High-performance CO oxidation catalysts engineered for CO₂ Lasers


Material Information

High-performance CO oxidation catalysts engineered for CO₂ Lasers
Physical Description:
vi, 172 leaves : ill. ; 29 cm.
Gardner, Steven Dwayne, 1961-
Publication Date:


bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1990.
Includes bibliographical references (leaves 162-171).
Statement of Responsibility:
by Steven Dwayne Gardner.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001638756
notis - AHR3696
oclc - 24160888
System ID:

Full Text








I wish to express my sincere thanks to Professor Gar

B. Hoflund whose instruction and guidance made this re-

search possible. We have shared countless discussions

through which his knowledge and optimism have served to

motivate me both personally and professionally. I would

also like to acknowledge the efforts of David R. Schryer,

Billy T. Upchurch, Erik J. Kielin and Jacqueline Schryer at

NASA Langley Research Center who provided the catalytic ac-

tivity measurements. As exemplified during several NASA-

sponsored workshops, their contributions have had an enor-

mous impact on the quality of this entire research project.

Support of this research under NASA grant NAG-I-794 is also

gratefully acknowledged. Finally, I wish to thank my wife,

Carmen, for her encouragement and forbearance during the

seemingly endless years of graduate study. Throughout the

exhausting hours of research, she has been a constant

reminder of what is most important in my life.



ACKNOWLEDGMENTS...................................... ii

ABSTRACT .............................................. iv


1 INTRODUCTION.................................... 1

COz Lasers in Remote Sensing Applications... 1
__,COz Lasers and Low-Temperature CO Oxidation
Catalysis ................................ 5


Introduction ................................. 10
Experimental................................ 11
Results and Discussion...................... 13
Summary ..................................... 23


Introduction................................. 24
Experimental................................ 25
Results and Discussion ...................... 26
Summary.................................... 37

CO OXIDATION ................................ 39

Introduction................................. 39
Experimental................................ 41
Results and Discussion...................... 42
Summary..................................... 61

PRETREATMENT TEMPERATURE.................... 63


Introduction ................................ 63
Experimental .................................. 65
Results and Discussion................. ..... 66
Summary .................................... 80

PERFORMANCE................................... 82

Introduction................................ 82
Catalyst Preparation........................ 85
Experimental ...... ........................ 87
Results and Discussion.... ......... ......... 88
Summary..................................... 105

TEMPERATURE CO OXIDATION.................... 107

Introduction............................... 107
Experimental ................................ 108
Results and Discussion..................... 110
Summary ........................................ 121

Au/MnOx .................. .................... 123

Introduction................................. 123
Experimental ................................ 126
Results and Discussion...................... 128
Summary..................................... 153

9 SUMMARY............................................ 154

FACE ANALYSIS SYSTEM.................... 158

REFERENCES..... ........... ........................... 162

BIOGRAPHICAL SKETCH................................. 172

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy




December, 1990

Chairman: Gar B. Hoflund
Major Department: Chemical Engineering

The low-temperature CO oxidation activity of numerous

materials has been evaluated in order to develop efficient

catalysts for use in COz lasers. The materials were

screened for activity in small, stoichiometric concentra-

tions of CO and Oz at temperatures near 55 C. An Au/MnOx

catalyst was synthesized which exhibited exceptional CO

oxidation activity while maintaining negligible performance

decay over a period of at least 70 days. The data suggest

that Au/MnOx has potential applications in air purification

and CO gas sensing as well.

Extensive surface characterization data from Pt/SnOx

and Au/MnOx catalysts are reported which relate surface

composition and chemical state information to corresponding

CO oxidation activity data. Ion__scattering spectroscopy

(ISS), Auger electron spectroscopy (AES), angle-resolved

Auger electron spectroscopy (ARAES) and X-ray photoelectron

spectroscopy (XPS) were utilized to observe the behavior of

these surfaces as a function of numerous pretreatments

which alter their catalytic activity. The results suggest

that Pt(OH)2 and Pt/Sn alloy formation may play a key role

in the CO oxidation mechanism on Pt/SnOx surfaces. A PtsSn

alloy was subsequently characterized before and after Hz

reduction to study its surface characteristics.

Surface characterization of Au/MnOx and MnOx was per-

formed in order to elucidate the CO oxidation mechanism.

The spectral data yield evidence that the enhanced CO

oxidation activity of Au/MnOx is related to Mn present

primarily as Mn3sO with substantial amounts of water or

hydroxyl groups. The spectra are consistent with very

small Au particles which may exist in an oxidized state.

The behavior of Au/MnOx and MnOx toward an inert pretreat-

ment suggests the possibility of a Au-MnOx interaction.


This research is a collection of studies whose main

objective is to develop low-temperature CO oxidation

catalysts which can be effectively utilized in COz lasers.

The studies themselves represent systematic efforts to syn-

thesize and evaluate new catalysts for this application and

to gain fundamental information about the low-temperature

CO oxidation mechanism. It will become apparent during the

course of the investigations, however, that the results

have potential applications beyond COz lasers.

In order to provide an understanding of the basis for

this research, this introductory section highlights impor-

tant aspects of COz laser operation and identifies the need

for low-temperature CO oxidation catalysts. This is fol-

lowed by a review which summarizes previous, related re-

search on COz lasers utilizing CO oxidation catalysis. Af-

terwards the individual studies themselves are presented.

The research concludes with a summary of the overall

results with recommendations for future research.

COz Lasers in Remote Sensing Applications

The advent of pulsed lasers has made it possible to

utilize pulsed laser energy in a radar fashion (laser radar

or lidar) with applications in altimetry, ranging, pollu-

tion detection, atmospheric chemistry, and weather monitor-

ing [1-51. Recently, the National Aeronautics and Space Ad-

ministration (NASA) has undertaken a project to develop a

laser atmospheric wind sounder (LAWS) which will incor-

porate a COz Doppler lidar [6]. The LAWS system will

operate from a satellite platform in earth orbit and

measure global wind velocities for a period of up to three

years. It is anticipated that the data obtained from LAWS

will enhance the understanding of atmospheric dynamics and

aid in weather prediction. The same techniques may be

employed to detect dangerous wind shear near airports. A

COz lidar was chosen because the technology is mature and

the corresponding COz laser radiation is safe to the eye.

However, as described below, there are significant problems

which need to be solved before COz lidars can be operated

in space or other remote locations.

i Carbon dioxide lasers are commonly operated in an

open-cycle mode. That is, fresh COz laser gas flows con-

tinuously through the laser from an external storage tank

and is subsequently exhausted to the environment. Unat-

tended operation for an extended period of time, therefore,

requires large, massive reservoirs of COz laser gas. Con-

sequently, this method of operation is undesirable for COz

lidars which will be deployed in remote locations such as

earth orbit. An additional complication arises for COz

lidars which must transmit signals long distances through

the atmosphere. The atmosphere contains a significant con-

centration of COz (approximately 300 parts per million) and

HzO both of which absorb COz laser radiation. As a result,

the COz lidar beam may experience considerable attenuation

as it propagates through the atmosphere [4,6]. A solution

is to use a CO2 laser gas which contains a COz isotope such

as C*'Oz. Relative to a common-isotope COz (C'60) laser

which operates at a wavelength of 10.6 micrometers, a laser

containing C1'Oz emits radiation at 9.1 micrometers which

interacts less with atmospheric constituents. However,

C'1Oz is much more expensive than COz and therefore open-

cycle laser configurations which use C'0Oz are not economi-


An alternative to open-cycle COz laser operation is

known as closed-cycle operation [7]. A COz laser operating

in a closed-cycle mode is initially charged with a COz gas

mixture and subsequently sealed. This configuration is ad-

vantageous because it eliminates the need for an external

CO2 storage tank and the COz gas mixture itself is con-

served. Unfortunately, there is a serious problem with COz

lasers which precludes closed-cycle operation. During the

lasing process, COz in the laser gas is dissociated into

stoichiometric amounts of CO and Oz. The gradual loss of

COz steadily decreases the power output of the laser. Per-

haps most critical, however, is the increase in Oz con-

centration which can lead to arcing and ultimately to laser

failure. An overall solution is to incorporate a CO oxida-

tion catalyst into the working confines of the COz laser to

recombine the CO and Oz into COz. Successful closed-cycle

operation is achieved by continuously recirculating the

laser gas across a suitable catalyst bed. The resulting

configuration is compact and portable and permits the COz

laser to be utilized in remote locations for an extended

period of time.

The requirements for an effective COz laser catalyst

are rather stringent [7,8]. An appropriate catalyst must

exhibit exceptional CO oxidation activity under the condi-

tions which correspond to steady-state COz laser operation.

That is, the catalyst must maintain activity at tempera-

tures near or below 50 C in a gas mixture which contains

small, stoichiometric concentrations of CO and 02 and as

much as 50% COz. In addition, the use of C18Oz requires

that the catalyst preserve the integrity of the isotopic

laser gas mixture. The catalyst should also be resistant

to dusting which could damage the delicate optics within

the laser.

COz Lasers and Low-Temperature CO Oxidation Catalysis

Few catalysts have been researched and targeted

specifically for use in COz lasers. Ordinarily, CO oxida-

tion catalysts are screened for activity using test gases

which consist of small concentrations of CO in air (excess

Oz). As a result, the performance of most CO oxidation

catalysts remains to be evaluated in the COz laser environ-

ment. Excess Oz facilitates CO oxidation and therefore

numerous catalysts perform well under these conditions even

at low temperatures (less than 100 OC) [9-171. However, CO

oxidation in stoichiometric concentrations of CO and Oz is

much more difficult. Consequently, under these conditions

only a limited number of catalysts are available which ex-

hibit significant low-temperature CO oxidation activity.

The following paragraphs summarize previous research

directed towards testing and evaluating low-temperature CO

catalysts in COz laser environments.

Hopcalite (a mixed-oxide catalyst consisting of ap-

proximately 60% MnOz and 40% CuO with trace amounts of

other oxides) has been tested in a sealed CO2 laser with

limited success [18]. Appreciable CO oxidation activity

was realized only when excess CO (approximately 0.5%) was

introduced into the COz laser gas mixture. This is due

largely to the fact that the CO oxidation rate over

hopcalite is approximately first-order with respect to CO

concentration [11,12,19]. Contrary to the COz laser's low

tolerance for Oz, a small excess in CO concentration is not

harmful to the COz laser performance. The hopcalite

catalyst was able to sustain laser performance for a period

of up to 105 pulses with no indication of failure. In or-

der to satisfy the requirements for LAWS, the laser must be

capable of at least 109 pulses [6]. Therefore, more exten-

sive testing is needed to verify the performance of hop-

calite. However, considering the activity decay which has

previously been observed for hopcalite catalysts [19], it

is doubtful that hopcalite will provide a sufficient

lifetime for the laser.

It has been shown that a heated Pt wire may be used to

recombine CO and Oz in atmospheres similar to those in

sealed COz lasers [20,211. Activity measurements between

700 and 1300 OC indicated significant recombination rates

which were independent of the CO and Oz partial pressures.

Nevertheless, this overall approach to CO and 02 recombina-

tion is undesirable in that it is fragile and requires ad-

ditional energy input to heat the Pt catalyst. In addi-

tion, the power dissipated in the laser may cause thermal

expansion of the laser components resulting in decreased

performance. It is much more advantageous to utilize a

catalyst which is effective near the steady-state

temperatures of the COz laser, particularly for space

applications where there are weight and power constraints.

Catalysts consisting of Pt and/or Pd supported on SnOx

(Pt/SnOx, Pd/SnOx, Pt-Pd/SnOx) have received considerable

attention with respect to CO oxidation in COz lasers

[7,8,22-26]. As a result, several interesting observations

have been made regarding their performance characteristics.

It has been demonstrated that optimum Pt/SnOx performance

results from Pt concentrations between 15 and 20 wt% (27].

Further enhancements in activity are realized by pretreat-

ing the Pt/SnOx surfaces with CO or Hz at elevated tempera-

tures (near 125 to 225 C) [26]. It has also been deter-

mined that relative to Pt/SnOx and Pd/SnOx alone, bimetal-

lic catalysts consisting of Pt-Pd/SnOx yield better perfor-

mance [27,281. The performance of a 2 wt% Pt/SnOx catalyst

has been verified during actual closed-cycle COz laser

operation. Utilizing an external catalyst bed containing

150 grams of 2 wt% Pt/SnOx at 100 C, the COz laser main-

tained 96% of its maximum power output for at least 28

hours with no indication of failure [29].

Although Pt/SnOx and Pd/SnOx appear to be promising

catalysts for the COz laser application, they are not yet

totally compatible with long-term laser operation in space.

Isotopic reaction studies [30] have indicated that about

15% of the total amount of Oz which participates in CO

oxidation over Pt/SnOx originates from the Pt/SnOx surface

itself. This is likely to be the case for Pd/SnOx and Pt-

Pd/SnOx as well. Since these SnOx-based surfaces or-

dinarily contain *Oz, the isotopic integrity of a C*"Oz

laser gas mixture would be destroyed. However, a procedure

has been developed to label Pt/SnOx with 1Osz yielding

Pt/Sn1XOx [30]. A preliminary test (lasting 17 days) has

indicated that C'O oxidation over the *'80-labelled

Pt/SnOx catalyst yields essentially 100% C"Oz within ex-

perimental error. Nevertheless, the tests need to be per-

formed over longer periods to ensure isotopic com-

patibility. The long-term performance characteristics of

Pt/SnOx, Pd/SnOx and Pt-Pd/SnOx represent another area for

concern. In the presence of stoichiometric CO and Oz near

ambient temperatures, all of these SnOx-based catalysts ex-

hibit significant decay in CO oxidation activity over ex-

tended time [27,28]. Consequently, their performance is

currently unacceptable in long-term COz laser applications

such as LAWS.

A study has recently evaluated the performance of a

sealed COz laser containing a Au catalyst [311. The inter-

nal wall of the laser's discharge tube was coated with a

thin, discontinuous Au film to recombine the products of

COz dissociation. The activity of the Au film appears to

result from the presence of atomic oxygen in the laser

discharge. However, the presence of a conductive Au film

in the laser amplification volume can create serious

problems under certain conditions. Care must be taken to

ensure that the Au film is sufficiently discontinuous (Au

islands) so as to avoid interference with the laser's

electrical requirements. In addition, the Au surface area

must be kept low to minimize stray reflection lasing which

reduces power output. Therefore, the disadvantages as-

sociated with coating such a large surface with Au under

conditions which minimize (but not eliminate) the drawbacks

noted above render this approach highly cost-ineffective

and inefficient.

Considering the previous research on CO oxidation

catalysis in COz lasers, much of the present research has

been directed toward SnOx-based materials such as Pt/SnOx.

Chapters 1 through 4 describe surface characterization

studies of Pt/SnOx and Pt-Sn alloy surfaces. These studies

were performed to learn more about the low-temperature CO

oxidation mechanism over Pt/SnOx catalysts. Chapter 5

presents results of activity screening experiments designed

to identify new low-temperature CO oxidation catalysts for

the CO2 laser application. Finally, Chapters 6 and 7

report activity and surface characterization data for a

Au/MnOx catalyst whose exceptional activity was identified

in the former screening experiments.



Platinized tin oxide surfaces (Pt/SnOz or Pt/SnOx)

have been proven to be efficient, low-temperature, CO

oxidation catalysts which makes them useful in COz laser

applications [7,8,231. However, relatively little is un-

derstood about the oxidation of CO over Pt supported on

reducible oxides. In order to understand the mechanism, it

is necessary to determine the surface composition, chemical

states of the surface species, and types of interactions

between the surface species. It also appears that surface

hydrogen has a large influence on the catalytic behavior of

these surfaces (26].

Alumina-supported Pt/Sn bimetallic catalysts are im-

portant in hydrocarbon reforming processes [32-34]. The

addition of Sn generally leads to prolonged catalyst

lifetime through an improved resistance toward coke forma-

tion. Furthermore, the reaction selectivity is enhanced

toward aromatization which results in an increase in the

reformate octane number.

It is important to understand the nature of the Pt/Sn

interaction in both the CO oxidation catalysts and the

reforming catalysts. This point has been controversial for

many years with regard to Pt/Sn reforming catalysts. Many

studies suggest that a Pt/Sn alloy is formed during reduc-

tion [35,36] while others suggest that tin is present as an

oxide [37-42]. Recent studies by Davis and coworkers

(43,44] are particularly convincing in demonstrating that

an alloy forms during reduction using X-ray photoelectron

spectroscopy (XPS) and in-situ X-ray diffraction. The dif-

fraction patterns indicate that the composition of the al-

loy is PtSn, but other alloys such as PtsSn may form at

different metal loadings.

The purpose of the present study is to investigate the

Pt/Sn interaction on a platinized tin oxide surface similar

to those used in CO oxidation. Auger electron spectroscopy

(AES), ion scattering spectroscopy (ISS), and XPS have been

used to examine the platinized tin oxide surface before,

during and after reduction by vacuum annealing and after

exposure of the reduced surface to oxygen at room tempera-



A tin oxide film was prepared by thermal hydrolysis of

Sn4* from a solution containing 3 M SnCl4 and 1.5 M HCI

1451. The solution was sprayed onto a titanium foil which

had been heated and maintained at 500 C in air. The

resulting tin oxide film was about 5000 angstroms thick and

it contained significant amounts of hydrogen [46-50].

Platinum was deposited by impregnating the tin oxide film

with a saturated chloroplatinic acid solution. The film

was then rinsed with distilled water and calcined in air at

500 OC for one hour.

The sample was inserted into an ultrahigh vacuum (UHV)

system (base pressure of 10-s1 Torr) which is described in

detail in the Appendix. The UHV system contained a double-

pass cylindrical mirror analyzer (CMA) (Perkin-Elmer PHI

Model 25-270AR) used for AES, ISS and XPS. The AES experi-

ments were performed in the nonretarding mode using a 3-

keV, 10-pA primary beam with a 0.2-mm spot diameter. The

ISS spectra were collected in the nonretarding mode using a

1470-scattering angle and pulse counting detection [51]. A

1-keV, 100-nA 4He' primary beam was defocused over a 1-cm-

diameter area to minimize sputter damage. The XPS spectra

were collected in the retarding mode using a 25-eV pass

energy and Mg Ka excitation. The sample was heated

radiantly using a tungsten filament and its temperature was

measured with an optical pyrometer.

Results and Discussion

An AES spectrum and an ISS spectrum taken from the as-

prepared surface are shown in Figure 2-1. The AES spectrum

exhibits peaks due to Pt, Sn, O, C1 and C. The large Pt

feature at 64 eV indicates that the surface contains a high

concentration of Pt which is consistent with the ISS

spectrum shown in Figure 2-1(b). Several features appear

in the ISS spectrum. The Sn peak is a shoulder on the Pt

peak and the peak at 0.6 E/Eo is due to contamination

(probably Na). The secondary ions apparently desorb with a

threshold at 0.12 E/Eo. This suggests that charging is oc-

curring on this surface which shifts the elemental features

to higher E/Eo than predicted by the binary collision model

[52]. A very small O peak is at 0.49 E/Eo and the peak at

0.28 E/Eo may be due to C. A crude estimate of the elemen-

tal cross sections [53] and surface composition suggests

that the outermost layer of atoms contains about 60 at% Pt.

Next the sample was annealed under vacuum in incre-

ments of 50 C from 250 to 450 C for 30 minutes at each

temperature. The ISS spectra taken after annealing at 300

and 450 C are shown in Figures 2-2(a) and 2-2(b), respec-

tively. Several changes have occurred after annealing the

as-prepared sample at 300 OC. The Sn shoulder has become a

more prominent peak and its size has increased relative to

the Pt peak. Charging is no longer apparent on this

(3)N 0


d Q 0


Ldk 0
4J 4-JI.

FA ac


0 Z


V- k
3P /-ed
z 4


P4 -P wka

3P/ONP ked d)

3 4J

- C,

- U

- U


















.> >
0 -


rT4 4

surface so the peak positions are approximately those pre-

dicted by the elastic binary scattering equation. Both the

contamination peak at 0.52 E/Eo and the O peak at 0.39 E/Eo

are reduced in intensity, and the C peak is no longer ap-

parent. Furthermore, the background intensity is sig-

nificantly decreased from that observed in Figure 2-1(b).

A possible explanation is discussed below.

The trends observed in annealing to 300 OC are con-

tinued by annealing to higher temperatures. Figure 2-2(b)

shows that the surface formed after annealing at 450 OC is

very different from the as-prepared surface. In comparing

Figures 2-2(a) and 2-2(b), the Sn peak is increased with

respect to the Pt and is even more well defined. The con-

taminant peak is no longer present and the O peak ap-

parently is further decreased in intensity. Also, the in-

elastically scattered ions composing the background are

greatly decreased in intensity. Several similarities are

observed between the ISS spectrum shown in Figure 2-2(b)

and the spectra taken from PtsSn surfaces in previous

studies (see Reference 54, Figure 5(b)]. Both surfaces

produce comparable inelastic backgrounds and display

similar Pt, Sn and O features with regard to peak shapes

and relative peak heights. The small inelastic background

is characteristic of spectra taken from metallic surfaces

where the electron mobility is sufficiently high to rapidly

neutralize inelastically scattered ions. The fact that the

ISS spectrum taken from a reduced platinized tin oxide sur-

face closely resembles an ISS spectrum taken from a Pt/Sn

alloy surface (rather than an ISS spectrum taken from a

nonreduced platinized tin oxide surface) supports the as-

sertion that a Pt/Sn alloy forms during reduction of the

platinized tin oxide surface.

The AES spectrum shown in Figure 2-3(a) and the Sn 3d

XPS spectrum shown in Figure 2-3(b) were also taken from

the sample after annealing at 450 OC. Comparing Figures 2-

l(a) and 2-3(a) shows that the Pt-to-Sn peak height ratio

essentially remains constant at a value of 1.12 during the

annealing. However, the Sn-to-O peak height ratio in-

creases from 1.10 to 1.37 during annealing by either migra-

tion of O below the near-surface region or desorption of 0

from the surface. These facts suggest an increased inter-

action between Sn and Pt in the near-surface region. The

Sn 3d XPS spectrum shown in Figure 2-3(b) has a dominant

peak at a binding energy of 486.4 eV. This feature is due

to Sn present as an oxide, but it is difficult to determine

the precise nature of this oxide from the Sn 3d features as

discussed previously [32]. The arrows in Figure 2-3(b) in-

dicate shoulders which are apparent at an approximate bind-

ing energy of 484.5 eV. This binding energy suggests that










0 9


B a

m ,

0' ^
*^ fd
Cu *P


~- co


mL 6

Sn is also present in metallic or alloyed form. Only the

oxidic feature at 486.4 eV is present before annealing the

sample so tin oxide is reduced to metallic or alloyed Sn

during the vacuum annealing, which is a reductive process.

A study by Paffett and Windham (55] provides evidence that

the form of the Sn is alloyed rather than metallic. They

deposited metallic Sn on a Pt(lll) surface and showed that

an alloy forms during annealing at or above 150 C. The

450 OC annealing temperature used in the present study then

should result in alloy formation. Thus, the ISS, AES and

XPS results are all consistent with the assertion that a

Pt/Sn alloy is formed during the reduction of platinized

tin oxide.

Similar annealing treatments have been carried out on

tin oxide films which did not contain Pt [45,48,50,56]. In

these studies vacuum annealing to 450 C or above did not

result in reduction of the tin oxide to metallic Sn. Water

of hydration or hydroxyl groups are removed from the sur-

face by vacuum annealing and a portion of the SnOz is

reduced to SnO, but metallic Sn does not form. In fact, a

tin oxide surface which has been reduced by argon-ion sput-

tering is actually enriched in 0 by vacuum annealing

through migration of bulk O to the surface [56]. These ob-

servations demonstrate the important role which Pt plays in

reducing the Sn to metallic form which then alloys with the


Next, the oxidative behavior of the surface was ex-

amined by exposing the sample to 10-7 Torr of oxygen for 1

hour at room temperature. The ISS and XPS spectra taken

from this surface are shown in Figures 2-4(a) and 2-4(b),

respectively. The expanded O feature in the ISS spectrum

indicates that the amount of 0 in the outermost surface

layer of atoms increases by about 30% during the oxygen ex-

posure. By comparison of Figures 2-2(b) and 2-4(a), it can

be seen that the oxygen exposure has increased the height

of the Sn peak with respect to the height of the Pt peak.

This behavior is characteristic of Pt/Sn alloys as

described in previous studies [54,571. The differences be-

tween the ISS spectra shown if Figures 2-2(b) and 2-4(a)

are quite similar to those obtained by exposing a Pt3Sn al-

loy surface to low-pressure oxygen at room temperature [see

Reference 57, Figures 5(a) and 5(b)]. Also, there is a

very slight reduction in the size of the shoulders on the

XPS Sn 3d peaks. The extent of this reduction is so small

that it is necessary to hold an expanded version of Figure

2-3(b) over Figure 2-4(b) in order to observe the decrease

in the size of the shoulders. The nonreactive behavior of

this surface toward oxygen as detected by XPS is also

characteristic of Pt/Sn alloy surfaces [57,58]. Using XPS,


* o
4 X





S4 o
ra 4J


C o.
U g

w col






Cheung [58] found that Sn alloyed with Pt exhibits greater

resistance toward oxidation than metallic Sn. It appears

that the Pt/Sn alloy bond causes the Sn to be less suscep-

tible to oxidation. Therefore, this finding also supports

the assertion that the shoulders observed in the XPS Sn 3d

spectra are due to alloyed Sn rather than elemental Sn, but

quantification of the rates of oxygen adsorption on a Pt/Sn

alloy and polycrystalline Sn has not yet been carried out.

It is interesting to compare the oxidative and reduc-

tive properties of platinized titania (Pt/TiOz) with those

of platinized tin oxide presented in this study since both

titania and tin oxide are reducible, conductive oxides. A

recent study by Hoflund and co-workers [59] of platinized

titania shows that the outermost layer always exhibits a

large ISS peak due to 0. Reduction of this surface results

in encapsulation of the Pt by TiO, and oxygen exposure

forms TiOz which migrates back off the Pt. This behavior

is quite different than that for the platinized tin oxide

surface for which the O lies underneath the Sn and Pt. It

is interesting to note that oxygen exposure of a polycrys-

talline Sn surface results in incorporation of 0 beneath an

outermost layer of Sn atoms [60]. One possible explanation

for the dramatically different behavior of platinized tin

oxide and platinized titania is that tin oxide is more

readily reduced to metal than titania in the presence of

Pt. Reduction of a platinized tin oxide surface by anneal-

ing under vacuum then results in the formation of a Pt/Sn

alloy rather than encapsulation of the Pt by reduced oxidic

support species as observed on platinized titania.


The Pt/Sn interaction at a platinized tin oxide

catalyst surface has been examined using ISS, AES and XPS

after reduction of the surface by vacuum annealing and ex-

posure of the reduced surface to oxygen. After reduction

XPS shows that metallic or alloyed Sn is formed, AES shows

an increase in the Sn-to-O peak height ratio and the ISS

spectrum becomes similar to that taken from a Pt/Sn alloy

surface. Furthermore, exposure to oxygen at room tempera-

ture induces changes similar to those found for a PtsSn al-

loy surface. These facts all support the assertion that a

Pt/Sn alloy is formed during reduction.



Platinum/tin systems are important catalytic materials

for hydrocarbon reforming (33-40,61-66] and low-temperature

CO oxidation [7,8,22,23]. The presence of Sn alters the

catalytic behavior of Pt so it is important to understand

the nature of the Pt-Sn interaction. In the previous chap-

ter it was shown that a Pt/Sn alloy forms when a platinized

tin oxide film is reduced by annealing in vacuum. This

result suggests the importance of characterizing Pt/Sn al-

loy surfaces since these surfaces may be responsible for

the unique catalytic properties of Pt/Sn systems. Early

studies by Bouwman and co-workers (67-70] show that the

surface compositions of PtSn and PtaSn samples are readily

altered by ion sputtering, annealing in vacuum, exposure to

oxygen or exposure to hydrogen. More recent studies by

Hoflund and co-workers [54,57,71,72] have focused primarily

on the surface enrichment in Sn of PtsSn alloy surfaces

caused by annealing in vacuum or oxygen exposure. These

studies utilized angle-resolved Auger electron spectroscopy

(ARAES) [731, high-energy-resolution Auger electron

spectroscopy (HRAES) [74], scanning Auger microscopy (SAM),

ion scattering spectroscopy (ISS) and angle-resolved X-ray

photoelectron spectroscopy (ARXPS).

Relatively little work has been done on the reduction

of oxidized Pt/Sn alloy surfaces. The purpose of the

present study is to examine an air-exposed PtsSn alloy sur-

face before and after reducing the surface by annealing

hydrogen. Results from ISS, ARAES and XPS are presented.


Details of the method used to prepare the PtsSn sample

appear in a related study [71]. The sample was stored in

argon until use. It was inserted into the UHV system (see

Appendix) and sputter-cleaned until all C and 0 contamina-

tion was removed. Then the sample was exposed to air and

reinserted into the vacuum system. After the oxidized

sample was characterized, it was moved into a preparation

chamber attached to the UHV system and reduced under 1 Torr

of hydrogen at 300 OC for 1 hour. The heating was carried

out using a special heating system [75] which did not ex-

pose any hot spots to the hydrogen which would cause dis-

sociation to atomic hydrogen. Thus, the effects of the

reduction are due only to exposure to molecular hydrogen.

The sample temperature was measured using an optical

pyrometer. After reduction the sample was moved back into

the UHV system without air exposure for further charac-


The ARAES, ISS and XPS experiments were performed

using a Perkin-Elmer PHI Model 25-270AR double-pass

cylindrical mirror analyzer (CMA) containing an internal

electron gun and movable aperture as the charged particle

energy analyzer. The ARAES experiments were performed in

the nonretarding mode using a 3-keV, 10-pA primary beam

from an external, glancing incidence electron gun. The

electron beam incidence angle was approximately 20 off the

alloy surface plane. Rotation of the 900-slotted aperture

allowed for selection of emission angles of 75 to obtain

more bulk-sensitive spectra and 200 to obtain more surface-

sensitive spectra. The ISS spectra were also collected in

the nonretarding mode using a 1470-scattering angle and

pulse counting detection [511. Sputter damage was mini-

mized through the use of a 1-keV, 100-nA 4He* primary beam

defocused over an area of about 1 cm2. Both survey and

high-resolution XPS spectra were collected with Mg Ka ex-

citation in the retarding mode using 50- and 25-eV pass

energies, respectively.

Results and Discussion

An ISS spectrum taken after sample cleaning and air

exposure is shown if Figure 3-1(a). This spectrum contains

a predominant peak due to Sn and very small peaks due to Pt


---H i
o a 0
M .g O


0 AI

o *X

M. 4J

ci M

to a

o a

In i m 0


o C', 00

and 0. The O peak is shifted from its appropriate E/Eo

value of 0.4 to 0.5, the background is fairly large and no

ions are detected below an E/Eo of 0.26. These facts indi-

cate that the surface is oxidic in nature and that charging

is occurring. This ISS spectrum is quite similar to that

taken from an oxygen-exposed Sn surface [601.

Annealing in hydrogen dramatically alters the composi-

tion of the outermost layer of surface atoms as seen in the

ISS spectrum of Figure 3-1(b). The Pt peak has grown in

size nearly to that of the Sn peak, the O peak is sig-

nificantly decreased in size, the 0 peak appears at its

predicted E/Eo value of 0.4 indicating that charging is no

longer a problem and the inelastic background is reduced

indicating that the outermost surface layer is more metal-

lic in nature. Figures 3-1(a) and 3-1(b) were collected

using the same instrument settings and are scaled ac-

curately with respect to each other. Thus, the Sn peak

height does not change significantly during the reduction.

The ISS cross section of O is much smaller than that of Pt

[53]. This suggests that the decrease in the number of 0

atoms may roughly correspond to the increase in the number

of Pt atoms in the outermost surface layer. It is probable

that the O atoms are removed from the surface layer through

water formation during the reduction. Although

speculative, it is possible that the Pt which migrates to

the surface fills the vacancies left by the 0. The chemi-

cal interaction between the Pt and hydrogen provides a

driving force for the Pt migration as discussed by Bouwman

and co-workers [671 much like the chemical interaction be-

tween the Sn and O provides a driving force for Sn migra-

tion to the surface during an oxygen exposure

[54,57,72,73]. These studies also demonstrate that anneal-

ing a Pt/Sn alloy surface in vacuum rather than hydrogen

enriches the near-surface region in Sn and not in Pt. A

hydrogen environment is necessary to enrich the near-

surface region in Pt.

Figure 3-2(a) shows a bulk-sensitive AES spectrum and

Figure 3-3(a) shows a surface-sensitive AES spectrum taken

from the cleaned and air-exposed sample. There are Sn, C

and 0 peaks appearing in these spectra but no peaks due to

Pt appear. This fact provides clear evidence that a fairly

thick (>30 angstroms) layer of tin oxide forms over the Pt-

rich region during the air exposure. The same conclusion

has been reached in a previous study by Hoflund and Asbury

[72] using angle-resolved XPS. The O/Sn peak-height ratios

in Figures 3-2(a) and 3-3(a) are 0.61 and 0.58, respec-

tively. These comparable values indicate that the composi-

tion of the tin oxide layer is fairly uniform with depth.

The ratios are also typical of those obtained from tin

oxide surfaces (50]. The C peak is probably due to

- m.D








m l







m e


* 01


0 -4





- -b



8 UJ








0 4

> m


$4 0
a -

rl 0
M 0

* M

94 4

adsorption of hydrocarbons during the air exposure and al-

ways appears after exposing these samples to air.

Figures 3-2(b) and 3-3(b) show the bulk-sensitive and

surface-sensitive AES spectra, respectively, taken after

the reduction. Two changes due to the reduction are ob-

served by comparing the bulk-sensitive AES spectra shown in

Figures 3-2(a) and 3-2(b). The O/Sn ratio is substantially

reduced and Pt migrates to the near-surface region during

the reduction. Both of these facts are consistent with the

ISS spectrum shown in Figure 3-1(b) even though the bulk-

sensitive AES probes more deeply beneath the surface than

ISS. Comparing the surface-sensitive spectrum shown in

Figure 3-3(b) and the bulk-sensitive spectrum shown in

Figure 3-2(b), it is seen that more Pt and less O are con-

tained in the near-surface region than at greater depths.

This observation supports the assertion that Pt replaces O

which has been removed during the reduction. It is also

possible to qualitatively describe the depth composition

profiles of the Pt and 0 from these spectra. Initially,

the O had a fairly uniform concentration through the oxidic

region (of greater depth than that probed by the bulk-

sensitive AES). After the reduction the 0 concentration is

low at the surface and increases at greater depths. The Pt

is more concentrated at the surface and less so at greater

depths. It is seen that the increase in Pt and decrease in

O in the near-surface region is large by comparing the

surface-sensitive spectra shown in Figures 3-3(a) and 3-


The XPS survey spectra taken before and after the

reduction are shown in Figures 3-4(a) and 3-4(b), respec-

tively. Peaks due to Sn, O, Pt, C, and Ta are observed in

the spectra. The Ta peak appears because a Ta strip was

used to hold the sample on the mounting block. As observed

by the presence of the large Pt 4f peaks in Figure 3-4(a),

XPS probes more deeply than bulk-sensitive AES. This Pt

signal originates from a Sn-depleted region which lies

beneath the tin oxide layer [57,72]. After the reduction

the Pt peaks are increased in size and the height of the O

Is peak is decreased. These results are consistent with

both the ISS and AES spectra even though XPS probes more

deeply than the other two techniques.

High-resolution XPS spectra of the Sn 3d, Pt 4f and 0

Is features before (A(a), B(a) and C(a), respectively) and

after reduction (A(b), B(b) and C(b), respectively) are

shown in Figure 3-5. The Sn 3d peaks shown in A(a) have a

binding energy of 486.4 eV which corresponds to that of

oxidic Sn (SnO, SnOz or Sn hydroxides) [32,501. However,

these peaks are broad and contain a shoulder at about 485.0

eV which is due either to alloyed or to metallic tin lying

beneath the oxide layer [72]. After the reduction most of

U -









- U

- a:






0 :

0 f

o w


x 03



-V 4







x zP




X 7






0 E-4

4r ) -4 r


01 0




-rl 0d ld

4J 0 0
:3 :3o
43 3

I o d a

m 0 4) N

-H -IF a RC~
14 e~d W 0

the Sn is in the form of alloyed or metallic Sn as

evidenced by the peak position of about 485.3 eV. This

value is larger than the Sn 3d binding energy of 484.6 eV

for metallic Sn. Since a detailed deconvolution of these

peaks has not been attempted in this study, a rigorous as-

signment of binding energies to species is not made. Fur-

ther work on this topic is in progress. A higher binding

energy shoulder at approximately 486.4 eV (shown as a

dotted line) is also present after the reduction. This is

consistent with the AES spectra shown in Figures 3-2(b) and

3-3(b), which show that most of the O lies beneath the sur-


The Pt 4f peaks shown in Figure 3-5B(a) lie at a bind-

ing energy of 71.5 eV which is about 0.6 eV larger than

that of metallic Pt. The width and shapes of these peaks

suggest that only one chemical form of Pt contributes to

this spectrum. As discussed above, this Pt is probably an

Sn-depleted alloy, but the reason for the binding energy

difference from metallic Pt is not currently understood.

After reduction the Pt 4f binding energy is 71.7 eV.

Again, it appears that only one chemical form of Pt is

present. One possibility is that the Sn-depleted Pt alloy

has a binding energy of 71.5 eV and that this binding

energy increases to 71.7 eV as the Sn concentration in-

creases after reduction of the alloy. Two studies [55,58]

suggest that the Sn and Pt peaks obtained from an alloy are

shifted to higher binding energy. This and other pos-

sibilities are being investigated.

Figure 3-5C(a) shows the XPS O Is peak taken from the

air-exposed alloy surface and Figure 3-5C(b) shows the O Is

peak taken from the reduced alloy surface. The width and

shape of the peak in Figure 3-5C(a) indicates that two

forms of O are present on the air-exposed sample: an

oxidic form with a binding energy at 530.5 eV and hydroxyl

groups with a binding energy at about 531.5 eV. The reduc-

tion process removes the hydroxyl groups leaving only the

oxidic form beneath the surface.


An air-exposed, polycrystalline Pt3Sn surface has been

examined using ISS, ARAES and XPS before and after reduc-

tion by heating at 300 OC under 1 Torr of hydrogen for 1

hour. Before reduction the surface was covered with a

fairly thick (about 40 angstroms), uniform layer of tin

oxide and hydroxide. Beneath this layer was a Pt-rich

region. The ISS data shows that the outermost atomic layer

contains mostly Sn and O with only a very small amount of

Pt present.

Reduction results in loss of 0 and enrichment in Pt of

the near-surface region. The ISS data shows (1) that the

outermost atomic layer is strongly enriched in Pt through

migration of Pt to the surface, (2) that the outermost

atomic layer becomes more metallic and (3) that O is

removed from the outermost atomic layer during the reduc-

tion. The ARAES data suggest that the Pt migrates to the

surface from the Pt-rich region by moving through vacancies

left by the O which reacted with the hydrogen and desorbed,

thus strongly enriching the outermost layer in Pt. The XPS

data shows that most of the tin oxide is reduced to metal-

lic form and probably is present as alloyed Sn. The reduc-

tion results in loss of hydroxyl groups but a subsurface,

oxidic form remains under the reductive conditions used.

It is likely that reduction under the conditions used in

this study but for longer periods would result in complete

removal of the O leaving a Pt-rich alloy at the surface.



It has been observed that the performance of

platinized tin oxide catalysts used for low-temperature CO

oxidation can be enhanced through reductive pretreatments

prior to reaction [26]. For the pretreatment and reaction

conditions used (see Figure 4-1), the unpretreated catalyst

exhibits the lowest long-term activity. Pretreatment in CO

at 100 OC greatly enhances the long-term activity, and a

further enhancement occurs following CO pretreatment at

125 C. Utilizing higher pretreatment temperatures up to

225 C yields no change in the catalytic activity compared

to the pretreatment at 125 OC. In order to understand how

platinized tin oxide catalysts function in converting CO

and Oz _to COz at low temperatures, it is necessary to

characterize the composition and chemical states of the

species present at these surfaces. This has been ac-

complished in this study using several surface charac-

terization techniques, including ion scattering spectros-

copy (ISS), X-ray photoelectron spectroscopy (XPS) and

Auger electron spectroscopy (AES) to examine the surface of


a cd


\ *

: ?\r ;

s U


* U



CL 0



e ^ (


Z .


o E




. mm

OO PleA %



.ma a. a

... .. a a

platinized tin oxide before and after pretreatment in CO as

a function of the pretreatment temperature. It is an-

ticipated that these studies will lead to a better under-

standing of this catalytic process and eventually to the

development of improved catalysts for this application.


The platinized tin oxide catalyst was acquired from

Engelhard Industries in powdered form. It has an average

particle size of 1 ijm and a BET surface area of 6.9

m2/gram. For the characterization studies, the powder was

pressed into thin disks approximately 1 cm2 in diameter.

These were inserted directly into the ultrahigh vacuum

(UHV) system (see Appendix) for pretreatment and charac-

terization. Pretreatments were performed by moving the

sample into a preparation chamber (base pressure of 10-7

Torr), introducing CO at a pressure of 40 Torr and heating

the sample at the prescribed temperature for 1 hour. Then

the sample was allowed to cool while the preparation cham-

ber was pumped down to 10-7 Torr before moving the sample

into the main analytical chamber (base pressure of 10-1

Torr) for analysis by ISS, XPS and AES. The samples were

heated at 75, 100, 125 and 175 OC in the preparation cham-

ber using a custom sample heater [75], which did not expose

the reducing gas to hot spots that would have dissociated

the CO. Two different as-prepared samples were analyzed

without prereduction and the results were found to be

reproducible. A new sample was prepared and introduced

into the analysis chamber for each reduction temperature


The ISS, XPS and AES data were taken using a double-

pass cylindrical mirror analyzer (CMA) (Perkin-Elmer PHI

Model 25-270AR) as the charged-particle energy analyzer.

The ISS spectra were collected in the nonretarding mode

using a 1470-scattering angle and pulse counting detection

[51]. A 1-keV, 100-nA 4He* primary ion beam was defocused

over a 1-cm-diameter area and spectra were collected as

quickly as possible (typically 90 s) to minimize damage

from sputtering. The AES experiments were performed in the

nonretarding mode using a 3-keV, 10--A primary electron

beam with a 0.2-mm spot diameter. The XPS experiments were

performed in the retarding mode using Mg Ka excitation.

Survey XPS spectra were collected using a 50-eV pass energy

whereas a 25-eV pass energy was utilized for obtaining

elemental lineshape information.

Results and Discussion

An XPS survey spectrum taken from the as-received

Engelhard platinized tin oxide catalyst is shown in Figure

4-2. The peaks of significant size are due only to 0 and

Sn, and no peaks due to C or other typical contaminants ap-

pear. In fact, this spectrum is essentially identical to










. 0

0 '


- a




40 Wj



an XPS spectrum obtained from a very clean tin oxide sur-

face. It is most interesting that the predominant Pt 4f

peaks are so small that they cannot be discerned readily in

this spectrum. A corresponding AES spectrum taken from the

same surface is shown in Figure 4-3. Again, the

predominant features are due to Sn and 0, and this spectrum

is similar to one obtained from a clean tin oxide surface.

However, a feature due to Pt appears at a binding energy

near 64 eV. It can be described as an edge rather than a

typically shaped AES peak. Similar Pt features have been

observed in a study of the electro-chemisorption of Pt on

tin oxide (76]. This feature is more prominent than the Pt

XPS peaks in Figure 4-2. In agreement with the XPS data,

no contaminant peaks are apparent in this AES spectrum. As

the catalyst is reduced at various temperatures, very small

changes are observed in the XPS and AES spectra correspond-

ing to Figures 4-2 and 4-3, but taken from the reduced sur-

faces. Therefore, these survey spectra are not shown.

Ion scattering spectroscopy is a particularly useful

technique for examining catalytic surfaces because it is

very highly surface sensitive (outermost one or two atomic

layers). The ISS spectra taken before (a) and after (b-e)

reduction are shown in Figure 4-4. The spectrum shown in

(a) consists of peaks due to O, Sn and Pt, and the high in-

elastic background is characteristic of ISS spectra taken

C -





co is

- i




a -

0 4


3 ,

4J 0


10 0

v U


1. o
Ie an


from nonmetallic surfaces. In such a case, inelastically

scattered ions are not efficiently neutralized since the

electron mobility at a non-metallic surface is so low that

these ions contribute to the background signal. A previous

study by Asbury and Hoflund [60] showed that O penetrates

beneath the surface during room-temperature oxygen exposure

of polycrystalline Sn. This suggests that the fairly large

O peak in (a) is due to O associated with the Pt and/or

perhaps with hydroxyl groups attached to Sn that may lie

above the surface [50].

The ISS spectra shown if Figure 4-4(b-e) were obtained

from samples reduced in 40 Torr of CO for 1 hour at 75,

100, 125 and 175 C respectively. All four spectra have

two characteristics in common. Firstly, the pretreatments

have resulted in negligible inelastic backgrounds, which is

indicative of the formation of surfaces with a metallic na-

ture. Secondly, the O peak is no longer discernible after

the reductions, which is also consistent with the observa-

tion that the surfaces appear to be metallic. The reduc-

tive pretreatment results in an increase in the Sn/Pt

ratio, and the extent of this increase is greater at higher

reduction temperatures. As shown in chapter 1, an increase

in the ISS Sn/Pt ratio during reduction of platinized tin

oxide surfaces has been found to be indicative of alloy

formation. All of the ISS spectra shown in Figure 4-4 were

taken using the same instrument settings, but the maximum

peak heights vary considerably. Although the variation is

not understood, it could be due to changes in ion

neutralization probability, surface morphological changes

or changes in the concentration of surface hydrogen, which

have been shown to alter the ISS signal strength [77,781.

The Sn 3d XPS spectra and Sn MNN AES spectra taken

before and after reduction are shown in Figures 4-5 and 4-

6, respectively. Before pretreatment (air-exposed sample)

the Sn 3ds/2 lineshape and peak position (486.4 eV) indi-

cate that Sn is present in the +2 or +4 oxidation states

most likely as SnO, Sn(OH)z, SnOz or Sn(OH)e and that

metallic Sn is absent. As discussed by Hoflund and co-

workers [32], it is not possible to distinguish between

these species based on the XPS Sn 3d peaks, but more

specific information can be gained about these species

using electron energy-loss spectroscopy (ELS) [48,561,

valence band XPS [48,50], electron-stimulated desorption

(ESD) [49,50] or secondary ion mass spectrometry (SIMS)

[46,47,79]. The metallic XPS Sn 3d5/z peak appears at an

energy of 484.6 eV [80]. When a small amount of metallic

Sn and a relatively large amount of tin oxides or

hydroxides are present, a slight broadening appears on the

low-binding-energy sides of the oxidic XPS Sn 3d features.

This is the case for the 175 OC-reduced sample as shown in

X r



.'4 0
in -r4
.!o -,.t

0 *

4 0) -r4
S- 0 4P

| ,4 w


0 0 0
| o -r4


Fi D
| *M *01

S14 0
o a go

,4 g- 4
'0 V -r

*_ 0
LO 43 U4

WN $4 $C
t0 0 wl

0 i3 o














Figure 4-5. Samples reduced at lower temperatures also

yield this metallic shoulder, and the amount of metallic Sn

produced generally increases as the annealing temperature

increases with the most being produced at 175 C. However,

the amounts of metallic Sn produced at 75 and 100 C ap-

pear similar. Similar observations can be made by con-

sidering the AES Sn MNN peaks shown in Figure 4-6. The

high-kinetic-energy oxidic peak lies at 430 eV, while the

high-kinetic-energy metallic peak lies at 423 eV. The

peaks shown in Figure 4-6(a) are characteristic of oxidic

Sn. When metallic Sn is present, the height of the split-

ting between the two primary peaks decreases. This is ob-

served in spectra (b)-(e) taken from the reduced samples.

In agreement with the XPS Sn 3d spectra, the extent of

metallic Sn formation is greater at elevated reduction tem-

peratures with the maximum amount of metallic Sn being

produced at 175 OC. Further reduction, either for pro-

longed periods or at higher temperatures than used in this

study, would undoubtedly result in the production of in-

creased amounts of metallic Sn.

The XPS O Is peaks are shown in Figure 4-7. These

were taken before (a) and after (b) reduction at 175 OC.

The peak shown in (a) exhibits a distinct asymmetry on the

high-binding-energy side due to the presence of hydroxyl

groups, which are responsible for a shoulder at










0 1

4.) r
A 00

0 a





k.- E4-


approximately 531.8 eV. It is also possible that the very

small shoulder near 533.0 eV is due to adsorbed water.

This assignment is consistent with results obtained in the

study of hydrated polycrystalline tin oxide films by

Tarlov and Evans [81]. Pretreatment at 75 OC slightly

reduces the amounts of adsorbed water and hydroxyl groups

present while pretreatment at 100 C or above eliminates

the adsorbed water and further reduces the concentration of

hydroxyl groups. The 175 C pretreatment results in the

lowest surface hydroxyl group concentration. It is inter-

esting to note that these hydroxyl species are strongly

bonded to the surface and require annealing at 600 C in

vacuum for nearly complete removal [48,50,82].

The O content of the near-surface region is decreased

by the pretreatment process. The AES and XPS O/Sn atomic

ratios obtained at the various pretreatment temperatures

are listed in Table 4-1. A decrease in the O/Sn ratio is

caused by loss of adsorbed water, a decrease in hydroxyl

group concentration, reduction of tin oxides and hydroxides

to metallic Sn and reduction SnOz to SnO. However, the

relative importance of these factors cannot be assessed

completely from the types of data taken in this study. The

amount of Pt present on these surfaces is so small that

changes in the Pt oxidation state, which are discussed

below, would not affect the O/Sn atomic ratios presented in

W) tn qds v) nr
0 -,4 n -,4 r4 r-4 r

No 0 o o o
.* I omm
m aC C na


SUo ln olno

5 t^ ^




" t

Table 4-1. The AES and XPS results in Table 4-1 are dif-

ferent with respect to both magnitude and trend. The AES

data indicate that the near-surface region loses about 20%

of its 0 regardless of the reduction temperature. The XPS

O/Sn atomic ratios are considerably larger than the AES

values and decrease monotonically as the pretreatment tem-

perature increases. Either the XPS values differ from the

AES values due to inaccuracies in the tabulated cross sec-

tions or the variation is due to the fact that AES and XPS

probe different volumes of the near-surface region. If the

difference were due only to inaccurate cross sections, then

the XPS O/Sn ratios obtained from the reduced samples would

not show such a large variation with reduction temperature.

Thus, the variation is due to the fact that XPS probes more

deeply than AES, which is consistent with mean-free-path

arguments also. The kinetic energies of the XPS O Is and

Sn 3d electrons are 723 eV and approximately 770 eV,

respectively, while the kinetic energies of the AES O and

Sn electrons are near 510 eV and 430 eV, respectively. The

corresponding mean free paths (T) are about 6 angstroms for

XPS and 4 angstroms for AES [833. Since most of the XPS

and AES electrons originate within a depth of approximately

3(T), XPS probes about 18 angstroms beneath the surface

while AES probes at a depth of about 12 angstroms. Then,

the data in Table 4-1 indicate that the near-surface region

probed by AES contains less O than the region probed by XPS

for the untreated sample and samples reduced at 75-175 OC.

Reduction at any of the temperatures used causes the AES

O/Sn ratio to drop from 1.32 to about 1.06 whereas the XPS

O/Sn ratio decreases monotonically as the reduction tem-

perature increases. The essentially constant AES value

probably results from competing processes, that is, O leav-

ing the surface as CO2 during the reduction and O migrating

to the near-surface region from farther beneath the sur-

face. This is consistent with the trend of the XPS data

implying that subsurface reduction takes place to a greater

extent at higher temperatures. A similar phenomenon has

been observed previously for the reduction of a TiOz (001)

surface [84].

The XPS Pt 4f peaks obtained from the unpretreated

samples are shown in Figure 4-8(a), and the peak assign-

ments used in this study are listed in Table 4-2. Most of

these assignments were taken from a standard reference

[801, but the 72.3 eV feature has been assigned as Pt-O-Sn

in previous studies of platinized tin oxide surfaces

[32,561. The spectrum shown in Figure 4-8(a) indicates

that very little metallic Pt is present and that the Pt

species consist mostly of Pt-O-Sn, Pt(OH)z and Pt oxides.

In fact, a spectrum very similar to this one has been taken


0r .0
S I n


a a0
00 14
S4J 0
0 0 I

t o qv 9

lu r)

-i Co

$4 0





0w N
4 wN oU n

0% .00
4 a $4

SMr-i A
* -4 o) 4)

40-O 0 0 4

C 4 o a 3 4-
0 0 0

i ._ .4M M

E01 41 *
0 01C 00

'- I^o S
a 'U a

from a co-deposited platinum/tin oxide film after calcining

in air at 725 K for 1.5 hours [32].

The XPS Pt 4f spectra taken after pretreating at 75

and 100 C are shown in Figures 4-8(b) and 4-8(c), respec-

tively. These two spectra are quite similar in that very

little metallic Pt is present and the predominant species

consist of Pt-O-Sn and Pt(OH)z. The metallic Pt apparently

is in the form of small crystallites. A shoulder due to

PtOz may also be present in both spectra, but this shoulder

is smaller after the 100 C reduction. The relative

amounts of Pt-O-Sn and Pt(OH)z are similar in both spectra.

Pretreating at a temperature 25 C higher produces a sig-

nificant change in the Pt species present, as shown in

Figure 4-8(d). The predominant Pt species after the 125 C

reduction are Pt(OH)z and Pt-O-Sn in approximately equal

amounts. Also, a small amount of metallic Pt is present

after this pretreatment, and features due to Pt oxides do

not appear.

Reduction at 175 C produces a further and more

pronounced shift in the Pt chemical state toward Pt(OH)2 as

shown in Figure 4-8(e). The Pt-O-Sn feature is now a

shoulder on the Pt(OH)2 peak, and the amount of metallic Pt

present is decreased compared to the lower temperature

reductions. The spectrum shown in Figure 4-8(a), taken

from the unpretreated sample, exhibits a prominent feature

at 79.8 eV binding energy. Since the splitting between the

Pt 4f7/2 and Pt 4f5/2 peaks is approximately 3.35 eV, this

feature does not correspond to the Pt 4f5/z peak of any of

the species listed in Table 4-2. The 75 OC pretreatment

reduces the size of this feature, and it does not appear

after the 100 or 125 OC pretreatment. Therefore, it be-

haves like an oxidic feature and reappears in Figure 4-


It is interesting to compare the spectral information

contained in Figure 4-8 with the kinetic information con-

tained in Figure 4-1. The nonpretreated catalyst exhibits

a low activity compare to any reduced catalyst. This

correlates with the facts that the Pt is predominantly

oxidic on the unpretreated catalyst and that significant

changes in the Pt chemical state occur during the reductive

pretreatments used in these studies. As the pretreatment

temperature increases, more of the Pt is converted into

Pt(OH)z. This fact suggests that the Pt(OH)z plays an im-

portant role in the conversion of CO and Oz into COz at

low temperatures. In this discussion a relationship is

being drawn between the long-term catalytic behavior (past

the first 500 minutes of reaction) and the chemical state

of the Pt after the pretreatment but before the onset of

reaction (time=0 in Figure 4-1). The initial catalytic be-

havior is quite complex [261, and changes may occur in the

Pt chemical state during this period. Consequently,

characterization studies are in progress in which the state

of the catalytic surface is being examined as the reaction

is run for various periods of time. These studies should

lead to an understanding of the chemical changes respon-

sible for the unusual, initial catalytic behavior and the

long-term decay in catalytic activity.

A small amount of metallic Pt forms during both the

125 and 175 OC pretreatment, and metallic Sn is also

present as stated above. Paffett and Windham [55] have

deposited layers of Sn on Pt(lll) and found that alloy for-

mation during annealing at or above 150 C is strongly sug-

gested by their data. Also, Fryberger and Semancik [85]

have found that Pd deposited on SnOz (110) alloys with the

Sn even at room temperature. It is anticipated that alloy

formation occurs in the Engelhard catalyst under the

pretreatment conditions used in this study. Since the

amount of Pt present on this surface is small, it is likely

that most of the metallic Sn is not alloyed while probably

all of the Pt is alloyed.


A platinized tin oxide catalyst (commercially avail-

able from Engelhard Industries for low-temperature CO

oxidation) has been examined using surface analytical tech-

niques, including ISS, AES and XPS, before and after

pretreatment by annealing in 40 Torr of CO for 1 hour at

75, 100, 125 and 175 C. The results have been correlated

with catalytic activity data. The nonpretreated sample

consists primarily of oxidic Sn (SnOz, SnO and Sn(OH)x)

with a very small amount of Pt present as Pt-O-Sn, Pt(OH)z,

PtO and PtOz species. Reduction results in loss of both 0

and OH- from the Sn and produces metallic Sn. The extent

of these processes increases as the pretreatment tempera-

ture increases. The chemical state of the Pt changes with

pretreatment temperature. At or below 100 OC, the

predominant forms are Pt-O-Sn and Pt(OH)2. As the reduc-

tion temperature increases, more Pt(OH)2 forms. This fact

suggests that Pt(OH)z plays an important role in the low-

temperature catalytic oxidation of CO. The results

demonstrate that the application of surface analytical

techniques in studies of real catalysts can provide infor-

mation that is useful in understanding catalytic behavior.



Research has indicated that platinized tin oxide

(Pt/SnOx) is an efficient CO oxidation catalyst at condi-

tions which correspond to steady-state CO2 laser operation

[7,8,23,28]. As discussed in the previous chapter, the

performance of Pt/SnOx may be further enhanced following

pretreatment in CO at elevated temperatures. However, for

CO reduction temperatures above about 175 C, a sharp, tem-

porary decrease in activity is observed initially [26].

This induction period is believed to be the result of sur-

face dehydration caused by combination of surface hydroxyl

groups and desorption of water. No significant induction

period results when Pt/SnOx is humidified either after CO

pretreatment or during the reaction itself and optimum

pretreatment times exist which are consistent with the

hypothesis involving surface dehydration. The data ob-

tained thus far suggest that COz is produced through decom-

position of a bicarbonate species formed from adsorbed CO,

a surface hydroxyl group and lattice oxygen [86].

Acknowledging the importance of surface hydroxyl

groups, there has been considerable effort directed towards

the development of a platinized tin oxide catalyst which is

supported on a silica substrate (Pt/SnOx/SiOz) [27,87].

Hygroscopic silica may improve the performance of Pt/SnOx

surfaces by preventing extensive surface dehydration and

consequent activity loss. Experiments have been conducted

wherein the catalyst performance has been optimized with

respect to several variables including pretreatment proce-

dures [27]. Superior performance is realized by using a

reductive pretreatment in 5% CO/He at 125 C for 1 hour.

As pretreatment temperatures approach 250 C, there is a

significant decrease in the observed activity. The op-

timized Pt/SnOx/SiOz catalyst represents a significant im-

provement over commercially available Pt/SnOx with respect

to low-temperature CO oxidation activity and performance


In order to understand these experimental observa-

tions, a characterization study has been carried out on

silica-supported Pt/SnOx before and after reduction in CO

at 125 and 250 OC. This study utilized ion scattering

spectroscopy (ISS) and X-ray photoelectron spectroscopy

(XPS) to examine the concentrations and chemical states of

species present at these surfaces and the changes which oc-

cur during reduction.


The catalyst prepared for this study consisted of a

thin layer of platinum and tin oxide dispersed on a silica

gel substrate [27,87]. The silica gel was impregnated with

tin oxide by evaporation of a stirred solution of tin metal

powder and silica gel in concentrated nitric acid at

150 C. Subsequently, Pt was precipitated from an aqueous

solution of platinum tetraamine dihydroxide and formic acid

with heating followed by drying at 150 OC. The final

product was heated in air at 150 C for 4 hours.

As-prepared, silica-supported Pt/SnOz samples were in-

serted into an ultrahigh vacuum (UHV) system (see Appendix)

which has an ultimate pressure near 10-11 Torr. After ini-

tial surface characterization, the samples were transferred

into a preparation chamber connected to the UHV system and

reduced in 10 Torr of CO for 2 hours at 125 or 250 OC.

Heating was accomplished using a platform heating element

[75] which did not dissociate the reducing gas. Sample

temperatures were measured using a thermocouple attached to

the stainless-steel sample support block. After reduction

each sample was returned to the UHV analytical chamber

without air exposure for further characterization.

Energy analysis for the ISS and XPS experiments was

accomplished using a Perkin-Elmer PHI Model 25-270AR

double-pass cylindrical mirror analyzer (CMA). The CMA

contained an internal, movable aperture which varied the

polar acceptance angle for incoming particles. The ISS

spectra were collected in the nonretarding mode using a

1470-scattering angle, which was fixed by the experimental

geometry, and pulse counting detection [51]. A 100-nA, 1-

keV 4He* primary ion beam was defocused over a 1-cm2 area

to minimize sputter damage. Survey and high-resolution XPS

spectra were recorded with Mg Ka excitation in the retard-

ing mode using 50- and 25-eV pass energies respectively.

Results and Discussion

Survey XPS spectra taken from one silica-supported

Pt/SnOx catalyst sample before and after reduction in 10

Torr of CO at 125 C appear in Figure 5-1, and XPS survey

spectra taken from another sample before and after reduc-

tion at 250 OC appear in Figure 5-2. The two catalyst

samples examined were similar in that both were randomly

dispensed from the same catalyst batch. The spectra shown

in Figures 5-1(a) and 5-2(a) were taken from the air-

exposed surfaces. They exhibit predominant peaks due to Sn

and O while peaks due to Pt are present but less discern-

ible. Differences between Figures 5-1(a) and 5-2(a) with

regard to the O/Sn ratio suggest that the as-prepared

catalyst samples lack uniformity in average surface com-

position. Nevertheless, important information may still be

obtained with respect to data taken from each catalyst


0 4J



X o

,e Q


0 fn$4

1k 0 5







--o c

08 4

3 a

z 01

o 0

o W U .
0 10

t 4o 4

% 3u 0
to C14

0 -4

4 a S0
:3l 0 3
W $4m

sample before and after reduction. A thick tin oxide sur-

face layer (> 50 angstroms) is indicated by the lack of Si

XPS signals arising from the silica substrate. Only trace

amounts of C and surface contaminants are detected. Reduc-

tion in CO results in changes in the Sn and O signals as

shown in Figures 5-1(b) and 5-2(b). The spectra indicate

that the surface region of the sample reduced at 125 OC be-

comes oxygen-enriched while depletion of surface O occurs

in the sample reduced at 250 oC. A possible mechanism

responsible for this result is discussed below.

An ISS spectrum taken from an air-exposed, silica-

supported Pt/SnOx surface before reduction is shown in

Figure 5-3(a), and ISS spectra taken after reduction in CO

at 125 and 250 C appear in Figures 5-3(b) and 5-3(c),

respectively. Figure 5-3(a) reveals distinct Pt and O fea-

tures and a Sn peak which appears as a shoulder on the Pt

peak. Peaks due to small amounts of contaminants such as C

and Na are also present as are some peaks due to charging.

Peaks due to charging shift in energy as the sample is

rotated so they can be distinguished easily from elemental

peaks, which appear at energies near those predicted by the

elastic binary collision model [52,88]. After catalyst

reduction at 125 OC, Figure 5-3(b) indicates that the sur-

face composition has changed considerably. The Sn and Pt

peaks have increased in size significantly, and the Sn peak





6 a Id
4 J0

0 40
W 14





z tOW 0

oiia~ 0

has become more prominent and well defined in size with

respect to the Pt. As shown in Chapter 1, this behavior is

characteristic of Pt/Sn alloy formation. A few peaks due

to migration of substrate impurities such as Na and Mg also

appear after the reduction as does a Ti peak from the

sample holder. After reduction at 250 C the Sn and Pt

peaks are greatly reduced in size due to coverage by con-

taminants (primarily Na and Mg) in the silica which migrate

to the surface during reduction. Due to the stability of

silica at these fairly low reduction temperatures, silica-

containing species most likely do not migrate over the Pt

and Sn. Since ISS is essentially sensitive to the outer-

most layer of surface atoms, the data in Figure 5-3 suggest

that the 250 OC reduction results in physical coverage

encapsulationn) of most of the surface Pt and Sn. This en-

capsulation hypothesis is confirmed by ISS data taken after

the reduced surface was lightly sputtered. As shown in

Figure 5-4, the sputtering process uncovered significant

amounts of underlying Pt and Sn. Although peak energy

shifts and surface charging features appear in these

spectra, the peaks which correspond to Pt and Sn remain

prominent and meaningful.

The data presented thus far suggest that reduction at

both 125 and 250 C promotes movement of O, Sn and

subsurface impurities contained in the silica. The O


C%4 -r4 14

0 >0
Id 0

43 MY o

.0 0

4-) ON $4



U+, 1



M 4J~
0% 4

P4. ~- 44 -

concentration in the near-surface region is probably

determined by the rates of several mechanisms. One is

migration of subsurface 0 to the near-surface region during

the reduction at elevated temperature under the chemical

potential produced by the presence of the reducing gas.

Another is loss of surface O through reaction with the

reducing gas and subsequent desorption as COz. As stated

above, O loss may also occur through OH- combination fol-

lowed by desorption of HzO. At 125 C the rate of migra-

tion of subsurface O to the surface apparently is greater

than the rate of O loss by reaction to form COz and H20,

which results in an increase in the near-surface O con-

centration as observed in the XPS data. Similar observa-

tions have been made during the reduction of a TiOz(001)

surface [841. However, at 250 OC the rate of O loss

through reaction is greater than the rate of 0 migration to

the surface, resulting in a decrease in the O concentration

in the near-surface region.

As shown in Figure 5-5, the Pt 4f XPS spectra exhibit

broad features suggesting that multiple Pt oxidation states

are present both before and after reduction. According to

a standard reference [80] and previous studies from this

laboratory [32,89], Figure 5-5(a) indicates that a mixture

of primarily PtO2 and PtO with smaller amounts of Pt(OH)z

and metallic Pt is present on the air-exposed surfaces.





Id 43


C, U243
14 W
4J 0.

>- .43 9

*5 44.


ri4 0.5
04 Wo

Ln V 0
0 -r4
14 43
:1 1 8
tf 01
-r $ 4
1 4 f

After reduction, peak positions shift toward lower binding

energies as the surface Pt species are partially reduced.

After reduction at 125 OC the predominant species present

is Pt(OH)z with an increased amount of metallic Pt.

However, only very small amounts of PtO and PtOz remain

after the 125 C reduction. The extent of reduction is

much greater after the 250 OC reduction. The Pt is

primarily in metallic form with a relatively small amount

of Pt(OH)z. Little or no PtO or PtOz is present. There

has been some debate concerning the stability of PtO, but

its existence is well substantiated in numerous studies


Tin can be present in several forms which include

SnOz, SnO, Sn hydroxides, Sn suboxides, Sn-O-Pt species

132,891, metallic Sn and Sn alloyed with Pt. Using XPS, it

is not difficult to distinguish between metallic Sn and Sn

oxides because their Sn 3d binding energies differ by about

1.8 eV, and Asbury and Hoflund have observed XPS features

due to suboxides [601. However, it is difficult to distin-

guish between SnOz and SnO using XPS, but this can be done

using valence-band XPS [50,953 or electron energy-loss

spectroscopy (ELS) [56,96]. Paparazzo and co-workers 197]

claim that there is a 0.18 eV difference in the Sn 3d bind-

ing energies of SnO and SnOz. Valence-band XPS (50,95] and

electron-stimulated desorption (ESD) [501 can be used to

identify Sn hydroxides. These species also yield a high-

binding-energy shoulder on the O Is XPS peak [50,81], but

no studies have reported any influence on the Sn 3d XPS

peaks. Studies of Pt/Sn alloys [55,58] suggest that the Sn

3d XPS alloy features differ from those of metallic Sn, but

a systematic study of Sn 3d shifts due to alloying has not

been reported.

The Sn 3d XPS spectrum taken before reduction is shown

in Figure 5-6(a), and spectra taken after reduction in CO

at 125 and 250 OC are shown in Figures 5-6(b) and 5-

6(c), respectively. Small changes in the Sn 3d peak shapes

occur during annealing in CO. Since a large number of Sn

species may be present and their contributions to the Sn 3d

XPS features are not well understood, it is only possible

to analyze these peaks in a qualitative manner. Before

reduction, the Sn 3d XPS peaks are due predominantly to

SnOz. However, they are not symmetrical, and the symmetry

on the low-binding-energy side indicates that some of the

Sn on the as-prepared catalyst may be present as suboxides

and SnO and possibly as metal. The Sn 3d peaks become

slightly wider with increasing reduction temperature as in-

dicated in Figure 5-6. Based on the data of Paparazzo and

co-workers [97], this is probably due to partial conversion

of SnOz to SnO. After reduction at 250 C it appears that

the low-binding-energy shoulder near the baseline increases



a oz




r-4 0 in


0 4

$4 .4J$
44 E4
$4 l

4J d)


X go
14 I

o Ln

0 0O

0$4 4J

~W $4

as well suggesting that some of the Sn oxides are reduced

to metallic Sn. Metallic Sn probably forms during reduc-

tion at 125 OC, but the amount is so small that it is not

detected by XPS. Recall that similar changes in the Sn 3d

peak shape were observed during reduction of platinized tin

oxide surfaces in Chapters 1 and 3. Some, but probably not

all, of the metallic Sn formed alloys with Pt. Paffett and

Windham [55] have found that Pt/Sn alloy formation occurs

at temperatures as low as 175 OC in their study of Sn

deposited on Pt(lll). Recently, tin oxide films (no Pt)

have been reduced in CO under conditions similar to those

used in this study [98]. The XPS data indicate that metal-

lic Sn does not form in this case. Thus, it appears that

Pt promotes the formation of metallic Sn during reduction

in CO. The XPS data in Figure 5-6 also indicate that the

extent of reduction is greater at 250 OC than at 125 C.

It is interesting to relate the results of this

characterization study with those obtained from the commer-

cial Pt/SnOx catalyst (not supported on silica) inves-

tigated in Chapter 4. Both catalysts require a reductive

pretreatment for optimization of catalytic performance, but

increasing the reduction temperature affects the two

catalysts differently. The silica-supported catalyst ex-

amined in this study yields an optimum CO oxidation ac-

tivity after a 125 C pretreatment. The commercial Pt/SnOx

catalyst exhibits the same activity over a broad range of

pretreatment temperatures (125 to 225 oC) [26]. The XPS

data in this study and the characterization study in Chap-

ter 4 agree that Pt is present predominantly as hydroxides

after a pretreatment which yields high activity. This fact

is consistent with the observation that the catalytic ac-

tivity of the commercial Pt/SnOx catalyst is enhanced by

the presence of water vapor in the reaction gas mixture

[26]. Another similarity is that the data obtained from

both catalysts indicate that a Pt/Sn alloy forms during

reduction. This fact apparently explains why Pt supported

on tin oxide is greatly superior as a catalyst for CO

oxidation at low temperatures compared to Pt supported on

alumina or other nonreducible oxides.

Another factor relating to catalytic activity which

must be considered is encapsulation of the active Pt and Sn

species. This process does not occur when reducing the

commercial Pt/SnOx catalyst at temperatures up to 175 OC.

The impurities in the silica-supported Pt/SnOx catalyst

which cover the Pt and Sn species during the 250 OC reduc-

tion may be responsible for the loss in catalytic activity

since the encapsulation does not occur during the 125 C

reduction after which the catalytic activity remains high.

The catalyst surface produced during the reduction

step is not the same surface responsible for the long-term

catalytic behavior. Significant changes occur at the

catalyst surface during the initial reaction period. Based

on activity data [26], transient behavior may occur for

hours after which time the activity goes through a maximum

and then slowly decays. The changing activity must corre-

late with changes in composition and chemical states of

species at the surface, but this relationship has not been

determined yet.


The changes induced at an optimized, silica-supported

platinized tin oxide surface by annealing at 125 and 250 C

in 10 Torr of CO for 2 hours have been determined using ISS

and XPS. The reduction at 125 C results in enrichment of

the near-surface region in O, reduction of Pt oxides with

the formation of Pt hydroxide, and the reduction of SnOz to

SnO and metallic Sn which mostly alloys with Pt. Different

results are obtained from the surface reduced at 250 OC in

that the extent of the Sn oxide reduction is greater, the

predominant form of the Pt is metallic, impurities in the

silica migrate over the active Pt and Sn species, and the O

content of the near-surface region is reduced. The dif-

ferences found after reduction at the two temperatures may

explain the fact that a 125 OC pretreatment results in a

surface which is active for CO oxidation at low temperature


whereas a 250 OC pretreatment results in a catalytically

inactive surface.



The catalytic oxidation of CO near ambient tempera-

tures has many important applications. As described pre-

viously, closed-cycle COz lasers produce CO and Oz in the

laser discharge resulting in a rapid loss of output power.

This problem can be overcome by incorporating a low-

temperature CO oxidation catalyst into the laser system

which converts the dissociated products back into CO2

17,8,22,23]. Also, air purification devices often contain

catalysts to oxidize dangerous levels of toxic CO [17,99-

101]. Such devices are utilized in fire safety equipment

and in underground mines as respiratory aids.

Consequently, the development of low-temperature CO

oxidation catalysts has received considerable attention [7-

171. Although significant progress has been made with

regard to understanding the reaction mechanism, there

remains a need for the development of catalysts which ex-

hibit higher activities for prolonged periods at low tem-

peratures (typically less than 100 C) and in the diverse

range of oxidation environments which are encountered.

Factors which determine oxygen availability and gaseous im-

purities often have a pronounced effect on catalyst perfor-

mance. CO oxidation in ambient air has the advantages of

excess Oz and low COz concentrations which facilitate the

reaction considerably. Consequently, numerous materials

are known to oxidize CO in excess Oz at low temperatures

[9-11,13,141, but complications due to the presence of

humidity and/or air pollutants are often detrimental to

their activity. In COz lasers, CO and 02 are present in

small stoichiometric quantities in a large amount of COz.

Although the catalytic reaction benefits from the fact that

the lasers usually operate at temperatures somewhat above

ambient (25-100 OC), catalytic CO oxidation is difficult

under these conditions.

Recently, Haruta and co-workers [15,161 prepared

supported-gold catalysts on various base-metal oxides in-

cluding MnOx, FezOs, CoasO, NiO and CuO and determined

their catalytic activities toward the oxidation of Hz

and/or CO. Most of the CO oxidation tests were performed

using 1 vol% CO in dry air. At 0 C, Au/FezOs and Au/NiO

maintain essentially 100% CO conversion under the flow con-

ditions used over a 7-day test period. Similar performance

was also observed for Au/FezOs and Au/CosO* at 30 C in 76%

relative humidity. Therefore, these catalysts appear to be

quite useful in air purification devices, but activities in

the presence of air contaminants were not determined.

These catalysts may be useful in COe lasers even though the

reaction conditions are quite different as described above.

It is interesting to note that Haruta and co-workers

[15,161 apparently did not examine the behavior of Au/MnOx

toward CO oxidation.

Catalysts consisting of Pt and/or Pd supported on tin

oxide have been researched extensively for use in COz

lasers [7,8,22-26]. Although these materials can exhibit

considerable CO oxidation activity in this application,

there are complications which must be overcome. Acceptable

activity is observed only after the catalyst undergoes a

reductive pretreatment. Unfortunately, such pretreatments

may lead to considerable induction periods often lasting

several days during which the observed activity declines

before reaching a maximum [26]. Even after acceptable ac-

tivity is recovered these materials exhibit a steady decay

in performance over time.

The purpose of the present study is to explore the be-

havior of materials other than platinized tin oxide as

catalysts for low-temperature CO oxidation, particularly

with regard to COz laser applications. Several materials

were synthesized and screened for CO oxidation activity

using small concentrations of stoichiometric CO and z0 in

He and temperatures between 30 and 75 OC. The tests were

run for periods as long as 18000 minutes in order to ob-

serve the induction and decay characteristics of the


Catalysts Preparation

A review of the literature provided a basis for selection

of support materials examined in this study which include

iron oxide (Fe2Os), nonstoichiometric manganese oxides

(MnOx), and ceria (CeOx) where x is between 1 and 2. The

materials investigated were synthesized using established

impregnation and coprecipitation techniques [1021. The

samples prepared include MnOx, Pt/MnOx, Ag/MnOx, Pd/MnOx,

Cu/MnOx, Au/MnOx, Ru/MnOx, Au/CeOx and Au/FezOa.

The MnOx was used as-received from the Kerr-McGee Com-

pany, USA. It was prepared by the electrolytic oxidation

of manganous sulfate and has BET surface area of 74

m2/gram. The MnOx served as a sample itself as well as an

impregnation support for other materials.

Two Pt/MnOx samples (0.2 wt% Pt) were prepared by im-

pregnation of MnOx using an aqueous solution of NazPt(OH)e.

Sample #1 was dried in air at 280 OC for 4.5 hours whereas

sample #2 was dried in air at 75 C for 3 hours. A Pd/MnOx

catalyst (0.2 wt% Pd) was prepared by impregnating MnOx

with an aqueous solution of PdClz. The product was dried

in air at 280 C for 4.5 hours.

Two Ag/MnOx samples were prepared, one containing 0.2

wt% Ag and another containing 1.0 wt% Ag. Impregnation of

MnOx with Ag was accomplished using a solution prepared by

dissolving AgO in NH4OH. The products were dried in air at

280 C for 4.5 hours.

A sample which contained admixtures of CuO and MnOx

was prepared from the products of several procedures. Pro-

cedure A involved coprecipitation from aqueous solutions of

CuSO4 + sucrose and KMnO4. The precipitate was washed with

water and dried in air at 105 OC for 15 hours. The Cu:Mn

molar ratio was approximately 1.4. In procedure B, a por-

tion of the former product was dried in air at 280 C for 2

hours. In procedure C, MnOx was precipitated from aqueous

solutions of sucrose and KMnO4. The precipitate was washed

and dried as outlined in procedure A. The final product

consisted of an admixture of 0.4 grams from procedure A,

0.7 grams from procedure B, 0.3 grams from procedure C and

0.2 grams of commercial CuO powder.

A technique in which Mn(OH)z was precipitated in the

presence of fine Ru powder was utilized to prepare a 2 wt%

Ru/MnOx sample. An aqueous solution of Mn(NOs)2 was added

dropwise to a stirred mixture of Ru powder in NHeOH. The

resulting product was dried and calcined at 400 OC for 2


Three supported Au samples were synthesized via

coprecipitation from aqueous HAuClI and the nitrate of the

corresponding support metal. The composition of the

materials is approximately 5 at% Au/MnOx, 20 at% Au/CeOx

and 5 at% Au/FezO3 on a Au/metal basis. In each case the

appropriate precursor solutions were added dropwise to a

stirred solution of NazCOs at room temperature. After

washing with hot water (80 oC) and drying, the pre-

cipitates were calcined in air at 400 OC for 4 hours. Two

Au/FezOs samples were prepared which differed only in the

temperature of the wash water utilized (25 C and 80 C).


The reactor used to test the CO oxidation activity of

the catalysts is located at NASA Langley Research Center in

Hampton, VA, and has been described previously [27,103].

Screening of catalysts for CO oxidation has typically been

performed using a test gas consisting of a few percent CO

in air (excess Oz), and the catalytic behavior under

stoichiometric CO and Oz and in the presence of COz has of-

ten not been determined. Since the catalytic behavior can

vary considerably under different environments as described

above, it is necessary to perform such experiments. All

tests were conducted using 0.15 grams of catalyst and a

reaction gas mixture consisting of 1 vol% CO, 0.5 vol% 02

and 2 vol% Ne in He at a total pressure of 1 atmosphere

flowing at 10 standard cm3 per minute (sccm). The reaction

temperatures investigated were 75, 50 and 30 OC as noted.

The conversions are quite high under these conditions which

corresponds to operating the reactor in an integral mode.

In most cases the catalysts were tested as prepared

without additional pretreatments. Unless noted otherwise,

each catalyst was loaded into the reactor and exposed to

flowing He for about 1 hour as the reaction temperature

stabilized. Then the He flow was changed to the reaction

gas mixture and product sampling was begun. At predeter-

mined time intervals, an automated sampling valve directed

a small fraction of the reaction products to a gas

chromatograph (GC) for analysis of %COz yield, %CO loss and

%0 loss, and the results were plotted versus time.

Results and Discussion

During these initial activity screening experiments,

emphasis is placed upon characteristics of the overall CO

oxidation activity curves with respect to temperature and

time. An appropriate catalyst for use in COz lasers must

not only exhibit high activity at low temperatures (25-

100 C) but also maintain acceptable activity over a

lifetime of up to 3 years [6]. Since a catalyst cannot be

practically tested for a 3-year lifetime, its activity

profile must be extrapolated with reasonable confidence.

Nevertheless, it is necessary to exercise caution when

evaluating potential catalysts for COz lasers because a

catalyst which exhibits the best activity initially might

succumb to decay mechanisms which render it inferior after

extended use. Consequently, a catalyst exhibiting only

marginal activity initially may become the optimal choice

if the corresponding activity decay remains negligible.

Carbon monoxide oxidation activity curves for several

MnOx-based catalysts appear in Figure 6-1. Initially, MnOx

and Cu/MnOx exhibit the highest CO oxidation activities al-

though their performance rapidly deteriorates. However,

after about 2000 minutes the reaction curve for Cu/MnOx ap-

pears to approach a steady-state conversion with negligible

activity decay. The MnOx sample may approach a more active

steady-state conversion but more extensive testing is

required to be certain. Even though the Pt/MnOx #1 and

Ag/MnOx samples display superior activity throughout most

of the test period, extrapolation of the data in Figure 6-1

indicates that Cu/MnOx may be the optimal catalyst in a

long-term test.

The data in Figure 6-1 also depict an interesting com-

parison between the catalytic activities of Pt/MnOx #1

(dried for 4.5 hours at 280 C) and Pt/MnOx #2 (dried for 3

hours at 75 OC). The poor activity exhibited by Pt/MnOx #2

may be the result of incomplete removal of surface im-

purities (such as Na, Cl or OH-) associated with





a <3

aI 4

'1 0

i- CM
S x x %
"O O

E 525

0 o <
0~ I ;r

0I Q~







I I I r


4 d
N 4J

p x



0 x
0 1S

a -,-
4Jd 0


E- tPu
C. 4J

43 < a


M -


us en

the impregnation step. However, as found in previous

studies of MnOz and MnOz-CuO catalysts [104-1063, the inac-

tivity is most likely the result of incomplete surface ac-

tivation. Many MnOx-based catalysts usually require heat-

ing between 100-200 OC in air or Oz to produce an active

surface. The heat treatments apparently activate the sur-

face through the creation of reactive sites via partial

surface reduction, depletion of adsorbed water or surface

hydroxyl groups, and/or concurrent micropore generation.

An interesting observation is that the reaction

profiles of Pt/MnOx #1 and Ag/MnOx are remarkably similar.

This is unexpected based on the different catalytic

properties of Pt and Ag. It is possible that this behavior

results primarily from exposure of MnOx to similar basic

solutions followed by drying in air at 280 OC for 4.5

hours. The activity curve for pure MnOx (as-received and

common to both samples) is quite different in character,

which is consistent with this hypothesis. Nevertheless,

both materials performed well during the 10000-minute test

period oxidizing 70-80% of available CO at 75 OC.

As mentioned above, Pt/SnOx catalysts have received

considerable attention for use in COz lasers. A comparison

of CO oxidation performance between Pt/MnOx #1 (see Figure

6-1) and a commercial Pt/SnOx catalyst manufactured by

Engelhard Industries (see Chapter 4) is shown in Figure

00 Pl"eA %



4 o"






* u


4 *


0 4P


6-2. At 75 OC, the Pt/MnOx #1 sample exhibits superior

activity after approximately 2500 minutes of reaction. Due

to the limited reaction data for the Pt/SnOx sample, fur-

ther comparisons require data extrapolation. Assuming that

the indicated trends continue, Pt/MnOx #1 represents the

optimal catalyst over an extended time period. This also

is true for Ag/MnOx which behaves identically to Pt/MnOx


Figure 6-2 represents a valid comparison because the

sample size and experimental parameters used were identical

in both tests. It should be noted, however, that the

Pt/SnOx was pretreated in a 5% CO/He stream at 225 OC for 1

hour prior to activity testing. As discussed in previous

chapters of this research, such reductive pretreatments

significantly enhance the performance of Pt/SnOx catalysts

[26]. The fact that no pretreatments were used for the

MnOx-based catalysts is an advantage. Furthermore, since

the precious metal loading for the Pt/MnOx #1 and Ag/MnOx

samples is only 0.2 wt% (compared with 2 wt% for the

Pt/SnOx catalyst), there also appears to be an economic ad-

vantage over the Engelhard Pt/SnOx catalyst. Of course,

the Ag/MnOx catalyst is the least costly.

The fact that reductive pretreatments activate Pt/SnOx

catalysts provided motivation to investigate the effects of

similar pretreatments on Pt/MnOx catalysts. Two

pretreatment conditions were used in which the Pt/MnOx #1

sample was exposed to 5% CO/He for 1 hour at 125 and

225 C. The effects on catalytic performance are shown in

Figure 6-3. It is clear that the pretreatments are

detrimental to the CO oxidation activity of Pt/MnOx. In

fact, the observed activity of Pt/MnOx decreases with in-

creasing pretreatment temperature, a trend opposite to that

which is observed for Pt/SnOx catalysts [261. A possible

explanation may involve the reducibility of the MnOx and

SnOx supports. It appears that catalysts based on these

materials require a certain degree of surface reduction for

optimal activity. There is evidence that a completely

dehydroxylated or an entirely oxygenated MnOx surface is

not active toward low-temperature CO oxidation [104,105].

Similarly, surface hydroxyl groups are believed to be in-

strumental in the CO oxidation mechanism over Pt/SnOx [26].

Given the relative instability of MnOx with respect to

SnOx, such an optimum degree of surface reduction most

likely results from milder pretreatments than those used to

generate the data shown in Figure 6-3. In fact, heat

treatments in air or Oz appear to be more beneficial for

MnOx CO oxidation catalysts [12,104,105,107,108]. Although

the CO reductive pretreatments at 125 and 225 OC are ap-

propriate for Pt/SnOx, they apparently are too severe for