J . Phys. Chem. 1991, 95. 3432-3435
3432
the flow rate than that in the latter mode. 5 . Finally we comment on the new bistability between the thermodynamic and oscillatory branch and the resulting mushroom structure15 (Figure 2c,d) found in mixing mode (A + C) B. While the limits of the original hysteresis a t high flow rate show the customary sensitive dependence on stirring, the new dynamical structure at low flow is strikingly insensitive to stirring. This is true for the oscillation amplitude and period, but the upper limit of the oscillations shows a slight S dependence. This robustness under stirring probably means that the structure in question arises primarily from concentration changes due to the prereaction rather than from heterogeneity and micromixing effects. A similar mushroom structure is predicted1ssfrom simulations of the autocatalator model and has subsequently been found experimentally and computationallyls,bin the iodate/arsenous acid reaction.
+
Conclusions We have shown that the dynamics of the BrOf/Br-/Ce2+ system shows great sensitivity toward the mode of reactant mixing as well as to the rate of stirring. This system provides a striking example that reactant premixing and enhanced stirring are generally not equivalent and that the dynamical responses can be mutually opposed. We have explained qualitatively the SM effects on the bistability transitions with the help of the NFT mechanism. The sensitivity of the experiment to the details of SM raises a number of related issues, the first being that of the wellantrolled and reproducible experiment. The difference between the (A + (B C)) results (Figure lb) and the 'old premixed" results4 illustrate the difficulty of exactly duplicating the hydrodynamic and kinetic parameters of mixer, stirrer, and reactor, particularly when different setups are used. It was observed that apart from
+
the gross aspects of SM described here, the dynamics can depend sensitively on the point of injection of the feedstreams into the reactor and on the contact time between premixing and entry into the CSTR. Indeed it is doubtful whether the premixed feedstreams are homogeneous on the molecular scale and whether they are relaxed chemically, given a transit time of the order of 0.5 s, while it is knowne that even in the turbulent environment of the CSTR, the time for perfect micromixing is typically of the order of seconds. More reproducible, standard experiments are clearly desirable and feasible, based on more efficient stirrer design utilizing high-speed impellers and baffles9 to maximize turbulence and on more refined premixers. However, it should be kept in mind first that the perfectly micromixed CSTR will probably remain an idealization attainable only for asymptotically high turbulence, and second that nonlinear dynamical phenomena occurring in nature, e.g., in biosystems, are likely to occur under nonideal inhomogeneous conditions. It is important to distinguish between the robust dynamical phenomena due to the ideally homogeneous medium, and to tailor the reaction mechanism accordingly, and between those heterogeneity-induced dynamical structuresZl0 that may be driven even by small deviations from ideality. Appropriate theoretical descriptions and computational algorithms that go beyond the ideal limit have been described." In light of the present results, the good agreement between the early experimentd4 and homogeneous kinetic simulations13 is probably fortuitous. A more realistic simulation of the present experiments is being undertaken.12 Acknowledgment. This work is supported by the NSERC of Canada.
Liquid-Liquid Partition and Hydration of Cobalt( I I I ) Acetyiacetonate and Cobalt( I I I) Monothioacetyiacetonate Jerzy Narbutt Department of Radiochemistry, Institute of Nuclear Chemistry and Technology,03- 195 Warsaw, Poland (Received: May 7 , 1990; In Final Form: October 22, 1990)
Partition coefficients of tris(acetylacetonato)cobalt(III) and tris(monothioacetylacetonato)cobalt(III) in the system waterln-heptane were determined over the temperature range from 5 to 45 OC, and standard thermodynamic functions of .transfer have been computed. Salting-out coefftcients of the chelates in NaCIO, solutions were calculated. The much lower partition coefficients for Co(acac), than those for Co(Sacac)3, as well as higher standard enthalpy and entropy of transfer and salting-out coefficient for the former chelate, are explained in terms of its stronger outer-sphere hydration in the aqueous phase, in accordance with greater tendency to form hydrogen bonds by oxygen than by sulfur atoms.
Introduction Hydration of coordinatively saturated metal &diketonates in solution has been established by numerous authors using various techniques,'-" but no consistent model has been generally accepted (1) Hopkins. P. D.; Douglas, B. E. Inorg. Chem. 1964, 3, 357. (2) Davis, T. S.;Fackler, Jr., J. P. Inorg. Chem. 1966, 5, 242. (3) Allard, B.; Johnson, S.; Narbutt, J.; Lundqvist, R. Prm. Inr. Solvent Extr. Conf. ISEC'7R Can. Inst. Merall. Spec. Vol. 1979, 21, 150. (4) Narbutt, J. J. Inorg. Nucl. Chem. 1981, 43, 3343. (5) Yamamoto, M. J . Phys. Chem. 1984,88, 3356. ( 6 ) Imura, H.; Suzuki, N . Talanra 1985, 32. 785. (7) Ycshimura, Y.;Sato, N.; Kikuchi, hi. Bull. Chem. Soc. Jpn. 1986,59, 2135.
0022-365419 112095-3432302.50/0
till now. The author suggested4 that chelate complexes form hydrogen bonds with water molecules in aqueous solutions, and that the donor oxygen atoms from coordinating ligands make hydrophilic centres in the chelate molecules. The same conclusion was then drawn by Imura and Suzuki.6 Such specific outer-sphere hydration of coordinatively saturated chelates in aqueous solutions is responsible for much higher values of standard thermodynamic functions of transfer of metal chelates from water to organic solvents as compared with the functions for homomorphous hy(8) Darus, H. B.; Meloan, C. B. Solvent Exrr. Ion Exch. 1986, 4, 495. (9) Saitoh, K.; Tsukahara, S.; Suzuki, N. Anal. Lett. 1988, 21, 599. (IO) Narbutt, J. Prm. Int. Solvenr Exrr. Conf. ISEC'88 1988, I , 104. (11) Moore, P.; Narbutt. J. J . Solution Chem., in press.
0 1991 American Chemical Society
Partition Coefficients of Metal Diketonates drocarbons. The reason is that hydrogen bonds between hydrophilic groups (heteroatoms) of the ligands and water molecules in the aqueous phase are broken in the transfer process. Further evidence for outer-sphere hydration of metal chelates can be obtained from studying the effect of replacing at least one oxygen atom in a chelating ligand, e.g., acetylacetone, by a sulfur atom. When compared to oxygen, sulfur has a lower electronegativity and a weaker tendency to form hydrogen bonds, which appears, among others, in much lower solubility in water of thioethers than homomorphous ethers. Taking that into account, one can expect higher and less temperature dependent partition coefficient for the thioacetylacetonate. However, the corresponding changes in the entropy of transfer and changes in the salting-out coefficients for the chelates are more difficult to predict because of anomalous contributions to these quantities from hydrophilic groups.5J2 Although monothio-p-diketones are widely used as metal extractants for analytical purposes,13 no data on the liquid-liquid partition (transfer) of their metal chelates, except for those of nickel and zinc,I4 have been reported. The reason is that aliphatic monothio-p-diketones and their metal chelates are unstable under conditions of transfer (partition equilibri~m)’~,’~ and that the partition coefficients of such chelates are very high. Moreover, the data reported14 can hardly serve our purpose, because acetylacetonates of nickel and zinc are coordinatively unsaturated, so that the expected effect may be obscured by changes in the inner-sphere hydration of the central metal ion. Therefore, to provide further arguments in favor of the model of outer-sphere hydration, the thermodynamics of transfer and salting out of cobalt( 111) chelates with acetylacetone and monothioacetylacetone were studied in this work. Trivalent cobalt was chosen as the central metal ion because it forms stable, coordinatively saturated chelates with both ligands. Special attention was paid to accurate determination of partition coefficients.
Experimental Section Acetylacetone (2,4-pentanedione; acacH) (Fluka, pure) was purified by shaking with carbonate buffer solution, then dried, and distilled. Monothioacetylacetone (4-thioxo-2-pentanone; SacacH) was synthesized in a low-temperature reaction of hydrogen sulfide with acetylacetone in the presence of hydrogen chloride, followed by purification via lead(I1) chelate.I6 The product obtained (bp 40 OC at 4-5 mmHg) was then transformed into sodium salt” in order to increase its stability during prolonged storing. n-Heptane (Reachim (USSR),standard) was used as supplied. Doubly distilled water was used throughout. The radionuclide purity of radiotracer W o was checked by high-resolution yspectrometry. Tris(acetylacetonato)cobalt(III) labeled with 6oco was prepared by the reaction of cobalt(I1) acetate with acetylacetone in the presence of hydrogen peroxide, according to a slightly modified standard procedure,’*then recrystallized from benzene, washed with n-heptane, and vacuum dried. Tris(monothioacety1acetonato)cobalt(III) labeled with was precipitated from aerated aqueous solutions of cobalt(I1) acetate and sodium monothioacetylacetonate, recrystallized from acetone and vacuum dried.I9 The specific activities of the prepared chelates were 11 kBq/mg and 1.1 MBq/mg, respectively, and their melting points (nonlabeled samples) were 21 1 OC (lit. 213 OC18) and 191 “C (lit. 180-190 oC1e2’), respectively. (12) Wilcox, F. L.;Schrier, E. E. J. fhys. Chem. 1971, 75, 3757. (13) Uhlemann, E.;Morgenstern, R. Z . Chem. 1977, 17, 405.
(14) Leban, M.: Jeffries, D.;Fresco, J. Can. J . Chem. 1979, 57, 3190. (15) Leban, M.;Fresco, J.; Livingstone, S.E. Aus?. J . Chem. 1974, 27, 2353. (16) Chaston, S.H.H.;et al. Aus?. J . Chem. 1965, 18, 673. (17) Siimann, 0.;Fresco, J. J . Chcm. Phys. 1971.54, 734. (18) Bryant, B. E.; Fernelius, W. C. Inorg. Synth. 1957, 5, 188. (19) Yokoyama, A.; Kawanishi, S.;Tanaka, H. Chem. Pharm. Bull. 1970, 18, 356. (20) Yokoyama, A.; Kawanishi. S.; Chikuma, M.;Tanaka, H.Chem. Pharm. Bull. 1967, IS, 540.
The Journal of Physical Chemistry, Vol. 95, No. 8, 1991 3433
i -t
3.2
3.3
3.4
3.5
3.6
10OO/T, K-’ Figure 1. Temperature dependence of partition coefficients (P,,,)of Co(acac), (lower set) and Co(Sacac), (upper set) in the system waterln-heptane. For the sake of clarity, only the extreme experimental points for each T are presented.
Partition measurements were carried out with freshly prepared heptane solutions of the chelates. To avoid the decomposition of the chelates, both systems also contained the extractant, previously dissolved either in heptane (acacH) or in water (SacacNa) at the original concentration of 10” mol dm-,. Equal volumes of both phases were shaken for 20 min at a temperature measured with the accuracy of 0.1 OC and then separated. The aqueous phase containing Co(Sacac), was centrifuged. The specific activities of two or three samples of each phase were measured with the use of a well-type NaI(T1) detector. In the case of monothioacetylacetone which is easily oxidized,I5especially at pH lower than 7, two consecutive equilibrations of the same organic phase containing Co(Sacac), were carried out with freshly prepared aqueous phases, and the average P M value from both determinations was calculated, provided that the difference was less than 10%. The results were occasionally checked by carrying out the process under anaerobic conditions (glovebox with argon atmosphere and argon-saturated solutions). Transfer equilibria of the chelates from aqueous solutions of sodium perchlorate (0.5-3 M) to n-heptane were studied in the same way.
Results and Discussion The experimental conditions have been chosen so that the experimental values of distribution ratio (D)are equal to the partition coefficients (PM)of the chelates, defined as the ratio of molar concentration (or specific activity) of the chelate in the organic phase to that in the aqueous phase. That was achieved by working in the region where PM is independent of pH and extractant Concentration: i.e., at pH from 4 to 7 for Co(acac), and 8 to 9 for Co(Sacac), with the concentration of the extractants of about lo-’ mol dm-,. The equilibrium was reached after about 5 min shaking, and then the D values remained constant for further 30-40 min. However, with no extractant added to the system, a slow dissociation of the chelates resulted in a noticeable decrease of the distribution ratio. The PM values did not depend on the original chelate concentration which varied from 6 X lo4 to 8 X lo4 mol dm-’ for Co(acac), and from 2 X to 6 X lo-’ mol dm-’ for Co(Sacac),. Total concentration of electrolytes in the aqueous phase was usually less than 1 X lo-, mol dm-’ (except for the experiments on salting out). Such low concentration does not affect the properties of water in respect to the thermodynamics of chelate transfer: and therefore the aqueous phase can be considered as pure water. Figure 1 shows the partition coefficients of the chelates, plotted in the form of van’t Hoffs relation. The parameters of the best fit parabola^,^ log PM = a T 2 + b T 1 + c, fitted by means of least-squares regression to 54 experimental data points for Co(acac), and to 37 ones for Co(Sacac),, were found to be a = (21) Chaston, S. H.H.; Livingstone, S.E. Aus?. J . Chem. 1967, 20, 1065.
3434 The Journal of Physical Chemistry, Vol. 95, No. 8, 1991
Narbutt
TABLE I: Partition Coefficientso of Cobalt(II1) Chelates and the Relevant Standard Tbermodynrmic FuacHolrp of Transfer from Water to n-Heptane chelate temp, OC PM AGO, kJ mol-’ AHo, kJ mol-’ ASo,J mol-’ K-’ Co(acac), IO 0.024 f 0.001 8.77 f 0.03 61.3 f 1.7 185 f 6 25 0.081 f 0.002 6.22 f 0.05 52.7 f 0.6 156 f 2 40 0.209 f 0.003 4.08 f 0.04 44.9 f 2.1 130 f 7 Co(Sacac), 10 72 f 2 -10.07 f 0.05 29.0 f 2.4 138 f 8 25 122 f 3 -1 1.90 f 0.06 20.0 f 0.8 107 f 3 40 166 f 4 -13.31 f 0.06 11.9 f 2.5 81 1 8 O P M = [chelate], /[chelate],,, where expressions in brackets denote molar concentrations of the chelate in the organic and aqueous phase, respectively. Standardstates: ideal dilute solutions; based on molarities. The error limits denote two standard deviations.
-1.262 X IO6 and -1.317 X IO6 K2, b = 5.713 X IO’ and 7.786 X IO3 K,c = -6.0566 and -9.218, and the correlation coefficients were equal to 0.999 and 0.989, respectively.
After adopting the ideal dilute solution of the chelate in each phase as the reference solution and taking into account that the PMare independent of chelate concentration, the standard free energy of transfer from water to n-heptane can be calculated as AGO = -RT In PM. The standard enthalpy of transfer can be determined as AHo = -R[d In P M / d ( l / n ] and , the standard entropy as ASo = (I/T)(AHo - A G O ) . The values of thermodynamic functions of transfer computed4 for various temperatures, are listed in Table I. No correction for the difference of thermal expansion coefficients of both phases was made.4 It is worth noting that the P,,,value of Co(acac), at 25 OC is based on about 10-2096 lower than the literature spectrophotometric measurements. As expected, the PMvalues for Co(Sacac), are much higher (3 orders of magnitude) and less temperature dependent than those for C ~ ( a c a c ) ~In. principle, interactions in both organic and aqueous phase can be responsible for this difference, but it can easily be proved that the role of the organic phase is less important. It has been pointed outU that solutions of coordinatively saturated metal &diketonates in hydrocarbons can be considered as regular, Le., having the ideal entropy of mixing. Therefore, the only difference between the two chelates with respect to the organic phase may manifest itself in the enthalpy of mixing. However, because of a close similarity of the molecular structure and of the identical hydrocarbon residues of both chelates, the difference is not significant. Accordingly, the experimentally observed differences in the thermodynamic functions of transfer must result from differences in interactions of the chelates in the aqueous phase. Two disparate regions, differing from each other with respect to hydration in aqueous solutions, can be distinguished in the molecules of the chelates studied: the hydrophilic center composed of six donor atoms from the ligands, and the hydrophobic region composed of hydrocarbon radicals. The hydrophobic part of the solute promotes the hydrogen-bonded structure of liquid water in its vicinity.” Such hydrophobic hydration of Co(acac)’ in aqueous solution was evidenced by Yoshimura et al.’ Because the hydrocarbon radicals in both cobalt chelates studied are identical, one can reasonably assume that their contributions to the thermodynamic functions of transfer are almost the same for both chelates. Thus the only difference with respect to intermolecular interactions is due to the hydrophilic centers of the chelates and to their different degrees of hydration. Much higher values of thermodynamic functions of transfer for Co(acac)’ than those for Co(Sacac)3 must be attributed to stronger specific (hydrophilic) hydration of the former in the aqueous phase. In fact, the breaking of more hydrogen bonds, accompanied by liberation of more water molecules which originally hydrated the chelate molecule in the aqueous phase, is a process more endothermic and involving a more positive entropy change. As this is the case of hydrogen bonding, this large positive enthalpy of dehydration is much higher than the respective entropy contri(22) Awano, H.; Wstarai, H.; Suzuki, N. J. Inorg. Nucl. Chem. 1979,41, 124. ( 2 3 ) Koshimura, H. J. Inorg. Nucl. Chem. 1978, 40, 865. (24) Franks, F. Wate-A Comprehenriue Treatise; Plenum Press: New York. 1975; Vol. 4. Chapter 1.
3
2
1
0
C,
Relationships between partition coefficients (PM)of Co(acac)3 (lower set) and Co(Sacac), (upper set) and molar concentration (C)of NaC10, in the aqueous phase at 25 OC. The points denote the mean log PM value for each C, together with their 95% confidence intervals. F-2.
bution, and therefore the free energy of transfer also increases-the more the stronger the hydration is. The differences in standard thermodynamic functions of transfer of Co(Sacac)’ and C ~ ( a c a cmay ) ~ be attributed to the differences in hydrating three oxygen and three sulfur atoms in the hydrophilic centers of both chelates, i.e., in the formation of water adducts to the given atoms in the chelates dissolved in the aqueous phase. Assuming that the hydration of sulfur is negligible when compared to that of oxygen, and that the effects of hydration of two oxygen atoms in one acetylacetone ligand are additive, one can evaluate standard thermodynamic functions of the outer-sphere hydration of C o ( a ~ a c as ) ~ twice the respective differences: AGho = -36.2 kJ mol-’, AHh”= -65.4 kJ mol-’, Mho= -98 J mol-’ K-I. These large values show that a number of water molecules are engaged in hydrogen bonding to the solute molecule in the aqueous phase. According to Zolotov et al.,2Sthe enhanced extraction of zinc chelates with ligands containing sulfur donor atoms, when compared to the extraction of chelates with ligands containing oxygen donor atoms only, is due to much weaker hydration of the former chelates in the inner coordination sphere of zinc. This is caused by the lower coordination number of zinc in sulfur-containing chelates. The present work points to another, more general reason of the enhanced extraction of metal complexes with sulfur-containing ligands, i.e., to the decrease of their outer-sphere hydration in the aqueous phase. The stronger hydration of Co(acac)’ than of C ~ ( S a c a c is ) ~also manifested in salting out the chelates from aqueous solutions of an inert electrolyte. Figure 2 shows that linear relationships exist for both chelates between the logarithm of partition coefficient and the molar concentration of NaCI04 in the aqueous phase at 25 OC, log P M = k,C + const known as Setschenow’s equation.26 The straight lines were fitted to 30 experimental data points for Co(acac), and to 51 points for Co(Sacac),, with the correlation coefficients of 0.999 and 0.989, ~
~
~
~~
~
~
(25) Zolotov, Yu,A.; Petrukhin, 0. M.;Gavrilova, L.G. J . Inorg. Nucl. Chem. 1970, 32, 1679. (26) Long, F. A.; McDcvit, W.F. Chem. Reu. 1952, S I , 119.
J. Phys. Chem. 1991,95, 3435-3437
and Co(Sacac), are not only the evidence for a stronger outersphere hydration of the former chelate, but they also make it possible to estimate the magnitude of the hydration. It may be expected that similar contributions to thermodynamic functions of transfer and to salting coefficients should be observed in the case of extraction of other metal complexes with organic ligands containing sulfur as donor atoms.
respectively. The salting-out coefficients, k,, are equal to slopes, 0.284 k 0.003 and 0.214 f 0.005 dm3 mol-', respectively. The former value is the same as that for Cr(acac)3,'o and it is nearly equal to 1.5 times the value for Be(a~ac)~.'It may be assumed that each acetylacetonate ligand in the molecule of coordinatively saturated chelate contributes the same share into the total salting out of the molecule, independently of the central metal ion. The greater salting-out effect exerted on Co(acac), than on Co(Sacac), reflects stronger outer-sphere hydration of the former chelate. The difference, Ak, = 0.070dm3 mol-', is comparable with the respective difference for acetylacetonates of zinc and beryllium,'+27Ak, = 0.077 dm3 mol-'. The latter difference can be attributed to the inner-sphere hydration of Zn(acac)z, which engages a number of water molecules per chelate molecule.z7 Therefore, both the differences in the respective functions of transfer and the differences in salting-out coefficients for Co(acac), ~~~
~~
(27) Narbutt. J.
3435
Acknowledgment. The author is grateful to Professor S. Siekierski for his valuable comments and to Mrs. B. BartoS for synthesizing monothioacetylacetone. The work was supported from the Central Research Programme CPBP 01.09, and by a grant from the International Atomic Energy Agency, which is acknowledged. Financial support from the donors of the Petroleum Research Fund, administered by the American Chemical Society, is also appreciated. Registry No. acacH, 123-54-6; SacacH, 14660-20-9; Co, 7440-48-4; Co(acac),, 21679-46-9; Co(Sacac),, 15491-68-6; heptane, 142-82-5.
~~
To be published.
Gaseous Specles In the TI-AI-CI System and Reaction with H20 D. L. Hildenbrand,* K. H. Lau, SRI International, Menlo Park, California 94025
and S . V. R. Mastrangelo E. I. du Pont de Nemours & Company, Edge Moor, Delaware 19809 (Received: October 15, 1990)
Chemical species in the T i - A I 4 system at elevated temperatures were studied by effusion-beam mass spectrometry up to 1400 K. Gaseous TiAlC15 and TiAICls were identified as products of the reaction of TiCI,(g) and a Ti-AI mixture above about 900 K, for which AlC13, TiCl2,and TiC13 were also present. No evidence was found for TiAIC1,. When gaseous TiCI4, AIC13, and H20and admitted simultaneously, TiOClz, AI(OH)CIZ,and TiAI(OH)C15were observed. All ion species showed the correct isotopic distribution. The results confirm earlier indications that AICI, forms gaseous complexes with divalent and trivalent Ti. but not with tetravalent Ti.
Introduction The gaseous complexation of the titanium chlorides with AlCl, was studied by Sorlie and Oye' by means of absorption spectroscopy at temperatures of 500-900 K. Samples of TiCl,(s) and TiC12(s) were contained in sealed tubes with AICl, at pressures ranging from 0.3 to 4 atm; under these conditions the dimer AlZCb is the major A 1 4 species present. From measurements of the absorption spectra as a function of temperature, reaction stoichiometric coefficients were obtained and were used in identifying the gaseous complexes formed. Complexation was observed with gaseous TiClz and TiCl,, but not with TiCII. However, the derived stoichiometric coefficients were nonintegral, indicating the possibility of more than one complex for each Ti-Cl species. With TiCl,, the dominant gaseous complex was assumed to be TiA1Cl6, with a smaller amount of TiAl2Cl9,while, for TiClZ,the results indicated TiA13ClI was dominant, with a possible contribution from TiAl4ClI4. Our interest in the high-temperature chemistry of the Ti-A14 system and the products of reaction with H20led us to investigate this system by a more direct method, effusion-beam mass spectrometry. Use of a molecular effusion source necessarily limits the total pressure to the region below IW5 atm, but it does provide a direct sampling method that eliminates secondary reactions with surfaces and other gas molecules, thereby giving unambiguous information about vapor composition. In this work, the gaseous products generated by reaction of TiC14with a Ti-AI mixture and by direct reaction of TiC14, AlCl,, and H 2 0 were examined by (1)
Sorlie, M.;Oye. H. A. Inorg. Chem. 1978. 17, 2473. 0022-3654/91 /2095-3435$02.50/0
mass spectrometry, with the results reported below.
Experimental Section All measurements were made with the magnetic deflection mass spectrometer system and experimental techniques described in previous publications.2J For studies of the Ti-Al-Cl system alone, TiC14(g) was admitted to a graphite effusion cell containing pieces of Ti and AI wire. To provide inert conditions with gas mixtures containing HzO,a platinum effusion cell with two concentric gas-inlet tubes was utilized. In order to prevent prereactions in the inlet system, H 2 0 was added through the small central Pt inlet tube extending into the cell hot region, while TiC14and AlCl, were admitted through the annular space between the inlet tubes. The Pt cell contained pieces of Ti wire supported on a Pt screen, giving a less reducing environment than that of the Al/Ti sample in the graphite cell. Each reagent gas was added from a separate reservoir containing the solid or liquid, and the flows were controlled with precision leak valves. Ionization efficiency curves were recorded automatically,) and threshold appearance potentials were evaluated by the vanishing-current method. All reported ion intensities were subjected to the beam-defining slit test to eliminate potential background contributions. Temperatures were measured by optical pyrometry, sighting on a blackbody cavity in the cell lid; for the Pt cell, a graphite cover containing the cavity was placed over the cell body. (2) Hildenbrand, D. L. J . Chem. Phys. 1968,18, 3657; 1970, 52, 5751. (3) Hildenbrand, D. L. Inr. J . Muss. Spectrom. Ion Phys. 1970. 4, 75; 1971, 7, 255.
0 1991 American Chemical Society