ETHANE-ETHYLENE AND PROPANE-PROPYLENE EQUILIBRIA BY G. B. KISTIAKOWSKY AND A. GORDON NICKLE* Received 23rd January, 1951 Accurate measurements of the above equilibria are described, in which the equilibria were approached from both sides on a reduced chromic hydroxide catalyst and corrections were made for side reactions. For the reaction C,H, + C,H, + H, the equilibrium constants were found to be and 5-13 f 0.13 x I O - ~ atm. at 723.2' K . . . 4-04 f 0.17 x I O - ~ atm. a t 6 5 3 . ~ ~ K. For the reaction C3H8 -+ C,H, + H, the equilibrium constants were found to be 5-17 & 0.15 x ~ o - ~ a t m . a t 648.2' K and 3-67 f 0.17 x I O - ~ atm. a t 583.2O K. From these equilibrium constants the heats of reaction are calculated which are in satisfactory agreement with the calorimetric data.Using statistical methods on accepted molecular constants and calorimetric heats of reaction, equilibrium constants are calculated for the ethane system which are in excellent agreement with the present data. On the basis of these calculations and of other considerations, i t is suggested that the best value for the heat of hydro- genation of ethylene a t 298' K is AH = - 32,600 f 50 cal. The agreement of statistical calculations and the present data in the propane-propylene reaction is not quite as good, but the discrepancy may be due to the deviations of real molecules from the harmonic oscillator and rigid rotor models used in statistical calculations. An analysis of systematic errors in calorimetric work indicates that the best value for the heat of hydrogenation of propylene is AH = - 29,850 f 50 cal.a t 298" K. Quite some time ago a note appeared in print giving the results of our measurements of the above equilibria.1 Since they have some bearing on other thermodynamic properties of lower hydrocarbons, we present now a more detailed account of the measurements and their interpretation. and one of propane-propylene * have been described in the literature. The results are not internally consistent and are not fully in accord with the statistical calculations of equilibrium constants utilizing direct measurements of the heats of these reactions. A critical study of these measurements suggests two major sources of error. One arises from the methods chosen for the determination of the small amounts of olefins in the equilibrium mixtures, the other from the neglect of side reactions.We have attempted to eliminate both these errors in the present work. Four major investigations of the ethane-ethylene equilibrium 2s 4 e Experimental The equilibria were attained by repeated passes over a catalyst of gaseous mixtures of near-equilibrium composition a t controlled rates of flow and were measured by the determination of total pressure in the system and by analysis for hydrogen, total olefins and by-product olefins resulting from side reactions. * Standard Chemical Co. Ltd., Toronto, Canada. 1 Kistiakowsky and Nickle, J . Chem. Physics, 1942, 10, 78. Pease and Durgan, J . Amer. Chem. Soc., 1938, 50, 2715 Travers and Pearce, J . SOC. Chem. Ind., 1938, 53, 322. Frey and Huppke, Ind.Eng. Chem.. 1933, 25, 54. Pease and Byers, J . Amer. Chem. Soc., 1938, 60, 2489. Smith and Vaughan, J . Chem. Physics, 1935, 3, 341. I751 76 ETHANE-ETHYLENE, ETC., EQUILIBRIA Reaction System.-The reaction system consisted of a catalyst chamber with a thermocouple in a well inside the catalyst mass and a preheating chamber, placed in a metal furnace with manual temperature control. The temperature was measured by a chromel-alumel thermocouple, read on a Leeds and Northrup type K potentiometer with a HS galvanometer. The thermocouple was cali- brated a t several temperatures within the range used against a Bureau of Standards calibrated platinum resistance thermometer. The absolute tem- peratures given in the following are accurate to 0.5'.On a relative scale, the temperatures are accurate to 0.2'. The gas mixture was forced from a litre flask through a cold trap, the pre- heater chamber, and catalyst chamber into another litre flask by mercury, driven by a circulating pump patterned after the design of Urey.7 This pump permitted steady xates of flow of 4 to 250 ~m.~/min., but only Aows in the middle of this range were employed during equilibration runs. The all-glass system permitted multiple passes of the gases over the catalyst and was provided with the usual vacuum pumps and an accurate manometer. Catalyst.-Two catalysts were tried. The one described by Taylor and Joris,8 reduced copper on magnesia, failed to maintain high activity even a t 450° C. Good results were obtained with the catalyst described by Frey and Huppke,*~ and we followed their method of preparation except that the chromic oxide gel was washed by centrifugation rather than filtration.This catalyst retained its activity so well tbat one portion of the material was used in all the runs described. The precipitated and thoroughly washed chromic hydroxide was placed in an oven whose temperature was raised to 150' C during an hour and was dried a t that temperature for 5 hr. This material was ground and some of the -20 f 4 0 mesh material dried in vucuo a t 250°C for 18 hr. About 3 ~ m . ~ of the catalyst was now put in the catalyst chamber and hydrogen passed over i t a t 400' C for 12 hr. and then a t 490" C for 12 hr. Finally the catalyst was pumped out and flushed with ethylene, but still it retained much hydrogen which ap- peared in the gases of the first few equilibration runs.The same catalyst was used for propane after flushing i t with hydrogen a t 450' C and evacuation. There was always either a loss of olefin or an excessive production of hydrogen in the equilibration runs. To have the final gas mixture contain approximately equal amounts of olefin and hydrogen, it was necessary t o add olefin to the alkane on approaches from the low side, and to have olefin in excess of hydrogen on approaches from the high side. Materials .-HYDROGEN.-The hydrogen used in the catalyst treatment and in all equilibration experiments was Ohio Chemical Company's water-pumped hydrogenation grade gas. For use in equilibration runs i t was purified by passing over heated platinized asbestos and drying over Dehydrite.ETHANE.-Ethane was prepared by hydrogenation over copper of Ohio Chemical Company's 99-5 yo ethylene and was purified by repeated condensation, pumping off, and slow " bulb-to-bulb " distillations. One sample, which was used in Runs 36 to 44 inclusive, showed a difference of vapour pressures of 9.9 mm. in 490 when the last 5 yo was compared with the first 5 yo after a slow isckhermal distillation. 1 0 Most of the impurity was undoubtedly ethylene, since this material con- tained also the first fractions from the distillations of the second sample. This one was purified after hydrogenation by passing through 96 Yo sulphuric acid, potassium hydroxide solution and pellets, and a dry ice trap. The first 15 yo of this material, distilled off in vacuum, was combined with the first sample.On comparing the first and last 5 yo of the remainder, only a 2.2 mm. in 490 pressure difference was observed. Olefins were determined by bromination as described below, and found to the extent of 0.012 mole yo. This is not sufficient to account for the vapour pressure difference, and the impurity is most probably a paraffin formed by slow polymerization on the hydrogenation catalyst. Using the integrated form of the Rayleigh distillation equation,ll its concentration is computed as 0.02 mole Yo if the impurity is butane. Thode and Urey, J . Chem. Physics, 1939. 7, 34. Taylor and Joris, Bull. SOC. chim. BeZg., 1937, 46, 241. See also Lewis and Taylor, J . Amer. Chem. SOL, 1938, 60, 877.9 Frey and Huppke, U.S. Put. 2,098,959. loshepherd, J . Res. Nut. Bur. Stand., 1934, 12, 185. 11 Walker, Lewis, McAdams and Gilliland, Principles of Chemical Engineering (McGraw-Hill Book Company), 3rd edn., p. 532.G. B. KISTIAKOWSKY AND A. GORDON NICKLE 177 ETHYLENE.-Ethylene used in the control analyses and in equilibration runs was Ohio Chemical Company’s 99-5 yo material, purified somewhat by several slow bulb-to-bulb distillations with large rejections a t both beginning and end of each distillation. A difference in vapour pressure of 1-2 mm. in 490 was obtained when the first and last 5 yo of a slow isothermal distillation were com- pared. Assuming the impurity to be ethane, a similar calculation to that de- scribed above for ethane shows an impurity here of about 0.06 mole yo.PRoPANE.-Propane was Ohio Chemical Company’s 99.9 yo pure material purified by passing through 96 yo sulphuric acid , potassium hydroxide solution and pellets and collected in a dry ice trap. The product was further purified by filtration through glass wool at liquid nitrogen temperature to remove water, according to Roper’s method,12 then distilled bulb-to-bulb several times. A test for olefin showed none to the sensitivity of the method, viz., about 0.002 yo. The difference in vapour pressures of the first and last 5 yo was zero to the sensitivity of the method, viz., a few tenths of a millimetre at 600 mm. total pressure. Because of a crack in the apparatus, part of the sample later became mixed with a little air and was re-purified only by bulb-to-bulb distillation.Thus i t may have been contaminated with a trace of water, but no difference was noted in runs made with the slightly impure gas and those with the best material. PRoPYLENE.-Propylene used as a standard for analytical work was given us by Dr. F. D. Rossini, and was labelled by him as the remainder from their best sample used in the determination of the heat of combustion of propylene. A full discussion of the preparation and purity is given in the paper where the heat of combustion is reported.13 The propylene used for addition, when ap- proaching the equilibrium irom the high side, was from the Linde Company, labelled 99.9 % pure. The only purification was condensation in liquid air and pumping to remove any non-condensable material. Analyses.-Various precautions recommended in precision analytical work were applied in making, standardizing and preserving reagent solutions.Analytical weights and volumetric apparatus were recalibrated. The procedure €or the determination of the composition of equilibrated mixtures was: (i) combustion of hydrogen over copper oxide, (ii) photo- chemical bromination in the vapour phase of a portion of the hydrocarbon fraction with a subsequent determination of excess bromine to determine total olefins, and (iii) catalytic bromination of another sample of the hydrocarbons, followed by a determination of the freezing point of the dibromides to evaluate the amount of the pertinent olefin in the total olefins. HYDROGEN.-After the run, about 200 ~ m . ~ of the gases were transferred into an all-glass system containing hand-operated Toepler’pumps, a gas burette, a heated copper oxide tube, a phosphorus pentoxide tube, a cold trap and suitable connections with numerous stopcocks.The hydrocarbons were frozen out in the trap by liquid nitrogen, hydrogen was pumped off, the hydrocarbons melted and re-frozen, then pumped off again. The hydrogen-rich fraction was measured, then repeatedly passed over copper oxide at 250° C and water removed by phosphorus pentoxide. The volume of the residual gases was again measured and hydrogen determined by difference. The residual gases (largely methane, with a little ethane in the C , equilibria) were then combined with the material in the cold trap. This procedure was tested for the following sources of error : (u) oxidation of hydrocarbons on copper oxide, (b) incomplete combustion of hydrogen, (c) sorption or desorption of gases on the solids in the system.None was found to affect the results by more than 0.1 yo. A control mixture con- sisting of hydrogen, ethane and ethylene was analyzed and the result agreed within 0-1 yo with the actual composition. This is slightly better than the estimated probable error of 0.2 %. TOTAL OLEFINS.-About 500 ~ m . ~ of the hydrocarbons from the run were introduced into flask D shown in schematic manner in Fig. I. After the intro- duction the stopcock was closed and by means oi an external magnet the dry ground-glass plug B was pushed into place, protecting bromine vapour from interaction with stopcock grease.Then, with another magnet, the glass-enclosed iron piece E was lifted and dropped on the thin-walled, sealed glass capsule F, containing a weighed amount of bromine, breaking it. The breaker was moved up and down a, number of times to mix bromine vapour with the contents of the flask, and the latter was then exposed for 10 min. a t 15 cm. distance to the l2 Roper, Ind. Eng. Chem. (Anal.), 1937, 9, 414. 13 Rossini and Knowlton, J . Res. Nat. Bur. Stand., 1937, 19, 249.1 78 ETHANE-ETHYLENE, ETC., EQUILIBRIA light of a 500-W incandescent lamp. Neutral potassium iodide solution was poured into bulb 2, the ground glass plug lowered, and the solution was run into flask D without admitting air, and shaken well to scrub the bromine from the vapour phase. The upper connections and top part of the flask were washed down by admitting water through bulb A, then the ground glass joint C (held together with a low-melting thermoplastic cement applied only to the outer edge of the joint) was separated and the solution titrated with thiosulphate using starch as indicator.When 0.897 3'418 3'542 0.458 FIG. I . 0.900 0-3 3'452 3'4 3.586 4'4 0.496 3'8 the end point was reached, neutral potassium iodate solution was added and the liberated iodine titrated again. In a well-known manner the second of these titrations measures the quantity of acid present and so the extent of the sub- stitution reaction of bromine. The quantity of olefins present was evaluated from these titrations by assuming that the reactions occurling were : C,H,, + Br, + ~ , H , , B r , C,H,,+, + Br, -+ C,H,,+,Br -t HBr CnH2a + HBr -+ C,H2,+1Br C,H,,,Br, + Br, -+ CnH2,1+lBr3 + HBr, etc.Hence moles of biomine consumed in the reaction less moles ot acid formed equal moles of olefins. Since the amounts of acid involved are small, the flask had to be scrupulously free from traces of acid ; in particular, i t was found that cleaning solutions containing divalent acids could not be used because these acids were strongly adsorbed ; prolonged subsequent cleaning was required before the flask was sufficiently free of acid. to be serviceable. Much work was done on this procedure to obtain an exact equivalence between theoretical and experimental results, but the quantity of olefins found was always slightly less than that introduced.The magnitude of this loss, shown in Table I, was tound to depend on the quantity of bromine introduced and on the olefin being determined. No wav was found to prevent this loss and an empirical correction, obtained by plotting the data of Table I , was applied to a d runs. In the worst case (C, hydrocarbons at 450" C) the correction changed the equilibrium constants by about 1-5 %. The difficulty is probably due to a free radical induced polymerization of the olefins, and it is interesting to note that work on this subject l4 has shown that the rate is four rimes as fast with ethylene as with propylene, in qualitative agreement with the results obtained by this analytical method. TABLE 1.-Loss OF OLEFIN IN PHOTOCHEMICAL BROMINATION Olefin Ethylene Propylene I Bromine Added (Moles x 104) 1'427 5'582 5'957 6.133 1.849 5.952 6.116 Bromine Found (Moles x 104) ~~ 0.530 2.164 2.415 5'675 0'547 I 1.898 2.016 Olefin Total Olefin Olefin Found 1 Added 1 Loss (Moles x 104) (Moles x 104) (Moles x 10-6) I I PURITY OF 0LEFINS.--The extent of the substitution reactions taking place on illumination is such that the recovered bromides are not suitable for the determination of by-products.Therefore a portion of the equilibrium mixture was broniinated catalytically under conditions which minimize the substitution reaction. Slightly more than the required amount of bromine vapour was picked l4 Reeck and Rust, J , Chenz. Physics, 1941, 9, 480.G. 13. KISTTAKOWSKY AND A. GORDON NICKLE 179 up by a stream of nitrogen and mixed with a stream of the hydrocarbons just before the catalyst chamber.The catalyst was made by saturating chips of porous plate with a strong calcium bromide solution and drying, and was operated a t a temperature of about 50°C. Excess bromine was removed by passage over copper, the products were frozen out in a cold trap, lower-boiling constituents were removed by pumping and the high-boiling material was then used for melting point determinations which were carried out in a micro-adapta- tion of the melting curve technique.15. l6 Approximately 0.01 ml. of the dibromides was introduced into a thin-walled capillary, about the size of a large melting point tube, and a copper-constantan thermocouple of fine wires inserted into the caplllar?, well under the surface oi the liquid.By friction, the thermo- couple suspended the capillarv in the centre of an electrically heated cylindrical shield, closed at the top and bottom by corks. The whole was mounted in an enclosing glass shield and immersed into a constant temperature bath slightly colder than the melting point of the dibromides (melting ice for C,, melting chloroform for CJ. Because of the small size of the sample, the absence of stirring 400 MOLE % IMPURITY FIG. 2.-The effect of impurities upon the melting point of ethylene dibromide. 0 Propylene dibromide ; 1-butylene dibromide ; A 2-butylene di- caused no difficulty, but the magnitude of thermal leaks, as represented for instance by the size and length of the thermocouple wires, had to be carefully and reproducibly adjusted.This done, the device functioned very satisfactorily and permitted the determination of freezing points by observing melting curves almost as accurately as the larger devices. Fig. 2 shows the effect of several likely impurities on the freezing point of ethylene dibromide ; Fig. 3 provides similar data for propylene dibromide. In these figures, temperatures are plotted as microvolts, the reference junction being in an ice bath. The reliability of this technique may be. appreciated from the following. Brominations of mixtures of ethane and " pure " ethylene, and propane and " pure " propylene gave dibromides which analyzed 99.9 yo pure ; when I mole % of ethylene in the ethane-ethylene mixture was replaced by propylene, the dibromide was determined as 99.1 mole yo pure.Such tests indicate that i t is possible to determine the concentration of ethylene 01 propylene dibromides in the mixtures t o about 0.2 mole yo accuracy. ALKmES.-The partial pressure of ethane or propane in the equilibrium mixture was determined by difference. The quantity of methane found in bromide ; x 68 yo I-butylene dibromide, 32 yo propylene dibromide. 15Kistiakowsky. Ruhoff, Smith and Vaughan, J . Amer. Chem. SOC., 1935, 16Roper, J . Amer. Chem. SOC., 1938. 60, 1693. 579 876.I 80 ETHANE-ETHYLENE, ETC., EQUILIBRIA the gaseous residue after the combustion of hydrogen on copper oxide was always far too small to account for the olefinic impurity plus its alkane, if they were formed by disproportionation reactions such as and zC,H, -+ CH, + C3H8 2C3H8 -+ CH, $- C,H12.The methane is evidently formed by cracking reactions such as and and the olefinic impurities by independent polymerization reactions such as and No attempt was made to determine the composition of the impurities, as this information is not necessary for the interpretation of the freezing point data. > -'950r------- a t- z 0 - a -2000 c3 z t- -I W - -2050 MOLE % IMPURITY FIG. 3.-The effect of impurities upon the melting point of propylene dibromide. a Ethylene dibromide ; 0 2-methyl-2 : 3-dibromopentane. The alkane impurity in equilibrium with the olefin impuiity must be present in relatively very small amounts since one alkane is in equilibrium with several isomeric olefins. and of Kilpatrick, Frosen, Pitzer and Rossini 18 to calculate the concentrations of butanes and hexanes in equilibrium with the pertinent olefins, i t is found that the partial pressure of the alkane impurity would not be more than 0-2 yo of the total pressure under the conditions of this research.Rather than introduce a doubtful correction of such small magnitude, i t has been ignored. The difference between the total pressure prevailing during the last pass of the equilibration run and the sum of the partial pressures of the other constituents discussed above was assumed to be ethane or propane as the case might be. It was assumed that all the gases concerned act ideally a t the temperatures near 400° C a t which equilibration runs were made. The pressure and volume values obtained a t room temperature were converted to molal quantities using equation of state data from the following sources : hydrogen, ethane and ethylene from International Critical Tables ; *Q propane from Kemp and Egan ; 2 O propylene from Roper.21 The equilibrium constants were then calculated in the usual manner.The individual values were converted to stated temperatures by the van't Hoff isochore, using 34.0 kcal. per mole for the heat of reaction in the C, system and 30.8 kcal. per mole for the C , system. The maximum temperature correction was 1*5OC, and i t was usually only a few tenths of a degree. 17 Kilpatrick, Prosen, Pitzer and Rossini, J . Res. Nut. Bur. Stand., 1946, 1* Prosen, Pitzer and Rossini, J . Res. Nat. Bur. Stand., 1945, 34, 403. 1Q Int. Crit. Tables, 1st edn., 3, 3. 2 O Kemp and Egan, J .Amer. Chew SOL, 1938, 60, 1521. 21 Roper, J . Physic. Chem., 1940, 44, 835. Using the data of Frey and Huppke 36, 559-G. B. KISTIAKOWSKY AND A. GORDON NICKLE 181 L RUI No - 29 30 32 33 34 35 42 43 44 36 37 38 39 40 41 Results ETHANE.-Tabk I1 summarizes the results of all ethane runs after No. 28, with the exception of Run 31 in which the sample of dibromides was lost and hence no purity determination could be carried out. Runs preceding No. 29 were either of preliminary nature or were incomplete for one reason or another. The data a t 450' C excepting Run 30, are extremely self-consistent, the mean deviation being only 1.4 yo. These runs differed in direction of approach, contact time, total pressure and in composition. The reproducibility of the equilibrium constants is therefore the best proof that equilibrium was really attained and was correctly measured. The reason for the failure of Run 30 to agree with the rest remains unknown; this run was not included in the average since i t deviates by almost seven times the mean deviation of the others.The data a t 380'C are less extensive and accurate, the mean deviation being about 3 %. The data suggest that the equilibrium was not quite attained from either side, but against this are the data of Runs 36 and 37 which agree with the rest even though the contact time in these runs was only half as long. TABLE II.-SUMMARY OF DATA FOR ETHANE-ETHYLENE-HYDROGEN EQUILIBRIUM Approx. Initial Pressures - C2H4 (Atm X 10-2) nil 4.1 4' 2 5'0 4'1 5'2 1'5 4'9 1'1 0-6 1.1 0.3 nil 1'2 1'2 - HI 10-2) ( Atm X nil 3'1 nil 2.5 3'1 nil 3'8 nil 3'9 nil 0.9 nil nil 1'1 1'2 - %Ha Atm.0.80 0.71 0.80 0.84 0.97 0-71 0.63 0.76 1'00 0.86 0.78 1-02 1'01 0.80 0.78 - Con tacl Tim Per Pas: :Set, 2 2 2 2 I 2 2 2 2 2 2 2 2 2 2 - No. o Passe 5 5 4 3 5 5 7 8 6 5 5 I0 I1 I0 11 Im- purity % of Total Olefin - 3.1 6.3 9.6 9'0 7-7 7.3 6.0 8.6 8.3 6.0 5'0 6.6 6.3 6.2 3'1 C2H4 (Atm. 10-2) X 0.894 0.887 1.404 1'774 1.967 1.440 I -5 06 1.941 2.062 7.78 7'23 6.93 6-20 5'39 5'77 H4 (Atm. 104) X 4'495 3'97 2'994 2'557 2.762 2'591 2.364 2.728 2.088 4'35 4'72 5'79 6-55 5-91 543 - GH6 (Atm.] - 0.780 0.726 0.803 0.838 0'993 0.695 0.663 0.976 0.790 0.85 I 0.783 1.006 1.007 0.780 0.778 KT (Atm.) 5-15 X 10-4 4-85 5'23 5-41 5'47 5'37 5'37 5'43 5'45 3-98 x I O - ~ 4-36 3'99 4'03 4-08 4'33 Temp.("C) 451-2 451'3 451.1 451.5 451'4 451'3 451'4 451'3 451'3 380.5 380.6 380.6 380.6 380.6 380.5 Ko23.2 (Atm.1 4.95 x 10-4 4.63 5-04 5'15 5'23 5'15 5'13 5-20 5.22 K653*2 3.90 x 10-5 4-26 3-91 3'95 3-98 4-23 A careful consideration of several sources of systematic errors, as well as of accidental errors, the magnitude of which is inferred from Table 11, leads us to the conclusion that the standard error of the average value of the equilibrium constant a t 4jo' C, whch is K,,,., = 5-13 x I O - ~ , is less than 2-5 yo. The standard error of the value a t 380° C, K653.2 = 4-04 x I O - ~ , is less than 4 yo. Propane.-The results of all completed runs are shown in Table 111, and i t will be noted that the values obtained in the first few runs are quite low.The catalyst was pumped extensively before the runs previous to No. 52, and an examination of the table shows a distinct separation in values for the equilibrium constant between approaches from the high olefin side and approaches from the low side. In subsequent runs this pumping was dispensed with and the difference has disappeared. The reasons for this effect are not clear, but the important fact is that with the cessation of pumping, practically identical results were obtained on approaches from both sides in several runs. This seems a re asonable criterion of equilibrium. From the data of Table 111 we compute K,,,., = 5-17 x I O - ~ and. K683.2 3-67 x 10-6. Taking into account systematic errors as well as the accidental errors shown by the scatter of the results, we believe that the standard errorI 82 Vo. of 'asses 3 3 3 5 5 g 4 6 5 7 8 ETHANE-ETHYLENE, ETC., EQUILIBRIA Im- purity yo of Tot a1 -- 2-8 4'5 3'3 3'5 2-8 2.5 ' 2'1 2'1 2'0 2'0 1.5 of the average value of the equilibrium constant at 3 7 5 O C is less than 3 yo ; the value at 310' C has a standard error of less than 4-5 yo.Discussion Ethane.-Fig. 4 shows a plot of the equilibrium constants determined here, together with those given in the literature. Since a wide range of values of equilibrium constants is covered, it has been necessary to add a function of temperature (essentially a term of the type AHIT) to the logarithm of the equilibrium constant to get a sufficiently large ordinate scale to allow critical comparison of the results. Their agreement is not perfect, but in view of several sources of error in the earlier work it is surprisingly good.TABLE III.-SUMMARY OF DATA FOR PROPANE-PROPYLENE-HYDROGEN 0.828 0.982 0.984 0.647 0.664 0.814 0.838 0.884 0.918 0.920 04g1 EQUILIBRIUM -___ 5-06 5.01 5-32 5.09 5-24 5.28 5'30 5-21 5'29 3.66 3.66 Run No. - 45 47 48 49 50 51 54 55 56 52 53 Approx. Initial Pressures C3H6 X [Atrn. 10-2) 2'4 3'2 5'6 2'4 3'2 3'6 nil 4'4 0.6 nil 1'2 - H2 10-2) Atm. X nil nil 3'0 nil 3'0 3'1 nil 4'4 nil ni 1 1'2 - C3H8 htm.) 0.86 0.96 0' 66 0.65 0.80 0.87 0.86 0.94 I '00 0.93 0.89 - Con- tact rime per Pass [Sec,) 3'3 3'3 3'3 3' 3 3'3 3'3 3'3 3'3 3'3 3'3 3'3 CsH6 X Atm. xo-2) 2'033 1.942 2.693 2.107 1.649 1.899 1.996 1.962 2-3 2 2 0.653 0.641 H2 (Atrn. xo-2) X 2.061 2.533 x-944 1.563 2.108 2.263 2-225 2.349 2.093 0.516 0.509 Temp.("C) 375'4 375'2 375'6 375'0 375'0 375'2 375'0 375'0 375'0 309'9 310.0 Km-9 (Atm.) 4.98 x I O - ~ 4'97 5-21 5'09 5'24 5-23 5'30 5.21 5'29 K803.2 3.68 x IO-~ 3.66X 1o-I' In the work of Pease and Durgan, Travers and Hockin,22 and Travers and Pearce,s equilibrium was obtained by the heating of near-equilibrium mixtures in silica bulbs for appropriate times. In all this work there are two main potential sources of error. The first is the method of analysis for ethylene, which in no case is entirely above suspicion, but errors would be such as to yield high values for the equilibrium constant. The second is the extent of side reaction to produce (mainly) methane, and the dis- turbing effect this might have on the equilibrium. At 973'K, Pease and Durgan found more than 30 yo methane in the reaction mixture and comment that " the most serious source of error is indicated by the presence of considerable amounts of methane among the products of some of the experiments.The corresponding side reaction undoubtedly has displaced equilibrium to some extent." As low as 863" I<, Travers and Pearce frequently found more than 20 yo methane, but claim that " it is a fact very definitely established by experiment that the equili- brium relations between ethane, ethylene and hydrogen in a mixture initially in equilibrium, and undergoing pyrolysis, remain undisturbed ". They carried out experiments under widely different conditions, and indeed no trend is observable between the value for the equilibrium constant and any other measured quantity.However, at 883OK, the diffsrence between the equilibrium constant obtained from Fig. 4 and 22 Travers and Hockin, Proc. Boy. Soc. A , 1932, 136, I,G. F,. KISTIAKOWSKY AND A. GORDON NICKLE 183 that reported by Travers and Pearce is only about 7 yo. Detailed exam- ination of their results shows variations of f 5 yo in the individual values for the equilibrium constant ; these variations being of about the same magnitude as the effect in question, they could mask a trend between the extent of side reaction and the value of the equilibrium constant. In the work of Frey and Huppke,4 the high values obtained by extra- polation of equilibrium data to infinite contact time have already been recalculated.23 Since total olefins were determined by fuming sulphuric acid and were reported as ethylene, the agreement of the recalculated values with the present work is satisfactory. The work of Vvedenski and Vinnikova 2 4 is subject to many criticisms, their high results being attributable to the inclusion of higher olefins as ethylene. Thus, while the equilibrium constants determined here are not in perfect agreement with those given in the previous literature, we believe that the fault is not ours. 7.32 7.28- FIG. 4.-Comparison of equilibrium data on ethane -ethylene reaction from various sources. Curve I : statistical cal- culations based on AH of Kistiakowsky e f al. ; Curve 2 : statistical calculations based on AH of Prosen and Rossini ; A this re- search; 0 Frey and Huppke, original ; 0 Frey and Huppke, recalculated ; x Travers and Pearce ; @ Travers and Hockin; Pease and Durgan.- 7 24- 7.20 + y 716- - ; I 7.12 7.08 0 I I 7 - x 0 0 700: 1.0 1 1 1 2 1.3 1.4 1.5 T 6 By the use of the van’t Hoff isochore, the heat of hydrogenation of ethylene is calculated from the data of Table I1 as AH,,, = - 34-03 kcal. per mole with an uncertainty of 1-2 kcal. p-r mole. This value agrees extremely well with the value calculated for this temperature from the heat of reaction determined directly at 355°K and the heat capacity data discussed below, which gives AH,,, = - 33-95 kcal. per mole. A more sensitive test of the self-consistency of equilibrium data is obtained by comparing the best value of the equilibrium constant determined here with that calculated statistically from the directly deter- mined heat of reaction and an assortment of molecular constants.These calculations are by now so well known that we shall merely state the authorities whose molecular constants we have used and give the results. 23 Kistiakowsky, Romeyn, Ruhoff, Smith and Vaughan, J . Amer. Chem. SOL, 193.5, 57, 65- 24 Vvedenski and Vinnikova, J . Gen. Chem. (U.S.S.B.), 1934, 4, 120.184 ETHANE-ETHYLENE, ETC., EQUILIBRIA Over several years there have been published by the National Bureau of Standards extensive and self-consistent tables of thermodynamic functions of hydrocarbons. Of these the tables by Rossini, Pitzer, ct aLZ5, 2 6 ~ 27 and by Brickwedde, Moskow and Aston are appropriate for the calculation of the above-mentioned equilibrium constant. Both sets of data give essentially the same results and indeed are based on almost the same physical constants.A new assignment of vibrational frequencies of ethylene which was recently proposed by Arnett and Crawford 20 reduces all calculated equilibrium constants by 4 to 5 yo. In addition to the statistical data, the heat of hydrogenation of ethylene is required. Three values for this, based on modern data, should be considered : (i) - 32,575 cal./mole at 298.16~ K from the direct determination of the heat of hydrogenation of ethylene at 82' C 23 with the original data corrected for the new atomic weight of carbon. (ii) - 32,777 cal./mole at 298.16~ K from the heats of combustion of ethane,a0, 31 ethylene (iii) - 32,732 cal./mole at 298.16~ K estimated by Prosen and Rossini 33 to be the best value for the heat of hydrogenation.In Table IV are shown the values for the equilibrium constants using these heats of reaction and the tables published by Brickwedde, Moskow and Aston 2* for the statistical calculations. The first choice would ob- viously be the correct one, were it not that anharmonicity of vibrations TABLE IV.--COMPARISON OF EXPERIMENTAL AND CALCULATED EQUILIBRIUM CONSTANTS FOR THE REACTION C,H, --f C,H, + H, and hydrogen.27, *2 Heat of Reaction a t 298'i', A H (kcal. ) - 32'575 321777 32,732 Heat of Reaction: at 688'. A H (kcal.) 34'0 f 1'2 33'95 34'15 34-11 I Source Present data Kistiakowsky et al. Rossini et al. Prosen and Rossini 4-04 f 0.17 4-17 3'57 3-70 and the stretching of molecules by rotation were neglected in the statis- tical calculations for ethane and ethylene.The magnitude of the necessary corrections is unknown and hence the final selection among the three choices of Table IV must be postponed, although certain plausi- bility considerations, which we shall now give, indicate the best choice. When the simplifications of rigid rotors and harmonic oscillators are eliminated, the calculated h2a.t capacity of ethane should suffer a greater increase than that of ethylene because of looser structure of ethane and a larger number of vibrations. Thus the chemical potential of ethane will decrease more than the sum of the chemical potentials of ethylene 25 Kilpatrick and Pitzer, J. Res. Nal. Bur. Stand., 1946, 37, 163 ; 1947, 38, 191.26 Pitzer, Ind. Eng. Chem , 1944, 36, 829. 27 Wagman, Kilpatrick, Taylor, Pitzer and Rossini, J . Res. Nut. Bztr. Stand., 1945,34, 143. 28 Brickwedde, Moskow and Aston, J . Res. Nut. Bur. Stand., 1946, 37, 263. 29 Arnett and Crawford, J . Chem. Physics, 1950, 18, 118. 30 Prosen and Rossini, J . Res. Nut. Bur. Stand., 1945, 34, 263. 31 Rossini, J . lies. Nut. Bur. Stand., 1934, 12, 735. 32Rossini, J. Res. Nat. Bur. Stand., 1939, 22, 407. 33 Prosen and Rossini, J . Res. Nut. BUY. Stand., 1946, 36, 269.G. B. KISTIAKOWSKY AND A. GORDON NICKLE 185 and hydrogen and the net result should be an increase in the value of AGO, thus a smaller equilibrium constant for the reaction under discussion. The values calculated for the equilibrium constant using the heats of reaction proposed by Rossini et al.are already too low, so that a more refined treatment of the vibrations and rotations could not improve the agreement but would make it worse. The same reasoning applies a fortiori when the Arnett-Crawford assignment for ethylene is used, since it drops all calculated K's by a few per cent. On re-studying the papers 23, 3% 3% on the heat of hydrogenation of ethylene, we find ourselves unable to agree with Rossini's arguments against the lower value. The same catalyst as was used for ethylene was used for very extensive work with higher olefins, and in no case was any polymerization noted. The extent of polymerization of the higher olefins was certainly less than 0-5 yo and to suppose that ethylene did polymerize to the extent of 3 yo is quite unjustified.As regards the samples of ethylene studied by Kistiakowsky', Romeyn, Ruhoff, Smith and Vaughan,s6 the data on their own purified ethylene still seem to us the best because this ethylene was prepared by a method which excluded ethane, because it was repeatedly distilled, but mainly because the middle and last fractions gave heats of hydrogenation which were identical, within the very small average deviations of the two series of measurements, 44 cal./mole. If ethane was present in the material which was subjected to the final distillation, it would certainly be enriched in the " last fraction. In deference to Rossini's higher figure, we suggest, therefore, the rounded value of AH,,, = - 32,600 (f 50 ?) cal./mole, thus almost within Rossini's own estimated " precision uncertainty " which is in the nature of a probable error and hence leaves considerable probability for a slightly larger deviation than the precision indicated.It also places the Harvard figure, for the ethylene sample obtained from the Linde Air Products Company, as recalculated by Rossini allowing for 0.25 yo ethane, within the probable deviation range of the final value. Using this reaction heat, the calculated equilibrium constant at 723.2" K comes out to be 5-15, or 4.89 if Arnett and Crawford's assignment is used, in good agreement with the experimental value unless the anharmonicity correction is unexpectedly large. Propane .-Fig. 5 shows a plot of experimental and calculated equili- brium constants. As before, skew co-ordinates have been used.The work of Frey and Huppke 4 is obviously in disagreement with this research, and the same objections to their work are applicable here as for ethane, namely, uncertainty in the analytical method for olefin and no proof of the purity of the olefin found. In connection with some work primarily concerned with the rate of polymerization of propane, Travers 36 reported the value 0.085 for the C, equilibrium constant at 826°K. Careful examination of his data shows numerous discrepancies between the equilibrium constant which he gives and those which can be calculated from his reported analytical data. Presuming the analytical data to be correct, an average value of 0-077 is obtained for the equilibrium constant at 826' K and is the value plotted in Fig.5. The individual values scatter considerably, and the mean value could easily be in error by 5 yo. As in ths ethane work, it is possible that polymerization processes have had a disturbing effect on the equilibrium. By use of the van't Hoff isochore, the heat of hydrogenation of propylene is calculated from the data of Table I11 as AH,,, =- 30.5 f 1.2 kcal. per mole. This value agrees well with the value calculated from the heat of reaction determined directly at 355" K. 34Ro;sini, J . Res. Nat. Bur. Staf-zd., 1936, 17, 629. 35 Kistiakowsky, Ruhoff, Smith and Vaughan, J . Amcr. Chem. SOC., 1936, 58, 137. 38 Travers, Trans. Faraday SOL, 1937, 33, 751.I 86 ETHANE-ETHY LENE, ETC., EQU ILIBR IL4 For statistical calculations, adequate tables have been published by Pitzer et aLza, a3, a4 Three values are available for the heat of hydro- genation : (i) - 29,877 cal./mole at 298.16~ K from the direct determination of the heat of hydrogenation of propylene at 355037 with the original data corrected for the new atomic weight of carbon.(ii) - 29,532 cal./mole at 298.16~ K from the heats of combustion of propane,29, 31 propylene l3 and hydrogen.27, aa (iii) - 29,699 cal./mole at 298.16~ K estimated by Prosen and Rossini 33 to be the best value for the heat of hydrogenation. 321 7 28 1 y ! I 7 I 7 121 7 08 1 0 10 1 5 1.8 7 04 1x0 T FIG. 5.-Comparison of equilibrium dath on propane-propylene reaction from various sources. Curve I : statistical calculations based on AH of Prosen and Rossini ; Curve z : statistical calculations based on AH oi Kistiakowsky et al.; A this research; Travers; 0 Frey and Huppke, recalculated. In Table f a r e shown the values for the equilibrium constants cal- culated using these heats of reaction and the above-mentioned tables. None of the values gives as close a check with the experimental value as does ethane, but here the cal- culated values are higher than the experimental. As in the ethane reaction, this is the direction in which errors would be expected in the calculated values arising from the deviations of real molecules from the rigid rotor, harmonic oscillator models. Furthermore, th? greater uncertainty in the molecular models, arising from doubtful frequency assignments and inaccuracies in internal rotation barriers, makes the discrepancy between the cal- culations and the present measure- ments understandable. It prevents a positive identification of the best ” value for the heat of this reaction. The following considera- tions, however, suggest that the best value is close to - 29,877 cal. determined in hydrogenation ex- periments. Since Rossini’s value is lower, its difference from the hydrogenation experiments cannot be explained by such assumptions as the polymerization of olefin or the presence of saturated impurity in hydrogenation work, but can only be due to the presence of a more unsaturated impurity in the latter. In view of the two different methods of preparation, the extensive purification and the close agreement between values obtained for the heat of hydrogenation of the two samples of proyylene, this source of error does not seem reasonable. The clean- ness of the roaction and freedom from cracking during hydrogenation were amply demonstrated by appropriate methods. Systematic error of the calorimeter or calorimetric technique is ruled out by the close agree- ment between the values obtained by Rossini et al. and Kistiakowsky et U Z . ~ ~ for the formation of water from the elements. Thus, since we can find no adequate source of error in the hydrogenation value, and since 37 Kistiakowsky, Ruhoff, Smith and Vaughan, J . Amer. Cjiem, Sac., 1935, 58,876.G. B. KISTIAKOWSKY AND A. GORDON NICKLE 187 this value is within the estimated uncertainty of that determined by Rossini et al., we believe the former to be more nearly correct and recommend AH,,, =- 29,850 f 50 cal. as the final choice. KSP 9. I , Heat of Reytion Heat of Reaction at 298.1 at 616O, Source AH (kcal:) AH (kcal.) Atm. x 104 - 30.5 & 1-2 Present data 5-17 0.15 29377 30.85 Kistiakowsky 6.04 299532 30.50 Rossini ef al. 7-89 29,699 30'67 Prosen and 6.92 et al. Rossini TABLE V.-~OMPARISON OF EXPERIMENTAL AND CAIXU LATED EQUILIBRIUM CONSTANTS FOR THE REACTION C,H, -+ C,H, + H, Kba3.2, Atm. x 105 3-67 + 0-17 4-15 5-58 4-85 Gibbs Chemical Laboratory, Harvard University, Department of Chemistry, 12 Oxford Street, Cambridge 38, Mass., U.S.A.