Anal. Chem. lQQ2,6 4 , 1144-1 153
1144
factant and labeled analytes, the behavior of the system appears to parallel that of the solution-initiated reaction.
ACKNOWLEDGMENT This research was supported by grants from Merck Sharp & Dohme Research Laboratories and from the Biotechnology Research and Development Corporation. We also thank Bioanalytical Systems for the loan of a potentiostat used in the early phase of this work. Registry No. Cetylammonium bromide, 68810-16-2; phenyl lO-methylacridinium-9-carboxYlate, 123632-55-3; 4-(2-(succinimidyloxycarbony1)ethyl)phenyl 10-methylacridinium-9carboxylate, 87198-88-7. R li!li'RR li'.NP.li!!S L l Y I YIlYI. V Y U
Weeks. I.; Beheshtl, I.; McCapra. F.; Campbell, A. K.; Woodhead, J. S . Clh. Chem. 1983. 29. 1474-1479. Weeks, 1.; Sturgess, M.;Brown. R. C.; Woodhead. J. S. Methods EnZymOl. 1988, 133, 366-387. McCapra, F.; Beheshtl, I . Bldomnescence end Chemllumlnescence: Instnrments and Appllcetions; CRC Press: Boca Raton, FL, 1985, pp 1, 9-42.
(4) McCapra, F. Acc. Chem. Res. 1878, 9 , 201-208. (5) Llu, H.; Yu. J. C.; Blndra, D. S.; Wens. R. S.; Wilson. G. S. Anal. Chem. 1981, 63, 666-669. (6) Downey. T. M.; Nleman, T. A. Anal. Chem. 1802, 64. 261-268. (7) VanDyke, D. A., Ph.D. Thesls, University of Illlnois at Urbana-Champalgn. 1986. (8) Nieman, T. A. Mkfochlm Act8 W88, 111, 239-247. (9) Haapakka, K. E.; Kankare, J. J. Anal. Chlm. Acta 1981. 130. 415-4 18. (10) Murphy, R. J.; Svehla, G. Anal. Chlm. Acta 1081, 125, 73-83. (11) Taylor, R. J.; Humffray, A. A. J . Electraenal. (2".Interfaclel Elechochem. 1075, 64, 63-84. (12) Taylor, R. J.; Humffray, A. A. J . Elechoanal. Chem. InterfeclelElech o d " . 1875, 64, 95-105. 63, 586-595. (13) Hage, D. s.; Kao, p. C. Anal. Chem,, (14) Bagazgohia, F. J.; Garcia, J. L.; Diequez. C.; Weeks, I.; Woodhead, J. S . J . Biolumln. Chemllumln. 1988, 2 , 121-128. (15) Taylor, D. W.; Nleman, T. A. Anal. Chem. 1084. 56, 593-595. (16) Skooa. D. A.: West. D. M.: Holler. F. J. Fundementals of AnaMicel CheGshy, 5th ed.; Saunders: New Yolk, 1988 pp 25-26. (17) Box, G. E. P.; Hunter, W. 0.;Hunter, J. S. Statistics for Ekperhentters,
....-, . ..-.. . _... .- .
. - -. .- - .-.
Wilav. Naw Vnrk , 197R. -, nn rr 76 and 919
RECEIVED
for review September 307 lggl*
6, 1992.
Vaporization and Atomization of Lead and Tin from a Pyrolytic Graphite Probe in Graphite Furnace Atomic Absorption Spectrometry Glen F. R. Gilchrist, Chuni L. Chakrabarti,*Jeff T. F. Ashley, and Dianne M. Hughes Centre for Analvtical and Environmental Chemistrv. DeDartment of Chemistry, Carleton University, Ottaua; Canadd KlS 5B6
A novel technique has been used to lnvestlgate the proceeses of vaporlzatlon and atomlzatlon of lead and tln In graphtte probe furnace atomic absorption rpectrometry. Udng this technique lt ha8 been found that, for lead and tln, vaporlzatbn and atomlzatlon are separate processes and lead and tin are vaporized a8 molecular specks. Furthermore, rate constants and acthratlon energies for the vaporization of the molecular specks of lead and tln have been detennlned. The acthratlon energies for vaporization have been found to be 86 f 4 kJ mol-' for the lead molecular specles and 120 f 6 kJ mol-' for the tin molecular specles. It Is wggested that these acthratlon energks represent the energy for desorption of PbO and SnO from the graphlte surface. Experhrental resulte have been compared wlth those predicted by an earlier model based on homogeneow gas-phase thermodynamic equlllbrlum. The etfects d the gas-phase chemlcal modifiers, H, and 0,, on the atomlc absorption a n a l proflk d lead and th are presented and dlscu88ed. Atomlzatlon meChanl8m8 have been proposed for lead and tin and compared wlth tho8e proposed by other Workers. PbO and SnO have been proposed as the mort Ilkely gaseous molecules of lead and tin, respectively, formed by vaporlzatlon.
INTRODUCTION Several have studied mechanisms of formation and loss of analyte atoms in graphite furnace AAS. Fuller2% *To whom all correspondence should be addressed.
has introduced the kinetic approach to predict the shape of copper atomic absorption signal profiles. The kinetic approach2s3is based on the supply and removal of atoms by consecutive first-order reactions. Thus, the rate of change in the number of gas-phase atoms is equal to the difference between the supply function and the removal function. Sturgeon et aL4have proposed a combined thermodynamic and kinetic model and have used the leading edge of the atomic absorbance signal profie to determine the activation energy of the rate-determining step for atom formation. To elucidate reaction mechanisms, Sturgeon et al.' have compared experimental activation energies with the standard enthalpy of some appropriate reactions. On the basis of the standard enthalpy of reaction, they4 have concluded that lead and tin are atomized direct from the surface of the graphite atomizer after PbO(1) and SnO(s) have been reduced by carbon at high temperatures to Pb(1) and Sn(s). L'vov et aL5 have used a kinetic approach and have obtained results similar to those of Sturgeon et al.4 for atomization of lead and tin. Smeta6has applied a kinetic approach which has allowed use of the whole atomic absorbance signal profie to determine the activation energy of atom formation process. Sturgeon and Arlow' have used a kinetic approach, similar to that of Smetd to investigate vacuum atomization and have concluded that PbO(g) is reduced on the wall of the graphite atomizer to yield Pb(g) and CO(g). Using the kinetic approach, Akman and co-workers8 have proposed that the condensed-phase lead monoxide is vaporized directly from the graphite surface and is then dissociated in the gas phase. The kinetic approach has proved useful in predicting the behavior of the atomic absorption signal profdes of many elements under a variety of conditions.
0003-2700/92/0364-1144$03.00/00 1992 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 64, NO. 10, MAY 15, 1992
Using the kinetic approach, some authors have reached a consensus on the atomization mechanism for several elements. However, methods used up to now1-*have three limitations: (i) measurement of the rate of atomization is done by an indirect method involving convolution of the supply and the removal function to yield the atomic absorption signal profile; (ii) the rate of vaporization and the partial pressure of the molecular species of the analyte cannot be directly measured; (iii) if vaporization and atomization are separate processes they become lumped together and cannot be separated and studied as individual processes.
VAPORIZATION AND ATOMIZATION Aasuming the absence of crystallites, condensed-phase analytes may be vaporized direct from the graphite surface as atoms or as molecules. The molecules may dissociate in the gas phase or dissociate concurrently upon vaporization, or they may be reduced by carbon at high temperatures to form gaseous analyte atoms:
where M is an analyte atom, MX is a molecule (X may be any atom including an oxygen atom in eq 31, MO is an oxide molecule, k14 are rate constants, and K is the equilibrium constant. The subscript c is used to denote a species that exists in a solid, liquid, chemisorbed, or physisorbed state. The condensed-phase metal atoms in eq 1are reduced from a higher oxidation state (as a molecule or a compound) prior to vaporization. If the analyte is present in amounts equal to nanograms or less, the condensed-phase analyte species might be distributed on the graphite surface in monolayers or submonolayers. In the case of monolayer or submonolayer distribution of analyte atoms or molecules, their release from the graphite surface occurs by desorption9 at a rate given by Holcombe et al.l0 as
--auM - - a&v exp(-E,/RT) at where UM is the surface coverage by M at any time, n is the order of desorption, v is the frequency factor, E, is the activation energy of the desorption process, R is the gas constant, and T is absolute temperature. If the analyte atoms are distributed on the graphite surface as a monolayer or submonolayer, then the order of desorption can be l6and the surface coverage term may be replaced with the number of analyte atoms present on the graphite surface, [M(c)]. One can then write the rate of desorption as
where k is a fmt-order rate constant for desorption at constant temperature. Rearranging the terms of eq 6 and integrating it, one gets
-h
[E] = kt
(7)
where [M(c)], and [M(c)], are the number of atoms on the surface of the graphite probe at time t and zero, respectively. The rate constant, k,is related to the activation energy of the rate-determining step by the Arrhenius equation:
1145
k = v exp(-E,/R!!')
(8) where the various terms have been defined earlier. For a first-order desorption, a plot of -In ([M(C)],/[M(C)]~) vs t should be a straight line with slope equal to the rate constant, k. From eq 8, a plot of In k vs 1/T should be a straight line with slope -EJR and intercept In v. If an analyte is dispersed on the graphite probe surface as a monolayer or submonolayer then the activation energy for release from the graphite probe surface will not be the heat of vaporization; it will be the energy required to remove the analyte atom from the graphite probe surface.1° In such a case, any agreement between this activation energy and the heat of vaporization of the analyte would be merely coincidental. Equation 2 represents a case where vaporization and atomization are separate processes. In the mechanism represented by eq 2, molecular species are vaporized direct from the graphite surface and are then dissociated. In the case where the analyte is desorbed in a molecular form, MX(c) or MO(c) can be substituted for M(c) in eqs 5-7 so that eq 7 may be rewritten as
where molecules, instead of atoms, are vaporized. Equations 3 and 4 represent the cases where condensedphase molecular species are thermally decomposed, or are reduced by carbon and vaporized in a single step, and therefore represent thermal decomposition and carbon reduction, respectively, of the analyte oxide. A commercial, pulse-heated graphite furnace (which is almost universally used for quantitative analysis) presents a formidable experimental problem to the study of kinetics in graphite furnace AAS. The temperature of a pulse-heated graphite furnace increases throughout most of the duration of the atomic absorption signal, and atomization under increasing temperature is nonisothermal. This problem of nonisothermal atomization is circumvented in this study by using a commercial graphite furnace that has been converted into a graphite probe furnace to provide near-isothermal atomization. In the graphite probe furnace, a graphite probe containing dried residues of aqueous solutions of analytes is inserted at preset times after the graphite furnace has attained a preselected, steady-state temperature. This technique will be henceforth called the graphite probe technique. The objective of this study is to differentiate, experiment i d y , the processes of vaporization and atomization, to make direct measurement of the rate of vaporization of analyte atoms and molecules from the surface of the graphite probe, to investigate formation and dissociation of molecules in the gas phase, and to study analyte atom formation and loss in the graphite probe furnace AAS.
GRAPHITE PROBE FURNACE Isothermal atomization in GFAAS has been the objective of several innovative furnace designs including the capacitive discharge furnace,l' the platform furnace,12and the probe furnace.13-16 Isothermal atomization provides experimental conditions for minimizing matrix interferences in GFAAS. In addition to the above advantage, isothermal atomization allows considerable simplification in the study of kinetics in GFAAS since the rate constant is a sensitive function of temperature and remains constant only at constant temperature. At the beginning of the atomization cycle the probe is positioned outside of the furnace and the temperature of the graphite probe is approximately 300 K. The probe is inserted into the graphite furnace only after the graphite furnace has attained the preset, constant temperature.l3-lBUpon insertion of the probe into the furnace the temperature of the probe
1146
ANALYTICAL CHEMISTRY, VOL. 64, NO. 10, MAY 15, 1992
increases rapidly to a temperature just below that of the furnace waIl.l4 The probe therefore is not an isothermal atomization devise. However, if the time over which the probe is heated is short compared to the time for atomization then the atomization process may be treated as nearly isothermal. Giri et al.13 made optical-pyrometric measurements of a pyrolytic graphite probe inserted into a graphite furnace at the steady-state temperature of 2573 K. They13found that the portion of the probe inside the furnace was heated to 2473 K within the first 0.5 s-a heating rate of 4400 K ms-l. As explained later on, optical-pyrometric measurement of the temperature of a heated graphite probe located inside a heated graphite tube may give results of questionable validity. However, the works of Chakrabarti et al.14 and Wu et al.15 seem to confirm the work of Giri et al.13 that a thin graphite probe head (-0.35 mm thick), as used in this work, is very rapidly heated to a temperature approaching the heated tube wall temperature. Chakrabarti et al.14 and Wu et d.16have used the two-line atomic absorption method to determine the temporal distribution of the gas-phase temperature in the graphite probe furnace. Their works14J5have provided an indirect measure of the heating rate of the graphite probe inside the graphite furnace. When the probe is inserted in the furnace there is a drop in the temperature of the gas phase, followed by a relaxation to a steady-state gas temperature as shown by the two-line method. This perturbation and relaxation process can be understood in terms of heat transfer between the have furnace wall, the probe, and the gas phase. Wu et shown that because of the low heat capacity of the gas phase the probe is heated primarily by radiation from the furnace wall and not by conduction from the gas phase. The gas-phase temperature, however, is quenched by insertion of the cold probe; the gas is thereafter heated by conduction from the tube wall and (after the probe has been heated by radiation from the furnace wall) from the probe. The relaxation of the gas phase to a steady-state temperature after the perturbation caused by the insertion of the cold probe into the furnace14 shows that the probe is no longer removing heat from the gas phase and has reached a state of quasi-thermal equilibrium with the gas phase and the furnace wall. Quasi-thermal equilibrium in this case means that the rate of heat stored in the probe equals the rate of heat lost by the probe. The data published by Chakrabarti et al.14show a relaxation time between 0.2 and 0.6 8, depending on the thickneas of the probe. The measured temperature of the gas phase14 is below that of the furnace wall. Thii difference occurs partly because the two-line method determines the mean gas temperature which is affected by the difference in the wall temperature of the graphite tube in ita horizontal axis, whereas the furnace wall temperature is measured at the middle of the horizontal axis, which is the location that attains the highest tem~erature.'~ They14J6have also shown that for a furnace heated to 2780 K, the drop in the furnace wall temperature caused by insertion of the cold probe is between 30 and 70 K, depending on the thickness of the probe.
EXPERIMENTAL SECTION Apparatus. An atomic absorption spectrophotometer,Model 603,fitted with a deuterium-arc background corrector, a Model 2100 programmable power supply and a Model 400 Heated Graphite Atomizer (all from Perkin-Elmer Co., Norwalk, CT, USA), was modified to make it a graphite probe furnace.13J4The above modification included adding an automated probe insertion device which allowed precise timing of the probe insertion into the graphite furnace. The injection port of pyrolytically coated graphite tubes (Perkin-Elmer part No. 091504)was enlarged to make a rectangular slot of height 2 mm, and width 4.5 mm. Pyrolytic graphite probes were fabricated from a block of anisotropic pyrolytic graphite (Airco Speer Carbon Graphite,
Pennsylvania, PA). The pyrolytic graphite probes were cut so that the flat sample head was parallel to the crystallographicAB
plane. This orientation was chosen because the AB plane is relatively inert and the thermal conductivity perpendicular to the AB plane (the C direction) is approximately 300 times less than that in the AB plane." The graphite probes were 25 mm long, 4 mm wide at the head, and 2 mm wide at the base, with thicknesses ranging from 0.39 to 0.42 mm. The probes were mounted on a ceramic holder, which was fixed on the shaft of a two-way, gas-operated cylinder."l6 The probe insertion device consisted of a microprocessor-controlledtimer, which was activated by a trigger pulse from the programmable power supply. Once activated, the time counted down the preset time and then sent an electronic pulse to a relay circuit. The relay circuit converted the timer pulse into a 120-V ac pulse, which opened a solenoid switch to admit high-pressure (45 psi) argon into the above gas-operated cylinder. A detailed description of this device was given in previous publi~ations.'~J~ Reagents and Standards. Stock solutions containing loo0 mg/L of each anal@ were separately prepared from pure metals or compounds as follows. For lead, Pb(N0J2 was dissolved in 2% (v/v) nitric acid (ULTREX) and the solution diluted with ultrapure water. For tin, the metal was dissolved in 50% (v/v) hydrochloric acid (ULTREX) and the solution diluted with ultrapure water. All stock solutions were made to contain 1% (v/v) nitric acid for lead and 5% (v/v) hydrochloric acid for tin. The ULTREX brand acids were manufactured by Baker Chemical Co. Ultrapure water of resistivity 18.3 MQcm was obtained direct from a Milli-Q2 water purification system (Millipore Corp.). All test solutions were prepared daily by serial dilution of the above stock solutions with ultrapure water. Gases and Gas Flow Control. High-purity argon gas (99.995% pure, Matheson Gas Products, Canada), to be called the Ar purge gas, was used as both purge gas and sheath gas for the graphite probe furnace. Two other gas mixtures were occasionally used as purge gas and sheath gas. One gas mixture contained 5.00% (v/v) hydrogen in argon (to be called the H2/Ar purge gas), the other contained 2.00% (v/v) oxygen in argon (to be called the 02/Ar purge gas)-both were prepared and supplied by Matheson Gas Products, Canada. The flow rate of the sheath gas and purge gas was 1.5 L min-' and 325 mL m i d , respectively. Under the interrupt mode of purge gas flow which was used throughout this work, the purge gas flow was interrupted during the atomization cycle but was restored immediately on the termination of the atomization cycle in order to provide an Ar gas atmosphere in which the probe was allowed to cool to room temperature inside the graphite probe furnace before it was manually withdrawn from the graphite probe furnace. Experimental and Operating Conditions. Signals were recorded with a two-channel, programmabledigital oscilloscope, Model 4094,fitted with an XF-44 disk recorder (Nicolet Instrument Corp., Madison, WI). Integrated absorbance was measured with the above-mentioned programmable oscilloscope using a software package supplied by Nicolet Instrument Corp. Commercial hollow cathode lamps were operated at 5mA lamp m e n t The analysis wavelengths were 283.3 nm for lead and 224.6 nm for tin. A spectral band-pass of 0.7 nm was used. Test solutions (1pL) were deposited on the sampling head of the graphite probe using a Hamilton microsyringe, fitted with a Chaney adapter and then dried using an infrared lamp. The analyte mass used for atomization was 1 ng of lead and 100 ng of tin. The temperature of the inside surface of the graphite tube wall was measured with an automatic optical pyrometer, Ircon series 300 (Ircon Inc., Skokie, IL). The pyrometer was focused on the inside surface of the graphite tube wall just below the injection port; the pyrometer signal was recorded with the above-mentioned Nicolet osilloscope and the temperature was read off the temperature calibration curve supplied by the manufacturer. The heating rate of the graphite tube was 450 K s-l. In all cases, the graphite furnace had attained a steady-state temperature before the probe was inserted into the furnace. Test solutions were deposited on the head of the probe and dried as described earlier. The oscilloscope was triggered at the start of the atomization step. After the furnace had attained the steady-state temperature, the probe was automatically inserted into the heated graphite furnace with the automatic probe in-
ANALYTICAL CHEMISTRY, VOL. 64, NO. IO, MAY 15, 1992
Pb
1147
T
Pb 2300
1800 1300 800
6( 0.00
'
400
300 0 1000
1600
2200
Temperature (K) Flgure 1. Charring and atomlration curves for lead and tin for atomization from the graphite tube wall. The charring curve (left-hand-side cuve)shows the peakaea absorbanceat tube waH temperatwe 2200 K as a functlon of charring temperature (40-9 char time). The atomization curve (right-hand-sidecwve) shows the peak-area absorbance at charring temperature 800 K as a functlon of atomization tempera-
ture. sertion device. To ensure attainment of the preselected, steady-state temperature by the graphite furnace, a delay time of 10 s was used prior to insertion of the probe into the graphite furnace. RESULTS AND DISCUSSION Charring Curve and Atomization Curve for Lead and Tin. Figure 1shows the charring and the atomization curves for atomization from the tube wall. The charring curve (left-hand-side curve) was constructed by holding the atomization temperature at a constant, optimum value and varying the charring temperature. The atomization curve (righthand-side curve) was constructed by holding the charring temperature at some constant, optimum value and varying the atomization temperature. With Pb and Sn the loss of these elements begins in the charring cycle at a temperature which is about 300 and 550 K lower than the appearance temperature of lead and tin, respectively. The appearance temperature is defined as the lowest temperature at which the anal@ signal becomes just visible above the baseline noise. The curves in Figure 1can be interpretted as follows. Both the lead and the tin in their respective samples start being lost when the charring temperature exceeds 800 K, whereas the atomic absorption signals for lead and tin are not detectable till the furnace has attained a temperature of 1150 K for lead and 1400 K for tin. This loss of lead and tin must be in the form of molecular species when the temperature exceeds 800 K, and these molecular species do not yield Pb(g) and Sn(g) till the furnace has attained a temperature of 1150 and 1400 K,respectively. Since the lead and tin are not known to form refractory, condensed-phase molecular species at the above temperatures, the most probable lead and tin molecular species that can be formed in the graphite furnace are PbO(g) and SnO(g), respectively, as has been reported in our previous publications1p20and by Frech et al.21 The above formation of PbO(g) and SnO(g) will be treated as an assumption at this time-this assumption must be shown to be reasonable and consistent with further experimental evidence before an atomization mechanism based on this assumption can be accepted as satisfactory. Temperature of the Graphite Probe. The temperature of the graphite probe surface could not be measured by optical pyrometry because an optical pyrometer, the calibration of which is based on blackbody radiation, cannot be used to measure the temperature of the graphite probe without making uncertain corrections, since the probe is located inside
5
10
15
20
Time (s)
2800
Flgure 2. Temperature program for exposurstlme experlments with the graphite probe. T , was the temperature of the graphite tube wail when the probe was Initially Inserted at time t,; t , was the time from the insertknof the probe to the cutoff of power to the gaphb furnace; T, was the temperature of the graphite tube wail when the probe was
reinserted at tlme t,. the graphite tube which is at a much higher temperature and the probe is not in thermal equilibrium with its surroundings during most of the heating cycle. Hence, the temperature of the probe has been estimated by taking into account the temperature of the tube wall and of the gas phase, the width and the mass of the probe, and the time elapsed from the insertion of the (cold) probe into the steady-state temperature of the graphite tube w a l l 5 As can be seen from the Results and Discussion, the estimated temperature is indeed a good one, and gives self-consistent values of the Arrhenius parameters derived from the atomic absorbance signal profiles using the estimated temperature in the Arrhenius equation. Exposure-Time Experiment. The kinetics of lead and tin loss from the furnace was studied by doing exposure-time experiments as described below. The exposure-time experiment was carried out using aqueous solutions of lead and tin. For these experiments, test solutions were deposited on the head of the graphite probe and dried with an infrared lamp. The graphite furnace was then heated to a constant temperature, T,(Figure 2), and the probe was automatically inserted, as described earlier. At the end of the programmed exposure time, t,, the power to the graphite furnace was automatically cut off; the furnace and the probe were then rapidly cooled in a forced, convective flow (325 mL min-') of Ar gas. The probe was then withdrawn from the furnace, the furnace was reheated to a high temperature, T,,the probe was reinserted, and the atomic absorption signal profile was recorded. Figure 2 shows the heating program for the exposure-time experiment. The atomic species giving the atomic absorbance signal is designated [M(c)], to denote that it was the analyte atomic species that remained on the probe at the end of the exposure time. The designation [MX(c)], is used for the molecular species of the analyte element. A reference atomic absorbance signal profiie was obtained by omitting the first, low-temperature exposure; the species giving this atomic absorption signal profile was designated [M(c)], or [MX(c)], to denote that it was the analyte species present on the probe at the start of the atomization cycle-the subscript zero stan& for the exposure time equal to zero. As shown in Figure 3, in the graphite probe furnace, loss of lead from the probe occurs at a tube wall temperature, T,, 11OOO K, but the lead atomic absorption signal is not observed until T, > 1340 K is attained; loas of tin from the probe occurs at T, 1 1200 K,but the tin atomic absorption signal is not observed until T, 1 1800 K is attained. For the initial (low) exposure temperature, T,,no atomic absorption signal is observed (curve A); subsequent atomization at a much higher
1148
ANALYTICAL CHEMISTRY, VOL. 64, NO. 10, MAY 15, 1992 0.30
Table I. Results from Exposure-Time Experiments, for Lead, Where the Loss Function Is Equal to -In ([MX],/[MX],) loss exposure abscieea firat datum exposure point, s rime, a function temp, K intecept, s
Q)
g
0.20
f"m
I
4
2
I
0.10
r
e
' \
I\ A
1120
5.33
6.98
1230
2.75
4.11
1340
1.08
6.98 8.83 12.30 4.11 5.17 5.17
6.06 6.93 1.44 1.44 1.84 2.24 2.24
1.44
0.00
0
10
5
15
20
Time (s) -3.
Lossofleedandthfromthegephiteprobe: (A)slgnelwhen the probe was inserted at T,; (e) slgnai when the probe was reinsetted at T, after Mal exposwe at temperaare Tofor time t,; (C) -1 when the probe was inserted at T,, omitting the lower-temperature exposue step.
2 3
\
r
2.00
1340 K
xv
z
w C
50
1.00
150
1120 K
!
1.05 f 0.10
T,K
k(1000 K)/k(TJ ratio
700 800
13.1 4.5 2.0 1.4 173.3 20.2 3.8 1.9 2280.7 91.0 7.5 2.6 30020.4 409.1 14.5 3.5
900
1.50
0.50
0.56 f 0.06
Table 11. Effect of Temperature and Activation Energy on Rate Constant Compared to the Rate Constant for a Temperature of 1000 K
100 0
0.230 f 0.005
OThe numbera after i represent uncertainties in the reported values.
activation energy, kJ/mol
i
0.39 0.79 1.61 0.71 1.24 1.56 1.90 2.30 0.43 0.34 0.76 1.14 1.31
calcd k value," l's
950 700 800 900 950 700 800 900
T 950 0.00
0
2
4
6
8
1 0 1 2 1 4
Time (s)
=io01 / E 1440 K
1.00
0.00
0
2
4
700 800 900
Flgm 4. Resub of the exposwe time experiments for lead wing the graphlte probe. The abscissa is the exposure time to.
X-
200
6
8
Time (s) Flgure 5. Resuits of the exposure time experiments for tin using the graphite probe. The abscissa was the exposure time t,.
temperature, T,,produces an atomic absorption signal (curve B) that is smaller than what is obtained if the atomization is done without the low-temperature exposure step (curve C ) . Systematic change of the exposure time and the exposure temperature can yield useful information about analyte vaporization. The results of the exposure-time experiments are shown in Figures 4 and 5 as plots of -In ([MX(c)],/[MX(c)lo)vs time for lead and tin at three temperatures. If the vaporization of MX is a fit-order process, then according to eq 9 the rate constant is equal to the slope of the appropriate line in Figure 4 or 5. Exposure-time experiments were also conducted using the H2/Ar purge gas and the 02/Ar purge gas. The H2/Ar
950
purge gas gave straight line plots for the rate constant, k. The 02/Ar purge gas gave results having unacceptably high irreproducibility. The irreproducibility that occurred with the 02/Ar purge gas was probably due to rapid degradation of the graphite probe and the graphite tube because of their oxidation by O2of the purge gas at the temperatures employed. Equations 6-9 are formulations of the first-order rate law. The exposure-time experiments will approximate vaporization under isothermal conditions provided that the time required for the surface of the probe to reach the steady-state temperature, the induction time, is short compared to the time over which vaporization occurs. The existence of the induction time can be seen in Figures 4 and 5, where extrapolation of the fitted linea intercept the abscissa at positive values of time. Figures 4 and 5 and Table I show that the data used in calculation of the rate constant and collected well after insertion of the probe, when the function -ln ([MX(c)],/[MX(c)l0)vs time was linear, were obtained after the surface of the probe head had reached a steady-state temperature. It is a simple matter to calculate the amount of analyte lost during the induction time; Table I shows that,on average, it was less than 30% of the total. With at least 70% of the analyte being vaporized after the induction time had elapsed, the exposure-time experiment approximates isothermal conditions. The effect of temperature on the rate constant, kT,may be calculated if the activation energy is known. Table I1 shows the effect of temperature on kT for activation energies of 50-200 kJ mol-'. From Table 11, one can see that compared to kT for a process at lo00 K, the kT values for temperature below 900 K are very small. The amount of analyte vaporized over any time interval is related directly to the rate constant
ANALYTICAL CHEMISTRY, VOL. 64, NO. 10, MAY 15, 1992
I
I
3*0 2.0
, 1.0
-2.0
t'
Table 111. Experimental Activation Energies for the Vaporization of PbO and SnO in the Graphite Probe Furnace with Ar and H,/Ar Purge Gases and in the Graphite Tube Furnace with A r Purge Gas'
Y
Byrne et al.
1
activation energy for vaporization, kJ mol-' 100% AI purge gas
PbO
9
11
SnO
1/T x104 (K") Figure 6. Arrhenlus plots for the vaporization of lead species using the graphite probe furnace. Purge gases: 100% Ar; 5 % (v/v) H2/Ar. The percent lead kss w e of Byme et al.18for atomkatbn from the tube wall has been plotted using their data.
-0.5
(v/v)
H2/Ar purge gas
T 86 f 4 T-100 73i4 T-200 60i3 T 120 f 6 T-100 102f5 ~-200 a7i5
105 i 15
230 i 30
150 f 20
180 i 20
graphite probe
a For the Ar purge gas, results are given for tube wall temperature (T),tube wall temperature less 100 K (T- 100), and tube wall temperature less 200 K ( T - 200); see text for further details. bThenumbers after f represent uncertainties in the reported val-
Table IV. Thermochemical Data for Relevant Reactions of Lead and Tin Species
, -1.0
std enthalpy
&
W
-1.5
-2.5
5%
graphite tube wall
ues.
t
-2.0
a
graphite probe
oxide
Y
7
3
1149
reaction
1
'
6
\ 7
8
9
Flguro 7. Anhenlus plots for the vaporization of tin species uslng the gepMe probe furnace. FWge gases: 100% Ar; 5 % (v/v) HdAr. The percent bad loss curve of Byme et ai." for atomlratlon from the tube wall has been plotted using thelr data.
------ -
change, kJ mol-'
temp: K
ref
178 2023 28 Pb(1) Pb(d 230 2473 28 Sn(1) S4g) PbO(1) PbO(g) 213 1745 28 251 1800 28 SnO(1) SnO(g) 314 2273 28 SnOdl) Sn02(g) 629 1100 29, 30 PbO(1) Pb(d + O(g) 753 1300 29, 30 SnO(1) Snk) O(g) 29,30 Sn02(l) Sn(d + O h ) 756 1800 PbO(1) + C(s) Pb(g) + CO(g) 263 1100 29, 30 SnO,(l) + 2C(s) Sn(g) + 2CO(g) 507 1800 29,30 Temperatures listed here are those given in the respective ref-
-
O
and the length of the time interval. Thus, a moderate heating rate (a 1 2 K ms-l) combined with a relatively small rate constant will cause only a small amount of the anal@ to be vaporized before the temperature of the probe is within 50 K of its final, steady-state temperature (SOthat the time over which the probe head is heated is short compared to the time for vaporization). Our calculations (not given here) have shown that in Figure 1 80% of the analyte is vaporized if the tube is held at 1050 K for 40 s. The rate of loss at 1050 K is therefore too slow to account for the observed analyta loss at 1120 K during the exposure-time experiments (80% loss after 12 s at 1120 K). This observation suggests that vaporization of analyte during the exposure-time experiment occurred, for the most part, at temperatures well within 70 K of the final,steady-state temperature. The exposure temperature 1120 K was selected as the lowest temperature for the exposure-time experiments because lower temperatures required extremely long (>30s) exposures and produced results having high uncertainty. Quation 8 shows that a plot of In k vs 112' (the Arrhenius plot) should be a straight line with slope equal to the activation energy for the vaporization process. Figure 6 shows the Arrhenius plots for lead monoxide molecules vaporized using the Ar and the H2/Ar purge gas. Figure 6 also shows the curve (straight line) obtained when the regults of percent loss of lead (using the Ar purge gas) as a function of temperature published by Byrne et al.18 are (as explained below) fitted to the Arrhenius plot. Figure 7 shows the Arrhenius plots for tin with the Ar purge gas, and with the H2/Ar purge gas and the reworked data of Byrne et al.18 As mentioned earlier, the
erence.
steady-state temperature of the probe is probably below the steady-state temperature of the graphite furnace wall,lg16 so the Arrhenius plots for the exposure-time experiments have ale0 been constructed using temperatures lower than the tube wall temperatures shown in Figures 6 and 7. Table I11 lists the activation energy, E,, obtained using the tube wall temperatures along with the E, values calculated using temperatures 100 and 200 K lower than the tube wall temperatures. The calculated E, value decreases and the standard deviation increases as the effective temperature is decreased. The standard deviation of the slope of the least-squares-fit plot increases substantially if the effective temperature is lowered by 300 K below the tube wall temperature. Table 111lists the range of possible E, values for the vaporization process ob~ervedin the exposure time experiments. The E, values within the tabulated range are normally attributed to chemosorption. Chemical reactions involving the breaking of chemical bonds generally have E, values in the range 200-500 kJ mol-' and higher, whereas physisorption is generally characterized by E, values below 40 kJ The activation energies for the vaporization of PbO and SnO from the probe in the Ar purge gas generally agree with those for vaporization of PbO and SnO from the tube wall. Table N shows thermochemicalproperties of the relevant lead and tin species. Comparison of Table III with Table N shows that the activation energies obtained show no correlation with vaporization from the bulk oxides of Pb or Sn. If, as we have assumed, the small amomta of analytm used form a monolayer
1150
ANALYTICAL CHEMISTRY, VOL. 64, NO. 10, MAY 15, 1992
or a submonolayer of analyte molecules on the surface of the graphite probe, then the experimentiallymeasured activation energies should reflect the energies of desorption of the oxide molecules from the surface of the graphite probe. When hydrogen is present in the purge gas, the activation energies for demrption of both PbO and SnO are higher, which suggests that hydrogen may alter the desorption energy of these oxides. The effect of chemisorbed hydrogen on the atomic absorption signal profile of lead and several other metals have been discussed by Sturgeon and BermanSz3Bansal et aLZ4have shown that hydrogen chemisorb on carbon surfaces and forms a stable surface complex. Probably, the graphite surface with hydrogen chemisorbed on its forms bonds with the lead and tin oxides which are different from those formed by the graphite surface without chemisorbed hydrogen. Cutoff Temperature Experiments in the Graphite Tube Furnace. As mentioned earlier, activation energy values for vaporization of lead and tin from the graphite tube wall were calculated using the data published by Byrne et al.,18 who conducted experiments in which test solutions containing lead or tin (as the nitrate and as the chloride, respectively) were dried and charred and then heated rapidly to temperatures up to and including the appearance temperature of lead and tin in the graphite furnace. At a preselected cutoff temperature, the power to the furnace was cut off and the furnace was purged with 100% argon gas. The sample was then rapidly heated to 2300 K, and the lead or tin atomic absorption signal was recorded. Byrne et al.18 concluded that lead and tin were lost as PbO(g) and SnO(g), respectively, at temperatures lower than their appearance temperatures. From the data on the loss of lead and tin vs the cutoff temperature reported by Byrne et al.lS the activation energies for the desorption (vaporization) of PbO or SnO have now been calculated.16 Since the initial temperature of the furnace and the heating rate are known, it is possible to convert the temperature scale into the time scale. The temperature change during a small time increment is small and thus the error involved by assuming a constant temperature over the small increment in time is small. Division of the mass of lead lost during the small time increment (-A[PbO(c)]) by the time increment (At) gives the rate of vaporization during that time increment. For this small time increment, assuming a first-order process, the following approximation may be made -A[PbO(c)] -d[PbO(c)] = kl[PbO(~)] (10) At dt or -A[PbO(c)] kl = At [PbO (c)] where kl is a first-order rate constant, and [PbO(c)] is the amount of lead in the form PbO(c) at the midpoint of the small time increment. If one assumes that PbO(c) exists as a monolayer or submonolayer on the graphite tube surface (a valid assumption for the small mass of lead used for this experiment), then the mass is proportional to the surface coverage. A plot of In k l vs 1/T should then be a straight line with slope equal to the activation energy for vaporization of the oxide, not the atom, since no atomic absorption signal is observed during this experiment. Any observed uncertainty in the E, values would be due to uncertainty in the value of k,. The uncertainty in the kl value may be calculated by considering the minimum and the maximum value of the rate constant over the small temperature interval, assuming that the temperature is a linear function of time over the small temperature range involved:
-
o'20
I
1620 K
A
0.00 11
14
13
12
15
Time (s) Flgure 8. Atomic absorption signal proflles of lead when the graphite probe was Inserted at the fdlowlng atomization temperatures: 1340, 1440, and 1620 K ( M n g the bwer-temperature exposue step). The atomlzatkm time was 10 s. The absdssa is t h e h from the beglnnhg of the atomlzatlon cycle. The probe was inserted 10.83 s after the beginning of the atomization cycle.
o'20
1
11
1890 K\k/1800
12
K
13
14
15
Time (s) Figure Q. Atomic absorptlon signal profiles of lead when the probe was inserted at the following atomlration temperatures: 17 10, 1800, and 1890 K (omhing the lower-temperature exposure step). The atomlzatkn time was 10 s. The abscissa Is the time from the beglnnhg of the atomlzatlon cycle. The probe was Inserted 10.83 s after the beginning of the atomlzation cycle.
where T1> T2.For example, using eq 12 one finds that for E, = 100 kJ mol-l, T1= 990 K and Tz= 1010 K, the ratio kTI/kT2= 0.79. The data published by Byrne et al.18 and eq 11were used to calculate the data points shown on Figures 6 and 7; eq 12 was used to calculate the uncertainty of these points; a 50-ms interval was used for lead, and a 10-ms interval was used for tin. The activation energies obtained for desorption of PbO and SnO from the graphite tube wall, calculated by using this method, are presented in Table 111. Rayson and Johnsonz6 have used an experimental technique similar to that of Byrne et al.18 combined with an interpretation similar to that of Gilchrist16 to calculate the activation energy of copper desorption in the graphite tube furnace. Atomization from the Graphite Probe. Figures 8-14 were obtained by atomization from the probe (omitting the low-temperature exposure step) at the indicated atomization temperature, using the purge gas flow interrupt mode of furnace operation. Figure 8 shows the effect of tube wall temperature, T,, on the lead signal for T, = 1340,1440, and 1620 K. The peak height and the peak area increase with increasing temperature, whereas atomization time, 71 (defied as the time from the appearance of the atomic absorption signal to the peak of the signal profile) decreases with increasing temperature. The 10-fold increase in the peak area for T, = 1620 K over T, = 1340 K is due to greater degree of dissociation of PbO(g) molecules at 1620 K than at 1340
ANALYTICAL CHEMISTRY, VOL. 64, NO. 10, MAY 15, 1992
Pb
h
v
[
0'30
1620 K
x
Sn
Q)
0 0.25
d
a
\o
B
1111
0.20
e
0.20
2
0.15
0 m
Y
0.10
g-J0.10 .+ Q) m
0.05 0.00 0
% .
1
3
2
4
(d
B
Time, s
cy
F@m 10. Calculated temporal dlstrlbutk~of the molecular and atomic lead specleg in the graphite probe fumace at the foibwing atomizatkn temperatures: 1340 and 1620 K. Time on the abscissa was counted from the appearance time of the respecthe species. 0.30
0.00 1200
1600
2000
2400
2800
Temperature (K) Figure 13. Peak-height absorbance curves of lead and tin atomized from the graphtte probe as a function of atomization temperature. Purge gases: 5 % (v/v) H,/Ar; 100% Ar; 2% (v/v) OJAr.
I
Sn
Pb 2.50
Q)
m
2
W
(d
Q)
2.00
E..
1.50
E .+
9
h
53
Q)
V
3
1.00
k
cd --
& 0.50
a 4
Time (s) Flguv 11. Atomlc absaptkn signal profiles of lead when the graphite probe was inserted into the furnace at tube wail temperature 1620 K for 5 % (v/v) H,/Ar and for 100% Ar purge gases. Time on the absdssa was counted from the beginning of the atomization cycle. The probe was inserted 10.83 8 after the beginning of the atomization cycle.
0.00 1200
I 1600
2000
2400
2800
Temperature (K) Figwe 14. Appearance time of lead and tln atomic absorption signals for atomlzation from the graphite probe as a function of atomization temperature. Purge gases: 5 % (v/v) HJAr; 100% Ar; 2% (v/v)
O,/Ar.
1200
1600
2000
2400
2800
Temperature (K) Figure 12, Peak-area absorbance curves of lead and tin atomized from the graphite probe as a function of atomizatlon temperature. Purge gases: 5 % (v/v) H,/Ar; 100% Ar; 2% (v/v) OJAr.
K. Figure 9 shows the effect of T , on the lead atomic absorption signal profile for T , from 1710 to 1890 K. The peak height increases with increasing T,, whereas the peak area remains constant and T~ decreases. If the dissociation of PbO is complete at 1710 K, then there will be no increase in the peak area with increasing T,, but the condensed-phase lead monoxide will vaporize at a greater rate, giving the observed increase in the peak height and decrease in the atomization
time. Figure 10 shows the calculated distribution of the lead species for two different temperatures, assuming that the lead is vaporized as PbO and that there is thermodynamic equilibrium in the gas phase. Figure 10 was drawn using experimentally determined rate constants and activation energies, which were used to model the supply of PbO, and the diffusion coefficients given by L'vov et al.%to model the removal of analyte molecules and atoms from the furnace. A heating time of 0.5 s for the probe to attain the final temperature was included in the calculation of k. This heating time represents the induction time discussed earlier, and its inclusion has improved the fit of the calculated curves to the experimental curves. A Po,scale similar to that used by Frech et which excludes heterogeneous equilibrium between solid graphite and oxygen,2I and the free energy minimization routine of Frech et al.,2l along with published thermodynamic data,were used for the equilibrium calculations. Comparison of Figure 8 with Figure 10 show8 that the experimental results are consistent with the hypothesis that PbO vaporizes from the graphite probe and is dissociated in the gas phase and that gas-phase thermodynamic equilibrium exists since the changea in peak heights, peak areas, and appearance times with T , are what would be expected from the gas-phase thermodynamic equilibrium The inclusion of a 0.5sheating time for the probe thus gives better fit of the experimental curve to the calculated curve and also agrees with the observation made by Giri et al.13
1152
ANALYTICAL CHEMISTRY, VOL. 64, NO. I O , MAY 15, 1992
Earlier it was mentioned that hydrogen gas, chemisorbed on the graphite surface, resulted in a different activation energy for the vaporization of PbO and SnO. Figure 11shows that when the H2/Ar purge gas is used, the atomic absorption signal profile for lead has an earlier appearance time and a greater peak height but an equal peak area, compared with the case when the Ar purge gas is used. The effects on the atomic absorption signal profile of lead are consistent with what has been reported by us and by other authors.lS2lP This point will be discussed further in the next section. Figures 12 and 13 show the peak area and the peak height absorbance for lead and tin atomized from the graphite probe (omitting the low-temperature exposure step) as a function of atomization temperature. The Ar purge gas and the H2/Ar purge gas were used for lead, and the Ar purge gas, the H2/Ar purge gas, and the 02/Ar purge gas were used for tin. The same trends can be seen for lead and tin with respect to change in peak area, peak height, and appearance time with increasing atomization temperature (T,).Figure 14 shows appearance time (defined as the time from the insertion of the graphite probe to the first appearance of the atomic absorption signal) of lead and tin as a function of atomization temperature. The dependence of peak area on temperature and purge gas composition (Figure 12) can be explained in the following way. L'voV3l has shown that the peak area of an atomic absorption signal is proportional to the total number of atoms, No, introduced into analysis volume, multiplied by the residence time of atoms in the furnace, r2 (defined as the time taken by the atomic absorption signal to decay to l / e of its value by a first-order loss process)
where N ( t ) is the number of atoms in the analysis volume at time t. For both lead and tin the peak area is greater when the H2/Ar purge gas is used than when the Ar purge gas is used. Presumably, the added hydrogen lowers the partial pressure of oxygen, thereby shifting the oxide dissociation equilibrium toward greater dissociation of the oxides. The observed peak-area enhancements for the H2/Ar purge gas compared to the Ar purge gas for T, < 1600 K occur primarily because the lower Po,shifta the dissociation equilibrium of PbO(g) to a greater degree of dissociation of PbO(g). At temperatures >1600 K the oxygen levels are lowered by reactions between the gas-phase oxygen and the graphite tube. This, combined with the higher equilibrium constant at a higher temperature, gives virtually complete dissociation of the oxides for the Ar and the H2/Ar purge gases; hence, no further increase in the peak area occurs with further increase in the temperature. The greater peak area for the H2/Ar purge gas compared to the Ar purge gas in the plateau region of the atomization curve is due to the greater rate of vaporization of PbO(c) when the H2/Ar purge gas is used, as discussed earlier. Since, for atomization of lead from the graphite probe at a given atomization temperature, the mean residence time ( T ~ of ) lead atoms in the furnace is virtually the same for the Ar purge gas as it is for the H2/Ar purge gas, but the atomization time (rl)is longer for the Ar purge gas, the peak area of the atomic absorption signal profile is lower for the Ar purge gas than for the H2/Ar purge gas. The effect of the 02/Ar purge gas on the peak area of tin is shown in Figure 12. The added oxygen has an effect opposite to that of the added hydrogen, presumably, because the added oxygen shifts the SnO(g) dissociation equilibrium toward a lesser degree of dissociation of SnO(g). Figure 13 shows that for lead with the Ar or the H2/Ar purge gas and for tin with the Ar, the H2/Ar, or the 02/Ar purge gas, peak height increases with increasing temperature. L'vov (see ref 1,p 117) has shown that the peak height of an
Table V. Effect of a Forced, Convective Flow of Ar Purge Gas during the Initial Exposure Time on the Atomic Absorption Signals of Lead and Tin
element
purge flow during temp, K initial initial final exposure
Pb
1515
2200
Sn
1710
2430
on off on off
peak-area absorbanceo initial final
0.00 0.077 f 0.004 0.03 f 0.02 0.044 f 0.007 0.00 0.057 f 0.005 0.00 0.090 f 0.010
The numbers after f represent uncertainties in the reported values.
atomic absorption signal is dependent on the rate of atomization of the analyte through the dependency of peak height on the r1/r2ratio when the r 1/ r 2ratio is not very much smaller than 1. The observed increase in peak height with increasing graphite tube wall temperature is due to an increase in the rate of vaporization of PbO(c) and SnO(c) with increasing temperature. Peak height initially increases rapidly because of an increase in the degree of dissociation of PbO(g) and SnO(g), coupled with an increase in the rate of vaporization of PbO(c) and SnO(c). At high temperatures where the vaporization of PbO(c) and SnO(c) is complete and the dissociation of PbO(g) and SnO(g) is nearly complete, peak height increases less rapidly with increasing temperature and approaches a maximum value, N-, when r1
