorption S~ctrochemical Analysis, Hilger, London, 1970. 18. R. E. Sturgeon and H. Falk, J. Anal. Atom. Spectrom., 1988, 3, 27. 19. W. Huettner and C. Busche, 2. Anal. Gem., 1986,323, 674. 20. R. C. Weast, Hamibook of Chemistry and Physics, 66th Ed., CRC Press, Boca Raton, 19854986. Comprehensive Inorganic 21. A. F. Trotman-Dickenson, Chemistry, Vol. 3, p. 1055. Pergamon Press, Oxford, 1973. of Chemist 22. I. S. Kulikov, Thermal foliation Cornpour& p. 106. Israel Program for Scientific Translations, Jerusalem, 1967. 23. M. G. Zhuravleva and 0.1. Chufarov, Tr. fnst, Metali. Akad. Nat& S.S.S.R. Ural Filial, 1959, 3, 63.