864
Anal. Chem. 1982, 5 4 , 864-869
to be a suitable substrate for a mercury film electrode, as applied to stripping analysis (9, 16). However, the results obtained with the present configuration (defined cadmium and lead peaks only at the M concentration level, employing 2-min deposition, 500 rpm stirring speed, and the differential pulse stripping mode (not shown)) are inferior when compared with those of previous studies at in situ plated thin mercury film electrodes. This may be attributed to the increased background current and nonuniform mercury plating due to the rough surface of the graphite-epoxy that provides the electrical contact to the RVC.
(2) Lingane, J. J. Anal. Chim. Acta 1948,2, 584. (3) Bard, A. J. Anal. Chem. 1963, 35, 1125. (4) Clem, R. G. Anal. Chem. 1971,43, 1853. (5) Wang, J. Nectrochlm. Acta 1981,26, 1721. (6) Strohl, A. N.; Curran, 0.J. Anal. Chem. 1979, 51. 353. (7) Blaedel, W. J.; Wang, J. Anal. Chern. 1979, 57, 799. (8) Strohl, A. N.; Curran, D.J. Anal. Chem. 1979, 51, 1050. (9) Blaedel, W. J.; Wang, J. Anal. Chem. 1980,52,76. (IO) Nowell, V. E.; Mamantov, G. Anal. Chern. 1977,49,1470. (11) Wang, J. Anal. Chem. 1981, 53, 2280. (12) Wang, J. Anal. Chlm. Acta 1981, 129, 253. (13) Lingane, J. J. J. Am. Chem. SOC.1945, 67, 1916. (14) Mlller, 6.: Bruckensteln, S. Anal. Chem. 1974,46. 2026. (15) Blaedel, W. J.; Wang, J. Anal. Chem. 1980,52. 1697. (16) Blaedel, W. J.: Wang, J. Anal. Chem. 1979, 51, 1724.
LITERATURE CITED (1) Bard, A. J.; Faulkner, L. R. “Electrochemical Methods, Fundamentals and Appllcations”; Wiley: New York, 1980; Chapter 10.
RECEIVED for review
November 20, lg81*Accepted February
16, 1982.
Mechanisms of Vaporization of Vanadium Pentaoxide from Vitreous Carbon and Tantalum Furnaces by Combined Atomic AbsorptionIMass Spectrometry D. L. Styrls” and J. H. Kaye Pacific North west Laboratory, P. 0. Box 999, Richland, Washington 99352
High-temperaturevaporlratlon of vanadium pentaoxlde (V,O,) samples from vitreous carbon and tantalum furnaces that are heated reslstively in ultrahigh vacuum is lnvestlgated with the slmultaneous atomic absorptlon/mass spectrometric technique. The advantage of thls method Is that if allows the observation of lntermedlates In the vapor phase. The correlated experimental resuits are related to thermodynamically feaslbie reactlons which are consistent with the appearances of these Intermediates and the temperatures at whlch they appear. The atomlzatlon mechanlsms that are indicated by these reactions are developed.
Fundamental understanding of furnace atomic absorption spectroscopy ( U S ) depends largely on what is known of the mechanisms which control atomization processes that take place within the furnace. Surprisingly few workers have investigated these mechanisms, which are responsible for the signal amplitudes and shapes on which AAS analyses are based. Some of the earliest investigations were made by Aggett and Sprott ( 1 ) and by Campbell and Ottaway (2) in 1974 by correlating appearance temperatures and temperatures a t which reductions of metal oxides by carbon are thermodynamically favorable. Aggett and Sprott compared their results from a carbon rod with results from a tantalum strip. They concluded that (i) reduction from a graphite atomizer occurred for oxides of Co, Fe, Ni, and Sn but not for the other 12 elements investigated and (ii) formation of gaseous atoms from the reduced metals (Co and Fe) is controlled by the vapor pressure of those metals. On the other hand Campbell and Ottaway concluded that the majority of the 27 elements they investigated formed gaseous metal atoms by reduction of the oxides in a condensed phase in a carbon furnace. Eklund and Holcombe (3) have recently shown by differential thermal analysis and X-ray photoelectron spectroscopy that a metal oxide such as CuO can be reduced by graphite at a temperature that is lower than the appearance temperature, which implies that the vapor pressure of the
analyte can indeed be controlling the appearance temperature. Sturgeon, Chakrabarti and Langford ( 4 ) used a thermodynamic approach (assuming analyte solid-gas phase equilibrium) combined with a kinetic approach whereby the atomization process is characterized by a rate constant. Activation energies involved in the primary atomization process were determined from the resulting Arrhenius type equation and were related to dissociation energies and heat of atomization of the metal. Johnson, Sharp, West, and Dagnall (5) used a Boltzmann distribution to describe a vaporization rate characterized by the heat of vaporization of the metal or dissociation energy of the metal-oxide bond. Their resulting model of atomization from a carbon filament atomizer worked well for molybdenum but it did not appear to be applicable to metals which vaporize a t lower temperatures. Sturgeon, e t al. ( 4 ) , point out that it may not always be valid for reductions by tantalum to be neglected as was done by Aggett and Sprott (1)in analyzing their tantalum strip data. At high temperatures the Ta20, surface layer may not be sufficiently stable to prevent reduction from occurring. The results of the tracer work of Maessen, Balke, and Massee (6) to investigate thermochemical reactions in graphite furnaces led to the conclusion that reactions with graphite furnaces are so complex that atomization behaviors of arbitrary compositions of analytes cannot be predicted reliably. Most of the above approaches attempt to resolve the problem by thermodynamic or kinetic considerations. The concern regarding validity of these methods is that thermodynamic equilibrium or isothermal conditions are not achieved during the heating pulse. Techniques which will allow direct observations of intermediate phases and interferents, or the effects thereof, must be used to obtain additional data. Eklund and Holcombe (7, 8), for example, have monitored the absorbance under steady-state and transient conditions at various hieghk above a graphite filament atomizer. They showed that (i) oxidation of the analyte can occur in the gas phase via reactions with the thermal decomposition products of the interferent and (ii) gas-phase oxidation of the analyte can be depressed by a preferential binding of oxygen with the interferent in the
0003-2700/82/0354-0864$0 1.251’0 0 1982 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 54, NO. 6. MAY 1982
I O CI1"OI"UP 16,AOS I,
mUn 1. Schematic of t b AASIMS aooaratw. The Gashed se~ment illistrates the furnace in the loading po&tion. When the furnaci is in contact with the power contacts, the top (loading)hole is positioned directly under the wnductance limiter orifice to the QMA.
gas phase. Most recently, Wahah and Chakraharti (9) have shown that yttrium atom formation from pyrolitic carbon, tantalum, and tungsten surfaces can he explained by a single mechanism. These investigators attribute atomization as primarily due to thermal decomposition of Y,O,(s) and thermal dissociation of the resulting YO(g). Observations by Styris and Kaye (IO),using a multiple probe (AAS combined with mass spectrometry) technique, have provided strong evidence for interactions between Talos on the surface of a tantalum atomizer and samples of RhCI. T h e value of this technique is that i t allows observations of vaporized intermediates while simultaneously monitoring absorption of a particular atomic species. By correlating these events, it is poasihle to obtain a hetter understanding of the vaporization processes involved. The results in this case support the Aggett and Sprott hypothesis (I) that a metal salt in a tantalum furnace will first dissociate to the oxide and finally to metal atoms. I t is advantageous to apply this multiple probe method (AAS/MS) to a given analyte in carbon 88 well as in tantalum furnaces. The vaporization mechanisms which are determined for each furnace material must then exhibit characteristics consistent with both sets of data. Some of the ambiguities which normally prevail in proposed vaporization mechanisms might therefore he eliminated. The purpose of the work described in this paper is to apply the combined AAS/MS method to the investigation of vaporization processes which occur when vanadium pentaoxide is heated in vitreous carbon and in tantalum furnaces. EXPERIMENTAL SECTION Apparatus. High-temperature vaporization of V,O, samples from vitreous carbon and tantalum furnaces that are heated resistively in ultrahigh vacuum is investigated with the simultaneous AAS/MS technique. The apparatus and methodology have heen detailed in a previous paper (IO)and will therefore be described only briefly in this section. The all-metal, hakeahle vacuum system used in these experiments contains the furnace in a reaction chamber and an Extranuclear quadrupole mass analyzer (QMA) in an adjoining chamber. A schematic of this arrangement is shown in Figure 1. The system is pumped differentially across these chamhers with two closed cycle, 1200 L/s He cryopumps. Gaseous species which escape through the loading hole in one side of the furnace enter the $MA chamber through an orifice which separates the two chambers. The species are then ionized by electron bombardment and mass analyzed. The ahsorption of light from a hollow cathode source, focused through sapphire vacuum windows and through the furnace, is monitored with a photomultiplier on an Ehert-type monwhromator. Electrical power to the furnace is supplied hy an Instrumentation Laboratory (Lexington, MA)
865
Model 455 atomizer power supply A loading chamber connects the reaction chamber through an isolation gate valve. The furnace can be retracted into the loading chamber by means of a linear translator. Loading can therefore be accomplished in an inert atmosphere while maintaining vacuum within the reaction and QMA chambers. Rough pumping of the loading chamber is accomplished by cyrogenic adsorption. The vitreous carbon furnaces used in this work were fabricated by Fluorocarbon Process Systems (Anaheim, CA) in the shape of 4.69 mm i.d. X 6.45 mm 0.d. X 38.94 mm long cylinders. The tantalum furnaceswere shaped into 6.4 X 6.4 X 50 mm right angle parallelopipeds from 0.127 mm thick, 99.99% pure tantalum foil; details are descrihed in ref 10. Procedure. The V,O, samples used in this work contain 500, 375,250,50,25,and 5 mg L-' V in 2 pL of aqueous solution. The more dilute solutionshelp determine analyte concentration effects. The sample is loaded into the furnace in an AI atmosphere and vacuum dried hy slowly evacuating the loading chamber with the cryosorption roughing pumps. The rough-pumped (0.1 Pa) loading chamber is opened to the evacuated Pa) reaction chamber and vacuum pumping is continued by means of the helium cryopumps. The furnace is then brought into the reaction chamber hy extending the linear translator. Furnace heating far atomization is initiated only after achieving a reaction chamber pressure of less than 6.5 X lo* Pa. Heat treatment ("ashing") sequences are not used in this work. The QMA is usually adjusted for a single maw of interest. This mass is then monitored continuously during the furnace heating time. Argon, which is used to backfill the loading chamber, and krypton and xenon calibration gases are used to cslihrate the QMA. Preliminary seanning over a series of masses at a scan rate of 1amu s-' during furnace heating helps identify all masses that are included in the vapor phase. These masses are then investigated separately as functions of time. A vanadium hollow cathode lamp (318.4 nm) is used as the radiation source for the absorption spectroscopy. Outputs from the QMA and from the monochromator photomultiplier tube are each filtered with Spectrum 1021A electronic filters and then monitored simultaneously and photographed fmm a multiple beam oscillmpe trace. Background corrections are made from results obtained when the furnace is loaded with a deionized water sample. The inside surface temperature of the vitreous carbon furnace is monitored with a 0.076 nun W 5% h W 26% Re thermocouple (TC)which a n t a & this su-face through the fumace loading hole. The TC output is monitored for a furnace heating cycle which uses the same power settings and the same furnace as that used for sample vaporization. A temperature w. time profile (without TC response corrections) is then obtained from calibration data supplied by Omega Engineering (11). The response time ( 7 ) of indications the TC is determined from surface temperature (TAU) obtained by observing the appearance time of the atomic ahsorption signal for gold when the furnace is loaded with a gold wire sample. The value of 7 is the difference between this appearance time and the time the TC registers a temperature equal value determined from vapor pressure data and the to a TAU furnace pressure. This pressure in the furnace, at the gold appearance time, is evaluated as a function of furnace gas conductance, pumping speed, and the recorded reaction chamber preasure by aasuming that throughputs of the furnace and pump are nearly equal, to within a usually negligible term V dp/dt (V is the volume of the reaction chamber and dp/dt is time rate of change of pressure at the appearance time). This assumption is reasonable since pressure in the reaction chamber is observed to he relatively stable when the gold signal appears. Since gas conductanceof the furnacecan be written in terms of temperature, atomic weight of the gas, and furnace dimensions, the pressure within the furnace is also a function of these variables. The magnitude of TAUis obtained by assuming equivalence of this function with the vapor pressure approximation B PEA+-
TA"
(1)
where A and B are empirically determined constants (12). These = 1234 K and that r a t this temexperiments indicate that TAU perature is 0.5 s. Finally, the TC temperature curve for T 2 TAU is corrected for this response time. This curve is included with
800
ANALYTICAL C M M I S T R Y , VOL. 54. NO. 6. M A Y
1982
4. Dual beam oscilloscope trace of AAS/MS signals for V,O,
(182amu) in a viireous carbon furnace: (lower trace) VO,' (83amu) MS signal lor IO' R input resistor. 5 mV1divisian. ionizer electron
energy is 43 ev; (upper trace) vanadium AAS signal, 0.2 Vldivision: sweep rate is 1 sldivision.
FURNACE HEaITINO ,,ME.
s
Composite of AASIMS signal and fumace temperature as funcibns of furnacs heating tlme for V,Os (182amu) in a vlireous carbMl hmaca. The fmbcu!ar ion s@waklw ihe VO;' (83amu) and VO+ (67 amu) are magnified for clarny. Flgun 2.
FURNACE HEATlNO TIME. I
ncE*. 3. Composne of AASIMS dgfmkand funace temperatue as hmcilons of furnace heanng @mefw V,O, (182amu) in a tantalum hmace. The only molecular Ion signal ObJeNed for this case WRBspnds to vO+ (67 amu). the speetrosmpy data shown for the carbon furnace case in Figure 2 and is discussed in the next section. The temperaturetime profile for the Ta furnace was found solely by monitoring appearance times of Au, Pt, and Co along with the reaction chamber presaure and by ealculating the temperaturea in the manner described above. This gives three temperatures (1334,1543,and 2061 K at 3.2,3.5,and 4.6 8 , respectively) which determine the temperaturetime curve. The 100 K difference in the gold appearance temperatures for the two furnace materials is due to differences in chamber pressures at the respective appearance times. Thia pressure for the tantalum furnace case was about 25 times that monitored for the carbon fumace. The temperature curve is included with the speetrosmpy data shown for the tantalum furnace case in Figure 3. RESULTS AND DISCUSSION An example of the ~ b s e ~ simultaneous ed AAS/MS signals is shown in Figure 4. These particular traces are from the
vitreous carbon furnace experiments. In this case the MS system is calibrated to monitor V02+ as a function of time following initiation of the heating pulse. The AAS output (upper trace) is due to gaseous vanadium absorption at 318.4 nm. The initial MS precursor is a background signal which is also observed during heating after a 'blank loading" with deionized water. Composites of such signals from a given furnace for each mass observed, minus hackgrounds, provide the necessary information for correlating these vaporization signals. Figures 2 and 3 show composites as functions of heating times for the vitreous carbon and tantalum furnaces, respectively. Furnace temperatures are superimposed, and for the carbon furnace, the molecular mass signals are shown magnified for clarity. Molecular s p i e s appear from both furnaces, hut both VO,' and VO+ are seen for the case of vitreous carbon whereas only VO+ is observed for the case of tanatlum. Appearance times of the molecular ions, the Vc ions, and the V absorption signals are observed to be equal to within 0.2s for both cases. Uncertainty in the simultaneity of the appearance of these signals for the case of the vitreous carbon furnace occurs because of the low rise times of the relatively weak V+ and AAS signals during their appearance phases. The molecular signals for the cases of both furnaces begin to decay prior to the time that their respective AAS signals (from V) reach full magnitude. Temperature rise times are similar for the two furnaces: i.e., 2000 K is reached in 4-4.5 s. The appearance temperatures are all in the neighborhood of 1500 K, with those from the vitreous carbon furnace being lower than for tantalum by about 100 K. Within the certainty of the temperature determinations, and considering that neither furnace provides an isothermal environment, it is reasonable to use a value of 1500 K for the appearance temperature. The maxima in the AAS traces in both cases appear to lag the V+ ion maxima, probably because of the geometry of the furnaces. Vanadium vapor concentrations can become depleted in the hot central region of the furnace due to diffusion of the vapor toward the ends, which are cooler due to thermal conduction through the electrical contacts. The $MA primarily monitors the vapor which exits in the central hightemperature region of the furnace through the loading hole. Hence the MS signal can indicate a vapor depletion rate while the AAS signal is indicating that the vapor concentration is increased a t the cooler ends. This results in a separation in time of the AAS and MS peak amplitudes. The relative magnitudes of theae amplitudes are also considerably different for the two furnace types. Ratios of AAS (V) to the MS (V+) amplitudes are about 3 times greater for the case of the vitreous carbon furnace.
ANALYTICAL. CHEMISTRY, VOL. 54, NO. 6, MAY 1982
8167
-.
Table I. Free Energy of Reduction oxide
mp, K
V2C’,
943+ 10
V2C’,
23405 20
V,C’,
18185 5
--
reaction temp, K
-AGreactio:,
kcal mol-
800 800 800 800 800 2000 1500 1500
Effects due to the initial quantity of analyte in the furnace are observed in the V02+ data. In both type furnaces the quantitative molecular VO+ data are too imprecise to be correlated with the quantity of analyte. Peak amplitudes of the V02+ signalai (carbon furnace) are enhanced when the vanadium content is increased from 10 to 100 ng. These peak signal amplitudes then exhibit a decreasing trend when the vanadium content is increased from 100 to 500 ng and from 500 to 750 ng. Carbides of vanadium which could conceivably appear in the gas phase within the vitreous carbon furnaces do not appear in the ME; spectra. This alone is not sufficient, however, to rule out the possibility of carbide formation since the furnace temperatures are likely to be too low to produce significant vapor pressures of carbides. Similarly, there is no evidence that TaO and TaOz are vaporized from the tantalum furnaces during vaporization of any of the vanadium produch. The data from Figures 2 and 3 are compared most easily if they are converted to a temperature representation. When this is done the dlata appear as shown in Figure 5. Keep in mind that temperature in this case can be considered the maximum temperature within the furnace at a given time; the furnace is not heated isothermally. This temperature occurs in the vicinity of the central region of the furnace where the sample is loaded. Hence it is representative of the temperature of the sample only prior to vaporization. The appearance of the atomic vapor at temperatures well above the 943 K melting temperature of the Vz05sample is indicative of reductions to species having higher melting points. Hence the gaseous metal oxides observed in these experiments are either the first generation products of these reductions or products that are generated as a result of further increases in temperature and associated reductions and/or dissociations. Which of these applies depends an the thermodynamic characteristics of the oxide. The solid-solid transition processes which may be precursors to the appearances of the gaseous oxides can be evaluated by assuming localized thermodynamic equilibrium at the position of the solid sample; i.e., this localized region can be considered isothermal. Application of thermodynamic equilibrium calculations involving gas phases are, on the other hand, of questionable validity because of the potential problems encountered in establishing chemical and physical equilibrium when temperature rise times are short and thermal gradients are relatively large. Reactions indicated by these experiments are checked for feasibility by calculating the overall free energy changes of the reactions using the free energy values from the JANAF thermochemical tables (13). These reactions are discussed in the following paragraphs and are tabulated in Table I in terms of the reaction identification number (column 6), reaction temperature (column 3), the oxide which is reduced (column l),and its melting temperature (column 2). Note that all of these reductions are thermodynamically feasible < 0) as indicated in column 4. It should be kept (i.e., AGm,, in mind, however, that the reductions may be limited by the reaction kinetics. The relative extent of occurrence of these reactions is found by comparing the equilibrium constant Keq
17.49 81.37 69.37 29.37 31.61 18.89 3.98 93.97
-
reaction no.
log K,, 5.21 22.23 18.95 8.02 8.63 2.17 0.58 14.69
2
3 5 7 8 4 6 9
t-
!I VI
t
s5
6 1
E
“L
1500
2000
FURNACE TEMPERATURE K
Flgure 5. Composites of AAS/MS signals for vitreous carbon arid tantalum furnaces as functions of furnace temperature. The moleculiir ion signals corresponding to V02+ (83 amu) and VO’ (67 amu) are shown magnified for clarity.
as determined from AGT = -RT log Keg. Values of Kegare tabulated in column 5 . As discussed above, the reduction to lower oxides occurs at T < 943 K. The actual temperature at which this reduction occurs is unknown, but some reduction reactions that are 800 K and that have thermodynamically feasible at T reactions products which can be associated with the products of vaporization observed at the higher temperatures are Vzo,(S)
f
c(S)
and
V205(rs)+ 2C(s)
- vzo,(S)
-
V203(s)
-I- c o ( g )
+ 2CO(g)
(2)
(3) The vanadium oxide products of these reactions are converted to lower oxides and vanadium at the respective temperatures indicated by the data. For example, the appearance of VOZ at around 1350 K can be explained as the incongruent sublimation of the V 2 0 4 produced by reaction 2. Farber et al. ( 1 4 ) report that Vz04 decomposes to V02(g)and V203(s)at temperatures in the range 1000-1200 K. Fwthermore, if it is assumed that sublimation is not complete during the 1-s rise time from 1325 to 1800 K, the remaining V204will melt at 1818 K and vaporize to VOz(g) (13). This would account for the second V02+peak in Figure 5. The early, weak vanadium and vanadium oxide signals are attributed to thle initiation of congruent sublimation of the V203produced by reaction 3 and by incongruent sublimation of V,O,. Farber (14) has shown that V, VO, and VOz are products of such
868
ANALYTICAL CHEMISTRY, VOL. 54,
NO.6,MAY 1982
sublimation. Formation of VO(g) due to interactions between V(g) and VO,(g) might then explain the observed decreasing trend in the VOz(g)concentration when quantities of analyte are increased above 100 ng. If sublimation of Vz03is incomplete, the rapid increase in density of V(g) at 2000-2100 K can be accounted for by reduction of the remaining Vz03(s). This reduction reaction is given by Vz03(s)
+ 3C(s)
-
2V(s)
+ 3CO(g)
(4)
When this occurs the V(s) from this reaction sublimes with a vapor pressure of 2.7 X 10-1 Pa. The AAS/MS outputs for V appear to peak approximately 100-200 K above the V melting temperature (2190 K) (13). Hence the appearance of the V signal is controlled by vapor pressure and the time a t which the signal reaches peak amplitude is controlled by melting. The close resemblance in the shapes of the VO+ and VOz+ signals implies a strong correlation. A portion of the VOZ is probably dissociated by the 40-eV electrons in the ionizer. This resulta in the appearance of VO+ as well as V02+. These results are in agreement with the conclusion of Campbell and Ottaway (2) regarding reduction of oxides in a condensed phase. The reduction at temperatures below the appearance temperature also supports the observations made by Eklund and Holcombe for CuO (3). However, the present findings imply that the reduction process for the higher oxides is considerably more complicated, in that initial reductions do not produce the analyte metal in a free or condensed state. Incongruent sublimation and further reduction of the resulting lower oxides, at higher temperatures, occur to yield the analyte in these states. In earlier work on Vz05 by Sturgeon et al. ( 4 ) , the appearance of vanadium absorption at 2200 K is attributed to initial decomposition to Vz03 which then decomposes to VO(g) and Oz(g). Activation energy determinations suggest that a decomposition of the VO(g) yields the free vanadium. This vaporization pathway, and that described by reaction 4, may differ because of differences in experimental conditions used in defining these pathways. A lo5P a inert gas environment, ”ashing” procedures, and a graphite furnace are used on one hand, while on the other hand, the environment is vacuum, the furnace material is vitreous carbon, and “ashing“ procedures are not used. Confinement times on the surfaces of the furnaces will differ because of the environments, and reactions related to the different surface characteristics of the furnaces cannot be expected to be identical. Under the inert gas atmosphere congruent sublimation of Vz03 from reaction 3 will not be a contributing factor to the appearance of V. However, reaction 4 remains feasible at 2000 K and consequently V(g) should appear at a temperature close to melting (2190 K). This would be in agreement with the appearance temperature in ref 4. The different mechanisms must therefore be reconciled in terms of the differences in furnace materials and “ashing”. Experimental comparisons between the two materials have not been made. “Ashing”at temperatures between 800 and 1000 K should not change the results of this work since Vz03would remain in the furnace and reduction would still occur by reaction 4. However, “ashing” at temperatures well below 800 K could lead to different reaction pathways, since oxides differing from those in reactions 2 and 3 might be formed. The results from the tantalum furnace also imply reduction to lower oxides at a temperature below the melting point of the Vz05 sample. A thermodynamically feasible reaction which results in a lower oxide product at 800 K is given by 2V205(s) 2Ta(s) Ta~05(s)+ 2 V b ) + VZO&) + OAg) (5)
+
-+
The appearance of VO+ at 1520 K is then explained by reduction of the Vz03(s)produced in reaction 5 and sublimation
of the resulting VO(s). This reduction is given by 2Vz03(s)
+ 2Ta(s)
TazO,(s)
-,
+ VO(s) + 3V(s)
(6)
Unfortunately, vapor pressure data for VO are not available. However, Berkowitz et al. report on sublimation of VO in the temperature range of 1680-1950 K (15). The V(s) from reactions 5 and 6 sublimes at 1500 K with a vapor pressure of P a to give the resulting absorption and ion signals for vanadium. These signals peak at 1800-1900 K. Appearances of these vanadium signals are therefore controlled by the vapor pressure of vanadium. This is similar to what was found for the vitreous carbon furnace except that vanadium peak heights for the carbon case appear nearer the melting temperature of vanadium. The results also provide evidence that the TazOS on the surface of the tanatalum furnace is not a stable, coherent film. These reactions imply that for a lo5 P a inert gas environment the appearance temperature for V will increase to a value near that of melting; sublimation will not occur. The VO in reaction 5 will also be retained until its melting temperature (2336 K) is reached. Reactions 4 and 5 are not necessarily unique. Other thermodynamically feasible reactions (see Table I) which can occur at 800 K are given by 2V206(s) + Tab) and V,O,(s)
+ Ta(s)
-
2Vz04(s)
+ TaO,(g)
V204(s)+ TaO(g)
(7) (8)
Note that the TaO and TaOz will be produced only in the gaseous phase (16). Appearance of VO and V at 1520 K might then be explained by the reduction of the V204(s)products of reactions 6 and 7. This reduction is given by 2V,04(s)
+ 2Ta(s)
-,
TazOds) + V O k )
+ 3 V g ) + Oz(g) (9)
However, such reductions evidently do not occur in this case since the lower oxides of tantalum produced by reactions 7 and 8 are not evident in the mass spectra. Reactions 5 and 6 are therefore considered the most likely mechanisms to control the vaporization and atomization processes. The principal mechanisms proposed for the tantalum and vitreous carbon furnaces are similar. That is to say, below 943 K the Vz05is reduced to V203(s),and a t 1520 and 2000 K, respectively, V(s) is produced by the reduction of this V,O,(s). For the tantalum case, however, the initial reduction reaction also produces V(s) along with V203(s);sublimation of the V(s) produced in both of the tantalum reduction reactions follows. For the carbon case the initial reductions do not produce V(s). Consequently the free vanadium appears a t a lower temperature for the tantalum furnace case, and complete sublimation occurs before the melting temperature of V(s) is attained. Finally, the reactions suggested by this work indicate that reductions to a gas phase of vanadium do not occur for either of the systems investigated. These results may be useful in helping choose between tantalum for some AAS investigations since the prediction of reduction temperatures for both materials is possible by judicious use of thermochemical tables (15). LITERATURE CITED (1) Aggett, J.; Sprott. A. J. Anal. Chim. Acta 1974, 72, 49. (2) Campbell, W. C.; Ottawa, J. M. Ta/anfa 1974, 21, 837. (3) Eklund, R. H.; Holcombe, J. A. Talanta 1979, 26, 1055. (4) Sturgeon, R. F.; Chakrabarti, C. L.; Langford, C. H. Anal. Chem. 1976, 48, 1792. (5) Johnson, D. C.; Sharp, B. L.; West, T. S.; Dagnall, R. M. Anal. Chem. 1975, 4 7 , 1234. (6) Maessen, F. J. M. J.; Balke, J.; Massee, R. Spectrochim. Acta, Part6 1978, 338,311. (7) Eklund, R. H.; Holcombe, J. A. Anal. Chim. Acta 1979, 108, 53. (8) Eklund, R. H.; Holcombe, J. A. Anal. Chirn. Acta 1979, 109, 97.
Anal. Chern. 1982, 5 4 , 869-872 (9) Wahab, H. W.; Chakrnbarti, C. L. Spectrochim. Acta, Part B 1981, 368, 475. (10) Styris, D. L.; Kaye, J. H. Spectrochim. Acta, Part 8 1981, 368, 41. (11) "The OMEGA Temperature Measurement Handbook"; OMEGA Engineering, Inc., Stamford, CT, 1981. (12) Roth, A. "Vacuum Technology"; North Holland: New York, 1978. (13) Chase, M. W., Curnutt, J. L.; Prophet H.;McDonald, R. A,; Syverud, A. N. J . Pbys. C:hem.Ref. Data 1975, 4, 136-175. (14) Farber, M.; Uy, 0. M.; Srivastava, R. D. J. Cbem. W s . 1972, 56, 5312. (15) Berkowitz, J.; Chupka, w. A.; Ingram, M. G. J . Chem. PbyS, 1957, 67, 27.
869
(16) Ingram, M. G.; Chupka, W. A.; Berkowitz, J. J. Cbern. Pbys. 1957, 27, 589.
RECEIVED for review December 11, 1981. Accepted January 28, 1982. This paper is based on work sponsored by the Division of Chemical Sciences of the Department of Energy and performed under DOE Contract NO. De-AcO6-76RLO1830.
Determination of Alkyl Anilines and Alkyl Pyridines in Solvent Refined Coal Distillates and Aqueous Extracts by Gas Chromalography/Mass Spectrometry Lawrence J. Felice Pacific Northwest Laboratoty, P. 0. Box 999, Richland, Washington 99352
Derhratlzatlon wHth acetb anhydrlde In aqueous media coupled wlth caplllary gas chromatography/mass spectrometry was used to dlstlngulrsh alkyl anlllnes from alkyl pyrldlnes. By use of thls approach anlllne, C1 anlllnes, and C-2 anilines, as well as lesser quantltles of C-2 pyrldlnes and C-3 pyrldlnes, were podtlvely Identifled In a solvent refined coal (SRC)-I1 blended distallate and In an aqueous extract of the dlstlllate. The extractlon procedure and acetylation condltlons employed were examined In detail and found suitable for quantltatlve analysis when comblned wlth capillary gas chromatography. The SRC-I I blended distillate contalned 8200-9025 pg/mL anlllnes and 1335 pg/mL pyridines. Dlstllled water squlllbrated wlth the cllstlllate (1:100, SRC-II:HpO) contalned 65.9-78.9 pg/mL anlllnes and 5.6 pg/mL pyrldlnes.
As the synthetic fuels industry develops, environmental studies are needed to provide information for the growth of an environmentally acceptable industry. Of particular concern is the relatively high quantity of nitrogen- and oxygen-containing compounds found in coal-derived liquids (I) and the fate of such components if spilled into the environment. Many of these polar water soluble species, such as anilines and phenols, are toxic and have a high potential for environmental mobility, thus posing a threat to surface and ground waters. In aquatic toxicity studies of aqueous extracts of an SRC-TI distillate for example, phenols were found to contribute to more than 50% of the organic carbon in the extracts (2). In order to accurately assess the potential hazards associated with a spill of coal-derived liquids, analytical methodology to characterize the polar constituents of these liquids and to follow these compounds through environmental systems must be available. In this report an analytical approach based on chemical derivatization with acetic anhydride in aqueous media followed by capillary gas chromatography (GC) and capillary gas chromatographyJmass spectrometry (GC/ MS) is presented for the identification and quantitation of trace quantities of alkyl anilines andl pyridines in SRC-I1 distillates and aqueous solutions contacted by SRC-I1 liquids. Both anilines and pyridines have beten previously analyzed in coal-derived liquids by GC/MG (3-5). However, the similarity of the electron impact mass spectra of alkyl anilines and pyridines and the 0003-2700/82/0354-086cl$0 1.25/0
lack of a complete set of standard compounds prompted the development of alternative analytical method that could unambiguously differentiate anilines from pyridines. Derivatization of alkyl anilines with acetic anhydride occurs rapidly in aqueous media and results in acetanilide derivatives which are easily distinguished from the underivatized alkyl pyridines by GC or GC/MS. In addition to providing a useful qualitative tool, the acetylation approach was found to be quantitative and was used for the measurement of alkyl anilines a d pyridines in an SRC-I1 blended distillate and aqueous extracts of the distillate. The approach presented is thought to have broad applicability and to be useful for distinguishing primary aromatic amines from nitrogen heterocycles. EXPERIMENTAL SECTION Chemicals. All solvents used were Burdick and Jackson distilled in glass. Standard pyridines, anilines, and acetanilides were purchased from Pfaltz and Bauer and Aldrich Chemical Co. 4-Ethylacetanilideand 2,4,5-trimethylacetanilidewere synthesized as described below. Synthesis o f Acetanilides. An aqueous solution of 4ethylaniline (6.2 g), 0.42 M (103 mL), was warmed to 50 "C, acetic anhydride (6 mL) was added followed by 2.4 M aqueous sodium acetate (25 mL). The mixture was stirred and then cooled in tin ice bath until the product crystallized. The acetanilide was recrystallized twice from water/ethanol. 2,4,5-Trimethylacetanilide was prepared in the same manner and recrystallized twice from ether/ethanol. The mass spectra and infrared spectra of both compounds were consistent with the desired products. SRC-I1 Distillate. The SRC-I1distillate used throughout this study is a blend of middle and heavy distillates (2.9:l.O) produced at the SRC-I1 pilot plant at Fort Lewis, WA, operated by the Pittaburg & Midway Coal Mining Co. The blend is not necessarily representative of a future commercial product. An aqueous extract of this blend was prepared by mixing 1 part SRC-I1 blended distillate with 100 parts water (v/v) for 24 h. After allowing the SRC to settle, the aqueous layer was separated and passed through a glass wool plug. The aquatic toxicity of aqueous extracts of the SRC-I1blended distillate, prepared in a similar manner, have been studied extensively at the Pacific Northwest Laboratories (i?). Capillary Gas Chromatography (GC) and Capillary Gas Chromatography/Mass Spectrometry (GC/MS). Gas chromatographic analyses were performed on a Hewlett-Packard 5880A equipped with flame ionization (FID) and nitrogen/ phosphorus (N/R) detectors. A 60-m SP2100 capillary column (J&W Scientific) was used for all experiments with the following temperature program: initial oven temperature at 7 5 "C for 4 0 1982 American Chemical Society
