editor's note

92
SCOPE OF REVIEW In 1983 we reviewed the applications of temperature-programmed desorption and reaction (TPD, TPRx) to supported metal catalysts Ill. With these transient techniques the products leaving a surface are measured as a function of temperature for a known heating schedule. As we noted at that time, the full potential of the TPD/TPRx apparatus had not been utilized. Recent studies, however, have expanded the capabilities of the technique by syste- matically varying experimental parameters. For example, interrupted reaction sequences have been used, and the adsorption temperature, carrier gas concen- tration, and system pressure have been varied. Isotope labeling has also been used to provide additional information about surface processes. With some modifications, the TPD/TPRx apparatus can also be used for other transient techniques. These techniques include temperature-programmed re- duction, temperature-programmed oxidation, transient isotopic tracing, and transient oscillations in concentration and temperature. Our interest in this review is the use of transient techniques to understand the properties of supported Ni catalysts. A useful probe reaction, which has been used in a number of studies of Ni catalysts, is CO hydrogenation and this will be our focus. A comprehensive presentation of the application of these techniques to Ni catalysts provides a basis for discussing both supported metal catalysts in general and transient techniques. We will extend our earlier discussions of experimental procedures that can be conducted in a TPD/TPi?x apparatus and describe other transient procedures that have been used to study supported Ni catalysts. The effects of supports and promoters will also be discussed. Two important features of Ni catalysts can be studied by transient techniques. The first is catalyst preparation and its effects due to the variety of procedures that comprise forming a supported metal catalyst. The second is reaction pathways as revealed in mechanistic studies based on data from transient experiments. These areas of preparation and reaction pathways are intrinsically related. We begin this review with a general discussion of transient techniques. We then provide a description of the experimental procedures for carrying out transient studies. Experiments are included that can be conducted in a TPD/TPRx apparatus as well as those that require modifications of the apparatus. Particular consideration is placed on describing the flexibility obtained by changing the experimental conditions in a systematic manner to reveal details of surface reaction processes.

Upload: julian-ross

Post on 26-Jun-2016

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Editor's note

SCOPE OF REVIEW

In 1983 we reviewed the applications of temperature-programmed desorption

and reaction (TPD, TPRx) to supported metal catalysts Ill. With these

transient techniques the products leaving a surface are measured as a function

of temperature for a known heating schedule. As we noted at that time, the

full potential of the TPD/TPRx apparatus had not been utilized. Recent

studies, however, have expanded the capabilities of the technique by syste-

matically varying experimental parameters. For example, interrupted reaction

sequences have been used, and the adsorption temperature, carrier gas concen-

tration, and system pressure have been varied. Isotope labeling has also been

used to provide additional information about surface processes.

With some modifications, the TPD/TPRx apparatus can also be used for other

transient techniques. These techniques include temperature-programmed re-

duction, temperature-programmed oxidation, transient isotopic tracing, and

transient oscillations in concentration and temperature. Our interest in this

review is the use of transient techniques to understand the properties of

supported Ni catalysts.

A useful probe reaction, which has been used in a number of studies of Ni

catalysts, is CO hydrogenation and this will be our focus. A comprehensive

presentation of the application of these techniques to Ni catalysts provides a

basis for discussing both supported metal catalysts in general and transient

techniques. We will extend our earlier discussions of experimental procedures

that can be conducted in a TPD/TPi?x apparatus and describe other transient

procedures that have been used to study supported Ni catalysts. The effects of

supports and promoters will also be discussed.

Two important features of Ni catalysts can be studied by transient

techniques. The first is catalyst preparation and its effects due to the

variety of procedures that comprise forming a supported metal catalyst. The

second is reaction pathways as revealed in mechanistic studies based on data

from transient experiments. These areas of preparation and reaction pathways

are intrinsically related.

We begin this review with a general discussion of transient techniques.

We then provide a description of the experimental procedures for carrying out

transient studies. Experiments are included that can be conducted in a

TPD/TPRx apparatus as well as those that require modifications of the

apparatus. Particular consideration is placed on describing the flexibility

obtained by changing the experimental conditions in a systematic manner to

reveal details of surface reaction processes.

Page 2: Editor's note

In all transient studies, transport effects must be minimized. Mass and

heat transfer limitations influence the rate of formation of products during

steady-state reaction and will likewise affect transient responses superimposed

on steady-state conditions. Mass transfer affects the rate of desorption of

reactants. Both heat and mass transfer have been analyzed, and we will briefly

revlew these in the next section.

Our focus In this review Is the properties of supported Ni catalysts as

revealed by transient techniques. We use preparation and mechanistic studies

as an outline for our discussion. Emphasis is placed on recent studies from

our laboratories. We conclude this review with a summary of the types of

Information provided by transient techniques and the role played by supported

Ni catalysts In the CO hydrogenation reaction.

INTRODUCTION TO TRANSIENT TECHNIQUES

Chemical relaxation techniques have been employed to study the rates of

elementary reaction steps. The use of relaxation techniques calls for an

experimental variable to be changed, and two of the most useful variables to

control are the concentration of reactants and the temperature of the reacting

system. The dynamic response of the system to changes In these variables Is

related to the rate determining step in the reaction CZI.

Chemical relaxation techniques might be conveniently divided Into two

general types. In the first, the reaction system, operating at steady state is

changed, e.g., by a change in the composition of the reactants or a programmed

temperature change. The system relaxes to a new steady state and analysis of

this phenomenon furnishes Information about Intermediate species. We refer to

this relaxation technique as single cycle transient analysis (SCTA). The basic

ideas of SCTA applied to heterogeneous catalysis were set forth by Tamaru C31

and later reviewed by Bennett 141. We refer to the second general classlfi-

cation for relaxation techniques as multiple cycle transient analysis (MCIA).

Here, the system Is periodically switched between two steady states, e.g., by

periodically changing the reactant concentration. This later technique was

originally applied to heterogeneously-catalyzed chemical reactions by use of

molecular beam relaxation spectrometry (BARS) by Schwarz and Radix CSI and more

recently to supported catalytic systems by Dautzenberg et al. C61 and Hegedus

et al. CTI. HCTA offers the advantage of being able to systematically vary

the frequency at which the reacting system Is forced to relax. Reaction path-

ways leading to products with characteristic times that are long compared to

this modulation frequency are effectively shut off, and these reaction products

are not detected because their signals are severely demodulated.

Page 3: Editor's note

3

RKPRRIHENI'ALDESCRIPTION

The same reactor and detection system can be used for all the transient

techniques to be described, but the gas feed system and the procedures must be

modified for each technique. Thus, the reactor and detection system will be

described first, and the details of each experimental setup and procedure will

then be presented. Note that the system can also be used as a differential

reactor for steady-state kinetic measurements with essentially no

modifications.

Reactor and Detector System

A small catalyst sample (typically 100 mg or less, 80-100 mesh) is located

in a quarts downflow reactor, which contains a sintered quartz disk (00 poro-

sity) to support the catalyst. A typical reactor diameter Is 1 cm, and down-

stream of the catalyst the reactor diameter can be smaller in order to minimize

the transit time between the reactor and the detector.

The reactor is heated by an electric furnace, whose power is controlled by a

derivative-proportional temperature programmer. A homemade quartz furnace with

Kanthal Al wire (liWS Wire Industries) has been found effective for obtaining

temperatures up to 1400 K in the catalyst bed. The exterior of the Kantal wire

oxidizes and passivates the wire against further oxidation. A small shielded

thermocouple (0.25 mm OD). located in the catalyst bed, measures the catalyst

temperature and provides feedback to the programmer so the catalyst can be

maintained at constant temperature or a temperature ramp can be applied to the

system. A derivative-proportional controller has been found necessary to

obtain a linear ramp over the entire temperature range needed for TPD/TPRx

experiments.

The detector is located immediately downstream from the reactor so that the

delay time between reaction processes and their detection Is mlnlmized. For

most of the experiments to be described, a mass spectrometer detector is

necessary. The mass spectrometer is located in a small ultrahigh vacuum

chamber, which is pumped by a turbomolecular pump so that gases are removed

rapidly after analysis, and thus an accurate measurement of the Instantaneous

concentration in the effluent stream is obtained. By using a small ultrahigh

vacuum chamber (just large enough for a UII quadrupole mass spectrometer) and a

110 L/s turbomolecular pump, we have been able to obtain a rapid response in

the system. Since large quantities of low molecular weight gases, He and 4,

are used in many of the transient experiments to be described. an ion pump

would not be as effective.

The effluent gas stream can be sampled by a Nupro sampling valve that allows

only a small fraction of the effluent stream to be introduced into the ultra-

high vacuum chamber. An alternate approach is to bleed gas through a small

hole drilled into a copper conflat gasket. By using two gaskets in series,

Page 4: Editor's note

4

with a mechanical pump In between, the pressure can be decreased from atmos-

pheric to 10-e TOW. Throttling the mechanical pump then allows control of

the pressure in the ultrahigh vacuum chamber. For either type of sampling,

accurate calibrations are needed since gases with different molecular weights

diffuse at different rates.

A computer must be interfaced to the mass spectrometer so that multiple mass

peaks can be detected simultaneously. The ability to detect several mass peaks

at the same time Is needed for a number of reasons:

I) For most desorptlon and reaction processes, more than one product Is

formed.

2) Because the ionizer of a mass spectrometer cracks molecules into

smaller fragments, some of which have the same mass as other products of lnter-

rest, corrections must be made to accurately determine how much of each product

is formed.

3) Isotope labeling, which provides a wealth of information about surface

processes, Is only effective if each of the Isotopes can be detected.

4) Impurities can have a large effect during experlments where high gas

flow rates are used over a small amount of catalyst Ill. and the gas stream purity needs to be monitored. One approach to determine the effect of impuri-

ties Is to flow the gas over a catalyst for an hour and then carry out a TPD to

determine what gases desorb.

The temperature Is also recorded by the computer. A typical procedure

cycles through the temperature measurement and the amplitude of the mass peaks

of interest in one second. Thls sampling rate has been found effective for

obtaining data with high signal to noise during TPD/TPkx.

The computer also greatly simplifies data analysis. Cracking corrections

are easily made to the original mass signals, which are then corrected using

calibrations, so that the resulting signal amplitudes are proportional to the

gaseous concentrations. Spreadsheet programs such as Lotus l-2-3 are Ideally

suited for such corrections and for subsequent plottlng of data. Data from the

mass spectrometer programs are imported into Lotus so that cracking correct-

ions, calibrations, and comparisons can be easily carried out. Areas under

the TPRx curves, which are proportional to the amount of gas formed, are also

easily calculated_ Data in the Lotus files can then be readily plotted for

further analysis. High quality plots, such as many of those in this paper,

can then be directly obtained from the Lotus files.

Downstream from the mass spectrometer sampling valve, a back pressure

regulator is used to maintain a constant flow rate in the reactor as the

temperature Is changed In TPD experiments_ The back pressure regulator Is also

necessary to equalize pressures In transient tracing experiments so that the

feed can be switched from one gas stream to another without disturbing the

Page 5: Editor's note

5

steady-state reaction conditions. lloreover, TPD/TPRx experiments at higher

pressures have advantages In some cases, and a back pressure regulator is

necessary to raise the reactor pressure above atmospheric. A schematic of a

TPD/TPRx apparatus is shown in Figure 1.

The reactor and the detection system just described are essentially the same

for each of the following techniques. The conditions under which the system is

used differ, but most techniques require the ability to detect more than one

mass peak, to obtain an accurate measure of instantaneous concentration In the

reactor effluent, and to manipulate the data for subsequent analysis.

Single-Cvcle Transient Analysis

Temperature-Programmed Reaction (TPRx). A detailed description of this type

of system was presented previously Cl1 and thus only a brief overview, which

describes improvements, will be presented. In addition, a description of the

parameters that can be varied during TPD/TPRx experiments will be presented.

The flow system for TPRx allows several gases to flow through the reactor,

either individually or at the same time. Mass flow controllers are used to

maintain the constant flow rates needed to detect small changes in mass

signals. In a typical TPRx experiment for methanatlon, the catalyst is reduced

in H,, and CO is then adsorbed by injecting pulses of CO, or a Co/He mixture,

into the carrier gas that flows through the catalyst bed. Though a H, carrier

or a H&He mixture Is used during the TPRx experiments, CO is often adsorbed in

He instead of H, flow, and the flow then changed to H, after CO adsorption Is

complete.

For some of the slower adsorption and transfer processes we have studied

C81. an air-actuated, computer-controlled pulse valve is useful. The

computer can control the frequency of injection, count the number of pulses,

and also observe the amount of gas in the reactor effluent and thus determine

how much did not adsorb. As Indicated in Figure I, more than one pulse valve

is ncccssary so that Isotopic gases can be adsorbed in succession. This allows

one valve to be devoted to an expensive, isotopically-labeled gas. Gases can

also be adsorbed by continuous flov of the reactant gas over the catalyst, and

this procedure is necessary In some cases to obtain high coverages. However, a

long time is then required to flush the gas out of the system and return the

background signal in the mass spectrometer to an acceptable level for sensitive

detection of the desorbing gases.

For most TPD/TPRx experiments, gases are a&orbed to saturation at a given

temperature. Lower initial coverages are obtained by saturating the surface

and then desorblng or reacting away some of the adsorbed gas. Thls procedure is

more likely to yield a uniform lover coverage throughout the bed than can be

obtained by lower-than-saturation exposures. For TPRx mathanatlon, CO is

Page 6: Editor's note
Page 7: Editor's note

7

adsorbed and sufficient time is allowed for the excess CO to flush out of the

flow system and for the mass spectrometer chamber to return to a low background

partial pressure of CO. The catalyst temperature is then raised linearly in

time in H, flow, and changes in concentrations in the effluent stream from the

reactor are continuously measured by the mass spectrometer and recorded by the

computer. The catalyst temperature is also recorded by the computer. As the

temperature is raised, the rate of CH, formation initially increases exponen-

tially. As reaction depletes the surface of CO, the rate of CH, formation

goes through a maximum. The rate decreases to zero if the temperature is

raised enough to remove all the adsorbed CO.

A high flow rate of H, carrier gas through the reactor bed is used during

TPRx to rapidly flush the system. Thus, the CH, concentration measured by the

mass spectrometer is a good measure of the rate of reaction. For methanation,

the CHr product does not readily readsorb and thus is rapidly removed from the

bed. The mass signal at m/e=15 is used to detect CH, because of CO and He0

cracking products at m/e=l6. The carrier gas flow rate depends on the rate of

heating and the amount of catalyst in the bed. For a low flow rate, the

transit time between the catalyst and the mass spectrometer detector may be

long and prevent determination of an accurate relation between the catalyst

temperature and the rate. For a high flow rate, the transit time is reduced

but the concentrations of desorbed species are also reduced and thus their

detection is more difficult. Moreover, the cumulative exposure to impurities

increases with the flow rate. Thus, leaks and impurities in the ppm range can

yield a significant impurity concentration on the surface If these impurities

preferentially adsorb. For reproducible results, the catalyst may need to be

reduced between experiments.

A number of variations in the TPRx experlment for methanation can increase

the information obtained about catalytic surface processes:

I) The adsorption temperature can be varied. We have used temperatures

from 260 to 385 K for CO adsorption. The low temperature was obtained by cool-

ing the reactor with vapors from liquid nitrogen. Temperatures above 385 K

have also been used, but the CO dissociation rate increases significantly and

carbon can deposit on Ni surfaces.

2) Carrier gases other than H, can be used during CO adsorption. Signifi-

cant differences are observed between CO adsorption in He and in H.!., even

though the heatlng was in H, in each case C91. Moreover, how the H, is

exposed to the catalyst before CO adsorption can also make a difference.

Cooling the catalyst in H, prior to CO adsorption is different from cooling In

He and then adsorbing CO in H, because of activated adsorption of H, [El.

3) A range of H, partial pressures can be used during heatrng CIOI.

Page 8: Editor's note

8

Pressure3 below atmospheric can be obtained by dilution with He, though the

system can also be run at a partial vacuum. Pressures above atmospheric are

mainly limited by the capabilities of the reactor and the feed system. For

quartz reactors, we have used pressures up to 2 atm. For low H, partial

pressures, the change in the H, concentration can be monitored during TPRx.

Though the H, signal could also be monitored at high H, partial pressures, the

fractional change in the H, signal is small for 100 mg of catalyst and thus

difficult to detect; H, usually has the poorest signal to noise of any of

the masses of interest. At low H, partial pressures, a TPRx experiment is

similar to TPRd except only the H, uptake is monitored during TPRd.

4) In order to vary the initial coverage, reaction can be stopped after

some CH, has formed. The catalyst is heated in H, to a given temperature,

cooled to room temperature, and then heated again in a standard fashion to

obtain an interrupted TPRx. A3 described below, this procedure can be used for

isotope labeling of sites.

5) Varying the heating rate over a large range allow3 activation energies

to be measured without assuming preexponential factor3 Cll. A3 will be dis-

cussed, when surface transfer processes compete with CH, formation, a variation

in heating rate can allow one process to dominate and thus be studied Gill.

6) The carrier gas can be D, instead of H, in order to study kinetic iso-

tope effects and to observe exchange processes that occur. For example, the

catalyst can be cooled in D,, CO adsorbed in D,, and the TPRx carried out in

H3 c121.

7) Part of the surface can be covered with one isotope of carbon monoxide

and another part of the surface with a different isotope. This type of experi-

ment yields additional information about surface processes i8,11,13-151. Two

approaches have been used for isotope labeling in TPRx methanation. For one

approach the surface is saturated with 13C0, and interrupted reaction or de-

sorption is used to remove IWO from some surface sites. Subsequent adsorption

of *3CO or Cl30 then occupies those vacated sites Clll. An alternate approach

is to adsorb '3CO to saturation at elevated temperature and then expose the

catalyst to 13CO at room temperature 30 that exchangeable IWO is replaced by

13CO C81.

Tetnnerature-Programmed Surface Reaction (TPSR). In our earlier review we

attempted to adopt terminology for each TPD technique that would clearly

describe the processes occuring during the transient. The designation

temperature-programmed surface reaction (TPSR) refers to a form of TPRx in

which a surface layer, such as carbon, is reacted. The emphasis here was that

the adsorbate was irreversibly bound to the catalyst surface (metal and/or

support) and could not be removed by TPD. In most of the earlier studies, TPSR

Page 9: Editor's note

9

experiments were directed towards examining the relative rates of formation of

CH, derived from carbon deposited by exposure of CO to the catalyst at elevated

temperature. Recent studies have shown that results from TPSR experiments

conducted subsequent to the steady-state reaction are sensitive to method of

preparation of the catalyst C161. In addition, the inventory of carbonaceous

residues, as revealed by TPSR, can be used to delineate between rate determin-

ing steps in a proposed mechanism for the reaction Cl'71. We feel that TPSR,

although similar to TPRx, provides complimentary information and thus should be

differentiated. A standardization of notation in literature reports can serve

to help in clarification, especially when a number of TPD/TPRx techniques are

used in a single study.

Temperature-Programmed Desorption (TPD). The experimental setup for TPD is

identical to that for TPRx. The basic difference between the two techniques

is that the catalyst is heated in an inert gas during TPD. Usually He is used,

but Ar is useful when D, must be detected, because He and Da are both seen at

m/e=4. However, detection of Da0 is difficult in Ar flow because of the

doubly ionized signal at m/e=ZO, and thus experiments are necessary in each

carrier gas for reaction processes where both D, and De0 are involved Cl2,181.

Several types of TPD experiments have been carried out:

1) Desorption of a gas that does not decompose. For example, adsorption

and desorption of H, from supported Ni catalysts has been used to measure sur-

face areas and measure changes in H, binding energy with catalyst properties

c19-211. This has been the most common use of TPD.

2) Reactive desorption of an adsorbed gas. For example, when CO is

adsorbed on Ni catalysts, both CO and CO, products form during heating in He

c221.

3) Desorption of coadsorbcd gases. Intermediates can form from coadsorbed

CO and Ha on Ni catalysts and desorption of CO and H, during TPD can provide

information on the nature of the intermediates and their interaction with the

catalyst surface C12,131. Stoichiomctrics of the intermediates can be

inferred from the quantities of desorbing gases, as determined from the areas

under the TPD curves.

Some of the same variations used for TPRx experiments can also be used for

TPD. The adsorption temperature is an important variable because it can affect

the form and amount of adsorbed species. Interrupted desorption or reaction

can also be used prior to TPD. For example, when two types of species with

different reactivlties in TPRx are located on a catalyst, removal of one

species by TPRx prior to removal of the other species by TPD allows adsorption

and reaction sites to be related to each other. Removing one species by TPRx

and replacing it by an Isotopically labeled species prior to TPD also allows

Page 10: Editor's note

10

adsorption and reaction sites to be related and has been used successfully for

Ni/Al,O, catalysts f12.131.

Temperature-Programmed Reduction (TPRd). Temperature-programmed reduction

(TPRd) is used to characterize the reducible components of a catalyst. This

technique, which has been reviewed in detail by Hurst et al. C231, has been

applied to studles of catalysts or catalytic precursors at different stages of

preparation. During TPRd, a hydrogen-containing gas mixture (typically less

than 107. H, in an inert carrier) continuously perfuses the catalyst bed while

the temperature of the bed is raised linearly with time. A TPRd spectrum is

obtained by measuring the consumption of Ha as a function of temperature.

During the TPRd process, the oxidation state or states of the catalyst/support

are lowered and Ha is consumed In the process. The reduction process ceases

after all the components that are reducible in the temperature range used have

been reduced.

Analysis techniques to determine kinetic parameters have been proposed by

Gentry et al. C241 and Mont1 and Baiker f251. Gentry et al. C2.41 studied

the effects caused by variation of the maximum Ha consumption rate. They pro-

posed a method to estimate the activation energy using profiles obtained with

different heating rates. Monti and Baiker CZSI proposed that a peak shape

analysis leads to more accurate determination of the kinetic parameters than

an estimation based on the shift with heating rate of the tcmpcrature of the

maximum reduction rate.

The experimental geometry and the mass-spectrometer detection system

described earlier, though sufficient to carry out such experiments, is not.

optimal. The main reason for this limitation is the inherent instability of

quadrupoles for the detection of Ha and the abundance of H, as a background

gas in most vacuum systems. Collectively, these contribute to poor signal to

noise and ultimate loss,in sensitivity. The water product, which results from

the reduction reaction, can be monitored for pre-calcined catalysts C261.

Water, however, is strongly adsorbed on most supports and on the walls of the

equipment downstream of the reactor. These effects lead to broad peaks at best,

and prevent quantitative analysis of the reduction profile. For quantitative

measurements a thermal conductivity detector (TCD) has been used in most

studies. However, TCDs also have limitations because they are extremely

sensitive to flow patterns and back-pressure variations during temperature-

programming and they require calibration on a regular basis. If a TPD/TPRx

apparatus were used in TPRd studies, the mass 2 peak would be monitored con-

tinuously, as has been done in TPRx for methanation at low H, partial

pressures.

The effects of operating variables on the reduction profiles was investigat-

ed by Ho&i and Balker CZSI. The heating rate and the H, concentration in the

Page 11: Editor's note

11

reducing gas were found to have a more pronounced effect on the temperature of

the maximum rate of reduction than either the total flow rate or the amount of

reducible sample. Improved sensitivity in the TPRd slgnals is obtained by

using low feed rates of He. However, total flow rate is the parameter that

dominates in the time lag of the temperature measurement and in the dispersion

of the signal between the reactor and detector. Thus, significant errors can

result if the temperature of maximum reduction rate is used to evaluate kinetic

parameters for the reduction reaction C271.

In addition to H, consumption reflecting the reduction of metal cations.

other processes occur, as pointed out by Huang et al. C281. Reduced metal is

capable of adsorbing H, from the carrier gas, and the adsorbed H, can subse-

quently desorb. In some cases, the H, may diffuse along the surface of the

catalyst and spill over to the support. The spillover H, can then desorb from

the support at higher temperatures C293. Furthermore, some supports are

reducible and their reduction can contribute to the TPRd profile. In general,

the measured net consumption of H, is composed of some chemical processes that

do not involve changes in oxidation state. The following mass balance accounts

for these contributions I.281:

heasured H, consumption/emission = metal reduction + support reduction + H,

adsorption + H, spillover - desorption of

chemlsorbed H, - desorption of spillover H,.

The metal reduction term In this equation consists of an isothermal contribu-

tion and a temperature-programmed component. When a dehydrated sample is first

exposed to H, at the initial temperature of the experiment, some metal can be

reduced. Procedures are described below to obtain the TPRd profile, which Is

defined as that signature reflecting reduction processes that occur during

temperature programming.

(i) Experimental Sequence during Temperature Proprammim.

1) bhvdration. The sample is flushed with inert carrier gas as the tempera-

ture is increased, held at elevated temperature for a desired period of time,

and then cooled in the inert gas flow.

2) Reduction. The carrier gas is switched to a Hz/inert gas mixture, the

temperature is raised at a constant rate, and the changes in H, signal in-

tensity monitored. The resulting spectrum is designated as the first TPRd

spectrum. The catalyst is then flushed with the inert gas at the final temper-

ature and cooled in the inert gas.

3) Adsorntion/Desorption. The inert gas is replaced by the Hz/inert gas

mixture and the procedures of step 2 are repeated. The resulting spectrum is

designated the second TPRd spectrum. Subsequently, the catalyst Is cooled in

Page 12: Editor's note

12

the s/inert gas. The resulting Ha consumption durfng cooling is designated

as the adsorption spectrum. For a completely reduced catalyst, the second

TPRd spectrum should be a mirror image of the adsorption spectrum if neither Ha

spillover nor activated adsorption of Ha makes a contribution. When this is

not the case, step 2 should be repeated as many times as necessary until the

adsorption spectrum is the mirror image of the second TPRd spectrum.

4) Once step 3 has been completed, a TPD spectrum of Ha can be obtained to

retrieve additional information regarding the binding strength of Ha on the

surface and to estimate the metal's dispersion.

(iI) Experimental Sequence for Isothermal Reduction.

1) Dehydration.

2) Reduction/Adsorption. Pulses of Ha are injected into the inert carrier

gas while the catalyst is maintained at constant temperature. The injections

are repeated until the breakthrough of pulses occurs. The portion of sample

that is reduced can adsorb Ha during the isothermal exposure. The amount of

Ha adsorbed can be assessed by the following procedures.

3) TpD. The sample's temperature is programmed in the inert gas, the Hx

emission is recorded, and the sample is then cooled In the inert gas. The Hz

emission may not reflect a TPD spectrum from a completely reduced sample

because adsorbed Ha can react with the unreduced part of the sample during

heating. An additional TPRd experiment is necessary to assess this.

4) Temperature-Programmed Reduction. The inert gas Is switched to the

H&inert gas and the procedures outlined in step 3 during temperature-

programming are repeated. The difference between the resulting TPRd spectrum

and the 1st TPRd spectrum is the amount of Ha consumed during the previous TPD

experiment. The sum of the H, consumption in steps 3 and 4 is the total amount

of Hs adsorbed during isothermal reduction. Subtraction of the amount of Ha

consumed during isothermal reduction (step 2) yields the amount of Ha consumed

due to the reduction of the sample under the isothermal conditions employed.

It is noted that room temperature is only a convenient starting temperature,

and any temperature may be used.

On-line data acquisition facilitates the analysis of TPRd data. Not only

can reduction profiles be constructed by appropriate addition and subtractlon

of spectra, but quantitative calculations such as percent metal exposed can be

obtained by suitable integration.

A reduced catalyst, when exposed to an Oa/inert gas stream, will adsorb 0,

as a function of the temperature/reactivity propertles of the reduced speclcs.

When the temperature is ramped while the sample is exposed to Ox flow, this

transient technique has been designated as temperature-programmed oxidation

(TPO). A useful sequence to characterize the catalyst system Is TPRd/TPD/TPO

c301.

Page 13: Editor's note

The above procedures are the same whether a mass spectrometer or a TCD is

wed as the detector. Other reactive gases such as H.$ can replace the active

gas in the carrier stream, although few examples have been reported C311. One

advantage of using the same apparatus for the above studies and TPRx is that a

catalyst can be characterized initially by TPRd/TPD/TPO, transient reaction

experiments can be carried out by TPRx. and steady-state studies to assess

catalyst performance can be carried out serially on the same catalyst without

removing it from the apparatus.

Transient Isotom Tracing. Transient isotope tracing provides a means of

studying the kinetics and mechanism8 of a catalytic reaction without perturbing

the steady-state conditions of the reaction system. At steady-state conditions,

an isotopically labeled reactant is substituted for an unlabeled reactant In

the feed stream, and a mass spectrometer is used to observe the transient

responses of labeled and unlabeled reactants and products. This technique has

been described in detail in a recent book E321.

The objective is to rapidly switch from one feed stream, which Is flowing

through the reactor, to another feed stream, which is flowing to a vent. Con-

tinuous sampling of the effluent from the reactor, as a function of time, with

a mass spectrometer yields the transient data. A computer is necessary so that

several mass peaks can be monitored simultaneously. To simplify analysis, a

differential reactor or a continuous stirred tank reactor Is used. High gas

velocities have been used to obtain a well-mixed reactor C33-351. though

channeling is possible at these conditions.

A step function change in the feed stream is desired, and a motor-driven,

low dead volume, four-way switching value Is used to obtain a fast and repro-

ducible switch from one feed stream to the other. The most difficult aspect

of the switch is maintaining the same pressure drops, flow rates, and concen-

trations in the two streams. Slight differences in the flow rates or the

pressure drops of the two stream8 can result in additional perturbation8 in

the concentration signals, and analysis become8 impossible. Thus, back

pressure regulators are necessary to adjust the pressure drops downstream of

the reactor for the reactor stream and downstream of the switching valve for

the bypass stream. Pressure gauges downstream are useful for this adjustment.

Kass flow controllers maintain the constant flow rates needed. Transient

tracing experiments can be run with the reactant gases diluted with He In

order to discern small changes in the gas stream8 and to minimize the amount

of isotopically labeled gas used. A small concentration of Ar Is used in one

of the stream8 as a tracer, so that the responses of the valve and the system

are accurately known.

Once the isotope stream reaches steady-state, the feed stream is switched

back to the unlabeled flow in order to minimize isotope gas usage. After this

Page 14: Editor's note

14

second switch, transient data are obtained a8 the unlabeled gas replace8

the labeled gas in the reactor.

Sten Concentration Char-ares. In the previous section, the procedures to

accomplish step changes in concentration have been described. The same

method8 are used vhen the relative concentration of reactant8 are changed

abruptly during steady-state reaction. For example, In an experiment at

steady-state with H&O flow, a common transient experiment is a switch to H8

flow. The H8 can react with carbon-containing intermediate species and produce

CD,. Bennett C43 has pointed out that if this experiment is not done by a

svitch from X,/CO to Hg, but rather by suddenly stopping the CO supply in an

apparatus using continuous gas blending, a CH, peak is seen which has no rela-

tlon to the H, titration of surface carbon. The reason for this is that the

Hz/CO mixture is still in the feed line between the blending valve and the

reactor, and when the flow rate is reduced because the CO flow is stopped,

the conversion to CH4 is increased. This result8 in a "false" peak at the

beginnlng of the CH, transient. The same effect is obtained by a switch from

Hz/CO to He flowing at a lower rate. Other combinations are posslble. but

momentarlly stopping flow to a reactor at the time of a switch will introduce

spurious peaks in transient response data. Ways to minimize this effect

include using low dead volume valves, short gas lines, and preblended H,/CO

stream8 instead of continuously blending two streams.

Another variation of a step change in concentration has been used by Mori

et al. C363. The method consists of injection of a pulse of CO into a

carrier gas stream comprised of H, and an inert gas. The.CO is Immediately

adsorbed on the catalyst surface and subsequently hydrogenated to yield CH,

and HaO. The analysis of the transient response of products was called pulse

surface reaction rate analysis. The CH, concentration was measured by a flame

ionization detector (FID) during steady-state reaction, but when the rate of

CH,, formation was measured during transient studies, the column, upstream of

the FID, was by-passed to minimize the dispersion of CH, between the reactor

and FID. The Ha0 formed by the reaction was quantitatively measured by

trapping the reactor effluent after a pulse of CO. The response time of the

system was determined by injecting a pulse of CH, instead of CO; the CH, was

eluted from the bed immediately. On the other hand, after a CO injection, the

CD, signal rose rapidly and then decayed exponentially. According to the

theory developed to quantify pulse reaction rate analysis C371, the slope of

the straight line obtained by plotting the logarithm of the rate versus time

Is the rate constant of the surface reaction. The surface reaction step that

bra8 being measured. according to Mori et al., was the carbon-oxygen bond dl8-

soclation of an adsorbed CO molecule or a partially hydrogenated CO species.

Page 15: Editor's note

15

Hultlple-Cycle Transient Analvsls (MCTA)

The dicussions thus far have been confined to what we referred to as

single-cycle transient analyses. Another class of transient techniques,

referred to as multiple-cycle transient analysis (BCTA), can be conducted with

a TPD apparatus. In these experiments one can study kinetics under conditions

that are close to steady-state operation of the catalyst. During DICTA, the

response of the system to small perturbations is used to obtain information

about reaction kinetics.

The results of ETA, In conjunction with kinetic data from other transient

techniques, can be used to develop a self-consistent reaction model. However,

the kinetic model may not be unique. tlany mechanisms may give the same model,

and many such models may give adequate agreement with the experimental data.

Therefore, the results of experiments should be critically analyzed on the

basis of a reasonable kinetic model, which Is selected according to the

following criteria:

I) A minimum number of adjustable parameters.

2) Both qualitative and quantitative features of the experimental data, over

a range of experimental conditions, should be consistent with the model.

3) The resulting rate constants should be chemically reasonable.

Either the reactant concentration or the temperature can be perturbed during

DCTA experiments. Though the experimental geometry and the detection system

are the same as for TPD/TPBx experiments, certain additional design considera-

tions are necessary, as discussed in the sections below.

A variation of ECTA experiments involves simultaneous changes in inlet

composition to the reactor and temperature of the catalyst. One such sequence,

which was used to study chain growth during Fischer-Tropsch synthesis C63,

Involved contacting the reduced catalyst with a He/CO mixture for a period of

time, switching to a pure H, steam, and simultaneously Increasing the catalyst

temperature. The temperature ramp was imposed to enforce chain termination

and produce desorption. Since the amount of hydrocarbons obtained in one cycle

was insufficient for analysis, the products from a large number of cycles were

collected and analyzed. The use of this transient procedure to study the

methanation reaction does not appear to be as suitable as those offered by the

other DCTA methods outlined In this section.

HCTA-Concentration (HCTA-C). Programmable switching values are employed

to adapt the TPD apparatus for MCTA-C studies. During methanatlon, switching

the inlet flow between Hs and a s/He mixture while the CO inlet flow is kept

constant perturbs the H, reactant concentration. Switching times for valves

with a low dead volume can be less than 0.25 s. and under these conditions, a

H, perturbation reaches 90% of its new value within 2 s C381. Precision

cross-pattern valves (Swagelok, SS-AEX) are used to feed CO into the H, con-

Page 16: Editor's note

16

taining streams. Tylan mass flow controllers (FC-2601 allow the %/CO ratio

and the magnitude of the perturbation to be varied. A sampling port above the

reactor allows one to verify that the CO concentration is constant at the

reactor inlet.

To conduct HCTA experiments, a reactant feed of Ii, and CO is Introduced,

the temperature is increased and the system is allowed to reach steady-state

conditions. While the total pressure is held constant, the Hz/CO ratio is

perlodically perturbed by switching to a second feed stream with a slightly

higher Ha/CO ratio. The wave forms above the reactor should be routinely

checked to ensure a periodic input function. With this configuration the flow

rate, the H&O ratio, the perturbation, the weight of catalyst, the cycle time,

and the temperature can be varied.

UCTA-Temuerature (UCTA-T). In kinetic studies designed to elucidate the

mechanism of catalytic reactions, it is common to observe abrupt changes in the

activity pattern for the catalyst as a function of temperature. This generally

signals some type of irreversible alteration of the catalyst surface at a trans-

ition temperature. For CO hydrogenation over Ni such effects have been attri-

buted to the deposition of unreactive carbon. A transient temperature technique

has been proposed to follow activity pattern changes during steady-state opera-

tion T391.

Multiple-cycle transient analysis-temperature (HCTA-T) periodically changes

a reaction system between two temperature limits in a sinusoidal wave while

maintaining a constant reactant feed. The time response of the product (CH,)

signal Is periodic, but its symmetry is a function of experimental conditions.

A TPD apparatus with a microprocessor-controlled infrared oven (bicricon

823) has been employed in HCTA-T studies C391, and the temperature of the

100 rug bed of Ni/SiO, catalyst with a 0.225 mm particle diameter was measured

at the center. A uniform temperature distribution was demonstrated to exlst

through the catalyst bed (403. A Ha/CO gas mixture Is introduced and the

temperature is increased. Throughout the experiment the m/e=lS peak Is con-

tinuously monitored by an on-line mass spectrometer. After the system reaches

steady-state the rate of methanation is recorded. The temperature is modulated

in a sinusoidal wave with an amplitude of f S K about this reaction temperature.

Accomplishing the desired temperature variation requires a balance between the

power input to the oven and the water flow rate in a cooling co11 surrounding

the reactor. The data from such experiments are used to test deactivation

models, which are constrained to agree with the experimental results over the

time domain of the perturbation.

TRAHSPORT EFFECIS

In this section we review the recent studies ,that have analyzed transport

Page 17: Editor's note

17

effects during TPD/TPRx experiments. The small beds generally used should

minimize transport effects. However, if the dead volume of the reactor is

large, then the delay time resulting from that volume will be a significant

contribution to the relaxation time of the total system. When this is the

case, the prediction of the concentration of an Intermediate will be in error,

although the concentration of reactants might be uniform throughout the

reactor. Hass transfer effects during TPD and TPRx and heat transfer effects

during other types of transient experiments are also discussed in this section.

Absence of Transport Limitations

In general the application of transient techniques requires the absence of

transport effects. Under this assumption the performance of a catalyst bed can

be described with a continuous stirred tank reactor (CSTR) approximation. Such

a model consists of a perfectly mixed fluid phase with a continuous flow and a

stationary catalyst phase. This approximation was used by Happel et al. C411

to describe the performance of a gradientless recirculating reactor. Biloen

et al. C331 used the same approximation for a differential catalytic bed. Lee

et al. C221 examined the differential bed requirement and demonstrated uni-

formity In adsorbate concentration within a bed of loosely packed catalyst

particles when subject to a flow of reactants. The advantage of the CSTR

approximation comes from its mathematical simplicty. A set of ordinary differ-

ential equations with constant coefficients describes the reaction system.

Happel et al. C341 and Blloen et al. C331 used transient isotope tracing

subsequent to a step change of reactant to study methanation and Fischer-Tropsch

synthesis over Ni catalysts. They suggested that the coverage of surface

intermediates (Bi) could be related to the relaxation time (~1 and the turnover

number of products (TON) by the relation

TON = 0i ~-1 (1)

Experimental design is important to insure that Equation 1 retains its

validity even if the CSTR model is a good approximation. Huang et al. C421

analyzed the CSTR approximation under transient operation by considering the

volume occupied by a CSTR. They reasoned that for adsorbed species that are

uniform in concentration throughout the catalyst bed, the number of active

sites on the catalytic surface can be considered equivalent to the volume of

a homogeneous CSTR. Therefore, the hold volume of the CSIR is not the total

volume of the catalyst bed but is the volume of the gas that can be adsorbed

on the catalytically active surface. Thls quantity can, in principle, be

determined by selective chemisorption. Under these assumptions the residence

Page 18: Editor's note

18

time is the time required for this equivalent volume of the heterogeneous CSTR

to ba replaced with the volume of new adsorbate whose concentration is evalu-

ated under the reaction conditions of pressure and temperature.

Several reaction models were examined by simulation, subject to the assump-

tion that mass transfer effects were absent. First-order relaxation analysis

to obtain intermediate concentrations is valid only over a limited range of

experimental conditions, and it overestimates the intermediate concentration.

Based on the hold volume of a heterogeneous CSTR, four experimental factors

contribute to the actual residence time. These are the weight of catalyst,

the metal surface area, the volumetric flow rate, and the reactant that acts

as a perturbation. From these experimental parameters the residence time

constant can be estimated and compared to the experimentally measured relax-

ation spectra. To demonstrate their analysis, Huang et al. considered two

reported studies using transient experiments designed to study the Hz/CO

reaction over Ni. Table 1 presents the pertinent data. Previous studies

TABLE1

Residence Times for Ref Ref Laboratory Reactors c333 c221

Catalyst Weight (g) 4.4 Gas phase hold up (cm31 SO Nickel surface area (me/g) 22 Total pmoles adsorbed 2.5 Heterogeneous CSTR hold volume occupied by CO (cma) 28

Volumetric flow rate of the perturbed species (laCO)(cms/s) 2.0

Residence time (V/F) based on total volume (s) 40

0.03 0.009 60 4.6

O-L'0

4.4

0.04

L43.443 have shown that the surface coverage of CO under steady-state condi-

tions is 5 - 6 times that of H,. Then, for example, if the perturbation to the

reaction system is CD, -857. of the calculated heterogeneous CSTR volume is

required to compute the residence time; if the perturbation is H,, -19 of the

heterogeneous CSTR volume is required to compute the residence time. When the

results shown in Table 1 and those given in Ref. 42 are used in Equation (l),

the magnitude of the intermediate concentration rB1 is greatly overestimated.

Hass Transfer Effects

The intrinsic kinetic parameters for surface processes can often be obscured

by transport effects. A number of papers have appeared in the past six years

that provide guidelines for experimental design to minlmize this phenomenon in

transient temperature experiments carried out in a TPD apparatus. The influ-

ence of mass transfer on TPD spectra has been examined from the standpoints of

nondimensional analysis C45,461 and numerical simulation of specific models

Page 19: Editor's note

19

c47-491. Gorte E451 showed that four characteristic groups determine the

influence of experimental variables on TPD spectra from a packed bed of porous

catalyst particles. Convective and diffusive lags, and particle and bed con-

centration gradients were analyzed. Readsorption effects were later considered

by Demmln and Gorte C461, who evaluated two additional parameters that indicate

the importance of readsorption at low and infinite carrier-gas flow rates. A

summary of these parameters is given in Table 2.

TABLE 2

Influence of Characteristic GrOuDS on TPD Spectra

Parametera Definition Observed effect Ideal requirement

E Rz$

Dp(Tf-To)

&&iii P

apsFRz szD

P

apsFV(l-En)

Q

Residence time of carrier gas

Time constant for diffusion out of an individual particle

Ratio of carrier- gas flow rate to rate of diffusion

Ratio of carrier- gas flow rate to axial mixing

Ratio of adsorption rate to diffusion

Ratio of adsorption rate to carrier-gas flow rate

Convective lag

Diffusive lag

Particle concen- tration gradients

Bed concentration gradients

Readsorption at infinite flow rate

Readsorption at low flow rate

Bust be to.01 for negligible lag

Must be <O.Ol for negligible lag

Must be <O.OS for negligible gradient

Bust be <O-l for CSTR

Must be <l for negligible readsorption

Xust be <I for negligible readsorption

%@lbols: Es, bed porosity; V, bed volume; 8, heating rate (K/s); Q, carrier gas flow rate; T,, final temperature; To, initial temperature; E , particle porosity; R, radius of particles: D,, particle diffusion coeffic ent; P N, number of particles in bed; L, bed length; Da. bed dispersion coefficient; a, particle surface area; p, particle density; s, sticking coefficient; F, (RT/2nM)l/z cm/s;

In a later paper Ricck and Bell C481 provided a theoretical analysis for

TPD in which the bed was modeled either as a single CSTR or as multiple CSTR's

connected in series. They showed that the position and shape of the spectra

are functions of catalyst particle size, bed depth, carrier-gas flow rate, and

composition. Since the position and shape of spectra are used to determine

Page 20: Editor's note

kinetic parameters, their work provided estimates of the errors involved when

readsorption and mass transfer effects are not considered. Huang and Schwarz

CSOI included the effects of temperature-dependent transport parameters in

their analysis of the mixing cell model of Rleck and Bell. They concluded that

mass transfer effects on TPD spectra require solution of a continuity equation

that includes both axial dispersion and convective transport.

The general conclusions from these theoretlcal studies have provtded useful

information and guidelines for the interpretation of experimental TPD spectra:

1) Extensive readsorptlon of gas occurs within the catalyst particles and

the local adsorbate coverage Is governed by equilibrium adsorption.

2) Intraparticle concentration gradients can be mintmized by reducing

catalyst particle size and carrier gas flow rate.

3) Axial gradients in the gas phase concentration of adsorbate can be

minimized by the use of shallow catalyst beds.

4) Upon heating the catalyst, nonuniform Initial adsorption profiles become

uniform before significant loss of adsorbate from the particles.

S) Reasonable estimates of the adsorption enthalpy can be obtained using a

theoretical relationship based on the assumption of equilibrium adsorption in

the absence of intra- or interparticle gradients in the adsorbate concentra-

tion.

An analog to the Weiss-Prater method was used by Huang et al. CSII to

assess the importance of mass transfer limltatlons during a TPD experiment.

Weiss and Prater CS21 proposed that the effect of lntraparticle diffusion can

be contained in a separate factor, n, the so-called effectiveness factor in

the general rate equation

R = kVc CE n

where R is the observed desorption rate (mol/s), V, is the bed volume fcms),

Co is the concentration of the reactant at the particle surface (mol/cms), and

n is the order of the reaction. The reaction rate constant, k, has units of

(concentration) . I-ns_i If the reactant is supplied by the outside environment,

and its diffusion into the particle is rapid compared to the reaction rate,

the entlre Internal surface area within the catalyst particle will contain a

uniform concentration profile. As a consequence, the whole catalyst surface is

effective In promoting the reaction. On the other hand, during a TPD experiment

the reactant (i.e., adsorbate) is supplied by the reservoir inside the particle

and it diffuses outward through the particle. To preserve the simple form of

Equation (2), Huang et al. proposed an effectiveness factor for TPD experiments

in which readsorption effects were considered. They found that the effective-

ness factor was determined by:

Page 21: Editor's note

21

I) observed desorption rate

2) particle size

3) surface concentration of adsorbate

4) diffusivlty of the adsorbate in the carrier gas/catalyst particle

5) catalyst bed volume

6) heating rate

When the effectiveness factor is much less than unity, intrapartlcle diffusion

dominates and the experimental conditions need to be altered. Their character-

istic plots for first- and second-order desorptlon, shown in Figure 2, can be

used to select the appropriate experimental conditions to minimize the effects

of intraparticle gradients during the TPD process.

First-order desorDtion

0.65Llllllnl _7 ' ' ' ' fl,*,,I ' ' ' ' III11 &O

r3 U6 IO-’ < R/VcDeff

Second- order desaption

‘0.80

0.65

!O-e lo-6 ri R/V, Dew

Figure 2. Characteristic plots of effectiveness factor, 0. versus RrE/V,*D,ff parametric in surface concentration of adsorbate (mol/cm3).

Page 22: Editor's note

Heat Transfer Effects

Some of the transient techniques that were described in the previous section

operate by superimposing a transient input on a steady-state condition. There-

fore heat transfer effects, which could give rise to anomalous temperature

profiles within the bed, need to be considered. Calculation of the temperature

profile requires a knowledge of the effective thermal conductivity of the bed.

A transient technique was proposed to determine the mean effective thermal

conductivity of small beds of supported metal catalyst such as those used in

TPD experiments E401. The perturbation was a sinusoidal temperature wave

symmetrical about an average temperature. An unreactive carrier gas was used

to eliminate heat effects due to reaction. A comparison was then made between

the temperature measured from thermocouples imbedded in the bed and the

temperature calculated by solving the appropriate energy balance parametric in

the mean effective thermal conductivity. The minimum deviation over a cycle

between temperatures provides a lower bound for the mean effective thermal con-

ductivity. To demonstrate the effect of operating conditions on temperature

uniformity within the bed, the methanation reaction over Ni/SiO, was considered.

A rate expression in conjunction with the suitable energy balance was used to

simulate the radial temperature gradient as a function of the mean effective

thermal conductivity. As the mean effective thermal conductivity decreases,

which occurs as the metal weight loading is lowered on insulating supports, the

temperature gradient increases. The gradient also increases dramatically If

the operating temperature increases, even while maintaining the conversion

below 10%. Therefore, evaluation of rate data based on temperature uniformity

must be carefully considered, even in the typically small beds used in trans-

ient studies, because of the heat transfer resistance of the catalyst bed.

CATALYST PREPARATION AND CHARACTERIZATION

Preparation

The attainment of high activity, selectivity, and longevity of supported

metal catalysts is directly related to the steps in catalyst preparation. The

variables that determine catalyst performance Include the method of metal salt

deposition, the drying procedure, calclnation conditions, reduction time and

temperature, and flow rates. Therefore, these must be uniformly and repeatedly

controlled.

One distinct advantage of the transient techniques described in this review

is that they can be carried out in one apparatus, on the same sample, without

exposing it to the atmosphere. This is a desirable feature because the impact

of each stage of the preparation sequence can be studied, and the Influence of

the preparation steps can be evaluated.

Page 23: Editor's note

23

In this section ue will discuss the application of transient techniques to

study variations in preparation procedures of Ni catalysts. Our focus will be

coprecipitated and impregnated catalysts. Coprecipitated catalysts are used In

commercial methanation. but impregnated catalysts have been studied extensively

in research laboratories because of ease of preparation. A general description

of these catalyst preparation methods will be given first and then transient

studies that have been applied to characterization of the catalyst and/or cata-

lytic precursor will be reviewed.

Precipitated Catalysts

Precipitated catalysts are usually prepared in three sequential steps:

1) precipitation and washing

2) calcination of the precipitate

3) reduction, which Is usually followed by calclnation.

Precipitation is carried out in an aqueous environment In which Ni cations

and/or support ions are brought in contact with suitable counterions such as

present in carbonate, ammonia, or hydroxide solutions. Above a certain pH,

precipitation occurs. hany variables, such as choice of anions/cations, pH and

temperature of the precipitation process, stirring rate, and aging conditions

can influence the composition and structure of the precipitate and ultimately

the catalytic properties.

Precipitated catalysts can be formed elther sequentially or as a coprecipi-

tate. In the former case one of the components is precipitated first and the

second is precipitated on top of the first. In the coprecipitation route both

components are precipitated at the same time using either a rising pH method or

a constant pH method. In the rising pH method the precipitating agent is added

at a constant rate to the solution, and the increasing pH causes a precipitate

to form. In the constant pH method, solutions of the metal and support ions

and the precipitating agent are added simultaneously. The pH of the combined

solution is controlled by the rate at which each of the individual solutions

are added. Coprecipitated catalysts are less likely to lead to inhomogeneous

precipitates ES3a.bl.

It is important to note that at some point in the preparation sequence, the

precipitate must be washed to remove undesirable cations (e.g., Na+) that come

from the prccipltating agent. On the other hand, additional cations that are

active promoters can be added during the precipitation process CS3c,dl.

Impregnated Catalysts

The first step in the preparation of impregnated catalysts is impregnation,

which generally involves contacting a support with a solution containing an

active precursor salt. The various species in solution can interact wlth the

support in several ways, ranging from ion exchange and coordination or complex-

Page 24: Editor's note

24

ation reactions to precipitation of the metal salt in support pores. The type

and strength of interactions may be affected by the pH and ionic strength of

the solution, the point of zero charge of the support, which corresponds to

the pH of the electrolyte when there is no net charge on the support surface,

and the charge and concentration of the adsorbing species.

Two general procedures, wet impregnation and Incipient wetness, are used to

prepare impregnated catalysts. An amount of impregnant in excess of the pore

filling volume of the support is used for wet impregnation. The volume of

impregnant used corresponds to the pore volume of the support during incipient

wetness. For nonporous supports, the pore volume is not meaningful and since

the total impregnant voluma used can affect the final catalytic behavior CS41.

the volumes used should be specified.

The weight loading of the metal Is an Important parameter. For catalysts

prepared by wet Impregnation, the weight loading Is Increased by using a higher

concentration of metal salt, but this also changes the pH of the impregnating

solution. Any added acid or base affects both the ionic strength and the pH of

the aqueous phase. Increasing the ionic strength controls the amount adsorbed

on the support by shrinking the electrical double layer thickness ESSI. The

metal ion activity is likewise changed CSSI. Varying the pH of the impregna-

tion solution changes the surface charge of the support CS61 and the amount

of AlgO, dissolution CS'II. Due to these factors, the weight loading of

catalysts prepared by wet Impregnation must be determined experimentally. Use

of incipient wetness procedures results in no ambiguity in the weight loading;

it is determined by the pore filling volume and the concentration of metal salt

In the impregnant. For wet Impregnation, the metal loading depends on, among

other things, the extent of adsorption, the attainment of adsorption equill-

brium, and the amount of metal precursor contained in the pore filling volume

CS81.

Catalyst Characterization

Temperature-Prom-and Reaction (TPRx). Since our last review a large

number of studies using TPkx have appeared. Carbon monoxide hydrogenation has

been the most extensively studied system. Since the objective of this section

is to relate how variations in catalyst preparation are reflected In the

results from transient studles in a TPD/TPRx apparatus, we will only focus on

those reports where variables involved in preparation have been examined using

TPRx.

Huang and Schwarz t16,59-611 studied the effect of catalyst preparation on

the catalytic activity of Ni/Al,O, catalysts prepared under controlled con-

ditions. They proposed that 1orpH impregnation solutions tend to facilitate

the formation of NiA1,04-like species C621. They found that high weight load-

ing catalysts prepared k-om impregnants whose pH was adjusted to 1 have two

Page 25: Editor's note

25

types of reaction centers during TPRx, whereas low weight loading catalysts

prepared at the same initial pH of I have a single type of reaction center.

Figure 3 shows their TPRx spectra for low and high weight loading catalysts

prepared by two preparation methods: wet impregnation (W) and incipient

wetness (I).

IO- , I I I J 530K (01

400 600 773 -

Temperafure(K)

IOr , 534L I 1

(cl u a-

fE - 5 6- .- ‘5 -

g 4- I

z

I I I 400 600 7?3-

Temperature(K)

457K (b)

aI E 2.0-

a

g 1.5- .- 5

2 l.O- ZJ r

0.5-

.s! d: c .o z s

_c B E

Tempemfure(K)

I 8 1 6- I

(d) 432K n

4- 508K

, 600

Temperolure tz3

Figure 3. TPRx spectra for Ni/Al,O, catalysts: (a) 0.9 wtX (I) (b) 6.23X (I), (c) 0.841: (W), and (d) 6.53% (W). W = wet impregnation; I = incipient wetness. The rate is in pmol/g Nibs.

XPS studies of these catalysts showed the existence of two types of Ni

species. On the low weight loading samples, only features similar to NiAl,O,

were found. On the high weight loading sample, both Ni and NtAl,O, were pre-

sent. Their TPRx and XP'S results are summarized in Table 3. They proposed

that the reaction centers on the low weight loading catalysts prepared by

either method are due to NlAl,O,-like species only C17.621. They also

Page 26: Editor's note

26

proposed that nickel from NiAl,O,,-like species produced the high-temperature

TPRx peak and reduced nickel produced the low-temperature peak C17,621.

TABLE 3

Comnarlson between the ESCA and TPRx results Nickel weight Catalysta ESCA TPRx loading preparation surface nickel peak temperaturesb (Z) method speciation (K)

A B

0.84 W NiAl,O, _-- 534 6.58 W Ni+NiAlaO, 432 SO8 0.90 I NiAl,O, _-_ 530 6.23 I Ni+NiAlsO, 457 538

"w: wet impregnation; I: incipient wetness; b(A) low-temperature peak, (B) high-temperature peak.

Huang et al. explained the Ni speciation on low and high weight loading

catalysts on the basis of solution/support interactions during catalyst

preparation. During wet Impregnation, the precursor dlstrlbutlon consists of

an adsorbed component and a pore-filling component. Initial contact of the

impregnant with the support leads to an exchange of Nia+ cations with adjacent

hydroxyl groups, thus forming a bidentate cluster. This cluster is converted to

a NiAl,O, type species during catalyst activation. The weight loading of these

precursors is close to the hydroxyl group density on an Al,O, with a BET area

of 180 ms/g C631. For the catalysts that gave two CtI4 peaks in their TPRx

spectra, the precursor loading was much higher than the cation-exchange

capacity of their Al,O,. On the other hand, the precursor loadings for those

catalysts with a single methane peak in the TPRx spectrum were comparable to

the cation-exchange scheme.

The above results demonstrate how transient experiments can be used to

study the impact of preparation on catalyst performance. However, only a

single heating rate was used in the TPRx experiments. Falconer and co-workers

C8,11,131 have shown that at heating rates higher than those used by Huang

et al.. two CH, peaks are observed in TPRx experiments, regardless of the

weight loading. They attributed the higher temperature peak to methanation of

a methoxy species that is formed due to splllover of CO from Ni to the A1,Os

support. This peak could be saturated by sequential exposure of the catalyst

to CO at elevated temperatures.

Temperature-Programmed Surface Reaction (TPSR). The performance of cata-

lysts has been shown to be strongly dependent on their methods of preparation.

In a series of recent reports C16,17,!58-623 the effects of preparation Pro-

cedures (incipient or wet impregnation) as well as metal concentration, pH,

Page 27: Editor's note

27

and ionic strength of the lmpregnant solution were studied for Ni/Al,O, cata-

lysts. The amount of carbon-containing residue left on the metal or support

surface after steady-state CO hydrogenation was determined by TPSR. The

results show that catalysts prepared by incipient wetness tend to have more

carbon-containing residues left on the catalyst after a constant exposure to a

H&Z0 mixture. Furthermore, the carbon inventory for catalysts prepared by

incipient wetness is more complex than that for catalysts prepared by wet

impregnation. A8 many as four types of carbon-containing species, as evident

from their different peak temperatures. are released as CH, during temperature

programming in Hx. For the catalysts prepared by wet impregnation, a maximum

of three types of carbon-containing species was observed. In the earlier TPSR

studies, the surface carbon species were formed by dissociation of CO at ele-

vated temperatures. It is perhaps more relevant to study the amount and type

of carbon residues left on the catalyst after steady-state reaction. The

carbon containing specie8 may serve as Intermediates In methanatlon C39,43,601

and/or as site-blocking poisons C641. Detailed information about the carbon-

containing species thus obtained can be Used to guide catalyst design, to

define regeneration conditions for the used catalysts, and to obtain additional

insight into the mechanism of the methanatlon reaction C6Sl. Indeed, criteria

for the design of NI/Al,O, catalysts that would carry a given level of carbon-

containing residues have been established C601.

Temperature-Programmed Desorvtion (TPD). A critical factor In catalyst

preparation Is the resulting dispersion of the metal. Selective chemisorption

under static conditions Is generally employed to obtain a measure of the dis-

persion. This, however, requires a separate apparatus and transfer of the

sample in the atmosphere. Instead, TPD can be used to estimate the catalyst

dispersion Immediately after decomposition of the precursor. Typically, H, is

the gas used and the adsorption stolchiometry on Nl is one H atom per Ni atom.

Generally, uniform adsorption is assumed.

A number of assumptions are inherent in using TPD or static chemisorption

to obtain metal dispersions. Following sample reduction, If all the H, 18 not

desorbed from the metal surface, the dispersion will be underestimated. High

temperature treatment in an inert carrier gas flow can minimize this possi-

blllty, but may result in catalyst sintering. Under flow conditions, a

portion of the adsorbate might desorb even at room temperature and be convect-

ively transported through the bed. This is particularly true if the heat of

adsorption is strongly coverage dependent C661 and the dispersion would be

underestimated. For Hi/SiOz catalysts, calculations based on a specific

adsorption-desorption model indicate the effect can be as large a8 a 303: under-

estimate based on static chemisorption result8 I491.

Page 28: Editor's note

20

The influence of the time of contact between Ha and the sample on the amount

of H, adsorbed is another consideration. As can be seen from Figure 4 for

coprecipitated Ni/Al,Oa catalysts, a peak develops at lower temperature as the

adsorption time is increased. These results were explained on the basis that

the rapidly-adsorbed Hz, which resulted from short exposures, is adsorbed

more strongly than is the slowly-adsorbed H, CS3a,671.

Figure 4. TPD spectra for H, from NI/AL,O, after: (a) pulsed adsorption, (b) continuous adsorption for 16 min (130 Torr H, in Ar). and (c) continuous adsorption for 1S.S h (130 Torr PI in At-). Reproduced from the Ref. CS3al.

The adsorption temperature and the H, pressure can also influence the re-

sulting TPD. These effects can only be assessed experlmentally, although it

has been proposed that the time dependence of theLadsorption process during

static chemisorption experiments can be used as a guide for determining contact

time effects during TPD experiments E681. At high coverages, if the equillb-

rium adsorbed amounts remain constant as the $ exposure is increased, then no

additional adsorption will occur for longer contact times during a TPD experi-

ment.

Use of adsorbates other than H, for TPD to determine Ni dispersion is

generally difficult. For example, CO disproportionates at elevated temper-

atures and releases CO and CO,. Carbon can be left as a residue on the

surface and a meaningful mass balance to assess total uptake is difficult

c491. In addition, the adsorption stoichiometry of CO on Group VIII metals

(in particular Ni) Is subject to controversy. Multiple adsorption sites and

Page 29: Editor's note

29

the possibility of carbonyl formation have been reported by several lnvestl-

gators C69-711. On the other hand, TPBx of CO to yleld CH4 has been wed to

provide an estimate of metal dispersion. In thls case, the adsorption

stolchlometry Is still required for an unambiguous determination of metal dls-

perslon and experimental data are needed to measure the amount of CO desorbed

and the amount of higher hydrocarbons produced. In addition, carbon residues

must not be left on the surface.

The objective of this section of the review is to place in perspective how

TPD experiments to determine metal dispersions can be used to study the effect

of catalyst preparation on the properties of the finished catalyst. In a

recent series of papers, Huang et al. C16,61,623 used TPD of H, to assess

variations In the metal dispersion as a function of weight loading. solution

pH, and method of Impregnation for a series of impregnated Ni/Al,O, catalysts.

Catalysts prepared by lnclplent wetness (2 g Al,Os/S curs of electrolyte) and

wet Impregnation (2 g AlaO,/ cm3 of electrolyte/l hr contact) were studied.

Table 4 compares the dispersion for catalysts prepared at the sama Initial

lmpregnant pH with approximately equal Nl weight loadings for the two

TABLE 4

Effect of Catalyst Prenaratlon on Dispersion

Nickel weight Inpregnant Bethod ofa Dispersion (%jb loading (I!) PH preparation

0.90 1 0.84 1 6.23 1 6.58 1 1.30 3 1.21 3 5.00 3 4.12 3 2.87 s 3.04 S 6.23 S 7.27 S

I 89.0

W 93.5

I 11.0 W 11.7 I 27.3 W 70.1 I 9.4 W 20.7 I 13.3 W 29.4 I 4.5 W 10.7

?I: wet Impregnation; I: incipient wetness b ticasured by H, chemlsorption and assuming complete reduction

Impregnation methods. The dispersions were obtained by TPD of H,. Based on

these results a rationale for the processes occuring during Impregnation was

proposed Cl61. Huang et al. argued that the contact time between the electro-

lyte solution and support for incipient wetness IS longer than that involving

wet Impregnation. Thus, during Incipient wetness a sedimentation type of pro-

cess occurs for the Nl precursors and dissolved Also, species. Dissolution of

Page 30: Editor's note

30

~1~0, becomes more dominant at lower pH’s CSII. For vet impregnation, the

relatively short contact time only allows the electrolyte to be adsorbed on

specific active sites on the support and thus results in a crystallization pro-

cess. For catalysts prepared by incipient wetness, the sedimentation tends to

yield bigger particles than the crystallization process because of the longer

contact time.

Temperature-Programmed Reduction (TPRd). Differences in the preparative

routes for precipitated and Impregnated catalysts might be expected to yield

catalysts with different structures and activities. Indeed, this has been

observed and one of the most effective ways to study and compare the properties

of such catalysts is by TPRd. We will first review some recent reports, which

have used TPRd to study the effects of variations In preparation procedures of

precipitated Ni catalysts. This will be followed by a complimentary discussion

of TPRd results for impregnated Ii catalysts. We conclude this section with an

example of how TPRd can be used to test the results of modeling the adsorption

process during wet impregnation.

Ross et al. C72-743 developed a two-phase model for the oxide form of

copreclpitated catalysts of high Ni content. They proposed that one phase

consisted of poorly ordered spine1 (NIAl,O,) and the other of NIO with some

dissolved AleO,. They suggested that these structures gave rise to two distinct

types of Ni upon reduction; the NiO was easier to reduce and generated bulk Ni

crystallites, and the NiAl,O, phase produced monodispersed Ni atoms that

remained closely associated with the AleO, structure. The H, adsorptlon

behavior and cata1ytl.c activity of these two types of Ni sites are different

and they found that the activity and selectivity of the catalysts varied

considerably as a function of the reduction temperature of the catalysts.

Precipitated catalysts have been studied primarily by examination of the

bulk structure of the catalyst by XRD. The metal content of these catalysts

is high, and interactions between the metal and oxide phases are so extensive

at all stages of preparation that bulk techniques are far more Informative

than they are for impregnated catalysts. However, in conjunction with XRD,

TPRd studies provlde additional information about the influence of preparation

procedures on the structure of the catalyst. For a series of copreclpitated

catalysts with Identical compositions, but calcincd at different temperatures,

the peak temperature maxima In TPRd experiments shifted to higher temperature

as the calclnation temperature Increased CS3e3. Puxley et al. C7Sl showed that

a sample calclned at ISSO X had a reduction peak close to the temporature of

pure, bulk NiAl,O,. A small peak was also seen at 723-773 K, which is just

above the reduction temperature of pure NIO.

The calcination temperature also causes catalyst structural changes, which

Page 31: Editor's note

31

are indicated by the lattice parameters determined by XRD C761. In Figure S,

from the work of Doesburg et al. (761, the lattice parameter is shown to

increase as the calclnatlon temperature is increased. They suggested that,

because the lattice parameters of the NiO-like phase are a function of the

calcination temperature, the amount of Ala+ present in the NiO-like phase is

important in the reduction process. Alzamore et al. C773 suggested that

reduction takes place within the NIO crystallites and Al Ions diffuse at the

same time to the surface of the crystallite8 and promote nucleation of the

AlsO, phase.

a(ii) 4.18

‘NiO-

4.17

I I 1 I

800 IO00 I200 T WI

Figure S. The a-lattice parameter of a calcined, coprecipitated Ni/Al,O, catalyst as a function of calcination temperature.

Copreclpitated NI/Al,O, catalysts are highly resistant to sintering. and

this has been explained on the basis of XRD and TPRd experimental results.

The presence of small Also, crystallites outside the Ni crystallites have been

proposed to physically block transport of Ni ions L771, but other explana-

tions of the same data have been proposed. Puxley et al. C7SI suggested that

a small fraction of the Al ions stay with the Ni crystallites, possibly as

aluminate groups. This results in slightly distorted lattice dimensions.

Such crystallites. termed paracrystalline, are more resistant to sintering.

Page 32: Editor's note

32

The argument is based on the opposing factors of strain-energy and surface

energy. Strain arises because of the presence of aluminate groups within the

NiO crystallites. The strain caused by aluminate groups near the surface can

be relaxed to a greater extent than strain caused by those groups present

within the bulk structure. Consequently, for smaller crystallites a larger

portion of this strain is relaxed and the strain energy lowered. However,

smaller crystallites have larger surface energies and these arc more

susceptible to sintering. These two phenomenon act in opposing directions

and equilibrium results when the sum of both energies is at a minimum C681.

A comprehensive study designed to compare the properties of sequentially

precipitated and coprecipltated catalysts of high Ni/Al ratlos was recently

reported IS3bl. The seqentially precipitated catalysts were formed by first

precipitating the Also, component and then the Ni component. During TPRd, two

reduction peaks were found for the sequential precipitates, whereas only one

was found for the coprecipitate. These results led to the proposal that the

most likely structure for coprecipitated Ni/Al,O, catalysts of high Ni/Al

ratios Is one in which Als+ ions are included in the NiO phase; i.e., no

separate pure NiO phase exists.

Zielinskl CT81 found that nickel oxide appears in impregnated Ni/Al,O,

catalysts in two forms, as "free" and "fixed" oxide. This finding is similar

to those for precipitated catalysts and occurs after high temperature

calcination of the fresh catalyst_ The fixed form of the oxide was connected

with the formation of NIAlZ04. which was difficult to reduce. The TPRd spectra

of his catalysts showed two peak features; the NiO reduced before the NiAl,O,.

Similar results were reported by de Bokx et al. I791 for Ni precipitated on a

nonporous Also, after high temperature oxidation of the precipitate. In both

cases the peak temperatures in the TPRd spectra were higher than those found

for samples calcined at lower temperatures.

Quantitative measurements of TPRd and TPO of 2X Ni/Al,O, at temperatures up

to IS00 K were recently reported by Kadkhodayan and Brenner CSOI. They found

that supported Ni (as well as Fe, ho, and W) could be reduced to metal only at

temperatures near 1400 K. The reduction of the supported metals always

occurred at temperatures higher than those of the bulk oxides. The authors

suggested that Interaction of the metal oxides with the support inhibited

reduction. On the other hand, TPO of the supported metals usually occurred at

a temperature lower than those for the bulk metals, which suggested to them

that oxidation diminishes metal-support interaction and the higher rates of

oxidation of supported metals reflect their smaller crystallite size. They

also found that sequences of TPRd and TPO showed changes particularly notice-

able in TPRd spectra. Specifically for Ni. they reported that the higher

Page 33: Editor's note

temperature peak obtained during the second TPRd was shifted to lover tempera-

tures after an intermediate TPO. The lower temperature peak shifted to higher

temperatures. They attributed these effects to changes in Ni crystallite size

due to the hlgh temperature cycling in H, and 0,.

The temperature-programmed reduction method is particularly useful as a

diagnostic tool. However, assigning peaks of a TPRd profile to definite

chemical species or to the same species located in different sites on the

support is difficult. A common feature in most TPRd spectra for impregnated

NI/Al,O, catalysts is the existence of two reducible states separated by - 100

K. A remarkably similar observation was made by tllle et al. C811 for Ni/SIO,

catalysts prepared by impregnation. They found that a sample, oxidized at

873 I< and then subjected to TPRd up to 900 K revealed two TPRd peaks. Reoxl-

dation of this sample at different temperatures and for varying times also

showed two TPRd peaks separated by * 100 K. These two TPRd peaks resembled

the two peaks of the original sample but were displaced to much m tempera-

ture. bile et al. proposed that both peaks were due to the reduction of NiO.

The more reducible NiO was located In the small pores of the SiOx support, and

the less reducible oxide was located in the larger pores. They concluded that

even reduction of surface NiO Is controlled by the environment of the original

oxide. These findings suggest that peak temperature maxima are not a reliable

fingerprint for specific identification of a reducible state of Ni. On the

other hand, the peak temperature separation C-100 K) is Indicative of a common

chemical difference between reducible Nl species.

The effects of thermal treatments on the properties of impregnated Ni cata-

lysts are not confined to reactive gases. The reducibility of a series of

Ni/Al,Os catalysts was studied subsequent to various thermal treatments carried

out within a TPRd apparatus C821. After temperature programming to 773 K

under Ar, the reduced and passivated catalyst showed no reduction profile

during TPRd. On the other hand, the catalyst reduced normally during TPRd

conducted either immediately after room temperature exposure of the reduced

catalyst to 0, or after thermal treatment at 393 K under Ar.

A model to account for these behaviors was proposed. Support species,

designated as AlxOy. were mobile, and could decorate the reducible components

on the catalyst surface. Thermal cycling to 773 K was sufficient for complete

decoration of all reducible Ni phases and accounted for the absence of a TPRd

profile. Interaction of the surface of the non-reducible catalyst with a

dilute 0, mixture at room temperature was necessary in order to displace the

support species that were bound to the Ni phases Cl7.821.

TPRd results from precipitated and impregnated catalysts demonstrate that

the properties of supported-metal catalysts are not uniquely determined by the

metal/support combination, but depend strongly on the preparation/pretreatment

Page 34: Editor's note

of these materials. Copreclpitated catalysts are made via a mixed solution of

Ni and Al precursors to finally yield a mixed oxide. In the preparation of

impregnated catalysts, the two metals are separate and only a limited amount of

mixed oxide forms; the amount depends on the extent of calcination. For

Ni/Al,O, catalysts, thermal treatment during catalyst activation can severely

alter both the chemisorptive and reduction properties of the active phase(s),

and the effects are sensitive to the preparation procedure.

Although support effects are apparent for impregnated as well as preclpl-

tated catalysts, little attention has been given to examining the effects of

different Al,O, supports on the properties of the finished catalyst. Prellm-

inary TPRd results for a series of Ni/Al,O, catalysts with different Al,Os

supports have been obtained C831. Ethane hydrogenolysls was used as a test

reaction and care was taken to ensure that all conditions during characteriza-

tion and performance studies were identical; the only variable in this system

was the source of the AlgO, support material. Table S summarizes the pertinent

physical properties of the supports.

TABLES

Physical Properties of Al,O, Supports

Surface

Al,03 area Particle Void volume Source material supporta (mx/g) size (pm) (ems/g)

A 200 0.012 Nonporous Aluminum Chloride B 75-90 o.oso Nonporous Aluminum Sulphate C 19s 22s 0.54 Not reported D 200 22s o-so Pseudoboehmite

aA: Degussa (Lot URVO969). B: Union Carbide (Lot IfB601) C: Englehard (Lot I Unknown), and D: Kaiser 202 (Lot ly40620)

The catalysts were prepared by incipient wetness from NitNO,), hydrate at

sufficient concentration to mount 6X by weight of metal. The dried catalysts

were activated and passivated by established procedures C841. Although the

TPRd profiles, Figure 6, were all obtained from Ni/Al,Oa catalysts, the

profiles are distinctly different.

Figure 7 shows that the hydrogenolysis activity of these catalysts are

also quite different. The rates are reported on a per gram total Ni basis.

Comparison of Figures 6 and 7 clearly demonstrates that TPRd results can

distinguish between differences in catalyst performance.

An examination C833 of the properties of the supports used In this study

revealed the crystallinity (determined by XRD) varied considerably from one

Page 35: Editor's note

35

--A -- B

2 0.006 ---c 5 -.._ D

w

-0.001 I I I I I

200 400 600 800

Temperature (K 1

Figure 6. TPRd spectra for 6% Ni catalysts supported on different aluminas. A: Dcpunsa (Lot #RV0969), B: Union Carbide (Lot #B601). C: Englehard (Lot # unknown), and II: Kaiser 201 (Lot W40620).

I

0

c .1. z 8 -I

u

5 -2 B E

0) ‘j -3

a

s ._ _ -4 ” 0

B -5

-6

I I I I I

I I I I 1 , I.4 1.6 I.8 2.0

I/Tx103 (K-I)

Figure 7. Hydrogenolysis activities for 6Z Ni catalysts supported on different aluminas. Support designation same as Figure 6.

Page 36: Editor's note

36

Al,Os to the next. The greater the crystallinity, the less Al,Os dissolution

during preparation, the smaller the H/B value determined by TPRd, and the

smaller the Ni uptake during wet impregnation.

Temperature-programmed reduction studies of catalytic precursors formed

subsequent to impregnation but prior to activation can be used to assess their

extent of interaction with the support. Data such as these can be useful in

developing first principles models of the adsorption process during lmpregna-

tlon. Recent TPRd studies of Ni/C catalysts have revealed some encouraging

results CESI. A low ash content activated carbon was subjected to controlled

oxidation by HNO,. The physical properties of the carbon, such as surface area

and pore structure, were not changed by these treatments, but the surface

properties were dramatically affected, as shown In Table 6. The point of

TABLE 6

Characterization of Carbon Samples

HNO, Concentration (M) BET area, (ms/gl Total Ash, (X) PZCP Neutralization

(meq/g) HCl NaOH NaHCO, Na,CO,

Ammonia Acidity (mmol/g)

None 0.2 0.4 1.0 2.0 1100 10% 1020 101s IISO 1.1 1 0.6 0.8 0.6 IO 7.8 6 s.s 3.5

0.738 0.599 0.314 0.04 0.276 0.478

0.127

2.35 3 4.24 5.18 6.41

0.163 0.143 0.887 1.709 0.351 0.503 0.334 0.117

apH corresponding to zero net surface charge

zero charge shifted to lower pH values as the extent of oxidation increased.

This was a direct consequence of increasing the surface acidity of the carbon,

as evident by an increase in their base neutralization uptakes. These effects

are due to the introduction of a higher density of oxygen functionalities due

to oxidation of the carbon. These oxygen functlonalitles serve as exchange

sites for cations. A model based on a single type of amphoteric hydroxyl

group has been applied to those data and results in the pH dependent density

of surface charge carriers. The aqueous phase speciatlon of the catalytic

precursor, Ni.(NO,),, can be determined (20). In conjunction with experi-

mental data on the pH dependence of Ni2+ exchange and a suitable complexation

reaction scheme, the thermodynamics of the exchange reaction can be obtained.

The conplexation model assumes the following mechanism

Page 37: Editor's note

37

Kc -COH + Ni*+ d x (CO)NI + x H+ (3)

where x. a stoichlometric coefficient, and Kc, the equilibrium constant, can

be evaluated by numerical procedures. As the extent of pre-oxidation of the

carbon increased, the value of Kc decreased or the absolute magnitude of pKc

increased. Figure 8 shows the relationship between pKc and the Ha consump-

tion at the peak temperature of the TPRd spectra of the precursors for this

series of M/C catalysts. As the extent of oxidation increased the H,

consumption at the peak temperature decreased. Results, such as shown in

Figure 8, can be interpreted as a decrease in the reducible species/support

Interaction with increasing oxidation of the support, and is consistent with

the variation in Kc. A combination of modeling the adsorption process and

TPRd studies as outlined above can be extremely useful in the development of a

scientific basis for catalyst preparation.

160. , , , , , , a , , I ’ I

140-

120-

lco-

80-

60 -

-6.4 -6 -5.6 -5.2 - 4.8 - 4.4 -4 -36

pKc

Figure 8. Hc consumption at peak temperature of TPRd spectra of dried Ni/C precursors versus surface stability constant. Lower absolute values of pKc reflect smaller magnitudes of Kc.

REACTION PATHWAYS

Background

Nearly ninety years ago, Ni was discovered to be an effective catalyst for

the synthesis of CH,, from CO and He C861. A quarter of a century later,

Fisher and Tropsch C871 proposed that carbon was an intermediate in the

Page 38: Editor's note

38

reaction pathway leading to CH, formation. At nearly the same time, Elvlns

and Nash [881 suggested the existence of an oxygen-containing ComPlex, CHnO,

as the Intermediate In the reaction. Despite the numerous studies that have

been conducted since the discovery of the catalytic properties of Ni in the

methanation reaction, the mechanism of the synthesis is still controversial-

Numerous reviews on the mechanism of the reaction have appeared [S9-921.

They can be generally divided lnto two schools of thought. One involves the

formation of an intermediate complex, CH,O, on the catalyst surface. The other

considers surface carbon, which is formed by dissociation of CO, as the active

Intermediate. A consensus that emerges from a study of the various proposed

mechanisms is that CO dissociation yields an active carbon intermediate, which

is hydrogenated to CH4. Questions that often arise are which of the reaction

steps is rate-determining, and does the rate-determining step change with

reaction conditions. The answer to these questions cannot be determined solely

on the basis of steady-state reaction kinetics. Transient techniques can be

used to obtain kinetic information about reaction mechanisms and possible

active intermediates for catalytic surface reactions.

One indication of the validity of this assertion was proposed In our earlier

review. We noted that CH, peak temperatures from TPRx data could be correlated

with turnover frequencies (TOF's) from steady-state experiments. Temperature-

programmed reaction has a number of advantages over steady-state kinetics, as

has been discussed. However, these advantages are of little value If the same

reaction processes are not measured by both techniques. For the methanation

reaction, in which the CO surface coverage is high under steady-state condl-

tions and at the start of TPRx, the correlation between results appear to be

good. Falconer et al. C931 saw a good correlation between the relative

activities as measured during steady-state reaction and as estimated from TPRx

data. The ratio of potassium-promoted rate to the unpromoted rate was deter-

mined for Wi on four supports (SiO,, Al,Os, TiO,, SiO,-AlzO,), and similar

ratios were obtained from both measurements.

Similarly, Bailey et al. C213 compared Ni/Al,O, catalysts with different

potassium loadings and found that when the peak temperature Increased during

TPRx (an indication of decreased activity), the activity decreased in steady-

state measurements. A larger increase in peak temperature was seen as the

potassium loading Increased, and the steady-state activity decreased more for

the higher potassium loading. On a IX Ni/Al,Oa catalyst, whose steady-state

activity was not decreased by the addition of potassium, the peak temperature

did not decrease.

Thus. qualitatively, TPRx and steady-state kinetic measurements show good

agreement. tftxe detailed comparisons have been made by Hieth and Schwarz C941

Page 39: Editor's note

39

for supported Ru catalysts. In a recent study of Ru/AleOs catalysts whose dis-

persions varied from 1 to SO%. Increases in methanatlon activity were accom-

panied by decreases in the TPRx peak temperatures C941. Figure 9 shows

the TOF at 523 K as a function of the CR4 peak temperatures. When TOF's at

various temperatures were plotted against the TPRx peak temperatures, the

shapes of the curves generated were similar to that shown in Figure 9,

suggesting that this correlation between TOF and TPRx peak temperature is

generally valid.

420 440 460 480 500 Peok Temperature ( K)

Figure 9. TOF versus TPRx peak temperature for methanation. A, Ru(NOO)(NO~).JA~~O~ by incipient wetness; v, Ru(NO)(NO,),/Al,Os by wet impregnation; 0, RuCl,/Al,O,; Cl Ru(II)/Al,Os.

We will now review some recent attempts using transient techniques to study

possible reaction pathways In the methanation reaction.

Page 40: Editor's note

40

Single-Cycle Transient Analysis

Temperature-Pronrad Reaction (TPRx). The application of temperature-

programmed reaction (TPRx) to methanation on Ni catalysts has yielded a wealth

of information about the reaction process and the properties of the catalysts.

tloreover, the methanation reactlon demonstrates the ability of TPRx to study

catalyst properties in general, and TPRx for methanation has been applied to a

number of other metals besides Nl C14,9S-1031. The properties of NI/SiO,

catalysts, as measured with TPRx, will be reviewed first. The effects of the

support on the catalyst propertles will then be considered. The presence of

two sites, the transfer processes observed between sites on Ni/Al,Os and

NUTlOg catalysts, and the use of isotopes to study sites will be discussed in

some detail because these studies demonstrate the ability of TPRx to separate

reaction sites on the surface and determine information not obtained from

steady-state kinetic measurements. The combination of Isotope labeling and

TPRx IS particularly powerful in these studies because the sites react at

significantly different rates. By labeling each site with a different isotope

of carbon monoxide, the interactions between sites can be clearly studied.

Though isotopes are particularly useful in TPRx studies, this advantage has

been fully exploited.

i. Nickel/silica catalysts. Rethane, the dominant product formed during

TPRx, forms over a narrow temperature range on Ni/SiO, catalysts. In addition

to cne. CaHo and CO, are observed in small quantities. Unreacted CO also

desorbs from NI/Si02 catalysts, and most of the CO desorbs before the methana-

tion rate is large. Because CH, does not readily readsorb on Ni catalysts,

the CH, peak Is relatively narrow and thus the temperature of its peak maximum

can be used to compare the effects of preparation, promoters, and supports on

the rate of methanation, as has been discussed.

Because CO adsorbs more strongly than does H, on Ni, the TPR experiment for

methanation is carried out by adsorbing CO, usually at room temperature or

belov. and then heating the catalyst In flowing H,. However, as will be

discussed, higher adsorption temperatures can also be used to occupy sites for

which adsorption is activated or to deposit carbon on the surface.

To study the effect of CO surface concentration on methanation rates, the

Initial CO coverage on the Ni is varied before the TPR experiment. This can

be done in several ways. Perhaps the best vay is to saturate the catalyst

with CO and then react off a fraction of the adsorbed CO by interrupted

reaction. That is, after adsorbing CO to saturation coverage, the catalyst is

heated, as in a normal TPRx experiment, until a given amount of CH, forms; the

furnace is then turned off and the catalyst rapidly cooled. A TPRx experiment

carried out after cooling thus corresponds to an initial coverage that is below

saturation. By using a higher interruption temperature, a lower initial CO

Page 41: Editor's note

41

coverage can be obtained. An alternate approach that has been successful for

methanation Is to adsorb CO, on the catalyst surface. Carbon dioxide adsorp-

tion on Ni Is activated, and the coverage obtained at 300 K is low. Higher

covet-ages can be obtained by using higher adsorption temperatures L1041.

Carbon dioxide dissociates on Ni and the subsequent TPRx spectrum is the same

as that observed for CO adsorption. Another approach, which is less useful,

is to adsorb CO to saturation coverage and then carry out an interrupted

desorptlon (heating in He) to remove some of the CO. A subsequent TPRx will

then hydrogenate the remaining CO. However, if too high a temperature is

required to obtain a low coverage of CO, carbon forms on the surface by

disproportionatlon of the adsorbed CO. Since some types of carbon

hydrogenate at faster rates than adsorbed CO, the subsequent TPRx will not

correspond to a lower-than-saturation coverage of CO. Instead, a broader CHe

peak results due to the different rates of hydrogenation of CO and carbon. For

a ten-fold variation In CO coverage on Ni/SiO,, the CH, peak temperature does

not change significantly ClO41. The rate of methanation thus appears to be

first order In CO coverage under TPRx conditions.

To study the effect of Ha concentration on the catalyst surface, the H,

partial pressure is varied. To do this, the surface Is saturated with CO and

TPRx is carried out at a given Ha partial pressure. A series of TPRx experi-

ments, each at a different H, partial pressure, is then used to determine the

effect of H, pressure. For a Ni/SiO, catalyst, the H, partial pressure was

varied from 0.2 atm to 2 atm ClOl. Pressures below atmospheric were obtained

by flowing a %/He mixture over the catalyst at I or 2 atm total pressure.

On a Ni/SiO, catalyst, as the H, partial pressure decreased, the CH, peak

shifted to higher temperature: i.e., the methanation rate decreased as the H,

pressure decreased (Figure 10). The initial rise of the CH, curve is propor-

tional to the methanation rate at high CO coverages, which corresponds to the

conditions present during steady-state methanation. Since the gas phase CO

pressure is quite low during TPRx, CO apparently does not Inhibit methanation,

and the methanation reaction under TPRx conditions Is found to be approximately

first order In Ha, except at higher pressures, where it appears to approach

zero order CiOl.

As described below, on NI/Al,O, and Ni/TiO, catalysts, more than one

reaction site exists for methanation. Since the single TPRx peak on NI/SiO,

could consist of two reaction sites with similar activities I8.9,141, Sen

and Falconer [El carried out isotope studies on two Ni/SiO, catalysts to

determine if two types of sites were present. This was done by adsorbing

IV,0 to saturation at 300 and 385 K and then reacting off sorae of the i2C0 by

interrupted TPRx. The catalysts were then exposed to 'X0 in an attempt to

Page 42: Editor's note

42

420 500 Temperature (K)

580

Figure 10. TPRx methane spectra on a S% Ni/SiO, catalyst. The CO was adsorbed to saturation at 300 K In pure H, flow at ambient pressure. TPRx was carried out in a H,/He mixture at ambient pressure. (a) 0.82, (b) 0.64, (c) 0.34, (d) 0.14.

Partial pressures of H, (atm):

adsorb 13CO on the sites vacated by the interrupted TPRx, and a TPRx was then

carried out. These studies showed that only one type of site appeared to be

present on NI/SiQ, catalysts since the 13CH4 and the 13CH4 peaks in TPRx had

the same peak temperature; that is, l3CO and 13CO reacted at the same rate to

produce methane. hot-cover, most of the methane was I3CH4. even when only a

small fraction of the original i3CO was reacted off the surface. The 13CO in

the gas phase readily displaces 1X0 on the Hi surface during the adsorptlon

procedure at room temperature, so that most of the surface was covered with

i3C0 before the start of the TPRx experiment.

Page 43: Editor's note

43

2. Hultiple reaction sites. Perhaps the best examples of the abilities of

TPRx are for Ni supported on Also, and TiO,. For all catalysts studied on these

supports, two distinct CH, peaks have been observed during TPRx. Though two

CH, peaks were first observed by Zag11 LIOSI, only more recently has a

satisfactory explanation been presented for the occurrence. Because the

Ni/Al,Os catalysts have been most extensively studied, they will be discussed

in some detail. A brief description will then be given of NUTlO, catalysts,

since they exhibit the same surface process as Ni/Al,Os catalysts. The

presence of multiple pathways for methanation on supported metals is a general

phenomena that has also been observed on Ru/Al,O, C141, Pt/Al,O, C963,

Pt/TiO, CYST, and Co/AlsOs Cl061 and thus the explanations presented here are

of interest for other supported metals.

3. Nickel/alumina catalysts. The TPRx results on Ni/Al,Os demonstrate the

care that needs to be taken in interpreting TPRx spectra. When Ni/Al,Os

catalysts are heated in Hs during TPRx experiments, several competing

processes occur. The CO can desorb, react to form CH,, or transfer to another

site on the surface. host of the CO eventually reacts to form CH4 on Ni

catalysts in atmospheric pressure Hsr so the desorption of unreacted CO is of

less importance for the present discussion. However, a significant amount of

the CO originally adsorbed on the Ni of NI/Al,O, catalysts can transfer to the

support during TPRx. The CO that transfers to the Also, support does not

react as rapidly to form CH4 and thus two CHo peaks are seen in the TPRx

spectrum.

Kester and Falconer Cl071 first studied this process in detail. For a 4.7X

NI/Al,Os catalyst prepared by impregnation to incipient wetness, they observed

two distinct CH, peaks (Figure 11) for CO adsorbed at 300 K. Because the peak

temperatures did not change wlth catalyst weight or H, flow rate (when delay

times between reactor and mass spectrometer were taken into account) CIOEI,

readsorption did not appear to be responsible for the high-temperature peak.

Since CHq 1s weakly adsorbed on Nl or A&O, surfaces, such readsorption would

not be expected. By studying the hydrogenation rate of carbon deposited by CO

disproportionation, Kester and Falconer showed that neither of the two peaks

was due to direct hydrogenation of carbon. Methane was the main product on

NIiA1,0,, as it ~8s for Ni/SiO, catalysts; the unreacted CO spectrum is shown

in Figure 11 but for clarity it will not be presented in most of the subsequent

figures.

A partial explanation of the surface process that occurred during TPRx was

presented by Kester and Falconer ClO71. The two CH4 peaks were attributed to

two distinct sites on the surface. The low temperature peak was designated as

site A and the high temperature peak as site B. The A site was assigned to CO

Page 44: Editor's note

44

I I I I I I I I I I

300 400 500 600 700

Temperature (K)

Figure ii. Hs at 300 K.

TPRx spectra for a 4.7% Ni/Al,Os catalyst. The CO was adsorbed in The rate is per gram of catalyst.

adsorbed on Ni particles and the B site to CO adsorbed on Ni atoms bonded to

Nl ions or aluminum ions. Hare recent studies CE,lO-14,96,1091 have shown

instead that the B site is due to CO adsorbed as a H-CO complex on the Also,

surface.

A key to understanding the process was the observation CiO73 that CO moved

between the two sites on the surface and that H, facilitated the conversion

from the A sites (Ni) to the B sites (Al,O,). The rate of the reverse conver-

sion was slow under TPRx conditions. Varying the heating rate was shown to

be an effective means to verify the movement between sites. For a higher

Page 45: Editor's note

45

heating rate, more CH,, was seen In peak A and less in peak B; this result

implied that the activation energy for methanation of CO on Ni was greater

than that for conversion of CO from Ni to Also,. Since higher heating rates

shift reactions to higher temperatures In TF'Rx [II, and higher temperatures

favor reactions with the higher activation energies, the higher heating rates

should favor methanation on Ni, as was observed ClO71. Indeed, as shown in

Figure 12 for a 5.1% Ni/Al,O, catalyst, at a sufficiently low heating rate

(0.07 K/s) transfer to AlgOa completely dominates methanation on Ni, and

essentially all the CO originally adsorbed on Nl transfers to the AlgO, Clil.

Thus, only one CHe peak is seen in Figure 12. This peak was clearly Identified

as CH, from the Also, site (B site) because it was at a higher temperature

than CH, from CO on Ni for 0.1 K/s heating rate, and lowering the heating rate

lowers the peak temperture for all peaks.

Temperature (K)

Figure 12. tlethane TPRx spectra for CO adsorbed at 300 K in H, on a 5.1% Nl/hl,Os catalyst. Heating rate was 0.07 K/s. The rate is per gram of catalyst.

Page 46: Editor's note

The effect of heating rate on the TPRx spectra for NI/Al,O, catalysts may

explain some discrepancies In the llterature. Kester el al. Cl101 and Bailey

Cl083 reported that at low Al loading (IX Ni), most of the CH, formed in the

high temperature peak, and the peak was at much higher temperatures than on a

5% NI/Al,O, catalyst. Huang and Schwarz 1623 observed for some low Nl load-

ing catalysts that only a high temperature peak existed; all the CO may have

been transferred from Ni to AlaO, for the low heating rate they used, though

additional phenomena may occur because of the small fraction of Ni that is

reduced.

The use of isotopes in combination with TPRx clarified the processes first

studied by Kester and Falconer. Glugla et al. Cl11 showed clearly that CG

adsorbed on Nl at 300 K in He and then during TPRx, more than half the CO that

adsorbed on a S.lX NI/Al,O, catalyst transferred from the Nl. The TPR spectrum

showed two CH, peaks, similar to those in Figure Ii. The combined area under

both CH, curves (plus the unreacted CO) corresponds to the amount of CO

adsorbed on the Ni surface (60 pmol/g cat). When TPR was interrupted after

20 pmol of CH,,/g cat had formed, no CO remained adsorbed on the Nl surface.

That is, of the 60 umol/g cat originally adsorbed, 20 umol/g cat reacted to

CH,, and the remaining 40 pmol/g cat transferred to the AlaOs surface. This

was clearly demonstrated by Isotope labeling experiments. After 20 pmol of

CH,/g cat formed in the low temperature CH, peak, heating was stopped and the

catalyst was cooled to 300 K. An additional 65 pmol of 13CO adsorbed on the

catalyst at 300 K. as shown by the resulting TPRx in Figure 13. As the catalyst

was heated during the TPRx following adsorption of IsCO. more than half of the

WI0 transferred to AlaO, and appeared in the high temperature peak. Thus,

the total amount of CH, in the high temperature peak was significantly larger

than in the original TPRx.

These results show that the transfer process Is activated, and thus CO can

be adsorbed to saturation on the AlgO, surface by adsorption at 385 K instead

of 300 K. Hydrogen was required to saturate the AlaO surface, however, since

the amount of CO adsorbed in He at 385 K was the same as that adsorbed at

300 K in He. When CO was adsorbed at 385 K in He, much more CO was taken up,

and a comparison of the amount of CH, formed for CO adsorbed at 300 and 385 K

in H, is presented In Figure 14. To verify the presence of two sites, the

surface was saturated at 385 K with IZCO, an Interrupted TPRx was used to

remove iaCO from the Ni surface, and 13CO was then adsorbed on the vacant

sites. The resulting TPRx showed that the A120, was essentially saturated

since not much laC0 transferred during this TPRx. Most of the 13CO reacted

to form the low temperature CH, peak and most of the laC0 reacted to form the

high temperature peak.

Page 47: Editor's note

47

‘bs b ij 0.6 -

3. 9 - e e 2 - z z

0.3 -

I 0 350 450 550 650

Temperature(K)

Figure 13. Methane (WIH, and IsCH,) TPRx spectra for sequential adsorption of WI0 and 13120 in H, on a S.lX Ni/Al,O, catalyst. The surface was saturated with 12120 at 300 K and the low temperature CH, was removed before 13C0 adsorption at 300 K. The rate is per gram of catalyst.

I

9-

9 Y B 6-

s \ ’ 350 450 550 650

2 - L Ql is _ I I c I I I I

3- /

I I I I

I 0

350 450 550 650

Temperature (K)

Figure 14. Methane TPRx spectra for CO adsorbed at 300 and 385 K in H, on a S.lZ Nt/A1,03 catalyst. The insert shows, on an expanded y axis (and reduced X-axis), the CH, spectra for CO adsorption at 300 K. The rate is per gram of catalyst.

Page 48: Editor's note

48

The presence of two distinct peaks was a general phenomenon that was alSO

observed for a 192 Ni/Al,O, catalyst C81. The value of isotope labeling was

demonstrated for the 192 NI/Al,O, catalyst because the rates of methanation of

the two types of adsorbed CO were similar to each other and isotopes were

required to clearly demonstrate that two peaks existed (Figure 1s). The number

of A1203 sites was the same on the 19% NI/Al,O, catalyst and the S-12 Ni/AlxO,

catalyst. Moreover, the CO adsorbed on the Ni surface readily exchanged with

CO in the gas phase, but CO on the support did not exchange at a significant

rate at room temperature. This was the same type of exchange observed for CO

adsorbed on the Ni surface of Nf/Si02 catalysts and was a further demonstration

0 1 I I

300 400 500 600

Temperature (K)

0

Figure 15. Methane (IZCH, and IsCH,) TPKx spectra from 19% Ni/Al,03. The i2C0 was adsorbed in 4 at 385 K and the catalyst was then heated to 485 K to remove some 1X0. The catalyst was then cooled in H, and 13CO was adsorbed at 300 K. Note that the mass 17 signal above SO0 K is mostly due to cracking of the Hz0 product.

Page 49: Editor's note

that the two sites had dlffetent catalytic properties and that the low tenper-

atute CH, peak resulted from hydrogenation of CO adsorbed on the Ni surface.

Studies on the 19% NI/Al,O, catalyst also demonstrate the ability of isotope8

49

to separate peaks that have similar activities and might not be detected

otherwise. Similar separation of overlapping peaks has been seen for other N1

catalysts 11111.

One of the reasons for using a Ni/Al,O, catalyst with 19% N1 ia that at low

Ni loadings, Ni IS not readily reduced on AlaOs C781. but the degree of

reduction increases with Ni loading and with reduction temperature CSo.34,112~

1137. .~. Thus, the 19% Ni/Al,Oa catalyst was also reduced at 97s K to obtain

almost complete reduction Cll41. Two distinct CH4 peaks were still present

after 975 K reduction (Figure 16), and the amount of CH, formed from CO

adsorbed on N1 was significantly smaller than that seen for reduction at 775 K.

At 975 K, sintetrng reduced the Ni surface area, and TPRx is effective at

observing these changes in catalyst structure with pretreatment conditions.

The other observation of interest is that the rate of methanation of the CC

adsorbed on AlaO, decreased as the Ni surface area decreased: the 19% Ni/Al,Os

catalyst reduced at 975 K behaved like a lower loading catalyst reduced at

775 K.

20

15

%H,xlO

300 400 500 600 7

Temperature (K)

Figure 16. bethane (IaCHe and *%Ztl,) TPRx spectra from 19% Ni/AlZ03 after reduction at 975 K for 3 h. The WZO was adsorbed at 385 K and the catalyst was then exposed to i3CO at 300 K. The laCH,, curve Is displaced vertically.

Page 50: Editor's note

50

A series of Ni/Al,O, catalysts with a range of Ni loadings C9.1101 also

exhibited two CH, peaks during TPRx, and the relative amounts of transfer that

occurred during TPRx were a function of Ni loading. The rate of methanation of

the CO adsorbed on the AlgO, was also a function of the Ni loading, though a

completely satisfactory explanation for this occurrence has not been presented.

At a Ni loading of 0.74 wt Z, the CH,, peak tenperature for hydrogenation of

the CO adsorbed on AleO, was at 715 K. For 19% Ni loading, the CH,, peak

temperature for hydrogenation of CO adsorbed on AlaOs was at 480 K- Figure 17

shows that peak temperature monotonically decreased as the Ni loading increased

c91. As observed in separate studies for a number of Ni/Al,Os catalysts pre-

pared by both incipient wetness C9,16,21,61,1101 and by wet impregnation CS9,

601, two CH, peaks occur on almost all Ni/Al,O, catalysts. At low weight load-

ing, a third peak was also observed, but it was not studied ClO83.

60

420 500 580 Temperature (K)

Figure 11. Methane spectra for TPRx of CO adsorbed on a S.SZ The CO was adsorbed at 300 K in ambient pressure H, flow and out in s/He mixtures. Hydrogen partial pressure (atm): (a) (c) 0.81. (d) 0.71, (e) O.SO. (f) 0.34. (g) 0.21.

Ni/Al,Os catalyst. TPRx were carried 2, (b) 1.75.

Page 51: Editor's note

51

Running a series of TPRx experiments on Ni/S102 catalysts at different H,

partial pressures showed that CO hydrogenation on Ni is approximately first

order in H, pressure during TPRx. Similar experiments are more valuable for

NI/Al,O, catalysts. As higher Ii, pressures were used, more CO transferred onto

the Al,O, surface during TPRx and the CH, peaks became more separated In

temperature CIOI. In contrast, as shown in Figure 18, at H, pressures below

60

50

40 g

Y e z 30 z- u u”

20

10

0

T

I I I I I

400 500 600 700

Temperoture (K)

Figure 18. tlethanc spectra from TPRx on Ni/Al,Os catalysts for CO adsorbed In He at 300 K. Nickel weight loadtngs are Indicated.

one atm, the peaks started to overlap and less CO transferred to Also,. That

is. hydrogenation of CO adsorbed on Ni has a positive order In H, pressure,

but hydrogenatfon of CO adsorbed on AlsOa has a negative order in H, pressure.

This result further confirm that two distinct surface precesses are occurring

on Ni/Al,O, catalysts during TPRx. This type of experiment also points out

that the choice of gas phase partial pressure during TPRx can be critical in

distinguishing surface processes. Peaks that are not separated at one pressure

might be readily distinguished at another.

Page 52: Editor's note

52

Kester and Falconer Cl073 first reported that CO also moved back from the

Also, to the Ni surface under some conditions. In attempts to vary the inttlal

surface coverage of CO before a TPRx experiment, they carried out a series of

interupted TPKx experiments. When a TPkx experiment was interupted after the

low-temperature CH, peak had formed, a subsequent TPRx showed only the high

temperature peak, as expected. However, if the carrier gas was switched to He

and the catalyst held in He for 10 min at SO0 K after the low temperature CH4

peak had formed, the results were different. The subsequent TPRx, obtained

after cooling in He to 300 K, showed both CH, peaks present. That is, holding

the catalyst in He for 10 min apparently caused CO to transfer back from the

AlaOs to the Ni. The ability of TPkx to directly observe the amounts on each

site allows this type of study.

Isotope labeling experiments were used to verify this reverse transfer

process [Ill. A 5.1% NI/Al,O, catalyst was exposed to *sCO at 300 K and an

interrupted TPRx was carried out to 460 K to remove isC0 from the Ni site. At

460 K the carrier gas was then switched to He and the catalyst was cooled In

He. The flow was switched back to H, at 300 K, iaC0 was adsorbed to satur-

ation, and a standard TPRx was carried out. The resulting laCH4 and t3cH4

spectra, shown in Figure 19, had the same shape and were quite different from

Figure 19. Methane ('aCH4 and IaCH,) TPRx spectra for sequential adsorption of 12C0 and lsCO in He on a 5.1X Ni/Al,O, catalyst. The surface was saturated with 12C0 at 300 K and the low temperature CH, peak was reacted away by interrupted TPRx to 460 K. The catalyst was then held at 460 K and the carrier gas changed from H, to He before cooling. H

At 300 K. the flov was switched to i3C0 was adsorbed, and a TPRx vas carried out.

Salyst. The rate is per gram of

Page 53: Editor's note

53

the interrupted TPRx with isotope labeling experiments that were discussed

previously (Figure 13). The total amount of tgCH4 plus WZH,, was approximately

the same as that seen for 1%X adsorption alone at 300 K in a standard TPKx.

This demonstrates that most of the CO transferred back to the Ni surface.

At the start of the final TPRx. the Ni surface was covered by igC0 and '3co.

and thus both isCO and 1sCO transferred to the Al,O, during TPRx. The amount

of 17-CH4 was small because lsC0 transferred back to the Ni surface during the

IO min In He and then exchanged vith gas phase WZO.

bore recent experiments Cl151 also show that this reverse transfer occurs

In He at 400 K. In those studies, CO was adsorbed at 300 K and the catalyst

was held in Hs at 385 K in order to transfer all the CO to the Al,O:,. as shown

by a subsequent TPRx. In repeat experiments, after transferring the CO to

AlgO, by holding in H, at 385 K, the CO was transferred back to the Ni by

holding in He. Some CO vas lost to the gas phase, but holding the catalyst In

He, at temperature from 400-425 K, allows the CO to transfer back to the Nl,

as evidenced by the TPKx spectra. This ability of TPRx to determine the

location of the adsorbed surface species after a series of treatments allows

kinetic studies of the transfer processes to be made.

4. tlethoxv species on nickel/alumina catalysts. A combination of TPRx and

TPD experiments on Ni/Al,O, and TPRx and IR studies on Pt/Al,O, C961 have

clearly demonstrated that the high temperature CH4 peak in the TPRx experiments

on Ni/Al,O, is due to CO adsorbed on the support as a methoxy species. Lu

et al. Cl161 observed two CH, peaks on a 201: NI/Al,O, catalyst, and by combin-

ing IR studies with TPRx, they also concluded that the low temperature CHH4

peak (at 448 K for a heating rate of 3.7 K/min) was due to hydrogenation of CO

on Ni. Their CH, peak at 500 K was assigned to hydrogenation of a formate

species that formed by splllover from Ni to Al,O,. Indeed, a formate species

has been observed by IR on a number of catalysts vith A1203 supports Cil7-1191,

and Lu et al. Cl161 and hlradoatos et al. Cl203 observed a for-mate species on

their NI/Al,O, catalyst. However, as a result of more recent experiments [II,

13,961, the formate species was concluded to be a minor part of the reaction

sequence and a methoxy species on the Al,O, was concluded to account for the

high temperature CH4 peak.

In the TPD experiments Cl31 described later, CO was adsorbed at 385 KS

using H, as the carrier gas during adsorption. However, instead of being

heated in H,, the catalyst was heated in He and desorption of CO, H2, CO,.

and CR, was observed. The CO and H, desorbed simultaneously when the surface

was saturated at 385 K, and the H/CO ratio was approximately three f131.

This approximate ratio led to the initial conjecture that a CH,O species was

on the Al,O, surface.

Robbins and haruchl-Soos C961 also observed two CH, peaks in TPRx studies

Page 54: Editor's note

54

on Pt/Al,Os, but because Pt is a poor methanation catalyst, the high temper-

ature CH, peak was from CO adsorbed on Pt and the low temperature CH, Peak

resulted from hydrogenation of the CHoO species that was adsorbed on Al,Os.

direct relation between the IR and TPR results, obtained by interrupted TPRx,

alloved them to make this identification directly. The strong similarity

between the studies on Ni/Al,O, and Pt/AlaOo catalysts argues that a CHoO

species is also present on the AlaO, surface of Ni/Al,O, catalysts.

The CH,O species was concluded to be on the Al,O, surface rather than on

the Ni surface, Ni atoms dispersed on the AleO,, or a NiAl,O, for a number of

reasons:

1) A CHoO species was identified on Pt/Al,O, by IR C961 and Ru/Al,Os by

TPD C141, though these metals have not been shown to form an aluminate.

2) The number of CHaO species was 3 times the number of Ru atoms in

Ru/Al,O, Cl41.

31 The number of CH,O species per gram of AlaO, was the same at saturation

for S.l% Ni/Al,O,, 197. Ni/Al,O,, and IX Ru/AlaO,.

4) TPD experiments on single crystal Ni in WV indicate that a CHaO species

is not stable on Ni surfaces above 300 K C121,1223.

S) When a 19X Ni/Al,O, catalyst C81 was reduced at 975 K. the Ni surface

area decreased due to sintering, and the catalyst was almost completely

reduced, but the number of CHsO species vas unchanged at saturation. That is,

the unreduced Ni concentration changed and the Ni surface area also changed,

but the CH,O concentration did not.

61 On Ni/SiO, catalysts, CH,O species are not observed, even for extended

exposure of CO in H, at 385 K C81.

Perhaps the most convincing result, however, is that the concentration of

CHaO species was increased (per gram Nil by intimately mixing Ni/Al,O, and

A&O, in a physical mixture (crushed, wetted and crushed together, and then

dried), and the resulting concentration CHoO was the same per gram of total

AlaOs ClO91. tlorcover, a high temperature CH4 peak was also formed during TPRx

of a physical mixture (prepared in the same way) of Ni/SiO, + Al,O,. On

Ni/SiO, alone, Only a single CH, peak formed. Since the AlaO, alone does not

form a CHaO species from CO and s, gas phase or surface transport must be

involved. Since the spillover process does not occur from a Ni/SiOa bed on top

of a AlaOs bed, the spillover does not appear to be by gas phase transfer.

However, the spillover species can form on Ni/Si02 and transfer to AlaO,, even

though a CHaO was not seen on SiO, under these conditions. The species that

spills over from Ni to A&O, could be CHsO. or it could be CO, which then

reacts with spilled-over Hz_ Hydrogen adsorption is activated on Ni/Al,Oo and

the activated formation of CHaO appears related to this activated adsorption.

Page 55: Editor's note

55

The decomposition of CH,O appears to limit the rate of formation of the CO

and H, peaks during TPD, since they are seen simultaneously. Since the CO and

H, peaks are at similar temperatures in TPD to the high temperature CH, peak in

TPRx, decomposition of the CH,O appears to limit methanatlon. When TPRx

experiments were run at higher H, pressure, the high temperature CH, peak moved

to higher temperature: the formatlon of CH, from CH,O has a negative order in

H, CIOI. Thus. the decomposition of CH,O may be limited by the availability of

sites on the Ni for decomposition. However, direct hydrogenation on Al,O,,

with a negative order in $ pressure, cannot be ruled out.

S. Titration of surface spe&les on nickel/alumina catalvats. Steady-state

hydrogenation of CO Involves adsorption and dissociation of CO molecules and

hydrogenation of the products. TPRx experiments subsequent to steady-state

reaction can be used to study reaction pathways that result from dlfferent

reactive centers on the catalyst surface. Huang et al. showed C16,17,59-621

that the properties of Ni/Al,O, catalysts are strongly dependent on their

method of preparation. These results were tested by a titration procedure

using TPRx data from a series of catalysts prepared by incipient wetness; the

weight loading ranged from 0.9-6.231. These data were obtained subsequent to

an exposure of each catalyst to a H&O (3:l) mixture under reaction conditions.

After the last measurement for steady-state reaction was made, each catalyst

was flushed with He at 553 K for S min and cooled to room temperature in a He

flow. An amount of H, necessary to achieve saturation was then pulsed onto the

catalyst surface, providing a limited source of H, for removal of other species

that remained on the surface. The catalysts were heated in a He flow to 753 K

and the amounts of CO, CO,, CH,, and H, evolved were quantitatively measured.

Several of these sequences were performed and a final TPSR experiment was then

employed to determine the amount of carbon residue that could not be removed by

the TPD procedures used. Representative .spectra of each of the species evolved

during the TF'Rx-titration experiment are shown in Figure 20. A comparison of

400 500 600 700 773- Temperature(K)

I I I I I (bl

560K n

400 500 600 700 i73- Temperature(K)

Page 56: Editor's note

/ ! I I I J c

400 500 603 xx) ?i3- Temperoturel K 1

- I I I I

400 500 603 700 ?73- Temperoture (K)

I I I I I (e)

2-

0 400 500 600 700 773-

Temperoture 1 K)

Figure 20. Product spectra for: (a) &, (b) CH,, (c) CO, and (d) CO, generation for a 6.23% NI/Al,O, catalyst during first titration experiment, and (e) CH, TPSR after titration experiments. Note that the temperature-ramp was stopped at 713 K and held at 773 K for 1 h. The rate is in pmol/g Ni*s.

the desorption rates of individual products at comparable temperatures shows

that an excess amount of H, desorbed after the methane peak had ceased. No

product signals were observed during the second titration. The data were

analyzed by considering the reactive pathways for carbon/oxygen removal under

the non steady-state conditions of these experiments. The objective was to

evaluate the average concentrations of reactive intermediates

abundances to TOF's representative of each of the catalysts.

and relate their

The reactive pathways durlng the titration experiments are shown in Scheme 1.

CO(s) --+ CO(g)

CO(S) -+ C(s) + O(s)

(4)

(S)

Page 57: Editor's note

57

C(s) + O(s) + CO(s) (6)

CO(s) + O(s) -+ co,(g) (7)

CO(s) + (2+x) H(s) + CHx(s) + Ha0 (8)

2 CO(s) --+ C(s) + con(g) (9)

CHx(sl + (4-x) H(s) --, CH4(g) (IO)

Scheme 1

It was proposed that the CH, produced during the first titration is due to

hydrogenation of CHx in the carbon pool depicted in Scheme 1. This assertion

was supported by the facts that excess H, desorbed after CH, generation had

ceased, and TPRx evidence that the hydrogenation of surface carbon inter-

mediates is faster and reacts at lower temperature than surface adsorbed CO(s)

C123-1251. On the other hand, the amount of CO, CO,. and CH, produced

during TPSR was proposed to be due to the CO(s) in the carbon pool. In accor-

dance with their assumptions, mass balances for the amount of CH, and CO(s)

were used to estimate the concentration of these species for each catalyst

after steady-state reaction.

The variations in the TOF with the average number of CH, and CO(s) species

in the reaction pool are shown in Figure 21. The group of catalysts is

divided into two sets regardless of the basis for the abscissa, and the trends

within each set were attributed to differences in reaction mechanisms over

their catalysts.

10-l - :.l-woTFRxPeaks ’ (b) - n SingleTPRxFbks \

‘;i 4 c f c

\

ci f al

; IO”:

b

\ m

Figure 21. Plot of TOF versus average number of: (a) Cl&,(s) and (b) CO(s) species in the reaction pool for a series of Ni/Al,O, catalysts.

Page 58: Editor's note

58

The high weight loading catalysts have two reaction centers, as shown by

TPRx experiments. These catalysts also are more poorly dispersed than their

lover weight loading counterparts that show only one reactive center in TPRx

experiments. Huang and Schvarz El71 proposed that for the high weight loading

catalysts the dominant reaction center under steady-state conditions was a

Ni-like species similar to those found on Ni/SiO,. Evidence from the

literature C1261, which indicates the interaction between a CHx species and an

adsorbed H atom is the slow step in the mechanism, was used by the authors to

argue that, if this were the case, the methanation rate should Increase with

increasing CHx on the surface. Indeed, this is what is found, and shovn in

Figure 21. Low weight loading catalysts were analyzed In a similar fashion.

Here, however, it was postulated that the activity of this reaction center was

controlled by its ability to dissociate CO(s). The results for the low weight

loading catalysts shown in Figure 21 support, but are too limited to confirm,

the hypothesis. The methanation rate does increase with increasing CO(s) for

the two catalysts studied. Kore convincing, however, is the fact that the data

shown in Figure 21 also demonstrate that the TOF decreases with lncrcasing

CO(s) for the group of catalysts designated as high weight loading. These

results are consistant with the majority of mechanistic studies, which conclude

that CO inhibits the TOF on high weight loading catalysts.

6. Nickel/titania catalysts. Oxdogan et al. Cl271 reported TPRx on NI/TIO,

catalysts and Wilson I201 carried out more extensive studies of NI/TiOe

catalysts using TPRx. As shown in Figure 22, except at the lowest coverages,

the peak temperature for CH, formation was approximately independent of

Initial CO coverage, which was varied by Interrupted TPRx. A more detailed

analysis of the TPRx spectra shoved, however, that CH,, formed at a slightly

faster rate, at the lover temperatures, as the Interruption temperature was

Increased, and thus the reaction process appears more complicated than that

observed on Ni/S102 catalysts. Wilson C201 also carried out one of the first

studies of TPRx to look at the effect of H, pressure during TPRx. A 10X Hz/He

mixture was used and the CH, peak was at higher temperatures than seen for

TPRx in pure H,. As seen for NI/SiO,, the methanation rate is positive order

In H, pressure. The consumption of H, was also measured in this study by

recording the H, signal In the carrier gas.

On Ni/TiO,, C,H, was formed in small quantities during TPRx, but the amount

of CzHe was larger than seen on other catalysts. As shown in Figure 23, CzHe

formed at a lover temperature than CH,, and C,H, formation also appeared to be

first order in CO Initial coverage.

Page 59: Editor's note

59

Temperature (K)

Figure 22. Methane product during TPRx following CO adsorption at 300 K on a 102 NI/Ti02 catalyst and interrupted TPRx. Interrupted temperatures of: (a) 305 K, (b) 395 K, (c) 407 K, (d) 416 K, (e) 425 K, (f) 430 K, (g) 434 K- The rate is per gram of catalyst.

Page 60: Editor's note

60

I I I I I 1

II.1 ,d

Temperature (K)

)O

Figure 23. Ethane spectra during TPRx following CO adsorption in H on a 10% Ni/TtSi2 catalyst and interrupted TPRx to successively higher tempefature. rate is per gram of catalyst.

The

Page 61: Editor's note

61

The multiple CH, peaks observed during TPR are not unique to Ni/Al,O,, and

similar processes have been observed on Ni/TiO, ClS,20,1281. During the

TPRx following CO adsorption at 300 K, CH, forma with a peak temperture of

455 K, which is significantly lower than that for NI/SIO,; the specific

activity of Ni/TiOg catalysts Is higher than that of Ni/SiOg catalysts in

steady-state experiments also C933. In addition, as shown in Figure 22, a

small high temperature peak was seen 115,20.127,1281. This high temperature

peak results from hydrogenation of CO adsorbed as a t&O species on the TlOe

surface. The transfer process to form the CHsO species is slower on Ni/TiOe

than on NI/Al,O,, but CO adsorption at 385 K successfully occupied the TlOe

surface, as shown in Figure 24 for a series of exposure times. Even after

3 h of CO exposure (one CO pulse every 30 8) at 385 K, the CH,O species was

still forming at a constant rate; the coverage of CHaO on TiOx was far from

saturation. Because of the slower transfer rate, separation by isot&

labeling was distinct. As shown in Figure 25, the CO adsorbed on the two sites

did not intermlx significantly during TPRx and thus the lxCH4 and t3cH,, peaks

are well separated.

Though the transfer process is slower on NI/TIO, than on Ni/Al,Os.

isothermal experiments at 385 K Cl281 showed that the CO adsorbed on Ni trans-

ferred onto the TiOe in an activated process. The reverse transfer occurred

at higher temperatures and thus the behavior of NUTlO, catalysts is similar

to that of Ni/Al,Os catalysts. The TPD experiments described below also

showed that H, and CO desorbed simultaneously from the Ni/TiO, catalyst and

they indicated the presence of a methoxy species on the TiO, surface. These

results on NI/T102 show that the transfer rates between metal and support are

dependent on the support as well as the metal.

Running a series of TPRx experiments at different H, pressure shows clearly

the distinct nature of the two methanatlon sites on Ni/TiOe catalysts CIOI.

Similar to the results on Ni/Al,O,, the two peaks were clearly separated at l-2

atm H, pressure, but at low 9 partial pressures the two peaks coalesced into

one peak. Hydrogenation of CO on Ni was approximately first order in Hg, but

the hydrogenation of the CHsO species that was adsorbed on TlOx was negative

order in He; this is the same behavior observed on Ni/Al,Os catalysts, and

indicates that the same processes occur on both catalysts.

Tenwerature-Programmed Surface Reaction (TPSR). Since most of the proposed

mechanisma for CO methanation on Ni conclude that CO first dissociates to form

carbon and oxygen on the surface, studying the rate of carbon hydrogenation

provides a means to understand one of the steps in the reaction sequence.

Temperature-programmed surface reaction is particularly suited for studying

this step, and a number of studies have deposited carbon and then hydrogenated

Page 62: Editor's note

62

lB1

15

6

0

300 400 500 600 700

TEMPERATURE (K)

Figure 24. Methane spectra from TPRx of CO adsorbed In H, on a 5% NUTlO catalyst. Adsorption temperature, number of CO pulses: (a) 300 K, 40 pulses; (b) 375 K. 40 pulses; 360 pulses.

(c) 385 K, 120 pulses; (d) 385 K, 240 pulses; (e) 38s K,

Page 63: Editor's note

63

TEMPERATURE (K)

Figure 25. Methane ("CH, on a SZ Ni/TiO, catalyst.

and IaCH,) spectra for interrupted TPUx experiment The leC0 was adsorbed at 333 K in H,, the catalyst

was heated to 490 K to remove some 1sCO (as IQH,), and IWO was adsorbed at 300 K in Hs. The catalyst was then heated in H,.

It during heating in H, C108.123,12?,1291. In most cases, the carbon was

deposited by exposing the catalyst to CO in He flow at elevated temperatures.

As carbon was deposited, CO, formed and unrcacted CO desorbed if the catalyst

was held In He flow for sufficient time. The catalyst was then cooled and the

Page 64: Editor's note

64

carbon reacted during programmed heating in Hs flow. The resulting CH4 peak

was at a lower temperature than that from CO hydrogenation, as shown in

Figure 26. That is, the rate of methanation of carbon on Ni/SiO, is faster

than the rate of CO methanation under these conditions. The rate of carbon

hydrogenation has been studied with TPRx on other metals with similar results

C98,101.102,1301.

050

h

‘” 0

\

5

5 0.25 fu - z a

I .

I . I . I . I . I . I .

- I . I

.a

0 300 400 500 600

Temperature (K)

Figure 26. TPSR on a 162 Ni/SiO, catalyst. Curve (a) is the CH, signal observed during TPSR following carbon deposition by CO disproportionation at 573 K. The corresponding CzHe signal is also shown. Curve (b) Is the contribution to the CH,, signal from coadsorbed CO. The difference between curves (a) and (b) is shown as a dashed line, which corresponds to carbon hydrogenation.

Perhaps the most extensive studies of carbon hydrogenation on Ni catalysts

were carried out by McCarty et al. C123.1291 on Ni/Al,Oa catalysts. McCarty

and Wise deposited carbon by exposure to CO (and to C,H,) at SO0 to 625 K and

they observed four forms of carbon during programmed heating in Hs. The most

Page 65: Editor's note

65

reactive carbon was chemisorbed carbon atoms (a-carbon), and they concluded

that a-carbon was a likely intermediate in forming CH, from CO and H,. They

also detected a more reactive a' state of carbon at low coverages, and some

CH, was produced at 300 K following carbon deposition. Bulk nickel carbide

and amorphous carbon (g-carbon) were also seen. The least reactive carbon was

concluded to be crystalline elemental carbon. tfccarty and Wise observed the

slow conversion of a- and g-carbon to graphite.

The rate of hydrogenation of the a-carbon was studied by Ozdogan et al.

Cl273 as a function of support. The rate of CO hydrogenation during TPRx was

found to be dependent on the support, with Ni/TiO, being the most active of

the Ni catalysts studied and NI/SiO,-AlgO, being the least active. On the

same four Ni catalysts (TiOg, A&O,, Slog, SiO,-Al,Os supports) the rate of

carbon hydrogenation, as measured by TPSR at low carbon coverages, was

essentially independent of the support. On all the catalysts, the rate of

carbon hydrogenation was greater than the rate of CO hydrogenation, as shown

in Figure 26. For both CO and carbon hydrogenation, C,H, was also observed as

an hydrogenation product, and it formed at a significantly lower temperature

than CH,. Though the amounts of C,H, were small (l-27. of the CH, amount),

C,H, was readily detected with the mass spectrometer, as shown in Figure 26.

McCarty et al. Cl313 used TPSR to further study carbon hydrogenation by

both H, and HsO on NI/Al,O, catalysts. They combined these studies with x-ray

diffraction and electron microscopy in order to relate the peaks seen in TPRx

to the type of carbon. Seven reactive carbon states were seen, depending on

the deposition temperature; carbon was deposited by either CO or C,H, decompo-

sition. They found a number of less-reactive forms of carbon that formed by

exposure to C,H, above 713 K. One of the Interesting aspects of their study,

vhich has wider implications for TPRx in general, is that above 1100 K the

rate of methanatlon by H, decreased with increasing temperature, but the rate

increased again as the catalyst was cooled. They concluded that the concen-

tration of CH* in the effluent gas vas limited by equilibrium and thus much of

the carbon that was only hydrogenated at,high temperatures (E carbon) was not

gasified for a typical TPSR experiment. tlore complete gasification of this

carbon was obtained by using steam instead of H,. since the steam-carbon

reaction was not equilibrium limited under these conditions.

Temperature-Programmed Desorption (TPD).

1. Hydrogen TPD. Hydrogen desorptlon from supported Ni/SiO, was studied in

detail by Lee and Schwarz Cl91 in one of the earliest studies to use mass

spectrometer detection during TPD from supported metal catalysts. They also

studied the adsorption process by pulse chemlsorptlon, but only the desorption

of H, will be discussed here. In contrast to the TPRx experiments discussed

Page 66: Editor's note

66

above, in which CH, did not readsorb, H, readsorbs and desorbs over a wide

temperature range. The initial coverage was varied by increasing the H,

exposure. The peak temperature shifted to higher temperatures as the Initial

coverage decreased because of readsorption. The TPD spectra were found to

depend on the weight of catalyst in the reactor and this was attributed to

readsorptlon; interparticle diffusion effects were concluded to be unimportant

under the operating conditions. The effect of readsorption was confirmed by

analyzing the curves by the desorption rate isotherm method [il. The desorp-

tion energy was found to be a function of coverage and decreased from 89 to

SS kJ/mol as the coverage increased.

Temperature-programmed desorption of H, from supported Ni catalysts was

extended to other supports by Weatherbee and Bartholomew Cl32.1. The heats of

adsorption were found to be a function of support tSiO,, Al,Os, TiO,). An

unsupported Ni sample was used for comparison, and its desorption spectrum was

similar to that of their Ni/SlO, catalyst, whose TPD spectra were in agreement

with those of Lee and Schwarz Cl93. The H, desorption peaks for Ni/Al,O, and

NI/TIO, were at higher temperatures than those from Ni/SiO,. Moreover, as the

adsorption temperature Increased for Ni/Al,O,, the amount adsorbed increased;

some H, adsorption on Ni/Al,O, was activated.

Hydrogen desorptlon on NI/Al,O, was studied by Stockwell et al. Cl331 for a

10X Ni/Al,O, catalyst. They used a high resolution mass spectrometer as a

detector, and thus were able to eliminate background due to Hea+, which

resulted from their He carrier gas. In addition, after H.. adsorption at 300 K

or above, they cooled their sample to 77 K before carrying out TPD. The H,

desorbod in three distinct peaks at 167, 280, and 446 K, and the amount of &

that adsorbed at 300 K increased with exposure time even after one hour In

atmospheric pressure He. Moreover, the amount adsorbed at 423 K. with cooling

to 300 K, was a rather slow process, and thus the strong chemisorption they

saw was an activated process Cl331. The cause of activated adsorption was

suggested to be decoration of the Ni by Al,O,, which dissolved in the impreg-

nating solution and accumulated on the Ni surface during preparation.

Bailey et al. C211 also observed that H, desorbed over a wide temperature

range from supported N1 catalysts and they obtained good agreement in the

amounts adsorbed when measured by TPD and by static chemisorption if the

catalyst was cooled in H,_ from 773 K before the TPD. Adsorption at 300 K by

injecting pulses of H, through the catalyst bed was not effective in saturating

the surface Cl081 and less than half as much H, was adsorbed on Ni/Al,O, by

this method before the rate of adsorption dropped significantly. As a result,

the TPD following pulse adsorption at 300 K was stgnificantly different from

that following cooling in Ha.

Page 67: Editor's note

67

The effect of reduction temperature on the TPD spectra of H, from Ni/TiO,

was also studied by Weatherbee and Bartholomew C1321. In agreement with

previous static chemiaorptlon studies, the amount of He that deaorbed decreased

as the reduction temperature increased.

Two Important points to emphasize here are that good agreement can be

obtained from TPD and static chemiaorption and that the deaorption of He from

supported Ni is significantly Influenced by readsorption.

Recent results on promoted copreclpitated catalysts showed, using TPD. the

effects of promotion on the heat of adsorption of H, on these materials

cs31. The heats of adsorption of H, using desorption rate isotherm

analysis and an isosteric method were determined for several toprecipitated Ni

catalyst systems: NI/Al,O, and La,Os- and TiOe-promoted Ni/Al,O,. The NI/Al

ratio of the materials used was 2.5 for the unpromoted and LaxOs-promoted

samples and 3.0 for the TiOs-promoted samples. Deaorption rate isotherms were

constructed from a series of TPD spectra in which the initial coverage of H,

was the variable. Hydrogen chemisorption. as revealed by TPD on the La,Oa

promoted sample, was slightly suppressed compared to the unpromoted sample and

desorptlon occurred at slightly lower temperatures. The TiOe-promoted samples

showed a much stronger suppression in H, adsorption; the extent of this

suppression Increased with an increase in the amount of TiOx. Analysis of the

data showed that the heat of adsorption of He increases in the order

(Ni/Al,Os + S% TiOx) < (NI/Al,Os + 3% LaeO,) < Ni/Al,O,. These promotors also

increased the catalytic activity. For TiO, the Increase was accompanied by a

decrease in the activation energy, but for LaeOs an increase in activation

energy was observed. It was proposed that the relative changes in the heats

of adsorption of H, caused by the presence of a promoter accounted for changes

in the ratio of adsorbed reactants and this, In turn, accounted for the

activity differences of each of the promoted catalysts.

2. Carbon monoxide TPD. Zag11 et al. Cl341 studied TPD of CO from two

commercial Ni catalysts and one Ni/Al,O, catalyst prepared in the laboratory.

Kore than half of the CO appeared to disproportionate so that both CO and CO,

deaorbed from the catalyst and the rate of CO deaorption was still significant

at 875 K on the commercial catalysts. Falconer and Zagli Cl043 likewise

observed that CO deaorbed in three peaks over the range from 300 to 8SO K on

Ni/SiO, while CO, formed from 450 to 100 K.

Lee et al. C221 studied CO desorption from Ni/SiO, in more detail. They

found that readsorption dominates the CO desorption spectra, as it does for He

desorptlon, and thus new pathways for the disproportionation reaction are

opened. A low temperature CO peak was attributed to decomposition of nickel

car-bony1 and the high temperature peak was attributed to carbon-oxygen

recombination. as had been suggested previously.

Page 68: Editor's note

TA

BL

E ‘f

&m

mar

y of

T

PD

an

d T

PR

x R

esu

lts

C19

,22.

381

Car

rier

R

eact

ive

Kin

etic

P

aram

eter

s A

nal

ysis

an

d M

eth

od

gas

spec

ies

E (

kca

l/m

ol)

A (

s-1)

as

sum

pt to

n

TP

D

He

prea

dsor

bed

H,

prca

dsor

bed

CO

TP

Rx

H,

prea

dsor

bed

C

He

prea

dsor

bed

C(s

) an

d H

(s)

% (a

) 0

k-2

(b)

2

22(1

-0.5

9~,)

k

(a)

i 0

k

(b)

-1

26(1

-0,-

J

k,

18 f

2

k4

(c)

17 f

3

H2

prea

dsor

bed

C(s

1

kaC

24

f

2

He

prea

dsor

bed

k3

2s

f 2

CO

(s)

and

H(s

)

l.sx

lo-~

~P

,[2

l.S

xlO

~~

sxio

- 12

Pc0

3.S

xlO

l3

” sx

ios

-107

-ioi

o

-sxi

o’o

deso

rpti

on

rate

is

oth

erm

s

hea

tin

g ra

te

vari

atio

n

hea

tin

g ra

te

vari

atio

n

deso

rpti

on

rate

is

oth

erm

s as

sum

ing

no

read

sorp

tion

deso

rptl

on

rate

is

oth

erm

s as

sum

ing

no

read

sorp

tion

hea

tin

g ra

te

vari

atio

n

assu

min

g ps

eudo

fi

rst-

or

der

reac

tion

(a)

Ads

orpt

ion

ra

te

con

stan

t.

(b) D

esor

ptio

n

rate

co

nst

ant.

(c

) R

ate

con

stan

ts

use

d in

si

mu

lati

on.

Page 69: Editor's note

69

Values of the kinetic paramters for adsorption and desorption of the

reactants, Hz and CO, are important data for mathematical modeling of the

methanation reaction. The activation energies and preexponential factors can

be estimated from TPD data of the individual adsorbates. An important assump-

tion is that the values obtained are applicable to Hz and CO coadsorbed on the

surface during steady-state reaction.

The kinetic parameters for adsorption/desorptlon of Hz and CO from a

NI/SiOz catalysts were determined using TPD techniques C19,22,383. Hydrogen

adsorption was found to be nonactivated. The adsorption preexponential factor

was estimated from statistical mechanics considerations. Desorption rate iso-

therm analysis was used to determine the activation energy for desorption and

its coverage dependence. Readsorption. which played a dominant role during

the desorption process, was considered in the analysis.

Carbon monoxide desorption from this catalyst results in a complex spectrum

c221. hultlple peaks and the evolution of CO, made direct analysis to deter-

mine desorption kinetic parameters impossible. However, one peak, although

shifted to higher temperatures and broadened due to readsorption effects,

resembled spectra of CO desorption from single crystal Nl surfaces C135-1371.

These data, in conjunction with a convective transport model, demonstrated that

the desorptlon energetlcs of molecular CO(s) from NI/SIOz was In reasonable

agreement with values reported in the surface science literature. Adsorption

was nonactivated. The values of the kinetic parameters for adsorption/

desorption of the reactants are reported in Table 7. These values, along

with those determined by TPRx experiments, were used to parameterize a model

for the methanation reaction over this catalyst. The methodology and results

will be reviewed later.

As indicated, CO desorbs over a wide temperature range from supported Ni

catalysts and the desorptlon spectra are quite different from those observed

from single crystal and polycrystalline Ni surfaces in ultrahlgh vacuum. This

indicates one of the weaknesses of TPD on supported catalysts. However,

additional information can be obtained from TPD when isotopes are employed, as

shown by Galuszka et al. Cl381. They adsorbed Ciao and found that the high

temperature desorption consisted almost exclusively of 060. They also saw

that a considerable fraction of the adsorbed CO underwent disproportionation.

By combining IR with TPD they concluded that carbon deposited as a result of

CO dissociation and was then oxidized at the higher temperature. The oxygen to

oxidize the carbon was conjectured to come from the AlzO, since the desorbing

CO was labeled almost exclusively with 160. They also deposited 13c from *3cO

to confirm that carbon was oxidized to yield the hLgh temperature CO peak.

Similar results for CisO adsorption have been obtained for other supported

Ni catalysts. Wilson ~201 observed that following Cl80 adsorption on NI/TiOz.

Page 70: Editor's note

70

the CO that desorbed up to SSO K was all Ciao, the CO above SSO K was EiOStly

Ci60, and the CO, that formed was mostly labeled with 160, as shown in

Figure 27. Bailey Cl081 obtained similar results for TPD from a 19% Ni/AI,O,

0.2

300 400 500 600

Temperature (K)

Figure 27. TPD spectra for Cl80 adsorption at 300 K on a 101: N1/TiO, catalyst.

Page 71: Editor's note

71

catalyst. As successive TPD experiments were carried out for ClsO adsorption,

without intermediate exposure to other gases, the amount of Cl*0 in the high

temperature peak increased significantly and the amount of CWI, decreased

significantly, as the 1% source in the catalyst was replaced by IsO. These

studies demonstrate that isotope labeling must be used to understand TPD of CO

from supported metal catalysts; the processes that occur during TPD are more

complicated than those observed In UHV studies on single crystals.

Transient data can be analyzed to determine kinetic parameters that appear

In individual steps of a proposed reaction mechanism. For example, the

activation energy and preexponential factor for a surface reaction between CO

and H, can be estimated from TPD data. In this case the preadsorbed reactants

and the products are removed from the surface in an inert gas. Lee and Schwarz

C381 used heating rate variation, assumed a pseudo first-order reaction. and

found these values to be 2Sf2 kcal/mol and .. SxlOlo s-l. While these isolation

techniques can greatly simplify the study of complex reaction systems by pro-

viding kinetic parameters, their values are not unambiguous. Surface concen-

trations of intermediates are not obtained, and the experimental conditions do

not correspond to steady-state reaction conditions. Adsorption of CO and H,

may not be independent, as discussed in the next section.

3. Desorntion of coadsorbed CO and h. Zagli et al. Cl341 carried out TPD

following CO and Hx coadsorption at 300 K on several Ni catalysts and they saw

similar results whether CO or H, was adsorbed first. The catalysts, after

being saturated with CO, could still adsorb a large amount of H,. On some

catalysts a large fraction of the adsorbed CO reacted to CH4, even though the

only H, source was the H, adsorbed on the surface at the start of the TPD

heating. The CH, formed over a much larger temperature range than it did

during TPRx In flowing H,. In addition to CH,, CO, CO,, and Ha0 were detected,

but experimental difficulties prevented observation of H, desorption.

tlore recent studies on Ni/Al,O, C8,131 and NI/TiO, Cl283 carried out TPD

following CO and H, coadsorption at 300 K and at 385 K. As shown in Figure 28,

CO desorption In the presence of coadsorbed H, on a S.lX Ni/Al,Os catalyst is

similar to that observed for CO adsorption alone (Figure 291, but the lower

temperature CO desorption and the CO, formation are clearly affected by the

presence of H,. Host noticeable is that CO and H, appear to desorb

simultaneously at higher temperature, as If some of the CO and H, interacted

to form a complex. Also, in contrast to the catalysts studied by Zagli et al.

Cl341, not much CH, formed. Similarly, on a SX Ni/TIO, catalyst Cl281, when

CO and H, were a&orbed at 300 K, CO desorbed in several peaks up to high

temperatures, CO, also formed below 600 K, and only a small amount of He

desorbed; no CH, was seen. On both of these catalysts, the CO was adsorbed

after the catalyst was cooled in H, from 775 K. This step is important because

Page 72: Editor's note

72

0.1 !

I I I I I I I I

300 400 500 600

Temperature (K)

700 800

Figure 28. Carbon monoxide, CO,, % on a 5.1% Ni/Al,Os catalyst.

and H, spectra for TPD of coadsorbed CO and The CO was adsorbed after the catalyst was

cooled to 300 K in H,. Less than 1 vmol CH,/g cat was seen.

Page 73: Editor's note

73

I I I I I I I I I

0.X

0 ! I

300 400 500 600

Temperature (K)

700 800

Figure 29. Solid lines: CO and CO, spectra for TPD of CO adsorbed at 300 K In He on a 5.1% Ni/hl,O, catalyst (no H, exposure). Dashed line: TPD spectrum of H, that was adsorbed on a S.lX Ni/Al,O, catalyst while cooling In H, from 775 K (no CO exposure).

Page 74: Editor's note

74

some of the H, adsorption is activated. However, CO readily adsorbs on these

catalysts that are originally saturated with H, and adsorption at 380 K can

provide information about the ability of CO to displace H, and the interaction

between H and CO on Ni.

A more interesting use of TPD in these studies was to study the desorption

following adsorption at elevated temperatures. As described in the TPRx

section, on both Ni/Al,O, and Ni/TiO, catalysts, much more CO adsorbs at

elevated temperatures in the presence of H, [8,11,1Sl. Glugla et al. Cl13

carried out TPD on a 5.1% Ni/Al,O, catalyst following CO adsorption at 385 K

in the presence of He. As shown in Figure 30, the subsequent desorptlon of CO

and H, was quite different from that observed following adsorption at 300 K.

Several differences are significant:

1) Huch more CO and H, adsorbed at 385 K (note the scale differences for

Figures 28 and 30).

2) The CO and H, desorbed simultaneously.

3) The CO, and CH, amounts were only one-tenth of the CO amount.

4) Small amounts desorbed at masses 30, 45, and 46.

The H/CO ratio was approximately 3, though the values ranged from 2.6 to

3.5, depending on coverage. The simultaneous desorption of H, and CO was

associated with the decomposition of a H-CO complex and the stoichiometry

indicated that a CHaO species might be present. hethoxy species have been

observed on A&O, by IR C961, but so have formate species [II?-1191. One

of the disadvantages of IR alone is that it cannot determine the amounts of

species on the surface. Since TPD can accurately measure amounts that desorb,

TPD provides complimentary information to IR. In particular, IR is less

sensitive to CHaO species, but studies on Pt/Al,O, C961 clearly related the

disappearance of CHaO from the surface to the formation of one of the CH, peaks

in TPRx.

A more accurate measure of the stoichiometry of the H-CO species was

obtained by saturating the surface with CO and H, at 385 K and then hydrogen-

ating the CO from the Ni. By taking advantage of the separation of sites in

TPRx, the low temperature CH, peak was formed without reacting a large amount

of the high temperature CH,. The subsequent TPD yielded a H/CO ratio of 3 for

a SZ Ni/Al,O, catalyst 11151. For Ni/TiO,, however, H, adsorbed on the Ni

surface after CO removal. Because the amount absorbed on the TiOx surface was

a smaller fraction of the total amount adsorbed on the catalyst, an accurate

measure of stoichlometry was not obtained. For TPD when CO was present on

both Ni and TiO,, the H/CO ratio was close to 4, but COe also formed in

significant amounts.

In analogy to the TPRx experiments employing isotopes, TPD experiments

were carried out on NI/Al,Oa 1131 and Ni/TiOg Cl281 in which iW0 was adsorbed

Page 75: Editor's note

75

I-

i-

,-

-’

I I I I I I

300 400 500 600

Temperature (K)

700 800

Figure 30. Desorption and reaction products for TPD of coadsorbed CO and H, on a 5.1X NI/Al,O, catalyst. Carbon monoxide was adsorbed at 385 K in 4. The catalyst was then cooled to 300 K and the dcsorption was carried out in He.

Page 76: Editor's note

76

on the support at 385 K and ‘3cO on the Ni after interrupted reaction. The use

of isotopes allows TPRx and TPD sites to be related. Part of the dlfflculty of

measuring the stoichiometry of the species on the A&O, is that CO from Ni

desorbs at a similar temperature to CO from the oxide, as shown by the isotope

labeling experiments for TPD (Figure 31). Note that 13C0 and 12C0 desorbed

with different peak temperatures. What is most interesting about Figure 31 is

that the IsCO, which desorbs from Ni, desorbs in a manner that is quite

different from that in Figure 29, where CO was only adsorbed on Ni. Tenta-

tively, this difference has been attributed to reverse spillover from the

A1303 9 which keeps the Ni surface covered with adsorbed CO and thus limits the

amounts of readsorption CIISI. A similar separation of CO desorption sites

was obtained on a NI/TiO, catalyst C1283.

Because CO adsorption in H, to form a CH,O species Is an activated process,

coverage variation experiments can be carried out to obtain a uniform coverage

throughout the catalyst bed. Figures 32 and 33 show TPD for CO and H, as a

function of exposure CIZI. A D-CO complex was also formed on Nl/Al,O, by

adsorbing CO at 385 K in D, flow. The D, and CO desorbed simultaneously and

with a peak temperature (D2, CO) that was S K higher than that seen when a CHsO

species was decomposed. A shift to higher temperature would be expected if a

CD30 complex formed and either C-D bond breaking or a diffusion step was

limiting decomposition.

Transient Isotope Tracti. Transient isotope tracing has been applied to CO

hydrogenation on Ni catalysts to obtain information about reaction mechanisms

and the concentration of active sites on the surface (or the fraction of the

adsorption sites that are active for methanation). Happel C321 pointed out

that discrimination among a number of reaction models is often not clear and

several models may fit a given set of data equally well. He presented some

methods to distinguish models. Happel et al. C341 used a 60% Ni/kieselguhr

commercial catalyst and carried out steady-state reaction while replacing

i2Cl60 with either 13Cl60 or iQls0. Similarly, the H, feed was replaced with

D, in other transient tracing experiments. They modeled the reaction system by

differential equations that correspond to the mass balance for each species and

then determined the coefficients in the equations by parameter estimation in

order to model the observed transients.

At steady-state in a &/CO flow over a Ni catalyst, when H, flow was

replaced by D,, transients were observed in the various deuterated methane

species C321. Since replacing H, by D, causes a kinetic isotope effect, this

experiment does not correspond to a true transient isotope tracing experiment

at steady-state, but it does yield some information about the reaction process.

The relatively short delay time before CH,D appeared, as compared to the

Page 77: Editor's note

77

- _/I] I-

I I I I I I 300 400 500 600 700 800

Temperature (K)

Figure 31. Isotopic CO and H, desorption spectra for TPD of CO and H, coadsorbed on a S.lZ NI/Al,O, catalyst. The izC0 was adsorbed to saturation at 385 K In H, flow. The catalyst was then heated to 460 K in H, to react off the tzC0 that forms the low temperature CH, peak in TPRx. The catalyst was then cooled to 300 K and exposed to 13CO until saturation before heating in He.

Page 78: Editor's note

Rat

e of

CO

Des

orpt

ion

P

P

-I 0

.b

m

b

Page 79: Editor's note

79

4

3

2

1

60 min A

300 400 500 600 700 800 900

Temperature (K)

Figure 33. TPD spectra for H, desorption following coadsorption of CO and s

at 385 K on a 5.1% Nl/Al,O, catalyst. same conditions as Figure 29.

Page 80: Editor's note

80

methane molecules with two or more D atoms, indicated that a partially hydrogen-

ated intermediate on the surface was being hydrogenated by D,. The De tracing

showed that CHg and CHs concentrations were smaller than the CH concentration

on the surface, and thus Happel C321 concluded that hydrogenation of CHx is a

controlling step In methanation. Both active and inactive carbon were assumed

in order to model the experimental results and the concentration of active

carbon was much smaller than inactive carbon.

To study the pathways of water, Happel et al. C341 replaced Cl60 by CisO

in the Hz/CO feed and they observed the appearance of isO in the HgO and CO,

products, Because of the long delay before the IsO concentrations In the CO,

and HgO products reached steady-state, they concluded that oxygen in the SIO,

support can exchange wlth these components. They were able to accurately fit

their transient data with a model that involved CO dissociation and subsequent

reaction of the adsorbed 0 atoms to form HgO and COe.

Their most extensive studies were carried out for i3cO replacement of IeCO

in the H,/CO feed stream; Happel et al. said that 13C tracing yields the best

results because they do not present problems of kinetic isotope effects or

lattice exchange. They were able to correlate their 13C transient data with a

single mechanistic model that involved CO dissociation and hydrogenation of

carbon to CH, and then CH,. They concluded that the rate determining step

Involves hydrogenation of chemlsorbed CH, (x = O-3) and they estimated concen-

trations of intermediates and velocities of individual steps in the mechanism.

Biloen and coworkers carried out a series of studies using transient isotope

tracing on Nl catalysts C33,65,1391. The turnover number (TON) was assumed

to be a product of a reaction rate per site (ITON) and the fraction of the

surface atoms that are active sites (es):

TON = 0, - ITON

The isotope transient tracing technique measures 81 (fractional coverage of

intermediates) and r (the time constant for reaction of these intermediates):

TON = Bi - f-1

(11)

(12)

Blloen et al. C331 pointed out that a purely exponential curve is expected for

a unidirectional reaction with a single intermediate X (referred to as a single

pool) on the surface;

A(g) + X(s) + B(g) (13)

Page 81: Editor's note

81

when the reactor Is modeled as a single CSTR. When more than one intermediate

In series Is on the surface, deviations are expected from purely exponential

curves. By treating their data as if they were generated by one pool, they

obtained 81, which IS an upper limit to the total coverage of intermediates,

since the effect of several pools Is to Increase T beyond the value given by

the same number of intermediates in one pool t331. When Blloen et al. C331

replaced i2C0 by 13C0 in their feed stream, they also observed a chromato-

graphic effect because time was required to replace the original taC0 by the

new adsorbed laC0.

They measured fractional coverages of reactive Intermediates from 0.08 to

0.13 for CH4 formation. In later studies, Yang et al. C6S3 found that the

coverage of reactive Intermediates was essentially independent of the s/CO

ratio, but the coverage of intermediates Increased with temperature. They con-

cluded that the low temperature coverage in intermediates is low because CH,

coverage Is restricted by the availability of active sites, due to surface

heterogeniety. They carried out transient isotope tracing on both Raney Ni and

a 60% Ni/S102 catalyst and observed a ten times larger fraction of inter-

mediates on Raney Nl than on NI/SiO,, but the same intrinsic rate constant was

seen for each catalyst. When 1sCO replaces leC0 in the feed, the amount of

adsorbed carbon monoxide present at reaction conditions can be measured. For

a large range of reaction conditions at one atmosphere total pressure, 9,o was

between 0.8 and 1.0. That is, the surface is covered with CO at reaction

conditions.

Soong et al. Cl391 used transient isotope tracing to study the changes in Ni

catalysts with time on stream. They found that the 1'3cH,, transients observed

after 120 s exposure to a HJiaCO stream were different from those observed

after 2000 s exposure to the H,/~sCO stream. They concluded that this result

indicated two pools of intermediates that reacted in parallel, and led to

relaxation curves that consist of two exponentlals. If the exposure time of

13CO was short relative to the time constant of the intermediate, then that

intermediate did not contribute to the subsequent transient when exposed to a

H2/isC0 stream.

Transient isotope tracing for CO, hydrogenation on NI/SiO, Cl401 yielded

results that were similar to those reported by Yang et al. C651. A H2/12C0 2

feed was replaced by a H,J'sCO, feed and the response of the 13CH,, signal was

used to measure the site-specific rate constant and the fraction of sites that

were active. The fraction of sites active was similar to that reported by

Yang et al. and the fraction increased with temperature, as reported by Yang

et al. Thus, the active surface appeared to be similar for both CO and CO,

hydrogenation.

Page 82: Editor's note

82

Step Concentration Changes.

Underwood and Bennett Cl411 studied methanation over a Ni/Al,O, catalyst

using step changes in reactant concentration. They found that when the

steady-state reaction in He/CO was switched to H, alone, a large CH, peak and

a smaller peak of higher hydrocarbons was produced. Additional transient

experiments involving Ha/CO, He/CO, and He&H, indicated that the large CH4

peak included a contribution due to hydrogenation of chemisorbed CO and a

contribution due to hydrogenation of surface carbon. These studies were

extended in a later paper to include additional transient experiments using

isotopic labeling Cl421. The reaction pathways leading to CH,, were found to

depend on both the rate of CO dissociation and the rate of CH, hydrogenation

without a well-defined rate-controlling step.

hethanation kinetics and Ha-D, isotope effects on the rate were studied by

Horl et al. C361 using pulse surface reaction rate analysis. A CO pulse,

when adsorbed on a Ni/SiO, catalyst, was rapidly hydrogenated to CH,, and HaO.

It was found that CH4 and Hz0 were produced at the same rate. The rate-

determining step of the reaction was found to be C-O bond dissociation of an

adsorbed CO molecule or a partially hydrogenated CO species. The ratio of the

rate of this process, for H, and Da reactants, was less than one. They con-

cluded that adsorbed CO is not directly dissociated to surface carbon and

oxygen atoms under their experimental conditions, but is partially hydrogenated

before the C-O bond dissociates.

FIultiule-Cycle Transient Analvsis

PJCTA-Concentration (ETA-CJ. The model depicted In Scheme 2 wae used by

Lee et al. C383 to compare simulated results with those determined in KTA-C

experiments.

CO(g) + CO(s) (14) t

k2 H,(g) -

k-2 2 H(s) (IS)

k3 CO(s) + 2 H(s) - C(s) + I!,0 (16)

C(s) + 2 H(s) -% CH,(s)

k, CH2(s) + 2 H(s) + CH, (g)

In this model, s represents a surface site for the competitive adsorption of

(17)

(18)

Scheme 2

Page 83: Editor's note

83

H, and CO. Subsequent steps are not written with the intent to demonstrate

the details of elementary reaction steps. The time dependence of surface

species based on this model were solved numerically; the kinetic parameters of

the model were determined by TPD/TPRx procedures outlined earlier. Asummary

of the parameter values used in the simulation was given In Table 7. The per-

turbation parameter in the HCTA-C experiments was a periodic increase in the

H, concentration; the resulting time dependent CH, response was determined

experimentally and compared to the results of the simulation. Good agreamenl

was found for temperatures from 423 to 500 K, H./Z0 ratios from 1.34 to 10, and

flow rates from 88 to 170 cms/min. Table 8 shows a comparison of the simulated

TABLE 8

Exverlmental and Simulated TOF (Flow Rate = 170 ems/s)

Temperature (K) HJCO 430 440 450 470 490

Experimental TOF x 103 (s-1)

10 2.2 4.3 7.7 26 31 S 1.8 3.3 5.7 18 4s 2 1.1 1.7 2.9 8.5 20

Simulated TOF x 103 (s-1)

10 2.0 4.1 7.9 27 80 S 1.1 2.3 4.7 17 so 2 0.04 1.0 2.0 7.7 2s

and experimental TOF of CH, for different experimental conditions. It is

important to note that In this approach no assumption was made as to the rate-

determining step. However, such assumptions are required to simplify analysis

of steady-state models of the methanatlon reaction.

One distinct advantage of these procedures is that steady-state surface

concentrations of intermediates can be determined from the numerical solutions.

This allows for additional consistency checks with other's data. For example,

the simulation showed that the surface coverages of hydrogen and carbon

increased with the HJCO ratio, and thls resulted in an Increase in the rate of

methanatlon. The steady-state results of Polizzotti and Schwarz C441, Vannice

C1431, and others Cl44,14Sl have also shown that the TOF over Ni catalysts

increases as the H,/CO ratio increases.

Page 84: Editor's note

a4

IfCiCTA-Temperature (MCTA-Tl . Steady-state methanation studies over both

supported and unsupported Ni catalysts have been reported by numerous investi-

gators. One intriguing feature of these studies is the observation of a change

in the slope of the Arrhenlus plot. Depending on the reaction conditions, two

kinetic regimes, characterized by different activation energies, are found.

Provided the reaction temperature is maintained low enough, the transition

between the two kinetic regimes is reversible. Table 9 summarizes the results

TABLE 9

Summarv of Deactivation Hodels

Reaction Conditions

P,&Torr) Pse(Torr) T (K) Catalyst Interpretatlon References

i-180 180-160

0.2-2 0.8-8

2 8

53 760

190 570

190 570

0.62-2.5 1.5-9.4

413-613

450-800

300-800

400-800

433-800

550-800

350-800

4.5-231 WSIO,

Nl(lOO)

Ni(lOO)

2-102 Ni/Si02

0.84-8.291 Ni/A1,0,

Ni powder

Ni foil

change In apparent activation energy due to change in surface coverage of CO

surface carbide and graphite formation

surface carbide and graphite formation

lnhlbitation of s adsorption by CO due to change in temperature

change in activity pattern

change in rate constant

competitive adsorption between 11, and CO on surface

Cl461

Cl473

Cl481

(431

Cl73

Cl491

L441

from previous investigators, defines the reaction conditions, and provides

their proposed interpretation for this behavior. Two conclusions emerge to

explain this phenomenon. The first proposes that an increase In surface

carbon either limits the number of active sites or is converted to a less

active carbon 143,147l. The second proposes a change in the rate constant of

one of the pathways leading to CH, formation Cl461.

HCTA-T techniques were used to study the methanation reaction over a NUSIOa

catalyst t391. The time dependent C4 signal was found to be periodic in

response to the periodic perturbation of the temperature. However, the

symmetry of the product response was found to be a function of the reaction

Page 85: Editor's note

85

temperature. In light of these observed effects, the tKTA-T technique was

used to distinguish between the models proposed to explain the changes in the

observed activity pattern. A high H,/CQ ratio was chosen to ensure that carbon

accumulation would not lead to irreversible changes in the steady-state activity

of the catalyst over the range of temperatures studied C391. The basic model

depicted In Scheme 2 was analyzed with additional steps to account for the

effect of surface carbon [Equation (1611.

Carbon precursors were assumed to be irreversibly adsorbed on active sites

and thus the activity decreases because of direct subtraction of active sites

from the catalytic cycle. The additional step to the basic model (Scheme 2)

was written as

C(reactlve)Ni +

The parameter x

deactivate more

ka x Nl --+ C(less active)Nl(x+i) (19)

was introduced to test the possibility that surface carbon can

than one Ni surface site ClSO,lSil.

A modified form of Equation (16) in Scheme 2 was consldered

change in the mechanism of-the reaction due to a change in the

k3- It was written as

CO(s) + 2 H(s) L C(s) + fl,O(g)

to account for

rate constant,

(20)

a

The kinetic parameters for the additional rate constants ki and k, were

determined in the following manner. At each temperature studied, the steady-

state TOF was measured. Each model vas solved numerically at that temperature,

and the best fit (to within 1X) value of the unknown k(l) was determined

Iteratively. The parameter “1” refers to the temperature “1”. This operation

was carried out at each temperature for which experimental steady-state data

were available. Thus, for each model tn k(i) versus i/T was obtained. The

slope of this plot gave E(l) ; A(i) was determined by A(i) = k(i)*exp(E(i)/RT).

These rate constants were then used in the numerical solution of the time

dependent mass balances for each model.

The results of these simulations were compared vith the HCTA data. When the

temperature was lower than the transition temperature, the CH4 rate wave form

was symmetric and similar to the imput sinusoidal temperature wave. Assymetric

wave forms of the CH, rate occurred above the transition temperature. The model

based on direct surface suppression by carbon predicts a lower rate than that

measured. On the other hand, the model based on Scheme 2 predicts a higher rate

than found experimentally because no steps in the overall model account for a

transition In the kinetics. The deactivation model that considers a change in

rate constant was found to predict the experiPaenta1 results well.

Page 86: Editor's note

86

We have reviewed recent publications that have used transient techniques to

study the properties of supported Ni catalysts and the mechanisms of CO

hydrogenation on these catalysts. Our earlier review considered temperature-

programmed desorption (TPD). temperature-programmed reaction (TPRx). and

temperature-programmed surface reaction (TPSR). In this review we have

limited the application of these techniques to the CO hydrogenation reaction

(a useful and widely studied probe reaction) on supported Nl catalysts and

have mostly considered studies that have been published since 1982. The

emphasis in this review is on using transient techniques to characterize

supported metal catalysts. By choosing Ni catalysts and the CO hydrogenation

reaction, we have limited the scope to a manageable size.

The majority of results presented are for two techniques that can be

carried out in a TPD/TPRx apparatus:

l temperature-programmed reduction (TPRd)

l temporature-programmed reaction (TPRx)

However, one of the advantages of a TPD/TPRx apparatus with mass spectrometric

detection is that with modifications, a range of transient techniques can be

carried out. Applications of the following techniques to supported Nk cata-

lysts were discussed:

l temperature-programmed desorption (TPD)

l transient isotope tracing

l multiple cycle transient analysis in temperature (tICTA-T) and

in concentration (DATA-C)

l step changes in feed concentrations

The TPD/TPRx apparatus can also be used as a differential reactor, and

though transient reaction techniques have many advantages, steady-state

kinetic measurements are critical for complete catalyst characterization.

In this review we have presented recent results in which the effects of

catalyst preparation on catalyst properties and the CO hydrogenation reaction

have been studied extensively by these techniques. One of the distinct

advantages of most of these techniques is that a number of catalysts can be

studied rapidly and samples can be characterized at various stages during

their preparation without exposure to the atmosphere. Transport effects can

limit the information obtainable with these techniques, however, and a

discussion of criteria to avoid such limitations was presented.

Essentially all the techniques dlscussed can be used to discern the effects

of support, preparation, and promoters on catalyst properties. In particular,

TPRd Is one of the most effective ways to compare catalyst properties of

supported Ni, which is often not completely reduced. The specific informatlon

that can be obtained from transient techniques includes:

Page 87: Editor's note

87

dispersion (TPD, TPRx, transient isotope tracing)

extent of reduction (TPRd)

reaction pathways (TPRx, TPSR, transient isotope tracing, HCTA,

step changes in concentration, TPD of coadsorbed species)

spillover (TPD, TPRx)

fraction of active sites (transient isotope tracing, TPRx)

multiple reaction sites or species (TPRx, TPSR, transient isotope tracing,

TPD)

adsorption site concentration and strength (TPD)

Systematic variations in experimental parameters have been taken advantage of,

particularly for TPRx, and the recent use of Isotopes has allowed labeling of

sites and studies of transfer between sites. horeover, comparisons between

steady-state kinetic measurements and TPRx have shown quite good correlations.

Recent studies using transient techniques have provided the following

information about supported Nl catalysts:

I) The extent of Ni interaction with supports and the extent of Ni reduction

depend on catalyst preparation, weight loading, the detailed properties of the

support, and the extent of thermal treatment. On Also, the presence of NIAlsO,,

species can be important.

2) Catalysts with Nl weight loadings of 'l-102 and higher have catalytic

properties similar to those of bulk Ni. At lower loadings, the support has

a more pronounced influence.

3) The support can trap reactive intermediates that are generated on the Ni

surface, and these intermediates can migrate back to the metal under some

conditions.

4) Catalyst performance strongly depends on preparation, and transient

techniques are particularly effective at characterizing catalyst properties

and catalyst performance.

Though these results are for supported Ni catalysts, the techniques and

approaches described in this review are generally applicable to supported

metal catalysts and to reactions other than CO hydrogenation.

ACKNOWLEDGKENTS

We gratefully acknowledge support by the National Science Foundation,

Grant CBT-8616494 (J.L.F.), and by the Department of Energy, Basic Energy

Sciences, Grant DE-FG02-87ER136SO (J.A.S.). We also want to thank the

graduate students, post-doctoral visitors, undergraduate students, and faculty

who collaborated on our work described in this review. Our appreciation is

expressed to LaNita Jansen (L.J.) for the typing of this manuscript.

Page 88: Editor's note

TERhINoLCGY

Desorption spectra. A plot of desorption rate versus catalyst temperature.

This corresponds to the form of data obtained during a TPD experiment.

Flash desorption. The analog to TPD for single or polycrystalline nonporous

materials in ultrahigh vaccum.

Interrunted TPD or TPRx. A method used to obtain partial surface coverages.

An adsorbent that is saturated with an adsorbed gas is heated in an inert

gas (TPD) or in a reactive gas (TPRx) until some gas desorbs; the absorbent

is then cooled to obtain a surface with less than saturation coverage. A

TPD or TPRx experiment is then carried out on the resulting adsorbent.

Multiple cycle transient analysis (ETA). A transient technique during

which the response of the system to small periodic perturbations is used to

obtain information about reaction kinetics.

Hultiple cycle transient analysis-concentration (HCTA-C). An tfCTA technique

in which the relative concentration of the reactants are perturbed.

tlultlple cycle transient analysis-temperature (lETA-T). An HCTA technique

In which the temperature is varied periodically about a mean temperature.

Temperature-programmed desorntion (TPD). A technique for the study of

desorption kinetics in which an adsorbed gas is desorbed from an adsorbent

surface by increasing the adsorbent temperature with time. The temperature

rise is usually linear with time. For porous catalysts, inert carrier gas

flows over the catalyst and the rato of desorption is determined by

measuring the resulting gas-phase concentration. The temperature at which

the maximum desorption rate is observed (the peak temperature) is an

indication of the strength of the surface bond.

Temperature-programmed reaction (TPRx). A technique for the study of

catalytic kinetics similar to temperature-programmed desorption. A reactive

carrier gos flows over the catalyst during heating and the rate of reaction

of this gas with an adsorbed gas is determined by continuously measuring the

concentrations of products in the gas phase, usually with a mass spectro-

meter. TPRx can also be carried out using two coadsorbed gases and an inert

carrier gas.

Temperature-programmed reduction (TPRd). A technique for catalyst character-

ization in which an unreduced catalyst is slowly heated at a constant rate In

a %/Inert gas stream with a low concentration of Ha. The uptake of Ha is

recorded as a function of temperature.

Temperature-orogrammed surface reaction (TPSR). A form of TPRx in which the

adsorbed gas is replaced by a surface layer such as carbon.

a: Kultiple cycle transient analysis

HCTA-C: bultiple cycle transient analysis-concentration

Page 89: Editor's note

89

HCTA-T: Hultiple cycle transient analysis-temperature

TB: Temperature-programmed desorption.

TPRd: Temperature-programmed reduction

m: Temperature-programmed reaction.

TB: Temperature-programmed surface reaction.

I. J.L. Falconer and J.A. Schwarz, Catal. Rev.-Sci. Eng., 2S (1983) 141. 2. tf. Eigen and L. de Baeyer, in Techniques of Organic Chemistry, Vol. 8.

Partz (A. Weissberger et al.. Editors). 3. K. Tamaru, Adv. Catal., IS (1964) 65. 4. C.O. Bennett, Catal. Rev.-Sci. Eng., 13, (1976) 121. s. J.A. Schwarz and R.J., Radix, Surf. Scl., 46 (1974) 317. 6. F.H. Dautzenberg, J.N. Helle. R.A. van Santen and Ii. Verbeek, J. Catal.,

SO (1977) 8. 7. L.L. Hegedus. C.C. Chang, D.J. &Even and E.H. Sloan, General Botors

Research Labs. Report-3039, (1979). 8. B. Sen and J.L. Falconer, J. Catal., 117 (1989) 404. 9. K.B. Bailey, G.-Y. Chai and J.L. Falconer, "Proceedings, 9th Inter-

national Congress on Catalysis" tIl.J. Phillips and M. Ternan. Editors). The Chemical Institute of Canada, Ottawa, Vol. 3, Calgary (1988) 1090.

10. B. Sen and J.L. Falconer, Submitted to J. Catal. (1989). 11. P. G. Glugla, K.H. Bailey and J.L. Falconer, J. Phys. Chem., 92 (1988)

4474. 12. R.L. Flesner and J.L. Falconer, in preparation. 13. P.G. Glugla, K.H. Bailey and J.L. Falconer, J. Catal., IIS (1989) 24. 14. B. Sen and J.L. Falconer, J. Catal., 113 (1988) 444. 1s. B. Sen and J.L. Falconer, in Proceedings of 2nd International

Conference on Spillover, Leipzig, GDR (1989) 106. 16. Y.-J. Huang and J.A. Schwarz, Appl. Catal., 32 (1987) S9.

17.' Y.-J. Huang and J.A. Schwarz, Appl. Catal., 36 (1988) 177. 18. J.L. Falconer. L.C. Burger, I.P. Corfa and K.G. Wilson, J. Catal.. 104

(1987) 424. 19. P.-I. Lee and J.A. Schwarz, J. Catal., 73 (1982) 272. 20. K.W. Wilson, Ph.D. Thesis, University of Colorado, Boulder (1987). 21. K.B. Bailey, T.K. Campbell and J.L. Falconer, Appl. Catal., in Press

(1989). 22. P.I. Lee, J.A. Schwarz and J.C. Heydweiller, Chem. Eng. Sci., 40 (198s)

so9. 23. N.W. Hurst, S.J. Gentry, A. Jones and B.D. tic Nicol. Catal. Rev.-Scl.

Eng., 24 (1982) 233. 24. S.J. Gentry, N.W. Hurst and A. Jones, J. Chem. Sot. Faraday Trans. I.,

7s (1979) 1688. 2s. D.A.D. Mont1 and A. Baiker, J. Catal., 83 (1983) 323. 26. X.-K. Wang and J.A. Schwarz, Appl. Catal., 18 (198s) 147. 27. A. Brenner and D.A. Hucul, J. Catal., 56 (1979) 134. 28. Y.-J. Huang, J. Xue and J.A. Schwarz, J. Catal., 111 (1988) 5.9. 29. J.R. Anderson, K. Foger and R.J. Breakspere, J. Catal., 57 (1979) 458. 30. H. Boer, W.J. Boersma and N. Wagstaff, Rev. Sci. Instrum., 53 (1982) 349. 31. J.P.R. Vissers, B. Scheffer, V.H.J. de Beer, J.A. Houlijn and R. Prins.

J. Catal.. 10s (1987) 277.

Page 90: Editor's note

90

32. J.

33. P.

34. J.

35. J.

36. T.

37. Y.

Happel. Isotopic Assessment of Heterogeneous Catalysis, Academic Press, 1986.

Biloen, J.N. Helle, F.G.A. van Den Berg and W.H.H. Sachtler, J. Catal. 81 (1983) 450.

Happel, I. Suzuki, P. Kokayeff and V. Fthenakis, J. Catal., 65 (1980) s9.

Happel, S. Kiang, L. Spencer, S. Oki and H.A. Hnatow, J. Catal., SO (1977) 429.

Hori, H. Masuda, H. Imai, A. Niyamoto. S. Baba and Y. Hurakami, J. Phys. Chcm., 86 (1982) 2753. Murakami, Some Theoretical Problems of Catalysis (T. Kwan,

G.K. Boreskon and K. Tamaru, Editors) University of Tokyo Press, Tokyo 1973, 203.

38. 39. 40.

41.

42.

43. 44. 4s. 46. 47. 48. 49.

50. Sl. 52. 53.

54. ss. 56. 57.

SE. s9. 60. 61. 62.

63.

64. 65.

P.-I. Lee and J.A. Schwarz. I&EC Proc. Res. Dev., 25 (1986) 76. Y.-J. Huang and J.A. Schwarz, I&EC Proc. Res. Dev., 26 (1987) 379. Y.-J. Huang. J.A. Schwarz and J.C. Heydweiller, I&EC Fund., 25 (1986)

402. J. Happel, H.Y. Cheh, H. Otarod, S. Ozawa, A.J. Severdla. T. Yoshida

and V. Fthenakis. J. Catal., 75 (1982) 314. Y.-J. Huang, P.-I. Lee, J.A. Schwarz and J.C. Heydweiller, Chem. Eng.

Commun., 39 (1985) 355. S.V. Ho and P. Harrlott, J. Catal., 14 (1980) 272. R.S. Polizzotti and J.A. Schwarz, J. Catal., 77 (1983) 1. R.J. Gorte, J. Catal., 75 (1982) 75. R.A. Demmln and R.J. Gorte, J. Catal., 90 (1984) 32. C.H. Herz, J.B. Kiela and S.P. Hartin, J. Catal., 73 (1982) 66. J.S. Rieck and A.T. Bell, J. Catal., ES (1984) 143. P.-I. Lee, Y.-J. Huang, J.A. Schwarz and J.C. Heydweiller, Chem. Eng.

Commun., 63 (1988) 205. Y.-J. Huang and J.A. Schwarz, J. Catal., 99 (1986) 249. Y.-J. Huang, J. Xue and J.A. Schwarz, J. Catal.. 109 (1988) 396. P.J. Weisz and C.A. Prater, Adv. Catal., 6, (1954) 143. (a) H.G.J. Lansink Rotgerink, Ph.D. Thesis, University of Twente (1988). (b) H.G.J. Lansink Rotgerink. J.G. van Ommen and J.R.H. Ross, in

Preparation of Catalysts IV (B. Delmon, P. Grange. P.A. Jacobs, and G. Poncelet, Editors) Elsevier, Amsterdam, 1987 795.

(c) H.G.J. Lansink Rotgerink, R.P.A.M. Paoiman, J.G. van Ommen, Appl. Catal., 45 (1988) 257.

(d) H.G.J. Lansink Rotgerink, J.C. Slaa, J.G. van Ommcn, Appl. Catal., 45 (1988) 281.

(e) H.G.J. Lansink Rotgerink, H. Bosch, J.G. van Ommen, J.R.H. Ross, Appl. Catal., 27 (1986) 41.

J-A. Schwarz, unreported results. H.S. Heise and J.A. Schwarz, J. Coll. Interf. Sci., 107 (1985) 237. H.S. Heise and J.A. Schwarz, J. Coil. Interf. Sci., 113 (1986) 44. J.A. Hieth, Y.-J. Huang and J.A. Schwarz, J. Coll. Interf. Sci., 123

(1988) 366. Y.-J. Huang, B.T. Barrett and J.A. Schwarz. Appl. Catal., 24 (1986) 241. Y.-J. Huang and J.A. Schwarz, Appl. Catal., (1988). Y.-J. Huang and J.A. Schwarz, Appl. Catal., 30 (1987) 255. Y.-J. Huang and J.A. Schwarz, Appl. Catal., 32 (1987) 4s. Y.-J. Huang. J.A. Schwarz, J.R. Diehl and J.P. Baltrus, Appl. Catal. 36

(1988) 163. R-W. Vaughan, L.B. Schreiber and J.A. Schwarz. Hagnetic Resonance in

Colloid and Interface Science (H.A. Restng and C.G. Wade, Editors) ACS Symp. Ser., 34 (1976) 275.

E-E. Wolf and F. Alfani, Catal. Rev.-Sci. Eng., 24 (1982) 329. C.-H. Yang, Y. Soong and P. Biloen, 8th International Congress on

Catalysis, Berlin (1984) Vol. IL, p.3.

Page 91: Editor's note

91

66. C.N. Satterfleld, Heterogeneous Catalysis in Practice, NcGraw Hill,

67. J.T. Richardson and T.S. Cale, J. Catal., 102 (1986) 419. 68. J.W. Schultz, J. Catal., 24 (1972) 64. 69. C.H. Bartholomew and R.B. Pannell, J. Catal., 89 (1984) 380. 70. M. Prim&, J.A. Dalmon and G.A. Martin, J. Catal., 46 (1972) 25. 71. D.G. Blackmond and E.I. Ko, J. Catal.. 94 (1985) 343. 72. E.C. Krulssink, L.E. Alzamora, S. Orr, E.B.M. Doesburg, L-L. van Reijen,

J.R.H. Ross and G. van Veen, in Preparation of Catalysts II (B. Delmon, P. Grange, P.A. Jacobs, G. Poncelet, Editors) Elsevier, Amsterdam, 1979, 143.

73. E.B.H. Doesburg, S.Orr, J.R.H. Ross and L.L. van Reijen, J. Chem. SOC.

74. J.R.H. Ross, M.C.F. Steel and Z. Zeinl-Isfahanl, J.Catal., 52 (1978) 280. 75. D.C. Puxley, I.J. Kitchener, C. Komodromos and N.D. Parkyns, in

Preparation of Catalysts III (G. Poncelet, P. Grange and P-A. Jacobs, Editors) Elsevler, Amsterdam, (1983) 237.

76. E.B.M. Doesburg, P.H.N. de Korte, H. Schaper and L-L. van Reijen, Appl.

77.

78. 79. 80. 81.

Sot. Faraday Trans. I, 77 (1981) 665. J. Zielinski, J. Catal., 76 (1982) 157. P.K. de Bokx, W.B.A. Wassenberg and J.W. Geus, J. Catal., 104 (1987) 86. A. Kadkhodayan and A. Brenner, J. Catal., 117 (1989) 311. B. Mile, D. Stirling, M.A. Zammitt, A. Love11 and N. Webb, J. Catal., ii4

(1988) 217. 82. J. Hu, J.A. Schwarz and Y.-J. Huang, Appl. Catal., 51 (1989) 223. 83. S.-L. Chen, M.S. Thesis, Syracuse University (1989). 84. C.H. Bartholomew and R.J. Farrauto, J. Catal., 45 (1976) 41. 85. J-S. Noh, Ph.D. Thesis, Syracuse University (1989). 86. P. Sabatier, J.B. Senderens and C.R. Hebd, Seances Acad. Sci., Ser. C,

134 (1902) 514. 87. F. Fischer and H. Tropsch, Brennst. Chem., 7 (1926) 97. 88. O.C. Elvins and A.W. Nash, Fuel, 5 (1926) 263; and Nature, 118 (1926)

154. 89. V.N. Vlasenko and G.E. Yuzetopvich, Russ. Chem. Rev., 38 (1969) 728. 90. G-A. Nllls and F. Steffgen. Catal. Rev.-Sci. Eng., 8 (1973) 159. 91. M.A. Vannice. Catal. Rev.-Scl. Eng., 14 (1976) 153. 92. V. Ponec, Catal. Rev.-Sci. Eng., 18 (1978) 151. 93. J-L. Falconer, P.D. Gochls and K.M. Bailey, Catalysis of Organic

Reactions, 22 (1985) 135 94. J.A. Mieth and J-A. Schwarz, J. Catal., 118 (1989) 203. 95. T.F. Mao and J.L. Falconer, Submitted to J. Catal. (1989). 96. J-L. Robbins and E. Marucchi-Soos, J. Phys. Chem., 93 (1989) 2885. 97. J.D. Way, Masters Thesis, University of Colorado, Boulder (1980). 98. K.M. Larson, Masters Thesis, University of Colorado, Boulder (1982). 99. A.E. Zag11 and J.L. Falconer, J. Catal., 69 (1981) 1.

100. K. Fujlmoto, N. Kameyama and T. Kunugi, J. Catal., 61 (1980) 7. 101. J.S. Rleck and A.T. Bell, J. Catal., 96 (1985) 88. 102. J.S. Rieck and A.T. Bell, J. Catal., 99 (1986) 262. 103. W.H. Lee and C.H. Bartholomew, J. Catal., 120 (1989) 256. 104. J-1.. Falconer and A.E. Zagli, J. Catal., 62 (1980) 280. 105. A.E. Zagli, Ph.D. Thesis, University of Colorado, Boulder (1981). 106. W.H. Lee and C.H. Bartholomew, Submitted to J. Catel., (1989). 107. K.B. Kester and J.L. Falconer, J. Catal., 89 (1984) 380. 108. K.M. Bailey, Ph.D. Thesis, University of Colorado, (1988). 109. B. Scn, T.F. Hao and J.L. Falconer, Submitted to J. Catal. (1989). 110. K.B. Kester, E. Zag11 and J.L. Falconer, Appl. Catal., 22 (1986) 311. 111. B. Sen, K.B. Kester and J.L. Falconer, In preparation. 112. G.A. Nartln and H. Praliand, J. Catal., 72 (1981) 394.

New York (1980) p. 131.

Chem. Comm., (1977) 734.

Catal., 11 (1984) 155. L.E. Alzamora, J.R.H. Ross, E.C. Kruissink and L.L. van Reijen, J. Chem.

Page 92: Editor's note

92

113.

114. Proc. Res. Dev., 20 (1981) 296.

11s. B.-S. Chen and J.L. Falconer, In preparation. 116. Y. Lu. J. Xue, X. Ll, G. Fu and D. Zhang, Cuihua Xuebao (Chinese

117. 118. 119. 120. 121. 122. 123. 124. 12s. 126.

121. 128. 129.

P.F.M.T. van Nisselrooy and J.W.E. Coenen, Appl. Catal., 3 (1982) 29. S.Z. Gsdogan, P.D. Gochis and J.L. Falconer, J. Catal., 83 (1983) 257. B. Sen and J.L. Falconer. J. Catal., in press (1990). P-H. Wentrcek, J.G. Hc Carty. C.M. Ablow and H. Wise, J. Catal., 61

(1980) 232. 130. G.G. Low and A.T. Bell, J. Catal., 57 (1979) 397. 131. J.G. McCarty, P.Y. Hou, D. Sheridan and H. Wise, AC'S Symposium Series,

202 (1982) 253. 132. G.D. Weatherbee and C.H. Bartholomew, J. Catal., 87 (1984) SS. 133. D.M. Stockwell, A. Bertucco, G.W. Coulston and C.O. Bennett,

134. 13s. 136. 137. 138. 139. 140. 141. 142. 143.

J. Catal., 113 (1988) 317. A.E. Zagli, J.L. Falconer and C.A. Keenen, J. Catal., 56 (1979) 453. B.E. Koel. D.E. Peebles and J.M. White, Surf. Scl.. 107 (1981) 1376. H.H. Madden. J. Kuppers and G. Ertl. J. Chem. Phys., S8 (1973) 3401. J.C. Bertolinl and B. Imelik, Surf. Scl., 80 (1979) 586. J. Galuszka, J.R. Chang and Y. Amenomlya, J. Catal., 68 (1981) 172. Y. Soong, K. Krishna and P. Biloen, J. Catal., 97 (1986) 330. D.A. Perkins and J.L. Falconer. in preparatlon. R.P. Underwood and C.O. Bennett, J. Catal., 86 (1984) .24S. D-M. Stockwell, J.S. Chung and C.O. Bennett, J. Catal., 112 (1989) 135. MA. Vannice, J. Catal., 37 (1975) 449. C. Kemball, Proc. Roy. Sot. Ser. A, 217 (1953) 376. J.R.H. Ross, J. Catal., 71 (1981) 205. J.A. Dalmon and G.A. Martin, J. Catal.. 84 (1983) 229. D.W. Goodman, R.D. Kelly, T.E. Madey and J.T. Yates, Jr., J. Catal., 63

144. 14s. 146. 147.

148. 149.

D-W. Goodman and J.T. Yates, Jr., J. Catal. 82 (1983) 255. R.S. Polizzotti, J.A. Schwarz and E.L. Kugler, in Proceedings of the

Symposium on Advances in Fischer-Tropsch Chemistry; American Chemical Society, Washington, D.C. 1978.

ISO. P.R. Wentrcek, B.J. Wood and H. Wise, J. Catal., 43 (1976) 363. lS1. G.A. Martin, M. Primet and J.A. Dalmon, J. Catal., 53 (1978) 321.

G. Poncelet. P. Galobs, F. Delannay. M. Genet, P. Geraed and A. Herbillon, Bull Miner., 102 (1979) 379.

C.H. Bartholomew, R.B. Pannell, J.L. Butler and D.J. Mustard, I&EC

J. Catal.), 6 (198s) 116. R.A. Della-Betta and M. Shclcf, J. Catal., 48 (1977) Ill. F. Solymosi, T. Bansagl and A. Erdohelyi, J. Catal., 72 (1981) 166. A. Palazov, G. Kudinov, C. Boney and S. Shapov, J. Catal., 84 (1982) 44. C. Mlrodatos, H. Prauliaud and If. Primet, J. Catal., 107 (1987) 275. J.E. Demuth and H. Ibach. Chem. Phys. Lett., 60 (1979) 395. S.R. Bare, J.A. Strosclo and W. Ho, Surf. Sci., IS0 (1985) 412. J.G. McCarty and H. Wise, J. Catal., 57 (1979) 406. J.A. Rabo, A.P. Rlsch and M.K. Poutsma. J. Catal., 53 (1978) 295. A.E. Zagli and J.L. Falconer, Appl. Catal., 4 (1982) 135. R.Z.C. van Heerten, J.G. Vollenbroek, M.H.J.M. de Croon,

(1980) 266.