Catalysis Today 54 (1999) 165–180

Catalytic combustion for the production of energy

Pio Forzatti ∗ , Gianpiero Groppi
Dipartimento di Chimica Industriale e Ingegneria Chimica ‘G. Natta’ del Politecnico, Pza Leonardo da Vinci 32, 20133 Milan, Italy

Abstract
In this paper, attention is focused on adiabatic lean-premixed catalytic combustion, which has attracted interest in the
last decades as an environmentally friendly and cost-effective alternative to flame combustion for power generation by gas
turbines.
The different configurations of the combustion systems for gas turbines are presented. and the results of pilot tests, full-scale
bench tests and field trials with retrofitted machines are illustrated. The relevant physico-chemical and catalytic properties
of highly active supported PdO catalysts, including reversible PdO–Pd transformation and low-temperature activity, are then
addressed. The structural and catalytic properties of highly stable metal substituted hexaaluminates are also discussed. The
use of fuel alternative to natural gas is briefly covered. The relevant features of mathematical models for both, the catalyst
section and the homogeneous section, that can be used in the design and analysis of the catalytic combustor, are described
and the major conclusions of the modelling activity are outlined. Finally, the research opportunities in the area are discussed.
©1999 Elsevier Science B.V. All rights reserved.
Keywords: Gas turbine combustors; NOx reduction; Methane combustion; Palladium catalyst; Hexaaluminates; Mathematical Modelling

1. Introduction 2. non-adiabatic premixed catalytic combustion,

which is under development for several applica-
In the last decades, catalytic combustion has been tions including premixed fibre burners, domestic
vigorously explored as a route to the production of boilers and compact chemical reactor heaters
heat and energy in view of its capability to achieve [13–17]; and
effective combustion at much lower temperatures than 3. non-adiabatic diffusive catalytic combustion, that
in conventional flame combustion, thus allowing for has been commercially applied in radiant heaters
the simultaneous ultra-low emissions of NOx , CO and [1].
Unburned Hydrocarbons (UHC) [1–12]. In this paper, attention is focused on the first con-
Various concepts of catalytic combustion have been cept. Adiabatic lean-premixed catalytic combustion is
investigated. and these can be classified into the fol- an environmentally-driven technology that has the po-
lowing three groups: tential to achieve NOx -emissions below 3–5 ppm from
1. adiabatic lean-premixed catalytic combustion, natural gas-fired turbines, without incurring additional
which has attracted attention for power generation capital and operating costs associated with the current
by gas turbines (GT) [8]; technology based on steam injection or lean-premixed
homogeneous combustion and SCR systems. The de-
∗ Corresponding author. Tel.: +39-02-23993238; velopment of the technology has required an integrated
fax: +39-02-70638173 approach based on the design of advanced combustor
E-mail address: [email protected] (P. Forzatti) configurations and on the development of catalytic

0920-5861/99/$ – see front matter ©1999 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 0 - 5 8 6 1 ( 9 9 ) 0 0 1 7 8 - 9
166 P. Forzatti, G. Groppi / Catalysis Today 54 (1999) 165–180

materials with superior activity and stability and

unique temperature self-regulating properties.
First, the different designs of the combustion sys-
tems for gas turbines are presented and the results
of pilot tests, full-scale bench tests and field trials
with retrofitted machines are illustrated. The relevant
physico-chemical and catalytic properties of highly ac-
tive supported PdO and highly stable metal-substituted
hexaaluminates are then addressed. The use of fu-
els alternative to natural gas, including low-heating
value (LHV) mixtures from biomasses gasification and
diesel fuels, are discussed. The use of mathematical
models as a tool in the design and analysis of the cat-
alytic combustion system is also covered. The relevant
features of mathematical models for both, the catalyst
section and the homogeneous section are described
and the major conclusions of the modelling activity
are outlined. Finally, the research opportunities in the
area are briefly illustrated.
Fig. 1. Schematic of conventional flame and catalytic combustion
systems [18]. C, Compressor; T, turbine; HC, homogeneous com-
2. Configurations and performances of the bustor; CC, catalytic combustor.
catalytic combustion systems

Table 1
The buildup of a gas turbine with a conventional
Design criteria and operating conditions of gas turbine combustors
flame combustion system and with a catalytic combus-
tion system are illustrated in Fig. 1 [18]. In a conven- Design criteria
tional flame combustion system, air enters the com- Emission targets NOx < 5 ppm
bustor at the compressor discharge temperature and, CO < 10 ppm
upon blending with fuel, the mixture is ignited and UHC < 10 ppm
<5%
burned at temperatures up to 1800◦ C. Since the turbine Pressure drops
Catalyst durability 8000 h
materials cannot withstand this temperature, part of
the compressed air is bypassed to the gas, exiting the Operating conditions
combustion chamber to reduce the temperature for de- Inlet temperature 300–450◦ C
livery to the turbine inlet. In the catalytic combustion Outlet temperature 1100–1300◦ C
system, the combustion is accomplished at lower tem- Pressure 10–20 atm.
Mass flow rate 100–200 kg/m2 s
perature, up to 1100–1300◦ C. Still, the design criteria Residence time 10–30 ms
and the operating conditions of a catalytic combustor
are very severe (see Table 1): ultra low single-digit
emissions for NOx , CO and UHC and catalyst dura-
bility exceeding 8 h must be ensured operating at very close as possible to that of compressed air, in or-
high gas velocities (in the range of 10–40 m/s referred der to minimise the temperature rise provided by
to the cross section immediately upstream of the cata- the pre-burner that is responsible for most of NOx
lyst section), with a small overall pressure drop and at formation;
very short residence times. This poses the following 2. high thermal stability of the catalytic materials
requirements: with respect to deactivation by sintering, phase
1. very high catalytic activity in methane complete transformation and volatilisation, and structural
oxidation to ensure ignition at temperatures as integrity upon thermal shocks;
 P. Forzatti, G. Groppi / Catalysis Today 54 (1999) 165–180 167

Fig. 2. CST combustor configuration [19].

bility of forming active monolithic honeycombs. The

applicability of this system has been demonstrated
by Osaka Gas through tests on a 160-kW prototype
device, which was operated under representative con-
ditions of a 1.5-MW gas turbine [12,19]. However.
the throughput of this system is so low that installing
it on a conventional compact engine package may not
be possible. In view of this, its commercialisation has
Fig. 3. Schematic of hybrid combustor configuration [21]. been suspended [12]. Furthermore. this system can
hardly be upgraded in view of the expected develop-
3. coupling of the heterogeneous and homogeneous ments of the GT technology towards higher turbine
reactions to guarantee complete conversion of the inlet temperature.
hydrocarbon and ultra low emissions of CO. In the second configuration (Fig. 3), only a frac-
Figs. 2–4 show the three alternative approaches that tion of the fuel is fed to the catalyst; here, the inlet
have attracted most attention. fuel-to-air ratio is carefully adjusted to limit the adi-
In the first configuration (Fig. 2), the premixed abatic reaction temperature, typically below 1000◦ C;
air/fuel stream is fed to multiple catalyst system thus reducing the thermal stresses and preventing cata-
where the combustion proceeds until complete fuel lyst deterioration [21]. The remaining amount of fuel,
consumption [12,19]. Typically, one or more mono- upon premixing with most of residual air (typically
lith honeycomb segments made of Pd on cordierite 65% of the overall air flow rate is split for the cata-
coated with stabilised alumina are positioned at the lyst, 30% for the premixing nozzles and 5% for cool-
entrance to ignite the methane combustion reaction, ing and leaking), is fed to a downstream homogeneous
whereas several less active Mn-substituted hexaalu- section, where combustion is assisted and stabilised
minate monolith segments sintered at higher tempera- by the hot gas exiting the catalyst. In addition to re-
tures (e.g. 1200◦ C and 1300◦ C) are used downstream. duction of catalyst thermal stresses and stabilisation
These latter materials, first developed by Arai et of homogeneous premixed combustion at lower tem-
al. [20], were selected in view of their combustion peratures, the following major advantages have been
activity, excellent high-T thermal stability and capa- claimed:
168 P. Forzatti, G. Groppi / Catalysis Today 54 (1999) 165–180

Fig. 4. Partial combustor configuration: XONON technology [26].

Table 2
The third configuration has been developed by Cat-
Tested performances of hybrid combustor [21]
alytica in co-operation with Tanaka Kikinzoku Kogyo
Full scale tests at base load conditions of a 10 MW GT KK and GT manufacturers (e.g. General Electric, So-
NOx < 5 ppm (at 16% O2 ) lar Turbines, Allison Engine) [22–24]. In this configu-
CO + UHC < 9 ppm ration (see Fig. 4) ,all the fuel, except that required for
Pressure drop < 4% the pre-burner, is fed to the catalyst section consisting
of an inlet stage designed to provide high catalytic ac-
1. stable combustion is realised under a wider range tivity, low light-off temperature and low catalyst wall
of combustor operating conditions by controlling temperature and an outer stage designed to operate at
the fuel split to the catalyst; higher wall temperatures in order to provide the re-
2. a certain degree of lack of uniformity in the quired high outlet gas temperature. In a downstream
fuel/air mixture to the catalytic section is allowed homogeneous section, the combustion of the fuel is
due to the low catalyst temperature. completed, CO and hydrocarbons are burned out to
This configuration was tested by CRIEPI and Kansai the levels required to meet the emission standards and
Electric in Japan at a scale equivalent to one com- the combustor outlet temperature is raised to the lev-
bustor of a 10-MW class, multi-can type GT [21]. At els of modern high efficiency GT, i.e. 1250–1450◦ C.
base load conditions, the design targets on emissions The control of the maximum catalyst wall temperature
(NOx < 5 ppm, CO and UHC < 10 ppm ) and pressure is achieved by means of a proprietary catalyst design
drops (<5%) have been demonstrated for combustor based on the temperature self-moderating properties of
outlet gas temperature of 1350◦ C (see Table 2). The PdO/Pd catalysts in methane combustion, the use of a
key factors in such achievements were reported as fol- monolithic metal support with integral heat-exchange
lows: capabilities between active and passive channels, and
1. development of a highly active catalyst made of the use of a diffusion barrier over the catalyst surface.
PdO, promoted with small amounts of Pt and Rh, These features will be discussed in a following para-
and supported on an Al2 O3 /ZrO2 washcoat de- graph.
posited on honeycomb type monolith made of During bench tests at full scale, performed in a
cordierite; catalytic combustor system developed by GE for its
2. arrangement of the catalyst section and the fuel MS9001E gas turbine, NOx , CO, and UHC emissions
distribution system in a configuration suitable were documented at levels below the emission targets
to optimise the interaction between the hetero- of 5, 10, and 10 ppm, respectively, at both base load
geneous and the homogeneous sections within (100%) and part load (78%) (see Table 3) [25]. The
acceptable pressure drops. tests also documented negligible pressure dynam-
Nevertheless, the long-term performance of the cata- ics. Besides, in a field test performed on a 1.5-MW
lyst and of the combustor has to be estimated and the Kawasaki M1A-13A machine, the system demon-
control method established. strated stable performance with a negligible loss of
 P. Forzatti, G. Groppi / Catalysis Today 54 (1999) 165–180 169

Table 3
3.1. Palladium-based catalysts
Tested performances of XONON combustor
Full scale tests at GE 9E GT conditions [25] Palladium oxide supported on alumina or zirconia
Emissions (15% O2 ) with various additives is the catalyst of choice for the
NOx 3.3 ppm combustion of methane primarily in view of the fol-
CO 2.0 ppm lowing properties:
UHC < 0.1 ppm
Pressure drop 2.6%
1. superior activity in the combustion reaction, which
Combustor dynamics results in low light-off temperatures;
good mechanical stability and low pressure dynamics 2. unique capability of temperature self-control as-
Field tests on 1.5 MW Kawasaki GT [10] sociated with the PdO–Pd reversible transforma-
tion;
Emissions (15% O2 )
3. low volatility of Pd species that may be present
NOx < 3 ppm
CO < 5 ppm under reaction conditions (metal, oxides, hydrox-
UHC < 5 ppm ides and oxyhydroxides).
Efficiency 23.0% (23.6%a ) The characteristics of the PdO–Pd transforma-
Load range 50–100% tion, its relevance to methane combustion and the
Durability > 1000 h
low-temperature activity of PdO catalysts are ad-
a Efficiency before introduction of the catalyst. dressed below.

efficiency and with ultra-low emissions of NOx , CO 3.1.1. Characteristics of the PdO–Pd transformation
and UHC during the 1000-h operation [10]. The PdO–Pd transformation has been investigated
As shown in Fig. 5, stable operation with low emis- in the literature primarily by means of TG analysis
sions are achieved within an operation window that is [27]. Fig. 6 shows the TGA profiles collected by heat-
bounded by the following conditions [25,26]: ing and cooling a PdO-on-alumina catalyst, previously
1. the inlet gas temperature must be high enough to calcined at 1000◦ C for 10 h. Above 800◦ C, PdO de-
ignite the catalyst; composition to Pd metal occurs and is complete at
2. the gas temperature at the catalyst exit must be around 900◦ C. The observed weight loss corresponds
high enough to stabilise homogeneous combustion well to the calculated value for the amount of PdO
and the adiabatic combustion temperature must be present. During the cooling ramp re-oxidation of pal-
high enough to ensure complete burnout of HC ladium metal occurs at about 680◦ C and is complete
and CO within the available contact time; at 550◦ C. However, the weight increase is lower than
3. the maximum catalyst wall temperature must be that observed during the decomposition. Different ex-
low enough to ensure long-term catalyst durabil- tents of re-oxidation of Pd metal have been reported
ity. in the literature, ranging from 30 to 80% by weight
Notice that all portion of the catalytic reactor must [26,27]. Such variations possibly depend on differ-
be within the operating window; accordingly, a good ent catalyst pre-treatment. The temperature difference
uniformity of inlet fuel-to-air ratio (e.g. ±5%) and of between decomposition and re-oxidation represents a
inlet catalyst temperature (e.g. 1T = 18◦ C) has to be hysteresis of about 150◦ C. In the second, and the fol-
ensured. Catalytica has recently announced the com- lowing cycles, PdO decomposes in two distinct steps
mercialisation of this technology. at 800–850◦ C and 850–900◦ C and re-oxidises below
650◦ C and its magnitude is comparable to those of the
first cycle. A third cycle provides essentially the same
3. Combustion catalysts results as the second one. It has been demonstrated
that re-oxidation is not simply a time–temperature rate
The catalysts that have been most extensively inves- process, but requires a specific temperature. Indeed no
tigated for GT applications are Pd-supported materials re-oxidation was observed when the sample, after de-
and metal-substituted hexaaluminates. composition, was hold for 30 min slightly above the
170 P. Forzatti, G. Groppi / Catalysis Today 54 (1999) 165–180

Fig. 5. Operation window in XONON technology [27].

temperature at which the onset of PdO reformation

was observed upon cooling [27]. On the other hand,
re-oxidation did occur at higher temperatures if only
the first reduction step has proceeded. In repeated ex-
periments, where the temperature ramp in the second
cycle was reversed after the first weight loss, a distinct
weight gain was observed even at 740◦ C [28]. Using
high-temperature X-ray diffraction PdO is the only de-
tectable palladium phase prior to decomposition. Pd
metal and PdO are both observed in the course of de-
composition and only Pd metal is present when de-
composition is complete. Following re-oxidation, both
PdO and Pd metal were detected, in line with incom- Fig. 6. TG and DTG profiles of Pd/Al2 O3 during heating and
cooling cycles.
plete PdO reformation during cooling shown by TGA
data. The X-ray calculated crystal dimensions of PdO
were found to be significantly smaller than those of Table 4
Temperatures of onset of PdO decomposition (TD1>: 1st stage;
Pd metal, indicating that palladium sinters upon de- TD2 2nd stage) and reformation (TR) on different supports
composition to the metal and re-disperses with refor-
Support TD1(◦ C) TD2(◦ C) TR(◦ C) TD1-TR
mation of the oxide. The tendency of metal palladium
particles to sinter easily and to re-disperse during oxi- Unsupported PdO 810 n.o. 785 25
dation is well documented in the literature [29–35]. In Al2 O3 795 840 690 105
La2 O3 /Al2 O3 800 835 690 110
addition to bulk PdO, evidences for other forms of Pd
CeO2 /Al2 O3 800 845 755 45
oxides and for strong palladium-support interactions CeO2 /La2 O3 /Al2 O3 800 840 750 50
have also been reported in the literature [36–40]. ZrO2 800 860 730 70
Table 4 shows that the temperatures of onset of the
decomposition of PdO to palladium metal during heat-
ing (TD), the temperatures of onset of the re-oxidation perature of decomposition of the oxide (TD) is only
of Pd metal during cooling (TR), and the extent of hys- marginally affected by the support, whereas the tem-
teresis (TD–TR), for unsupported PdO and for PdO perature of re-formation of the oxide from the metal
supported on several oxides. It appears that the tem- (TR) is markedly affected by the nature of the support;
 P. Forzatti, G. Groppi / Catalysis Today 54 (1999) 165–180 171

Fig. 7. TG and DTG profiles of bulk PdO during heating and cooling cycles.

accordingly, the extent of hysteresis (TD–TR) is also to 980◦ C at 16 atm. The kinetics of the decomposition
strongly influenced by the support. process and the related mechanism have been stud-
Fig. 7 has been generated according to data col- ied so far under conditions far from those of interest
lected in our laboratory in similar TG cycles using un- for GT applications, e.g. at low temperatures and un-
supported PdO powder. During the first heating ramp, der vacuum or reducing atmospheres containing H2
PdO decomposes to Pd metal with only one sharp or CH4 [36,41,42]. Accordingly, these points deserve
weight loss between 810◦ C and 870◦ C. In the cooling further investigation.
curve, only a small weight gain is observed at about The mechanism of Pd metal particle re-oxidation
780◦ C with a magnitude of about 10% of the ini- at low temperatures is believed to occur through
tial weight loss. In the second cycle, only one weight oxygen chemisorption to form a PdOx surface layer
loss is observed between 810 and 840◦ C correspond- followed by bulk oxidation of the metallic core
ing to reduction of the reformed PdOx . These results characterised by a high activation energy of about
closely resemble those previously obtained by Far- 100 kJ/mol [27,29–32,36,37,41,42]. The kinetics of
rauto et al. [27], but for the magnitude of the weight the re-oxidation process has been ascribed either to
gain in the first cooling stage that amounted to about electric field-driven transport of oxygen anions till
1% of the initial weight loss. However, the charac- the oxide layer is sufficiently thick and, thereafter, to
teristic of PdO/Pd transformation in bulk palladium the diffusion of oxygen through the oxide layer or to
oxide markedly deviates from those observed in sup- lattice rearrangement needed for the formation of the
ported catalysts, indicating a key role of the support oxide layer at the oxide-metal interface.
that deserves further investigation. On the other hand, the Pd metal re-oxidation dur-
The temperature of onset of PdO decomposition as ing the cooling cycles at high temperatures, that is of
a function of the oxygen partial pressure is plotted interest for GT applications, is a complex and poorly
in Fig. 8 for both, unsupported and supported PdO. understood process. The hysteresis between PdO de-
A good agreement is observed between experimental composition and Pd metal re-oxidation has been ten-
data and predictions provided by the relationship that tatively associated with the capability of Pd metal to
can be derived, assuming that the process is governed chemisorb O2 only below TD [27] or to the formation
by the thermodynamics of the decomposition of bulk of a passive layer of chemisorbed oxygen [43]. How-
PdO, namely PdO → Pd + (1/2)O2 . This implies that ever, the first explanation can hardly account for the
TD in air shifts from 790◦ C at atmospheric pressure large difference in the extent of hysteresis observed for
172 P. Forzatti, G. Groppi / Catalysis Today 54 (1999) 165–180

tion, when only Pd metal is present at high tempera-

tures, the catalyst is inactive. However, there are also
indications that the partial reduction of PdO results in
a significant enhancement of the methane combustion
activity; this was ascribed to the role of Pd metal in
the activation of methane through dissociative adsorp-
tion [44,45]. Lyubowski and Pfefferle [46], along sim-
ilar lines, associated the anomalous increase of CH4
combustion activity observed during cooling ramp in
CH4 air mixtures with the formation of a small clus-
ter of Pd oxide dispersed onto the surface of large Pd
metal crystallites. Recently, Fujimoto et al. [47] have
hypothesised that activation of methane does not in-
Fig. 8. Thermodynamic control of the first PdO decomposition volve Pd metal, but occurs on site pairs consisting of
step: ln (PO2 ) vs. 1/TD1. Data for bulk PdO derived from Ref. oxygen atoms and oxygen vacancies on the surface of
[83].
palladium oxide crystallites.
The variations of activity associated with the
PdO–Pd reversible thermal transformation are respon-
unsupported PdO (TD–TR = 25–40◦ C) and supported sible for the unique self-regulating capability of Pd
PdO (e.g. TD–TR = 120◦ C for PdO on alumina) and catalysts in methane combustion under adiabatic con-
its dependence on the nature of the support. The sec- ditions. However, runaway ignition during start-up
ond explanation as well has not been corroborated by has been reported in the literature [48,49], which in-
direct evidence for the presence of the passive oxide dicates that the dynamics of the combustion reaction
layer; besides, it is not clear why the passive oxide does compete with the dynamics of PdO–Pd trans-
layer should not be effective below the threshold tem- formation. The dynamics of this transformation has
perature of Pd metal re-oxidation. not been studied so far at the temperature of interest
for GT applications and deserve further investigation.
3.1.2. Relevance of PdO–Pd transformation in In order to control (or to mitigate) runaway combus-
methane combustion tion in the case of methane-to-air ratio corresponding
The methane conversion measured during a cat- to high adiabatic temperatures, the use of zirconia
alytic test accomplished by ramping the furnace to a as a support is claimed in the patent literature [49].
maximum temperature of 900◦ C, holding for a lim-
ited time and then cooling the reactor is shown in
Fig. 9. During heating, methane conversion increases
with temperature, and is complete above 500◦ C up to
900◦ C, well beyond the decomposition temperature of
PdO to Pd metal (solid line). When the temperature
ramp is reversed and the catalyst is cooled in CH4 /air
at about 780◦ C, the conversion reduces and follows
the pattern observed in the presence of the alumina
support only (not reported in the figure), thus demon-
strating that the catalyst has lost its activity upon com-
plete reduction of PdO to Pd metal. On further cool-
ing, there is a sudden restoration of activity, beginning
slightly before re-oxidation of Pd metal to PdO is ob-
served in TG experiments with similar oxygen partial
pressure. There is a general consensus in the literature Fig. 9. CH4 combustion activity during heating and cooling ramps
that while PdO is highly active in methane combus- over PdO supported on La2 O3 stabilised alumina.
 P. Forzatti, G. Groppi / Catalysis Today 54 (1999) 165–180 173

Fig. 10. Examples of coupling of active and passive channels [23].

Table 5
catalysts are widely scattered. Indeed, the conversion
Turnover rates (TOR) of CH4 combustion over palladium catalysts
levels depend on several critical factors that, in many
TOR (s−1 )a Crystal size (nm) Reference cases, have not been taken into consideration: e.g. the
7 × 10−3 –1 × 10−1 1.4–5.6 [50] catalyst pre-treatment procedures, the nature of the Pd
1 × 10-4 –2 × 10−2 1–30 [34] precursor, the presence of water in the reaction mix-
3 × 10−3 3 [51] ture, and several features of the PdO–Pd transforma-
3 × 10−2 16 [52]
tion. Ribeiro et al. [54] noticed that, provided that
1 × 10−4 –2 × 10−2 2–80 [53]
2 × 10−2 –8 × 10−2 2–110 [54] these factors are properly accounted for, the turnover
3 × 10−2 –1.8 × 10−1 3–10 [47] rate (TOR) of the reaction is only marginally affected
1 × 10−2 –1 × 10−1 10–15 Our laboratory when the support, Pd precursor, and Pd particle sizes
a At 550 K and 2% of CH4 . are changed. On the other hand, for PdO/ZrO2 , a slight
increase of TOR with PdO crystal size in the 3–10 nm
range has been recently reported [47].
A diffusion barrier added on the top of the catalyst A relevant feature of commercial Pd catalysts for
layer to limit the supply of fuel and/or oxidant to the GT application is represented by the high Pd load-
catalyst, which decreases the apparent activation en- ing, that is usually in the 10–15% w/w range. Table 6
ergy of the combustion reaction, is also recommended shows the temperatures where 20% conversion has
[49]. Besides, metal monoliths are preferred and the
catalyst is applied to only one side of the monolith
structure, as depicted in Fig. 10. In this way, the heat Table 6
generated inside the active channel is transferred to Effect of Pd loading on an La2 O3 /Al2 O3 support on PdO crystal
the nearby inactive channel by conduction through size and CH4 combustion activitya
the washcoat and the metal [23,24]. Pd load (w/w) T20% (◦ C) τ PdO (nm)

2.5 350 12
3.1.3. Low-temperature activity of Pd catalysts 5 340 10
The low-temperature activity represents an impor- 10 320 13
tant issue of PdO catalysts, since it affects the light-off 15 300 13
performances of the system. As shown in Table 5, the 20 300 12
activity data of methane combustion over PdO based a Test conditions: GHSV = 210,000 Ncc/gh; YCH4 = 0.005.
174 P. Forzatti, G. Groppi / Catalysis Today 54 (1999) 165–180

been measured over Pd supported on stabilised alu-

mina catalysts prepared in our laboratory, together
with dimension of the PdO crystallites calculated by
XRD. The activity of the catalysts increases markedly
with the PdO loading up to 15% w/w and then tends to
level off. Accordingly, high Pd loading is required to
ensure low light-off temperatures. On the other hand,
the size of PdO particles is not affected by the PdO
loading: the dimensions of the PdO crystallites com-
pare well with those of the catalyst pores, which indi-
cates that sintering of PdO is controlled by the mor-
phology of the washcoat.
The assessment of a reliable kinetics for the com-
bustion of methane over PdO catalysts is another im-
portant issue for the development of catalytic combus-
tors. Kinetic studies in the literature have been cus-
tomarily performed under conditions far from those of
Fig. 11. Ignition/extinction curves in pilot scale combustors [8].
real interest, namely at low GHSV, low temperatures,
atmospheric pressure, and using a micro-flow reac-
with a pre-burner, it has been demonstrated that the
tor operated with diluted gas feed at low conversions.
effect of water and CO2 produced by the pre-burner
Under such conditions, the methane combustion was
is small. The temperature shifts to higher values of
found to be of the first order with respect to methane
18◦ C for ignition and 12◦ C for extinction were taken
and zero order with respect to oxygen. Inhibition by
as an indication that water inhibition is significantly
CO2 at concentrations above 0.5% by volume has been
reduced under real operation.
measured by Ribeiro et al. [54], but it was not con-
In order to develop representative kinetics. however,
firmed by other authors [34,55]. On the other hand,
it is important to collect laboratory-scale kinetic data
strong inhibition by water has been observed, that is
under conditions as close as possible to the industrial
expectedly less important at high temperatures. Ap-
ones. In this respect, a promising device is the an-
parent activation energies of 70–90 and ∼150 kJ/mol
nular reactor proposed by McCarty [43]. It consists
have been calculated in the absence, and in the pres-
of two coaxial tubes: the inner ceramic one, is exter-
ence, of water in the feed respectively. Such values
nally coated with the catalyst; the outer one is made of
can be reconciled if water inhibition is accounted for
quartz. Such a configuration allows for operation at ex-
in the analysis [55].
tremely high space velocities and negligible pressure
As the kinetic studies are customarily performed
drops; it also favours heat dispersion through radia-
under conditions far from reality, extrapolation of the
tion and can provide data under controlled conditions,
kinetics to actual operating ranges may be critical and
e.g. with respect to transport limitations. Accordingly,
must be properly checked. This has been attempted
data can be collected in the temperature range close
for, instance, by performing ignition/extinction tests.
to those of interest for industrial applications. A char-
In these tests, the reaction was accomplished at
acterisation of this reactor from the engineering point
11 atm and with a flow rate of 9300 SLPM. The feed
of view is given in [57].
gas was heated and cooled while the fuel-to-air ratio
was varied in order to keep constant the adiabatic
reaction temperature [56]. Typical ignition/extinction 3.2. Metal substituted hexaaluminate catalysts
curves are shown in Fig. 11: the large hysteresis be-
tween ignition and extinction temperatures is typical Hexaaluminate materials first developed by Arai et
of strongly exothermic reactions. By comparing the al. [20,58] have been extensively investigated for GT
results obtained providing a 200◦ C temperature rise applications in view of excellent thermal stability and
either with only electrical preheating of the feed or catalytic activity [59].
 P. Forzatti, G. Groppi / Catalysis Today 54 (1999) 165–180 175

Fig. 12. Crystal structure of ␤-Al2 O3 and magnetoplumbite phases. (䊉), Alkali, alkali-earth or rare-earth ions; (䊊), oxygen ions; and ( ),
aluminium ions.

These materials, containing transition metal ions reaction occurs via a redox mechanism with variation
in the structure, were prepared by hydrolysis of the of the oxidation state of the metal. A volcano-type
alkoxides [23,60–62] or by co-precipitation from the correlation has been observed between catalytic ac-
nitrates of the constituents by using (NH4 )2 CO3 as tivity and difference of standard formation enthalpies
a precipitating agent [63–66]. Monophasic samples of trivalent and divalent metal-oxides, where Mn is
with a layered alumina structure and surface areas in the most active element. Mn was demonstrated to
the range of 10–15 m2 /g were obtained upon calci- enter the structure at low concentrations (x ≤ 1 in
nation at 1300◦ C. The sintering resistance of hexaa- BaMnx Al12−x O19 ), preferentially in tetrahedral Al(2)
luminates is strictly related to their layered structure sites with dominant oxidation state 2+, with con-
[58] that can be classified into two types: magneto- comitant reduction of the Ba vacancies in the mirror
plumbite and ␤-Al2 O3 . As shown in Fig. 12, they plane due to a charge compensation mechanism, and
both consist of closely packed spinel blocks separated at high concentrations (2 ≤ x ≤3 in BaMnx Al12−x O19 )
by mirror planes that contain large cations of alkali, in octahedral Al(1) sites with the dominant oxidation
alkali-earth and rare-earth ions and loosely packed state 3+ [67]. Introduction of even higher Mn amount
oxygen ions. Spinel blocks and mirror planes are al- results in the segregation of extra phases (BaAl2 O4 )
ternatively stacked, originating a hexagonal structure and in a marked drop of surface area.
with the c-axis parallel to the direction of stacking. The The catalytic activity of substituted hexaaluminates
two structures differ in the composition of the mirror has been investigated in the literature mainly over
plane. These materials crystallise as hexagonal planar powder catalysts in a lab scale micro-reactor [60–66].
particles with marked anisotropic shape because the Methane conversion curves are reported in Fig. 13 for
crystal growth along the c-axis is suppressed due to in- BaMnx Al12−x O19 catalysts with different Mn content.
hibition of ions diffusion along the stacking direction. The catalytic activity increases with Mn loading up
In view of close similarity of ionic radii, transi- to x = 2, keeps constant for x = 3 and then decreases:
tion metal ions (M = Mn, Cu, Fe, Cr, Co, Ni) can the temperature at which 20% conversion is obtained
partially substitute Al ions. The transition metal ions shifts from 660◦ C for x = 0.5 to 560◦ C for x = 2 and
provide significant activity in combustion, since this x = 3, and then back to 700◦ C for x = 4. The conversion
176 P. Forzatti, G. Groppi / Catalysis Today 54 (1999) 165–180

tars, biomass and wastes as alternative fuels or diesel

oils (primarily) as back-up fuels in catalytic combus-
tion for GT application.
The authors are being involved in an EU programme
aimed at the development of an Ultra-Low-Emission-
CATalytic combustor (ULECAT) for GT, able to run
with both gasified biomasses and diesel fuels. Cata-
lyst development and testing is the main target of the
project. The following conclusions have been drawn
so far from the study [72–74]:
1. Scale-up of the activity data collected at the
lab-scale by means of mathematical modelling
Fig. 13. CH4 combustion activity of BaMnxAl12-xO19 [59]. Re- showed that Mn-substituted hexaaluminates are
sults of tests over catalyst powders at GHSV=48000 h-1 with CH4
(1% v/v) in air.
not active enough to satisfy the operating re-
quirements of catalytic combustors fuelled by
gasified biomasses. This is due to the relatively
curves for CO and H2 combustion [68,69] confirmed a high methane content of the fuel that requires
much higher reactivity of Mn-substituted hexaalumi- noble metals based-catalysts for ignition at low
nates towards these fuels: 20% conversion were mea- temperature.
sured over BaMnAl11 O19 at T = 260◦ C for CO and 2. High catalytic activity has been measured for pal-
T = 370◦ C for H2 , by operating with the same fuel ladium based catalysts developed by IFP in both,
concentration and GHSV of Fig. 13. The catalytic ac- the combustion of gasified biomasses and syn-
tivity is also influenced by the composition in the mir- thetic diesel fuel.
ror plane: maximum catalytic activity is reported by 3. Both, palladium-based and Mn-substituted hex-
Arai et al. [70] for Sr0.8 La0.2 MnAl11 O19 . This activ- aaluminate catalysts are very active in the ox-
ity, however, is comparable to that of BaMn2 Al10 O19 . idation of ammonia to NO. However, at high
Mathematical models of monolithic combustor temperatures a decrease in the NO formation is
were used for extrapolation of the laboratory-scale observed, paralleled by the production of N2 ,
data to industrial operating conditions in the case of possibly through selective non-catalytic reduction
BaMn2 Al10 O19 . [71]. Simulation results pointed out (SNCR) mechanism. This aspect deserves further
that in CH4 combustion ignition occurs only above investigation, being relevant to the reduction of
750–800◦ C. Accordingly, hexaaluminates can be used undesired fuel-NOx .
as end-stage catalysts in segmented configurations to 4. Over Pd catalysts, combustion of CO and H2
provide high gas temperature at the exit of the catalyst is not affected by sulphur compounds, whereas
section. This is in line with the configuration adopted methane combustion is markedly depressed. Over
by Osaka Gas [12]. In the case of CO/H2 containing Mn-substituted hexaaluminates, a severe deacti-
mixtures, the ignition temperature over hexaalumi- vation was observed for all fuel components (CO,
nates is ca. 300◦ C lower than with CH4 , but it is still H2 , CH4 , C2 H4 ) and Ba-containing materials
not low enough to fulfill the requirements of gas tur- were irreversibly poisoned.
bine combustor in most demanding conditions (idle 5. Preliminary results, collected at the pilot scale
and partial load). under atmospheric pressure, documented light-off
performances in line with the predictions set forth
on the basis of laboratory tests and by using math-
4. Catalytic combustion of fuels other than ematical modelling.
natural gas 6. For combustion of diesel fuel, problems associ-
ated with slow rate of gas-solid diffusion and high
There is a growing interest in the use of low-heating- homogeneous reactivity should be carefully ad-
value (LHV) fuels obtained by gasification of coal, dressed for the different system configurations.
 P. Forzatti, G. Groppi / Catalysis Today 54 (1999) 165–180 177

Further work must be performed at conditions of ing simulation results of progressively simplified mod-
real interest for industrial applications to assess the els with those of a rigorous one under laminar flow
feasibility of the dual fuel catalytic combustor. conditions [71,76–78]. Such conditions correspond to
the lowest limit of operation of hybrid catalytic com-
bustors and to standard conditions of pilot catalytic
5. Mathematical models of catalytic combustors combustor operated at atmospheric pressure.
1. Inlet effects associated with the development of
Mathematical models represent a powerful tool for temperature and concentration profiles must be
the design, analysis and operation of catalytic com- accounted for due to the short axial length of the
bustor for gas turbines. In the following, the major catalyst segments.
features of mathematical models of the catalyst and 2. The temperature dependence of the gas proper-
homogeneous sections, that have been extensively re- ties must be included since it markedly affects the
viewed in [11,75], are briefly addressed. performances of the combustor.
3. The contribution of axial heat conduction in ce-
5.1. Modelling of the catalyst section ramic monoliths can be neglected as compared to
the extremely high convective transport.
Single channel models are customarily used since 4. For similar reasons and also in view of typical high
assuming global adiabaticity and uniform distribu- aspect ratios, the contribution of heat radiation can
tion of the variables at the catalyst inlet section, all be neglected, but for heat dispersion effects at the
the channels operate under identical conditions. Both catalyst ends [79].
these multi-dimensional models, accounting for the 5. Intra-phase diffusion has to be accounted for since
variable distribution over the channel cross section it does affect the ignition behaviour. E.g. diffusion
and periphery (distributed models), and 1-D models, barriers are used to control catalyst temperature
based on lumping of the cross section and peripheral runaway in methane combustion [49].
distributed variables into appropriate average values 6. The relevance of homogeneous reaction within the
(lumped models), of the monolith single channel have catalyst section depends markedly on the type of
been developed in the literature. fuel, and appears to be negligible for methane.
In principle, many phenomena have to be included However in order to quantify more precisely its
in the model: contribution detailed kinetic schemes have to be
1. heterogeneous reactions at the catalyst wall and considered.
homogeneous reactions in the gas phase; 7. Lumped models, incorporating correlations for lo-
2. heat, mass and momentum transfer by convection cal heat and mass gas-solid transfer coefficients
and diffusion in the gas phase and at the gas-solid that take into appropriate account boundary con-
interface, that are markedly affected by entrance ditions and channel geometry, can be adequate
effects; but for the accurate prediction of the light-off be-
3. mass and heat diffusion in the active catalyst haviour.
phase; and Under turbulent or transitional-flow conditions, that
4. peripheral and axial heat transfer in the solid phase prevail under most operation regimes of GT combus-
through active washcoat and substrate by conduc- tors, modelling of monolithic channels suffers from
tion and radiation. the great complexity of the hydrodynamic phenom-
In view of the number of complex phenomena to be ena. In particular, for a transitional regime neither ac-
accounted for and considering strong non-linearity as- curate correlations for heat/mass transfer coefficients
sociated with temperature effects, the computational in 1-D models, nor rigorous distributed descriptions
labour for solution of the governing equations of rig- are available.
orous models is tedious so as to make any adequate In addition to theoretical investigations on the rela-
approximation quite attractive. tive importance of the several physico-chemical phe-
The following aspects on the relative importance of nomena that might affect the combustor performances,
the above phenomena have been assessed by compar- some practical applications of mathematical models of
178 P. Forzatti, G. Groppi / Catalysis Today 54 (1999) 165–180

perature and the residence time. Considering that the

target CO emission is 10 ppm and the residence time
in a typical GT combustor is of 10–30 ms, the com-
bustor outlet temperature should be 1100◦ C or higher
(and the catalyst exit temperature sufficiently high).

6. Future research opportunities

Areas where future research activities could be rec-

ommended are specified below:
1. Additional tests at the pilot and full-scale labora-
tory and field level have to be done to improve the
reliability of the most promising system configu-
rations and to address the issues still open These
include long-term performances of the catalyst
Fig. 14. Calculated profiles of CO concentrations in the afterburner and of the different components of the combus-
section of a partial combustor [82]. tor, enlargement of the load range of operation,
improvement of coupling between the heteroge-
neous and the homogeneous sections, establish-
the catalyst section are documented in the literature. ment and validation of the control method, spec-
Examples are: prediction of the activity levels required ifications for natural gas and other fuels.
to achieve ignition at relevant operating conditions and 2. The use of fuels other than methane as alternative
scale up of laboratory-scale data [56,71]; evaluation fuels and/or back-up fuels has to be investigated
of thermal stresses of the catalyst [80]; determination more extensively. In the case of LHV fuels the
of the thickness of an inert diffusional barrier required identification of the optimal thermodynamic cycle
to effectively reduce the reaction rate at the catalyst and the definition of the combustor lay out may
inlet, thus mitigating runaway phenomena [81]. Nev- also be considered.
ertheless, the use of mathematical models for practical 3. The dynamics of the PdO–Pd transformation at
purposes is still limited and strong validation efforts the temperature levels of interest for industrial ap-
are required to establish simulation codes as current plications, which is of relevance for ignition run-
design tools. away, has to be studied more in details and com-
pared to the dynamics of the combustion reactions.
5.2. Modelling of the homogeneous section Besides, several other features of Pd catalysts, are
not yet fully understood, deserve additional fun-
The use of consolidated detailed kinetics for gas damental investigations.
phase combustion can provide guide-lines for oper- 4. There is still a need for catalyst with enhanced
ation in the homogeneous section in order to match activity to light off the fuel mixture at the temper-
emission targets within reasonable constraints on com- ature of air delivered from the compressor. Any
bustor size. improvement in the basic understanding of active
Dalla Betta and Loffler [82] simulated the homoge- Pd catalysts may provide a chance in this respect.
neous radical reaction downstream of the catalyst sec- 5. More reliable and comprehensive chemical ki-
tion using a simplified version of the kinetic model of netic rate expressions based on experimental
Miller and Bowman and assuming a plug flow reactor data collected under conditions as close as pos-
model. Simulation results reported in Fig. 14 show that sible to those of industrial interest are required
CO is an intermediate product and that, for a constant for a proper design of catalytic combustors. A
catalyst exit temperature of 950◦ C, the CO concen- contribution along these lines may be provided by
tration strongly depends on the combustor-outlet tem- the development and use of experimental devices
 P. Forzatti, G. Groppi / Catalysis Today 54 (1999) 165–180 179

specifically designed for this purpose, such as the [12] H. Sadamori, Catal. Today 47 (1999) 325.
annular reactor mentioned in the text. [13] J. Saint-Just, J. Der Kinderen, Catal. Today 29 (1996) 387.
[14] S. Ro, A. Scholten, Catal. Today 47 (1999) 415.
6. More sophisticated modelling of the catalyst sec-
[15] Buderus Heiztechnik, EP Appl. 93110601.9 (12.1.94).
tion, and of the homogeneous section, can be at- [16] G. Saracco, I. Cerri, V. Specchia, R. Accornero, Proceedings
tempted to provide an improved description of the of ISCRE15, Newport Beach, 13–16 September 1998, p. 290.
several complex phenomena that govern the pro- [17] Z.R. Ismagilov, M.A. Kenzhentsev, Catal. Rev. Sci. Eng. 31
cess. In any case, model validation by comparison (1990) 51.
[18] I. Stambler, Gas Turbine World, March (1993) p. 23.
of model predictions and experimental results is [19] H. Sadamori, T. Tanioka, T. Matsuhisa, Catal. Today 26
rather scarce so far. This is mainly due to the lack (1995) 337.
of detailed experimental data at full scale and to [20] M. Machida, K. Eguchi, H. Arai, Chem. Letts. (1987) p 767.
the fact that experimental and modelling works [21] Y. Ozawa, Y. Tochihara, N. Mori, I. Yuri, T. Kanazawa, K.
are carried out by different research groups. Joint Sagimori, ASME Paper 98-GT-381, 1998.
[22] R.A. Dalla Betta, N. Ezawa, K. Tsurumi, J.C. Schlatter, S.G.
efforts from experimental and theoretical research Nickolas, US Patent 5183401, 2 February 1993.
groups are strongly recommended. [23] R.A. Dalla Betta, K. Tsurumi, N. Ezawa, US Patent 5232257,
Other aspects that have not been addressed in this 3 August 1993.
paper need further study. These include investigation [24] R.A. Dalla Betta, F.H. Ribeiro, T. Shoji, K. Tsurumi, N.
on structural and catalytic durability of combustion Ezawa, S.G. Nickolas, US Patent 5250489, 5 October 1993.
[25] J.C. Schlatter, R.A. Dalla Betta, S.G. Nicholas, M. B. Cutrone,
catalysts at high temperatures and also development K.W. Beebe, T. Tsuchiya, ASME Paper 97-GT-57, 1997.
of improved pre-burned designs with lower formation [26] R.A. Dalla Betta, J.C. Schlatter, S.G. Nicholas, M.B. Cutrone,
of NOx . K.W. Beebe, Y. Furuse, T. Tsuchiya, ASME Paper 96-GT-485,
1996.
[27] R.J. Farrauto, M.C. Hobson, T. Kennelly, E.M. Waterman,
Appl. Catal. A: Gen. 81 (1992) 227.
Acknowledgements [28] Unpublished results from our laboratory.
[29] J. Chen, E. Ruckenstein, J. Catal. 69 (1981) 254.
The authors acknowledge the financial support of [30] J. Chen, E. Ruckenstein, J. Phys. Chem. 85 (1981) 1606.
MURST (Rome, Italy). Thanks are also due to Drs. [31] E. Ruckenstein, J. Chen, J. Catal. 70 (1981) 233.
[32] E. Ruckenstein, J. Chen, J. Coll. Interface Sci. 86 (1982) 1.
Dalla Betta and McCarty of Catalytica Inc. and to [33] H. Lieske, J. Volter, J. Phys. Chem. 89 (1985) 1841.
Dr. Sadamori of Osaka Gas for providing valuable [34] R.F. Hicks, H. Qi, M.L. Young, R.G. Lee, J. Catal. 122 (1990)
unpublished material. 295.
[35] E. Garbowski, C. Feumi-Jantou, N. Mouaddid, M. Primet,
Appl. Catal. A: Gen. 109 (1994) 277.
References [36] R. Burch, F.J. Urbano, Appl. Catal. A: Gen. 124 (1995) 121.
[37] K. Otto, C. Hubbard, W. Weber, G. Graham, Appl. Catal. B:
Environm. 1 (1992) 317.
[1] D.L. Trimm, Appl. Catal. 7 (1983) 249. [38] A. Ogata, A. Obuchi, K. Mizuno, A. Ohi, H. Obuchi, J. Catal.
[2] R. Prasad, L.A. Kennedy, R.E. Ruckenstein, Catal. Rev. Sci. 144 (1993) 452.
Eng. 26 (1984) 1. [39] J.W.M. Jacobs, D. Schryvers, J. Catal. 103 (1987) 436.
[3] J.P. Kesserling, in: F.J. Winberg (Ed.), Advanced Combustion [40] T. Fleish, R. Kicks, A. Bell, J. Catal. 87 (1984) 398.
Methods, Academic Press, London, 1986. [41] S.C. Su, J.N. Carstens, A.T. Bell, J. Catal. 176 (1998) 125.
[4] L.D. Pefferle, W.C. Pfefferle, Catal. Rev. Sci. Eng. 29 (1987) [42] E. Voogt, PhD Thesis, University of Utrecht, 1997.
219. [43] J. McCarty, Catal. Today 26 (1995) 283.
[5] M. Zwinkels, S. Jaras, P.G. Menon, Catal. Rev. Sci. Eng. 35 [44] M Lyubovski, R. Weber, L. Pfefferle, 26th International
(1993) 319. Symposium on Combustion, 1779, 1996.
[6] S.T. Kolaczkowski, Trans. I. Chem. Epart A 73 (1995) 168. [45] J.N. Carstens, S.C. Su, A.T. Bell, J. Catal. 176 (1998) 136.
[7] K. Eguchi, H. Arai, Catal. Today 29 (1996) 379. [46] M. Lyubovsky, L. Pfefferle, Appl. Catal. A: Gen. 173 (1998)
[8] R.A. Dalla Betta, Catal. Today 35 (1997) 129. 107.
[9] J.G. McCarty, M. Gusman, D.M. Lowe, D.L. Hildenbrand, [47] K. Fujimoto, F.H. Ribeiro, M. Avalos-Borja, E. Iglesia, J.
K.N. Lau, Catal. Today 47 (1999) 5. Catal. 179 (1998) 431.
[10] J.G. McCarty. 3rd Europacat, Krackow, September 1997, [48] T. Griffin, W. Weisenstein, V. Scherer, M. Fowles, Combust.
Plenary lecture. Flames 101 (1995) 81.
[11] G. Groppi, E. Tronconi, P. Forzatti, Catal. Rev. Sci. Eng. [49] R.A. Dalla Betta, K. Tsurumi, T. Shoji, R.L. Garten, US
41(2) (1999) 227. Patent 5405260, 11 April 1995.
180 P. Forzatti, G. Groppi / Catalysis Today 54 (1999) 165–180

[50] Y.Y. Yao, Ind. Eng. Chem Prod. Res. Dev. 19 (1980) 293. [68] G. Groppi, A. Belloli, E. Tronconi, P. Forzatti, Catal. Today
[51] S.H. Oh, P.J. Mitchell, R.M. Siewert, J. Catal. 132 (1991) 29 (1996) 403.
287. [69] C. Cristiani, G. Groppi, P. Forzatti, E. Tronconi, G. Busca,
[52] P. Briot, M. Primet, Appl. Catal. 68 (1991) 301. M. Daturi, Studies Surf. Sci. Catal. 101 (1996) 473.
[53] T.R. Baldwin, R. Burch, Appl. Catal. 66 (1990) 337. [70] M. Machida, K. Eguchi, H. Arai, J. Catal. 123 (1990) 477.
[54] F.H. Ribeiro, M. Chow, R.A. Dalla Betta, J. Catal. 146 (1994) [71] G. Groppi, E. Tronconi, P. Forzatti, Appl. Catal. A 138 (1996)
537. 177.
[55] J.C. van Giezen, PhD Thesis, the University of Utrecht, [72] G. Groppi, L. Lietti, E. Tronconi, P. Forzatti, Catal. Today
1997. 45 (1998) 159.
[56] R.A. Dalla Betta, J.C. Schlatter, D.Y. Kee, D.G. Loffler, T. [73] M. Johansson, PhD Thesis, KTH, Sweden, 1998.
Shoji, Catal. Today 26 (1995) 329. [74] J.H. Legal, G. Martin, D. Durand, ASME paper 98-GT-294,
[57] A. Beretta, P. Baiardi, D. Prina, P. Forzatti, Chem. Eng. Sci. 1998.
54 (1999) 765. [75] R.E. Hayes, S.T. Kolackzowski, Introduction to Catalytic
[58] H. Arai, US Patent 4788174, 29 November 1988. Combustion, Gordon and Breach, 1997.
[59] G. Groppi, C. Cristiani, P. Forzatti, Catal. 13 (1997) 85. [76] G. Groppi, A. Belloli, E. Tronconi, P. Forzatti, A.I.Ch.E. J.
[60] M. Machida, K. Eguchi, H. Arai, J. Catal. 103 (1987) 385. 41 (1995) 2250.
[61] M. Machida, K. Eguchi, H. Arai, Bull. Chem. Soc. Jpn. 61 [77] G. Groppi, A. Belloli, E. Tronconi, P. Forzatti, Chem. Eng.
(1988) 3659. Sci. 50 (1995) 2705.
[62] M. Machida, K. Eguchi, H. Arai, J. Catal. 120 (1989) 377. [78] E. Tronconi, P. Forzatti, A.I.Ch.E. J. 38 (1992) 201.
[63] G. Groppi, M. Bellotto, C. Cristiani, P. Forzatti, P.L. Villa, [79] R.E. Hayes, S.T. Kolackzowski, W.J. Thomas, Comp. Chem.
Appl. Catal. A 104 (1993) 101. Eng. 16 (1992) 645.
[64] G. Groppi, M. Bellotto, C. Cristiani, P. Forzatti, J. Mat. Sci. [80] Y. Tsujikawa, S. Fujii, H. Sadamori, S. Ito, S Katsura, ASME
29 (1994) 3441. Paper, 95-GT-351, 1995.
[65] G. Groppi, F. Assandri, M. Bellotto, C. Cristiani, P. Forzatti, [81] D. Leung, R.E. Hayes, S.T. Kolackzowski, Can. J. Chem.
J. Sol. Stat. Chem. 114 (1995) 326. Eng. 74 (1996) 94.
[66] G. Groppi, C. Cristiani, P. Forzatti, J. Catal. 168 (1997) 95. [82] R.A. Dalla Betta, D.G. Loffler, ACS Symp. Ser. Hetero-
[67] M. Bellotto, G. Artioli, C. Cristiani, P. Forzatti, G. Groppi, geneous Hydrocarbon Oxidation, 638 (1996) 36
J. Catal. 179 (1998) 597. [83] H. Kleykamp, Z. Physik. Chem. N.F. 71 (1970) 142.