Complete oxidation of methane at low temperature over noble metal based catalysts

Complete oxidation of methane at low temperature over noble metal based catalysts Quá trình oxy hóa của khí metan ở nhiệt độ thấp dùng các chất xúc tác kim loại quý Complete oxidation of methane at low temperature over noble metal based catalysts Quá trình oxy hóa của khí metan ở nhiệt độ thấp dùng các chất xúc tác kim loại quý

Complete oxidation of methane at low temperature

over noble metal based catalysts: a review

Patrick Gélin∗, Michel Primet

Laboratoire d’Application de la Chimie à l’Environnement, UMR CNRS 5634, Université Claude Bernard Lyon 1, Building Chevreul,

43 Boulevard du 11 Novembre 1918, 69622 Villeurbanne Cedex, France

Received 29 October 2001; received in revised form 20 March 2002; accepted 25 March 2002

Abstract

This review examines recent developments in the complete oxidation of methane at low temperature over noble metalbased catalysts in patents and open literature The abatement of natural gas vehicle (NGV) methane emissions is taken asone example among possible applications The review develops current ideas about the properties of palladium and platinumcatalysts supported on silica and alumina supports in the complete oxidation of methane under oxidising conditions, focusing

on low-temperature reaction conditions The influence of residual chloride ions on the catalytic activity, the kinetic aspects

of the oxidation of methane over these catalysts, the nature of the active sites, the influence of metal particle size and reactionproducts on the activity, the observed changes in catalytic activity with reaction time and the effect of sulphur containingcompounds are examined The latest studies concerned with improved palladium and platinum supported catalysts whichwould exhibit enhanced and stable catalytic activity at low temperature in the presence of water and sulphur containingcompounds are reported Possible routes for preparing catalysts able to meet future regulations concerning methane emissionsfrom lean-burn NGV vehicles are discussed

© 2002 Elsevier Science B.V All rights reserved

Keywords: Methane oxidation; Noble metals; Catalytic combustion; Low temperature; Lean-burn NGV; Natural gas; Emission abatement;

Platinum; Palladium; Catalyst poisoning; Water inhibition; Poisoning by sulphur and chlorine containing compounds; Kinetic studies; Silica; Alumina; Zirconia; Tin dioxide; Ceria; Ceria-zirconia solid solution; Zeolite; Aluminophosphate; Mixed oxide supports; Oxide additives; Bimetallic catalysts

1 Introduction

The catalytic combustion of methane has been

extensively studied as an alternative to conventional

thermal combustion and was reviewed [1–5] This

method was shown to be effective in producing energy

in gas turbine combustors, while reducing emissions

Many studies were devoted to the design of catalytic

∗Corresponding author Tel.:+33-4-72-43-11-48;

fax: +33-4-72-44-81-14.

E-mail address: patrick.gelin@univ-lyon1.fr (P G´elin).

materials able to withstand high temperatures in mospheres containing steam and oxygen Anothermain application of catalytic total oxidation of hy-drocarbons is the abatement of methane emissionsfrom natural gas or methane combustion devices, be-ing either catalytic or non-catalytic This would inturn cover a wide range of applications, such as e.g.the abatement of methane emissions from lean-burnnatural gas vehicles (NGVs)

at-Programs for the use of NGVs in urban areas,especially heavy-duty vehicles, are currently beingdeveloped very rapidly in most industrial countries.0926-3373/02/$ – see front matter © 2002 Elsevier Science B.V All rights reserved.

PII: S 0 9 2 6 - 3 3 7 3 ( 0 2 ) 0 0 0 7 6 - 0

Table 1

US federal heavy-duty emission limit values (g/hph)a (1 hph= 2.6856 × 106 J)

Effective date Vehicle type CO Hydrocarbon NMHC + NOx NOx Particulate

The need for governments to diversify energy sources

together with the huge world-wide resources of

natu-ral gas (much larger than crude oil) partially explains

this fact In addition to economical and/or political

reasons, the use of compressed natural gas (CNG)

for automotive applications offers significant

environ-mental advantages over gasoline and diesel Natural

gas engines can operate under lean conditions so that

the fuel efficiency can be much increased compared to

stoichiometric conditions Under lean-burn conditions,

nitrogen oxides (NOx) emissions of CNG engines

are much reduced This is due to cooler combustion

resulting from the high air to fuel ratios at which the

Table 2

European heavy-duty diesel and gas emission limit values (g/kWh)

matter

Smoke (m −1)

b Directive 1999/96/EC of 13 December 1999.

c CH 4 for natural gas engine only.

d A special low-emission vehicle class (environmentally enhanced vehicle, EEV) is defined Tax incentives can be granted for vehicles complying with these requirements.

lean engines operate Typically, NOx emissions of aEuropean diesel bus meet the final Euro III standard(since October 2000), i.e 5 g/kWh with either station-ary cycle European stationary cycle (ESC) or Euro-pean transient cycle (ETC) For a lean-burn CNG bussubmitted to the same test cycles, NOxemission fallsdown to less than 2 g/kWh CO2 emissions are alsoreduced because of the high H:C ratio of the methanemolecule, which is the main component (85–95%) ofnatural gas Because of the very low sulphur content

of natural gas, NGVs SOx emissions are very low

In addition, the crucial advantage of NGV enginescompared to diesel is definitely the very low amount

of particulates (oil derived) in the exhaust gases For

these reasons, urban CNG buses appear attractive to

solve transportation problems in cities and reduce

pollution As a matter of fact, NGV programs are

cur-rently developed very rapidly in European countries,

NGVs replacing diesel vehicles as a viable approach

to reducing NOx and particulates in urban areas

The NGV advantages are however partially

bal-anced by the emission of unburned methane The

hydrocarbon composition in the exhaust gases of

lean-burn CNG engines reflects the composition of

natural gas in methane and non-methanic

hydrocar-bons (NMHCs), typically 90–95% methane Methane

is a potent greenhouse gas, which is recognised to

contribute more to global atmosphere warming than

carbon dioxide at equivalent emission rates, all the

more since its lifetime is quite long The

environmen-tal impact of NGV methane emissions is now being

taken into account in present and future regulations in

many countries[6].Tables 1 and 2report heavy-duty

emission limit values in US and in Europe It should

be noted that, besides differing limit values between

different countries, engine testing modes might be

also very different This point was clearly addressed

by Lampert et al in their evaluation of palladium

cata-lysts for methane emissions abatement from lean-burn

NGVs [7] A steady-state cycle, such as the ECE

R-49 test used in Europe before 2000, was shown to

result in better performance for hydrocarbons

emis-sions abatement than a transient cycle, such as the

US federal heavy-duty FTP test This is due to higher

NGV exhaust gas temperatures in steady-state cycles,

which favour higher conversions While steady-state

type testing is still used in Japan for emission

certifi-cation of heavy-duty engines, the situation in Europe

changed recently with the new transient ETC cycle,

which has been adopted for testing NGV heavy-duty

engines and turns to give emissions levels similar to

the US Federal heavy-duty FTP test

Without any after-treatment device, the total

hy-drocarbon emission from a lean-burn natural gas bus

reaches typically 3 g/kWh on the ESC test cycle,

which can be extrapolated to 4 g/kWh on the ETC

test These values are much higher than the limit value

of 1.6 or 1.7 g/kWh in Euro III and US standards

re-spectively for heavy-duty gas engines methane

emis-sions CO and NOxemissions are lower than the limit

values of the present regulations, even without any

exhaust gases after-treatment Therefore, emissions

of methane from NGVs must be necessarily reducedand this can be achieved by catalytic after-treatment

of exhaust gases which would perform the completeoxidation of methane At least 60% methane conver-sion is required to meet the most stringent currentregulations (Europe) As an example, a Europeanbus-maker proposes currently NGV buses equippedwith a 19 l catalytic exhaust converter, i.e almosttwice the engine capacity, loaded with 250 g/ft3noblemetal, which induces an extra cost of ca 3000 withrespect to the equivalent Euro III diesel vehicle.The approach is more difficult than for NMHCs be-cause of the higher stability of the methane molecule.Additional obstacles arise from the reaction conditionsspecific to lean-burn NGV engine exhausts:

• low temperatures at which the catalyst must operate

(typically less than 500–550◦C),

• low concentrations of methane (500–1000 ppm),

(15%),

• large excess of oxygen,

For low-temperature combustion applications, such

as in the abatement of methane emissions fromlean-burn NGV, it is clear that the thermal stability

of the catalyst is out of concern The main objective

is rather to design catalytic materials exhibiting thehighest activity at the lowest temperature and the bestresistance to poisons present in exhaust gases.The complete oxidation of methane can be per-formed over either noble metals or transition metaloxides These two families of catalysts have been ex-tensively studied during last decades in view of devel-oping catalytic combustion applications The main ad-vantage of noble metal catalysts over metal oxides isdefinitely their superior specific activity, which makethem as the best candidates for low-temperature com-bustion of hydrocarbons This is particularly true inthe case of methane which is the hydrocarbon the mostdifficult to activate Among noble metals, platinumand palladium are the most commonly used and stud-ied catalysts They can be obtained in a high degree ofdispersion when deposited on conventional supportswith a high specific area like silica or alumina Theincrease of the metal dispersion allows to improve thecatalytic activity

The aim of the present contribution is to review

the recent developments in the complete oxidation of

methane over noble metal based catalysts at low

tem-perature The abatement of lean-burn NGV methane

emissions is considered as one example It should

be noted that, however, very few references

actu-ally concern NGV emission conditions Interestingly,

contributions found both in the patents and the open

literature provide useful material to understand the

catalytic behaviour and provide ideas to find a good,

durable formulation for NGVs but also more

gener-ally for all low-temperature applications First, we

will develop the current ideas about the properties of

palladium and platinum catalysts supported on silica

and alumina supports in the complete oxidation of

methane mainly under oxidising conditions In this

section, we will examine successively the influence

of residual chloride ions on the catalytic activity, the

kinetic aspects of the oxidation of methane over these

catalysts, the nature of the active sites, the influence

of the metal particle size and the reaction products on

the activity, the observed changes in catalytic activity

with reaction time and the effect of sulphur containing

compounds All the aspects concerning the application

of these catalysts to combustion reactions at

tempera-tures exceeding 600◦C will be discarded InSection 2,

we will review the latest studies concerning improved

palladium and platinum supported catalysts which

would exhibit enhanced and stable catalytic activity

at low temperature We will endeavour to compare the

catalytic properties of the ‘improved’ catalysts to that

of the ‘reference’ catalysts described in this section

2 Silica- and alumina-supported Pt and Pd

catalysts

2.1 Poisoning effect of chlorine

The inhibiting effect of halogenated compounds on

the catalytic activity of supported palladium and

plat-inum catalysts with respect to the total oxidation of

methane was first reported by Cullis and Willatt [8]

The reaction was carried out in a pulse-flow reactor

un-der stoichiometric O2:CH4conditions After the

injec-tion of a pulse containing 1.8␮mol CH4, 3.6␮mol O2

and 0.14␮mol halogenated compound at 377◦C, the

activity decreased in all cases and was restored or not,

depending on the nature of the metal and/or the port, by flowing in helium and pulses of the CH4:O2mixture free of halogen compound Rich mixtureswere found to be more effective to restore the activ-ity Pd catalysts appeared more sensitive to poisoningthan Pt ones Moreover, when supported on alumina,the catalytic activity of Pt was fully restored while thedeactivation of Pd was irreversible No change of theparticle size was observed during these experiments

sup-No clear explanation for the decrease of the catalyticactivity was given Auger electron spectroscopy indi-cated that after exposure to dichloromethane, chlorineand carbon were found where palladium was presentand palladium was no more in the form of PdO XPSshowed a reduction of palladium to the metallic stateand the partial recovery of activity was associated with

a partial re-oxidation of Pd

Supported Pd and Pt catalysts are usually prepared

by impregnation of the support with Cl-containingmetal precursors It turns out that conventional acti-vation treatments (calcination in oxygen followed byreduction in hydrogen) do not allow the complete re-moval of chloride ions originating from the metal pre-cursor Commercially available supports might alsocontain significant of amounts of chloride themselves.Therefore, the question of whether residual chlorinestill present on Pd and Pt catalysts after conventionalactivation could inhibit their catalytic activity for ox-idation reactions was of concern[9]

The effects of chloride originating from the cursor salts or other impurities from the alumina onthe activity of Pd/Al2O3for the complete oxidation ofmethane were first addressed by Simone et al [10]

pre-A commercial␥-alumina containing 700 ppm Cl was

used It was clearly evidenced that Pd(NO3)2 led tocatalysts having a much better activity than PdCl2, al-though the PdCl2 prepared catalyst exhibited higherdispersion (from CO chemisorption) and smaller crys-tallite size (from TEM) The presence of chloride inthe latter catalyst was revealed by chemical analy-sis and XPS, thus explaining its low activity (videinfra) Moreover, the catalyst prepared from PdCl2showed significant improvements in performance afterlong-term ageing at 600◦C in air This was related tothe removal of chloride ions as revealed by the chem-ical analysis of the aged catalyst Various treatmentswere studied in order to improve the catalytic perfor-mance of the Pd catalysts prepared with Cl-containing

salts In the same paper [10], a loss of Pd after

age-ing in air at high temperature via the formation of

PdxOyClz complexes was proposed

Very recently, the inhibiting effect of residual

chlo-rine ions either originating from chlorinated precursor

or subsequent impregnation of Cl-free catalysts with

HCl on the catalytic activity of 2 wt.% Pd/Al2O3was

proved unambiguously [11] Experiments were

car-ried out with Cl-free and Cl-containing catalysts

hav-ing constant and equal dispersions (ca 10%) so that

sintering effects could be ruled out As previously

ob-served, the catalyst prepared from chlorinated

precur-sor strongly activated with time on stream at constant

temperature (475◦C, 1 vol.% methane, 4 vol.%

oxy-gen in nitrooxy-gen/helium carrier) The key point of the

experiment was the detection of HCl at the reactor

out-let and the in situ direct measurement of HCl

depar-ture could be correlated with the progressive increase

of catalytic activity (Fig 1) The activity of the Cl-free

Fig 1 In situ measurements of Cl departure and catalytic activity

in methane oxidation over a fresh Cl-containing 2 wt.% Pd/Al 2 O 3

catalyst (prepared from H 2 PdCl 4 , treated in O 2 at 500 ◦C and

reduced in H 2 at 300 ◦C) The variations of HCl, CH4 and H2O

concentrations at the reactor outlet are plotted as a function of

time on stream at 475 and 600 ◦C The feed (1 vol.% CH4, 4 vol.%

O 2 , He balance; GHSV = 15,000 h −1) was first introduced on

the catalyst (200 mg) preheated in He at 475 ◦C The dash line

indicates the time at which the catalyst was heated up to 600 ◦C.

(Reproduced from Fig 3 of [11] , with permission from Elsevier

Science.)

catalyst was finally reached when chloride ions werecompletely removed from the catalyst surface Aftersubsequent impregnation of the Cl-free catalyst withHCl, the activity was strongly decreased to the samelevel as the catalyst prepared with the chlorinated pre-cursor

Similar to palladium, chlorine was proposed tostrongly inhibit the catalytic activity of platinum par-ticles supported on alumina supports in the completeoxidation of methane under oxidising conditions(1 vol.% CH4:4 vol.% O2)[12,13] Samples preparedfrom H2PtCl6 still retained 0.3–0.8 wt.% Cl after re-duction in H2 at 350 or 500◦C The Cl-containing

Pt catalysts exhibited a strong activation under actants in spite of the decrease of the number of Ptsurface sites due to particle sintering After reachingthe steady-state activity, residual chlorine was com-pletely removed from the catalysts and Pt dispersiondid not change any more On the other hand, Cl-free

re-Pt catalysts exhibited a progressive deactivation withtime on stream, which was related to the sintering ofmetal particles The authors concluded that residualchlorine originating from Cl-containing Pt precursor

is responsible for a strong inhibition in the completeoxidation of methane, this effect being larger than thedecrease of the activity due to particle sintering Thecatalysts prepared from H2PtCl6and stabilised underreactants at 600◦C were found to be more active thanthe ones prepared from Cl-free precursors, probablybecause of a higher dispersion

The mechanism by which Cl ions would act asstrong inhibitors on catalytic activity is not yet clearlyestablished In the case of palladium catalysts, it wasproposed[11]that Cl ions, regardless of the way theywere introduced, would be mainly localised on thesupport after activation treatments, since satisfactoryagreement between dispersions measured by electronmicroscopy and chemisorption was obtained Duringcatalytic reaction, Cl ions desorb in the form of HCland evolved HCl was proposed to compete with re-actants on the metal active sites, blocking these sitesand inhibiting the activity For Pt/Al2O3prepared withchlorinated Pt precursor, Marceau et al.[13]suggestedthat, in addition to chloride ions siting on the support,

a second type of chlorine species would exist, beinglocated at the platinum–alumina interface possibly asbridging species between Pt and Al, and directly in-fluencing the adsorptive and catalytic properties of the

metal Although being in a small amount, this second

type of species would predominate, masking the

in-fluence of chloride ions siting on the alumina alone

Electronic effects affecting the surface properties of

the metal are put forward to explain the inhibition of

the activity by Cl

Even if it is now clear that Cl ions play an

inhibit-ing role in the total oxidation of methane on platinum

and palladium supported catalysts, more experimental

work is needed to fully understand how it affects the

catalytic properties of the metal In the case of

plat-inum catalysts, recognised to easily sinter under

ox-idising conditions, it would be worthwhile to check

whether the presence of chloride ions originating from

metal precursor play a role in the sintering process

It must be kept in mind for the next sections that

a proper comparison of the catalytic behaviour

mea-sured with varying catalysts will require to know

whether some residual chlorine is present or not at

the surface of the studied catalysts Unfortunately,

this aspect was not considered in many of the studies

and this will be mentioned whenever possible

2.2 Kinetic studies

The catalytic activity of alumina-supported Pt and

Pd catalysts was shown to depend on the O2:CH4

mo-lar ratio In an oxidising CH4:O2feedstream (O2:CH4

molar ratio greater than 2), Pd/Al2O3 is more active

than Pt/Al2O3[14,15] Under these conditions, the

ox-idation of methane is complete and carbon dioxide is

the only carbon-containing product Typically, in a He

flow containing 0.2 vol.% CH4and 1 vol.% O2with a

space velocity of 52,000 h−1, the temperature at half

conversion (T50) of a 0.2 wt.% Pt/Al2O3catalyst was

found more than 100◦C higher than that of a 0.16 wt.%

Pd Al2O3catalyst The presence of CO in the feed had

no effect on the catalytic activity, which is expected in

view of the reactivity of CO towards O2much higher

than that of CH4

Several kinetic studies of the complete oxidation

of methane over Pt and Pd supported catalysts were

reported in the literature [16–21] Except for

refer-ence[21], catalysts were prepared from Cl-containing

precursors and the evolution of the Cl content during

the study was not addressed Another difficulty arises

from the fact that, in some cases, reaction conditions

were changed from oxidising to reducing (lean to rich),

which might also affect the nature of the active sites(vide infra)

When performed on Pd/Al2O3 catalysts under idising or stoichiometric conditions, it can be statedthat the reaction is first order with respect to methaneand 0 (or almost 0) order with respect to oxygen con-centration This was observed at very high space ve-locities with Pd on Si-stabilised Al2O3[20] This wasalso recently established by van Giezen et al.[21]intheir thorough kinetic study of methane oxidation over

ox-a 7.3 wt.% PdO-on-ox-aluminox-a cox-atox-alyst A Cl-free cox-atox-a-lyst was prepared by impregnation of an alumina withPd(NH3)4(NO3)2solution and further calcined in air

cata-at 450◦C The rate of the reaction was measured tween 180 and 515◦C, after stabilisation of the cat-alyst in a reaction feed consisting of 1 vol.% CH4,

be-4 vol.% O2in helium The apparent activation energywas measured between 200 and 320◦C Special at-tention was paid to avoid thermal effects and diffu-sion limitations and it was checked that no deactiva-tion of the catalyst occurred While CO2was shown

to have no influence on the reaction rate (between 0and 5 vol.% CO2), a strong inhibition by H2O was ob-served When operating under dry conditions (withoutwater added except that produced by the reaction), theinhibition by H2O was shown to depend on conversion

An apparent activation energy of 86 kJ/mol was found,very close to those ranging between 70 and 90 kJ/molreported previously in the literature [22,23] Whenadding 2 vol.% H2O to the dry feed the water contentwas very slightly affected by the reaction and differ-ential conditions were reached An apparent activationenergy of 151± 15 kJ/mol was found Kinetic mea-

surements were carried out under wet conditions TheCH4concentration was varied between 0 and 6 vol.%and O2between 2 and 7 vol.% while maintaining sto-ichiometric or lean conditions, and 2 vol.% H2O Theorders with respect to methane and oxygen pressureswere 1.0 ± 0.1 and 0.1 ± 0.1, respectively.

In a study of 0.5 wt.% Pd catalysts on SiO2, Al2O3and silica-alumina, Muto et al reported slightly dif-ferent values of reaction orders [18] The catalyticactivity was measured at 400◦C after the activity wasstabilised at 450◦C for 12 h The CH

4 concentrationwas varied from 2 to 15 vol.% with 10 vol.% O2 andthe O2amount varied between 10 and 60 vol.% with

10 vol.% CH4 So that experimental conditions ied from rich to lean mixtures (reducing to oxidising

var-compositions) The orders with respect to methane

were 0.46, 0.53 and 0.58 for Pd/SiO2, Pd/Al2O3 and

Pd/SiO2-Al2O3 respectively while the orders with

respect to O2 were 0.14, 0.18 and 0, respectively It

is difficult to explain the low values of the order with

respect to methane It can be suspected that, in this

case, varying conditions from rich to lean mixtures

has some influence on the nature of active sites,

lead-ing to changes of the order with respect to methane

Concerning platinum based catalysts, the trend is

the same as for palladium catalysts: the order of the

reaction with respect to oxygen mostly tends toward

zero while the order with respect to methane would

be close to unity

This was first claimed by Cullis and Willatt in a

pioneering study of Pt (and Pd) catalysts on various

supports (Al2O3, TiO2, SnO2, ThO2) at temperatures

in the range 300–440◦C with O

2:CH4 ratios varyingfrom lean (10:1) to rich (1:10) values[16] The same

conclusion was reached by Niwa et al for 2 wt.%

Pt/Al2O3 under stoichiometric conditions[17] More

recently, Ma et al [19] reported the kinetics of

ox-idation of light hydrocarbons (methane, ethane and

propane) on Pt/␦-Al2O3 In this work, the catalyst was

prepared by impregnation of the support by

chloropla-tinic acid and kinetic studies were carried out under

differential conditions (less than 10% conversion) at

a space velocity of ca 35,000 h−1 The orders with

respect to methane and oxygen pressures were

re-spectively 0.95 and−0.17[19] Several models were

developed and the kinetics of methane oxidation was

best described by a Langmuir–Hinshelwood model in

which adsorbed methane molecules react with atomic

oxygen atoms[19]

2.3 Nature of the active sites

It is well known that platinum and palladium

ex-hibit very different reactivities towards oxygen In the

presence of oxygen, palladium and platinum oxidises

into PdO and PtO2respectively PdO forms between

ca 300–400◦C, being stable in air at atmospheric

pressure up to about 800◦C Above this temperature,

the stable species is metallic palladium By contrast,

PtO2is highly unstable: compared to PdO, it

decom-poses at a much lower temperature, around 400◦C In

addition, PtO2 is highly volatile and this property is

often considered to explain reconstruction of platinum

surfaces under oxygen atmosphere by transport of Pt

in the form of PtO2 over nanometric distances Thissuggests that PdO forms easily in oxidising atmo-sphere, while Pt would mostly remain at the metallicstate under the same conditions It seems that thesame ideas would apply to supported Pt and Pd cat-alysts Supported Pd catalysts were shown to adsorbbetween 80 and 600◦C much higher oxygen amountsthan the corresponding Pt ones (e.g 100 times for analumina supported catalyst)[16] It is also establishedthat Pt0 forms even after calcination of supported Ptprecursors in oxygen (or in air) at 500◦C These ideasare addressed in this section

2.3.1 Palladium catalysts

When supported on carriers having a high cific surface area, the thermal stability of PdO underoxygen atmosphere exhibit significant variations de-pending on the nature of the support used Farrauto

spe-et al [24] studied the thermal stability of PdO ported on alumina in air (1 bar) by thermo-gravimetricanalysis (TGA) They found that a freshly preparedPdO/Al2O3(4 wt.% Pd sample prepared from the ni-trate salt) decomposed in air between 800 and 850◦C.However, a surprising feature is that, once PdO isdecomposed, temperatures well below 650◦C arerequired for its re-formation The influence of vari-ous oxide supports other than alumina, Ta2O3, TiO2,CeO2and ZrO2, on the thermal stability of supportedPdO and the reformation of PdO from Pd was ex-amined[25] The temperature of PdO decompositionwas found to vary from one support to another Forexample, PdO on ZrO2decomposes more than 100◦Cbelow the temperature of decomposition measuredfor PdO on the other supports (alumina included), all

sup-of which show only tiny variations sup-of PdO position temperatures The existence of significantsupport–metal oxide interactions was proposed Onthe other hand, regeneration of PdO from Pd metal inthe cooling step of temperature cycles was also found

decom-to be strongly dependent on the support TiO2 andCeO2were thus shown to induce a sharp increase inthe temperature of PdO reformation (ca 130◦C) withrespect to Al2O3 These supports were concluded toincrease the temperature domain for which PdO isstable, compared to Al2O3or ZrO2 It must be pointedout that for applications involving temperatures lowerthan typically 600◦C, which is the topic of the present

review, PdO is stable, so that its regeneration is out

of concern

In spite of these effects, it is clear that in the

pres-ence of 2–4 vol.% oxygen PdO is the

thermodynami-cally stable phase in the temperature range where the

catalyst is active, starting from 300◦C up to at least

600◦C However, the kinetics of the oxidation process

may limit the extent of oxidation Cullis and Willatt

[16]measured by a pulse method and the adsorption of

oxygen on Pd supported on various supports (Al2O3

being included in these supports) at various

tempera-tures Large amounts of oxygen were consumed,

com-pared to platinum catalysts and this was consistent

with the transformation of Pd metal into palladium

oxide to some extent The phenomenon was found

to be dependent on the support and the temperature

The amount of adsorbed oxygen increased with

tem-perature, e.g an optimum temperature for oxygen

ad-sorption on Pd/Al2O3being 600◦C The influence of

dispersion on the oxygen uptake of Pd/Al2O3catalysts

was studied by Hicks et al.[26] The O2 adsorption

was carried out at 300◦C for Pd particles supported on

alumina in which the dispersion varied from 3 to 80%

Bulk palladium oxidised and the extent of oxidation

was found to be dependent on the dispersion: the lower

the dispersion, the lower the Pd oxidation extent

The mechanism by which Pd oxidises into PdO is

still unclear For example, Chen and Ruckenstein[27]

and Jacobs and Schryvers[28]have studied the

reac-tion of oxygen with supported palladium particles by

electron microscopy At 350◦C, the particle size is

re-tained At 500◦C, extensive fragmentation and

spread-ing of all particles were observed Oxidised particles

contained mostly fcc metal and no crystalline PdO was

observed[28] A mechanism was proposed, initiated

by lattice imperfections and developing cracks and

fis-sures Once an oxide film is formed the reaction slows

down dramatically Metal particles are fragmented and

porous, covered with a thin layer of oxide not

de-tectable In a more recent study, Voogt et al.[29]

inves-tigated the oxidation of palladium model catalysts by

XPS analysis Palladium particles of 5 and 8 nm

sup-ported on SiO2/Si(1 0 0) were studied The oxidation

was modelled by a stepwise growing of the oxide layer

around the core of the metal particle The thickness of

the oxide layer formed during oxidation was found to

increase linearly with time The rate of the oxidation

was strongly dependent on temperature The

activa-tion energy for the oxidaactiva-tion would be at least equal to

100 kJ/mol It was proposed that the rate-limiting step

in the process is the lattice reconstruction needed forthe formation of a new oxide layer at the oxide–metalinterface A 8 wt.% Pd/SiO2catalyst behaved similar

to the model catalysts

It is now generally agreed that, under reactionconditions, especially in oxygen-rich atmosphere,palladium oxide is formed and represents the mainactive phase Several studies clearly establish the im-portance of PdO for methane oxidation [24,30–37].The occurrence of a reversible PdO ↔ Pd0 trans-formation in the presence of oxygen was clearlyestablished by TGA and temperature programmeddecomposition–temperature programmed oxidation(TPD–TPO) experiments [24,37] Fig 2 shows thetypical O2 concentration profile obtained in aTPD–TPO experiment upon heating (from 200 up to

900◦C at a heating rate of 15◦C/min in 1% O

2 inHe) a PdO catalyst supported on La-stabilised alu-mina and upon its subsequent cooling[37] At least

Fig 2 Comparison between the catalytic activity in CH 4 oxidation

of an alumina-supported Pd catalyst and the O 2 profile measured in

a TPO experiment The Pd catalyst was prepared by impregnation

of a La-doped Al 2 O 3 by Pd(NO 3 ) 2 (Pd loading = 5 wt.% Pd), calcined at 1000 ◦C and deposited as a thin layer on an annular tube

reactor For the reaction test, the feed composition was 5000 ppm

CH 4 , 2 vol.% O 2 , He balance, (GHSV = 1,100,000 cm 3 /g cat h) The conversion curves correspond to a second temperature cycle between 200 and 900 ◦C The TPO profile was obtained with

the same coated catalyst with 1 vol.% O 2 in helium (flow rate

120 cm 3 /min) (Reproduced from Fig 3 of [37] , with permission from Elsevier Science.)

two decomposition peaks (appearing positive) can be

observed above 700◦C on heating ramp while one

re-formation peak below 500◦C (appearing negative)

is observed on cooling This indicates that there is a

temperature range in the cool-down step of the cycle

where PdO is not re-formed, palladium being in the

metallic state This observation together with the

as-sumption that metallic palladium is much less active

than PdO was used to explain the catalytic behaviour

during the heating-cooling cycle, as reported inFig 2

for comparison A strong drop of activity during the

cooling-down step is observed in the range where Pd

is still at the metallic state until PdO re-forms, thus

leading to a catalytic activity identical to that obtained

during the heating-up step, before PdO

decomposi-tion It is clear that, depending on the experimental

conditions used for catalytic testing, the activity drop

can be more or less pronounced[24,37] Moreover,

the activity and TPO curves cannot be strictly

com-pared one to the other since being obtained in slightly

different conditions (different oxygen concentrations

and very different heating/cooling rates) This would

explain why the activity starts to decrease before PdO

decomposition in TPO on heating and, conversely, the

CH4 conversion is restored before PdO re-formation

on cooling Anyway, these results remarkably support

the idea that palladium in the metallic state has a

much lower activity than the oxide form

Burch and Urbano compared the reactivity of

oxy-gen chemisorbed on Pd metal to that of oxide ions for

Pd/Al2O3catalysts [31] A 4 wt.% Pd/Al2O3 sample

was prepared by impregnating the alumina with

pal-ladium nitrate, dried at 120◦C and then submitted to

a wide range of oxidation and reduction treatments at

500◦C before the catalytic activity (1 vol.% CH4 in

air) was measured as a function of time on stream at

300◦C Catalytic steady-state and pulse experiments

on the pre-reduced samples clearly indicated that

metallic palladium is not active while pre-oxidised

catalyst is active The study of the oxidation at 300◦C

of the pre-reduced samples revealed that the oxidation

always proceeds via a very fast formation of a

mono-layer of oxygen followed by a slower oxidation step

leading to almost complete oxidation of palladium

The question of what is the optimum state of PdOx

between chemisorbed oxygen on Pd metal, a PdO skin

on a Pd metal core or bulk PdO was further addressed

by comparing the evolution of methane oxidation

Fig 3 Comparison of the methane conversion at 300 ◦C vs time

and the oxygen uptake at 300 ◦C vs time for a 4 wt.% Pd/Al2O3

catalyst pre-reduced in H 2 at 300 ◦C Results were obtained in

separate experiments Composition of the reaction feed: 1 vol.%

CH 4 in air The oxygen uptake was measured by a pulse-flow experiment (Reproduced from Fig 1 of [33] , with permission from Elsevier Science.)

activity and oxygen uptake at the same temperature(300◦C) as a function of time[32,33] It was assumedthat the same Pd surface state was reached underreactants or in oxygen atmosphere The results areshown inFig 3 A chemisorbed monolayer of oxygenformed very quickly at 300◦C while the activity waslow It was deduced that oxygen chemisorbed on Pdmetal is poorly active Further exposure to oxygen led

to a slower oxidation of the Pd up to the almost plete oxidation to bulk PdO The activity increasedsimultaneously to reach a plateau at the point where70–75% of the complete oxidation of Pd into PdOwas achieved It was concluded that fully oxidisedbulk PdO is the optimum state for methane oxidationand the intermediate state corresponding to a ‘skin’

com-of PdO on a Pd metal core has no greater activitythan bulk PdO

For other authors, both PdO and Pd may be present

on the catalysts under reaction conditions Lyubovskyand Pfefferle[38]and Datye et al.[39]studied the de-composition of PdO into Pd in Pd catalysts supported

on␣-Al2O3plates and its reformation during ature cycling The formation of highly dispersed PdOclusters in thermodynamic equilibrium with the previ-ously formed metallic Pd surface after pre-treatment

temper-at tempertemper-atures above 800◦C was first suggested[38].

More recently[39], it was proposed that the PdO→

Pd transformation is initiated at the surface of PdO

and causes small domains of Pd metal to form on

the surface of PdO These small domains are easy to

re-oxidise upon cooling, which is not the case when

complete transformation into Pd metal is achieved

Strongly bound oxygen at the surface of Pd would

in-hibit bulk oxidation The re-oxidation of Pd when it

does occur would involve the oxide growing in patches

on the metal The re-oxidation via the growth of a

thicker oxide film (shrinking core of Pd metal) was

not retained as the most likely mechanism Complete

oxidation of the Pd metal would lead to the

forma-tion of polycrystalline PdO with a roughening of the

particle surfaces Re-activation of the Pd metal

cat-alysts upon cooling could be associated to two

pos-sible mechanisms: PdO re-formation (increase in the

fraction of Pd metal particles fully re-oxidised into

bulk PdO) or a reaction mechanism which involves

metallic Pd

It is worthwhile noticing a recent work studying the

influence of the oxygen content in supported Pd/PdO

particles on the activity in methane oxidation, even

though the conventional alumina support was

substi-tuted in this case by a ceria-zirconia support [40]

A 3 wt.% Pd/CeO2-ZrO2(containing 10% ceria) was

prepared by impregnation of the support with

palla-dium nitrate The catalyst was calcined at 500◦C in

O2 Catalytic testing was performed with pulses of

a 1:4 CH4:O2 mixture in helium after cycling up to

900◦C in a continuous flow of the same reaction

mix-ture The degree of reduction was varied by controlled

chemical reduction with methane Interestingly, slight

reduction was observed to improve the catalytic

activ-ity compared to either fully oxidised or fully reduced

metallic particles In addition, the partially reduced

sample was found to be easier to re-oxidise than the

completely reduced one

Even though there seems to be a general agreement

on the fact that PdO is formed under oxygen-rich

reaction conditions, some parameters such as the

pre-treatment history, the metal dispersion and the

composition of the reaction mixture seem to have

also some influence on the catalytic behaviour More

experimental work is still needed to fully understand

the exact nature of the active sites under working

con-ditions Attempts to correlate the activity in methane

oxidation with parameters such as the reactivity of

PdOx species towards methane or the extent of dation were unsuccessful so far

oxi-2.3.2 Platinum catalysts

Fewer studies were devoted to platinum catalystscompared to palladium ones For platinum catalysts, itseems that a distinction between large and small sup-ported particles should be made Experimental works

on Pt single crystals have shown that oxygen reactswith at most the top two layers of the surface, dissocia-tively chemisorbing onto the platinum surface below

500◦C[41] When supported, small and large Pt ticles may co-exist Hicks et al.[26]studied aluminasupported Pt catalysts prepared from Cl containingprecursors IR spectra of adsorbed carbon monoxide atsaturation coverage were used to probe the metal sur-face before and after exposure to reaction conditions It

par-is noteworthy that the catalysts were reduced in H2andevacuated in vacuo at 300◦C prior to CO adsorption.Two bands were observed at ca 2080 and 2070 cm−1,which were attributed to CO linearly bonded to sur-face platinum atoms These bands were associated

to the two phases of platinum on alumina proposed

in previous studies ([42]and references herein) Thehigh frequency band would be related to a crystallinephase weakly interacting with the support and respon-sible for high reaction rates The low frequency bandwould be due to a highly dispersed phase much lessactive for methane oxidation In the absence of any insitu characterisation of the catalysts, it was proposed

[26]that under reaction conditions, two types of idised species formed, associated to two TPR peaks:

ox-a peox-ak observed ox-at low temperox-ature would be due tooxygen dissociatively chemisorbed on large crystal-lites while a TPR peak at higher temperatures wouldcorrespond to a dispersed PtO2phase, less reactive to-wards methane oxidation The formation of PtO2wasreported above 300◦C in the case of 100% dispersedparticles[43]

Recently, Hwang and Yeh studied the various Pt-Ox

species formed on oxidation of reduced Pt/␥-Al2O3

[44] and Pt/SiO2 [45] by TPR technique Sampleswere prepared by impregnating the support with PtCl4solution Higher dispersions (100 and 65%) wereobtained with alumina compared to silica supportedcatalysts (30 and 50%) For Pt/Al2O3 catalysts, es-sentially four different forms of oxidised Pt specieswere proposed to form depending on the temperature

of oxidation: PtS–O at room temperature (PtSbeing a

surface platinum atom), PtO at 100◦C, PtO

2at 300◦Cand PtAl2O4 at 600◦C On SiO

2, a decrease of theO/Pt stoichiometry was observed and attributed to the

decomposition of PtO and PtO2 X-Ray photoelectron

spectroscopy confirmed the co-existence of these two

oxides below 400◦C The authors mentioned a loss of

Pt determined by ICP analysis upon oxidation above

400◦C This is unlikely since transport of platinum

should be restricted to atomic distances because of

the high volatility associated with the low stability

of PtO2 No such a loss of platinum was mentioned

elsewhere in the literature It is worthwhile to notice

that these studies were carried out on highly dispersed

samples which are known to sinter severely with time

on stream in the conditions of the reaction It would

be more convenient to study less dispersed samples

According to Burch and Loader, the extent of

ox-idation of the platinum surface would be a key

fac-tor on the catalytic behaviour, a less oxidised

plat-inum surface being more active compared to a more

oxidised surface [15] On this basis, operating under

methane-rich conditions is expected to lower the

ox-idation extent of Pt surface As a result, Pt/Al2O3

exhibited higher activity than Pd/Al2O3 catalysts in

the combustion of methane under reducing conditions

[15]

2.3.3 Mechanistic considerations

The different states of palladium and platinum

un-der reaction conditions being taken into account, two

different mechanisms for the oxidation of methane on

these two metals were proposed, involving two

dif-ferent ways for the activation of the C–H bond of

methane[46] It was proposed that Pt would activate

the almost non-polar C–H bonds of methane through a

homolytic mechanism (dissociative adsorption of CH4

at free metal sites), oxygen species acting as an

in-hibitor for the reaction at full coverage This would

agree with the optimum and very high activity under

rich conditions and a poor activity under lean

con-ditions By contrast, Pd is much more effective than

Pt under lean conditions Pd would be fully oxidised

and Pd2+ O2 − ion pairs at the surface of PdO would

activate the C–H bonds by a heterolytic mechanism

[46], similar to that proposed by Choudhary and Rane

on oxide catalysts [47] However, the metal and the

metal oxide were thought to be in dynamic equilibrium

depending on temperature, gas composition, the sition Pd/PdO being slow The activity of a 0.5 wt.%Pd/Al2O3 catalyst was exposed to a reaction mixture

tran-of composition oscillating between rich and lean ditions was found to be much lower than that reachedunder steady-state conditions (either rich or lean)

con-2.4 Particle size effect

Some early studies mentioned that the plete oxidation of methane over Pd and Pt catalystssupported on alumina could be structure sensitive

com-[9,23,26,48–50] since very strong variations of thecatalytic activity were observed with varying disper-sions These conclusions were drawn from the calcu-lation of turn-over frequencies (TOFs), the number

of active sites being measured by hydrogen uptake

of the reduced catalysts For example, Hicks et al

[26] measured the catalytic activity of supported Ptand Pd catalysts under slightly oxygen-rich condi-tions (O2:CH4 = 2.2) TOFs were calculated using

the initial dispersion determined by chemisorption

of hydrogen on the reduced samples before sure to reaction conditions The mean steady-stateTOFs at 335◦C were found to vary as follows forthe different catalysts: dispersed phase of platinum,

TOF= 0.08 s−1, small particles of palladium; TOF=

0.02 s−1, large particles of palladium; TOF= 1.3 s−1.The structure sensitivity was related to differences

in the reactivity of adsorbed oxygen A more carefulinvestigation of Pd catalysts seemed to confirm thestructure sensitivity of Pd towards methane oxida-tion [9] A series of Pd/Al2O3 catalysts of varying

Pd dispersions (2–74 nm average diameter from drogen chemisorption) were obtained by varyingthe calcination temperature and were tested undercontinuous-flow conditions, in the complete methaneoxidation in an oxygen rich atmosphere (1 vol.% CH4

hy-in air)[23] Wide activity variations which could not

be attributed to support effects were observed Noclear relationship between palladium particle size andreaction rate was established[23]

The ‘structure sensitivity’, proposed in these studiesmust be qualified, however, for two reasons All cat-alysts were prepared from Cl-containing precursors.Chlorine which is known to strongly inhibit the re-action of methane oxidation could affect the reaction

rate in a uncontrolled manner dependent on its

re-moval from the catalyst surface The second argument

concerns the determination of the number of active

sites from which TOF are calculated The number of

active sites is determined from the H2uptake by the

reduced metal particles For platinum particles, this

number is likely to be the same under working

condi-tions, at least for large particles, since the particle core

is likely to be at the metallic state However, for

palla-dium catalysts, the number of PdO sites accessible to

methane and involved in the reaction might differ from

that measured in the reduced metallic state (Section

2.3) Burch and Urbano[31]did not find any

correla-tion between either the initial or steady-state activity

and the amount of exposed palladium oxide surface,

as determined by chemisorption on reduced samples

But long-term morphological changes (especially in

the presence of water) are suggested to account for

varying catalytic behaviour

The opposite conclusion, i.e the structure

insensi-tivity, was drawn from activity and dispersion

mea-surements of other studies [15,16,51,52] After

pro-longed heating (40 days) in O2at 550◦C of 2.7 wt.%

Pd/␥-Al2O3, the sintering of particles was evidenced

by electron microscopy but this did not induce any

variation in the catalytic activity[16] With Pt/Al2O3

catalysts having particles in the range 1.4–3.7 nm

di-ameter, Burch and Loader[15]found no evidence of

a particle size effect in contrast to previous works

mentioning that the complete oxidation of methane is

structure sensitive[48,51,53] In the absence of any Cl

impurities originating from the Pd precursor, Hoyos

et al.[51]observed a slight increase of the catalytic

activity of Pd/SiO2(2.2 wt.% Pd, 1% CH4:4% O2in

nitrogen) in spite of the decrease of the metallic

dis-persion from 34 to 17% (from H2chemisorption): the

light-off temperature (temperature at half conversion,

T50) decreasing from 309 to 304 ◦C.

For the determination of how the turn over rate

varies with the structure of the catalyst, the average Pd

particle size on a series of Pd catalysts supported on

various supports (ZrO2, Al2O3and Si-Al2O3) was

var-ied between 2 and 130 nm[22] The turn over rate was

calculated at steady state (24 h reaction) on the basis

of Pd dispersion measured after reaction The values

ranged in the interval 2×10−2to 8×10−2s−1 It was

concluded that the reaction was ‘structure insensitive’

but this does not exclude some variation of the activity

with the particle size On two Cl-free Pt catalysts bilised under reactants at 600◦C for 60 h (1% CH

sta-4:4%

O2) and exhibiting 6 and 14% dispersion, respectively.Marceau et al.[12]found TOF of 0.15 and 0.08 s−1,respectively, suggesting an increase of TOF with themetal particle size

2.5 Influence of the reaction products on the activity

As already mentioned in Section 2.2, water eitherbeing present as a product of the methane conversion

or as an adduct to the reaction feed has an inhibitingeffect on the methane oxidation rate over palladiumcatalysts Cullis and Willatt[8]examined the catalyticactivity of a 2.7 wt.% Pd/␥-Al2O3 in a 2:1 O2:CH4mixture at 352◦C using pulse-flow experiments Wa-ter was added to the pulses containing 1.8␮mol CH4

in the range 0–55␮mol H2O, i.e 30 times the amount

of CH4 An apparent order of −0.8 with respect to

H2O was derived (Fig 4) but within experimentaluncertainties a variable negative order from −0.4 to

−1.3 could be also deduced from given data Ribeiro

et al [22] studied a 7.7 wt.% Pd/Si-Al2O3 prepared

by incipient wetness using an aqueous solution ofPd(NH3)2(NO2)2and calcined at 500◦C The depen-dence on H2O was measured at 277◦C by establishingexcess CO2(feed of 1% CH4, 0.25% CO2, balance air,with a CH4conversion level less than 3%) and vary-

Fig 4 The effect of H 2 O addition on the activity of a 2.7 wt.% Pd/Al 2 O 3 catalyst in CH 4 oxidation performed at 352 ◦C in a

pulse-flow reactor Composition of the reactant pulse: 1.8 ␮mol

CH 4 , 3.6 ␮mol O 2 (Reproduced from Fig 1 of [8] , with permission from Elsevier Science.)

ing the H2O concentration in the range 0.03–0.15%.

An order dependence of−0.98 was determined The

authors suggested a competition of H2O with CH4

for surface active sites leading to the formation of

Pd(OH)2 at PdO surface sites as first proposed by

Cullis et al.[54] The competition between methane

and water for the active sites was also suggested by van

Giezen et al.[21]in a kinetic study of the oxidation

of methane over a Cl-free PdO-on-alumina catalyst

In this study, the order with respect to H2O (within

the interval 0.6–3.1% H2O) measured in the

tempera-ture range 180–515◦C was found to vary with reaction

temperature in the range 0.8 ± 0.2.

The influence of water concentration on the rate of

oxidation of methane over Pd/Al2O3was also

inves-tigated by Burch et al.[52]at different temperatures

A 4 wt.% Pd/Al2O3was prepared from Pd(NO3)2and

calcined at 500◦C The feed consisted of 1% CH

4inair Prior to experiments the catalyst was held in the

reaction mixture at 300◦C for at least 12 h in order

to be sure that steady state had been reached An

in-hibiting effect decreasing with increasing temperature

was observed, becoming small above about 450◦C.

This effect was reversible Again, the existence of the

equilibrium PdO+ H2O = Pd(OH)2 was proposed,

where PdO represents the active phase and Pd(OH)2

an inactive state for methane oxidation On this basis,

it was postulated that the true rate determining step

could be the loss of H2O from Pd(OH)2instead of the

activation of the first C–H bond in methane Pulse

ex-periments in which small pulses of 1% CH4/air were

passed at 300◦C over 5% Pd/SiO

2maintained in driedair atmosphere revealed a very high and stable activ-

ity, much higher than the steady-state activity[32,33]

At 250◦C, some deactivation was observed with

in-creasing the number of pulses It was concluded that

the surface OH groups produced either by the reaction

or formed by H2O added to the feed stream are not

easily removed at 250◦C (respectively 300◦C) In this

process, the support might have some influence

Sup-ports with a greater affinity for H2O would initially

show higher activity because of the trapping of water

by the support But, it is also possible that at steady

state the local water concentration above PdO

parti-cles could be higher resulting in a lower reaction rate

The authors indicated that the activity of Pd/Al2O3

ini-tially higher becomes lower than the one of Pd/SiO2

at steady state

Water inhibition on the catalytic activity of nia supported Pd catalysts in methane oxidation wasalso observed [55,56] Reaching the water adsorp-tion/desorption equilibrium was found to be slow com-pared to the methane oxidation time scale, especially

zirco-at low temperzirco-atures In addition the time required forreaching the equilibrium strongly depended on tem-perature This was reflected in the strong water inhi-bition on catalytic activity observed in pulse experi-ments at lower temperatures

No influence of CO2 on the catalytic activity ofPd/Al2O3 catalysts could be observed [22,57,58] Inthe study of Ribeiro et al [22], the dependence ofthe activity of a 7.7 wt.% Pd/Si-Al2O3 on CO2 wasmeasured at 277◦C in the presence of added H

2O (feed

of 1 vol.% CH4, 0.05 vol.% H2O, balance air, with a

CH4conversion level less than 2.5%) and varying the

CO2 concentration (0.012–0.82 vol.%) No effect onreaction rate was observed up to 0.5 vol.% CO2abovewhich a strong inhibition occurred and did not findany explanation

2.6 Changes of activity under reactants (activation and deactivation)

Early studies mentioned large enhancements of theactivity of Pd/Al2O3 catalysts with time on stream

[9,23,50,59] On the other hand, the effect, if any, wasmuch less pronounced with Pt catalysts[48,49].Although morphological effects were invoked to ex-plain this phenomenon, it is now admitted[11,32,33]

that most of the activation of the catalysts with time

on stream has to be related to the slow removal ofresidual chlorine from the catalyst with time on stream(Section 2.1)

It is noteworthy that chlorine contamination could

be due to Cl-containing precursors of the metal, asfor instance in references[9,23,50,59,60] Burch[33]

suspected that the Pd catalysts studied in early works

by his group and prepared from Pd nitrate salt mighthave been contaminated by chlorine present as impu-rities in the salt Chlorine could be originally present

on the support as an impurity[51] In this study[51],the Pd/Al2O3catalyst was prepared by impregnating

a commercial alumina prepared by flame hydrolysis

of AlCl3(aluminium oxyd-C from Degussa) with ladium acetylacetonate in toluene A strong enhance-ment of the catalytic activity under reaction conditions

pal-was observed, which pal-was believed not to be due to

chlorine removal However, although the metal

precur-sor was free of chloride, the chlorine impurities of the

alumina itself should be considered and might play an

inhibiting role in the catalytic activity as well In this

study, the alumina contained high amounts of chloride

ions (ca 5000 ppm) This interpretation finds further

support in examining the data obtained with Pd/SiO2

prepared with Cl-free silica (Degussa Aerosil 200, Cl

content less than 240 ppm) and Pd(NH3)4(OH)2 In

this case, no activation could be observed and no

chlo-rine was present[51]

Although chlorine removal is likely to play the

ma-jor role in the activation of catalysts under reaction

conditions, this is not to claim the absence of

morpho-logical effects[22,48,49] For Pt/Al2O3, a change in

the reactivity of adsorbed oxygen was postulated[48]

Nanodiffraction and transmission electron studies[49]

revealed the preferential formation, under reactants, of

Pt(1 1 0) planes at the surface of the Pt particles, these

planes developing parallel to␥-alumina (1 1 0) planes

and allowing the epitaxial growth of Pt particles on the

alumina surface The increase of reactivity was related

by the authors to the preferential exposure of Pt(1 1 0)

planes at the surface of Pt particles, these planes being

thought to be more active towards methane oxidation

A similar study was carried out with a Pd/Al2O3

cat-alyst[61] A 1.95 wt.% Pd/Al2O3was prepared by

im-pregnating a transition alumina (SCM129 from Rhˆone

Poulenc) with an aqueous solution of

tetrachloropal-ladic acid The catalysts taken before and after

cat-alytic reaction (1 vol.% CH4:4 vol.% O2 in nitrogen)

were subsequently reduced in order to be studied by

nanodiffraction, electron spectroscopy and adsorption

of CO followed by FT-IR While the freshly reduced

sample exhibits mainly Pd(1 1 1) crystal planes

ex-posed to the reactants, less dense surface crystal planes

develop at the expense of Pd(1 1 1) ones after reaction

at 600◦C and the dispersion of the metal decreased.

This reconstruction of Pd particles was considered as

the main factor explaining the sharp increase of the

catalytic activity after ageing in the reaction mixture

Less dense planes were proposed to allow easier

re-versible transition between surface metallic palladium

and surface PdO because of only slight changes of

lat-tice parameters

Much fewer studies were devoted to the catalytic

behaviour of Pd/Al O and Pt/Al O in the complete

oxidation of methane under oxygen rich atmospherewith time on stream ([11,62]for Pd, and[12,13]forPt) In the presence of 1 vol.% CH4:4 vol.% O2 innitrogen (200 mg of 2 wt.% Pd/Al2O3, 6.5 l/h), Roth

et al.[11]have shown that, for reaction temperatures

in the range 350–450◦C, Cl free Pd catalysts exhibitslow deactivation with time on stream This deactiva-tion was quite severe since at 350◦C the conversiondecreased from initially ca 90–66% after 3 h reaction

It has to be mention that this behaviour was observedunder ‘dry’ conditions, that is in the absence of wateradded to the reaction feed except that produced by thereaction Purging the catalyst in gas carrier for 15 min

at the same temperature did not restore the activity generation of the catalyst could be however achieved

Re-by purging in dry carrier at temperatures above 500◦C.These results were tentatively attributed to the slowconversion of the active PdO phase into a less active

or inactive Pd(OH)2 phase This behaviour was tinguished from the inhibition by water also observedwhen water is added to the feed, as already shown inthe “kinetic studies” part Similar observations weremade by Mowery et al.[62]on 1 wt.% Pd/Al2O3pre-pared from Pd nitrate salt Deactivation and inhibitionwas also observed when water was added to the feed.Irreversible deactivation of the catalyst was observedeven at 520◦C Based on TPD experiments, it was pro-posed that water would be retained by the catalyst atthis temperature No explanation was given This de-activating behaviour of Pd/Al2O3catalysts is far frombeing understood Complementary experiments to elu-cidate this phenomenon and find out some relationshipwith the presence or the absence of residual Cl at thesurface of the support are required

dis-In the case of Pt/Al2O3catalysts, deactivation withtime on stream was also sometimes observed[12,13].According to Marceau et al [12,13], deactivationwould be due to sintering of the Pt particles under re-action mixture at 600◦C No data are given concern-ing the behaviour of these catalysts at intermediatetemperatures

2.7 Sulphur poisoning

The influence of sulphur compounds present in thereaction mixture on the catalytic activity of Pd and Ptcatalysts in the oxidation of methane was examined inthe literature Poisoning by sulphur compounds was

established, being more pronounced with palladium

than with platinum The influence of the reaction

con-ditions (concentration of sulphur compounds, presence

of water in the feed), or the nature of the support on

the poisoning are examined below

The influence of the nature of the support on the

poisoning of Pd catalysts by the addition of H2S

was examined by Hoyos et al [51] Alumina- and

silica-supported Pd catalysts were tested at 350◦C in

the oxidation of methane under oxidising conditions

(1 vol.% CH4:4 vol.% O2in nitrogen) in the absence

and in the presence of 100 vpm H2S Both catalysts

exhibited sharp deactivation in the presence of

sul-phur The fact that neither the Pd particle size nor

the apparent activation energy were affected by the

presence of sulphur in the feed suggested to associate

the deactivation to the decrease of the number active

sites by sulphur contamination without change of

their nature The extent of deactivation was the same

for both catalysts but the alumina support lowered

the rate of deactivation This property was attributed

to the ability of the alumina support to trap sulphate

species, H2S being fully oxidised into SO2/SO3 in

the reaction conditions In the case of Pd/SiO2, where

no surface sulphate species could form on the

sup-port, the formation of a palladium sulphate species

characterised by an IR band at 1435 cm−1 was

pro-posed These species can be totally decomposed upon

heating in vacuo or in flowing nitrogen at 600◦C, as

shown by the full depletion of the 1435 cm−1 band,

and the surface properties of palladium, together with

the catalytic activity, are fully restored A correlation

could be established between the 1435 cm−1

inten-sity obtained after increasing thermal regeneration of

the catalyst in nitrogen and the catalytic activity, the

higher the 1435 cm−1intensity the lower the rate of

methane oxidation Tentative regeneration of the

cata-lyst in hydrogen led to the decomposition of sulphate

species at temperatures as low as 350◦C But, the

catalytic activity was not regenerated in this case The

formation of highly stable surface palladium sulphide

species was suggested

The catalytic performance of Pd and Pt catalysts

supported on alumina or silica was also studied

un-der NGV engine exhaust gases or model reaction

mix-tures corresponding to the gas composition of NGV

exhausts (high redox ratios, total hydrocarbon

concen-trations less than 0.3 and 10 vol.% steam)[7] Under

real exhaust gases, the activity of Pd based catalystsfor methane oxidation declines rapidly It was estab-lished that this phenomenon is due mainly to sulphurcontained in the exhaust, even though its concentra-tion is very low, typically 1 ppm or less Exhaust sul-phur is derived from natural gas itself, i.e the odor-izer contained in the gas, or from engine lubricatingoil Ethane, propane and CO oxidation are also in-hibited by low SOx concentrations but to a lesser ex-tent than methane The authors developed the sameideas as in[51]concerning the mechanism by whichthe Pd catalyst deactivates and the role of the sup-port in the process (Fig 5) The deactivation curveobtained with Pd supported on a non-sulphating sup-port (ZrO2-SiO2) (110 g/ft3Pd, 320◦C, 800 ppm CH

4,

8 vol.% O2, 200,000 h−1, 0.1 or 0.9 ppm SO

2) wasconsistent with the 1:1 selective adsorption of SOxonPdO The deactivation is very rapid (2 h with 0.9 ppm

SO2) In contrast, palladium on sulphating supportssuch as␥-Al2O3, deactivates more slowly, which wasattributed to the adsorption of some SOx by the sup-port But in this case, the regeneration of the catalyst

is also more difficult than with non-sulphating ports Sulphated catalysts have equal activation ener-gies, suggesting identical sites irrespective of the sup-port In addition, the activation energy of Pd catalystspoisoned with SO2 increased significantly, which isconsistent with the transformation of active PdO sites

sup-to less active PdO-SOx sites The 0.5 eV increase of

Pd 3d5/2electron binding energies observed after low

SOx exposure would indicate an increase of the Pdoxidation state, possibly responsible for a decrease inthe availability of oxygen from PdO Compared to pal-ladium catalysts, platinum catalysts are much less ac-tive than Pd but also more resistant to deactivation by

SOx However, platinum is still less active than sulphurpoisoned palladium catalyst It is concluded from thiswork that NGVs equipped with palladium oxidationcatalysts can meet NMHCs and particulates standardsfor US heavy-duty transient test but not total hydro-carbons standards which require methane abatementduring cool operation

The effect of traces (20 vpm) of hydrogen sulphideand of sulphur dioxide on the catalytic activity of Pd,

Pd and Rh catalysts supported on alumina in the plete oxidation was examined by Meeyoo et al.[63].The nature of the metal salts was not indicated The re-action feed contained 1.8 vol.% CH and 21 vol.% O

com-Fig 5 Proposed mechanism for SO 2 inhibition of PdO activity in methane oxidation for PdO supported on sulphating or non-sulphating materials PdO converts SO 2 to SO 3 With a sulphating support, SO 3 adsorbs on both PdO and the support, retarding complete PdO poisoning On the contrary, a non-sulphating support cannot act as a sink for SO 3 which poisons PdO directly On suppressing SO 2 from the gas stream, SO 3 spills over from the sulphating support to PdO, while it desorbs directly from PdO with a non-sulphating support (Reproduced from Fig 7 of [7] , with permission from Elsevier Science.)

in helium Both gases (H2S, SO2) inhibited catalytic

oxidation over Pd and Rh The similarity of the

cat-alytic behaviour of Pd with respect to H2S and SO2can

be explained by the fact that H2S oxidises to sulphur

oxides above ca 350◦C And the sulphate formation

is proposed again to be responsible for deactivation

The same explanation was advanced to explain results

with the Rh containing catalyst On the contrary, the

activity of the Pt catalyst was found to be slightly

en-hanced by the pollutants Similar promoting effect of

SO2was previously observed by Burch et al.[46]in

the combustion of propane on 1 wt.% Pt/Al2O3but not

with Pt/SiO2 According to the results of Meeyoo et al

[63], the activity at low temperatures (below 550◦C)

is at least increased by a factor of 2 (with SO2)

com-pared to that obtained in the absence of pollutant The

formation of aluminium sulphates resulting in an

in-crease of acidity was thought to be responsible for the

enhancement of the activity In the case of platinum,

the sulphate formation on the metal is unlikely since

platinum oxide is not the favoured surface state

In the case of Pd/Al2O3catalysts, Yu and Shaw[64]

proposed an alternative explanation to the deactivation

by sulphur by formation of palladium sulphate The

catalyst (4 wt.% PdO supported on␥-Al O , 67 m2/g)

was prepared from the nitrate salt and calcined in air

at 500◦C before testing in 1 vol.% methane in air(dilution of the catalyst with ␥- or ␦-alumina) The

progressive inhibition of the conversion versus time

by adding H2S (80 vpm) to the feed at 400◦C wasagain shown Suppressing H2S allowed the activity to

be only slightly recovered Pre-exposure of the alyst to H2S in air (24 h) at increasing temperatures(100–400◦C) caused an increasing inhibiting effect onthe activity FT-IR data indicated the formation of alu-minium sulphate (1145 cm−1) and to a lesser extentsulphite (1060 cm−1) On the basis of the decrease ofthe BET area due to aluminium sulphate formation,the PdO occlusion was proposed to explain the de-crease of the activity instead of the formation of pal-ladium sulphate Another disagreement with Hoyos

cat-et al concerns the effect of H2S poisoning on the tivation energy, which decreased upon poisoning from

ac-ca 130 kJ/mol (without H2S) to ca 90 kJ/mol (withH2S) It must be noted that the value obtained withoutH2S differs significantly from those (85± 15 kJ/mol)

usually reported under similar experimental tions, which might possibly indicate thermal effects.The pre-exponential factor decreased by 4 orders ofmagnitude The decrease of the activation energy was

condi-attributed to changes in the kinetics from surface

con-trol to pore-diffusion concon-trol, the latter phenomenon

being due to the build-up of sulphate (sulphite) groups

at the surface of the alumina Most of these groups

were shown to be removed by H2treatment at 600◦C,

the activity for methane oxidation being thus

regen-erated This interpretation of the poisoning of

palla-dium catalysts by sulphur was not further considered

by other groups

The influence of sulphur poisoning (H2S) and

re-generation in reducing conditions on the activity for

methane oxidation of 2 wt.% Pt, Pd or Rh on

alu-mina was examined[65] The catalysts were either

re-duced in H2at 400◦C (fresh) or exposed to 100 ppm

H2S/900 ppm H2at 100◦C and regenerated in H

2 at

400◦C (regenerated) The reaction feed consisted of

4 vol.% CH4in air (6 l/h, 100 mgcat) For Pt and Rh

cat-alysts, regenerated samples were found to be slightly

more active than the fresh ones, while the reverse is

observed for Pd/Al2O3 For Pd/Al2O3, the activation

energy increased after pre-poisoning of the catalyst

(113 kJ/mol compared to 84 kJ/mol for the fresh

sam-ple) Based upon FT-IR of adsorbed CO, the

regener-ation seems to be successful for Pd/Al2O3 However,

an increase ofνCO on regenerated Pt catalyst

possi-bly indicated the presence of residual sulphur species

at the surface of Pt after regeneration

The mechanism of the deactivation of PdO/Al2O3

catalysts by SO2 was re-examined by Mowery et al

[62] focusing on the deactivation in the presence of

both water and SO2 Particular attention was given to

the ageing of the catalyst being placed in the exhaust

of a lean-burn spark ignited natural gas engine Two

types of catalysts were studied: a commercial catalyst,

100 g/ft3palladium/alumina deposited on a cordierite

monolith pre-calcined at 700◦C in air, and a model

catalyst, 1 wt.% Pd/␥-alumina (260 m2/g)

(impreg-nation with palladium nitrate) calcined at 500◦C in

air The catalytic activity in complete methane

oxi-dation was measured in a microreactor flowed with

800 ppm CH4, 6.5 vol.% O2, N2 balance (for dry

feed), 800 ppm CH4, 16.4 vol.% O2, 2–3 vol.% H2O,

N2balance (for wet feed) and 800 ppm CH4, 410 ppm

CO, 340 ppm NO, 2–3 vol.% H2O, 6 vol.% CO2,

16.4 vol.% O2, N2 balance (for simulated exhaust)

Rapid deactivation of PdO/Al2O3 catalysts was seen

as the result of engine ageing Phosphate (originating

from lubricant additives) and sulphate deposits were

detected, but only sulphates forming both on PdO andalumina surfaces are considered to be the main cause

of deactivation When the conversion of methane wasperformed in dry feed, fresh and engine aged samplesexhibited the same activity The presence of water

in the feed appeared necessary to reveal the vation of the catalyst SO2 was shown to cause bothinhibition (10% loss of activity with 10 ppm SO2 at

deacti-460◦C) and deactivation with time on stream activation was partly (or fully) reversible depending

De-on the temperature The presence of both water and

SO2in the feed caused the catalyst to deactivate morerapidly and the recovery of the activity was more dif-ficult than when either poison was added separately

On the basis of TPD experiments, it was proposedthat sulphation of alumina produced a more hy-drophilic surface, adsorbing water more strongly and

in larger amounts The presence of water was thought

to force spillover of surface SOx species from thealumina to PdO and enhance the rate of bulk PdSO4formation

Lee et al [66] merely addressed the oxidation ofH2S in the presence or not of methane over Pd- andPt-based monolith catalysts Both catalysts consisting

of ceramic monoliths coated with a washcoat of mina and the precious metal (1.4 g/l metal) were tested

alu-in the oxidation of H2S, methane and methane/H2S.Over Pd based catalyst, H2S (26 ppm) was shown tohave a strong inhibiting effect on the methane conver-sion (the light-off temperature being ca 200◦C higher

in the presence of H2S) On the contrary, with inum catalyst, the catalytic activity was slightly in-creased in the presence of H2S Its activity was higherthan that of the Pd catalyst under the same experi-mental conditions The Pt-based catalyst was alreadyused to design an industrial converter, which operatedsatisfactorily for 2 years

plat-3 Improved catalysts

3.1 Pd or Pt supported on sol–gel silica or alumina

Silica and alumina supported Pd and Pt catalysts areusually prepared by using commercial supports How-ever, some attempts were made to synthesise thesesupports by sol–gel method and the catalytic proper-ties were compared to that of conventional ones

Mizushima and Hori [67] prepared alumina

sup-ported Pd and Pt catalysts following two ways Pd and

Pt were supported on alumina aerogels by mixing the

metal chloride (H2PtCl6 and PdCl2) precursor to the

alumina sol (method A) or more conventionally

im-pregnating the calcined aerogel by the metallic salt

(method B) A commercial␥-Al2O3was also used as a

support for comparison A 1 wt.% metal catalysts were

thus obtained Pt catalysts, either supported on

com-mercial alumina or aerogel, exhibited approximately

the same catalytic activity On the contrary, Pd/Al2O3

were found more active when aerogel aluminas were

used than with the commercial␥-Al2O3 Method B

(impregnation of palladium after calcination of the gel

at 1000◦C) gave better results of activity

Unfortu-nately, the influence of residual chloride ions after

cal-cination of the catalyst at 500◦C was not examined.

It cannot be concluded whether aerogel has a

benefi-cial effect on the intrinsic activity of Pd/Al2O3in the

complete oxidation of methane or Cl departure

result-ing from metal precursor is facilitated on the aerogel

support

More recently, 0.5 wt.% Pd/SiO2catalysts were

pre-pared by sol–gel methods[68] The influence of the pH

of gelation and the way of Pd addition, before gelation

or by conventional impregnation of the sol–gel

cal-cined support, on the complete oxidation of methane

(stoichiometric conditions) was investigated

Differ-ent states of the catalysts were studied in 1 vol.%

CH4:2 vol.% O2 (He balance) (200 mgcat, flow rate

50 cm3/min), calcined at 450◦C, reduced at 500◦C or

aged under reactants at 100% conversion The metal

particle size was examined carefully, being measured

before catalytic testing and after reduction at 500◦C

by hydrogen chemisorption and TEM with a good

agreement between the two techniques For samples in

which Pd was incorporated before gelation (SG

sam-ples), the particle size did not vary much (2–3.3 nm)

with varying preparation conditions Approximately,

the same particle size was obtained for most of the

impregnated samples The light-off temperatures (T50)

were systematically lower for the impregnated

sam-ples compared to the sol–gel samsam-ples independently

on the activation procedure Moreover a strong

acti-vation under reactants could be observed for sol–gel

samples The trapping of Pd particles in sol–gel

sam-ples was suggested to explain their low activity but this

seems to contradict dispersion measurements It can

be concluded that the method consisting of ing Pd before gelation does not improve the catalyticactivity It is noteworthy however that the impregnatedsample corresponding to SiO2with the highest surfacearea exhibits interesting performances as indicated by

incorporat-a T50 equal to 300◦C.

A Pt/Al2O3 catalyst was prepared by sol–gelmethod and calcined in air at various temperaturesbefore reduction[69] The calcination of the catalystbetween 500 and 800◦C seemed to favour metallicdispersion A migration of platinum occluded in thealumina matrix to the surface has been proposed forthe increase of the catalytic activity in propene com-bustion as a function of the calcination temperature

3.2 Pd or Pt supported on ZrO2

Except silica and alumina supports, ZrO2 is ably the support of Pd and Pt catalysts which attractedthe largest number of studies in the most recent pastyears[34–36,70–72] But this concerns only Pd cat-alysts In the patent literature, Pd or Pt supported onZrO2, stabilised or not by Y or La, associated or notwith other refractory oxide supports, are designated

prob-as catalysts mainly used for combustion applications

[73–80]and scarcely for treatment of CH4-containingwaste gases[81]

Pd/ZrO2catalysts exhibit catalytic performances inthe oxidation of methane comparable to or higher thanPd/Al2O3catalysts[34,70–72,82] Fujimoto et al.[71]

prepared Pd/ZrO2 catalysts by impregnating the conia (16 m2/g) with solutions of Pd(NH3)2(NO2)2inHNO3 Steady-state turnover rates (280◦C, 2 vol.%

zir-CH4, 20 vol.% O2 in He balance) were calculated

as a function of crystallite size based on dispersionmeasured by H2O2 titration The maximum rate was0.18 s−1 An approximately linear dependence of therate versus crystallite size (3–11 nm) was observed.The same trend was also reported by Baiker andco-workers [34,72]with Pd/ZrO2 catalysts prepared

by oxidation of glassy Pd25Zr75alloy in air The ing catalysts consist of poorly crystalline palladiumoxide and monoclinic and tetragonal zirconia in inti-mate contact The main advantage of the preparationover other conventional methods is the very high level

result-of purity result-of the catalyst Based on the catalyst massand Pd surface area, this catalyst was claimed to ex-hibit a superior catalytic activity for complete methane