NOTES
4077
and 0.025 atm-l cm-', respectively.' The extinction coefficient of benzene was determined to be 707 atm-' cm-I for the 1849-A line as emitted by our lamp. All substances were deoxygenated prior to photolysis. The temperature was 27 f 2". Photolysis of both 1 and 2 torr of benzene gave an HZ quantum yield of 2.5 f 0.5 X This value is in good agreement with a value of 3 X reported in the earlier work over the pressure range from 200 to 760 torr.s I n Table I are presented the results of all experiments with added hydrocarbons. Table I: Quantum Yields of Hz and H from Benzene-Hydrocarbon Mixtures at 1849 A
and Teller" that H atoms may arise from benzene owing to absorption into a repulsive state underlying the T-T* El, + A1, transition. However, the major benzene disappearance route appears to be predissociative. (7) For an extremely wide range of olefins, Jones and Taylor ( A d . Chem., 27, 228 (1958)) report extinction coefficients a t 1850 A of ca. 7500-10,000 l./mole cm. If we therefore assume that all of the alkane optical absorption is due to olefin, we obtain upper bounds to the olefin concentration of ca. 0.001% in cyclohexane and isooctane and of ca. 0 . 0 0 5 ~ 0in propane. With the exception of one of our measurements at 0.0025% benzene, these olefin levels are considered sufficiently low to be neglected. (8) K. Yang, J. Am. Chem. SOC.,84, 3795 (1962). (9) H. Schiff and E. Steacie, Can, J. Chem., 29, 1 (1951). (10) D. F. DeTar and R. A. J. Long, J. Am. Chem. SOC.,80, 4742 (1958). (11) G. Nordheim, H. Sponer, and (1940).
E. Teller, J. Chem. Phys., 8 , 455
Partial Alkane
CsHe
4(Hz) bO(H2)
550 torr of propane 500 torr of propane 10 torr of propane 4 torr of propane 40 torr of isooctane 100 torr of cyclohexane Liquid cyclohexane Liquid cyclohexane
0.027 0.25 2.0 5.0 2.5 1.0 0.0025 0.025
0,010 0.015 0.005 0.002 0.003 0.007 0.006 0.006
vol.
%
4(H)
0.010 0.020 0.015
0.010 0.006
An Interpretation of the Concentration Dependence of Mobilities in Fused Alkali Carbonate Mixtures
0.009
0.006 0.006
The H atom quantum yields presented in Table I for benzene-propane mixtures were calculated with the value of k3/k4 = 100 as determined by Yang.* For both cyclohexane and isooctane, a value of k3/k4 = 33 was used based on a ratio of collision yields for H atom abstraction from propane and cyclohexane ofg = 3. ~~(CGHIZ)/JCI(CQHB) The low value of ca. 0.01-0.02 for +(H) indicates clearly that processes other than C-H bond rupture are mainly responsible for benzene disappearance at 1849 A. This is further confirmed by the fact that no biphenyl has been detected in the vapor-phase photolysis,2,3whereas it has been demonstrated, at least in solution, that reaction of phenyl radicals with benzene produces biphenyl and dihydrobiphenyl as major products. l o Owing to some H2 production from polymer buildup on the photolysis window (and this mas kept minimal by virtue of the low conversions), we feel that little significance can be placed upon the variation between individual yields in Table I. Within our uncertainties, therefore, there does not appear to be evidence for any important effect of total gas pressure on +(H). This contrasts markedly with a very high sensitivity to ) ~ tends foreign gas pressure exhibited by +( - C ~ H O and to support an early suggestion of Nordheim, Sponer,
by R. Mills and P. L. Spedding Diffusion Research Unit, Research School of Physical Sciences, Australian National University, Canberra, Australia (Received May 31, 1966)
I n a recently reported study from this laboratory' which was concerned primarily with the temperature dependence of tracer diffusion in alkali metal carbonates and their eutectic mixtures, we observed that diffusion coefficients at the eutectic compositions were considerably higher than in the component pure salts. With a view to exploring this behavior further, we have now made a more detailed study of the concentration and temperature dependence of the tracer-diff usion coefficients of Na+ and GO3*- ions in Li~C03-NazCOs mixtures. The experimental techniques used in this study have been described fully elsewhere.lv2 The data are tabulated in abbreviated form in Table I and for completeness we have included also the preliminary data obtained previously. 1;2 With these data we have calculated sets of isothermal diffusion coefficients and graphed them against composition as shown in Figure 1. Two not.able features of the graph which invite interpretation are the marked
~~
(1) P. L. Spedding and R. Mills, J . Electrochem. Soc., 113, 599 (1966).
(2) P. L. Spedding and R. Mills, ibid., 112, 594 (1965).
Volume 70, Number Id
December 1966
NOTES
4078
~~
~~
Table I: Tracer Diffusion in LipCOa-Nt&Oa Mixtures Trace
Temp range,
species
OC
LipC03
Na+
75 :25 mole % Li2COa-NasCOa 53.3:46.7 mole % L~zCOJ-N~ZCOS 25: 75 mole LizCOrN&COa Na2COa
Na + COS’ Na + Cos2Na+ COaZNa+ COaz-
809-905 849-991 757-938 757-938 580-863 569-842 778-932 778-932 910-1043 901-1062
Medium
’
cos*-
O
COMPOSITION
C
MOLE
%
Figure 1. Tracer-diffusion coefficient us. composition isotherms for labeled Na+ and CosP-in Li2COs-NazCOa mixtures.
positive deviations from linearity for each set of coefficients and the apparent maxima at the eutectic composition. It has been found that when the equivalent conductances of LizCOrNa&03 mixtures are graphed against composition, the curves show negative deviations from linearity. 3,4 Moynihan and Laity5 have interpreted similar deviations for the LiC1-KC1 system by invoking polarization effects. They implied also that in diffusion the presence of associated ions or groups of ions would tend to nullify the factors affecting the conductance and produce a more nearly linear dependence of the tracer-diffusion coefficients with composition. Instead, as Figure 1 shows, there is a large positive deviation for the two species investigated. The Journal of Physical Chemistry
D* X 109,cml/sec
(1.32 f 0.44) exp[-(9630 f 370)lRTI (1.35 f 0.06) exp[-(9740 f 40)/RT] (4.42 f 0.15) exp[-(10,060 f 90)/RT] (3.57 f 0.33) exp[-(10,920 f lOO)/RT] (9.81 f 0.14) exp[-(10,990 f 140)/RT] (7.36 f 0.02) exp[-(11,560 f 120)/RT] (9.40 f 2.08) exp[-(ll,glO f 240)/RT] (4.26 f 0.33) exp(-(ll,l30 f 90)/RT] (10.0 f 0.05) exp[-(12,170 f 130)/RT] (2.86 f 0.10) exp[-(10,620 f 180)/RT]
The Nernst-Einstein equation is often used in molten salts to compare conductance and diffusional mobilities. The substantial deviations from this equation which usually result can be interpreted in at least two ways. Thus, on the one hand, AngelP has presented evidence to support the contention that mutual ionic interference causes the breakdown of the equation in these systems. On the other hand, Bockris and Hooper’ have interpreted these deviations as being due to a paired vacancy process which allows diffusion to proceed by two mechanisms. More recently, Lantelme and Chemlas in their extensive work on the alkali metal nitrates have used the latter general approach and attributed their deviations to the presence of free ions and polyionic groups9 in the melts, so allowing differing conduction and diffusion modes. They were able to give quantitative expression to this concept and to calculate the percentages of free ions and associated species in their melts, the former usually being quite appreciable (>30%.) We have not been able to apply the Nernst-Einstein equation to the Li2C03-Na&03 systems as neither the tracer diffusion of Li+ ion nor transport numbers for the ions have been measured. However, the deviations for the pure Na&03 melt have been calculated and shown to be appreciable in the temperature range of this study.l If the presence of free ions and ionic aggregates is assumed in the Li2C03-i“\:a&03system, then (3) G. V. Vorobev, S. V. Karpachev, and S. F. Palguev, AEC-tr5948, 1963, p 167. (4) P. L. Spedding, unpublished work. (5) C. T. Moynihan and R. W. Laity, J. Phys. Chem., 68, 3312 (1964). (6) C. A. Angell, ibid., 69, 399 (1965). (7) J. 0. Bockris and G . W. Hooper, Discussions Faraday SOC.,3 2 , 218 (1962). (8) F. Lantelme and M. Chemla, Electrochim. Acta, 10, 663 (1965). (9) By “free ions” are meant single ions such as N a + or COaZ-. The polyionic groups include ion pairs of the type [M+C03*-]-, neutral groups such as MzC03, and possibly larger aggregates.
NOTES
4079
from consideration of all the data now available we put forward the following suggestions regarding transport in these melts. The ideas which we outline give a feasible qualitative explanation for the behavior of these data which is otherwise rather puzzling. We postulate that conduction proceeds predominantly by movements of free ions and tracer diffusion predominantly by groups of ions either neutral or charged. A corollary of this is that the concentration of free ions is very low since diffusion can proceed by both species. This first postulate asserting that essentially separate species are involved in conduction and diffusion is in accord with most of the known facts. Thus, we have shown previously’ that the activation energy for tracer diffusion for several ions in the alkali carbonates is about 11 kcal, whereas that for conductance is in the range 6-8 kcal. This can be explained in terms of the large activation energy needed for an aggregate of ions to diffuse and the small activation energy needed for free ions to migrate. Again, the equivalence of cation and anion activation energies in diffusion is strong evidence for an associated act such as aggregate movement. One can also explain why the equivalent conductance of Li2C03 is much higher than that of Nad203, whereas tracer-diffusion mobilities are in the reverse order, being higher in pure Na2C03. With the small size of the cation and its associated high polarizing power, Li2C03can be visualized as a structured meltlo with an essentially anionic lattice which would allow Li+ ion migration but inhibit aggregate movement. We next postulate that if the above picture of the two tranmort mechanisms is acceDted, the positive deviations for diffusion shown in Figure 1 Can be explained by some kind of structural breakdown as N h coois progressively added to the L ~ ~ Cmelt. O ~ hi^ loosening of the structure can be attributed basically to the differing sizes of the cations which have the effect Of providing asymmetry either by Of differing polarizing power or from purely geometrical considerations. Structural breakdown is also reflected in the negative deviation of the viscosity for many binary salt systems. In any event, the ion aggregates are able to diffuse faster in the looser melt reaching an apparent maximum at the eutectic composition. By analogy with other electrolyte systems,” the migration of free ions in the conduction process would be much less affected by structural changes. There is some controversy about t,he nature of the entities in melts at their eutectic composition, and this has recently been discussed by Antipin.12 In particular, the behavior of eutectic mixtures under pressure points to the fact that they should be regarded as the most loosely packed combinations of the *
ions or atoms which constitute them. It is also noteworthy that Jam and Saegusals found that the activation energy for viscous flow in a ternary eutectic of the alkali carbonates was 10 kcal, whereas that for the pure salts was of the order of 25 kcal. To explain the diffusion maxima one might postulate further, then, that there is maximal structural looseness at the eutectic composition. The measurement of the tracer-diffusion coefficients of Li+ ion in the above system is of course necessary to confirm the picture here presented. At the moment we do not have the facilities for mass spectrometric analysis. If our assumptions are correct, the lithium ion coefficients should show a positive deviation of the same type as Na+ and COS2- ions. These data would also allow a comparison of conductance and diffusion mobilities via the Nernst-Einstein equation and the extent of the deviations should then permit a quantitative test of the above ideas. Finally, it should be remarked that the positive deviations in the tracer-diffusion data appear to be most marked when the cations in the binary mixtures are fairly different in size, when the lithium ion is present, and when the anion is highly polarizable. Thus studies in the alkali nitrates show some evidence of the effects described but are not very definite. Similarly, our data for the K2C03-NaC03 systemJ2 though incomplete, indicate a more nearly linear relationship.
Acknowledgment. We wish to thank Dr. C. A. Angel1 for helpful discussions on this subject.
I
(10) The use of the term “structure” here and below should perhaps be qualified. Many models of molten salts assume the existence of interpenetrating anion and cation lattices which do not have the long-range order of the crystalline state yet are sufficiently real to the extent that each ion has, on the average, more nearest neighbors of opposite charge than of its own charge. This departure from a purely random state is what we imply by the term structure. (11) R. H. Stokes and R. Mills, “Viscosity of Electrolyte Solutions,” Pergamon Press Ltd., London, 1965, P 56. (12) L- N. Antipin, AECtr-594% 19639 Pa 123. (13) G. J. Janz and F. Saegusa, J. Electrochem. Soc., 110,452 (1963).
Deactivation in the Photolysis of Hexafluoroacetone at Low Pressure
by Gerald B. Porter and Kengo Uchida’ Department of Chemistry, University of British Columbia, Vancouver, Canada (Received June 13, 1966)
I n the gas phase, the dissociation of the vibrationally excited molecules in an upper electronic state Volume YO. Number 1.9 December 1966