The authors acknowledge the financial and technical assistance given by the many sponsoring organizations. Permeability coefficient in presence of counterelectropermeability of sodium ion, gram-mole per liter streaming potential, stat-volts void ratio, cc. Of void per cc of dry solid thickness of porous bed, om. Actual length of flow path in porous medium, cm. Superficial linear flow velocity in absence of counterlinear electro-osmosis flow velocity in
Original Description:
Original Title
+ Estimation of Diffusion Coefficients for Gases an Vopors
The authors acknowledge the financial and technical assistance given by the many sponsoring organizations. Permeability coefficient in presence of counterelectropermeability of sodium ion, gram-mole per liter streaming potential, stat-volts void ratio, cc. Of void per cc of dry solid thickness of porous bed, om. Actual length of flow path in porous medium, cm. Superficial linear flow velocity in absence of counterlinear electro-osmosis flow velocity in
The authors acknowledge the financial and technical assistance given by the many sponsoring organizations. Permeability coefficient in presence of counterelectropermeability of sodium ion, gram-mole per liter streaming potential, stat-volts void ratio, cc. Of void per cc of dry solid thickness of porous bed, om. Actual length of flow path in porous medium, cm. Superficial linear flow velocity in absence of counterlinear electro-osmosis flow velocity in
June 1955 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y 1253
program of fundamental research sponsored both by industrial
contributions and by the U. S. Army, Corps of Engineers Research and Development Laboratories. The authors gratefully acknowl- edge the financial and technical assistance given by the many sponsoring organizations, and the advice and criticism offered by members of the staff of the Soil Stabilization Laboratory. a H = e = K = K* = L = jy = u = u* = uom = gn == Ap = e = NOMENCLATURE activity of hydrogen ion, gram-mole per liter activity of sodium ion, gram-mole per liter streaming potential, stat-volts void ratio, cc. of void per cc. of dry solid permeability coefficient in presence of counterelectro- permeability coefficient in absence of counterelectro- thickness of porous bed, om. actual length of flow path in porous bed, cm. superficial linear flow velocity in presence of counter- superficial linear flow velocity in the absence of counter- linear electro-osmosis flow velocity in pores, cm. per pressure drop across porous bed, gram/cm.( sec.) dielectric constant osmosis, sq. cm. osmosis, sq. cm. electro-osmosis, cm. per second electro-osmosis, cm. per second second X T =zeta potential, stat-volts =specific conductance of liquid and surface of porous medium, stat-mhos per cm. =viscosity of liquid, gram/(cm.)(sec.) LITERATURE CITED (1) Booth, F. , Proc. Rou. Soc. (London), 203A, 533 (1950). (2) Carman, P. C., Trans. Inst. Chem. Engrs. (London), 15, 150 (3) Cornel1 University, Final Report, Soil Solidification Research, (4) Elton, G. A. H., Proc. Roy. SOC. (London), 194A, 259, 275 (5) Fuerstenau, D. W., So. D. thesis, Massachusetts Institute of (6) Grace, H. P., Chem. Eng. Prop. , 49, 303, 367 (1953). (7) Henniker, J . C., Reus. Mod. Phys. , 21, 322 (1949). (8) Kruyt, H. R., Colloid Science, Vol. I, Elsevier Publishing Go., New York, 1952. (9) Michaels, A. S., and Lin, C. S., ~ N D . ENG. CHEM., 46, 1239 (1954). (10) Muskat, M., Flow of Homogeneous Fluids through Porous fifedia, McGraw-Hill, New York, 1937. (11) Prutton, C. F., and Maron, S. H., Fundamental Principles of Physical Chemistry, p. 469, Macmillan, New York, 1950. (12) Ruth, B. E., ~N D . ENG. CHEM., 38, 564 (1946). R E C ~I V ~D for review J une 26, 1954. (1937). Vol. 11, 1951. (1948); 197A, 568 (1949). Technology, 1953. ACCEPTED December 23, 1054. Estimation of Diffusion Coefficients for Gases and Vapors c. R. WILKE AND C. Y. LEE Division of Chemical Engineering, University of California, Berkeley, Calif. S THE theory and engineering applications of diffusional A operations are advanced, increasing need arises for dif- fusion coefficients of gases and vapors. I n view of the scarcity of reliable experimental data, methods of predicting such informa- tion assume a useful role. The general objective of this paper is to review critically the existing methods for estimation of diffu- sivities. An effort has been made to establish therelativeaccuracy of various procedures by comparison of calculated results with a body of selected experimental data. REVIEW OF PREVIOUS CORRELATIONS Gilliland (7) developed an empirical formula based on the hard sphere model of the classical kinetic theory (25). Collision di- ameters were assumed proportional to the cube roots of the molar volumes at the normal boiling point as in the earlier method of Arnold. Examination of available data led to the relation: As used in this paper, the diffusion coefficient is the proportion- ality constant in the rate equation for mass transfer in absence of convection or other modes of transfer, defined as follows for unidirectional diffusion : where Nt =rate of diffusion of component 1 of a binary mixture, D12 =diffusion coefficient, sq. cm. per second A =cross-sectional area perpendicular to direction of pl gram-moles per second diffusion, sq. cm. =partial pressure of component 1, atm. R T =temperature, K. x =gas constant, cc.-atm./gram-mole-o K. =distance in direction of diffusion I n using Equation 1, values of V should be obtained in the same way as were those used by Gilliland-namely, using Kopps law of additive volumes, with the rules and values for each element given by LeBas (18). Values of atomic volumes have been sum- marized in several other sources ( 1 , b3,sO). Arnold Method. Arnold ( 2 ) adopted the general equation form proposed by Sutherland ( 25) containing an additional term involving the Sutherland constant, Clz, to account for deviations from the hard sphere model. Upon examination of available data to determine the best numerical constant, Arnold recom- mended the equation: As outlined by Arnold, the Sutherland constant, C12, may be calculated from the expression: The constants, Cl and CZ, are to be obtained from the absolute boiling temperature, ci =1.47T~, (4) From Equations 3 and 4 it can readily be seen that ci2 =1.47F( TB,TB,)~ ( 5) 1254 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y Vol. 47, No. 6 Oo5 mm 004 LINES OF CONSTANT 0 03 VAL UES O F A F OR XI.O A 002 001 0 001 002003 005 01 02 03 05 I O 20 30 50 MOL E C U L A R WE I GHT R AT I O, MI M2 Figure 1, Values of A as mole fraction of one component approaches zero I n absence of experimental data V is obtained by using Kopp's law of additive volumes, or calculated from viscosity data. Hirschfelder, Bird, and Spotz Method. If the proper function for potential energy of interaction between any two molecules is known, the more rigorous kinetic theory as developed by Chap- man and Cowling ( 6 ) may be employed to give improved expres- sions for D and their temperature properties. Hirschfelder, Bird, and Spotz (8-11) have carried out such calculations for nonpolar gases which have an energy of attraction varying with the inverse sixth power of the distance between centers of adja- cent molecules and a repulsive energy varying with the inverse twelfth power. For pairs of nonpolar gases the first approxi- mation for the diffusion coefficient is expressed as follows The numerical constant, B, is equal to 9.2916 X theory. felder, Bird, and Spotz as a function of k T/ . values are listed in Table I. according to The collision integral, W(I)(I), has been calculated by Hirsch- Representative Table I. Collision Integrals for Viscosity and Diffusion kT/ v/w(2)(2; k T/ r ) w'')(l; k T/ m) 0.3596 0.4432 0.5432 0.6301 0.7023 0.7613 0.8523 0.9172 0.9657 1.036 1.085 1.222 1.550 1.714 2.095 1.331 1.033 0.8335 0.7197 0.6479 0.6991 0.5373 0.4998 0.4745 0.4418 0.4211 0.3712 0.2878 0.2585 0.2085 For mixtures of two nonpolar gasee and 12 =dZ ( 8 ) For collisions between a polar and a nonpolar molecule, modi- fied forms of equations are suggested ( 14) . With these modifica- tions Equation 6 applies to systems containing a polar and a non- polar component. However, the energy of interaction among polar molecules, valence-unsaturated molecules, and cigar- shaped molecules is very different from that between nonpolar molecules. Before Equation 6 may be considered strictly appli- cable to systems containing such molecules, the collision integral should be evaluated according to their respective potential func- tions. The labor involved is formidable and as yet this has not been accomplished. The term A in Equation 6 is a correction factor, a small quan- tity usually less than 0.03, which varies with gas composition. Detailed equations for A have been given (8, 9). A is a com- plicated function of mole fraction, molecular weights, viscosities, the first approximation to the diffusion coefficient, and additional functions of kT/ eI 2. However, these equations are so lengthy that their use may not be justified for many engineering calcula- tions, particularly when the force constants have been estimated. As a first approximation to A for a given gas system it is suggested that A values for mole fractions of each component equal to zero be obtained and an average value calculated, assuming A to be linear in mole fraction. Values of A at zero mole fraction are pre- Rented in Figure 1. This procedure is not completely satis- factory, as A may deviate considerably from a linear function, but it should accomplish a major part of the correction. This pro- cedure was adopted in subsequent applications of Equation 6. ESTIMATION OF FORCE CONSTANTS FOR SIMILAR MOLECULES Both ~ / k and r , may be calculated from knowledge of gas vis- cosity at two temperatures by solving the equation for viscosity (9). x 107 = (9) where ?I =viscosity, grams per cm. second, and V/W2(2) =a function of kT/ e which has been evaluated by Hirschfelder, Bird, and Spotz. Representative values are given in Table I. Equation 9 has been well substantiated by experimental evidence and provides the most satisfactory basis for obtaining the force constants. I n absence of viscosity data Hirschfelder, Curtiss, and Bird ( I d ) recommend the following equations, which supersede those of the earlier work (8). el k =0.77TC (10) =1.15Tb (11) =1.92Tm (12) Prior to publication of the above equations, in the course of the present work examination of force constant data to check the earlier suggestions of Hirschfelder, Bird, and Spotz (8) indicated the following equations to be satisfactory: / I C =0.75Tc (13) / k =1.21Tb (14) Because the collision integrals are relatively insensitive to small errors in el k, Equations 10 and 11are considered to be equivalent to 13 and 14 within the limits of precision of subsequent compari- sons (Table I V). The relationships developed between T, and Tb by Meissner (2%') indicate that the value 1.21 in Equation 14 should be preferable to the value 1.39 ( 8) , assuming that Equa- tion 13 is valid, Unavailable critical and boiling data may be estimated with the equations of Meissner. The collision diameter may be calculated from a single value of viscosity in Equation 6 using / k from Equations 13 and 14. Alternatively Hirschfelder, Bird, and Spotz recommend: yo =0.833V:'a (15) where ro =A., and V, =critical volume, cc. per gram-mole Bromley and Wilke ( 4) have tested the general form of equa- tion proposed by Licht and Stechert ($0) and obtained a result which in terms of y o becomes June 1955 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y 1255 Table 11. Selected Diffusion Data for Test of Methods System Air-benzene Air-ethyl alcohol Air-ethyl ether Air-diphenyl Air-I2 Air-nitrobenzene Air-n-heptadecane Air-n-hexadecane Air-n-octadecane Air-n-octane Air-mater COz-ethyl alcohol COa-ethyl ether C0z-Hz C 02- C H4 C02-Nz COz-NZO COz-CsHs co2-I120 CO-CZHI CO-Hi CO-Nz co-01 Freon 12-benzene Freon 12-ethanol Freon 12-mater He-A He-benzene He-ethyl alcohol He-Ne He-nitrobenzene He-Hz He-HzO H2-NHa Hn-A Hz-benzene Ha-CClr Hi-CzHa Hz-ethyl alcohol Hz-CZHI Hg-ethyl ether HI-HC1 H2-CHI H2-h-i Hz-h-20 Hz-02 Ha-CaHE H2-SO2 HrHz0 h-2-NH1 Kz-CzHa Nz-Hs N2-I* N2-NO N2-02 NzO-CsHe Os-NHz Oz-CCh 02-C2H4 NHs-C1H4 A-Ne CsHs-CzHe CzHs-CHI 02-benzene Exptl. Diffusion Coefficient Sq Cm./Sec. (at i Temp., O K. Atm.) 298 298 293 49 1 298 298 298 298 298 298 298 273 273 273.2 273.2 288 298 298 297.7 273 273 288 273 298 298 298 273.2 298 298 293 298 293 298 293 293.2 273 296 273 340 293 293 294 273 2 293 273 2 273 2 300 273 293 293 300 288 273 293 273 298 293 296 298 293 293 293 293 293 0.0962 0.135 0.0896 0.160 0.0834 0.0868 0.423 0.0405 0.0397 0.0602 0.260 0.0693 0,0541 0,550 0.153 0.158 0.117 0.0863 0.164 0.151 0.622 0.211 0.185 0.0385 0,0475 0.105 0.641 0.384 0.494 1.23 0.372 1.64 0.908 0.849 0.770 0.317 0.345 0.439 0.578 0.483 0.354 0.795 0.626 1.205 0.535 0.697 0.450 0.483 0.850 0 241 0.187 0,743 0 070 0.232 0 181 0.086 0.253 0.0939 0.0731 0.182 0.177 0.329 0.085 0.163 Examination of data as in the case of e l k suggested an alter- native method of estimation of ro based on molal volume at the normal boiling point, which may be expressed as: ro =1.18V1ia (17) Equation 17 appears to be somewhat more satisfactory than Equation 15 or 16. The '/3 power on Ve has been retained in Equation 15, as it has a theoretical basis in the Lennard-J ones and Devonshire theory of gases and liquids ( 8) , although the data seem to correlate better us. V.O.sg. I n application of Equation 17 molal volumes may be estimated from Kopp's law and the rules of LeBas as in the methods of Gilliland and Arnold. COMPARISON OF EQUATIONS WITH EXPERIMENTAL DIFFUSION DATA To serve as a basis for comparison in determining the relative accuracy of the foregoing methods, experimental diffusion data were selected from reliable sources for 64 systems at atmospheric pressure and near room temperature. (All methods of this paper pertain to moderate pressure systems where ideal gas behavior may be assumed-Le., where the product DP is independent of Table 111. Force Constants and Collision Diameters for Components of Selected Systems TO, A. from Viscosity Gas qp&f;" Air Ammonia Argon Benzene COZ co CClr CzHa C2H30H Diphenyl Ethyl ether CzHa Freon 12 Helium n-Heptadecane H2 HC1 Iodine CHI Neon Nitrobenzene NO N2 NzO n-Octadecane n-Octane 0 2 Propane SO2 Water 97.0 315 124.0 440 190 110.3 327 230 391 600 350b 205 288 b 800 33.3 360 550 136.5 35.7 539 b 119 91.5 220 8200 320 113.2 254 252 363 356d 6. 03 3.617 2.624 3.418 5.270 3,996 3.590 5.881 4.418 4.455 6.223= 5,424'" 4.232 5,llOd 2.70 7.9235 2.968 3.305 4.982 3.882 2.80 4.931" 3.47 3.681 3.879 7.963a 7.451 3.433 5.061 4,290 2.655 2.649e All values obtained from viscosity are taken from Hirschfelder, Bird, and Spotz ( a) , unless otherwise noted. A mole complete table is presented by Hirschfelder, Curtiss, and Bird (I S). Values estimated from diffusion data. b Values estimated from Tc . C Values estimated from boiling temperature. d (6). e ( 94) . 3 0 20 0 w 1.0 I 0 m 0 05 e 02 i o 1 ; 03 l- a 0 -I DEVIATION 200 % 4 0 005 003 003005 01 02 03 05 10 20 30 Figure 2. Comparison of experimental data with results calculated by Gilliland equation OBSERVED D, SQ CM/SEC pressure.) These data are summarized in Table 11. I n view of the uncertainty and lack of precision in many investigations re- ported in the literature, use of this limited number of data from selected sources covering a wide range of diffusion coefficients and related properties was considered preferable to use of a larger body of data selected at random. Also, for a large majority of the systems force constants based on viscosity (see Table 111) were known for the individual components. Diffusivities calculated by Equations 1 and 2 and by Equation 6 with force constants from viscosity are presented in Figures 2, 3, and 4. Comparison between calculated and observed results from Table I1 are illustrated for these three methods. The methods of Arnold and of Hirschfelder, Bird, and Spotz gave nearly comparable results, with the latter somewhat better. The method of Gilliland gives poor agreement for these systems. Var- ious equations for estimating force constants were employed in connection with Equation 6 to obtain additional comparisons of 1256 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y Vol. 47, No. 6 ...v 2.0 v cn 1.0 i 7 0.5 . 0.3 Q 0.2 5 g 0.1 W \ 0 rn 0 W 0 a 0.05 0.03 003 005 01 0.2 0.3 0.5 I O 20 3.0 OBSERVED D, SQ GM/SEC. Figure 3. Comparison of experimental data with results calculated by Arnold equation 2,0c H.B.S. METHOD DEVIATION =7.0 % 0.03 0.05 0.1 0.2 0.3 0.5 1.0 2.0 3.0 OBSERVED 0, SQ. CM. / SEC. Figure 4. Comparison of experimental data with results calculated by Hirschfelder, Bird, and Spotz equation e/ k and r o from viscosity Table IV. Deviations Between Experimental Data and Calculated Values for 64 Systems Ay. % Max. ,% Deviation Deviation Item Method 1 Gilliland 2 Arnold 3 Hirschfelder, Bird and Spotz, 4 T O from Tc/ p?, (Eq. le), e l k 5 TO from V (Ea. 17), r / k from e / k and r o from viscosity from viscosity 6 ro 7 TO 8 M 9 ro 10 T O 11 ro 12 T O 13 TO 14 TO 15 TO viscosity from V. (Eq. 151, a/k from viscosity from V (Eq. 17), e / k =1.21 Tb (Eq. 14) [odified Hirschfelder Bird, Spotz, ,Eq., 18, e l k 'and ro from viscosity from Tc / pc (Eq. 16), c / k from viscosity from V (Eq. 17), a/k from viscosity from Ve (Eq. 15), r / k from viscosity from viscosity, c / k =0.75 Tc (Eq. 13) from viscosity, r / k =1.39 Tb (9) from viscosity, a/ k =1.21 from V (Eq. 17), e / k = 1.21Tb (Eq. 14) Ta (Eq. 14) Equation 1 2 18 18 18 18 1.8 18 18 18 20.0 8.4 7.0 10.1 10.3 11.8 12.0 3.9 9.4 7.0 1o.a 5.4 7.2 5.1 7.7 46.8 20.5 21.4 31.3 24.9 32.7 31.0 16.0 32.7 21.9 37.6 26.0 24.9 20.5 24.5 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 09 Figure 5. Variation of constant in Hirschfelder, Bird, and Spotz equation calculations with observed data. Average and maximum devia- tions between calculation and experiment are summarized in Table I V for all methods. It is of interest to compare the three methods with respect to the effect of temperature on the diffusion coefficient. The equa- tion of Gilliland is theoretically at variance with the Arnold and the Hirschfelder, Bird, and Spotz equations. The two latter methods consider possible molecular interaction upon collision and therefore in principle should give more satisfactory results. This conclusion is borne out by the relatively meager experimen- tal evidence available. Table V shows the temperature behavior of the carbon dioxide-air system over a range of 293 O to 1500' K., as observed by Kilbanova, Pomerantsev, and Frank-Kamenetskil (16). Results calculated by each method, assuming the value at 293' K. to be correct, are also given. The Arnold and the Hirschfelder, Bird, and Spotz methods agree closely, probably within the limits of experimental error, while the Gilliland equa- tion gives low values at higher temperatures. POSSIBLE MODIFICATION OF HIRSCHFELDER, BIRD, AND SPOTZ EQUATION As a means of studying deviations of Equation 6 from experi- mental results, values of the constant B were computed for each system given in Table I11 which would make the equation agree with experiment. Force constants were based on viscosity. Various attempts were made to relate these values of B to prop- erties of the systems. As shown in Figure 5, there appears to be some correlation of the constant with the molecular weight group [Mhky2] '". Assuming a linear relationship, the data of Figure 5 may be approximated by the equation: B =10.7 - 2.46 [ ML;Mfz] l I 2 Results obtained employing Equation 18 in connection with Equation 6 for the systems are shown in Figure 5 and Table I V. The average and maximum deviations between calculated and experimental values are significantly lese than those obtained with Equation 6, using the theoretical value of 9.2916 X lod4 for B. Table V. Variation of D with Temperature for Carbon Diffusion Coefficient, 9s. Cm./Sec. Dioxide-Air System Exptl. data Arnold H.B.S. Gilliland T, K. (smoothed) method method method 293 400 600 800 1000 1300 1500 0.151 0.273 0.555 0.915 1.32 1.97 2.45 0 : 267 0.543 0.905 1.28 1.97 2.49 0 : 266 0.523 0.883 1.97 2.52 1.28 0:iiz 0.448 0.690 0.970 1.42 1.77 ... June 1955 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y 1251 3 .O 2.0 MODIFIED H.B.S. METHOD OBSERVED D, SQ. GM. / SEC Figure 6. Comparison of experimental data with results calculated by modified Hirsch- felder, Bird, and Spotz equation 3.0 2.0 Y : 1.0 i c; 0.5 V v) o- 0.3 0 0.2 W -I 5i Y 3 0.1 a V 0.0 5 0.03 0.03 0.05 0.1 0.2 0.3 0.5 1.0 20 3.0 OBSERVED D, SQ. CM./SEC Figure 7. Results of modified Hirschfelder, Bird, and Spotz equation using estimated force constants Further results using estimated force constants are summarized in Table I V, items 9 to 15. Figure 7 shows results obtained with Equation 18 using force constants estimated from boiling point and molar volume. This procedure seems comparable in ac- curacy to the Arnold method for these systems. W()(I) =collision integral for diffusion, function of kT/e A =small correction term seldom exceeding 0.03 E =maximum energy of attraction 7 =viscosity, gram/cm.-sec. Subscripts 1 and 2 denote components 1 and 2, respectively. DISCUSSION One of the reasons for using Equation 18 was to obtain a best possible correlation of the data using force constants based on viscosity, and then to use this relation as a basis for comparing the various procedures for estimating the force constants. Methods of obtaining r o are compared in Table IV, items 8 to 11. Estimation of r o from molar volume (Equation 17) appears most satisfactory. Use of TJP, yields slightly better results than use of critical volume. However, experimental uncertainty in meas- urement of critical properties may contribute to the apparent in- accuracy. Methods of estimating ~ / k are compared in Table IV, items 12 to 14. Equation 13 ( / I C =0.755 T, ) and Equation 14 ( e l k =1.21 Ti ) show essentially equivalent behavior. As noted reviously, Equations 10 and 11 should be essentially equivagnt to 13 and 14. It is recognized that Equation 18 is highly empirical and that its use can be justified only on the basis of the improved corre- lation obtained for the present systems. It is believed that the deviations of B from the theoretical value as shown in Figure 5 represent some failure in the theory, which should ultimately be improved by a more fundamental approach. To seek modified relations for obtaining the force constants for dissimilar molecules from the values for the pure components would seem more romising ultimately than the more simple expedient of modi- rying the numerical constant in the diffusion equation. NOMENCLATURE B =numerical constant equal to 9.2916 X according C =Sutherlands constant for diffusion, O K. D =diffusion coefficient, sq. cm. per second k =gas constant/molecule, Boltzmanns constant M =molecular weight P =total pressure, atm. r =molecular diameter r12 or ro =collision diameter, A. 7 =absolute temperature, O K. Ta =normal boiling point, O K. T, =critical temperature, O K. T, =melting point, O K. =function of AT/ V =molar volume at normal boiling point, cc. per gram- Vc =critical volume, cc. per gram-mole to theory W2(2) mole (total molar volume of each component) REFERENCES (1) Arnold, J . H., IND. ENQ. CHEM., 22, 1091 (1930). (2) Boardman, L. E., and Wild, N.E., Proc. Roy. SOC. (London), (3) Bradley, R. S., and Shellard, A. D., Ibid. , A198, 239 (1949). (4) Bromley, L. A, , and Wilke, C. R., IND. ENQ. CHEM., 43, 1641 (5) Buddenberg, J . W., and Wilke, C. R., I bi d. , 41, 1345 (1949). (6) Chapman, Sydney, and Cowling, T. G., Mathematical Theory of NonuniformGases, Cambridge University Press, Ted- A162, 511 (1937). (1951). dington, 1939. (7) Gilliland, E. R., IND. ENQ. CHEM., 26, 681 (1934). (8) Hirschfelder, J. O., Bird, R. B., and Spota, E. L., Chem. Rev., 44, 205 (1949). (9) Hirschfelder; J. O., Bird, R. E., and Spotz, E. L., J . Chem. Phys., 16. 968 (1948). \ - - - , (10) Hirsohfelder, J. O., Bird, R. E., and Spotz, E. L., Trans. Am. SOC. Mech. Enors., 71, 921 (1949). (11) Hirschfelder, J . O., Curtiss, C. F., and Bird, R. B., Molecular Theory of Gases and Liquids, pp. 538-40, Wiley, New York, 1954. (12) Hirschfelder, J . O., Curtiss, C. F., and Bird, R. B., Trans. Am. SOC. Me&. Engrs., 71, 245, 8.6-4 (1949). (13) I bi d. , pp. 1110-12. (14) I b i d . , ~ . 6000, Eq. 8.6-3,8.6-4. (15) Xennard, E. H., Kinetic Theory of Gases, p. 195, McGraw- (16) Kilbanova, Pomerantsev, and Frank-Kamenetskir, J . Tech. (17) Landolt-Bijrnstein, Physikalisoh-chemische Tabellen, J. (18) LeBas, G., Chem. News, 99, 206 (1909). (19) Lee, C. Y., and Wilke, C. R., IND. ENG. CHEM., 46, 2381 (1954). (20) Licht, W., J r., and Stechert, D. G., J . Phys. Chem., 48, 23 (21) Mack, Edward, J r., J . Am. Chem. Soc., 47, 2468 (1925). (22) Meissner, H. P., Chem. Eno. Progr., 45, 149 (1949). (23) Perry, J . H., Chemical Engineers Handbook, McGraw-Hill (24) Rowlinson, J. S., Trans. Faraday SOC., 47, 120 (1951). (25) Sutherland, PhiL Mug., 36, 507 (1893). (26) Titani, T., Bull. Chem. SOC. (Japan), 5, 41 (1930). (27) Trautr, M., and Mtiller, W., Ann. Physi k, 414, 333 (1935). (28) Waldmann, L., Naturwissenschaften, 32, 223 (1944). (29) Wall, F. T., and Kidder, G. A., J. Phys. Chem., 50, 235 (1946). (30) Wilke, C. R., Chem. Eng. Progr., 45, 219 (1949). Hill Book Co., New York, 1938. Phys. (U.S.S.R.), 12, 14-30 (1942). Springer, Berlin, (1923-35). (1944). Book Co., New York. RECEIVED for review June 1, 1954. ACCEPTED December 4, 1954.