Molecular shape effects on electron and ion behavior in dielectric liquids: cis- and trans-butene-2 between the freezing

and critical points

SHINICHI SASAKAKI'
AND

GORDON FREEMAN R.

Depnrtrnerzt of Chemistry, University of Alberta, Ednzorzton, AB T6G 2G2, Cnnndn Received December 1 3, 1993

SHINICHI SASAKAKI GORDON FREEMAN. J. Chem. 72, 1625 (1994). and R. Can.
Radiolysis free ion yields, electron mobilities, and ion mobilities were measured as functions of applied electric field strength in the liquids cis- and trans-butene-2 at temperatures from near their freezing points to near their critical points. The dielectric permittivities were also measured, to permit extraction of electron thermalization distances from free ion yield data. The thermalization distance distribution in each liquid was found to be the Gaussian YGP, rather than the exponential YE or YEP. This indicates that electrons are not captured before thermalization in either liquid. The density-normalized therrnalization distances bGpd were -40% larger in cis- than in trans-butene-2; this correlates with the more globular structure of the cis- molecules, which decreases the efficiency of electron energy loss collisions in the liquid phase. Electron mobilities were not appreciably different from those reported earlier. Cation mobilities in the liquids from the freezing points (TIT, % 0.35, d/dc % 3.2) up to -400 K (TIT, % 0.92, d/dc % 2.0) are adequately described by the free-volume model. At densities d/dc < 2.0 the mobilities are smaller than predicted by the free-volume model.

SHINICHI SASAKAKI et GORDON FREEMAN. J. Chem. 72, 1625 (1994). R. Can. OpCrant 5 des tempkratures allant de leurs point de congClation 5 leurs point critiques, on a mesuri les rendements en ions libres, les mobilitCs des Clectrons et les mobilitts des ions produits, en fonction de la force du champ Clectrique appliquC, lors de la radiolyse des cis- et trans-but-2-knes liquides. On a aussi mesurC les permittivitCs afin de pouvoir diterminer les distances de thermalisation des Clectrons a partir des donnCes des rendements en ions libres. On a trouvC que la distribution des distances de thermalisation dans chaque liquide prCsente une forme gaussienne YGP plutat qu'une exponentielle YE ou YEP. Ce resultat indique que, dans chacun des liquides, les Clectrons ne sont pas capturCs avant la thermalisation. Les distances de thermalisation normalisCes pour la densite, bGpd, sont environ 40% plus grandes dans le cis- que dans le transbut-2-kne; ce rCsultat est en bon accord avec la structure plus globulaire des molicules cis qui, en phase liquide, diminue l'efficaciti des pertes d'Cnergie par collision. Les mobilitis des Clectrons ne sont pas trks diffkrentes de celles rapportCes anttrieurement. Les rnobilitis des cations dans les liquides, des points de congClation (TIT, % 0,35, d/dc % 3,2) jusqu'; environ 400 K (TIT, % 0,92, dld, % 2,O) sont dCcrites d'une facon adCquate par le modkle sans volume. A des densitts d/dc inf6rieures 5 2,0, les mobilitis sont plus faibles que celles pridites par le modkle de volume-libre. [Traduit par la redaction]

Introduction The present work continues the study of distribution functions of secondary electron thermalization distances in Xirradiated liquids (1). The scattering of low energy electrons in a liquid is strongly dependent on the molecular structure (2). Comparison of electron behaviors in isomeric hydrocarbons is particularly useful, because physical properties such as molar volume and average polarizability are similar, so differences in electron scattering properties can be attributed to differences in molecular shape. Isomers of special interest are cis- and trans-butene-2. Previous measurements showed remarkable differences in the radiolytic free ion yields (3, 4) and in the mobilities of thermal electrons (3, 5) in these liquids. It appears that electrons are captured to form transient anions in the trans- isomer. However, the free ion yields were measured over a smaller temperature range than more recent studies of other liquids, so it was desirable to extend the temperature range for these butenes. In the present work, samples of cis- and trans-butene-2 were sufficiently purified that electrons generated in them had half lives up to 300 ps at low temperatures. Electron and ion mobilities were remeasured as functions of electric field strength at temperatures from the freezing point to near the critical region.
'Permanent address: Radiation Safety Control Center, National Laboratory for High Energy Physics (KEK), Oho 1-1, Tsukuba, Ibaraki, Japan 305.

Experimental
A. Sarnple preparation cis-Butene-2 (Phillips Research Grade, 99.94%) or trans-butene-2 (Phillips Research Grade, 99.75%) was transferred from the cylinder into a grease-free vacuum line, and further purified. Air was removed and the liquid butene was stirred in a flask with lithium aluminum hydride powder for three or more days, then distilled under vacuum onto sodium hydroxide pellets and let stand for three days. Finally, the butene was transferred into a bulb containing sodium-potassium alloy, stirred gently for more than one week, then let sit on the alloy until transfer to the conductance cell. Between successive reagent treatments the butene was degassed by trap-to-trap distillation, with the receiver held at 77 K while pumping on the downstream side.
8. Apparatus and experinzental inethod Two types of conductance cell were used. The high-pressure conductance cell (HPC) was like that shown in ref. 6, and the lowpressure conductance cell (LPC) like that in ref. 7, except for the geometry of the metal pin lead-throughs. The separation between the electrode plates, 3.23 mm in HPC and 2.60 mm in LPC, were measured directly using precision calipers. The effective area of the collector plate, 2.62 cm' in HPC and 8.00 cm' in LPC, was determined from capacitance measurement (8) using liquid benzene and air as standards. The cell capacitances obtained using these two standards gave results that agreed within 0.2%. Before filling with a liquid, the cell was heated to 600 K while evacuating to 10-5 Pa. The temperature of the sample was controlled to within 0.2 K. Details of the cooling (9) and heating (10) systems, and of temperature control and measurement (1 1, 12) have been reported. The voltage sources and data collection system are described in ref. 12.

CAN. J. CHEM. VOL. 72, 1994

FIG.1. Relative permittivities for cis-butene-2 (8).

of cis- (0) trans-butene-2 (O), measured as a function of temperature. V, Previous experimental values and

To generate ion-electron pairs in the cell, a 0.10 or 1.0 p s pulse of 320 fJ (2 MeV) electrons from a van de Graaff accelerator was stopped in a 3 mm thick gold target to produce X rays. For each pulse a relative dose monitor measured the charge deposited on the gold target by the electron beam. The dose monitor was calibrated by LiF thermoluminescence (13) for each sample. Ion mobilities were determined by time of flight (12). Electron mobilities were measured either by time of flight or by the electron conductance transient (ECT) method (14). C. Physical properties Static relative permittivities E measured by the procedure of ref. 8 are plotted in Fig. 1 and listed in Table 1. The present values for cis-butene-2 are 1.0-2.7% lower than those at the same temperature reported in ref. 8. Values of the liquid densities d of the 2-butenes reported in ref. 15a are only estimates; they do not vary sufficiently rapidly with T in the vicinity of the critical temperature Tc, and also the reported value of Tc for cis-butene-2 is 3 K lower than a more recent value (15b). Measured values of the density of liquid n-butane under its vapor pressure have recently been reported for the entire liquid range (15c). The liquids cis- and trans-butene-2 and n-butane have similar values of Tc (435.6, 428.6, 425.2 K) and dc (240, 236, 228 kg/m3) (15b), so densities of the two butenes were calculated from a plot of d/dc against TITc made from the 11-butane data (15c). Values of the densities are listed in Table 1. They are within 1% of those given in ref. 15a for cis-butene-2 at 226 < T/K < 390, and for trans- at 240 < T/K < 400. Viscosities q were taken from ref. 15a, but they are unreliable near Tc .

applied field E(VIm) is given by

where G,,, is the total initial yield of ionizations, $ is fraction : of the initial ionxlectron pairs that escape geminate neutralization at E = 0, f (E, E, y, T) is a function that describes the field dependence of escape, and F(y)dy is the fraction of secondary electrons that have thermalization distances between y and y + dy. The model and eq. [1] are discussed in detail in ref. 2 and more briefly in ref. 17. The gas phase value of G,,, = 4.2 ionizationsll6a.J was used (2, 18). The electron thermalization distance distribution function that fits data from most liquids is a three-dimensional Gaussian with a power tail, and is denoted YGP (2):

Results and discussion Free ion yields and electron thermalization distances The electric field dependences of the free ion yields in cisand trans-butene-2 at 296 K are shown in Fig. 2. Data obtained from the two cells HPC and LPC agree within 1%. The lines in Fig. 2 were calculated from an extension (2) of the Onsager model (16). The model permits extrapolation of the yields to zero applied field strength and provides information about the distribution of thermalization distances of the : secondary electrons. The free ion yield G (ion pairsll6a.J) at

where 0.96 is a normalization factor, YG = (4y2/.n'/2bip)exp (-y2/b6p), and bGp is the most probable value of y. The lines in Fig. 2 were calculated with eqs. [1] and [2], using bGp = 6.46 nm in cis- and 4.72 nm in trans-butene-2. The fit is good for both compounds. The measured yields at the lowest temperatures, near the freezing points, are shown in Fig. 3, and those at the highest temperatires, near the critical points, are shown in Fig. 4. Several types of thermalization distance distribution function have been used from time to time (2, 19-24). The commonest alternatives to the three-dimensional Gaussian are onedimensional exponentials. The simple exponential is designated YE:

where yo is the reaction distance of the electron and ion and is

SASAKAKI AND FREEMAN

TABLE Relative permittivities 1.

T (K)
ea

and densities d
ea

d ( k g ~ r n ~ ) ~ T (K)

en

d ( k g ~ r n ~ ) ~ T (K)

d (kg~rn')~

'Measured. b ~ e text. e Treezing point from ref. 156; by extrapolation. d~ritical temperature and density from ref. 156, p. 642, and E calculated from the Clausius-Mosotti equation and the value of the molar polarization calculated from the other data: P (10" m3/mol) = 20.7 for trans-, 22 for cis-.

0 1 0

E (MVIm)
FIG2. Free ion yields in cis- (open symbols) and trans-butene-2 (solid symbols) at 296 K as a function of electric field strength. Results a) I E obtained in the LPC (0, and HPC ( A , A ) are shown. The lines were calculated using distribution YGP and G, = 4.2, with G = 0.170 and 0.068 ion pair/l6aJ, and bGp= 6.46 and 4.72 nm in cis- and trans-butene-2, respectively.

taken as 1 nm, and bE is the attenuation coefficient. Another distribution, designated YEP, has a power tail added to YE for the outermost few percent of distances: [41 0.97 YE 0.97tYE + 0.74{bkp/(y
-

Y -YO Y - Yo

< 3.5 > 3.5

~ E P ~ E P

where 0.97 is the normalization factor and b ~ p replaces bE in YE. A Gaussian distribution of thermalization distances might arise from random scattering of electrons by the molecules to which they lose energy. An exponential distribution might arise from electron capture prior to thermalization (2, 21, 24). Values of the yields calculated using these three distribution functions are shown in Figs. 3 and 4. Only YGP gave values

CAN. J. CHEM.

VOL. 72,

1994

FIG.3. Free ion yields in cis- ( 0 , 140 K) and tmns-butene-2 ( a , 174 K) near the freezing points, as a function of electric field strength in the LPC. The lines were calculated using G,,, = 4.2 and: -, YGP, bGp = 5.26 nm for cis-, 3.60 nm for trans-; . . ., YEP, yo = 1.0 nm, b ~ p 3.52 nm for cis-, 2.26 nm for trans-: - . - ., YE, yo = 1.0 nm, bE = 4.40 nm for cis-, 3.14 nm for trans-. =

FIG. Free ion yields in cis- ( 0 , 434 K) and trans-butene-2 ( a , 4. 426 K) near the critical temperatures, as a function of electric field strength in the HPC. The lines were calculated using G,,, = 4.2 and: -, YGP, bGP= 15.5 nm for cis- and 11.0 nm for trans-; . .. , YEP, yo = 1.0 nm, bEp= 18.2 nm for cis- and 11.2 nm for trans-: - . - ., YE, yo = 1.0 nm, bE = 19.7 nm for cis- and 12.4 nm for trans-.

that agreed with the experimental results over the entire temperature range. This indicates that the slowing-down electrons are not permanently captured before becoming thermalized in either cis- or trans-butene-2 (24). The values of the most probable thermalization distances bGPand the free ion yields extrapolated to zero applied electric field strength, G;, are listed in Table 2. Yields reported in 1973 (4) for these liquids at temperatures 183-293 K were about 25% higher than the presentvalues, due to less accurate dosimetry. If the efficiencies of the energy-loss collisions of the thermalizing electrons remained constant as the density of the liquid was changed, then the thennalization distance bGp would be inversely proportional to the density; that is, the density-

normalized thermalization distance bGpd would remain constant. The values of bGpd in cis-butene-2 are about 40% larger than those in the trans-isomer (Fig. 5 ) , which means that electron energy-loss collisions are more effective in the latter. The lower efficiency in cis- correlates with the greater sphericity of the molecules (4, 25). The efficiency of the energy-loss collisions increases (bGpd decreases) with increasing liquid density, except at the highest densities in cis-butene-2, near the freezing point, where the efficiency decreases (bGpd increases) again (Fig. 5). The behavior in cis-butene-2 is similar to that in propane, while that in trans-butene-2 is similar to that in ethene or cyclopropane (25). Transient anion formation has been suggested to occur in the latter three liquids (5,25). Anion formation would be more favorable at the higher densities near the freezing points. The decrease of efficiency in cis-butene-2 or propane at the highest densities, as the crystalization temperature is approached, is attributed to a beginning of the ordering of the molecules in the liquid structure. The value of bGPdis larger in the ordered crystal than in the disordered liquid of the same compound (26). The decrease of electron energy loss efficiency as the density decreases below 700 kg/m3 is attributed to the looser packing of the liquid molecules, the resulting lower frequency of the intramolecular oscillations, and the reduced magnitude of the short-range electron-molecule interactions. The value of bGPd is determined mainly by inelastic collisions of the electrons at energies <O. 1 aJ (2). Electron mobilities Electron mobilities in both liquids are independent of electric field strength up to the maximum used, 3.1 MV/m. The values obtained in the present work were virtually the same as those reported earlier ( 3 , and nothing can yet be added to the discussion there. Ion mobilities Most of the samples contained a trace of impurity that scavenged some of the electrons. In those cases the ion conduc-

SASAKAKI AND FREEMAN

TABLE Free ion yields and electron thermalization distances 2.

T(K) GO, " bGph bcdC T(K)

GO, "

bGph

b~#=

"Yield at zero applied field strength (ion pairsll6aJ). %lost probable thermalization distance (nm), distribution YGP, based on G,,, mensity-normalized thermalization distance kg/m2).

= 4.2.

d/d,
1.26 1.68 2.10 2.52 2.94 3.36

5 r
o

300

500

700
I

d(ka/m3\
\ c l

tance transient (ICT) consisted of a spike at the beginning due to electron conduction, followed by a "ramp" that corresponded to ion conduction, followed by a lower ramp that corresponded to slower ions. The ratio of the two ion mobilities obtained from the ratio of the ramp durations was 1.5 f 0.1 at all temperatures in cis-, and for trans- it was 1.5 in the low temperature cell and 1.3 in the high temperature cell. Anion and cation mobilities in several alkanes and cycloalkanes under irradiation have been measured separately, and ratios p-/p+ = 1.7-2.2 were found (27, 28). The negative ions in these liquids resulted from electron capture by traces of impurities. We therefore assume that the slowest ions in the butenes were also cations; their mobilities are listed in Table 3. The ion mobilities were independent of electric field strength up to the maximum used, 3.1 MVIm. Arrhenius plots of p+ are slightly S-shaped (Fig. 6). The temperature coefficients Ep between about 200 and 300 K are 4.9 kJ/mol in cis- and 4.6 kJ/mol in trnns-; they increase at lower and higher temperatures. However, these temperature coefficients are relatively small, and densities of the liquids decrease greatly as the temperature is raised from near the freezing point to near the critical point. We therefore apply the free-volume model of ion transport (14): [5]

FIG 5. Density-normalized thermalization distances bGpd plotted against density in cis- (0)and trans-butene-2 ( 0 ) .

1630 TABLE Cation mobilities p,+ 3. T (K)

CL+

CAN. J. CHEM. VOL. 72. 1994

TABLE Parameters for free-volume model (eq. [5]) of 4. cation mobilities in cis- and trans-butene-2
CL+

m2/V.s)

T (K)

CL+

m2/V.s) cis-Butene-2

T (K)

m2/V.s) Parameter cis-

Value trans-

"T, is the temperature in the middle of the range used to estimate E,.

trans-, T(K)
428.6
I

426
I

399
I

363
I

296
I

174
I

cis-, T(K)
435.6 434 428
I

399
I

307
1

195 140
I
I

FIG. Density-normalized mobilities as functions of relative num7. ber density (n, is the number density in the critical fluid) in cis- ( 0 ) and trans-butene-2 ( 0 ) - - -, calculated from the free-volume model . for both cis- and trans-.

FIG. Arrhenius plots of cation mobilities in cis- ( 0 ) and trans6. ) ( The lines were calculated with eq. [5], using the pabutene-2 . rameters listed in Table 4.

where A+ is a constant, V is the molar volume, Vo is the molar volume at which the liquid molecules are too crowded for diffusion to occur, E+ is a residual exponential temperature coefficient (see below), and R is the gas constant. Equation [5] provided an adequate description of the ion mobilities in the liquids from their freezing points up to 400 K, above which the calculated mobilities were too high (Fig. 6). The parameter

values obtained by fitting eq. [5] to the experimental results up to 400 K are listed in Table 4. The values A+ = 9.9 and 7.4 (lop6 K1/' . m2/V-s) for radiolytic cations in cis- and trans-butene-2 are similar to those in the C6 to C l l n-alkanes, which are 8-9 (loa6 K1I2 . rn2/v.s) (14). The value in a quite different liquid, CS2, is also similar, 11 x lop6 K'/' . m2/v.s (14). The similarity of A+ values over such a wide range of nonpolar liquids (cis-butene-2 is slightly polar, with dipole moment 1.0 x C.m (156)) indicates that the cations are monomeric. The values of Vo correspond to 17.9 and 18.7 cm3/mol per carbon atom in the molecules of cis- and trans-butene-2, respectively. They are similar to those for radiolytic cation mobilities in a series of n-alkanes, which range from 18.5 in C6

SASAKAKI AND FREEMAN

1631

to 18.3 (cm3/mol carbon) in C I I (14). If the sulfurs in CS2 are taken as equivalent in volume to methyl groups, the value Vo/3 = 17. 1 (cm3/mol C or S) is obtained. This similarity of values again suggests that the cations are monomeric. The values of E+ are smaller than the thermal agitation energy RT (Table 4), so they are not typical "activation" energies. The value for trans-butene-2 is slightly negative (Table 4), which produces the curvature of the plot near 180 K in Fig. 6. The small exponential term can be thought of as a modifier of the Vo term; it corresponds to a slight increase of Vo with T in cis-, and to a slight decrease of Vo with increasing T in trans-. This might be related to the stick-like structure of the trans- molecules and the more globular structure of the cis-. The flexible C6 to C I I n-alkane molecules all give positive values of E+, with E+/RT,,, = 0.4-0.6; the relatively globular CS2 has E+/RT,,, = 0.3 (14). This commonality of values again indicates a simplicity in the diverse systems. At T >N 400 K (TITc >N 0.92, in the liquids under their vapor pressure) the free volume model breaks down, which means that liquid viscosity ceases to be the controlling factor for ion mobility (14). In the low density gas phase, the controlling factor for ion mobility is the collision frequency with the gas molecules. Thus the product of ion mobility and number density n of the gas tends to be constant in gases (29). The density-normalized mobilities np+ in the present liquids are plotted against the relative densities nln,, where n, is the number density of the particular critical fluid, in Fig. 7. The dashed line in Fig. 7 indicates values estimated from the freevolume model for both liquids. The model predicts mobilities that are too high at n/n, < 2.0, which corresponds to T > 400 K in the liquids under their vapor pressure. The values of np+ pass through a maximum in the low 1.7 (Fig. 7). This is consistent with density liquids at n/n, earlier reports that the transition from the mechanism of ion transport in normal liquids, which is governed by viscosity or free volume, to the transport mechanism in normal gases, which is governed by collision rate or number density, occurs in the relative density region 2.0 > n/n, > 0.4 f 0.1 (29-3 1).

Acknowledgements We thank Norman Gee for helpful discussions and a c ~ i t i cal reading of this manuscript, and the staff of the Radiation Research Center for providing and maintaining the accelerator and electronic circuitry. One of us (S.S.) is grateful to the Japanese National Laboratory for High Energy Physics (KEK) and the Japanese Ministry of Education, Science and Culture for financial support during this work at the University of Alberta.

I. T. Yoshinari, N. Gee, and G.R. Freeman. Can. J. Chem. 70,327 (1992). 2. G.R. Freeman. 111 Kinetics of nonhomogeneous processes. Edited by G.R. Freeman. Wiley-Interscience, New York. 1987. Chap. 2. 3. J.-P. Dodelet, K. Shinsaka, and G.R. Freeman. J. Chem. Phys. 59, 1293 (1973). 4. J.-P. Dodelet, K. Shinsaka, U. Kortsch, and G.R. Freeman. J. Chem. Phys. 59, 2376 (1973). 5. J.-P. Dodelet and G.R. Freeman. Can. J. Chem. 55, 2893 (1977). 6. J.-P. Dodelet and G.R. Freeman. Can. J. Chem. 55,2264 (1977). 7. T. Wada and G.R. Freeman. Phys. Rev. Lett. 42, 715 (1979). 8. N. Gee, K. Shinsaka, J.-P. Dodelet, and G.R. Freeman. J. Chem. Thermodyn. 18, 221 (1986). 9. M.A. Floriano, N. Gee, and G.R. Freeman. J. Chem. Phys. 84, 6799 (1986). 10. M.A. Floriano and G.R. Freeman. J. Chem. Phys. 85, 1603 (1986). 11. M.A. Floriano, N. Gee, and G.R. Freeman. Can. J. Chem. 66, 1872 (1988). 12. N. Gee, M.A. Floriano, and G.R. Freeman. Z. Naturforsch. A39, 1225 (1984). 13. N. Gee, P.C. Senanayake, and G.R. Freeman. J. Chem. Phys. 89, 3710 (1988). 14. N. Gee and G.R. Freeman. J. Chem. Phys. 90, 5399 (1989). 15. ( a ) R.W. Gallant. Physical properties of hydrocarbons. Vol. 1. Gulf Publishing Co. Houston, TX. 1968; (b) R.C. Reid, J.M. Prausnitz, and T.K. Sherwood. The properties of gases and liquids. 3rd ed. McGraw-Hill, New York. 1977; ( c ) B.A. Younglove and J.F. Ely. J. Phys. Chem. Ref. Data, 16, 758 (1987). 16. L. Onsager. Phys. Rev. 54, 554 (1938). 17. G. Ramanan, N. Gee, and G.R. Freeman. Can. J. Phys. 68, 925 (1990). 18. R.M. Leblanc and J.A. Herman. J. Chem. Phys. 63, 1055 (1966). 19. G.C. Abell and K. Funabashi. J. Chem. Phys. 58, 1079 (1973). 20. J.-P. Dodelet and G.R. Freeman. J. Chem. Phys. 60,4657 (1974). 21. A. Mozumder and M. Tachiya. J. Chem. Phys. 62, 979 (1975). 22. P.M. Borsenberger and A.I. Ateya. J. Apply. Phys. 50, 909 (19791. 23. H.T. ~ h o i D.S. Sethi, and C.L. Braun. J. Chem. Phys. 77, 6027 , (1982). 24. G. Ramanan and G.R. Freeman. J. Chem. Phys. 87, 319 (1987). 25. N. Gee and G.R. Freeman. IEEE Trans. Electr. Insl. EI-23, 505 (1988). 26. K. Shinsaka and G.R. Freeman. Can. J. Chem. 52, 3556 (1974). 27. A. Hummel and A.O. Allen. J. Chem. Phys. 44, 3426 (1966). 28. P. Tewari and G.R. Freeman. J. Chem. Phys. 49, 4394 (1968). 29. I. Gyorgy, N. Gee, and G.R. Freeman. Can. J. Chem. 63, 1105 (1985). 30. N. Gee and G.R. Freeman. Can. J. Chem. 59, 2988 (1981). 31. N. Gee, G. Ramanan and G.R. Freeman. Can. J. Chem. 68, 1527 (1990).