KINETICSOF
THE
CYCLOPROPANE-HYDROGEN REACTION
1877
Kinetics of the Reaction of Cyclopropane with Hydrogen over a Series of Silica-Supported Metals
by J. H. Sinfelt, D. J. C. Yates, and W. F. Taylor Process Research Division, Esso Research and Engineering Company, Linden, N e w Jersey (Received December 4 , 1964)
The kinetics of the reaction of cyclopropane with hydrogen were investigated over a series of silica-supported metal catalysts : nickel, cobalt, platinum, and copper. The catalysts all contained 10 wt. % of metal, and values of the metal surface area were available from previously reported hydrogen chemisorption measurements. Over nickel and cobalt, the reaction products included methane and ethane in addition to propane. Over platinum and copper, the only product observed was propane. Since the surface areas of the metals were known, it was possible to determine specific catalytic activities and hence to allow for differences in activity arising from variations in the degree of metal dispersion on the support. The order of catalytic activity of the metals for the hydrogenation of cyclopropane to propane is Ni > Pt > Co > Cu.
The reaction of cyclopropane with hydrogen has been investigated over a variety of supported metal catalysts, primarily nicke11v2 and certain of the noble group VI11 metal^,^-^ including rhodium, palladium, iridium, and platinum. The reaction product observed in all these studies was limited to propane, there being no evidence for the formation of other products such as methane and ethane. However, recent studies on nickel films’ and platinum blackss have disclosed the formation of the latter products as well as propane. Recently, we have been interested in the kinetics of the reaction of cyclopropane with hydrogen over a series of silica-supported metals. Kinetic data hatie been obtained over two of the previously studied metals, nickel and platinum, and have been extended to include cobalt and copper. Values of the surface area of each of the supported metals have been obtained by hydrogen chemisorption measurements, and it has been possible to determine the specific catalytic activity (activity per unit surface area) of each of the metals. Consequently, the catalytic activities have been compared on a more fundamental basis than is commonly done since effects due to differing degrees of metal dispersion on the support are taken into account. A previous study of the reaction of cyclopropane with
hydrogen over supported nickel catalysts has shown that the products of the reaction include methane and ethane, as well as propane.1° Therefore, an item of some interest in the present work was the effect of varying the metal on the distribution of reaction products. In this paper the reaction leading to the formation of propane is termed hydrogenation, while the reaction yielding methane and ethane is termed hydrogenolysis. Kinetic information on both reactions, including pressure dependencies with respect to cyclopropane and hydrogen as well as the dependence on temperature, have been obtained. The results have extended our knowledge of the interaction between cyclopropane and hydrogen over metal catalysts and (1) G.C.Bond and J. Sheridan, Trans. Faraday SOC.,48, 713 (1952). (2) J. E.Benson and T. Kwan, J. Phys. Chem., 60,1601 (1956). (3) J. Addy and G. C. Bond, Trans. Faraday SOC.,53, 368 (1957). (4) J. Addy and G. C. Bond, ibid., 53, 383 (1957). (5) G.C. Bond and J. Turkevich, ibid., 50, 1335 (1954). (6) G.C. Bond and J. Newham, ibid., 56, 1501 (1960). (7) 2. Knor, V. Ponec, 2. Herman, 2. Dolejsek, and S. Cerny, J. Catalysis, 2, 299 (1963). (8) D.W. McKee, J . P h y s . C h m . , 67, 1336 (1963). (9) J. H. Sinfelt, W. F. Taylor, and D. J. C. Yates, ibid., 69, 95 (1965). (10) W.F. Taylor, D. J. C. Yates, and J. H. Sinfelt, J. Catalysis, in press.
Volume 69, Number 6 J u n e 1966
1878
have given some further insight into the nature of catalysis over supported metals.
Experimental Apparatus and Procedure. The kinetic data were obtained in a flow reactor system at atmospheric pressure, using a vertically mounted stainless steel reactor tube 1.0 em. in diameter and 8.0 cm. in length. Details of the reactor assembly, flow rate measurements, and the gas chromatographic analysis of the reaction products have been reported previously.ll The cyclopropane and hydrogen were mixed with helium and passed downflow through a bed containing 0.20 g. of catalyst diluted uniformly with 0.50 g. of ground Vycor glass. By appropriate adjustment of the helium flow rate, it was possible to vary the partial pressures of cyclopropane and hydrogen individually. The total gas flow was maintained at 1 l./min. throughout. In a typical run, the reactant gases were passed over the catalyst for 3 min. prior to sampling products for analysis. The cyclopropane was then cut out and the hydrogen flow continued for 10 min. prior to another reaction period. As an insurance against possible complications due to changing catalyst activity, most of the reaction periods were bracketed by periods at a standard set of conditions, so that the kinetic data could be expressed as rates relative to the rate a t the standard conditions. Prior to any reaction rate measurements, the catalysts were reduced overnight in 50 cc./min. of flowing hydrogen at 370" in the reactor. The surface areas of the supported metals used in this work were obtained from hydrogen chemisorption data and also carbon monoxide chemisorption in the case of the copper catalyst. The metal areas have been reported previously. Details of the chemisorption measurements used to estimate the metal surface areas can be obtained from previous papers by the authors. 9,12 The surface areas were determined after the catalyst had been reduced at the same conditions employed in reducing the catalyst in the reactor. Materials. The supported metal catalysts used in this work all contained 10 wt. % of metal and were prepared by impregnation of silica (Cabosil) with the nitrate salt pf the metal or, in the case of platinum, with Pt(NH3)2(N02)z. The details of preparation of all of these catalysts have been reported previou~ly.~ The cyclopropane was obtained from the Matheson Co. ; a chromatographic analysis showed no detectable impurities. It is estimated that an impurity, e.g., methane, would have been detected by the chromatographic analysis if it were present a t a concentration above 0.01 wt. %. High purity hydrogen was obtained from the Linde Co., Linden, PIT. J. It was furThe Journal of Physical Chemistry
J. H. SINFELT, D. J. C. YATES,AND W. F. TAYLOR
ther purified in a Deoxo unit containing palladium catalyst to remove trace amounts of oxygen. The water formed was then removed by a molecular sieve dryer.
Results As pointed out in the previous section of the paper, values of the metal surface areas of the catalysts were available from hydrogen chemisorption measurements. Expressed per gram of catalyst, the surface areas are: Nil 13.6 m.2; Co, 5.6 m.2; Pt, 4.4m.2; Cu, 3.3 m.2. The reaction of cyclopropane with hydrogen was studied at low conversion levels (0.1 to 10%). Rates were calculated from the relation
r
=
(F/W)z
(1)
where F represents the feed rate to the reactor in moles of cyclopropane per hour, W represents the weight in grams of the catalyst, and x represents the fraction of cyclopropane converted. In the determination of the rate of hydrogenation, x represents the fraction converted to propane, whereas, in the case of hydrogenolysis, x represents the conversion to methane and ethane. Reaction rates are correspondingly expressed as moles of cyclopropane converted per hour per gram of catalyst to propane or to methane plus ethane. In the case of the hydrogenolysis reaction, which was observed only over the nickel and cobalt catalysts, the methane and ethane were formed in equimolar portions. This is shown by the typical data in Table I giving the distribution of methane and ethane in the products. Since the methane and ethane are formed
Table I : Distribution of Methane and Ethane in Products Total conversion of cyclopropane,
Distribution of methane and ethane, mole %" Methane Ethane
Catalyst
Temp., OC.
%
Ni-Si09 Co-Si02
40 121
2.2
48.4
51.6
4.6
50.5
49.5
Cyclopropane pressure 0.030 atm. ; Hz pressure 0.20 atm.
in equimolar proportions, the rate of hydrogenolysis can be determined simply from the amount of either one of these compounds in the product. The methane and ethane appear to be primary reaction products, rather (11) J. H. Sinfelt, J . Phys. Chem., 68, 344 (1964). (12) D. J. C. Yates, W. F. Taylor, and J. H. Sinfelt, J . Am. Chem. Soc., 86, 2996 (1964).
KINETICS OF
THE
1879
CYCLOPROPANE-HYDROGEN REACTION
c
3
F: 10-4 2.0
~
, 2.2
2.4
1 2.6
, 2.8
1 3.0
3.2
,
,
1
~
3.4
1000/T (OK.).
Figure 1. Effect of temperature on rate of hydrogenation of cyclopropane to propane a t p~ = 0.20 atm., p c = 0.030 atm.: 0 , Ni; A, Pt; W, Co; 0, Cu.
10 -6 2.0
2.2
2.4
2.6
2.8
3.0
3.2
3.4
lOOO/T (OK,).
than secondary products arising from a further reaction of the propane formed. Data obtained on the hydrogenolysis of propane over the nickel catalyst lo have shown that the rate of conversion to methane and ethane is much too low to account for the rate of formation of these products from cyclopropane via a gas phase propane intermediate. In a run to measure reaction rates, the catalyst was first prereduced with hydrogen overnight a t 370". Then the temperature was lowered in flowing hydrogen, and at a standard set of hydrogen and cyclopropane ~ 0.20 atm., pc = 0.030 atm.) partial pressures ( p = the activity of the freshly reduced catalyst was measured. Then rates were measured a t a series of temperatures in a rising temperature sequence. The data for the four catalysts are shown in the Arrhenius plots in Figures 1 and 2. The data in Figure 1 are for the hydrogenation to propane, while the data in Figure 2 are for the hydrogenolysis to methane plus ethane. As already pointed out, the hydrogenolysis to methane and ethane was not observed over the platinum and copper catalysts over the range of temperatures studied. The order of catalytic activities for hydrogenation to propane is clearly Ni > Pt > Co > Cu. In the case of the hydrogenolysis reaction to methane and ethane, the order is Ni > Co > Pt or Cu. In the measurement of the catalytic activities of the various supported metals, there was some overlap in the temperature ranges in which the platinum, cobalt, and copper catalysts were investigated, so that part of the data on each of these catalysts was obtained a t about the same temperature. In the case of the nickel catalyst, however, there was no overlap in temperature range with the other catalysts, the temperatures being lower
Figure 2. Effect of temperature on rate of hydrogenolysis of cyclopropane to methane and ethane a t pa = 0.20 atm., p c = 0.030 atm.: 0, Ni; W, Co.
Table I1 : Effect of Cyclopropane and HP Pressures on Rates of Hydrogenation and Hydrogenolysis p T / T Q a -
PC, atm.
PH, a h .
Hydrogenation
Hydrogenolysis
Ni-Si02 (27')
0.030 0.030 0.030 0.010 0.030 0.100
0.10 0.20 0.40 0.20 0.20 0.20
1.13 1.00 0.95 0.41 1.00 2.59
1.21 1.00 0.77 0.63 1.00 1.36
Co-Si02 (121')
0.030 0.030 0.030 0.010 0.030 0.100
0.10 0.20 0.40 0.20 0.20 0.20
1.03 1.00 0.93 0.48 1.00 1.77
1.21 1.00 0.89 0.47 1.00 1.81
Pt-SiOz (79')
0.030 0.030 0.030 0.010 0.030 0.100
0.10 0.20 0.40 0.20 0.20 0.20
1.33 1.00 0.70 0.30 1.00 2.83
... ... ...
0.030 0.030 0.030 0.010 0.030 0.100
0.10 0.20 0.40 0.20 0.20 0.20
1.11 1.00 1.00 0.48 1.00 2.44
...
Catalyst
Cu-Si02 (146')
...
*.. *..
... ... ...
... ...
Rate relative to the rate a t standard conditions (PO = 0.030 atm., PH = 0.20 atm.) for the particular catalyst and temperature in question. The T / T O values cannot be used by themselves to compare the activities of the catalysts.
Volume 69,Number 6 June 1966
J. H. SINFELT,D. J. C. YATES,AND W. F. TAYLOR
1880
Table 111: Summary of Kinetic Parameters for the Reaction of Cyclopropane with HOover the Various Supported Metals Catalyst
Temp. range, OC.
Ni on Si02 Pt on Si02 Co on Si02 Cu on Si02
32-42 72-109 93-158 113-207
I
E6
nd
13.0 12.2 10.7 10.9
0.8 1.0 0.6 0.7
Hydrogenationu me
-0.1 -0.5 -0.1 -0.1
r‘f
2.8 X 1.0 x 1.6X 4.1 X
loz2 1021 IO’@ 1OI8
‘
EC
nd
16.0
0.4
Hydrogenolysisb me
-0.3
...
...
18.7
0.6
-0.2
*..
...
...
...
T’f
3 . 0 X lo2& 1.1 x 1028
a Hydrogenation of cyclopropane to propane. Apparent activation Hydrogenolysis of cyclopropane to methane and ethane. Pre-exponential factor in the exExponent on cyclopropane pressure. e Exponent on hydrogen pressure. energy, kcal./mole. pression, r = T ’ exp( - E / R T ) , which expresses the temperature dependence of the rate r a t standard conditions ( p =~ 0.20 atm., p c = 0.030 atm.). The units in which r‘ is expressed are molecules/(sec.
throughout the range investigated. This itself attests to the high activity of the nickel catalyst. If we compare the temperatures required to obtain a given measured rate over the various catalysts (by simply drawing a horizontal line across the Arrhenius plots in Figures 1 and 2 and noting the temperatures a t the points of intersection with the experimental lines for each of the catalysts), it is clear that nickel is by far the most active of the metals. After determining the effect of temperature on rates, the temperature was lowered to an intermediate value, and a series of measurements using the “bracketing technique” was made to determine the effects of the partial pressures of hydrogen and cyclopropane, p~ and pc, respectively, on the rates. For each set of conditions the rate r relative to the rate ro a t the standard conditions ( p c = 0.030 atm., p~ = 0.20 atm.) was expressed as a ratio r/ro. These data are presented in Table 11. For all four catalysts the data in Table I1 show that the rate of hydrogenation t o propane increases with increasing cyclopropane partial pressure but decreases slightly with increasing hydrogen partial pressure. In the case of hydrogenolysis, observed only over the nickel and cobalt catalysts, similar dependencies on the cyclopropane and hydrogen partial pressures are observed, except that the inverse effect of hydrogen pressure is somewhat greater than is found for the hydrogenation. The dependence of the rates of hydrogenation and hydrogenolysis on the partial pressures of cyclopropane and hydrogen can be expressed in the form of a simple power law, r = kpcnpRm. Approximate values of the exponents n and m are given in Table 111. The uncertainty in these exponents is estimated to be *0.1. Values of the apparent activation energy E and the pre-exponential factor T’ in the expression r = r’ exp(-E/RT), representing the temperature dependence of the rate r a t standard conditions ( p =~ 0.20 atm., pc = 0.030 atm.), are also The Journal of Physical Chemistry
’
listed in Table 111. The values of the pre-exponential factor r’ are expressed as molecules/(sec. cm.2). The apparent activation energies vary only slightly from metal to metal, and the differences may not be significant. The uncertainty in the activation energies is estimated to be about 1 kcal./mole except for the nickel catalyst. The activation energy in the case of the nickel is probably less precise since the data on nickel were obtained over a rather narrow temperature range compared to the data on the otherinetals.
Discussion The order of catalytic activities of the silica-supported metals for the hydrogenation of cyclopropane to propane is Ni > Pt > Co > Cu. While there is a slight variation in the apparent activation energy over these metals, the differences in activity are due primarily to differences in the pre-exponential factor of the rate expression, as shown in Table 111. This is analogous to results reportedI3 for the hydrogenation of ethylene over a series of silica-supported metals, in which the variation in activity was also due to differences in the pre-exponential factor. The order of activities of the metals for cyclopropane hydrogenation is roughly similar to the order observed for ethylene hydrogenation, except that nickel and platinum have been found to have about the same activity for the hydrogenation of ethylene. In the current work, it should be noted that the catalysts were prepared to contain the same amount of metal by weight, whieh means that the number of metal atoms per gram of silica support in the case of the platinum is less than one-third that of the other three catalysts. The nickel, cobalt, and copper catalysts all contain about the same number of metal atoms per gram of silica support, the variation among them amounting to less than 10%. While the catalytic activities have been determined per unit (13) G. C. A. Schuit and L. L. van Reijen, Advan. Catalysis, 10, 242 (1958).
KINETICSOF
THE
CYCLOPROPANE-HYDROGEN REACTION
metal surface area, it is possible that the activity per unit area increases with the amount of platinum present. This type of effect has been reported in the literature14 and has also been observed recently by us for similar supported metals. Thus, it may be that the specific activity of the platinum would be closer t o that of the nickel if the catalysts were compared a t the same concentration of metal atoms per unit weight of silica support. The observation that the order of catalytic activities of the various supported metals for cyclopropane hydrogenation is roughly similar to the order which has been reported for ethylene hydrogenation suggests that the underlying factors determining the catalytic activities of the metals for these reactions are similar. Cyclopropane thus appears to behave somewhat as an olefin in its properties, which is consistent with other evidence, both experimental and theoretical,15 attesting to the unsaturated nature of the cyclopropane ring. While the hydrogenation of cyclopropane is somewhat analogous to the hydrogenation of ethylene, there is a distinct difference in the detailed kinetics of the reactions, as pointed out by Bond and Sheridan.’ Thus, while ethylene hydrogenation is usually observed to be zero order in ethylene and first order in hydrogen, the hydrogenation of cyclopropane shows a definite positive dependence on cyclopropane pressure and is either independent of, or decreases slightly with, hydrogen pressure. These results indicate that cyclopropane is adsorbed less strongly than ethylene on metal catalysts. It seems reasonable that the lower strength of adsorption of cyclopropane relative to ethylene is related to the idea that the delocalized electrons of the cyclopropane ring are less available for bonding with orbitals of the metal than are the xelectrons of the ethylene. While the order of activities of the various metals for the hydrogenation of cyclopropane to propane is roughly similar to the order for ethylene hydrogenation, this is not the case for the hydrogenolysis reaction leading to methane and ethane. Here the cobalt catalyst ranks second to nickel, and the hydrogenolysis reaction is not even observed over the platinum and copper catalysts. These results closely parallel our previous findings on ethane hydrogenolysis, in which nickel was found to be more active than cobalt, which in turn was much niore active than either platinum or copper. The activity of nickel for ethane hydrogenolysis was found to be as much as lo5 to lo6 times as high as that of platinum or copper, but the relative activities did vary substantially with temperature. It would seem that for the hydrogenation reaction the cyclopropane behaves like an olefin with regard to
1881
the activities of the various metals, whereas, in the case of the hydrogenolysis reaction, it appears to behave like a paraffin. It is conceivable that the cyclopropane is adsorbed in different forms on the surface and that the hydrogenation and hydrogenolysis reactions involve different adsorbed species. Possibly, the hydrogenation reaction might proceed through a r-bonded intermediate,6 while the hydrogenolysis could conceivably involve a dissociative adsorption step prior to rupture of carbon-carbon bonds, analogous to the dissociative chemisorption which appears to be the initial step in the hydrogenolysis of paraffins such as ethane and propane over metals.I6 The possibility that cyclopropane is chemisorbed in several configurations on the surface is not unreasonable in view of the fact that evidence exists for both associative and dissociative adsorption of a molecule such as ethylene on metal surfaces.17 It is interesting that the supported copper catalyst shows significant activity for the hydrogenation of cyclopropane to propane although it is less active than the other metals. A commonly held impression of hydrogenation catalysis over metals is that the catalytic properties are due to the existence of partly filled dbands which are available for b ~ n d i n g ,which ~ ~ , ~ef~ fectively limits hydrogenation catalysts to the transition metals. Since copper does not have a partly filled d-band, its activity as a hydrogenation catalyst appears to be somewhat inconsistent with this hypothesis. However, copper immediately follows a transition metal series in the periodic table, and it may be that electrons can be promoted from 3d t o 4s states20g21 a t conditions under which chemisorption and catalysis take place. In previous studies1~2~6 on the hydrogenation of cyclopropane to propane over supported metals, it has been concluded that the rate-limiting step in the reaction involves the interaction of an adsorbed cyclopropane molecule with an adsorbed hydrogen atom. It has also been concluded that hydrogen is more strongly adsorbed than cyclopropane. In the present work it
(14) F. N. Hill and P. W. Selwood, J . Am. Chem. Sac., 71, 2522 (1949). (15) M. Yu Lukina, Russ. Chem. Rev., 31, 419 (1962). (16) A. Cimino, M. Boudrtrt, and H. S. Taylor, J . Phys. Chem., 58, 796 (1954). (17) G. C . Bond, “Catalysis by Metals,’’ Academic Press, New York, N. Y., 1962, pp. 229-236. (18) hl. Boudart, J . Am. Chem. SOC.,72, 1040 (1950). (19) 0. Beeck, Discussions Faraday SOC.,8, 118 (1950). (20) M. Boudart, J. Am. Chem. Soc., 74, 1534 (1952). (21) E. M. W. Trapnell, “Chemisorption,” Butterworth Scientific Publications, London, 1955, p. 174.
Volume 69, Number 6 June 1966
R. J. BERNI,R. R. BENERITO, AND H. M. ZIIFLE
1882
is interesting that the inverse dependence of the rate on hydrogen pressure is greater for the platinum catalyst than for the other catalysts. This suggests that the strength of adsorption of hydrogen on the platinum catalyst is greater than on the other catalysts. To summarize briefly, the results of the present work have made it possible to compare the catalytic
activities of a series of supported metals for the reaction of cyclopropane with hydrogen in a more fundamental manner than has been done previously since the metal surface areas were known from chemisorption measurements. In addition, it has been shown that the selectivity of conversion of the cyclopropane depends markedly on the particular metal chosen as a catalyst.
Kinetics of the Zinc Fluoroborate and Hydrogen Ion Catalyzed Hydrolyses of the Diglycidyl Ether of 1,4=Butanediol and of Diglycidyl Etherla
by Ralph J. Berni, Ruth R. Benerito, and Hilda M. Ziifle Southern Regional Research Laboratory,lb New Orleans, Louisiana
(Received December 7, 1964)
Kinetics of hydrolyses of the diglycidyl ether of 1,4-butanediol (DGEBD) and of diglycidyl ether (DGE) a t less than 1.0 M concentrations catalyzed by HC1 (pH 3.25) and by 0.05 M Zn(BF02 from 25 to 90" have been investigated. Semilogarithmic plots of changes in oxirane content with time indicated pseudo-first-order kinetics with both catalysts. Statistical evaluation indicated different rate constants for the opening of the first (kl) and second (k2)oxirane rings with Zn(BF& catalysis but equivalent rates with HCl catalysis. Analysis of data by the modified Swain method confirmed consecutive first-order kinetics for the two diepoxides with both catalysts. Further analysis of data with HC1 catalysis confirmed equivalent rate constants for both ring openings. With the exception of DGE a t go", relative rates (Ic2/kl)for the over-all reaction with Zn(BFS2 catalysis were less than unity at all temperatures. These over-all rate constants could not be corrected for the Hfion rate due to similarity in rates of ring opening with both catalysts. Enthalpies, entropies, and free energies of activation for ring openings were calculated. Large negative entropies ~ a t these low levels of catalysts. of activation indicated ring openings by S N mechanisms
Introduction Specific reaction rate constants for the hydrolyses of the diglycidyl ether of 1,4-butanediol (DGEBD) and of diglycidyl ether (DGE) at less than 1.0 M concentrations in the presence of 0.05 M zinc fluoroborate, Zn(BF4)2, were not available in the literature, nor were such data for the H + ion catalyzed hydrolyses of these diepoxides at pH 3.25, which was the pH value of the Zn (Bf)z-diepoxide solutions. Knowledge of these rates and their temperature coefficients was essential The Journal of Physical Chemistry
to the elucidation of certain cellulose diepoxide reactions which are catalyzed by dilute Zn(BFJ2 but not by dilute HCI. This paper presents a study of the kinetics of hydrolysis of the DGEBD and DGE when catalyzed by Zn(BF4)2 at pH 3.25 and by HC1 at pH 3.25. Rates (1) (a) Presented in part before the Division of Inorganic and Physical Chemistry at the Southeastern Regional Meeting of the American Chemical Society, Charleston, W. Va., Oct. 15-17, 1964. (b) One of the laboratories of the Southern Utilization Research and Development Division, Agricultural Research Service, U. S. Department of Agriculture.
