Applied Catalysis, 3 (1982) 321-346 327

Elsevier Scientific Publishing Company, Amsterdam -Printed in The Netherlands

BIMETALLIC CATALYSTS; APPLICATION IN CATALYTIC REFORMING

J.L. CARTER, G.B. McVICKER, W. WEISSMAN, W.S. KMAK and J.H. SINFELTa
Exxon Research and Engineering Co., Linden, New Jersey, U.S.A.

aJ.L. Carter, G.B. McVicker and J.H. Sinfelt are affiliated with the Corporate
Research Science Laboratories of Exxon Research and Engineering Company.
W. Weissman and W.S. Kmak are affiliated, respectively, with the Corporate
Research Technology Feasibility Center and the Exxon Engineering Petroleum
Department.

(Received 29 March 1982, accepted 6 April 1982)

ABSTRACT
The results of several types of studies on platinum-iridium and platinum-rhenium
reforming catalysts are reported in this paper. The studies include characterization
of the catalysts by chemisorption of hydrogen and carbon monoxide, comparisons
of the catalysts for the conversion of selected hydrocarbons and extended runs
on the reforming of petroleum naphtha fractions. The results of these studies reveal
differences in the nature of the catalysts and in their performance in reforming.
The differences in reforming properties are such that the two catalysts can be
effectively combined in a reforming operation. In this manner it is possible to
take advantage of the desirable properties of both catalysts.

INTRODUCTION
During the past three decades catalytic reforming has become one of the most
important industrial applications of catalysis [1,2]. In the process saturated
hydrocarbons (alkanes and cycloalkanes) in petroleum naphtha fractions are converted
extensively to aromatic hydrocarbons which have much higher "anti-knock" quality
for automotive fuels. The aromatic hydrocarbons produced are also of interest for
petrochemicals. The petroleum naphtha fractions consist of hydrocarbons having
boiling points within the approximate range of 50-200°C. Reaction temperatures of
425-525°C and pressures of lo-35 atm are employed in the process. In addition to
reactions leading to the formation of aromatic hydrocarbons, the process also
involves isomerization, hydrogenolysis and hydrocracking reactions. During the
1950s and 1960s platinum on alumina catalysts were employed almost exclusively in
commercial reforming units. The outstanding performance of these catalysts led to
the wide application of catalytic reforming in petroleum refineries.
From the late 1960s to the present the platinum on alumina catalysts have
largely been replaced in commercial practice by catalysts containing platinum and
a second metallic element. Two examples of such bimetallic catalyst systems which
have been employed extensively in commercial reformers are platinum-rhenium [3,4]
328

and platinum-iridium [5], both supported on alumina. The application of these

catalysts in reforming has resulted in major improvements in the process. While
in some respects these two catalysts exhibit similarities in performance relative
to the earlier platinum on alumina catalysts, there are differences in detail in
the nature of the catalysts and in their performance in reforming. In the present
paper we present the results of an extensive experimental program comparing the
reforming properties of the two types of catalysts.

EXPERIMENTAL
Stainless steel tubular reactors approximately 3.8 cm in diameter with a volume
of about 1200 cm3 were employed in all of the extended reforming runs. The
reactors were immersed vertically in electrically heated fluidized solids baths.
The naphtha fraction to be reformed was vaporized and heated to reaction tempera-
ture in a preheat coil located in the fluidized solids bath. A separate coil in
the bath was used to heat a hydrogen-rich recycle gas stream to reaction tempera-
ture. The preheated naphtha and recycle gas streams were combined and passed
through the reactor. The reactor was packed with catalyst in the form of extruded
particles approximately 0.16 cm in diameter and 0.5-1.0 cm in length. Inert
alundum beads comparable in diameter to the extruded particles were employed as a
catalyst diluent in all of the runs. An axial thermowell was located at the center
of the catalyst bed. It contained a number of thermocouples positioned to give
temperature readings throughout the bed. The effluent from the reactor was cooled
and separated into liquid and gaseous fractions. The latter fraction is the source
of the hydrogen-rich recycle gas stream already mentioned. The gaseous product
from the system was analyzed routinely by gas chromatography as were the C5 and
lower carbon number components of the liquid product. Periodically, more detailed
chromatographic and mass spectrometric analyses of the liquid product were also
made. Research clear octane numbers were determined routinely on the liquid
product.
Since platinum-rhenium and platinum-iridium catalysts are exposed to sulfur in
situ prior to use in commercial reformers (to limit the hydrogenolysis activity
of the catalyst), the catalysts in the present work were contacted with an H2S
containing gas [6] until H2S was detected at the reactor outlet. In all of the
naphtha reforming comparisons considered in this paper, the runs were made
simultaneously with naphtha taken from a common storage vessel. Isopropyl chloride
and water were introduced with the naphthas in controlled trace amounts to maintain
a constant Cl level on the catalyst. The naphthas were desulfurized to a sulfur
level of 1 ppm or lower.
Some reforming runs of shorter duration (about 500 h or less) were conducted
in a smaller stainless steel tubular reactor which was also heated in a fluidized
solids bath. The reactor volume was about 25 cm3. A once-through hydrogen stream
was used in place of hydrogen-rich recycle gas with this reactor. Otherwise the

procedures employed were similar to those used in the extended runs with the
larger reactors.
Chemisorption data on the catalysts were obtained with conventional high vacuum
apparatus which has been described elsewhere ['I]. Prior to the determination of
adsorption isotherms at room temperature, the catalysts were reduced in the ad-
sorption cell in flowing hydrogen at a temperature of 500°C. After the reduction
-5
step, the system was evacuated to a pressure between low6 and 10 Torr and cooled
to room temperature for the determination of the isotherm.
The platinum-iridium and platinum-rhenium catalysts employed in the work con-
tained approximately 0.3 wt % each of platinum and the second metallic element.
In both catalysts the alumina carrier was a gamma alumina with a total surface
2 -1
area of about 180 m g .
Various naphtha fractions were employed in the reforming studies reported in
this paper. The boiling ranges and compositions of the naphthas by hydrocarbon
type are specified in the presentation of the results of the various reforming
runs.

RESULTS
The work reported in this paper includes several types of studies on platinum-
iridium and platinum-rhenium catalysts. The studies include characterization of

the catalysts by chemisorption of hydrogen and carbon monoxide, comparisons of the

catalysts for the conversion of selected hydrocarbons, and extended runs on the
reforming of petroleum naphtha fractions.

Chemisorption Studies
Data on the chemisorption of hydrogen and carbon monoxide at room temperature
were obtained on the platinum-iridium and platinum-rhenium catalysts and on relat-
ed catalysts containing a single transition metal component, i.e., platinum,
iridium, or rhenium. Typical data on the chemisorption of hydrogen at room
temperature on the platinum-iridium on alumina catalyst containing 0.3 wt % each
of platinum and iridium are given in Figure 1. Two isotherms are shown in the
figure. Isotherm A is the initial isotherm obtained on the catalyst, while iso-
therm B is a second isotherm obtained after evacuation of the adsorption cell at
room temperature for a period of 10 min (to a final pressure of approximately
10-8 Tort-). The amount of hydrogen retained by the catalyst after the evacuation
step is taken as the difference between isotherm A and isotherm B, and is labelled
A-B in Figure 1. This difference isotherm is a measure of the strongly chemisorbed
hydrogen on the catalyst.
In Table 1 data on the chemisorption of hydrogen and carbon monoxide are summar-
ized for the various catalysts of interest in this investigation. The quantity
5 125-
s ._)*- A
I
____.-*-
E, loo- ___--*--
2
0.3% Pt, 0.3% Ir
!7J -*-_._
f 75_ -_-_-___ -A-S _ •-*-.-r,,.-*_._.L.~L -
\ . .
U
L.
z- %.
D 50- “‘K
:: o-0.0
-0-o 06%Pt -
P .-.-e-S
Y”-eio_, _
g 25- -_.-*-

0
>
I I I I goO 100 200
1 300 400
, 500
I

OO 50 100 150 200

Pressure, TORR Hours on Stream

FIGURE 1 Typical isotherms at room temperature for adsorption of hydrogen on an

alumina supported platinum-iridium catalyst containing 0.3 wt % each of platinum
and iridium. Isotherm A represents the total adsorption of hydrogen on the catalyst.
Isotherm B represents the weakly adsorbed hydrogen, determined by running a second
isotherm after evacuation of the adsorption cell for 10 min at room temperature
(to a pressure of approximately 10S6 Tort-) following the completion of isotherm A.
The difference between the isotherms, A-B, represents strongly chemisorbed hydrogen.

FIGURE 2 Comparison of alumina supported platinum-iridium and platinum catalysts

for the reforming of a 50-200°C boiling range naphtha at 487°C and 14.6 atm. The
naphtha weight hourly space velocity was 2.1 and the mole ratio of hydrogen to
naphtha at the reactor inlet was approximately 6. The research octane number of
the liquid reformate (C,+) is shown as a function of time.

H/M represents the number of hydrogen atoms chemisorbed per atom of metal in the

catalyst. The hydrogen chemisorption is expressed in this way to reflect the

dissociation of hydrogen molecules into atoms during chemisorption [8]. The
quantity CO/M represents the number of carbon monoxide molecules chemisorbed per
atom of metal in the catalyst. In the case of the carbon monoxide chemisorption
data all of the values in Table 1 correspond to differences of two isotherms such
as those labelled A and B for hydrogen in Figure 1. In the case of the hydrogen
chemisorption data, difference isotherms were not determined for all of the
catalysts of interest. Therefore, we report two quantities under the heading H/M
in the table. In the column labelled I the values were determined by extrapolating
the approximately linear regions of the isotherms back to zero pressure. This
 331

TABLE 1
Summary of chemisorption data on Pt-Ir, Pt-Re and related catalysts

Catalysta Calcination Temp. Reduction Temp. H/Me

/Yb /"CC CO/Md I II

0.3% Ir 270 500 1.8 2.5 2.0

500 500 0.3 0.5 -

0.3% Pt, 0.3% Ir 270 500 1.2 1.7 1.3

500 500 0.8 1.2 -

0.3% Pt 500 500 0.8 1.0 0.6

0.3% Pt, 0.3% Re 510 500 0.7 0.4 -

0.3% Re 510 500 0.7 0.0 -

aAlumina was the carrier in all of the catalysts.

bCalcinations conducted in an oxygen-helium stream (20% 02) for a period of 4 h.
'Reductions conducted in flowing hydrogen for 2 h.
d
Molecules of carbon monoxide chemisorbed per metal atom in the catalyst at room
temperature.
eAtoms of hydrogen chemisorbed per metal atom in the catalyst at room temperature.
Column I is determined by extrapolating the adsorption isotherm back to zero
pressure, while column II is the amount of adsorbed hydrogen retained when the
adsorption cell is subsequently evacuated at room temperature for 10 min to a
pressure of approximately IO+ Torr.

method of treating chemisorption data is frequently employed [9]. In the column

labelled II the values were determined from difference isotherms such as that shown
in Figure 1.
In the case of platinum dispersed on alumina, it is generally accepted that the
stoichiometry of hydrogen chemisorption corresponds to the adsorption of one atom
of hydrogen per surface platinum atom 181. In the chemisorption of carbon monoxide,
there is evidence of some contribution of an adsorbed species requiring two
platinum atoms per carbon monoxide molecule. However, the form requiring only
one surface platinum atom appears to be the dominant one [8]. On the basis of
these considerations, one would expect that values of CO/M and H/M would approach
a maximum value of unity when the dispersion of the platinum is very high, i.e.,
when the platinum atoms are virtually all surface atoms. The data in Table 1
are consistent with this expectation. Although data for platinum catalysts
calcined at different temperatures are not given in Table 1, we have found that
calcination temperature has virtually no effect over the range from 270 to 500°C.
In the case of iridium dispersed on alumina, the data on the catalyst calcined
332

at 270°C indicate that more than one atom of hydrogen or one molecule of carbon
monoxide is adsorbed per surface iridium atom. Observations of this kind have
been noted before [lO,ll]. Furthermore, the temperature of calcination of the
iridium catalyst has a marked effect on the chemisorption properties, which has
also been noted before [5,9,10]. Calcination in air or oxygen at 500°C leads to
formation of large Ir02 crystallites which yield large iridium crystallites on
reduction. The fraction of the iridium atoms present as surface atoms thus decreases
markedly, and this in turn is reflected in the chemisorption data.
In the case of the platinum-iridium catalyst calcined at 270°C the chemisorption
values of CO/M and H/M are intermediate between those for the platinum catalyst
and the iridium catalyst calcined at 27O"C, as might have been expected. Indeed,
the values for the platinum-iridium catalyst are very close to the average of the
values for the platinum and the iridium catalysts. When the data on the platinum-
iridium catalyst calcined at 500°C are examined, however, we find that the values
of CO/M and H/M are higher than the averages for the platinum and the iridium
catalysts calcined at 500°C. While calcination of the platinum-iridium catalyst
at 500°C also leads to formation of large Ir02 crystallites, and subsequently to
large iridium or iridium-rich crystallites on reduction [5,11], the deleterious
effect of the calcination is clearly not so pronounced as it is in the case of
the iridium catalyst which contains no platinum. The presence of the platinum
thus serves to inhibit the oxidative agglomeration of the iridium. This
observation is consistent with the view that the platinum and iridium in the
catalyst are present as bimetallic clusters rather than separate clusters of pure
platinum and iridium [53. The observation that the iridium is less susceptible
to oxidative agglomeration in the presence of the platinum has the important
practical feature that the platinum-iridium catalyst is more stable than a pure
iridium catalyst in the oxygen containing gases employed for the regeneration of
the catalysts. In general, the regeneration procedures employed with platinum-
iridium catalysts will differ from those employed with platinum catalysts because
of the greater susceptibility of iridium to oxidative agglomeration.
Data on the rhenium on alumina catalyst reduced in hydrogen at 500°C indicate
significant chemisorption of carbon monoxide, but no chemisorption of hydrogen.
Correspondingly, the platinum-rhenium catalyst exhibits a value of CO/M about
twice as high as the value of H/M. The latter approximates what one would expect
if hydrogen chemisorbed on only the platinum component of the catalyst.

Hydrocarbon Reaction Studies

Data on rates of dehydrocyclization rT, and cracking rC of n-heptane at 495°C
and 14.6 atm are given in Table 2 for platinum-iridium on alumina and platinum-
rhenium on alumina catalysts, and also for catalysts containing platinum or
iridium alone on alumina. The rate rb refers to the rate of production of
 333

TABLE 2
n-Heptane conversion on Pt-Ir, Pt-Re and related catalystsa

Catalystb Rates/(Mole/h/g catalystlc Selectivity

rD rC W'C

0.3% Ir 0.051 0.103 0.50

0.3% Pt, 0.3% Ir 0.051 0.092 0.55
0.3% Pt 0.021 0.050 0.42
0.3% Pt, 0.3% Re 0.031 0.070 0.44

aThe data were obtained with n-heptane containing 0.5 ppm sulfur after 40 h on
stream at 495°C and 14.6atmusinga5/1 mole ratio of hydrogen to n-heptane. The
catalysts contained 0.9 wt % Cl as charged.
b
Alumina was employed as the carrier in all of the catalysts.
'The quantity rD refers to the rate of dehydrocyclization of n-heptane to toluene
and C7 cycloalkanes, while rC is the rate of cracking to C and lower carbon
6
numberalkanes. The rates were determined at conversion levels of 7-12%.

toluene and C7 cycloalkanes, the latter consisting primarily of methylcyclohexane

and dimethylcyclopentanes. The rate of cracking refers to the rate of conversion

of n-heptane to C6 and lower carbon number alkanes. The reaction rates were
determined at low conversion levels (7-12%) in an attempt to minimize the effects
of secondary reactions. The inlet stream to the reactor contained five moles of
hydrogen per mole of n-heptane. The n-heptane contained 0.5 ppm sulfur and the
reaction rates were determined after 40 h on stream. Prior to the runs the
catalysts were pre-treated in the manner described in the Experimental Section
of this paper.
The rates of dehydrocyclization of n-heptane for the iridium and platinum-
iridium catalysts are more than twice as high as that for the platinum catalyst
and almost twice as high as that for the platinum-rhenium catalyst. The rates
of cracking are also higher for the iridium and platinum-iridium catalysts. The
selectivity, defined as the ratio rD/rC of the rate of dehydrocyclization to the
rate of cracking, is higher for the iridium-containing catalysts. It is significant
that the selectivity of the platinum-iridium catalyst is higher than that of the
catalyst containing iridium alone. The selectivity of the platinum-rhenium
catalyst appears to be slightly higher than that of the catalyst containing
platinum alone.
In Table 3 data are presented for the conversion of methylcyclopentane at
500°C and 14.6 atm over the same catalysts for which data on n-heptane reactions
are presented in Table 2 and also for a rhenium on alumina catalyst. The
selectivity to benzene and to C,-C6 alkanes, expressed as percentages of the
334

TABLE 3
Methylcyclopentane conversion over Pt-Ir, Pt-Re and related catalystsa

Catalystb Time/h % Conversion % Selectivity to

Benzene C,-C6 alkanes

0.3% Ir 0.5 33 53 47

24 42 39 61

0.3% Pt, 0.3% Ir 0.5 42 71 29

24 43 62 38

0.3% Pt 0.5 40 79 21

24 17 66 34

0.3% Pt, 0.3% Re 0.5 38 71 29

24 36 76 24

0.3% Re 0.5 17 16 84

24 13 25 75

aThe data were obtained with methylcyclopentane containing 1 ppm sulfur at 500°C
and 14.6 atm using a methylcyclopentane weight hourly space velocity of 40 and
a 5/l mole ratio of hydrogen to methylcyclopentane. The catalysts contained
0.9 wt % Cl as charged.
b
Alumina was the carrier in all of the catalysts.

total conversion of methylcyclopentane to all products, is shown after 0.5 h and

after 24 h on stream. The inlet stream to the reactor contained five moles of
hydrogen per mole of methylcyclopentane. The methylcyclopentane contained 1 ppm
sulfur, and the catalysts were pretreated in the same manner as they were in the
n-heptane conversion studies. The selectivity to benzene is much lower for the
iridium catalyst than for any of the other catalysts except rhenium on alumina.
The platinum-iridium catalyst is clearly more selective than the iridium catalyst
with respect to selectivity of benzene formation. However, it is less selective
for benzene formation than the catalyst containing platinum alone, although it
is possible that this debit may disappear as the catalyst ages during a run.
Also, except for the initial reaction period, the platinum-iridium catalyst is
less selective for benzene formation than the platinum-rhenium catalyst. The
selectivity of the platinum-rhenium catalyst for benzene formation is initially
lower than that of the platinum catalyst, but after 24 h on stream it is higher.
It may be noted that the maintenance of selectivity for benzene formation with
time on stream is better for the platinum-rhenium catalyst than for the platinum
catalyst or either of the iridium-containing catalysts. The better selectivity
maintenance may be associated with a higher retention of the sulfur incorporated
 335

during the pretreatment procedure.

From the foregoing studies of n-heptane and methylcyclopentane conversion
we can see some reasons why platinum-iridium and platinum-rhenium catalysts have
advantages over catalysts containing only one of the transition metal components,
i.e.,platinum, iridium or rhenium. If, for example, we consider an iridium-
alumina catalyst for the reforming of a petroleum naphtha fraction, we will find
that it produces a substantially higher octane number reformate than a platinum
on alumina catalyst under normal reforming conditions. The iridium-alumina
catalyst will also exhibit a lower rate of formation of carbonaceous residues on
the surface, with the result that the maintenance of activity with time is much
superior to that of a platinum-alumina catalyst. On the basis of these consider-
ations, the use of an iridium-alumina catalyst in reforming appears very
reasonable [12,13]. However, in our experience the yields of low molecular weight
alkanes (methane and ethane) are higher with an iridium-alumina catalyst than
with a platinum-alumina catalyst, with the result that yields of C5+ reformate
are lower. In view of the higher value of C5+ reformate relative to products
such as methane and ethane, the iridium-alumina catalyst is not used, despite
its higher activity and better activity maintenance.
The inclusion of platinum with iridium in bimetallic clusters [5] provides
a way of moderating the formation of the low molecular weight alkanes while still
retaining activity and activity maintenance much superior-to that of a platinum
catalyst. Our work has repeatedly indicated the desirability of this interaction
between platinum and iridium in obtaining a satisfactory product distribution,
i.e., a product distribution in which the yield of C5+ reformate of a given
octane number is equivalent to, or better than, the yield obtained with a
platinum-alumina catalyst. Experiments on reforming with physical mixtures of
platinum-alumina and iridium-alumina catalysts conducted in these laboratories
have shown that yields of C5+ reformate of a given octane number are several
percent lower than those obtained on a catalyst prepared by a procedure designed
to enhance the formation of bimetallic clusters of platinum and iridium. With
a satisfactory product distribution, the higher activity and improved activity
maintenance of the platinum-iridium catalyst make it very attractive relative
to platinum on alumina catalysts.
In Figure 2 data are shown for the reforming of a 50-200°C boiling range
Venezuelan naphtha over a bimetallic catalyst containing 0.3 wt % each of
platinum and iridium and a catalyst containing 0.6 wt % platinum. The naphtha
contained (on a volume percent basis) 47.3% paraffins, 42.2% naphthenes, and
10.5% aromatics and had a density of 0.7605 g cmq3. The data show the research
octane number of the C5+ liquid reformate as a function of time on stream. The
octane number is substantially higher for the platinum-iridium catalyst than for
the platinum catalyst, and the decline with time is much lower. The lower rate
of deactivation of the catalyst indicated by the lower decline of the reformate
336

octane number with time is also a key feature of platinum-rhenium catalysts.

The improved activity maintenance is accompanied by improved maintenance of
selectivity for the production of C5+ reformate. As a consequence, platinum-
iridium and platinum-rhenium catalysts both exhibit significant advantages
relative to the older platinum-alumina catalysts. Detailed comparisons of the
reforming characteristics of platinum-iridium and platinum-rhenium catalysts are
presented in the next section of the paper.

Extended Naphtha Reforming Runs

In this section some results of extended naphtha reforming runs are presented
to illustrate further the differences between platinum-iridium and platinum-rhenium
catalysts discussed in the previous section. The results presented should be
considered illustrative only, since they are limited to a series of Persian Gulf
naphtha fractions and do not encompass a wide range of operating conditions. The
reactors employed in the runs are nominally isothermal, but there is some variation
in temperature along the length of the catalyst bed. The temperature variation
is most pronounced at the reactor inlet because of the very rapid and highly
endothermic conversion of cyclohexanes to aromatics. The average reaction
temperature was generally about 5°C lower than the temperature of the fluidized
solids bath.
In the extended reforming runs for which results are presented in Figures 3-8,
detailed data were obtained for a 3 h reaction period once a day. During such a
period the amounts of liquid and gaseous products were determined, and the research
clear octane number of the liquid product was determined. An analysis was
routinely made of the components in the gaseous product (Hz and C,-C5 hydrocarbons)
and of any C4 or lower carbon number components retained in the liquid product
in the separation of liquid and gaseous fractions from the total reactor effluent.
In this way it was possible to obtain the yields of C5+ reformate and of the
individual components of the gaseous fraction. Material balances were excellent,
the recovery of the product streams ranging between 99 and 101% of the inlet
naphtha stream. Data were obtained for approximately 60 to 90 reaction periods
in the reforming runs considered here. To simplify the presentation of the data,
we have averaged the results of eight consecutive periods of operation in
obtaining the individual data points in Figures 3-8 and have taken the mid-point
of the overall time interval as the time for a given point. In comparing data
on product yields for the catalysts, it was found that the differences were very
small for some of the products. To minimize the confusion resulting from many
overlapping data points in the yield plots, the data on the two catalysts have
been averaged over time periods which are slightly displaced. For consistency,
this procedure has been adopted in the presentation of data on catalytic activities
as well as yields for the platinum-rhenium and platinum-iridium catalysts in
Figures 3-8. Thus the first point for the platinum-rhenium catalyst is obtained
 ;;_I
0 500 1000 1500 2000
Hours on Stream Hours on Stream

FIGURE 3 Comparison of activities of alumina supported platinum-iridium and

platinum-rhenium catalysts for the reforming of a 70-190°C boiling range Persian
Gulf naphtha to produce 98 research octane number reformate. The runs were made
at 490°C and 28.2 atm. The mole ratio of recycle gas to naphtha at the reactor
inlet was approximately 7. The catalysts contained 0.3 wt % each of platinum and
the second metallic element, iridium or rhenium.

FIGURE 4 Comparison of product yields obtained with platinum-iridium and

platinum-rhenium catalysts in the reforming runs for which activity data are shown
in Figure 3: platinum-iridium catalyst,@; platinum-rhenium catalyst,A.

by averaging data for reaction periods l-8, the second point by averaging data

for periods 9-16, etc., The first point fpr the platinum-iridium catalyst is
obtained by averaging data for reaction periods 5-12, the second for periods
13-20, etc. In Figures 7 and 8, where data are included for a run on a combined
catalyst system in which platinum-rhenium is followed by platinum-iridium, the
data for the combined system have been averaged over the same groups of eight
reaction periods that were used for the run on the platinum-iridium catalyst.
The initial weight hourly space velocity of naphtha fed to the reactor (g
338

-i I-35ATM
I', \... '. >.n.n .,, . n.n Hz
.n
05 1

%
5
4oor---7
zoo-
. . . . .
.

.
.PW
.

.
"
.
h
.
I3
. .
n
l

n
. n
*Pt.lr "ST
*Pt.Re

620 500 1000 1500 2000 2500 1

500 500 1000 1500 2000 2500
Hours on stream

FIGURE 5 Comparison of activities of alumina supported platinum-iridium and

platinum-rhenium catalysts for the reforming of a 65-150°C boiling range Persian
Gulf naphtha to produce 96 research octane number reformate. The runs were made
at 488°C and 28.2 atm. The data in the approximate time interval from hour 1460
to hour 1710 were obtained at 35 atm. The mole ratio of recycle gas to naphtha at
the reactor inlet was approximately 7. The catalysts contained 0.3 wt % each of
platinum and the second metallic element, iridium or rhenium.

FIGURE 6 Comparison of product yields obtained with platinum-iridium and platinum-

rhenium catalysts in the reforming runs for which activity data are shown in
Figure 5: platinum-iridium catalyst,*; platinum-rhenium catalyst,n.

of naphtha per h per g of catalyst) in the runs reported in this section ranged
approximately from 0.8 to 3.2 depending on the catalyst, the naphtha being
reformed, and the operating conditions. Compensation for variation of catalyst
activity with time on stream was made by changing the weight hourly space velocity.
In this way the research octane number of the liquid reformate, which is largely
determined by the aromatics content of the reformate, is held constant at the
desired value. Thus, if the octane number decreases, it can be restored to the
 I I
500 1000 1500 2000
500 1000 1500 2000
Hours on Stream Hours on Stream

FIGURE 7 Comparison of reforming activities of alumina supported platinum-iridium

and platinum-rhenium catalysts with that of a combined catalyst system in which
platinum-rhenium is followed by platinum-iridium. In the combined catalyst system
platinum-iridium constituted 65 wt % of the total catalyst charge. The data
represent catalytic activities for the reforming of a 65-190°C boiling range
Persian Gulf naphtha to produce 98 research octane number reformate at temperatures
of 488-500°C and a pressure of 28.2 atm. The mole ratio of recycle gas to naphtha
at the reactor inlet was approximately 7. The catalysts contained 0.3 wt % each
of platinum and the second metallic element, iridium or rhenium.

FIGURE 8 Comparison of product yields obtained in the reforming runs for which
activity data are shown in Figure 7: platinum-iridium catalyst,.; platinum-
rhenium catalyst,n; combined catalyst system,O.

desired value by a decrease in space velocity. The space velocity required to

produce a given octane number reformate is a direct measure of catalytic activity.
In devising an activity scale such as that employed in Figures 3, 5 and 7, one
assigns a value (say 100) to the activity of a reference catalyst which produces
a certain octane number product at a standard set of conditions (space velocity,
340

temperature, hydrogen partial pressure, naphtha partial pressure). The activity

of a catalyst being evaluated, either in the fresh state or after a substantial
time on stream, is obtained simply by determining the space velocity required
to produce the same octane number product at the standard conditions. Empirical
relations between reformate octane number and naphtha space velocity as a function
of temperature, hydrogen partial pressure and other process variables for the
reference catalyst make it possible to determine the activity of a test catalyst
from data at a variety of conditions. For the purposes of this paper the
absolute values shown for the activities are not important, since they simply
reflect activities relative to that of an arbitrary reference catalyst. We will
be concerned strictly with ratios of the activity values shown in Figures 3,5,
and 7 for the different catalyst systems.
In Figures 3 and 4 data on platinum-iridium and platinum-rhenium catalysts
are shown for the reforming of a 70-190°C boiling range Persian Gulf naphtha
to produce 98 research octane number product at a pressure of 28.2 atm and a
temperature of 490°C. The mole ratio of recycle gas to naphtha was approximately
7. The naphtha contained (on a liquid volume percent basis) 69.7% paraffins,
18.5% cycloparaffins , and 11.8% aromatic hydrocarbons. The density of the
-3
naphtha was 0.7414 g cm . From the data in Figure 3 we see that the platinum-
iridium catalyst is almost twice as active as the platinum-rhenium catalyst
throughout the run. Deposition of carbonaceous residues on the platinum-iridium
catalyst was less than half that on the platinum-rhenium catalyst (2.9 wt % vs.
6.9 wt % on catalyst at the ends of the runs). In general, we have observed the
rate of deposition of such residues on platinum-iridium catalysts to be lower
than on platinum or platinum-rhenium catalysts under conventional reforming
conditions. In Figure 4 the yields of the various products are shown. The
yield of C5+ reformate is higher for the platinum-rhenium catalyst, the average
difference amounting to about 1 vol % for the final two thirds of the runs.
Correspondingly, the yields of C, and C2 hydrocarbons (methane and ethane) are
lower for the platinum-rhenium catalyst. The yields of C3 and C4 hydrocarbons
are very nearly the same for the two catalysts, while the hydrogen yield is
higher for the platinum-rhenium catalyst. The higher hydrogen yield is consistent
with the lower yields of methane and ethane. As a result of the higher yields
of methane and ethane in the run on the platinum-iridium catalyst, the hydrogen
concentration in the recycle gas stream was lower than in the run on the
platinum-rhenium catalyst. The average inlet hydrogen partial pressures were
15.1 and 16.5 atm respectively, during the runs on the platinum-iridium and
platinum-rhenium catalysts. The difference in hydrogen partial pressure at a
fixed total pressure is a consequence of the different compositions of the
gaseous products, which in turn reflect differences in the kinetics of the
various reactions occurring over the catalysts. This feature is characteristic
 341

of all the comparisons we will be considering. Consequently, the C5+ yield

observed in a run on platinum-iridium catalyst in which the light gaseous pro-
ducts are recycled is higher than it would be if the hydrogen partial pressure
were the same as that characteristic of a similar run on platinum-rhenium
catalyst. In the case of the wide boiling range naphtha just considered, the
C5+ yield obtained with the platinum-rhenium catalyst was about 1 vol. %
higher than that obtained with the platinum-iridium catalyst despite the fact
that the hydrogen partial pressure was lower in the run on the latter. This
result is a typical one for such wide boiling range naphthas. In the case of

light naphtha fractions such as the one to be considered next, however, the
effect of the lower hydrogen partial pressure in the case of the platinum-iridium
catalyst can be large enough so that the C5+ yield becomes higher than that
obtained with the platinum-rhenium catalyst.
In Figures 5 and 6 data are presented for the reforming of a 65-150°C boiling
range Persian Gulf naphtha at a temperature of 488°C to produce 96 research
octane number reformate. The pressure was 28.2 atm except for a period at 35.0 atm
in the approximate time interval from hour 1460 to hour 1710 in each run. The
-3
naphtha had a density of 0.7243 g cm and contained (on a liquid volume percent
basis) 70.2% paraffins, 21.1% naphthenes, and 8.6% aromatics. The mole ratio
of recycle gas to naphtha was approximately 7 in these runs also. As shown in
Figure 5, the platinum-iridium catalyst is more than twice as active as the
platinum-rhenium catalyst for most of the time on stream. In Figure 5 there
appears to be no deactivation of the platinum-iridium catalyst out to hour 2000,
and in Figure 2 discussed earlier the rate of deactivation appears to be nil
after hour 200. However, it should be appreciated that the platinum-iridium
catalyst, like all reforming catalysts, deactivates with time. Frequently there
is an extended period of operation in which the extent of deactivation is

practically nil. However, the deactivation rate in such cases will eventually
increase significantly with increasing time on stream, and regeneration of the
catalyst will be necessary. Data on product yields are shown in Figure 6. In
contrast to the results on the 70-190°C Persian Gulf naphtha, the C5+ yield for
the platinum-iridium catalyst is at least 1 vol % higher than that for the
platinum-rhenium catalyst. The yields of Cl and C2 hydrocarbons are still
higher for the platinum-iridium catalyst than for the platinum-rhenium catalyst,
but the yields of C3 and C4 hydrocarbons are lower. The hydrogen yields are
about the same for the catalysts. The average hydrogen partial pressures were
13.5 and 15.3 atm, respectively, during the runs on the platinum-iridium and
platinum-rhenium catalysts.
Reforming data on a 65-190°C boiling range Persian Gulf naphtha similar to
the 70-190°C fraction considered earlier are given in Figures 7 and 8 for the
same platinum-rhenium and platinum-iridium catalysts and also for a combined
catalyst system in which a zone of platinum-rhenium catalyst is followed by a
342

zone of platinum-iridium catalyst. In the combined catalyst system the platinum-

iridium catalyst constituted 65% of the total catalyst charge. The runs were
conducted at a pressure of 28.2 atm and at temperatures of 488-500°C to produce
98 research octane number reformate. Again the mole ratio of recycle gas to
-3
naphtha was approximately 7. The naphtha had a density of 0.7421 g cm and
contained (on a liquid volume percent basis) 66.9% paraffins, 21.9% naphthenes
and 11.1% aromatics. In Figure 7 we see again that platinum-iridium catalyst
is approximately twice as active as platinum-rhenium. The activity of the
combined catalyst system is intermediate between the activities of the individual
platinum-iridium and platinum-rhenium catalysts. In Figure 8 we see that C5+
yields are equivalent for the platinum-rhenium catalyst and the combined catalyst

system and about 1.0 to 1.5 vol. % higher than for the platinum-iridium catalyst.
Methane and ethane yields for the combined catalyst system are higher than those
for the platinum-rhenium catalyst but lower than those for the platinum-iridium
catalyst. Yields of H2 are about equivalent for the combined catalyst system
and the platinum-iridium catalyst, and are lower than those for the platinum-
rhenium catalyst. Similarly, the yields of C3 and C4 hydrocarbons are about
equivalent for the platinum-iridium catalyst and the combined catalyst system,
but are lower than the yields for the platinum-rhenium catalyst. The concentrat-
ion of H2 in the recycle gas was higher in the run on the platinum-rhenium
catalyst than in the other two runs. Consequently, the hydrogen partial pressure
at the reactor inlet was higher, averaging about 16.5 atm during the run compared
to 14.2 atm and 14.4 atm, respectively, for the runs on the platinum-iridium
catalyst and the combined catalyst system. We emphasize that this comparison of
the combined catalyst system with either catalyst alone is just one example.
Data in Table 4 on average activities and C5+ yields obtained in a comparison
on a 75-150°C Persian Gulf naphtha fraction illustrate the sensitivity to the
boiling range of the naphtha. The data represent a period of 500 hours of
operation subsequent to five months of operation on the 65-150°C fraction. The
activity of the platinum-iridium catalyst or of the combined catalyst system
remains higher than that of the platinum-rhenium catalyst. However, the C5+
yield for the combined system is 0.6 vol % lower than the yield for either the
platinum-rhenium or the platinum-iridium catalyst. Note also that the C5+ yield
advantage observed for the platinum-iridium catalyst over the platinum-rhenium
catalyst in the particular case of the 65-150°C naphtha is no longer observed
when the boiling range is changed to 75-150°C. In general, comparisons of
either catalyst alone with a combined catalyst system will depend on the naphtha
being reformed, the reforming operating conditions, and also on the relative
amounts of the two catalysts employed in the combined system.
In the results presented thus far, comparisdns of catalysts have been made
with regard to activities and yields of various products. The products have
 343

TABLE 4
Comparison of Pt-Re and Pt-Ir catalysts with a combined Pt-Ir/Pt-Re catalyst
system in the reforming of a 75-150°C Persian Gulf naphthaa

Catalyst Activity C5+ Yield /vol %

Pt-Re 63 70.2
65% Pt-Ir, 35% Pt-Re 143 69.6
Pt-Ir 142 70.2

aReforming at 28.2 atm and 488°C to produce 96 research octane number reformate.
The data are average values for 500 h reforming periods after approximately five
months of operation on a 65-150°C Persian Gulf naphtha fraction.

included H2, C,-C4 hydrocarbons, and the C5+ liquid reformate. The yields of
specific aromatic hydrocarbons in the reformate are also of interest. Data on
yields of benzene, toluene, and total aromatic hydrocarbons in the reformate are
summarized in Table 5 for platinum-iridium and platinum-rhenium catalysts in the
reforming of the various naphtha fractions. In the case of one of the naphthas,
data are also shown for the combined catalyst system. The data shown are average
values for the runs. In comparing the platinum-iridium and platinum-rhenium
catalysts, we note that the yields of benzene, toluene, and total aromatics in
the C5+ reformate are consistently higher for the platinum-iridium catalyst.
In the case of the combined catalyst system the yields of benzene and toluene
are equivalent to those observed with the platinum-iridium catalyst.

DISCUSSION
Platinum-iridium and platinum-rhenium catalysts both exhibit lower rates of
deactivation in reforming than platinum-alumina catalysts. In the case of
platinum-iridium catalysts, the improved activity maintenance can be attributed,
at least in part, to the lower rate of formation of carbonaceous residues on
the catalyst. The decreased formation of such residues may be attributed to
the increased hydrogenolysis activity resulting from the presence of the iridium.
It is known that the hydrogenolysis activity of iridium is much higher than that
of platinum [14]. The lower rate of formation of carbonaceous residues relative
to that observed with a catalyst containing only the platinum is an indication
of the interaction between platinum and iridium in the catalyst. It seems likely
that hydrogenolysis provides a mechanism for removal of the residues or precursors
of the residues from the surface. In the case of platinum-rhenium catalysts,
the improved activity maintenance relative to platinum catalysts can not be
rationalized in this manner since the rate of formation of carbonaceous residues
is not significantly affected by the presence of the rhenium. This point is
344

TABLE 5
Summary of aromatic hydrocarbon yields in reforming with Pt-Ir and Pt-Re catalysts

Yields/wt % Catalyst
Pt-Ira Pt-Reb Combined catalystC

From 70-190°C Persian Gulf naphthad

Benzene 2.6 2.1

Toluene 13.0 11.2
Total aromatics 46.9 46.1

From 65-150°C Persian Gulf naphthae

Benzene 5.2 4.4

Toluene 18.7 15.5
Total aromatics 46.2 44.3

From 65-190°C Persian Gulf naphthaf

Benzene 4.1 3.3 4.1

Toluene 14.4 12.3 14.4
Total aromatics 50.2 50.0 51.7

aCatalyst contains 0.3 wt % each of platinum and iridium on alumina.

b
Catalyst contains 0.3 wt % each of platinum and rhenium on alumina.
'Pt-Re catalyst followed by Pt-Ir catalyst, the latter constituting 65 wt % of
the total.
dReforming at 28.2 atm and 490°C to produce 98 research octane number reformate.
The yields are averages for 1600 h on stream.
eReforming at 28.2 atm and 488°C to produce 96 research octane number reformate.
The yields are averages for 2100 h on stream.
fReforming at 28.2 atm and temperatures of 488-500°C to produce 98 research
octane number reformate. The yields are averages for 1500 h on stream.

interesting because metallic rhenium also has much higher hydrogenolysis activity
than platinum [14]. It is possible that the difference between platinum-rhenium
and platinum-iridium catalysts is due to the strong retention of sulfur by the
former. The inhibiting effect of sulfur on hydrogenolysis activity is well
known. The improved activity maintenance of a platinum-rhenium catalyst relative
to a platinum catalyst is achieved despite the fact that the formation of
carbonaceous residues is essentially unaffected.
When considering the beneficial effect of the high hydrogenolysis activity
of iridium in suppressing formation of carbonaceous residues, it must be remember-
ed that such activity may simultaneously have an adverse effect on the product
distribution in reforming. A catalyst containing iridium alone on alumina is
 345

limited in this manner. In the case of platinum-iridium catalysts, the inter-

action between the platinum and iridium moderates the hydrogenolysis activity of
the latter. By maintaining the proper level of hydrogenolysis activity in this

way, one can obtain a satisfactory product distribution and at the same time
realize large advantages in activity and activity maintenance over catalysts
containing platinum alone on alumina.
In comparing the reforming properties of platinum-iridium and platinum-rhenium
catalysts, we note first that the activity of platinum-iridium catalysts is
consistently higher than that of platinum-rhenium catalysts. The activity of a
0.3% Pt, 0.3% Ir catalyst is generally about twice as high as that of a 0.3% Pt,
0.3% Re catalyst for naphtha reforming. With regard to product distribution,

platinum-iridium catalysts usually give slightly higher yields of methane and

ethane than platinum-rhenium catalysts. Correspondingly, the yield of C5+

reformate obtained with a platinum-rhenium catalyst is usually higher, typically

about 1 vol % higher for wide boiling range naphthas. Under certain conditions
with light naphtha fractions this result is reversed, and the C5+ yield is about
1 vol % higher for the platinum-iridium catalyst. In the case of benzene and
toluene, which are of interest for petrochemicals, the yields are substantially
higher with platinum-iridium catalysts.
The attractive features of platinum-rhenium and platinum-iridium catalysts
can be combined in a reforming operation. Data on the reactions of selected
hydrocarbons on platinum-rhenium and platinum-iridium catalysts indicate that
the former is more selective for the conversion of cycloalkanes to aromatics while
the latter is more selective for the dehydrocyclization of alkanes. Since
cycloalkane conversion occurs primarily in the initial part of a reforming
system while dehydrocyclization is the predominant reaction after the cycloalkanes
have reacted, it is reasonable to use a platinum-rhenium catalyst in the front
of the system and to follow it with a platinum-iridium catalyst [15]. When this
is done, one can obtain the high activity and better yields of benzene and toluene
characteristic of a platinum-iridium catalyst and at the same time obtain a C +
5
reformate yield more nearly equivalent to that of a platinum-rhenium catalyst
in those cases where the latter has an advantage over a platinum-iridium catalyst
in this respect.
The choice of a catalyst system for a reforming unit is affected by a number
of factors, including the activity of the catalyst system and the yields of the
various products obtained with it. The relative importance of the different
factors, and hence the economic analysis, depends on the particular application.

CONCLUSION
The introduction of platinum-alumina catalysts for the reforming of petroleum
naphthas over three decades ago Cl&183 was a major advance in catalytic
technology, and was responsible for the widespread application of the catalytic
346

reforming process in petroleum refining. The introduction of bimetallic catalyst

systems about two decades later represented the first major improvement in re-
forming catalysts. The outstanding performance of these new catalysts has led
to their use in refineries throughout the world. While the bimetallic reforming
catalysts currently in use exhibit outstanding performance, it is highly possible
that improved modifications of the existing systems will emerge as a result of
future research in the area.

ACKNOWLEDGEMENT
The authors gratefully acknowledge the very capable assistance of A.E. Barnett,
P.J. Collins, J.J. Ziemiak, R.J. Draina, and R.H. Ernst in conducting the
experiments.

REFERENCES
J.H. Sinfelt, Adv. in Chem. Eng. 5 (1964) 37.
: F.G. Ciapetta and D.N. Wallace, Cat. Rev. 5 (1971) 67.
H.E. Kluksdahl, U.S. Patent 3,415,737, (1968).
R.L. Jacobson, H.E. Kluksdahl, C.S. McCoy, and R.W. Davis, Proceedings
of the American Petroleum Institute, Division of Refining, 49 (1969) 504.
J.H. Sinfelt, U.S. Patent 3,953,368, (1976).
J.L. Carter and J.H. Sinfelt, U.S. Patent 4,220,520, (1980).
J.H. Sinfelt and D.J.C. Yates, J. Catal. 8 (1967) 82.
J.H. Sinfelt, Progress in Solid State Chemistry, 10 (1975) 55.
R.L. Garten and J.H. Sinfelt, J. Catal. 62 (1980) 127.
G.B. McVicker, R.T.K. Baker, R.L. Garten and E.L. Kugler, J. Catal. 65
(1980) 207.
J.H. Sinfelt and G.H. Via, J. Catal. 56 (1979) 1.
J.H. Sinfelt, U.S. Patent 3,684,693, (1972).
J.H. Sinfelt, U.S. Patent 3,871,996, (1975).
J.H. Sinfelt, Adv.in Catal. 23 (1973) 91.
J.H. Sinfelt, U.S. Patent 3,791,961, (1974).
16 V. Haensel, U.S. Patent 2,479,109, (1949).
17 V. Haensel, U.S. Patent 2,479,110, (1949).
18 V. Haensel and G.R. Donaldson, Ind. Eng. Chem., 43 (1951) 2102.