HYDROGENATION

OF ETHYLENE
AND PROPYLENE

2543

Hydrogenation of Ethylene and Propylene over Palladium Hydride

by R. J. Rennard, Jr., and R. J. Kokes

Department of Chemistry, The Johns Hopkins University, Baltimore, Maryland
(Received February 26, 1966)

21218

The rate of ethylene hydrogenation as a function of hydrogen concentration, temperature, ethylene pressure, and hydrogen pressure has been studied over palladium hydride
and palladium deuteride. Similar (but less extensive) studies have been carried out with
propylene. The rate of hydrogenation at -78" is found to be nearly zero order in ethylene
and hydrogen pressure but first-order in the hydride concentration. Although the firstorder rate constant decreases with hydride concentration, the activity increases. There
is an inverse isotope effect with deuterium, and the principal deuterated product is CzH4Dz.
Analysis of the data suggests that the slow step is the addition of adsorbed hydrogen
atoms to adsorbed ethylene or adsorbed ethyl radicals.

Introduction

Experimental Section

Hydrogen, deuterium, and helium were purified by

passage through a charcoal trap a t -195". Ethylene
and propylene (CP grade) were fractionated prior to
use and checked for purity by gas chromatography.
Unless otherwise noted, before any of the experiments, the palladium sample was heated for 16 hr in
200 mm of hydrogen a t 450", degassed for 0.5 hr at
this temperature, and cooled in helium to -78".
At all times, the palladium was protected from mercury
vapor by a trap a t -78". Palladium hydride or deuteride was formed at -78" by sorption from the gas
phase.
Kinetic studies of the hydrogenation of ethylene
were carried out on 320 mg of palladium admixed with
1.5 g of powdered quartz, both 40-60 mesh. This
mixture was spread over the bottom of a 30-cc conical
flask connected via a capillary stopcock to a vacuuni
system. After pretreatment of the catalyst, prepara-

Palladium powder was prepared by a method similar

to that used by Gillespie and
An aqueous solution of palladium chloride (10%) was treated with
ammonia and then with hydrochloric acid. The salt
formed was reprecipitated several times and then
reduced to metallic palladium by slowly heating to 500"
in a stream of hydrogen. Then, the sample was flushed
with helium, cooled, and washed with hot distilled
water and concentrated ammonia. After this, the reduction at 500" was repeated. This procedure yielded
a palladium sponge with a B E T surface area of about
0.4 m2/g.

(1) (a) G. Reinacker and E. A. Bommer, 2. Anorg. Allgem. Chem.,

236, 263 (1939); (b) G. Reinacker, E. Muller, and R. Burmann.
ibid., 251, 55 (1943).
(2) D. A. Dowden and P. W. Reynolds, Discussions Faraday Soc.,
8, 184 (1950).
(3) A. Couper and D. D. Eley, ibid., 8, 172 (1950).
(4) M. Kowaka, N i p p o n Kinzoku Gakkaishi, 23, 625 (1959).
(5) R. J. Best and W. W. Russel, J . Am. Chem. Soc., 7 6 , 834 (1954).
(6) P. H. Emmett and W. K. Hall, J . Phys. Chem., 6 2 , 817 (1958).
(7) For a recent review see G. C. Bond, "Catalysis by IIetills,"
Academic Press Inc., New York, N. Y., 1962, pp 244-252.
(8) L. J. Gillespie and F. P. Hall, J . Am. Chem. Soc., 48, 1207
(1926).

Numerous studies'-? have been made on catalytically

active alloy systems in which the composition is systematically varied in an attempt to correlate activity
changes to known changes in solid-state properties.
In such studies, the activities of a series of different
preparations are compared (with6p6or without' corrections for differences in surface areas) on the assumption that the different preparative procedures, required
for different compositions, have only a trivial effect on
the activity. Similar studies on the palladium hydride system offer the possibility of carrying out such
comparisons on a single palladium sample. Such a
study of the hydrogenation activity of palladium as a
function of hydride concentration is the subject of this
report.

Volume 70, Number 8

August 1866

R. J. RENNARD,
JR.,AND R. J. KOKES

2544

tion of the hydride, and temperature adjustment, the

reactant gas was admitted. From time to time a small
sample of the reactant gas (1.35%) was withdrawn
for chromatographic analysis on an alumina column.
Preliminary tests showed that the sampling was representative.

Results
Adsorption Studies. Isotherms were determined
from 0 to 400" at pressures up to 500 mm. Above 100"
only the o phase was present and no hysteresis was observed; below 100" the a-p transition was present
and hysteresis was evident. These results were consistent with those reported by others.*-1 Even though
the hysteresis is pronounced a t room temperature,
however, the hydrogen could be completely removed
by room temperature evacuation. (For PdH0.28 complete degassing required 1 hr; for PdHo.42 complete
degassing required 3 hr.) Furthermore, even when
hysteresis occurs, the pressure rapidly adjusts to a
steady quasi-equilibrium value.
Hydride formation with palladium was also detected
at -78, -183, and even -195". At the lower temperatures, sorption is slow. At -78" with inlets of
hydrogen corresponding to compositions up to PdHO.60,
the half-time for sorption is of the order of 5 sec, the
residual pressure is negligible, and a 30-min evacuation
removes less than 0.05% of the hydrogen (from PdH0.24).
Surface area measurements on a pure palladium
sample yield a value of 0.81 m2, whereas those for a
PdH0.42sample prepared from the same sample yielded
a value of 0.79 m2. Further studies revealed that
neither the standard pretreatment nor reformation of
the hydride was accompanied by a detectable change
in area.
Figure 1 shows the pressure fall accompanying the
sorption of hydrogen at -78" by PdHo.24from an equimolar mixture of hydrogen and ethylene or hydrogen
and propylene. (Analysis of the gas phase shows that
about 10% of the pressure fall could arise from alkane
production.) I n the absence of olefin a t this hydrogen
pressure (107 mm), the half-time for sorption would be
about 5 sec; hence, the presence of olefin decreases the
rate by about two orders of magnitude.
Reaction with Hydrogen. Because of the rapid sorption of hydrogen compared to reaction it was only
possible to study the reaction of a hydrogen-olefin
mixture in which sorption was also occurring. To
this end, the amount hydrogenation of a 50:50 hydrogen-olefin mixture over PdHo.24was compared to
that of a 50:50 helium-olefin mixture. The results
are summarized in Table I. It appears from these
data that the rate of reaction decreases from 20 to
The JOUTnaE gf Physical Chemistry

I .4

12

16

Time (min.)

Figure 1. Adsorption from an olefin-hydrogen mixture at

-78O: open circles, 50:50 CaH&:Hz,P (total) = 214
mm; closed circles, 50:50 C2H4:H2, P (tot'al) = 214 rnm.

40% when the pressure of hydrogen decreases by four

or five orders of magnitude. In other words, the reaction is essentially zero order with respect to hydrogen
for reaction over palladium hydride.
Table I: Hydrogenation over PdHo.24
Initial

Final

Mixture

P H ~mm
,

P H ~m, m

CaHsHz
CtHe-He
CzHrHz
CzHa-He

107

30

. . .a

. . .a

107

34

. . .a

...

pmoles of
paraffinb

3.6
3.1
24
15

On the basis of the residual pressure after sorption, this

value would be of the order of 10-3 mm of Hz. Amount formed
after 18 min. In this time, about 200 pmoles of hydrogen was
taken up by the catalyst.

'

The effect of ethylene pressure on the reaction with

two samples of palladium hydride is shown in Table 11.
These results show that the reaction is nearly, but perhaps not quite, zero order in ethylene pressure.
The kinetics of the reaction of ethylene with palladium hydride can be represented by the equation:
In C/Co = -kt, where Co and C represent the hydrogen
content of the catalyst at t = 0 and time t, respectively,
and k is a pseudo-first-order rate constant that depends
(9) D. M. Nace and J. 0. Aston, J. Am. Chem. SOC.,79, 3619, 3623,
3627 (1957).
(10) D. P. Smith, "Hydrogen in Metals," University of Chicago
Press, Chicago, Ill., 1948.

HYDROGENATION
OF ETHYLENE
AND PROPYLENE

2545

Table I1 : Ethylene Hydrogenation over Palladium Hydride

Sample

PCZHII,
mm

rmoles of
ethane/hr

PdHo.396
PdHo.396
PdHo.nas
PdH0.m
PdH0.239

449.6
156.3
156.3
74.5
39.3

50.0
50.0
39.4
36.0
34.5

-+

*\

on Co. Figure 2 illustrates the fit of the kinetics to

this equation for several values of Co. It is evident
from these plots that this equation is valid for as much
as three half-lives.
Figure 3 illustrates the dependence of this rate constant on concentration by a plot (for convenience)
of the half-life vs. hydrogen concentration for two different samples of palladium. The results for the two
different batches of catalyst are nearly the same, and
for both samples the rate constant decreases with increasing hydrogen content. It should be noted, however, that the activity actually increases with hydrogen
content. This is illustrated in Figure 4 which shows
a plot of the ethane formed in 1 hr os. the hydride
concentration.
Reaction of propylene with palladium hydride was
determined a t -78" for PdHo.z3and PdHO.41. These
results could also be represented by the pseudo-firstorder equation. The ratios of rate constants for ethylene to those for propylene were 8 for PdHo.Z4and 9
for PdHo.41. Similar relative rates for propylene and
ethylene have been reported for conventional hydrogenation reactions."
Effect of Pretreatment. Kowaka4 reported that pretreatment of palladium with hydrogen reduces the
activity for ethylene hydrogenation. Details of the
pretreatment are not clear; hence, we investigated
the effect of several pretreatments on the activity.
The results are summarized in Table 111. Various
pretreatments were as follows.
A. The sample was degassed a t 400", then cooled
in sufficient hydrogen to form PdH0.23a t -78".
B. The sample was degassed a t 400" and cooled
to room temperature in sufficient hydrogen to form
PdHo.23. This hydrogen was removed by evacuation
a t room temperature and PdHo.23 was reformed a t
-78".
C. The sample was degassed a t 400" and cooled
to room temperature in enough hydrogen to form
PdHo.ol. It was then cooled to -78" and enough
hydrogen was added to form PdHo.23.

0.4
A\

0.3

A\

60

120

240 301

180

Time (min.)
Figure 2. Kinetic plots for ethylene hydrogenation: closed
circles, PdHo.40; open circles, PdHo.2,; triangles, PdHo.02,.

I
1000

l
/

0.12

0.24

0.36

0.40

0.6

H/Pd
Figure 3. Hydrogenation half-life as a function
of hydride composition: closed circles,
catalyst I; open circles, catalyst 11.

Table 111: Effect of Pretreatment of pdH0.2~

on Rate
Pretreatment

Standard
A (see text)

B (see text)
C (see text)

104k,
min-1

19.3
11.7
28.5
19.3

(11) K. N. Campbell and B. K. Campbell, Chem. Rev.,31,77 (1942).

Volume 70,Number 8 Auguat 1966

R. J. RENNARD,
JR.,AND R. J. KOKES

2546

1.8
t
Y
\
0

1.6

1.4
OY

0.10

0.40

0.50

0.30

0.20

1.2

H 1 Pd
Figure 4. Activity us. hydride composition. The ordinate
represents the amount of C2H6 formed per hour.

1.0

11:
0

0,I

0.2

0.3

0.4

0.

H / Pd

3'0

2.0

1.5

1.0

0.5

Y
0

ct

01
4.5

5.0

5.5

x to3

Figure 5. Arrhenius plot for ethylene hydrogenation: triangles,

PdHo.lt; open circles, PdH0.24; closed circles, PdHo+

If we compare the activity for a standard run to

that for these pretreatments, we find A reduces the
activity, B increases the activity, and C has no effect.
A comparable reduction in activity with pretreatment
A is also found for PdHo.12, PdH0.29,and PdHo.89.
E$ect of Temperature. The rate constant was determined for several hydride compositions over a
temperature range from -64 to -98". These results
are summarized by the Arrhenius plots in Figure 5.
Values of the apparent activation energies were 8.6,
7.7, and 7.5 kcal for PdHo.11, PdH0.24, and PdHO.40,
respectively .
The Journal of Physical Chemistry

Figure 6. Isotope effect us. hydride composition.

Reaction with Deuterium. The reaction of ethylene

or propylene with palladium deuteride followed the
first-order rate law. I n general, the rate of reaction
of the deuteride with olefin was greater than that
for the hydride, but the effect was most pronounced
a t lower temperatures and higher hydride concentrations. A systematic series of experiments with ethylene carried out alternately with the deuteride and the
hydride a t -78" yielded the value of k D / k H us. hydride composition. These results are summarized in
Figure 6. The ratio k D / k H is nearly unity a t very low
hydride concentrations, increases abruptly near PdHo,l,
and increases more slowly a t higher hydrogen concentrations to nearly 2. Near PdHoSlthe reproducibility
was far worse than a t higher or lower hydrogen concentrations; possibly this occurs because the surface
hydrogen concentration is often slightly above or below
the gross concentration with the results that near
PdHo.1 the surface can be in the high or low kD/kH
region of Figure 6.
The rate constant for reaction of propylene with
PdDo.24 was, as with the hydride, nearly an order of
magnitude less than that for ethylene. I n this case
also, an inverse isotope effect was found; k D / k H for
propylene was 1.9-2.0.
A series of alternate deuterium and hydrogen runs
was made between - 64 and - 98" for the single composition PdHo.zror PdDo.zr. These results are shown in
Figure 7 and indicate that k D / k H changes from 1.5 a t
-64" to 2.3 at -98". The activation energy for the deuterideis about 1 kcal less than that for the hydride (7.9
kcal) .

HYDROGENATION
OF ETHYLENE
AND PROPYLENE

1.6

1.2

s
0,

2547

0,s

t\I

0*4

4.5

' e

5.0

I
TX
x

5.5

No. of D-otoms / Molecule

io3

Figure 7. Arrhenius plot for reduction by hydride and

deuteride: open circles, PdDa.,c; closed circles, PdHo.la.

For the reaction with deuterium, it was possible to

estimate the deuterium content of the product ethane.
The fragmentation pattern for deuterated ethanes was
assumed to be that of the nondeuterated ethane (at
25 v) with appropriate statistical corrections and a
relative probability of 1.2 and 0.8 for cleavage of C-D
and C-H bands, respectively.12 The results at -78",
together with those reported by Bond and Wells13
for supported palladium a t -36O, are summarized by
the smoothed curves in Figure 8. Clearly there is
much less mixing of deuterium for ethane production
by palladium deuteride.

Discussion
Palladium hydride is a two-phase system; .both
phases, a and p, are cubic close packed with respect
to palladium atom^.'^^'^ At room temperature, the
CY phase alone is present for hydrogen concentrations
below PdHo,05and the /3 phase alone is present for concentrations above PdHO.jj. At intermediate concentrations both phases, as judged by X-rays, coexist.
At the temperatures of interest to us, the hydrogen
atoms are inobilelj and occupy the octahedral holes in
the close-packed lattice.16
The hydrogen-palladium system is not a typical
two-phase system. In the two-phase region the hydrogen fugacity increases with hydrogen content."
(Such behavior can be accounted for in part by analysis of stress eff ects.Is) Furthermore, physical properties such as magnetic susceptibility do not reflect the
phase transitions. The molar magnetic susceptibility

Figure 8. Isotopic distribution in product ethane: open

circles, PdDo.2aplus ethylene a t -78"; closed circles,
hydrogenation over supported palladium a t -36O.13

decreases linearly as the hydrogen concentration is increased and reaches zero near PdHo.6.19 This can be
rationalized by the assumption that each hydrogen atom
donates one electron to the existing holes in the d band
of palladium. Similarly, the relative resistivity increases linearly with hydrogen content up to PdH0.v6.
We are interested primarily in the relation of activity
to electronic structure, and the electronic structure,
as judged by susceptibility and resistivity, depends
on the hydrogen content, not on what phases are
present. Accordingly, we shall focus our attention on
the variation of activity with hydrogen content alone.
The very rapid uptake of hydrogen at - 78" suggests
a rapid transfer of hydrogen between the surface and
bulk palladium. Zero-order dependence on hydrogen
pressure for olefin hydrogenation is in line with this;
for, then, we would expect the surface concentration of
hydrogen to be controlled by the bulk concentration
and to be independent of the gas phase concentration.
The nearly zero-order dependence of hydrogenation
(12) D.0.Schissler, S. 0. Thompson, and J. Turkevich, Discussions
Faraday Soc., 10, 46 (1951).
(13) G . C. Bond and P. B. Wells, Advan. Catalysis, 15, 91 (1964).
(14) S. D. Axelrod and A. C. Makrides, J. Phys. Chem., 68, 2154
(1964).
(15) R.E.Norberg, Phys. Rev., 86,745 (1952).
(16) T.R. P.Gibbs, P r o p . Inorg. Chem., 3, 422 (1962).
(17) D.H.Everett and D. Norden, Proc. Roy. 8oc. (London), A254,
341 (1960).
(18) N. A. Scholtus and W. K. Hall, J. Chem. Phys., 39, 868 (1963).
(19) C. Kittel, "Solid State Physics," John Wiley and Sons, Inc.,
New York, N. Y.,1958,p 334.

Volume 70, Number 8 Auoust 1066

R. J. RENNARD,
JR.,AND R. J. KOKES

2548

rate on olefin pressure suggests the surface is nearly

covered with olefin. The reduction of hydrogen sorption rate by a factor of
to
when olefin is
present supports this view.
If the above be true, the rate-controlling step is
likely to be one of the following
C Z H

+ CZH4 I_ C2H5
H + CzHs
CzHs(g)
2CzH5
CzH4 + CZ&(g)
H

I n the above sequence, C represents the

concentration, C2Hs(g) represents gaseous ethane, and
all other species are assumed to be attached to the surface. Reactions 2 and 3 are usually assumed to occur
in olefin hydrogenation, and reaction 4 has been considered by BondeZ0 I n any event, if the steadystate approximation is applied to C2H6,we obtain

(It is, of course, conceivable that the rate of hydrogenation is controlled by diffusion of hydrogen from the
bulk to the surface. This can be ruled out on three
counts: (a) The kinetics are not consistent with diffusion. (b) The order of magnitude difference in rate
for ethylene and propylene is not consistent with a rate
controlled by diffusion. (c) The inverse isotope effect
for the reaction is not consistent with the normal isotope effect found for diffusion.)21
The marked difference in rate for ethylene and propylene rules out reaction 1 as the rate-controlling step.
The lack of isotopic mixing shows that the reverse of
reaction 2 can, in the first approximation, be neglected.
If we then make the assumption that the reverse reaction (1) is much more rapid than (2), we can write

On integration, with CzH4 constant, this yields the observed form

In C/C,

-kt

where the constant k is a composite quantity given by

The observed kinetic isotope effect is qualitatively

consistent with the conclusion that the slow step is
the rate of addition of a surface hydrogen atom to adsorbed olefin. Figure 9 shows on the left an energy
The J O U T T
of ~Phyaical Chemistry

Figure 9. Relative energies of PdH and PdD (see text).

diagram for gaseous hydrogen, deuterium, palladium

hydride, and deuteride. I n order to construct this
diagram, it was assumed that differences in the heat
of formationg of palladium deuteride and palladium
hydride, 8.6 and 9.6 kcal, respectively, stem primarily
from zero-point energy effects. From these data and
the zero-point energy of hydrogen vs. deuterium, we
find that the difference in zero-point energies for PdH
and PdD is about 0.8 kcal. If, consistent with the
kinetic analysis, it is assumed that the slow step is
the addition of a surface hydrogen to a carbon atom,
the activated complex will have a nearly normal
carbon-hydrogen band. Data for CC13D vs. CCl3HZ2
reveal the zero-point energy difference for this C-D
us. C-H bond is 2.1 kcal. With the same figure adopted
as the maximum zero-point energy difference for the
complex, we obtain the energy diagram on the right
of Figure 9. If we accept the foregoing qualitative
analyses as correct, the activation energy for reaction
with palladium deuteride is ut most 1.3 kea1 less than
that for palladium hydride. Thus, hydrogenation
would be expected to show the observed inverse isotope;
the agreement with the observed difference in activation energies (1.0 kcal) is regarded as fortuitous.
(We have no convincing explanation for the falloff
in isotope effect a t low concentrations of hydrogen.
Data are not available for low hydride concentrations
which would permit construction of a parallel t o Figure
9. We believe, however, that this change in isotope
effect may be indicative of a change in mechanism,
perhaps associated with the pure a phase.)
The first-order rate constant depends only on the
(20) G. C. Bond, Trans. Faraday SOC.,52, 1235 (1956).
(21) W. Jost, Diffusion, Academic Press Inc., New York, N. T.,
1952, p 308.
(22) G. Herzberg, hlolecular Spectra and Molecular Structure,
11, D. Van Nostrand and Co., Inc., New Tork, N. Y.,1945,p 316.

HYDROGENATION
OF ETHYLENE
AND PROPYLENE

initial hydrogen concentration and remains constant

as the hydrogen content is reduced by reaction with
ethylene. Michel and G a l l i ~ oreported
t~~
that although
sorption of hydrogen reduces the magnetic susceptibility, removal of sorbed hydrogen by reaction a t low
temperature does not restore the initial magnetic
susceptibility. Implications of this observation have
been recently discussed by Cribbs.I6 If this observation is correct, it would mean that the electronic properties of the catalyst are governed wholly by the initial
hydrogen content. Thus, provided kl/k-l and (C2H4) do not depend on the initial hydrogen content,
k2 decreases as the holes in the d band are filled.
Regardless of the validity of the analysis in the preceding paragraph, however, the following conclusion
can be stated without equivocation. The activity
increases as the holes in the d band are filled; the firstorder rate constant decreases as the holes in the d
band are filled. This raises questions about correla-

2549

tions attempted solely on the basis of activity without

kinetic analyses.
The analysis of the data obtained with the standard
pretreatment yielded a reasonable but admittedly
tentative interpretation. Effects of varying this pretreatment are too complex for detailed interpretation.
It is, however, worth noting that cooling in hydrogen
poisoned the catalyst as has been observed for nickeP4
and also palladium.26 Perhaps these effects are due
to changes in surface structure noted by Germer.26

Acknowledgment. Acknowledgment is made to the

donors of the Petroleum Research Fund, administered
by the American Chemical Society, for support of this
research.
(23) A. Michel and M. Gallisot, C m p t . Rend., 208, 434 (1939).
(24) W.K.Hall and P. H. Emmett, J.Phys. Chem., 63, 1102 (1959).
(25) A. Couper and D. D. Eley, Discussions Faraday SOC.,8 , 172
(1950).
(26) L. H.Germer, Advan. Catalysis, 13, 191 (1962).

Volume 70, Number 8 Auguet 1966