Applied Catalysis, 47 (1989) 229-242 229

Elsevier Science Publishers B.V., Amsterdam - - Printed in The Netherlands

Carbon M o n o x i d e H y d r o g e n a t i o n over Metal

Loaded A l u m i n o p h o s p h a t e s

S.T. WONG and R.F. HOWE*

Chemistry Department, Universityo/Auckland, Private Bag, Auckland (New Zealand)
and
N.B. MILESTONE
Chemistry Division, DSIR, Private Bag, Petone (New Zealand)
(This paper was presented at the Bicentenary Catalysis Conference in Sydney, 1-2 September
1988)

ABSTRACT

This paper describes the preparation of molybdenum, cobalt and iron loaded ALP04- 5 catalysts
via adsorption and decomposition of the corresponding carbonyl complexes. Mo (CO) 6and Fe (CO)
are adsorbed within the ALP04- 5 pores, and 30-50% of the transition metal is retained on sub-
sequent activation. Co2 (CO)s adsorbs, and decomposes also on the external surface. The molyb-
denum- and iron-loaded ALP04- 5 catalysts show activities and selectivities in carbon monoxide
hydrogenation similar to the corresponding zeolites. Broader product distributions from the co-
balt loaded A1PO4- 5 are consistent with external loading, but this catalyst does not maintain a
stable activity.

INTRODUCTION

Transition metal loaded zeolites have recently attracted attention as cata-

lysts for the Fischer-Tropsch synthesis of hydrocarbons. There are two incen-
tives for studying such catalysts. The zeolite matrix may restrict the growth of
metal particles, thereby altering the Fischer-Tropsch product selectivity due
to particle size effects [1]. Alternatively, the zeolite support may impose a
shape selectivity on the hydrocarbon products. Studies of this type have been
reported previously, for example with ruthenium [2,3 ], iron [4-9 ], cobalt [ 10-
12 ] and molybdenum [ 13-15 ] loaded zeolites.
Aluminophosphates offer an alternative to zeolites as catalyst supports
[16,17 ]. These materials, first prepared by Union Carbide [18], have pore
structures similar to zeolites but carry no nett negative charge on the frame-
work (unless substituted with other elements) and therefore contain no ex-
changeable cations. Introduction of transition metals by ion exchange and re-
duction in aluminophosphates is thus not possible. The alternative method

0166-9834/89/$03.50 © 1989 Elsevier Science Publishers B.V.

230

successfully employed with zeolites is to adsorb and decompose volatile tran-

sition metal carbonyl complexes within the pores. In this paper we report an
investigation of the metal carbonyl method to prepare metal loaded alumino-
phosphates. Our objectives were to determine the extent to which carbonyl
complexes could be adsorbed and retained within a large pore aluminophos-
phate, and to explore the catalytic properties of the resulting metal loaded
aluminophosphates for carbon monoxide hydrogenation. Comparison with re-
lated zeolite catalysts was a further objective.

EXPERIMENTAL

The A 1 P Q - 5 used in this work was prepared according to the method of

Wilson et al. [ 19 ], using triethylamine as the template. Unless otherwise stated,
the template was removed by calcining in air for 15 h at 500 ° C. Crystallinity
was checked by X-ray diffraction, and the calcined powder had a nitrogen ad-
sorption capacity of 384 ~3 per unit cell.
Gravimetric experiments employed a quartz spring microbalance in an all-
glass high-vacuum system. A1PO4- 5 samples were pressed into pellets at 4-107
N m - 2 and outgassed at 350 ° C. The carbonyl complexes Mo (CO) 6 (Riedel-de-
H~ian, Seelze-Hannover, F.R.G. ), Co2 / CO )8 (Merck-Schuchardt) and Fe (CO) 5
(Aldrich) were outgassed by freeze pumping at 77 K. Adsorption was carried
out at room temperature, and weight changes monitored during adsorption and
subsequent outgassing.
Samples for pore volume measurements were prepared by room temperature
adsorption of the carbonyls in a separate in situ cell. Nitrogen adsorption mea-
surements employed a Mictromeritics Accusorb instrument at 77 K. Samples
for chemical analysis were digested in hydrofluoric acid and analyzed by atomic
absorbtion spectrophotometry (Varian AA-1275). X-ray powder diffraction
patterns were recorded on a Philips PW 1130/000 X-ray diffractometer. X-ray
photoelectron spectroscopy (XPS) measurements were made on a Kratos
XSAM 800 instrument using unmonochromatized Mg K a radiation.
Carbon monoxide hydrogenation experiments were conducted in a recircu-
lation reactor, and a continuous flow microreactor. The circulation reactor and
analysis of products has been described elsewhere [13,20]. The flow reactor
consisted of a glass or stainless steel tube containing ca. 0.5 g of catalyst, cou-
pled to a Philips Pye Unicam PU4550 gas chromatograph; hydrocarbon anal-
ysis was performed on a 3-m Poropak Q column (thermal conductivity detec-
tor). Catalyst samples were prepared in situ in both reactors; in vacuo in the
static reactor and in flowing nitrogen in the flow reactor. Carbon monoxide
hydrogenation experiments were carried out at 250 ° C or 300 ° C using a carbon
monoxide-to-hydrogen ratio of 1.0. A flow-rate of 20 cm 3 min-1 in the flow
reactor corresponded to a GHSV of 1200 h-1.
 231

RESULTS

Adsorption and decomposition of carbonyl complexes in AIP04-5

Fig. 1 summarizes results of gravimetric measurements on the adsorption

and decomposition of Mo (CO) 6, Co2 (CO) s and Fe (CO) 5 in A1PO4-5. The three
complexes showed quite different behaviour. Exposure of dehydrated A1PO4-
5 to Mo (CO)6 vapour at room temperature (P/Po = 1, Po ~ 7.5-10-2 Torr [21 ] )
caused rapid uptake of 6-7 mg per 100 mg, corresponding to about 0.35
Mo (CO)6 molecules per unit cell, within i h, and no further weight increase
occurred after 20 h. Infrared spectra of adsorbed Mo (CO)6 showed the char-
acteristic bands of the intact molecule in a low symmetry site, similar to those
seen previously for Mo (CO)6 adsorbed in zeolites [22]. Evacuation at room
temperature caused a 17% weight loss and a slight decrease in intensity of the
infrared bands of adsorbed Mo (CO) 6. On subsequent heating in vacuo, a large
weight loss was observed at 100°C (Fig. 1), and at higher temperatures a fur-
ther slight decline to a final loading of 0.74 mg per 100 mg at 400 ° C. Chemical
analysis showed t h a t this loading corresponds to the molybdenum content of
the A1PO4-5, and infrared experiments confirmed that decarbonylation was
complete at 400 ° C. A major difference from the decomposition of Mo (C0)8
adsorbed in zeolites is that no intermediate subcarbonyl species were detected
by infrared spectroscopy in ALP04-5; spectra recorded after heating to pro-

28
26
24 "",\ R,L lOOC 208C 3ooc 4ooc

22 / ge(CO)5

~ ", i'
- 14

~ ~2

4 i.~ , ~ Mo (CO)G

2 z"
0
200 400 G00 800 1000 1200 1400 1600
TIME ( r a i n )

Fig. 1. Gravimetric data for carbonyl uptake and decomposition in ALP04-5. ( X )Mo (C0)6, ad-
sorbed 1600 min at room temperature, then evacuated at room temperature and above, as indi-
cated. (•) Co2(CO)s, adsorbed 79 h at room temperature, then evacuated at room temperature
and above, as indicated. (•) Fe (CO)5, adsorbed 700 min at room temperature, then evacuated at
room temperature and above, as indicated.
232

gressively higher temperatures in vacuo showed gradual disappearance of the

Mo (CO)6 bands with no new bands appearing. The final molybdenum loading
of 0.74 wt.-% corresponds to retention of ca. 30% of the Mo(CO)6 initially
adsorbed at room temperature.
As in the case of zeolite Y [22], higher molybdenum loadings in A1PO4-5
could be achieved by carrying out repeated cycles of adsorption and decom-
position. The amount of Mo(CO)6 adsorbed at saturation decreased as the
molybdenum loading increased, but loadings of up to 6 wt.-% molybdenum
could be achieved after 10 adsorption and decomposition cycles. An infrared
experiment was also carried out with an A1PO4-5 precursor which had not been
calcined i.e. in which the pore volume was largely occupied with the tri-
ethylamine template. Negligible adsorption of Mo(CO)6 was found on this
sample.
The uptake of Co2 (CO)8 in ALP04-5 at room temperature occurred much
more slowly than that of Mo (CO)6, and did not reach a limiting saturation
value after 24 h exposure (Fig. 1). The uptake increased further after brief
evacuation at room temperature, and for the experiment presented in Fig. 1 a
final uptake of 22.3 mgper 100 mgwas achieved after 79 h exposure. This figure
corresponds to 0.95 Co2 (CO)8 molecules per unit cell. The corresponding ex-
periment with ALP04-5 precursor gave however Co2 (CO) 8 uptakes > 50% of
those for calcined AIPO4-5, suggesting that a major part of the Co2 (CO)8 up-
take occurs on the external surface of the A1PO4-5. Infrared spectra of ad-
sorbed Co2 (CO) 8 showed clear evidence for the formation of Co4 (CO) 12during
adsorption.
Evacuation at 100 ° C of AIPO4-5 containing 22.3 mg per 100 mg of adsorbed
Co2 (CO)s caused a 70% weight loss (Fig. 1 ), but little further change occurred
on heating to higher temperatures. The final loading of 6.33 mg per 100 mg
after heating in vacuo to 400 °C agreed with the chemical analysis for cobalt,
and infrared spectra showed that all carbonyl ligands were removed after heat-
ing to 250°C. This final loading corresponds to retention of ca. 80% of the
Co2 (CO)s initially adsorbed.
Exposure of AIPQ-5 to Fe (CO) 5at room temperature (P/Po = 1,po .~ 28 Torr)
caused an immediate uptake of 8.5 mg per 100 mg, followed by a slow further
uptake which may correspond to condensation of liquid Fe (CO)5 on the exter-
nal surface of the ALP04-5 and the sample container. The extent of this second
uptake was diminished at lower Fe (CO)5 vapour pressures. Evacuation at room
temperature reduced the Fe (CO)5 loading to 6.5 mg per 100 mg, which corre-
sponds to 0.63 Fe (C0)5 per unit cell. Subsequent heating to 100°C caused a
large further weight loss similar to that seen with Mo (CO)6 (Fig. 1 ), and the
final loading of 1 mg per 100 mg after heating to 400 °C corresponds to the
chemical analysis for Fe. This figure represents retention of ca. 50% of the
strongly adsorbed Fe (CO)5 not removed at room temperature. Infrared spectra
showed that Fe (CO)5 is adsorbed in ALP04-5 as the intact molecule, and that
 233

TABLE 1

Pore volumes of ALP04-5

From nitrogen adsorption on pressed pellets at 77 K. Pore volumes from monolayer capacity
calculated by fitting data to Langmuir isotherm.

/~:*per unit cell

Before adsorption After adsorption Difference

and
decomposition
Mo(CO)6 302.3 274.8 27.5
Co2(CO)s 293.9 226.6 67.3
Fe (CO)5 310.2 289.1 21.1

decarbonylation is complete after heating to 400 ° C. As with Mo (CO)6, further

Fe(CO)5 could be adsorbed into the iron-loaded A1PO4-5 following decompo-
sition of the first dose of adsorbed Fe (CO)~.
Nitrogen adsorption measurements were carried out on pressed pellets of
ALP04-5 before and after adsorption and decomposition of the carbonyl com-
plexes. The pore volumes obtained, calculated from total nitrogen uptakes, are
given in Table 1. X-ray diffraction measurements showed that there was no
significant loss of crystallinity following adsorption and decomposition of the
carbonyl complexes.

Carbon monoxide hydrogenation

Fig. 2 shows conversion and product selectivity data for a molybdenum loaded
A1PO4-5 catalyst in the flow reactor at 300 ° C, as a function of time on stream.
The carbon monoxide conversion is expressed as total conversion to hydrocar-
bons, and the product selectivities as mole percentages of the carbon monoxide
conversion. Steady state conversions of ca. 2% were attained after 10-20 h on
stream. Methane is the major hydrocarbon product, but the selectivity to hy-
drocarbons >~C2 remained approximately constant at around 50% with time on
stream. Alkene yields were low but increased slightly with time on stream.
Carbon monoxide hydrogenation over molybdenum loaded A1PO4-5 in the
recirculation reactor gave similar carbon number product distributions to the
flow reactor. Alkenes (ethene and propene) formed initially decreased at longer
reaction times, and were presumably hydrogenated to alkanes. Similar con-
versions and product selectivities were obtained in subsequent runs provided
the catalyst was reduced in hydrogen at 400°C prior to admitting fresh
reactants.
Conversion and selectivity data for an iron loaded A1PO4-5 catalyst in the
flow reactor at 250 °C are shown in Fig. 3. Steady state conversions of 0.5-0.8%
234

8
z
o

8
o 0

100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500
TIME (min)

Fig. 2. Carbon monoxide conversion to hydrocarbons and selectivities {carbon atom % ) for Mo-
A1PO4-5 in the flow reactor at 300 ° C.

2.5

o
=
5~ R

1.5

o
1 2=23 = 3 2=23=3

0.5

0
120 240 360 480 600 720 840 960 1 0 8 0 1200 1320 1440
TIME (min)

Fig. 3. Carbon monoxide conversion to hydrocarbons and selectivities {carbon atom %) for Fe-
ALP04-5 in the flow reactor at 250 ° C.

were achieved after 10-20 h on stream at this temperature. This catalyst pro-
duced high relative yields of ethene and propene, but only small amounts of
butenes or hydrocarbons > C5 in the flow reactor. Data for an iron loaded A1PO4-
5 in the recirculation reactor at 300°C are shown in Fig. 4. The major products
formed in the recirculation reactor were methane, ethene and propene with
approximately equal selectivities; butenes, butanes and pentanes were also de-
tected but no pentenes.
 235

3.0

2.8
2.6
2.4
2,2
~ CH4
C3H6
c2-C4H8 (xl0)

2
,o
--1,8
@

I-C4H8 (xlO)
~1.4
@
~i.2
1
0.8
0.6 ' C2H6
0,4
w ~ n-C5H12
0.2 - - C3H8
0
120 240 360 480 600 720 840 960 1080 1200 1320 1440
TIME {min }

Fig. 4. Product distribution from carbon monoxide hydrogenation over F e - A 1 P 0 4 - 5 a t 3 0 0 ° C in

recirculation reactor.

Cobalt-loaded AIPO4-5 prepared from Co2 (CO)8 and reduced in hydrogen

showed a continuous decline in conversion in the flow reactor at 300 oC, from
ca. 3% after 1 h to 0.5% after 24 h. Methane was the initial major product, but
the selectivity to C3 and C4 alkanes increased with time on stream. The insta-
bility of the cobalt catalysts can also be seen in the recirculation reactor data
plotted in Fig. 5. The high initial rate of carbon monoxide conversion (major

1,0-
a
0.9

0.8

0.7

~0.5
o

0.3

0.2
O. 4 / C3H6

O. 1 t2-c4H8

0
120 240 360 480 G00 720 840 3GO 1080 1200 1320 1440
TiE (min)

Fig. 5. Product distribution from carbon monoxide hydrogenation over Co-AIPO4-5 prereduced
in hydrogen, at 300 ° C in recirculation reactor. ( a ) Alkenes, (b) alkanes. (continued on p. 236)
236
28 I-C4H10 P<IOD)

18
. . ~ CH4

16

14

I-C5H72 (×I DO)

• C2H6
4 C3H8

2
n-C4hl0
D n-C5H12
120 240 360 480 G00 720 840 960 1080 1200 1320 1440
TIME (min)

Fig. 5.

TABLE2

Specificactivities(turnover~equencies)~rhydrocarbon~rmation

Catalyst Moles CO converted per mole of metal per second

Flow reactor a Steady state Recirculation b

Initial reactor, initial

Mo-A1P04-5 (300°C) 4"10 -3 8.10 -4 1.10 -3

Fe -AIPO4-5 (250 ° C ) 8-10- 4 3.10 -4
(300°C) 1" 1 0 - 3
Co-A1PO4-5 (300°C) 2-10 -'~ 2.10-4~ 5.10-3

Total pressure 1 atmosphere.

b Total pressure 400 Tort.
" F r o m conversion after 24 h; steady state not reached.

product methane ) rapidly declined to almost zero after several hours, and the
yields of C2-C4 alkenes decreased.
Turnover frequencies calculated from total carbon monoxide conversion to
hydrocarbons for the three types of catalyst tested are presented in Table 2.
Data from the recirculation reactor are obtained from the initial rates of car-
bon monoxide hydrogenation. For data from the flow reactor, carbon monoxide
conversions after 2 h and 24 h on stream were used in the calculations. In all
cases total metal loadings was used in the calculations: this will underestimate
the true turnover frequency, but in the absence of active site densities does
allow at least comparison with other catalysts.
 237

TABLE 3

XPS analysis of used catalysts

Sample M : A1 XPS

Chemical analysis

Mo-A1P04-5 a 0.03 0.046

Fe-A1PO4-5 0.021 0.008
Co-AlP04-5 b 0.036-0.076 0.083

a Molybdenum of catalysts prepared from ALP04-5 powder were up to three times larger than those
of pressed pellets.
b As described earlier, cobalt contents obtained dependend on the time of exposure to Co~ (CO)s;
no saturation uptake could be achieved.

Used catalyst samples were examined by XPS to determine the distribution

of metal component throughout the A1PO4-5. Comparison of metal to alumin-
ium ratios determined by XPS with the corresponding figures from bulk chem-
ical analysis allows the extent of segregation of metal or migration of metal to
the external surface to be assessed. The data in Table 3 show that in no case is
there a large discrepancy between surface and bulk analysis. Similar conclu-
sions regarding uniformity of metal distribution have been reached from a pre-
liminary examination of catalyst samples by scanning transmission electron
microscopy.

DISCUSSION

Catalyst preparation

A1PO4-5 has unit cell composition AI12P1204s, and contains one-dimensional

channels with approximately circular cross-section, free aperture 7.4 A [18].
The theoretical pore volume available in the channels is 360 ~3 per unit cell,
and the figure calculated from nitrogen adsorption measurements on the pow-
der of 384 A3 is in close agreement with the theoretical value, allowing for
additional nitrogen adsorption on the external surface.
The lower pore volumes measured by nitrogen adsorption on pressed pellets
must result from partial pore blockage in these samples. The pore volume
available for carbonyl adsorption will also be attenuated in pressed pellets.
Mo ( CO ) 8 and Fe (CO) 5 have a free diameter of approximately 7 ,~; saturation
adsorption of these molecules in A1PO4-5 should amount to approximately 1
molecule per unit cell [ 18 mg per 100 mg for Mo (CO)8, and 13 mg per 100 mg
for Fe(CO)5]. The observed saturation adsorption for Mo(CO)8 in the gravi-
metric experiments of ca. 6 mg per 100 mg is less than half of the calculated
figure. This discrepancy is probably due to the same pore blockage causing a
238

20% decrease in adsorption capacity for the smaller nitrogen molecule. (We
note that up to two-fold higher molybdenum loadings were achieved in catalyst
preparations using loose powders. )
For Fe(CO)~, measurements of the saturation adsorption is prevented by
condensation of liquid carbonyl on the external surface.
The amount remaining adsorbed however after evacuation at room temper-
ature corresponds to ca. 0.5 molecules per unit cell, half the theoretical
maximum.
In the case of Co2(CO)s, complete filling of the pores should correspond to
0.5 molecules per unit cell, or 11.5 mg per 100 rag. As shown in Fig. 1, the
observed uptake of Co2(CO)s reaches this figure after ca. 30 h of exposure to
the vapour, but exceeds it by almost a factor of 2 after further prolonged ex-
posure (22.3 mg per 100 mg after 79 h). This fact, together with the observed
uptake of Co2 (CO)s by the A1PO4-5 precursor which has almost zero nitrogen
adsorption capacity, indicates that deposition of Co2 (CO)s occurs readily on
the external surface of A1PQ-5. Co2 (CO)s readily undergoes the dimerization
reaction:

2C0.~ (CO) s---~ C04 (CO) 12-I- 4CO

and Co4(CO)1~ if formed on the external surface is too large to enter the 7.4-
~, pores. Evidence has recently been obtained by XPS [23 ] for external loading
of zeolite Y with Co2 (CO)s in a similar manner.
Adsorption of Mo (CO) 6 and Fe (CO) 5 in zeolites is largely irreversible. For
example, zeolite L, which has a one-dimensional pore structure similar to
A1PO4-5, adsorbs 1.1 Mo (CO)6 molecules per unit cell at 25 ° C, with negligible
loss of molybdenum on subsequent activation [14]. For Fe(CO)~, 1.5 mole-
cules are adsorbed per unit cell in zeolite L, of which 1.0 remain on evacuation
[ 24 ]. In contrast, 70% of the adsorbed Mo (CO) 6 and 50% of the strongly ad-
sorbed Fe (CO) 5 (that which is not removed by room temperature evacuation )
are lost from A1PO4-5 on heating in vacuo. We attribute these differences to
the absence of exchangeable cations in the A1PO4-5 structure. Infrared evi-
dence for interaction of adsorbed carbonyl with zeolite cations has been pre-
sented elsewhere [14]. In A1PQ-5 such cation anchoring sites are not avail-
able, resulting in loss of the carbonyl complexes on heating. Some residual
carbonyl remains adsorbed, however. Although the framework of A1PQ-5 is
electrostatically neutral, the electronegativity differences between A13+ and
ps+ will result in alternating charge differences around the walls of the pores.
This effect has been invoked to explain the moderate hydrophilicity of A1PO4-
5; the affinity of A1PO4-5 for water is lower than that of faujasite but consid-
erably higher than that of the aluminium-free zeolite silicalite [24]. Similar
interactions may be responsible for the retention of Mo (CO)6 and Fe (CO)5 in
AIPO4-5. The decomposition of the residual carbonyl certainly differs from
 239

that in zeolites in that all six CO ligands appear to be lost simultaneously, with
no intermediate subcarbonyl species detected by infrared spectroscopy.
Decomposition of the involatile Co4 (CO),2 is almost complete at 100 ° C, so
that 80% of the cobalt is retained. The XPS measurements on cobalt loaded
catalysts exposed to Co2 (CO)s for shorter periods (and containing as a result
lower levels of cobalt) do not show evidence of extensive surface enrichment
in cobalt; this may indicate that Co2 (CO)s adsorbs first within the zeolite pores
and on the external surface subsequently, but that possibility requires further
investigation. The larger pore volume loss following adsorption and decom-
position of Co2(CO)s implies a greater degree of pore blockage than for
Mo (CO) 6 and Fe (CO) 5 (Table 1 ). The pore volume measurements on molyb-
denum and iron loaded A1PO4-5, together with the diminished capacity of the
loaded samples for a second dose of Mo (C0)6 or Fe (CO)5, suggests that the
metal remains within the pores following activation. This conclusion is sup-
ported by the XPS measurements on used catalysts.

Carbon monoxide hydrogenation

Molybdenum has not been traditionally regarded as an effective Fischer-

Tropsch catalyst. There have however been several recent reports of molyb-
denum based catalysts for production of LPG range hydrocarbons from syn-
thesis gas [25-31 ]. The product selectivities and conversion over molybdenum
loaded AIPQ-5 are broadly similar to those we have previously reported for
molybdenum loaded zeolites prepared from Mo (CO)6 [13-15]. For the zeolite
catalysts the activity and selectivity were found to vary with the nature of the
zeolite and the exchangeable cation. Zeolite L has a pore structure and size
closely similar to that of ALP04-5. Comparison of Mo-A1PO4-5 with Mo-KL
shows that the A1PO4-5 catalyst has both higher activity and selectivity to > Ca
hydrocarbons (ca. 50% compared with 30% for zeolite L). The flow reactor
data for molybdenum-loaded A1PO4-5 confirm the activity and selectivities
demonstrated in the recirculation reactor.
It is useful to compare the specific activity (turnover frequency) of the mo-
lybdenum loaded ALP04-5 with other molybdenum catalysts. Logan et al. [29]
investigated the hydrogenation of carbon monoxide over Mo (100) single crys-
tals and polycrystalline foils, and report typical initial turnover frequencies for
methane production of 0.11 s- 1at 300 ° C (CO/H2-- 0.33, total pressure = 4700
Torr) and 0.011 s -1 at 275°C (CO/H2=0.20, 1320 Torr), based on the geo-
metric surface area of the catalyst. Ranhotra et al. [28] studied molybdenum
carbide and nitride catalysts, and quote turnover frequencies in the range
20"10 -3 s - ' - 5 0 - 1 0 -3 s -1 (C0/H2=0.33, total pressure 400 Tort) at 300°C,
based on carbon monoxide chemisorption measurements of the number of ac-
tive sites. In both of these studies selectivities to methane were typically 80%
or higher. Sivasankar et al. [27] prepared highly dispersed zerovalent molyb-
240

denum catalysts by decomposition of Mo (CO)~ on fully dehydroxylated alu-

mina supports, and reported turnover frequencies of the order of 10 -2 s -1 at
350 ° C ( CO/H2 = 1, 23 atmospheres total pressure), based on total molybde-
num content. Conventional carbon-supported molybdenum catalysts, which
have been shown to contain a high concentration of zerovalent molybdenum
[25], give turnover frequencies (total molybdenum content) of up to 3-10 -3
s - 1 at 350 ° C (CO/H2 = 0.33, 1.2 atmospheres total pressure) [ 26 ].
The number of active sites in the molybdenum loaded A1PO4-5 is not known,
but comparison with the data of Logan et al. and Ranhotra et al. suggests that
a relatively high fraction of the total molybdenum is participating in carbon
monoxide hydrogenation, as would be expected from the high dispersion of
these catalysts. On a total molybdenum basis, the AlPO4-5 catalysts are at least
as active as the best conventionally supported molybdenum catalysts. The spe-
cific activity of molybdenum in A1PO4-5 is higher than that of molybdenum in
zeolite supports (maximum initial turnover frequencies of 10 -4 s-1 at 300 °C
in the recirculation reactor [20]), but the selectivities to hydrocarbons > C1
are comparable.
There is an extensive literature on carbon monoxide hydrogenation over
highly dispersed iron catalysts prepared from organometallic precursors on
high-surface-area supports [4-9, 33,34]. Hugues et al. [33] and Commereuc
et al. [34] reported that alumina or magnesium oxide-supported iron clusters
gave high selectivities to alkenes (mainly propene, up to 60% selectivity to
total alkenes at 265°C and 1-3% conversion). Zwart and Vink [9], in a de-
tailed study of iron clusters supported in zeolite Y, observed alkene selectivities
approaching 90% and increasing with time on stream for carbon monoxide
hydrogenation at 10 atmospheres pressure and 277 ° C. The activity levels quoted
by these authors correspond to turnover frequencies of ca. 1-10 -3 s -~ under
their experimental conditions (total iron basis), which are stated to be com-
parable with those of a fused magnetite catalyst, and the carbon number dis-
tributions showed no serious deviations from the Anderson-Schulz-Flory curve.
Bein et al. [7], on the other hand, showed that the carbon number distribu-
tions obtained from similar iron-loaded zeolite catalysts varied widely with
time on stream. These variations were shown to be due to migration of iron
out of the zeolite pores to the external surface and formation of a bulk iron
carbide phase; a simple Anderson-Schulz-Flory carbon number distribution
was obtained only when most of the iron was external to the zeolite pores. The
authors do not comment on alkene selectivity, and no absolute activity data
are given. This work suggests however that the catalysts used by Zwart and
Vink [9 ] consisted mostly of external iron.
The alkene selectivities observed for iron loaded A1PO4-5 in the flow reactor
are not as high as those reported for iron-loaded zeolite Y. Despite the differ-
ence in pressure, the specific activity levels are comparable. The iron-loaded
A1PO4-5 catalysts showed no large changes in carbon number distribution with
 241

time on stream, suggesting that migration of iron out of the pores of the type
reported for zeolite Y does not occur with the ALP04-5, and XPS analysis of
used catalysts confirmed this suggestion. We note however that migration of
iron may depend on reaction pressure. Higher alkene selectivities (up to 90%
for C3) were observed for iron-loaded A1PO4-5 in the recirculation reactor.
Cobalt-loaded zeolite catalysts in which the cobalt is dispersed within the
zeolite pores have been shown by several authors to produce large deviations
from the Anderson-Schulz-Flory distribution, with a predominance of light
alkenes [10-12]. These deviations were attributed to the extremely small size
of the cobalt metal clusters within the zeolite. The broad distribution of prod-
ucts obtained from the cobalt-loaded A1PO4-5 catalyst in the recirculation re-
actor suggests, in contrast, that external cobalt may be dominating the catal-
ysis. The rapid deactivation of the cobalt-loaded A1PO4-5 in both recirculation
and flow reactors is not understood at present.

CONCLUSIONS

The major conclusion of this study are that adsorption and decomposition
of Mo(CO)6 and Fe(CO)5 within the pores of A1PO4-5 produces highly dis-
persed metal clusters which appear to remain within the pores during carbon
monoxide hydrogenation at 1 atmosphere pressure. Relative to the corre-
sponding zeolites, ALP04-5 shows a lower capacity for adsorption of Mo (CO)6
and Fe (CO)5 from the gas phase and a markedly lower retention of the tran-
sition metal on subsequent activation. Both of these factors can be attributed
to the absence of exchangeable cations which can anchor t h e carbonyl com-
plexes. The metal loaded AIP04-5 catalysts show nevertheless specific activi-
ties for carbon monoxide hydrogenation at least as good if not better than the
corresponding zeolites, and the product distributions obtained are consistent
with high dispersions of the zerovalent metal. The attempted preparation of
similar cobalt-loaded A1PQ-5 from Co2 (CO)8 appears to result in high exter-
nal loading, and the resulting catalysts show poor carbon monoxide hydrogen-
ation performance.

ACKNOWLEDGEMENT

We acknowledge support of this work by a research grant from the New

Zealand Energy Research and Development Committee, administered cur-
rently by the Ministry of Energy.

REFERENCES

1 H.H.Nijs and P.A. Jacobs, J. Catal., 65 (1980) 328.

242

2 H.H. Nijs, P.A. Jacobs and J.B. Uytterhoeven, J. Chem. Sec. Chem. Comm., (1979) 1095.
3 H.H. Nijs and P.A. Jacobs, J. Catal., 66 (1980) 401.
4 D. Ballivet-Tkatchenko, G. Coudurier, Inorg. Chem., 18 (1979) 558.
5 D. Ballivet-Tkatchenko, N.D. Chau, H. Mozzanega, M.C. Roux and I. Tkatchenko, ACS
Symp. Ser., 152 (1981) 187.
6 D. Ballivet-Tkatchenko, G. Coudurier and N.D. Chau, in S. Kaliaguine and A. Mahay (Ed-
itors), Catalysis on the Energy Scene, Elsevier, Amsterdam, 1982, 123.
7 Th. Bein, G. Schmiester and P.A. Jacobs, J. Phys. Chem., 90 (1986) 4851.
8 T. Mitsudo, H. Boku, S. Murachi, A. Ishihara and Y. Watanabe, Chem. Lett., (1985) 1463.
9 J. Zwart and J. Vink, Appl. Catal., 33 (1987) 383.
10 K.C. McMahon, S.L. Suib, B.G. Johnson and C.H. Bartholomew, J. Catal., 106 {1987) 47.
11 D.K. Lee and S.K. Ihm, J. Catal., 106 (1987) 386.
12 R.K. Unger and M.C. Baird, Chem. Soc. Chem. Comm., {1986) 643.
13 Y.S. Yong and R.F. Howe, J. Mol. Catal., 38 (1986) 323.
14 Y.S. Yong and R.F. Howe, in Y. Murakami, A. Iijima and J.W. Ward (Editors), Proc. 7tb
Internat. Zeolite Conf., Kodansha, Tokyo, 1986, p. 883.
15 Y.S. Yong and R.F. Howe in, D.M. Bibby, C. Chang, R.F. Howe and S. Yurchak (Editors),
Methane Conversion, Elsevier, Amsterdam, 1988, p. 503.
16 N.B. Milestone and N.J. Tapp, in D.M. Bibby, C. Chang, R.F. Howe and S. Yurchak (Edi-
tors), Methane Conversion, Elsevier, Amsterdam, 1988, p. 993.
17 E.M. Flanigen, in Y. Murakami, A. Iijima and J.W. Ward (Editors), Proc. 7th Internat.
Zeolite Conf., Kodansha, Tokyo, 1986, p. 103.
18 S.T. Wilson, B.M. Lok, C.A. Messina, T.R. Cannan and E.M. Flanigen, J. Am. Chem. Soc.,
104 (1982) 1146.
19 S.T. Wilson, B.M. Lok and E.M. Flanigen, Eur. Pat. Appl., EPO 043 562 A1 (1982).
20 Y.S. Yong, Ph.D. Thesis, University of Auckland, 1987.
21 R.R. Monchamp and F.A. Cotton, J. Chem. Soc., (1960) 1438.
22 Y.S. Yong and R.F. Howe, J. Chem. Soc. Farad. Trans. I., 82 (1986) 2887.
23 S.L.T. Andersson and R.F. Howe, to be published.
24 S.T. Wilson, B.M. Lok, C.A. Messina, T.R. Cannan and E.M. Flanigen, ACS Symp. Ser., 218
(1983) 79.
25 S.B. Murchison in H.F. Barry and P.C.H. Mitchell (Editors), Climax Fourth International
Conference on the Chemistry and Uses of Molybdenum, Climax Molybdenum Company, Ann
Arbor, MI, 1982, p. 197.
26 J.W. Dun, E. Gulari and K.Y.S. Ng, Appl. Catal., 15 (1985) 247.
27 S. Sivasanker, S. Yesodharan, E.P. Sudhakar and A. Brenner, J. Catal., 87 (1984) 514.
28 G.S. Ranhotra, A.T. Bell and J.A. Reimer, J. Catal., 108 (1987) 40.
29 M. Logan, A. Gellman ad G.A. Somorjai, J. Catal., 94 (1985) 60.
30 H. Miura, M. Osawa,, K. Sugiyama and T. Matsuda, J. Catal., 101 (1986) 178.
31 I. Kojima and E. Miyazaki, J. Catal., 89 (1984) 168.
32 Y. Iwasawa and N. Ito, J. Catal., 96 (1985) 613.
33 F. Hugues, B. Besson, P. Bussiere, J.A. Dalmon, M. Leconte, J.M. Basset, Y. Chauvin and
D. Commereuc, ACS Symp. Set., 192 {1982) 255.
34 D. Commereuc, Y. Chauvin, F. Hugues, J.M. Basset and D. Olivier, J. Chem. Soc. Chem.
Commun., (1980) 154.