JOURNAL OF CATALYSlS 45, 41-53 (1976)

Chemistry of Nickel-Alumina Catalysts

CALVIN H. BARTHOLOMEW
Drpartnwnt of Chemical Engineering Science, Brigham Young University, Provo, Utah 84602

-4lvn

ROBERT J. FARRAUTO’

Research and Devrlopment Laboratory, Corning Glass Works, Corning, New York l&90
Received February 10, 1976

A study of the chemistry involved in preparation, pretreatment, and reduction of Ni/A1203

catalysts was carried out using hydrogen chemisorption, thermal gravimetric analysis, and
chromatographic analysis. Data showing effects on nickel surface area of calcination, heating
rate during reduction, nickel loading, and passivation are presented and discussed. The results
indicate that controlled decomposition of alumina-supported nickel nitrate in hydrogen atmo-
sphere (without prior calcination) tends to maximize nickel surface area, dispersion, and
reduction to nickel metal. Nickel surface area and the fraction of nickel reduced to the metallic
state both increase as the hydrogen space velocity during reduction is increased. Upon increas-
ing the reduction temperature from 300 to 5OO”C, the percentage reduction to nickel metal is
increased from 40 to 100% whereas nickel surface area changes relatively little.

I. INTRODUCTION the chemical reactions occurring in each

Supported nickel catalysts find wide preparation step and the effects of controll-
application in many important industrial able variables on the resultant catalytic
hydrogenation processes including meth- surface.
anation of coal synthesis gas. Yet there is In laboratory or commercial preparation
little quantitative information in the litera- of catalysts by impregnation or coprecipita-
ture dealing with the chemistry of nickel tion of metal salts, it is common procedure
catalyst preparation. In such preparations to decompose the metal salts to the oxide
one is confronted with the difficult problem by calcination in air at 400-500°C followed
of understanding and controlling in a by reduction in flowing hydrogen at 400-
reproducible manner many different steps 500°C to obtain well-dispersed metals.
such as drying, calcination, reduct,ion, etc., However, previous studies (l-11) indicate
and in each step such variables as time, that nickel supported on alumina is not
t,emperature, heating rate, and gas flow completely reduced under typical reducing
rate. Naturally the goal in such prepara- conditions t,o the metallic state, but rather
t’ions is to maximize the surface area and only after many hours of exposure to
activity of each catalyst. It would be hydrogen at very high temperatures (10,
highly desirable, therefore, to understand 11). The degree of reduction of nickel in
1 Present address : Engelhard Industries, Wood alumina-supported catalysts prepared in
Ave., Menlo Park, Edison, New Jersey 08817. various ways generally increases with

41
Copyright @ 1976 by Academic Press, Inc.
All rights of reproduction in any form reserved.
42 BARTHOLOMEW AND FARRAUTO

increasing reduction temperature and in- flowing hydrogen prior to the adsorption
creasing nickel content (9, 11). Samples measurement. The amount of gas adsorbed
prepared by impregnation generally achieve by the catalyst was determined by means of
a higher degree of reduction than samples a calibrated gas buret connected to a
prepared by precipitation (7, 8). Increasing manometer backed with a metrically cal-
the temperature of the calcination step ibrated mirror. Following reduction of
prior to reduction significantly decreases catalyst samples in flowing hydrogen (50
the reducibility of nickel on alumina (2, 4, cms/min) at 450-500°C and evacuation at
5, 7, 8). Indeed, data in this study suggest 400~500°C nickel surface areas were mea-
that more complete reduction to metallic sured using hydrogen chemisorption at
nickel is possible, if the supported nickel room temperature (12). The amount of
salt is not calcined at all but is decomposed hydrogen adsorbed by the catalyst was
in hydrogen rather than air. measured after allowing 45 min for the
This study of nickel-alumina catalysts adsorption to approach equilibrium. Nat-
was undertaken first, to determine the urally, this does not represent the maximum
effects of various pretreatment steps and possible Hz uptake since there is a slow,
controllable preparation variables such as activated or diffusion-controlled adsorption
reduction temperature, space velocity, over a period of hours. Nevertheless, it
and heating rate on nickel surface area and does serve as a reproducible property of
on the extent of reduction to nickel metal, these surfaces.
and second, to unravel the chemistry of Figure 1 shows a typical room tempera-
catalyst calcination and reduction. ture adsorption isotherm for hydrogen
adsorbed on Ni/A1203. The data show that
II. EXPERIMENTAL the gas taken up by the sample approaches
A. Apparatus and Procedure saturation at about 10 cm Hg pressure
(1 cm Hg = 1333 N m-“) and that physical
Adsorption measurements were per- adsorption on the support is small (in view
formed using a conventional Pyrex glass of the almost zero slope of the isotherm).
volumetric adsorption apparatus capable The uptake due to chemisorption was found
of lo+ Torr (1 Torr = 133.3 N m-“) by extrapolating to zero ‘pressure the
vacuum. The catalyst sample was placed in straight-line portion of the isotherm above
a flowthrough cell to permit reduction in the saturation pressure. Catalyst surface
areas were calculated assuming H/Ni(,, = 1
and a surface area of 6.5 A2 per nickel atom
(10 A = 1 nm), based on an average of the
areas for the (loo), (llO), and (111) planes.
The extent of reduction to nickel metal
(reduced catalysts) was determined after
hydrogen chemisorption by evacuating the
sample to 450°C and pressures of 10m4to
10e5 Torr followed by measuring the uptake
of pure oxygen at 400-430°C. It was
250 assumed that in the reduced catalyst all
I unreduced nickel was in the form of NiO
and that at 4OCM3O”C in the presence of
oxygen all nickel in the metallic form would
FIG. 1. Hz chemisorption 9 wt% Ni/A1201; be oxidized to NiO. Comparison of the
reduced, 7 hr, 420%; evac. 2 hr, 420°C. actual oxygen uptake with amount needed
 CHEMISTRY OF NICKEL-ALUMINA CATALYSTS 43

if all of the nickel were assumed to be sieve or a liquid nitrogen trap. The carbon
initially in the metallic state constituted monoxide (C. P. Grade, Matheson) was
the basis for calculating the percentage passed through a molecular sieve prior to
reduction to the metal. use.
Selected samples were also analyzed for
metallic nickel by passing CO over pre- III. RESULTS AND DISCUSSION
viously reduced catalysts at 80°C and a
space velocity of 300 hr-l (5, 6). The A. E$ect of Calcination on Nickel Surface
Ni(C0)4 thus formed was collected during Area
24 hr by decomposition to a nickel mirror
The effect of calcination at 400°C on
in a preweighed glass tube at 350°C. Two
nickel surface area is demonstrated by
of the samples were exposed to 7 and 25
hydrogen chemisorption data in Table 1.
ppm H2S in Hz at 450°C for 3-4 hr prior
Six different catalysts were prepared by
to cooling and exposure to CO.
impregnation of alumina with nickel nitrate
The decomposition products formed by
after which three of these catalysts were
passing air or hydrogen through alumina-
calcined in air at 400°C for 1 hr and then
supported nickel nitrate samples were
reduced in flowing hydrogen at 500°C for
analyzed with a Hewlett-Packard 5830
12 hr. The other three catalysts were
chromatograph using a Porpopak Q column.
reduced in hydrogen without prior calcina-
In each of the chromatographic runs
tion. The resulting nickel surface areas and
hydrogen or air was passed at a flow of 500
dispersions (dispersion defined as the frac-
cm3/min through a 10-g sample placed in a
tion of total nickel atoms found at the
stainless steel reactor as the temperature
surface) are significantly larger for samples
was slowly increased from 25 to 300°C.
reduced without prior calcination compared
Sample weight versus temperature mea-
to samples calcined prior to reduction. In
surements were performed using a DuPont’
other words, the calcination at 400°C is
951 Thermal Gravimetric Analyzer with a
clearly detrimental to preparing a high
990 control. Selected nickel samples were
nickel surface area. Similar observations
analyzed chemically by x-ray fluorescence
have been reported for Ni/SiOz catalysts
spectroscopy.
(13) and carbon-supported iron and plat-
inum-iron alloy catalysts (14).
B. Materials The observation of low-er hydrogen
surface areas for calcined samples has at
Analytically pure Ni (N03) 2. 6Hz0
least three possible explanations which are
(Baker) was used in the TGA experiments
and in preparation of supported nickel
catalysts. Solutions of this salt were TABLE 1
impregnated to incipient wetness into SA Effect of Calcination on Nickel Surface Area
Med. Alumina (Kaiser), SAS Med. Alumina
wt% Calcined Reduced Surface Percentage
Pellets (Kaiser), and 100 S Pellets Ni/AlrOa at 4OOT, at 5oo”c, area dispemion~
1 hr 12hr
(Houdry). After impregnation the samples W/d

were dried in air at 80-100°C followed by 9 NO Yes 10.0 15.9

9 Yes Yes 4.3 6.9
either calcination at 400°C for 1 or 2 hr
15 NO YES 14.6 13.8
and subsequent reduction in flowing hydro- 15 Yes Yea 5.2 5.0
gen or by direct reduction in hydrogen. 27 NO Yes 18.6 10.7
Hydrogen gas (99.96%) was purifed by 26 Yes Yes 10.3 5.7
passing through a palladium Deoxo purifier 0 Dispersion is the fraction of total nickel atoms found at the
(Engelhard) and dehydrated molecular BUTf&CB.
44 BARTHOLOMEW AND FARRAUTO

not mutually exclusive : of iron metal (200-400 A in diameter). On

the other hand, Bicek and Kelley (4) have
(i) The extent of reduction to nickel
suggested that in reduction of calcined
metal is less for a precalcined sample
nickel-alumina, nickel crystallites are
compared to a sample prepared by direct
formed by growth from nickel ions or
decomposition of the metal salt in hydro-
atoms; however, t’his particular inference
gen ; in addition, any unreduced NiO at
was not apparently supported by experi-
the surface does not chemisorb hydrogen.
mental evidence. Unfortunately, X-ray
(ii) Calcination produces via a sintering
diffraction scans of nickel-alumina catalysts
mechanism relatively large particles of
in this study showed considerable overlap
nickel oxide which are reduced (by some
of the lines for Ni, NiO, and alumina
unknown mechanism) to metal crystallites
making it practically impossible to diff eren-
of approximately the same size and are
tiate various phases and to estimate
larger than nickel crystallites produced by
crystallite sizes.
direct decomposition of supported nickel
The third explanation involving a high
nitrate in hydrogen.
temperature solid state reaction of nickel
(iii) In high temperature calcination of
oxide with AltO to form a Ni A1204 spine1
nickel-alumina catalysts NiO interacts with
is well documented in the literature (4,
ALO to form a very stable NiA1204 spinel,
15-19).
which cannot be reduced even under very
To better understand the chemical
severe, high temperature reducing condi-
changes in the calcination and reduction
tions.
processes, the decomposition of unsup-
The first explanation is reasonable in ported and supported Ni(NOS)z.6Hz0 in
view of previous studies (2, 4, 5, 7, 8) which air followed by reduction in hydrogen was
show that for progressively higher tempera- studied by means of thermal gravimetric
tures of calcination prior to reduction, a analysis (TGA). The weight loss versus
significantly lower fraction of the total temperature curve from the TGA experi-
nickel is reduced to the metallic state. In ment for the unsupported nitrate is shown
this study the percentage reduction to in Fig. 2. Decomposition of the nitrate to
nickel metal for noncalcined samples varied NiO in air is complete at 400°C as indicated
from 45-lOO$& whereas corresponding by an observed weight loss of 75% com-
values for calcined samples reduced under pared with a calculated value of 74.3% for
equivalent conditions of time and tempera- conversion of Ni (NOs)z. 6HzO to NiO. If
ture in other studies (2, 5) ranged from hydrogen is then passed through the sample
3-60% depending upon the calcination at 4OO”C, the oxide is completely reduced to
temperature as well as reduction time and the metallic state as evidenced by an
temperature. Moreover, it was observed in observed weight loss of 22.1% which
this study that hydrogen chemisorption on compares favorably with a calculated value
NiO is indeed negligible compared to of 21.5y0 predicted for the decomposition
adsorption on nickel metal. of NiO to Ni metal,
The second explanation of larger nickel
NiO + Hz - Ni + HZ0
oxide particles finds a precedent in a 4009 c
previous X-ray study (14) of carbon- (1)
AH025*o = 0.4 kcal/mole.
supported iron in which very large oxide
particles (500-1000 A) were observed In the corresponding TGA experiment
following calcination, whereas the decom- with supported nickel nitrate it was not
position of the supported nitrate in hydro- possible to determine quantitatively the
gen produced significantly smaller particles extent of oxidation and reduction in the
 CHENIRTI~Y OF NICKELALUMINA CATALYSTS 45

I
I
IIm lllri Jim 4x I r>
Trl’I,x’rdt,iri~ , I , Ti,,PlhlllJ

FIG. 2. Thermal gravimetric analysis of decomposition of Ni(NOa)2.6H#J in air or helium, then

hydrogen.

presence of air and hydrogen, respectively, trate or nickel oxide affects significantly
mainly because after impregnation to the amount of nickel surface area obtained
incipient wetness and drying in air, the for a given catalyst as illustrated by data in
catalyst when heated continued to lose Table 2. Nickel surface areas and disper-
physically held water from the support and sions are shown for two different catalyst
chemically held water from the hydrated pairs, for which the only difference in
nit’rate. Therefore, attempts to establish an preparation for a given pair was the rate
initial stoichiometric weight and quantita- of temperature increase during reduction.
tively differentiate between partially or Both 9% Ni/A120s catalysts were calcined
fully oxidized or reduced states were not at 400°C and then reduced in hydrogen at
successful. Nevertheless, it was determined, heating rates of approximately 15 and
based on experiments discussed in a later 5”C/min, respectively. The nickel surface
section of this paper, that supported nickel area and dispersion are about 25% greater
samples (in this study) exposed to air at for the catalyst reduced with a slower rate
400°C were oxidized to NiO. of temperature increase. Neither of the
15% Ni/A1203 catalysts were calcined, but
B. E$ect of Heating Rate during Reduction rather prepared by decomposition of the
on, Nickel Surface Area
nickel salt in hydrogen at heating rates of
The rate of temperature rise in hydrogen 5 and 15”C/min, respectively. The surface
reduction of alumina-supported nickel ni- area and dispersion arc t,wice as large for
TABI,E 2
Effect of Heating Nate in Heduction of Ni/AI,O:,

wt% Calcined Rate of temperature Surface area Percentage

Ni/A1203 at 400% Rise in reduction W/d dispersion

9 Yes Fast, -15’C/min 3.2 5.1

9 Yes Slow, -5’C/min 4.3 6.9
13 No Fast, -15’C/min 6.5 6.1
(No hold)
15 No Slow, -5’C/min 14.6 13.8
(4 hr hold at 230°C)
46 BARTHOLOMEW AND FARRAUTO

the catalyst reduced with a low heating hydrogen are :

rate. There is clearly a larger effect for
2H,+Ni(N03)2 --, Ni+2Hz0+2N02 (2)
catalysts reduced without prior calcination.
The chemistry of nickel oxide reduction 4Hz+Ni(NOs)z -+ Ni+4Hz0+2N0 (3)
(see Eq. 1 above) is fairly straightforward,
9Hz+Ni(NOa)z 3 Ni+6Hz0+2NH3 (4)
although the precise mechanisms are not
known. Hydrogen interacts with nickel The heats of reaction at 25°C are 11.9,
oxide to produce water, which if not - 86.5, and - 261.2 kcal/mole, respectively.
removed from the catalyst, aids in transport Since Reactions 3 and 4 are very exo-
and growth of nickel crystallites. The low thermic, a fast heating rate could possibly
enthalpy for reaction of nickel oxide with result in a temperature runaway and
hydrogen suggests that overheating effects extensive sintering at a temperature where
are not expected; however, a slow heating the rates of the reactions are significant.
rate does allow water to be driven off and This exothermic effect was actually
transported from the catalyst by the flowing observed during TGA measurements in-
stream of hydrogen, thus minimizing the volving the decomposition of 82 mg of
effects of sintering. In fact, it is possible to Ni(NOa)z.6Hz0 in flowing hydrogen. The
control catalyst dispersion by controlling thermogram in Fig. 3A shows a smooth
either the amount of water removed by weight loss with increasing temperature up
drying prior to reduction (20) or by control- to about 250°C (probably loss of water) at
ling the heating rate and partial pressure which point there was a rapid temperature
of water during reduction (19). rise in a matter of seconds to almost SOO”C,
In the decomposition of nickel nitrate in followed by a rapid weight loss at almost
hydrogen atmosphere, there is also con- constant temperature, and then a very
siderable water formed plus an additional rapid cooling at constant weight. Figure 3B
heating effect which can influence the shows the same reaction under similar
ultimate dispersion. Three possible decom- conditions for alumina-supported nickel
position reactions for nickel nitrate in nitrate. The exotherm is much smaller in

R7.C

A 6

\
49.2 \___-- ___--

101) 4””

FIG 3. Thermal gravimetric analysis of decomposition of (A) unsupported and (B) supported
Ni(NOa)*-6HtO in Hz.
 CHEMISTRY OF NICKEL-ALUMINA CATALYSTS 47

magnitude since the amount of nitrate is were undertaken to determine by inference

smaller and the support functions as a the most important reactions. Measure-
heat sink. Nevertheless, the heat effect ment’s of pH were made during reduction
might be of disasterous proportions in a of a dozen different 15 w-t% Ni/A1203
large bed of catalyst of industrial capacity. samples by placing Hydrion paper in the
It is possible, however, to minimize the gaseous product stream from a small Pyrex
exothermic heat release and obtain a highly reactor containing l-2 g of catalyst. The
dispersed nickel catalyst by controlled measurements showed a pH of 1 as the
temperature decomposition in hydrogen sample was heated from about 100 up to
atmosphere. TGA analysis and chemisorp- 250°C at’ which temperature the pH changed
tion measurements indicate that a low rapidly to 7, then very rapidly to S-10.
heating rate coupled with an intermittent Acidic products were heaviest in the
hold at about 230°C for 1 hr eliminates any temperature range from about 200 to
detrimental heat effects and assures reason- 250°C. The basic products were predom-
able dispersion of nickel on alumina. inant from about 270 to 360°C. In the
Diluting hydrogen with an inert gas to range from 369 to 420°C the pH decreased
obtain better heat removal during reduction to 7.
may result in lower surface areas unless Chromatographic analysis of the decom-
trace contaminants are removed from the position products from a stainless steel
inert gas. For example, reduction with an reactor indicated NO and NO2 formation
8% Hz, 92% Nz mixture consistently observed initially at 218°C which increased
produced catalysts with lower surface areas in magnitude up to about 250°C. Upon
compared to reduction of the same samples reaching 246’C there was a temperature
with pure hydrogen. Nevertheless, reduc- excursion to 275°C (duration of about 2-3
tion of alumina-supported nickel nitrate min), followed by a decrease in temperature
with a 20% Hz, 80% Nz mixture purified over 10 min to 247°C. Then as the tempera-
by passing through a palladium catalyst ture increased gradually from 250 to 275”C,
and zeolite trap produced a 15 wt% the concentration of NO and NOz de-
Ni/AL03, with 19.1 m”/g nickel area creased ; there was no significant NOz
compared to 16.2 m2/g for an identical formation above 275°C. The NO concentra-
sample reduced in pure hydrogen. tion remained small until the temperature
Although it is clear that the decomposi- reached 330°C and then increased again.
tion of nickel nitrate in hydrogen is very During the chromatographic measurements
exothermic and that the decomposition rate a strong acid smell was detected in the range
is very rapid at a temperature approaching of 218 to 230°C and a brown gas (presum-
25O”C, it is not obvious which of Reactions ably NOz) was observed at 240-250°C. A
2-4 predominates at any given set of con- smokey white vapor was observed in the
ditions. The stoichiometries of Reaction reduction gas products just prior to and
2-4 are such that differing amounts of during the transition from acidic to basic
hydrogen are required for each reaction. products. No NH, was detected chromato-
Hence, the decomposition rate and mech- graphically ; however, it was likely absorbed
anism may also depend upon the hydrogen by water in our reactor condenser trap.
flow rate, especially at temperatures where The pH and chromatographic data to-
the reactions are very fast. gether suggest that the reactions to produce
Chromatographic analysis and pH mea- NO and NO2 are important up to about
surement of the decomposition products 250°C at which temperature the reaction to
for nickel nitrate heated in flowing hydro- produce ammonia becomes important. Be-
gen as a function of time and temperature cause this latter reaction is so exothermic
 48 BAI~THOLOMEW AND FARRAUTO

the reaction can run away if the heating previously calcined at 45O”C, there is little
rate is sufficient. doubt that the 14.0yo Ni/ALOa catalyst
prepared (in this study) without precalcina-
C. Effects of Reduction Temperature on tion and reduced for 145 hr contained only
Nickel Surface Area and Percentage completely reduced nickel. In fact, the
Reduction to Nickel Metal percentage reduction to metallic nickel
The determination of percentage reduc- increased from 93yo (10 hr reduction) to
tion of nickel to the metallic state using O2 98yo (145 hr reduction) assuming stoichio-
chemisorption at 400430°C requires that metric formation of NiO upon exposure to
the stoichiometry of the oxide thus formed 02 ; the Hz adsorption uptake increased
be precisely known. Hill and Selwood (10) proportionately from 203.5 to 211.1 pmol/g.
explained Hz consumption values in excess These data are consistent within experi-
of those required for reduction of NiO by mental error with a stoichiometry for nickel
assuming the existence of higher oxidation oxide of NiO and an oxidation state for
states for surface nickel (e.g., Niz03). They nickel of 2.0, assuming complete reduction
argued that for small nickel oxide crystal- to nickel.
lites, present in dilute catalysts, the surface The second set of experiments involved
represents a significant fraction of the total independent determination of the extent
nickel oxide and consequently one measures of reduction using the selective removal of
an average oxidation state higher than +2. metallic nickel with CO (80°C) to form
Indeed, in nickel-alumina samples contain- nickel carbonyl (5, 6). The nickel carbonyl
ing 3-4 wt% nickel, the average oxidation thus formed was decomposed in a pre-
state after calcination at 450°C was 2.6-2.7 weighed glass tube at 350°C. The results
(10). Nevertheless, oxidation states of 2.0 including calculated extents of reduction
were observed for catalysts having nickel compared with calculations from the O2
loadings in excess of about 5 wt% (10). titration are listed in Table 3. The data
Moreover, Holm and Clark (7) observed Ni show good agreement between percentage
valences of 2.55, 2.14, 2.04, and 1.91 for a reduction values determined by nickel
3oj, Ni/Al,O, calcined in air at 300, 400, carbonyl extraction and by 02 uptake;
500, and 7OO”C, respectively, showing that the effective nickel valence for the oxide
NiO is formed above 400°C. Since bulk formed at 400-430°C is 2.12-2.16, assuming
phases of Niz03 or NiOz arc not stable in the Ni(C0)4 extraction technique is 100%
the anhydrous state (%‘I) and since the efficient. Apparently, presulfidation had no
nickel catalyst loadings in this study were effect on the results within the time frame
considerably greater than 5%, it is reason- of the extractions, although sulfur has been
able to assumr an average stoiahiometry reported as a catalyst for Ni(CO), forma-
of NiO. tion (B?).
Nevertheless the nickel oxide stoichiom- Data prescntcd in Figs. 4 and 5 illustrate
etry and the validity of the method used the cffccts of reduction temperature on
for measuring extent of reduction to metal- measured nickel surface area and percent-
lic nickel were checked in two independent age reduction to nickel metal, respectively.
experiments. First, a 0.52-g sample of Each of the impregnated and dried (but
14.0yo Ni/Al,Os which had been reduced uncalcined) samples of a 13.5 wt% Ni/A1,03
10 hr at 485V (space velocity of 2000 hr-1) catalyst was reduced approximately 16 hr
was reduced an additional 135 hr at 450°C in flowing hydrogen at a space velocity of
and 2000 hr-1. Since Hill and Selwood (10) 1500-2000 hr-1, and the heating rate was
found a reduction time of 130 hr adequate held to less than 5’C/min, interrupted only
for complete reduction of a 6.57, Ni/A1,03 by a 1 hr hold at 230°C to prevent undesir-
 CHEMISTRY OF NICKEL-ALUMINA CATALYSTS 4(3

TABLE 3
Extent of Reduction L)etermined by Nickel Carbonyl Extraction

Run Weight of Weight of Ni Weight of Ni Percentage Percentage Effective

no. sample (g) expected (g) measured (g) redltction reduction Ni valence
for complet,e based on 02
removal t,it.rat,ion

l’b 2.0373 0.2832 0.2427 8.i.l 92..i 2.16

2b 6.8852 0.9639 0.83.58 86.7 92.3 2.12
:Jr X9.506 o..i53 1 0.3889 70.3 74.2 2.12

u Catalyst presulfided with 7 ppm HzS for 260 min; CO exposure, 16.5 hr at 80%
b Catalyst presulfided with 25 ppm HzS for 105 min; CO exposure, 24 hr at 8O’C.
c No presulfidation; CO exposure, 26.5 hr at SO’C.

able heating effects otherwise due to the percentage reduction to nickel metal (from
exothermic decomposition reactions above 43 to 100yO) over the same range of
250°C. temperature (see Fig. 5). Dzis’ko et al. (9)
The surface area data plotted as a func- observed that in supported nickel catalysts,
tion of reduction temperature (see Fig. 4) the mean crystallite size of nickel is
from 300 to 500°C form a very flat convex increased as the degree of reduction to tho
parabola with a maximum in the nickel metal is increased. Our data are consistent
surface area of 20 m2/g occurring at about with their observation; that is, the rela-
3cjO to 400°C and minimum values of 14.3 tively small variation in surface area with
and 15.3 m2/g occurring at thr temperature
extremes, 300 and 5OO”C, respectively.
This modest variation in surface area is in
contrast to the relative drastic increase in

FIG. 4. Effect of reduction temperature on nickel FIG. 5. Effect of reduction temperature on

surface area, 13.5 wt% Ni/AlnOa (Hz GHSV percentage reduction to nickel metal, 13.5 wt’%
= 1500-2000 hr-I) . Ni/A1203 (Hz GHSV = 1500-2000 hr-I)
50 BARTHOLOMEW AND FARRAUTO

study is that for a given reduction tempera-

ture and space velocity approximately a
minimum of S-10 hr is required to reach
reasonably constant values of surface area
and fractional reduction to nickel metal
starting with a freshly impregnated and
dried sample. If the reduction is taken
beyond this minimum time requirement,
the surface area and extent of reduction
are expected to increase very little, assum-
ing temperature and space velocity are
fixed. This S-10 hr minimum reduction
time for uncalcined impregnated samples
is iapparently less than the 15-20 hrs
required for Ni/A1203 catalysts which
were previously calcined (2).

D. Effects of Hydrogen Space Velocity on

FIG. 6. Effect of hydrogen space velocity on Nickel Surface Area and Percentage
surface area 14.0 wt% Ni/AlzOz (reducedat 485’C,
lo-12 hr) . Reduction to Nickel Metal
Nickel surface area and extent of reduc-
increasing temperature is 8 result of two tion measurements are shown in Figs. 6
opposing effects : and 7 as a function of hydrogen space
(i) The fraction of nickel reduced to the
metal increases with increasing tempera-
, 11”

r A
ture. This metallic nickel thus formed
chemisorbs hydrogen whereas unreduced
nickel does not.
(ii) At the same time, however, the
average crystallite size increases with
temperature which means the fraction of
surface nickel metal available for hydrogen
chemisorption decreases.
Given a choice of reducing conditions for
a nickel-alumina catalyst, it appears advan-
tageous to reduce at 500°C so as to max- 50c
imize the extent of reduction to nickel
metal while at the same time assuring an t
equilibrium particle size which is stable
against sintering, especially at use tempera-
tures lower than 450°C. Accordingly, these
advantages are achieved without sacrificing
z”I------
appreciable catalytic surface area.
What is the time requirement to reach
FIG. 7. Effect of hydrogen space velocity on
a final state of reduction if no calcination is percentage reduction to nickel metal, 14.0 wt%
used in the preparation? The experience Ni/AleOs. l , reduced at 485’C, 10-12 hr; A,
from preparation of many catalysts in this reduced an additional 135hr at 450°C.
 CHEMISTRY OF NICKEL-ALUMINA CATALYSTS 51

velocity, The data correspond to various TABLE 4

samples of the same 14.0 wt% Ni/Al& Effect of Ni Loading on Ni Surface Area
catalyst, each reduced in a large ceramic
combustion tube approximately 10 hr at wt% Precalcined Surface Percentage
Ni/AltOs at 4OO’C area dispersion
490°C. Nickel surface areas were measured
W/d
for a portion of each sample after passiva-
tion, transfer to a glass cell, and rereduction 9 Yes 4.3 6.9
at 450°C for 6 hr. Nickel surface area I5 Yes 5.2 ri.0
increases significantly from 8 to 17 m2/g as 26 Yes 10.3 5.7
34 Yes IO.,> 4.5
the space veIocity is increased from 100 to
2000 hr-’ (see Fig. 6). This increase is 9 No 10.0 15.9
likely a result of two effects: (i) More 15 No 14.6 13.8
effective transport of wat#er (which aids in 27 No 18.6 10.7
the sintering process) from the sample at 36 No 17.2 6.8
higher space velocities and (ii) more
effective removal of heat generated by exo-
t’hermic decomposition reactions at higher and dispersions are larger for the latter
flow rates, thus minimizing temperature group. The results for either group of
catalysts are consistent with the generally
excursions and hot spots in the catalyst.
accepted observation that metal dispersion
The increased extent of reduction of
decreases with increased metal loading. The
nickel to the metal (from 74 to 93%) with
nickel surface area, on the other hand,
increasing space velocity (from 100 to
increases with increased nickel loading up
2000 hr-‘), as shown in Fig. 7, very likely
to a value of 25 wt% nickel and then
results from (i) increased transport of water
increases only very slightly or remains
and decomposition products, which are
constant. In other words, the nickel
oxidizing agents, from the nickel surface,
utilization is very poor for the incre-
and (ii) a related increased efficiency of
mental addition above 25 wt%.
hydrogen transport to the surface due to
increased turbulence and mixing. In other
F. Passivation of Reduced Nickel
words, there is an increased effective partial
pressure of hydrogen at the nickel surface. Finely dispersed nickel metal spontan-
An apparent saturation effect is observed eously oxidizes to nickel oxide (NiO and
in Figs. 6 and 7 in the higher space velocity possibly Ni20s) at room temperature with
range (1500-2000 hr-I). In other words, a large negative enthalpy of the reaction
there is not considerable improvement to (-58.4 kcal/mole for NiO). Therefore, one
be expected in either nickel surface area or should expect large temperature excursions
the extent of reduction by increasing the when a sample of reduced catalyst is
hydrogen space velocity during reduction exposed to air. These heat effects if not
beyond 2000 hr-l. controlled cause local and/or bulk sintering
of the catalyst. The process of preventing
E’. Bflect of N&&e1 Loading in Nickel the bulk oxidation of a reduced catalyst by
Surface Area oxidizing only the metal crystallite surface
Nickel surface areas and dispersions are to achieve a “passive” monolayer coating
shown in Table 4 for a variety of alumina- is referred to here as passivation.
supported samples, some of which were Possible effects of the passivation treat-
calcined at 400°C prior to reduction and ment on the surface area of a nickel catalyst
others prepared by hydrogen decomposition were determined in the following way:
of nickel nit)rate. Again the surface areas (i) a sample was prepared by hydrogen
 52 BARTHOLOMEW AND FARRAUTO

reduction of alumina-supported nickel ni- temperatures (above 600°C) in order to

trate at 480°C (the metal area was 15.32 form NiA1204 which itself is a very stable
m”/g), (ii) the sample was exposed to 10% noninteracting support for nickel (18, %Y).
air in helium at room temperature, and a (iii) A slow heating rate during reduc-
small heat effect was noted initially, (iii) tion of alumina-supported nickel nitrate
subsequent reduction under iden tical con- prevents exothermic temperature excur-
ditions followed by a surface area de- sions which may badly sinter the catalyst.
termination at room temperature gave The recommended preparative scheme to
14.94 mz/g. Other catalysts in this study obtain well-dispersed Ni/A1203 is to heat
exposed initially to 1% air in nitrogen, the supported nitrate slowly (less than 5”C/
showed no change in surface area within min) to 230°C in Hz (space velocity of
experimental error. Hence, passivation need 1500-2000 hr-1) holding at 230°C for 1 hr.
not lead to any alteration in exposed area This is followed by heating slowly to 480°C
due to the Ni/NiO/Ni cycle, provided the and holding there for at least 10 hr.
bed temperatures are controlled to avoid (iv) Oxygen titration at 450°C can be
sintering. This is particularly important used to measure the extent of reduction of
for large catalyst beds where considerably nickel to the metallic state in nickel
larger temperature excursions could occur catalysts having metal loadings greater
by rapid exposure to air. than about 5 wt%. This technique gives
results comparable to the Ni (CO)4 extrac-
CONCLUSIONS tion technique and is faster, safer, and
otherwise more convenient than the latter
The following conclusions are based on
method.
the results and discussion above:
(v) Nickel dispersion decreases with
(i) Decomposition of alumina-supported increased nickel loading. Surface area in-
nickel nitrate in pure flowing Hz (at 300- creases with increased nickel loading up to
500°C) results in about SO-100% reduction 25 wt% nickel and then increases only
to the metal and reasonably large nickel very slightly at higher loadings.
areas and dispersions. Nickel metal surface (vi) Nickel catalysts can be passivated
area and percentage reduction to nickel with no significant loss of surface area
metal are maximized by reducing at hydro- provided over-heating is prevented. Con-
gen space velocities above 1500-2000 hr-’ trolled exposure of reduced nickel-alumina
and at temperatures in the range of 450- catalysts to 1% air in nitrogen at room
500°C. These results are valid for nickel temperature prevents sintering and bulk
nitrate impregnated on a gamma-alumina oxidation upon further exposure to the
support which is a common starting mater- atmosphere.
ial for preparation of nickel catalysts. It
ACKNOWLEDGMENTS
would also be desirable t,o study other
The authors gratefully acknowledge technical
nickel salts and different supports as start-
assistance by W. L. Haynes, R. L. Johnson, R. B.
ing materials, although some of these results Pannell, E. R. Larsen, W. B. Barton, R. W. Fowler,
undoubtedly apply to other systems. and A. J. Perrotta; very helpful comments on the
(ii) Calcination in air prior to the reduc- manuscript by Professor P. W. Selwood; and
tion step is detrimental to obtaining high support in the form of grants to C. H. B. by Corning
Glass Works and the National Science Foundation
reducibility, metal surface area, and metal
(ENG 75-60254).
dispersion. This does not mean that calcina-
tion of nickel catalysts is always undesir- REFERENCES
able. It may be desirable, for example, to 1. Reinen, D., and Selwood, P. W., J. Cat&. 2,
heat a sample of Ni/A120a in air at high 109 (1963).
 CHEMISTRY OF NICKEL-ALUMINA CATALYSTS 53

2. Sieg, It. P., Constabaris, G., and Linquist, published in short form in Chem. Engr. Prog.
R. H., Amer. Chem. Sot. Meeting, New York, 71, 37 (1975).
September 1963. 13. Yates, D. J. C., Taylor, W. F., and Sinfelt, J. H.,
:3. Levinson, G. S., “Preprints,” p. 47. Division of J. Amer. Chem. Sot. 86, 2996 (1964).
Petroleum Chemistry, Amer. Chem. Rot., 14. Bartholomew, C. H., and Boudart, M., J. Catal.
Chicago, September (1967). 25, 173 (1972).
4. Bicek, E. J., and Kelly, C. J., “Preprints,” 2,5. Pattison, N. J., Keeley, W. M., and Maynor,
p. 57. Division of Petroleum Chemistry, Amer. H. W., J. Chem. Eng. Data 5, 433434 (1960)
Chem. Sot., Chicago, September (1967). 16. Branson, 1). I,., .J. Amer. Cer. Sot. 48, 591
6. Beuther, H., and Larson, 0. A., Ind. Eng. Chcm. (1965)
Proc. Design Develop. 4, 177 (1965). 17. Swift, H. Ii:.:., Lutinski, F. E., and Kehl, W. I,.,
G. Swift, H. E., Lutinski, F. E., and Tobin, H. H., J. Phys. Chem. 69, 3268, (196.5).
J. Catal. 5, 28.5 (1966). 18. Leech, C. A., III, and Campbell, L. El., AC8
Division INDE, 167th National Meeting of
7. Holm, V. C. F., and Clark, A., J. Catal. 11,
the American Chemical Rociet,y, Los Angeles,
305 (1968).
California, 1974.
8. Eischens, R. P., “Proceedings of the Battelle
19. Andrew, S. P. S., “Preparation of Cat,alysts,”
Conference on Surface Chemistry,” Gstaad,
(B. Delmon, P. A. Jacobs, and G. Poncelet,
Switzerland, Sept. 1974.
Eds.), pp. 429444. Elsevier, Amsterdam,
.9. Dzis’ko, V. A., Noskova, S. P., Borisova, M. S., 1976.
Bolgova, V. D., and Karakchiev, L. G., 20. Anderson, J. H., Jr., Conn, P. J., and Branden-
K&et. Catal. 15, 667 (1975). berger, S. G., J. C&Z. 16, 326 (1970).
10. Hill, F. N., and Selwood, P. W., J. Amer. Chem. 21. Remey, H., “Treatise on Inorganic Chemistry,”
Sot. 7 1, 2522 (1949). Vol. II, pp. 310. Elsevier, Amsterdam, 1956.
11. Martin, G. A., Ceaphalan, N., and Montgolfier, 22. Heinicke, G., Chem. Z’ech. 15, 197 (1963).
P. de, J. Chim. Phys. 70, 1422 (1973). 23. Bartholomew, C. H., Ind. Eng. Chem. Prod.
12. Farrauto, R. J., AZChE Symp. Ser. 70, 9 (1974) ; Res. Develop. 14, 29 (1975).