Joum~of_iolecutmC~hZty~ip.

9(1960)265 -274 265

@E%etierSequoiaSA_,Lausmne -Rint.edintheNetherhxis

L A. GlCRRETSEN, W. KLUT, 56. H. VR!XJGDENiiIL and J. d. F. SCEfOLTEN*

De~enCofChemi~Technolo~.DelfC UniuersibofTeehno~~.Jut~naLcan 136.
2628 BLDetf&(TheNethenbads}
(ReceivedSeptemberl3,1979)

We -died the kinetics of the heterogeneous hydroformylation of

propylene, using hydrido~bonyltris(tiphenylphosphine)rhodium(I), dis-
solved in triphenylphosphine and capillary condensed into the pores of a
support, as a catalyst_
The results can be described by the power rate equation:

r’=kaIRhl_nF$T P~,P& exp(-G/RT),

where E,, the apparent activation energy, is equal to 79.1 kJ/mol, end the
reaction orders a, b and c are 1.03,0.09 and 0.23, respectively. The reaction
order in carbon monoxide c is pressure dependent, and at carbon monoxide
pressures above 0.15 MPa equal to 0.08. The selectiv<ty towards n-butyr-
aldehyde is not influenced by the hydrogen and propylene pressures, but
varies strongly with the partial pressure of -bon monoside. When this
partial pressure is lowered tiom 0.52 to 0.05 MPa, the selectitity increases
from 10 to 30. Similarly, an increase in temperature from 70 to 1066 “c
raises the selectivity from 6.7 to 11.5. With KieseEguhr and polystyrene-20%
divinylbenzene (XAD-2) as catiyst supports, the rate of reaction is first
order in rhodium complex concentration in the solvent hgznd PPhs. The
kinetics differ significantly from those found in homogeneous hydroformyla-
tion in, for instance, toluene.

1. Introduction

In the preceding parts of this series [I - 41 and in the relative patents

[ 5,6] the preparation and characterization of supported liquid phase rhodium
catalysts have been described, tigether with their performance in the hetero-
geneous hydroformylation of ethylene and propylene at 90 - 199 “c and
266

12 - 1.57 MPa total pressure- We also discussed the influence of such para-
meters as the type of support, the degree of liquid loading, the type of ter-
tiary phosphine and the introduction of cerkin organic add&i- into the
catalyst solution.
The present paper deals with the kinetics of the hydroformylation of
propylene by means of SIP rhodium catalysts. Attention will be peid in
particular to the question in how far the kinetics differ hxn those found in
homogeneous catalysis with rhodium complex dissolved in, for instance,
benzene or toluene; in SIP catalysis using PPhs as a catalyst solvent we are
dealing with heterogeneous hydroformylation (see part IL of this series).

2. Experimental

The preparation and characterization of the catalysts, as well as the

continuous-flow apparatus for measuring the catalytic performance, have
been described in part I of this series_ For the chemicals and supports used
see part I and part III, respectively.

3. Results

3.1. Difjfus!obnal retm-dhtioon

To find out if, and to what degree, the reaction rate is retarckd by pore
diffusion, the activity and selectivity for propylene hydroformylation were
determined for two silica 0003E caalysts with different mean particle dia-
meters: 0.46 mm (the usual size) and 3.22 mm. The mean pore radius iv of
silica 0003E is 5.7 run, and the BET surface area 203 m2/g (see part III).
After being loaded with the catalyst solution (6 = 0.56), the BET surface
area of both catalysts is 39 m2/g cat. The results (Fig. 1) show that the cata-
lytic activity is not influenced by the particle diameter, proving that in these
mesoporous silica 0003E sup_ported catalysts the chemical reaction rate is
not slowed down by pore diffusional effects.
Also when the catalyst is supported on Kieselguhr MP-99 (6 = 0.44,
[Rh] = 5.7 mol/m3), the activity and selectivity in propylene hydroformyla-
tion are independent of the particle diameter between 0.46 and 2.61 mm;
this is in conformity with the macroporous texture of Kiese1guh.r (Fp =
248 nm, S,, = 18 m2/g; see part RI). Furthermore, calculations by Chu’s
method 171, demo_nstratz the nonoccurrence of diffusional retardation in
the film around the catalyst particle.
SIP hydroformylation being heterogeneous in nature (see part II),
diffusional retardation in the liquid catalyst solution need not be considered.

3.2. Relation between degree of conversion and nzfe of reaction

To find out in how fk the reactor operates differentially, the degree of
propylene conversion was determined as a function of the reactor space time.
 267

0 I I I I

0 100 200 300 433

- t.3 (hrs)

Pig. in propylene hytiroformylation

1. Actisitg and selectivity zs a fun&ion of time, for
two cat&& partide diameten. P = 1.57 l&Pa; Ci/H2/(=0 = l/l/l; c'= 90 'c; W/F =
0.987 x lo-" gFuLslQn3 Ci_ Catalyst,s~ca000-3E;6.0.56;[Rh],5.5 mol/m3;P/Rb.
744 mollmol;0,particIe diameter= 3_22mm~X,particle digmeter= 0.46 mm.

e (Y.1

Fig 2.Con~anofpmpyleneas afirnciion ofreacbrspacetime P= lS7MPa; C$/

H,/CKI = l/l/l; C = 90 "c; W/F = variabie. Catalyst, Kiesdguhr w-99; 6,0.44; [Rbl,
70mol/m3;P/Rh,61mot/mo1.

The straight Line fih.mugh the origin proves that the reactor operates
differentially below 6% conversion; in the int.avd from 0 to 6% conversion
the initial reaction rate can therefore be expressed as:

3.3_Itzfhence ofLherkodiumcomplexconce~fmtionon thekinetics

Therhdium complex concenkationinPPhBcapillarg condensedinto
the pores of JSiedguhr W-99 (8 = 0.44) vbasded from 5.7 to fOO mal/m3.
TABLE i
Influence of Che rhodium complex concentxation on the activity
and selectivity for propylere hydroformylationa

[Rtl]b P/Rh r90 S

(mol/n13 ) (mol/mol) (cm3 G/(g Rh- 5))

5.7 719 6.12 8.9

10-l 106 5-26 9.8
25.2 148 5.96 9.0
84-7 51 6.27 9.5
100.0 44 5.93 9.5

=Reaction conditior-sr P = 1.5i -!@a; C2-3 /Hz/CO = l/l/l; t = 90 ‘c; E/F = 0.981 x 1O-3
g Rh_ s/cm3 G_
“Concentrations bsed on the Eeight of rhodium complex used in the catalyst prepara-
tion; these are not equal to the actual concentrations becauze of adsorptive withbwal
of rhodium complexes onto the pore walls (see part III).

The results for propylene hydroformylation are summarized in Table 1. The

activities were measured at the final and stable activity level, which was
reached within 10 h (see also part III).
As is seen in Table 1, the activity per g of rhodium is independent of
the rho&rm complex ooncenIzation, which means that the activity per unit
volume of catalyst solution is first order in rhodium complex concentration;
furthermore, the selectivity also proves to be independent of the rhodium
comples concentration_
With macroreticular polystyrene-20% divinylbenzene XAD-2 as a sup-
port the activity and selectivity are neither influenced by change in rhodium
complex concentration between 5.5 and 50-l mol/m’.
When the rhodium complex concentration of a silica C003E supported
catalyst is increased over the fame range, no simple first-order relation is
found; at 50.1 mol/m3 the activity per g of rhodium is twice as high as
expected_ This is bemuse the adsorptive interaction between ‘the rhodium
complex and the surf&e of the silica support exceeds the interaction on the
ho supports mentioned above, as is illustrated by the different. shape of the
adsorption iso’herm of the rhodium complex (see part LG, Fig. 2). Whereas
in the tests w&31 Kieselguhr HP-99 and XAD-2 we o_peratzd in the linear
range of the isotherms, the measurements on the silica 0003E supported
catalysts were carried out also outside this range. Therefore, in the latter
case, a correction has to be made for the lower extent of adsorption of rho-
dium complexes at the litigher rh&.ium concentration_
The observed orders in rhodium complex concentration on XAD-2 and
Kieselguhr MP-99 point to a linear relationship between the concentration in
the bulk of the PPhs and at the phase boundary gas-PPh,.

3.4. Temper&we depe&ency

The apparent actktion energy was determined from an Arrhenius plot
(Fig. 3). The melting point of the catalyst solution in Kieselguhr MP-99 was
 269

6,i
I I I
32c 3GO 360 380
- +x 103 (K-1) -T (Kl

Fig. 3. .&&enius plot for propylene h~droformyI&ioa -P = 1.57 wa; G/&/m=

l/l/l;c= 50-5 -166.8 =C=;W/F= 0.981 x 1O-3 gRh.s/cd G. caL3lyst,Kieselguku
W-99; 6.0.44; [Rh],10.1mollm3; P/Rh, 406 mollmol.
Fig. 4. &lectivity as a function of temperature.Reaction c~~nditio~~~ ad CdAW- as in

Fig. 3.

found to be 74.7 =C (see part II), which points to a phase transition of the
catalyst solution within the temperature range inv~tigated (50.5 to 106.8 C).
Figure 3 shows that the apparent activation energy is constant (79.1 kJ/
mol) between 50.5 aud 106.8 %. The absence of an activity jump around
the melting point of the catalyst solution is to be ascribed to the hetero-
geneous nature of the hydroformylation reaction (see also Part II).
The temperature strongly influences the selectivity (Fig. 4): below 70 ‘c
the selectivity is temperature independent, whereas above 70 “c it skongly
increases. As only minute amounts of by-products are formed (see pa& I)
selectivity changes are related only to a change in the ratio normal/iso-
aIdehyde.

3.5. Influence of the partid pressures of hydrogen_ carbon monoxide and

propylene
The reaction rate and the selectivity were determined as functions of
the partial pressure of each reac’ant. By admixing the appropriate amount
of helium, the total flow rate was kept at 2.25 cm”/s and the total pressw~
at 1.57 MPa. In this way, one of the padid pressures could be varied, while
the other two were held constant at 0.52 MPa Similarly, the partial pmssures
of hydrogen and carbon monoxide were varied from 0.05 to 0.52 Ml%, and
that of propylene kom 0.20 to 0.52 MPa. The resul% are presented in Fig. 5.
By means of the ieast squares method, the results were fit&cl into a
power rate equation of the form:

r = kpg,= p$ p&

The reaction orders are compiled in Table 2.

270

- In (P,)

Fig. 5. Reaction rate for propylene hydrofo_=ybtion as a function of the partid pres-
SUITS 0-r the reacbntd x, caxbon monoxide; 0, hydrogen; q, propylene; P, 1.57 MPa;
t, 90 ‘2. Catalyst: Kiselguhr MP-99; 6.0.44; [Rh], 5.5 mol/m’ ; P/Rh, 744 molhol.

TABLE 2
Reaction orders for propylene hydrofonnylation at 90 Y=

PC: PLL pco (CO.15 MPa) pco (DO.15 Wa)

4 b - C C

r l-03 0.09 0.23 0.08

--II 1.04 O-09 0.22 0.03
‘iso 0.99 0.06 0.66 0.66

We also made a purely phenomenological distinction between +ke reac-

tion rates for n- and isobutyraldehyde formation in the following way:

and
1
ri_y)= -r
S+1

The reaction orders for r, and ri’iw,are presented in Table 2.

The reaction order in propylene is ahout one, that in hydrogen close to
zero_ The reaction order in carbon monoxide is a function of the carbon
monoxide partial pressum, uld changes from 0.23 at low carbon monoxide
pressure to 0.08 st high CO pressure.
The infiuence of the various partial p ressures on the seleckrity is shown
in Fig. 6_ It is seen that the selectivity strongly increases at low carbon mon-
oxide partial pressu res. The hydrogen and propylene partid pressure hardly
 271

k
1 I I
0 O-15 0 -30 o-45
- Pi (MPCI)

Fig 6.Selecti~ityasa~c;ionofthepartialpressuresoF~ereactankat90%. X,~a.rbon

moaoxide;o, hydrogen;l,propylene_Catal~tandreactioncr=nrtitiowasin Fig 5.

influence the selectivity. It was further found that at a hydrogen ptial

pressure of 021 MPa iustead of the usually employed 0.52 MFa, the reaction
orders in &on monoxide were equal to those given in Table 2. This fur-
nish= additional proof that the kinetics ze not influenced by the hydragen
partial pressure.
With XAD-2 being used as a support (S = 0.65; [Rh] = 5.5 mol/m3 and
P/Rh = 744 mol/mol) the reaction orders appeared to be equal to those for
a Kiese!guhr support&d cat&& (Table 2).

Diffusional retardation being absent (see section 3.1), we can +rectly

proceed with a discussion of the kinetics. Themost interesting observation is
the broken order in carhon monoxide pressure (c = 0.23). which tends to
the much lower value of 0.08 at CO partial presumes above 0.15 MPa. More-
over, the seIectivity tiwards normal butyraldehyde longly increases with
decreasing CO pressure, which is of great significance, both tirn a practical
and from 8 theoretical point of view. As we see it, these facts are to be
explained as follows:
In a catalyst solution several different rhodium complexes are in equi-
librium [S] :

=JS-hs )c -+ RhHCO(PPh3)3 ‘PP~ RhHCO(PPh& . m .

a 3
t-41 @I 0

=WC~)2(PPb 12 - PPh =W~hPPh3

3
(D) (E)
 Wilkinson [9 - ill, in studying the homogerxous hydroformylation
with RhHCO(PPhs)s d&solved in benzene or toluene, proved that the Com-
plexes, C, D and E axe catalytically active. For reasons to he advanced later
in this discussion, complex R may in the presence of a large excess of PPhs
and in the absence of other solvents also contribute considerably to the toti
activity _
When PPhs is used as a solvent, triphenylphosphtie strongly competes
with carbon monoxide for coordination with rhodium(I). The lower the CO
partial pressure, the higher the degree of PPh,coordination, and, in line with
what is know-n about the role of the PPhscoordination in homogeneous
tydroformylation [9,12] , the higher the selectivity towards normal butyr-
aldehyde and the slower the rate of reaction.
From spectroscopic studies [IO] we know that in homogeneous hydro-
formylation addition of CO +tothe alkyl complex, Rh(CsHi)CO(PPhs)a, and
subsequent insertion of the CO, is a very fast reaction, and that the order in
carbon monoxide is zero or even negative [9,12,13] _ In our heterogeneous
case the same will be true, but the zero-order dependency is obscured by the
positive influence the increase in CO partial pressure has on the formation of
lower PPh,coordinated complexes and, hence, on the rate of reaction_ Of
course, at higher CO p,artial pressures, this effect levels off, which is in line
with our observation above 2 CO partial pressure of 0.15 MPa.
The rate of reaction is about zero order in hydrogen, contrary to what
is found in homogeneous hydroformylation, where the oxidative hydrogen
addition to the acyl complex (Rh(COC,H,)CO(PPh,),, with x = I or 2) is
generally considered to be ratedete rmining, the reaction being first order in
hydrogen. Hence, in the case of the PPhs-rich complex- dissolved ti an
excess of PPhs (P/Rh = 744 mol/mol) with which we arc dealing in this
study, hydrogen addition has to be rejected as the rate~etermining step.
The rate of reaction was found to be fLrst order in propylene; therefore,
nddition of propylene to the PPhs-rich complexes is most probably rate-
determining. Obviously, for steric and electronic reasons, addition of propyl-
ene proceeds slowly in systems with a high PPhs coordination.
It is interesting to note that Cavalieri d’Oro et al. [14], in stud_ying
homogeneoirs hydroformylation with a relatively large excess of PPhs in
toluene (P/Rh = 150 mol/mol) also arrived at a rate of reaction that is zero
order in hydrogen pressure, whereas the orders in propylene and carbon
monoxide were 0.55 and -0.09, respectively_ Furthermore, within the range
of pressures investiga’t& the propylene and hydrogen pressures hardly influ-
ence the position of the equilibria between the :=uious complexes, and there-
fore have no effect on the selectivity_ This also accounts for the constancy cf
the reaction order in carbon monoxide upon a variation in hydrogen pressure.
In view of the remarkable difference with the kinetics of ho.mogeneous
hydrofo-myiation (where especially complexes D and E, and possibly C, are
known to be active) we presume that in the presence of a large excess of
PPhs and in the absence of other solvents, we are dealing with kinetio relati
to catalysis by higher PPhscoordinated complexes: complex C, but in partic-
 213

ular complex B, mi&t play a do minnnt ca~yCz roIe under these conditions.
However, definite clarib on this point CXIIonly.be obtained +tb.rougbcombi-
nation of more ext.e@ve kinetic measlrrements (espe&Uy covering a larger
range of -bon monoxide presmrres) with in S&Q@ectroscopic skidies of the
complex sfzuckrres.
_.

Acknowledgementr

The investigationswere supported (in pzt) by tie Netherlands Founda-

tion for Chemid Rm (SON) with financial aid from the Netherlands
Organization for the Advancement ef Pure Rm (ZWO).

List of symbols
c. h c reaction order
E, apparent activaticn energy, kJ//mol
F flow of o!eti at 0.1 MPa and 25 %, cm3/s
P tot2l pressuJz, MPa
P/Rh molar phosp_hineto rhodium ratio, mol/mol
P partizl pressure, M-Pa
EmI rhodium complex concenkation in PPh3 at 90 C, mo!/m’
R gas constant, J/mol K
r reaction rate, cm3 c3i/g Rh. s
-
TP mean pore radius, run
,
reaction rate, cm3G/m3 PPh3_ s
L selectkity = n/is0
T absolute tempera-. K
t kmpeiature, ‘c
td streuntime, h
W weight of rhodium in the reackr, g
6 degree of liquid loa&ng at 90 “c

References

L. A. Garikn, A mn Meerkerk, kI_ H. Vreugdenhil and J. J. F. Sddten, Part I,

J. Mol. Catd.. 9 (1980) 139.
L_ A_ Gerriken, J_ K Herman, W. EQut and J.J_ F. Schalten, part Ii, 6. hfoL Cut& 9
(1980)157.
L.A. Gerritsen,J.M.Herman ad J_J. F. Scfdkn, @ IQ J. MoL Cati., 9 (1980)
241.
L. GGercitsen, W_ Mut,M.H. VreugdenhilandJ.J_ F.Schoiteq partEV,d. Mol.
cu.b~_.9(1980)257.
Neth.PatentAppL7700554(1977),7712648(1977)and 7902964(1979),~L_A.
Gem&en and 6.J. F. Sdxolten (ZWO).
274

6 German Patent AppL 2802Z76 (1978), to L. k Gerritsen and J. d. F. Sd~olten

(St;lmicarban B-V.).
7 C. J. Chu, J. Kalil and-W. A. Wetteroth, Chem. Eng. .%gr., 49 (1953) 141.
8 D. Evans, G. Yagupsky and G. Wilkinson, J. Chem. Sot. (A).(1968) 2660.
9 C. K Brown and G. Wilkinson, J_ Chem. SIX. (AI.(IQ'iO)2753.
10 G. Yagupsky,C. K. Brown and G. Tabson, J. Chem. Sot (A). (1970) 1392.
11 D. Evans, J. A O&XII and G. MHcinsort, J. Chem. Sot. (A). (1968) 3133.
12 J. Hjortkjaer, J. MoL Cctal.. 5 (1979) 377.
13 K. L Olivier, B. L. Booth, Am. Chem. Sot.. Petr. Div. Prepr., 14(3) (1969) A?.
14 P. Cavalieri d’Gro et al., Rot. Symp. Komog. Catal., Vesspzm, 1978, p. 76.