2714
Anal. Chem. 1984, 56,2714-2720
enhancements observed in this study is drop size related, as the mixer paddle largely eliminates any enhancement. Such behavior is compatible with the AIR enhancement model. However, enhancements observed with nonionic surfactants are not explicable using an AIR model. This anomaly indicates that some mechanism other than AIR may in fact responsible for enhancements observed with both anionic and nonionic surfactants.
ACKNOWLEDGMENT The loan of spray chambers and nebulizers from PerkinElmer Corp. and Varian Techtron Pty. is gratefully acknowledged. Registry No. DOP, 14117-96-5;SDS, 151-21-3;Tween 20, 9005-64-5;Cu, 7440-50-8;Cr, 7440-47-3. LITERATURE CITED Pungor, E.; Mahr, M. Talanfa 1983, 10, 537. Dean, J. A. "Flame Emission and Atomic Absorption Spectrometry"; Dean, J. A., Rains, T. C., Eds.; Marcel Dekker: New York, 1969; p
(6) Korhnahrens, H.;Cook, K. D.; Armstrong, D. W. Anal. Chem. 1982, 54, 1325. (7) Borowiec, J. A,; Boorn, A. W.; Dillard, J. M.; Cresser, M. S . ; Browner, R. F.;Matteson, M. J. Anal. Chem. 1980, 52, 1054. (8) Smith, D. D. Doctoral Thesis, Georgia Institute of Technology, Atlanta, GA, 1983. (9) Smith, D. D.; Browner, R. F. Anal. Chem. 1984, 56, 2702.
(IO) Browner, R. F.; Boorn, A. W.; Smith, D. D. Anal. Chem. 1982, 5 4 , 141 1. (11) Browner, R. F.; Cresser, M.
S. Spectrocblm. Acta,
Part B 1980, 358,
73. (12) Smith, D. D.; Browner, R. F. Anal. Chem. 1982. 5 4 , 533. (13) Scott, R. H.;Fassei, V. A,; Kniseiey, R. N.; Dixon, D. E. Anal. Chem. 1974, 46,75. (14) Nukiyama, S.;Tanasawa, Y. "Experiments on the Atomization of Li-
quids in an Air Stream": Hope, E.;Transl.; Defense Research Board, Department of National Defense: Ottawa, Canada, 1950. (15) Gustavsson, A. Anal. Chem. 1983, 55,94. (16) Rosen, M. J. "Surfactants and Interfacial Phenomena"; Wiiey: New York, 1978: Chapter 3. (17) Boorn, A. W.; Cresser, M. S.; Browner, R. F. Spectrochlm. Acta, Part B 1980, 355, 832. (18) Boorn, A. W.; Browner, R. F. Anal. Chem. 1982, 54, 1402. (19) Thomas, W. D. E.; Potter, L. J . Colloid Inferface Sci. 1975, 50,397. (20) Cresser, M. S.;Browner, R. F. Appl. Spectrosc. 1980, 34,364.
306. Venable, R. L.; Ballad, R V. Anal. Chem. 1974, 46, 131. Venable, R. L. Report PB-264 184; U. S.Department of Commerce: Springfieldi VA, 1976. Kodama, M.; Miyagawa, S. Anal. Chem. 1980, 52,2358.
RECEIVED for review April 2, 1984. Accepted July 12, 1984. This material is based on work supported by the National Science Foundation under Grant CHE80-19947.
Atomization Mechanism with Arrhenius Plots Taking the Dissipation Function into Account in Graphite Furnace Atomic Absorption Spectrometry Chan-Huan Chung' Department of Chemistry, Faculty of Science, Hiroshima University, Higashisenda-machi, Naka-ku, Hiroshima 730, Japan
The mechanism of atom formatlon and dissipation for seven elements in graphlte furnace atomic absorptlon spectrometry has been studied through a combined ihermodynamlc and klnetic approach In which the rate of atom formatlon ( k ) at a given temperature was calculated by using the corresponding absorbance ( A , ) and the maximum absorbance a,) where kd IS the dlssipatlon ( A m a x ) :k = k&,/(A,,, constant. The Arrhenlus plot obtained for all the cases was linear over the temperature range from the appearance to the maximum absorbance. The activation energy E , of atom formation obtained showed that four atomlzation mechanisms are operative as the rate-determiningprocess: thermal dlssoclatlon of the oxide for Zn, AI, and Mg and of the hallde for Sb; vaporlzatlon of metal for Ag; carbon reduction for Fe; dissociation of the dimer for Cu.
-
The mechanism for atom formation in graphite furnace atomizers has been discussed by many workers (1-17). In one popular hypothesis for atomization, the metal oxide is reduced by the carbon of the furnace to the free metal. On the basis of this theory, Campbell et al. ( 1 ) found agreement between these theoretically predicted temperatures and the experiPresent address: Laboratory of Anhui Geological Bureau, Hefei, Anhui, People's Republic of China.
mentally observed temperatures in the reduction process for 27 elements, but there were a few exceptions. Quite recently, the problem of atomization was studied by considering effects of the partial pressure of oxygen (16,17),and the easiness of reduction on graphite for some metals such as copper and silver was explained. Fuller (2-4) has first described a kinetic approach to the atomization process in graphite furnace atomic absorption spectrometry under isothermal conditions. Sturgeon et al. ( 5 ) have studied the mechanism of atom formation in a graphite furnace through a combined thermodynamic and kinetic approach. They derived eq 1 and used it for obtaining E, values for the early stages of signal production of 15 elements. In A , = -E,/RT + A. (1) Two sequential atomization energies have been obtained for 12 elements such as copper with 340 and 187 kJ mol-'. In eq 1 A , is the absorbance a t time t ( s ) and A. is the constant. Using an equation by L'vov (8), Smets (12) derived the expression 1
A,
where /3 is the atomization efficiency and k the first-order rate constant for the formation of analyte atoms a t any temperAture. L'vov used eq 2 for macrokinetic studies (9) and an
0003-2700/84/0356-2714$01.50/00 1984 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 56, NO. 14, DECEMBER 1984
automatic analysis (13) for atomization. Akman et al. (14) derived the expression k = - 2t 72
- t2
(3)
where 7 is the time to transfer the total amount of analyte atoms into the system. They applied this model to nickel (141, arsenic (18), manganese, chromium, and lead (19). Recently, Frech et al. (10) have studied critically three methods utilizing the initial part of the absorbance signal for Ag, Cd, Cu, and P b and concluded that it is difficult to extract useful information from energy values obtained from these methods because of their large standard deviation which was attributed to involving several mechanisms during atom formation. However, our experimental results reveal that the large deviation is due not to complex mechanisms but to the difficulty in obtaining the linear Arrhenius plot with good precision in the narrow initial part of the absorbance signal. Based on the above instance, a different method with a combined thermodynamic and kinetic approach involving the maximum absorbance (Amm)has been developed for the investigation of atom formation processes under increasing temperature conditions. This study utilizes the possibilities of a versatile data acquisition system and attempts to provide all data on the nature of the atomization process from a absorbance-time profile, the data obtained being subsequently used for the calculation of atomization parameters which pertain to all time regions of the analytical signal with good precision. Therefore, an elaborate and detailed comparison of the present method with the hitherto proposed methods is possible.
THEORY Previous workers (2, 5, 8, 12) have found that atom formation in graphite furnace atomic absorption spectrometry may be described by a first-order rate process. Therefore, the atomization reactions of metals (M) that occur on the furnace tube wall may be described by the following models: (1) Reduction of metal oxide by the graphite surface or simply thermal decomposition on the surface
or
tomized analyte species MO and MX on the wall or in the gas phase a t any time t. The first and second terms on the right-hand side express the rate of supply and the rate of removal of the atoms, respectively. On the basis of the assumption of steady-state conditions (d[M](g)/dt = 0) in the initial part of the time absorbance signal when the activity of the condensed phase on the surface of the carbon tube is constant, Sturgeon et al. (5) derived eq 1by using eq 4: the rising slope of the absorbance profile is determined by the rate constant of atom formation. However, as pointed out by van den Broek and de Galan (15),this is true only if the removal term is negligible compared to the supply term or k d is much larger than k (20). If this condition is not fulfilled, a systematic error is made. To take into account the change in the activity of the condensed phase on a supply and removal balance for atomization, we also use the steady-state approximation in a different way. Equation 4 is rearranged to yield
In the atomization process, the measured absorbance A , a t time t is directly proportional to the number of analyte atoms in the vapor, and the maximum absorbance A,, is also proportional to the atom vapor concentration at the maximum absorbance [MI (g)max. We redefine [MI, as a unatomized analyte concentration which corresponds to a remainder of [M](g),,, and [M](g). Hence we have At = a[Ml(g)
(6)
Amax = a[Ml(g)max
(7)
[MI, = [MI k ) m a x - [MI (g)
(8)
where a is the proportionality constant. Substituting eq 6-8 to eq 5 yields (9)
When eq 9 is compared with eq 1and 2, it is readily seen that eq 9 is equivalent with eq 1 and 2 at the very beginning of atomization since the A , is very small and J;A, dt is almost constant. According to the Arrhenius equation, we have In k = -E,/RT
M(g)
(2) . , vaoorization of the oxide and thermal decomposition of oxide-in the gas phase
MOW -MO(g) or MO(s/l)
-
-
+ constant
(10)
where E, is the activation energy of the atom formation reaction and R is the gas constant. A plot of the In k vs. 1/T should yield a straight line with a slope of -EafR.
or MO(s)
2715
M(g)
M(g)
(3) vaporization of the halide and thermal dissociation of halide vapor
- -
MXs) MWg) M(g) where, for simplicity, subscripts in the oxide and halide compounds are omitted. In the atomization process, the net rate of appearance of metal vapor (M(g)) at any temperature is given by
(4) where k is the rate constant of atom formation, kd is the rate constant for the loss of atomic vapor, and [MI, is the una-
EXPERIMENTAL SECTION A Perkin-Elmer Model 5000 atomic absorption spectrometer equipped with a HGA 500 graphite furnace and a deuterium background corrector was used. It is possible to facilitate automatic sampling and programmed determination of up to six elements when combined with a Model AS-40 autosampler. The peak height and area were automaticallyprinted out and displayed by using a Perkin-Elmer Data System 10, the signals can be also stored on a diskette at 1/60-s intervals, and the signals stored on the diskette can be printed out. The furnace was operated in the 20 mL/min internal purge gas mode to give the greatest sensitivity (21). Hamamatsu TV hollow cathode lamps were used as the light source. The spectral lines and slit width were 324.7 and 0.7 nm (low) for Cu, 328.1 and 0.7 nm for Ag, 248.3 and 0.7 nm for Fe, 213.9 and 0.7 nm for Zn, 285.2 and 0.7 nm for Mg, 309.3 and 0.7 nm for Al, and 217.6 and 0.2 nm for Sb. The standard program for determination of these elements in aqueous solutions applying maximum power heating program is listed in Table I. A CHINO recording pyrometer was used to calibrate temperature settings. A stock solution, 1000 mg/L, for each element was prepared from chloride for Cu(II), Fe(III), Mg(II), and Al(III), from nitrate for Ag(1) and Zn(II), and from potassium antimony tartrate for
2716
ANALYTICAL CHEMISTRY, VOL. 56, NO. 14, DECEMBER 1984
Table I. Instrumental Parameters
step parameter temp, "C ramp time, s hold time, s
1,
2,
3,
ashing
atomiz.
conditioning
200 40 5
variable
2700 1 2
300
300
2450 0 5 -2 20
30 5
recorder
internal gas flow, mL/min
4,
drying
1.0
300 0.6
c"0
I
0.2
0.0
0.1
0.2
0.3
Time (s)
Flgure 2. Plot of log ( A -/A,) vs. time for seven elements. Condaions are the same as in Figure 1.
0
03
0.6
0.9
time ( s e c o n d ) Figure 1. Absorbance profiles for seven elements reproduced from data stored in the diskette. Pyrolytic graphite tube was used for all the elements except Sb. Amounts taken, ng: Cu, 0.7; Ag, 0.24; Fe, 0.6; Zn, 0.06; AI, 1.8; Mg, 0.032; Sb, 5. Temperature profile Is drawn in for 2450 OC. The ashing temperature was 550 O C .
Sb(II1). An aliquot of this solution was diluted with water to a suitable concentration before use. Pyrolytic graphite (PG) and nonpyrolytic graphite (NPG) tubes were used during the experimental work.
RESULTS AND DISCUSSION Typical absorbance-time profiles for seven elements were illustrated together with the temperature profile of the furnace in Figure 1in which the absorbance data stored on the diskette were plotted. For all the cases the atomization was completed before a steady-state temperature was reached. When calculating k using eq 9, we must consider the temperature dependence of kd. The loss of atom vapor from graphite atomizers is the complex result of element-specific removal processes such as element recombination in the gas phase and three different transport processes: diffusion, expansion, and convection (15). Of these factors diffusion is the dominant removal process (8, 15,23). L'vov (8) has reported the diffusion coefficient D of zinc atoms in argon over the temperature range 1100-2600 K; this temperature dependence of D for zinc contributes to the activation energy E, of atomization with 18 kJ mol-l under L'vov's conditions when using eq 10. However, this value is found to be less than 6% of E , values for most cases. Thus, temperature dependence of dif€usion loss is small. Furthermore, results obtained by an analysis of the decay portion of the signal in Figure 1 support the view that the temperature dependence of kd is negligible. Assuming that atom formation is negligibly small in the decay portion of the signal (23, 24), eq 4 is reduced to dtMl(g)/dt = -b[Ml(g) (11)
Since absorbance is proportional to the number of atom, we get the relation In A o / A , = kdt
+ constant
(12)
where A0 is arbitrary absorbance as a reference. The slope of the In Ao/A, vs. time plot should give an approximate value of k& Figure 2 shows the eq 12 plot for seven elements in which the maximum absorbance A,,, was selected as a reference absorbance. It is seen that, although the diffusion of atomic vapor occurs under increasing temperature conditions except for A1 (Figure l),a good linear graph was obtained for all the cases after a curve concave upward in a short initial period, thus showing the negligible temperature dependence of kd. The initial concave curve observed, more or less, for all the elements clearly indicates the continuing atom formation: Atomization is not complete at the maximum absorbance. The kd values calculated by using the linear portion with eq 12 are independent of the mass of atom or not consistent with diffusional loss mechanism. This situation has often been observed (24). Other factors such as convection and recondensation of gaseous atoms are considered. We also found that the shape of the removal function depends strongly upon the physicochemical properties of the graphite furnace, especially, for copper, iron, and aluminum. The absolute magnitude of k is very sensitive to the uncertainty of temperature. The minimum profile shift, '/SO s, between hydrochloric and nitric acid media was observed for zinc (Figure 3a). The ratio of the atomization rate constant in hydrochloric acid to that in nitric acid kHCl/k"03 is 3.2 for zinc (Figure 3b), 0.25 for aluminum, and 1.3 for copper under the same atomization conditions with an ashing temperature of 700 "C although both of the media provide nearly the same E, values for all the cases as will seen in Tables IV-VI. Thus, it is difficult to extract useful information concerning atomization from the absolute magnitude of k . In this study to get information, especially concerning the rate-determining step for atomization, our attention is placed on the slope of the Arrhenius plot for supply function and so the dissipation
ANALYTICAL CHEMISTRY, VOL. 56, NO. 14, DECEMBER 1984
1
b
a
2717
1.2
0.4
- 0.4 0.4
Y
m 4
1
-1.2
s" L
0
-2.c 0.1
8.2
7.6
0.3
-0.4
Y
9.0
lo4 l I T , K-l
tirne(s)
Figure 3. (a) Observed absorbance profile shift and (b) the corresponding changes in the atomization rate constant for zinc In (1) HCI and (2) HNO, with an ashlng temperature of 700 "C and the N f f i tube.
s" -1.2
4.8
5.2
5.6
lo4 I / T , tcl Flgure 5. Anhnius plots for iron (0.6 ng) in a HCI solution using various methods. Symbols and ashing temperature are the same as in Figure 4.
6.0
6.8
7.6
lo4 I / T , K - ~ Flgure 4. Arrhenlus plots for copper (0.7 ng) In a HCI solution using various methods. The ashlng temperature was 700 "C. (A) The proposed method (log k ) , (B) Sturgeon et al. method (log A), (C) Smets method (log k), and (D) Akman method (In k).
constant will not be considered in the following calculation of k with eq 9. The rate constant of atom formation at each temperature in Figure 1 was calculated by using eq 9. In Figures 4-6 Arrhenius plots for Cu, Fe, and A1 using eq 9 were compared with those of Sturgeon et al., Smets, and Akman et al. models. A good linear relationship is obtained for eq 9 with an error of slope and a coefficient of correlation obtained from the least-squares method: 1.3% and -0.999 for Cu, 3.0% and -0.996 for Fe, and 1.9% and -0.999 for Al. The time taken
3.8
4.0
lo4 I/T,
4.2
K'
Figure 6. Arrhenius plots for aluminum (1.8 ng) in a HCI solution using various methods. Symbols and ashing temperature are the same as in Figure 4.
to reach 2300 "C is 0.6 s for ashing temperature 1100 O C , 0.7 s for 700 O C , and 0.8 s for 400 "C. The results of this work and Slavin et al. (25) indicated that the temperature profile is very repeatable with a run, but it was difficult to get pre-
2718
ANALYTICAL CHEMISTRY, VOL. 56, NO. 14, DECEMBER 1984
E,, kJ mol-'
taken amt, ng
absorbance
HC1
HN03
cu
0.4 0.7 1.0 1.3
0.573 0.980 1.358 1.570
215 224 235 250
205 218 229 245
Ag
0.12 0.18 0.24 0.36
0.497 0.772 0.960 1.457
257 252 250 252
element
Temp. ( " C )
Table 11. Concentration Dependences of Activation Energy E , for Copper and Silver"
300
-/
" Ashing temDerature. 700 "C. ~~~~~
cisely the same profile each time. Calculations using the different temperature profiles showed that an error of *O.l s for the time taken to reach 2300 "C gives uncertainties of E, values less than 5%. Figure 7 shows relationships of E, values obtained from Arrhenius plots of Cu (Figure 4) to the atomizing time for the four methods. Each point in the Sturgeon et al. and Smets plots was estimated by using the four adjacent readings of absorbance or time and that in the Akman plot by using three adjacent readings. The present model gives a constant value of E, over a long period of atomization or wide absorbance range irrespective of the surface nature (PG or NPG) of the tubes used (ca. 75% plots for Cu and ca. 95% for Ag, Fe, Zn, Mg, Sb, and Al) whereas the Sturgeon et al. and Smets models give decreasing values with time. This difference between the models holds for all the analytes studied. The constant E, value obtained from eq 9 is in good agreement with only a value of the initial point for the Sturgeon et al. and Smets models. No significant value is obtained from the Akman et al. model. In view of the assumptions made, of course, the Sturgeon et al. model is effective only in the initial period of atomization, and they pointed out that the bending a t the higher absorbance values toward the higher temperature portion of the plot may result from the deviation of the atomic vapor temperature from the graphite surface temperature, the variation of the activity of the analyte on the furnace wall, and a temperature dependence of E,. However, the present results (Figures 4-6) seem to suggest that the major factor for the bending is dissipation of atomic vapor in the Sturgeon et al. model and is an underestimation of the removal effect in the Smets model in which the peak area absorbance is used for estimation of it (12, 13). The k values a t the beginning point of bending for the Sturgeon et al. model were compared with the kd value ob-
0 0.1
0.2
03
time ( second
Figure 7. Time dependences of E, values for copper obtained from the plots in Figure 4. Symbols are the same as in Figure 4.
tained by using eq 12. A linear relationship for the Sturgeon et al. model holds true for k values less than 2% of kd. Hence the influence of dissipation on the analysis of the atomization mechanism is possible from this model. Dependences of E, values on the initial amount of analyte for copper and silver are given in Table 11. Equations 9 and 10 assume that the E, value is independent of the initial amount of analytes, but despite the same absorbance region the E, value for copper increases with increasing the amount of analyte while that for silver is independent. This amount dependence for copper results from the fact that absorbances for an amount of copper of larger than 0.7 ng deviate negatively from the linear calibration graph, showing a complex mechanism of atom formation for copper: two atomization paths for copper compared to a single path for silver. Results for seven elements are summarized in Table 111. The difference in E, values between pyrolytic graphite (PG) and nonpyrolytic graphite (NPG) tubes was observed for copper, iron, and aluminum in which lower E, values were obtained for NPG tubes. Smets (12) also discussed that in contrast to the monolayer assumption, only a small fraction of the atoms are situated at the carbon surface, especially, in the NPG tube. The atomization of atoms that have diffused into the graphite is a slower or more delaying process than the direct atomization of surface atoms. However, it is unlikely that such a diffusion without any interaction with carbon plays an important role because atomization for zinc, silver, and
Table 111. Activation Energies of Atom Formation of the Elements
E,, kJ mol-' taken amt ng
element/ form Cu/HC1 Cu/HNOs Ag/HCl Ag/"O3 Fe/HCl Fe/HN03 Zn/HCl Zn/HN03 Al/HCl Al/HNOS Mg/HCl Mg/"O3 Sb/HCl Sb/HNOa
possible atomization state species kJ Ea mol-' ref
appearance temp, K
PG tube
NPG tube
0.7
1280
218
164
Cudg)
202
22
0.24
1120
269
281
Ads)
285
22
0.6
1700
518
276
FeO(s)
569
12
0.06
1100
326
324
ZnOW
289
22
1.8
2310
1730
620
AlO(s/l)
0.032
1760
688
674
MgO(s)
661
5
5 5
1470 1470
372 301
SbCW Sbz(g)
362 299
22 22
'
ANALYTICAL CHEMISTRY, VOL. 56, NO. 14, DECEMBER 1984
Table IV. Atom Formation of Copper under Various Conditions tube PG
NPG
ashing temp, O C
HC1
550 700 700 900 1100 700
227 218 224 282 366 166
E,, kJ mol-' HN03 av
Table V. Atom Formation of Silver and Iron under Various Conditions
appearance temp, K element
218 205 218 262 354 161
223 212' 221 272 360 164b
1280 1280 1280 1310 1450 1420
Ag
tube
ashing temp, O C
PG NPG
Fe
PG
"0.4 ng of Cu. b l . O ng of Cu. All the rest used 0.7 ng of Cu.
magnesium are independent of the texture of the tube. There is a general trend that a porous nature of the NPG tube leads to a broadened absorbance profile which results in lower E, and k d values for those elements which interact easily with graphite to form carbides or interstitial compounds (26). In this case a different rate-determining step for atomization occurs.
ATOMIZATION MECHANISMS Copper. Copper has been well studied by many workers: Fuller (2) obtained 138 kJ mol-' for the activation energy, Sturgeon et al. (5) 322 and 184 k J mol-', Smets (12) 310 kJ mol-', L'vov (9) 313 kJ mol-l, Suzuki et al. (11)347 kJ mol-', and Frech (10) 167 kJ mol-'. Table IV gives effects of ashing temperature and varying acid on E, values. When the ashing temperature is lower than 700 "C, E, values are independent of acids and the average value, 218 f 8 k J mol-', with an appearance temperature of 1280 K corresponds to the Cu-Cu bond energy (201.7 kJ mol-*) of the dimer. However, E, values increase with increasing ashing temperature above 900 "C and go up to 359 k J mol-' with a raised appearance temperature of 1450 K, corresponding to the heat of atomization of Cu(s) (337.6 kJ mol-') when ashing temperature is 1100 OC. This indicates that the two atomizing processes are present. Sturgeon et al. have also reported the two sequential atomization energies. They analyzed a Arrhenius plot which consists of two intersecting straight-line segments of different slopes: a larger slope a t a lower temperature rahge and a smaller slope a t a higher temperature range. It seems, however, unreasonable that of two parallel atomization paths the one of a lower E, value occurs at higher temperature. The segment of a lower slope may result from neglecting the bending of Arrhenius plots a t a higher temperature region due to the loss of atoms. In this study such plots were not obtained, but when ashing temperature is raised E , increases and Tappraises, showing the two processes in a different way. lower ashing temp: CuO(s)
- - Cub)
higher ashing temp: Cu(s)
Cu,(g)
338
202
Cu(g)
Cu(g)
When ashing is carried out at a lower temperature, atomization begins at a state of CuO(s) at 1280 K and the dimer is formed, the rate-determining step being the splitting of the dimer bond with an energy of 202 k J mol-'. This value is close to that of Frech (IO). When ashing is carried out at a higher temperature, atomization begins at 1450 K, and over this temperature the formation of the dimer is difficult. Copper oxide in the solid state is reduced to Cu(s) on the graphite surface and atomization begins a t Cu(s). The recent studies concerning the effect of the partial pressure of oxygen on the oxide decomposition in the graphite furnace (16,17) showed that copper and silver are reduced to the metal on graphite before the atomization stage.
2719
NPG
amearance -_ E,, kJ mol-' temp, K HC1 HN03 HC1 HN03
550 700 550 700
258 250 280 289
255 267 254 300
1100 1130 1060 1160
1060 1130 1060 1130
400 550 700 900 1100 550 700
536 555 519 484 522 272 269
522 511 505 505
1700 1740 1710 1710 1670 1720 1740
1660 1720 1730 1680
289
1760
Table VI. Atom Formation of Zinc, Magnesium, and Aluminum under Various Conditions
__
E,, kJ mol-' HC1 HN03
amearance temp, K HC1 HNOB
element
tube
temp, O C
Zn
PG NPG
550 400 550 700
309 343 339 332
326 324 316 322
1050 1110 1100 1130
1100 1070 1130 1160
Mg
PG
550 700 900 1100 550
68ga 651 697 670 674
697b 711 699 705 661
1800" 1810 1750 1710 1720
1760b 1770 1740 1710 1680
2350 2340 2340 2340 2320 2270 2260 2300 2310
2310 2340 2300 2310 2290 2270 2300 2260
NPG A1
PG
NPG
300 400 550 700 900 1100 400 550 700
1800 1750 1669 1734 1734 1750 670 590 680
1734 1749 1693 1678 1750 1750 590 590
'Column is 0.032 ng of Mg. Column is 0.052 ng of Ma.
Silver and Iron. Activation energies for Ag and Fe under various conditions are given in Table V which are independent of acids. The average, 269 f 18 kJ mol-', for Ag assumes the rate-determining step
For the iron atomization, the average value 518 f 20 k J mol-' was obtained using the PG tube. We have the reaction heats in kJ mol-': 410 for FeO(g) Fe(g) (22),498 for FeO(s) Fe(s) (12),416 for Fe(s) Fe(g) (22),and 569 for FeO(s) C Fe(g) + CO (12). The obtained value correlates well to the carbon-reduction reaction with 569 kJ mol-l. Therefore -+
+
-
-
-+
The reaction which corresponds to the lower E, value, 276 f 10 kJ mol-l, for the NPG tube is not found in the literature but may be a reaction concerning carbides. Zinc, Magnesium, and Aluminum. E, values of the three elements are tabulated in Table VI. The atomization behavior of Zn is not similar to Cu, Ag, and Fe. The average E,, 326 f 30 kJ mol-', is in close agreement with the dissociation energy of the Zn-0 bond. ZnO(s)
-
ZnO(g)
289
Zn(g)
2720
ANALYTICAL CHEMISTRY, VOL. 56, NO. 14, DECEMBER 1984
Table VII. Atom Formation of Antimony under Various Conditions
ashing temp, "C 550 700 550 700
appearance temp, K
E,,
soln
kJ mol-'
HC1 HC1
360
"03
312 290
1480 1470 1480
384
"0,
1440
The energy of atomization for Mg is 688 f 22 kJ mol-l regardless of the nature of the tubes and ashing temperature. Sturgeon et al. (5) reported 377 kJ mol-' corresponding to the bond dissociation energy of Mg-0. Thermogravimetric analysis (27) indicates that MgO(s) (mp 2852 "C; bp 5600 "C) is a product in the decomposition of both the chloride and nitrate below the appearance temperature (1750 K). On the other hand, it is known that Mg(g) and O2 are the principal gas-phase species in equilibrium with MgO(s) (28). Since the heat of vaporization of MgO(s), 661.1 kJ mol-l, is greater than the dissociation energy of MgO(g), 393.7 kJ mor1, dissociation of MgO(g) will not occur prior to any reaction (29). Therefore, the E, value for Mg should only correspond to the heat of sublimation of MgO(s), as the rate-determining step
- 861.1
MgO(s) M g W Mg(d Atomization energy for A1 is independent of ashing temperature and acids for the PG tube (Table VI), showing a very simple atomization process of a single E , value, although Sturgeon et al. (5) reported two sequential atomization energies (979 and 477 kJ mol-') as in the case of copper. The average E , value was 1733 f 34 kJ mol-' while E , values calculated by using the data in the initial period (Figure 6, lines B and C) with eq 1 and 2 were 1500 and 1420 kJ mol-', respectively. The lower E, value, 620 f 50 kJ mol-', was obtained for the NPG tube. The common product A1203(mp 2072 "C and bp 2980 OC (22))is very stable under these ashing conditions. Since reduction of this oxide is not thermochemically favorable at the appearance temperature (5), Al(g) must be formed by the thermal decomposition of the oxide. for PG:
A&O3(s/l)
1733
Al(d
512
for NPG: Al-Wg) --,AUg) Antimony. The energy of atomization for S b is 372 kJ mol-' for HC1 solution but 301 k J mol-' for H N 0 3 solution (Table VII). These values correlate with the dissociation energy of Sb-Cl(362 kJ mol-') and Sb-Sb (299 kJ mol-l). As a result, S b atoms in the vapor phase are formed by the gas-phase dissociation reaction in HC1 solution Sb-Cl(s)
-+
Sb-Cl(g)
362
Sb(g)
whereas in H N 0 3 solution the reaction is 299
Sb-O(s) Sb(s) Sb2(g) __* Sb(g) I t is concluded that a steady-state approximation was satisfactorily applied to the kinetics of atom formation over -+
the temperature range from the appearance to the maximum in absorbance time signals with a good linear Arrhenius plot by taking into account the dissipation effect on the change in the activity of analyte. The two sequential activation energies were not observed with each run for all the cases.
ACKNOWLEDGMENT I express my gratitude to Yuroku Yamamoto and Etsuro Iwamoto of Hiroshima University for suggesting this investigation and valuable discussion during the course of this study. Registry No. Zn, 7440-66-6;Al, 7429-90-5;Mg, 7439-95-4;Sb, 7440-36-0; Ag, 7440-22-4; Fe, 7439-89-6; Cu, 7440-50-8; CuO, 1317-38-0;FeO, 1345-25-1;ZnO, 1314-13-2;MgO, 1309-48-4;A1203, 1344-28-1; antimony chloride, 10025-91-9; antimony oxide, 1309-64-4;graphite, 7782-42-5. LITERATURE CITED Campbell, W. C.; Ottaway, J. M. Talanta 1974, 21, 837. Fuller, C. W. Analyst (London) 1974, 99, 739. Fuller, C. W. Analyst (London) 1975, 700, 229. Fuller, C W. Analyst (London) 1978, 707, 798 (5) Sturgeon, R. E.; Chakrabarti, C. L.; Langford, C. H. Anal. Chem. 1970, 48, 1792. (6) Chakrabarti, C. L.; Wan, C. C.; Teskey, R. J.; Chang, S. B. Spectrochlm. Acta, Part 8 1981, 368, 427. (7) Gregoire, D. C.; Chakrabarti, C. L. Spectrochim. Acta, Part 8 1982, 378, 611. (8) L'vov, B. V. "Atomic Absorption Spectrochemical Analysis"; Adam Hiiger: London, 1970. (9) L'vov, B. V.; Bayunov, P. A,; Ryabchuk, G. N. Spectrochim. Acta, Part 8 1981, 368,397. (10) Frech, W.; Zhou, N. G.; Lundberg, E. Spectrochim. Acta, Part B 1982, 37B,691. (11) Suzuki, M.; Yamakita, K.; Katsuno, T. Spectrochim. Acta, Part 8 1981, 368,679. (12) Smets, B. Spectrochim. Acta, Part 8 1980, 358,33. (13) Bayunov, P. A,; Savin, A. S.; L'vov, B. V. A t . Spectrosc. 1982, 3 , 161. (14) Akman, S.;Genc, 0.; Ozdural, A. R.; Balkis, T. Spectrochim. Acta, Part B 1980, 358,373. (15) van den Broek, W. M. G. T.; de Galan, L. Anal. Chem. 1977, 49, 2176. (16) L'vov, B. V.; Ryabchuk, G. N. Spectrochim. Acta, Part 8 1982, 375, (1) (2) (3) (4)
-. -.
673
(17) Sturgeon, R. E.; Siu, K. W. M.; Berman, S. S. Spectrochim. Acta, Part 8 1984, 398,213. (18) Akman, S.;Genc, 0.;Balkis, T. Spectrochim. Acta, Part B 1982, 378 -. - , 903 - -(19) Genc, 0.; Akman, S.; Ozdural, A. R.; Ates, S.; Balkis, T. Spectrochim. Acta, Part 5 1981, 368, 163. (20) Frost, A. A.; Pearson, R. G. "Kinetics and Mechanism"; Wiley: New York, 1961. (21) Chung, C. H.; Iwamoto, E.; Yamamoto, M.; Yamamoto, Y. Spectrochlm. Acta, Part 8 1984, 398,459. (22) Weast, R. C., Ed. "Handbook of Chemistry and Physics", 62nd ed.; CRC Press: Boca Raton, FL, 1981. (23) Sturgeon, R. E.; Chakrabarti, C. L. Anal. Chem. 1977, 49, 1100. (24) Akman, S.;Genc, 0.; Balkis, T. Spectrochim. Acta, Part 8 1981, 368, 1121. (25) Slavin, W.; Carnrick, G. R.; Manning, D. C. Anal. Chem. 1982, 54, 621. (26) Hennig, G. R. Prog. Inorg. Chem. 1959, 1 , 125. (27) Duval, C. "Inorganic Thermogravimetric Analysis" 2nd ed.; Elsevier: New York, 1963. (28) Aitman, R. L. J . Phys. Chem. 1983, 67,366. (29) Fuller, C. W. "Electrothermal Atomization for Atomic Absorption Spectrometry"; The Chemical Society: London, 1977.
+
RECEIVEDfor review February 22, 1984. Resubmitted and accepted August 10, 1984.
