Carbon dioxide reforming of methane to synthesis gas was studied over Nibased catalysts. It is shown that, in contrast to other Nibased catalysts which exhibit continuous deactivation with timeonstream, the rate over the Ni=La2O3 catalyst increases during the initial 2–3 h of reaction and then tends to be essentially invariable, displaying very good stability. Xray di5raction, hydrogen and CO uptake studies, as well as highresolution TEM indicate that, under reaction conditions, the Ni particles are partially covered by La2O2CO3 species which are formed by interaction of La2O3 with CO2. Catalytic activity occurs at the Ni–La2O2CO3 interface, while the oxycarbonate species participate directly by reacting with deposited carbon, thus restoring the activity of the Ni sites at the interface. XPS and FTIR studies provide evidence in support of this mechanistic scheme. It was also found that methane cracking on Ni sites and surface reaction between deposited carbon and oxycarbonate species are the rate determining steps in the reaction sequence. A kinetic model is developed based on this mechanistic scheme, which is found to predict satisfactorily the kinetic measurements. ? 2003 International Association for Hydrogen Energy. Published by Elsevier Scie
Trang 1International Journal of Hydrogen Energy 28 (2003) 1045–1063
www.elsevier.com/locate/ijhydene
Catalytic dry reforming of natural gas for the production of
chemicals and hydrogen
Xenophon E Verykios∗
Department of Chemical Engineering, University of Patras, GR 26500 Patras, Greece
Received 3 June 2002; accepted 16 September 2002
Abstract
Carbon dioxide reforming of methane to synthesis gas was studied over Ni-based catalysts It is shown that, in contrast
to other Ni-based catalysts which exhibit continuous deactivation with time-on-stream, the rate over the Ni=La2O3 catalyst increases during the initial 2–3 h of reaction and then tends to be essentially invariable, displaying very good stability X-ray di5raction, hydrogen and CO uptake studies, as well as high-resolution TEM indicate that, under reaction conditions, the Ni particles are partially covered by La2O2CO3species which are formed by interaction of La2O3 with CO2 Catalytic activity occurs at the Ni–La2O2CO3interface, while the oxycarbonate species participate directly by reacting with deposited carbon, thus restoring the activity of the Ni sites at the interface XPS and FTIR studies provide evidence in support of this mechanistic scheme It was also found that methane cracking on Ni sites and surface reaction between deposited carbon and oxycarbonate species are the rate determining steps in the reaction sequence A kinetic model is developed based on this mechanistic scheme, which is found to predict satisfactorily the kinetic measurements
? 2003 International Association for Hydrogen Energy Published by Elsevier Science Ltd All rights reserved
Keywords: Catalytic reforming; Methane; Natural gas; Synthesis gas; Hydrogen; Nickel; Lanthana; Carbon dioxide
1 Introduction
Conversion of methane and carbon dioxide, which are
two of the cheapest and most abundant carbon-containing
materials, into useful products is an important area of
cur-rent catalytic research In this regard, the process of
re-forming methane with carbon dioxide is of special interest
since it produces synthesis gas with low hydrogen-to-carbon
monoxide ratio, which can be preferentially used for
pro-duction of liquid hydrocarbons in the Fischer–Tropsch
syn-thesis network [1] This reaction has also very important
environmental implications because both methane and
car-bon dioxide are greenhouse gases which may be converted
into valuable feedstock In addition, this process has
po-tential thermochemical heat-pipe applications for the
recov-ery, storage and transmission of solar and other renewable
∗Tel.: +30-610-997-826; fax: +30-610-991527.
E-mail address: verykios@chemeng.upatras.gr
(X.E Verykios).
energy sources by use of the large heat of reaction and the reversibility of this reaction system [2,3] One of the major problems encountered in the application of this process is rapid deactivation of the catalyst, mainly by carbon deposi-tion [4,5]
During the past decades, the process of carbon dioxide reforming of methane has received attention, and e5orts have focused on development of catalysts which show high activity towards synthesis gas formation, and are also resis-tant to coking, thus displaying stable long-term operation Numerous supported metal catalysts have been tested for this process Among them, nickel-based catalysts [6 11] and supported noble metal catalysts (Rh, Ru, Ir, Pd and Pt) [12–22] give promising catalytic performance in terms
of methane conversion and selectivity to synthesis gas Conversions of CH4 and CO2to synthesis gas approaching those deDned by thermodynamic equilibrium can be ob-tained over most of the aforementioned catalysts, as long
as reaction temperature and contact time are suEciently high [8,10,12,13] The catalysts based on noble metals are reported to be less sensitive to coking than are nickel-based 0360-3199/03/$ 30.00 ? 2003 International Association for Hydrogen Energy Published by Elsevier Science Ltd All rights reserved PII: S0360-3199(02)00215-X
Trang 2catalysts [8,10,12,13,21–23] However, considering the
as-pects of high cost and limited availability of noble metals,
it is more desirable, from the industrial point of view, to
develop nickel-based catalysts which are resistant to carbon
deposition and exhibit stable operation for extended periods
of time Arakawa et al [24–27] used a Ni=Al2O3 catalyst
to obtain synthesis gas from a mixture of methane, carbon
dioxide and water They found that the catalyst deactivates
rapidly by carbon formation on the surface, but addition
of vanadium (5–10 wt%) can decrease, to a certain extent,
coke formation Rapid catalyst deactivation due to carbon
deposition on supported Ni catalysts during the CH4=CO2
reaction was observed by many investigators [6,7,16,23,28]
It is generally claimed that catalyst deactivation is due to
coke formation within the pores of the catalyst, which leads
to breakup of the catalyst particles Carbon dioxide
reform-ing of methane over Ni supported on di5erent carriers was
studied in detail by Gadalla and co-workers [8,10] They
found that no carbon deposition was obtained when reaction
temperatures higher than 940◦C and CO2=CH4ratios larger
than 2 were applied Due to the high temperature, however,
the support structure was found to be changing and the
activity to be decreasing with time on stream because of
reduction of surface area Rostrup-Nielsen [29,30] observed
that adsorption of sulphur atoms results in deactivation of
the neighbouring nickel atoms and that the rate of carbon
formation decreases more rapidly with sulphur coverage
than the reforming rate This suggests that the ensemble
for the reforming reaction is smaller than that required for
nucleation of carbon whiskers Based on this Dnding, the
SPARG (sulphur-passivated reforming) process has been
developed for CO2reforming of methane [31] By partially
sulDding the Ni catalyst, the sites for carbon formation are
blocked while suEcient sites for the reforming reaction are
maintained This permits the catalytic reaction to take place
without signiDcant coking problems However, catalytic
ac-tivity is sacriDced to a large extent Addition of an oxide of
strong basicity (e.g., alkali, alkaline oxide) to Ni-based
cat-alysts has been known to be an eEcient way for reduction of
coking Recently, Yamazaki et al [11] obtained carbon-free
operation of carbon dioxide reforming of methane at 850◦C
by addition of CaO to Ni/MgO catalyst Kinetic studies
showed that the CaO-promoted catalyst has higher aEnity
for CO2 chemisorption It was reasoned that the enhanced
CO2 chemisorption may promote the reaction with coke
precursors from methane, thus preventing accumulation of
coke However, a signiDcant reduction in activity of the
Ni/MgO catalyst was observed by addition of the strongly
basic CaO component Swaan et al [32] studied
deactiva-tion of supported Ni catalysts during reforming of methane
with carbon dioxide They found that Ni=ZrO2; Ni=La2O3,
Ni=SiO2and Ni–K=SiO2exhibit moderate deactivation with
zero order kinetics The deactivation was shown to be due
to carbon deposition on Ni from CO disproportionation
In the present communication, our work on the dry
re-forming of methane over the Ni=La2O3 [33–39] catalyst is
reviewed This catalyst, when properly prepared and acti-vated is capable of exhibiting good activity and, primar-ily excellent stability under conditions of CO2reforming of methane Particular attention is directed towards understand-ing the reasons for the unique behaviour of the Ni=La2O3
catalyst, in relation to a detailed surface mechanistic scheme for the reaction For this purpose, a number of experimental techniques are employed, including steady-state and tran-sient kinetic experiments, FTIR of adsorbed species, XPS, high resolution TEM, and others A fairly detailed descrip-tion of structural aspects and surface transformadescrip-tion steps emerges from combination of the results of these techniques
2 Experimental 2.1 Catalyst preparation Ni=La2O3, Ni=-Al2O3 and Ni/CaO catalysts were pre-pared by the wet-impregnation method, using nitrate salt
as the metal precursor A weighed amount of nickel ni-trate (Alfa Products) was placed in an 100 ml beaker and
a small amount of distilled water was added After 30 min, the appropriate weight of support (La2O3) was added un-der continuous stirring The slurry was heated to ca 80◦C and maintained at that temperature until the water evapo-rated The residue was then dried at 110◦C for 24 h and was subsequently heated to 500◦C under N2 Now for 2 h for complete decomposition of the nitrate After this treat-ment, the catalyst was reduced at 500◦C in H2 Now for at least 5 h
2.2 Kinetic measurements Kinetic studies under di5erential conditions, and stud-ies under integral reaction conditions were conducted
in a conventional Now apparatus consisting of a Now measuring and control system, a mixing chamber, a quartz-Dxed-bed reactor (ca 4 mm, i.d.), and an on-line gas chromatograph The feed stream typically consisted of
CH4=CO2=He=20=20=60 vol% For the kinetic studies under di5erential conditions, one portion of catalyst (5–10 mg) was diluted with 2–4 portions of -Al2O3 The solid mix-ture was powdered (d = 40 m) before being placed at the middle of the reaction tube Conversions were usually controlled to be signiDcantly lower than those deDned by thermodynamic equilibrium by adjusting the total Now rate (200–400 ml=min) Rate limitations by external or internal mass transfer, under di5erential conditions were proven to
be negligible by applying suitable criteria For the stud-ies under integral reaction conditions, one portion of the catalyst (10–50 mg) was diluted with up to 10 portions
of -Al2O3 so as to reduce the temperature gradient along the catalyst bed The temperature of the catalyst bed was measured by a chromel–alumel thermocouple, and it was
kept constant within ±2 ◦C Analysis of the feed stream
Trang 3and reaction mixture was performed using the TC detector
of a gas chromatograph Prior to reaction, the catalyst was
reduced again, in situ, at 750◦C in H2 Now for 1 h
2.3 Catalyst characterization
2.3.1 H2andCO chemisorption
H2 and CO chemisorption on Ni catalysts was studied
at room temperature H2 chemisorption was determined in
a constant-volume high vacuum apparatus (Micromeritics,
Accusorb 2100E) The adsorption isotherms were
mea-sured at equilibrium pressures between 10 and 300 mm
Hg CO chemisorption was conducted in a Now apparatus
which is connected to a quadrupole mass spectrometer
(Fisons, SXP Elite 300 H) Prior to adsorption
measure-ments, the samples were pre-reduced in H2 Now at 750◦C
for 2 h
2.3.2 XRD study
A Philips PW 1840 X-ray di5ractometer was used to
iden-tify the main phases of Ni=La2O3 catalysts, before and
af-ter reaction Anode Cu K (40 kV; 30 mA; = 1:54 QA) was
used as the X-ray source The catalyst which had been
ex-posed to reaction conditions for a certain period of time was
quickly quenched to room temperature and then transferred
onto the XRD sample holder for measurements The mean
nickel particle size was estimated by employing Scherrer’s
equation, following standard procedures
2.3.3 TPD
Temperature-programmed desorption (TPD) experiments
were carried out in an apparatus which consists of a Now
switching system, a heated reactor, and an analysis system
The reactor was a quartz tube of 0:6 cm diameter and 15 cm
length A section at the centre of the tube was expanded
to 1:2 cm diameter, in which the catalyst sample,
approxi-mately 300 mg, was placed The outlet of the reactor was
connected to a quadrupole mass spectrometer via a heated
silicon capillary tube of 2 m length The pressure in the
main chamber of the mass spectrometer was approximately
10−7mbar
The sample was Drst reduced in H2 Now at 750◦C for
more than 2 h After purging with He for 10 min, the sample
was cooled under He Now When the desired adsorption
temperature was reached, the He Now was switched to H2
or CO Now After 10 min, the sample was cooled to room
temperature under H2 or CO Now, and then the Now was
switched to He and the lines were cleaned for 2–5 min
Temperature programming was then initiated and the TPD
proDles were recorded
2.4 Surface analysis
A Nicolet 740 FTIR spectrometer equipped with a
DRIFT (di5use reNectance infrared Fourier transform) cell
was used for the measurement of surface species formed on
the Ni=La2O3catalyst The cell, containing ZnSe windows, which were cooled by water circulating through blocks in thermal contact with the windows, allowed collection of spectra over the temperature range 25–700◦C and at atmo-spheric pressure For all the spectra recorded, a 32-scan data accumulation was carried out at a resolution of 4:0 cm−1
An IR spectrum obtained under Ar Now (before the re-action) was used as the background to which the spectra, after reaction, were ratioed Because the surface species
on the working Ni=La2O3catalyst require a long time (ap-proximately 5 h) to reach a stable level and because the
IR cell cannot be exposed to the reaction conditions for
a long period of time, measurements were carried out ex situ to follow the change of the surface species with time
on stream, i.e., the treated sample was quickly quenched
to room temperature and transferred to the FTIR sample holder for measurements
XPS data were obtained with a Vacuum Science Work-shop X-ray anode using magnesium K radiation and a
100 mm hemispherical analyser The binding energies were corrected for charging by reference to adventitious car-bon at 284:8 eV, and signal intensities were corrected for cross-section and escape depth using Wagner’s sensitivity factors The La 3d, Ni 2p, O 1s, and C 1s signals were measured at a takeo5 angle normal to the sample
3 Kinetic behaviour of the Ni=La2O3 and other Ni-based catalysts
3.1 Catalytic performance of Ni-basedcatalysts Fig 1 shows the alteration of reaction rate, obtained under di5erential reaction conditions at 750◦C, over Ni=-Al2O3, Ni/CaO and Ni=La2O3 catalysts, as a func-tion of time on stream The feed stream consisted of
CH4=CO2=He = 20=20=60 vol%, while a total Now rate of
300 ml=min was used As shown in Fig 1, the intrinsic rates of methane reforming with CO2 over the Ni=-Al2O3
and Ni/CaO catalysts decrease continuously with time on stream In contrast, the rate over the Ni=La2O3 catalyst increases with time on stream during the initial 2–5 h of reaction, and then it tends to be essentially invariable with time on stream during 100 h of reaction, showing very good stability
Table 1 reports the reaction rates obtained over the Ni=-Al2O3, Ni/CaO and Ni=La2O3 catalysts at 550◦C,
650◦C and 750◦C Both reaction rates, measured initially and after 5 h of reaction, are presented For the Ni=La2O3
catalyst, the rate measured at 650◦C and 750◦C after 5 h of reaction corresponds to the rate at the stable level (Fig.1) The rate obtained over the Ni=La2O3at 550◦C shows a very slow increase with time on stream, which lasts for at least
10 h The rate at the pseudo-stable level at 550◦C amounts
to ca 0:18 mmol=(g s) which is signiDcantly lower than the one obtained at 550◦C, following Drst reaction at 750◦C
Trang 40 8 16
Time / h
0
0.5
1
1.5
2
2.5
3
RCO
Ni Catalyst, 750oC
100
CaO
Al2O3
La2O3
Fig 1 Alteration of reaction rate of carbon dioxide reforming
of methane to synthesis gas as a function of time on stream
over Ni=La2O 3 , Ni=-Al 2 O 3 and Ni/CaO catalysts Reaction
con-ditions: P CH4= 0:2 bar; P tot = 1:0 bar, CH 4 =CO 2 = 1; T = 750◦C,
W=F = 2 × 10 −3g s=ml, metal loading = 17 wt%.
for 5 h and decrease of temperature to 550◦C (Table1)
Apparently, the stable structure of the Ni=La2O3 catalyst is
favourably produced when the reaction temperature applied
is higher than 650◦C It is shown that the initial reaction rate
over Ni=-Al2O3 is ca 2 times higher than the respective
ones over Ni=La2O3 and Ni/CaO However, the reaction
rate over Ni=La2O3 at the stable level is higher than the
ones over the deactivated Ni=-Al2O3and Ni/CaO catalysts
It is well known that the stability of the catalysts may be
strongly a5ected by reaction temperature For this reason the
Table 1
InNuence of catalyst support on reaction rate at various temperatures over supported Ni catalyst
7 wt% Ni/
a Reaction conditions: P CH 4 = 0:2 bar; P tot = 1:0 bar; CH 4 =CO 2= 1; W=F = 2 × 10 −3 g s=ml.
b The data were obtained following initial reaction at 750◦C for 5 h and decrease of temperature from 750◦C to 650◦C and 550◦C.
Time / h
0 1 2 3
RCO
CH4/CO2 at 750˚C
750˚C
650˚C
550˚C
Fig 2 Alteration of reaction rate as a function of time on stream over the Ni=La2O 3 catalyst Reaction conditions: T =550◦C, 650◦C and 750◦C; P CH4= 0:2 bar, P tot = 1:0 bar, CH 4 =CO 2 = 1, metal loading = 17 wt%.
stability of the Ni=La2O3catalyst was investigated at 550◦C,
650◦C and 750◦C under di5erential reaction conditions and the variation of the rate of reaction with time-on-stream is shown in Fig.2 The Ni=La2O3 catalyst was Drst exposed
to the CH4=CO2 mixture at 750◦C until the reaction rate reached the stable level (Fig 1) It is shown that the re-sultant Ni=La2O3catalyst does not exhibit any deactivation during 20 h of reaction at these temperatures These results demonstrate the excellent stability of the Ni=La2O3 catalyst since it is known that even supported noble metal catalysts
do su5er carbon deposition and deactivation at reaction tem-peratures below 700◦C [12,13,20,21]
Trang 50 5 10 15 20
Time / h
0
10
20
30
40
50
RCO
Ni/La2O3, 1023K
3 wt.%
10 wt.%
17 wt.%
Fig 3 InNuence of Ni metal loading on reaction rate
and stability of the Ni=La2O 3 catalyst Reaction conditions:
P CH4= 0:2 bar; P tot = 1:0 bar; CH 4 =CO 2 = 1; T = 750◦C; W=F =
2 × 10 −3g s=ml.
3.2 In9uence of structural andoperating parameters on
kinetic behaviour
3.2.1 Ni metal loading
Fig.3shows the inNuence of metal loading (3–17 wt%)
on the reaction rate and the stability of Ni=La2O3
cata-lyst at 750◦C The reaction rate is expressed in units of
mmol=(gmetal s) It is observed that decreasing the nickel
loading on the Ni=La2O3 catalyst results in increase of the
reaction rate, presumably due to enhanced dispersion of Ni
on the Ni=La2O3support Regardless of di5erent metal
load-ings, a similar pattern, i.e the rate increasing with time on
stream during the initial several hours of reaction, is
ob-served After reaching a maximum level, the reaction rate
decreases gradually over the 3 wt% Ni=La2O3catalyst, but
tends to be essentially invariable with time on stream when
the nickel loading is increased to above 10 wt% It appears
that stable performance is favourably obtained over the
cat-alyst with large metal particle size
3.2.2 In9uence of contact time andtemperature
The inNuence of contact time on conversions of methane
and carbon dioxide over a 17 wt% Ni=La2O3 catalyst was
investigated at 750◦C The feed consisted of CH4=CO2=He=
20=20=60 vol% The alteration of contact time was
real-ized by adjusting both, the amount of catalyst (5–30 mg)
and the feed Now rate (30–300 ml=min) As shown in
Fig 4(a), both methane and carbon dioxide conversion
increases rapidly as contact time increases from 0.002 to
0:07 g s=ml Conversions approaching those expected at
(a)
(b)
0 0.02 0.04 0.06 0.08
Contact Time /gs/ml
0 25 50 75 100
Equilibrium Level
( CO2
CH4
500 600 700 800 900
Temperature / °C
0 25 50 75 100
CH4/CO2 =1
PCH4=0.2 bar
CO2
CH4
Fig 4 (a) InNuence of contact time on conversion obtained over the Ni=La2O 3 catalyst The dotted lines correspond to val-ues expected at thermodynamic equilibrium Reaction conditions:
P CH 4 = 0:2 bar; P tot = 1:0 bar; CH 4 =CO 2 = 1; T = 750◦C, metal loading = 17 wt% (b) InNuence of reaction temperature on con-version obtained over the Ni=La2O 3 catalyst using a constant contact time of 0:06 g s=ml The dotted lines correspond to version expected at thermodynamic equilibrium Reaction con-ditions: P CH4 = 0:2 bar; P tot = 1:0 bar, CH 4 =CO 2 = 1, metal loading = 17 wt%.
thermodynamic equilibrium (i.e the dotted lines) are al-ready achieved at contact times as low as ca 0:06 g s=ml, which correspond to a superDcial contact time of ca 0:02 s The conversions of methane and carbon dioxide obtained
at a contact time of 0:06 g s=ml was also studied at var-ious temperatures and the results are shown in Fig.4(b)
It is observed that the conversions obtained at various
Trang 6Table 2
InNuence of pretreatment of 17 wt% Ni=La2O 3 catalyst on reaction rates at the initial and stable levels
Exp Pretreatments Favourable compound a Rate for CO formation (mmol=g s) b
no.
then H 2 ; 1023 K; 2 h
a This compound is expected to be formed in preference following the stated pretreatment.
b Reaction conditions: P CH 4 = 0:2 bar; P tot = 1:0 bar; CH 4 =CO 2= 1; W=F = 2 × 10 −3 g s=ml; T = 750◦C.
temperatures, employing the speciDed contact time, are
approximately equal to those expected at thermodynamic
equilibrium (i.e the dotted lines) The high intrinsic
ac-tivity of Ni=La2O3 may be related to its absence of strong
alkali- and/or alkaline-promoter (La2O3 has only
moder-ate basicity) on the Ni catalyst It is well known [40] that
strong basic promoters help to inhibit accumulation of
sur-face coke but also result in signiDcant reduction of activity
or reforming-type reaction
3.2.3 In9uence of gas (pre)treatment
The inNuence of various gas pretreatments, including
heating under Now of O2, air, H2, CO2and CH4 at 1023 K
for 1–2 h, on the performance of the Ni=La2O3 catalyst
was investigated Table 2 reports the results obtained at
the initial state of the catalyst and after reaching the stable
level, following various pretreatments The pretreatment of
the Ni=La2O3 with CO2; O2 and air at high temperatures
(Experiments No 2, 3, 8, 9) would favour the formation
of La2O2CO3, NiO and LaNiO3, respectively From
ex-periments No 2, 3, 8 and 9, it is derived that none of the
compounds La2O2CO3, NiO and LaNiO3 is likely to be
solely responsible for the enhancement of the reaction rate
The results obtained in experiments No 5–7 indicate that
the increase of reaction rate during the initial several hours
of reaction is not due to in situ reduction of incompletely
reduced nickel since nickel is expected to be fully reduced
after exposure to pure H2 Now at 750◦C for 12 h
(experi-ment No 7) Although the pretreat(experi-ment a5ects the rate of
the initial state to a certain extent, it does not inNuence
sig-niDcantly the value of the reaction rate at the stable level
These results imply that there exists a strong tendency of
the Ni=La2O3 catalyst to form the stable surface structure
only under the working reaction conditions
Fig.5shows the inNuence of several treatments on the
reaction rate over the Ni=La2O3 catalyst, following the
Time / hour
0 0.6 1.2 1.8 2.4 3
Treatment
Š After H2 at 750˚C for 2h
/ After exposed to air at 30˚C After O2 at 750˚C for 2h
RCO
Fig 5 E5ect of various treatments on reaction rate over the Ni=La2O 3 catalyst, following establishment of the stable sur-face state Reaction conditions: P CH4= 0:2 bar, P tot = 1:0 bar,
CH 4 =CO 2 = 1, T = 750◦ C; W=F = 2 × 10 −3g s=ml.
tablishment of the stable surface state It is found that the stable surface is insensitive to exposure of the catalyst to air
at room temperature It is interesting to observe that when the catalyst is exposed to H2(or to O2) at 750◦C, following establishment of the stable surface state, evolution of CH4
(or of CO2) is registered Consequently, the stable surface structure is altered or destroyed, as indicated by the lower reaction rates which are obtained upon re-exposing the cat-alyst to the reaction mixture at the same temperature How-ever, the stable surface structure is found to be essentially
Trang 7retrievable after several hours of reaction (Fig.5) These
re-sults may imply that carbon itself may constitute an
imper-ative component contained in the stable surface structure
The results of Experiment No 4 (catalyst was pretreated
with CH4at 750◦C) given in Table2show the initial rate is
smaller but rather close to that at the stable level, suggesting
that the presence of a certain amount of carbon on Ni
crys-tallites favours the enhancement of the reaction rate The
higher initial rate might be due to accumulated carbon on
the surface which react with CO2to produce synthesis gas
3.3 Integral reactor performance
The results presented in the preceding sections were all
obtained using a dilute reaction mixture, i.e CH4=CO2=He
= 20=20=60 vol%, and the conversions were usually
con-trolled to be far below those expected by thermodynamic
equilibrium In this section, results of the long-term
sta-bility test of the Ni=La2O3 catalyst under integral reaction
conversions, with and without He dilution, are presented
Conversion somewhat lower than the equilibrium one was
achieved This allows to study the catalytic performance at
high conversions, while the catalyst deactivation, if there is
any, can also be easily detected
Fig.6(a) shows the alteration of conversion of methane
and carbon dioxide, and selectivity to carbon monoxide
and hydrogen as a function of time on stream, obtained
at 750◦C over the Ni=La2O3 catalyst using a feed mixture
of CH4=CO2=He = 20=20=60 vol% Both conversion and
selectivity increase during the initial several hours of
reac-tion After this, the conversion and selectivity tends to be
essentially invariable with time on stream during 100 h of
reaction Results of a similar long-term stability test,
con-ducted employing undiluted feed (CH4=CO2=50=50 vol%)
under otherwise similar conditions, are shown in Fig.6(b)
Even in this case, after several hours of reaction, both
con-version and selectivity tend to be rather stable Only a small
decline of activity with time-on-stream was observed during
the 100 h stability test It is found that the slow deactivation
which is observed in Fig.6(b) could be largely eliminated by
addition of small quantities (1–5%) of oxygen in the feed
4 Characterization of the Ni=La2O3catalyst
4.1 XRD study
The major crystalline phases of the Ni=-Al2O3 and
Ni=La2O3 catalysts were examined by XRD and are
de-scribed in Table 3 The results show that -Al2O3 and
NiAl2O4crystalline phases exist in the reduced Ni=-Al2O3
catalyst (fresh) The NiAl2O4 phase, which is not easily
reducible, should originate from the reaction between NiO
and Al2O3 No metallic Ni crystalline phase is observed in
Ni=Al2O3(Table3) Only metallic Ni and La2O3crystalline
phases are found in the reduced Ni=La2O3 catalyst (fresh)
(a)
(b)
Time / hour
40 60 80 100
, CO Selectivity
( H2 Selectivity
/ CO2 Conversion
CH4 Conversion
Time / hour
0 25 50 75 100
CO Selectivity
H2 Selectivity
CO2 Conversion
CH4 Conversion
Fig 6 (a) Alteration of conversion of CH 4 and CO 2 and selectiv-ity to CO and H 2 , obtained over a 17 wt% Ni=La2O 3 catalyst, as
a function of time on stream Reaction conditions: P CH4= 0:2 bar,
P tot = 1:0 bar; CH 4 =CO 2 = 1; T = 750◦C (b) Alteration of con-version of CH 4 and CO 2 and selectivity to CO and H 2 , ob-tained over a 17 wt% Ni=La2O 3 catalyst, as a function of time
on stream Reaction conditions: P CH 4 = 0:5 bar, P tot = 1:0 bar,
CH 4 =CO 2 = 1; T = 750◦C.
Since the most prominent peak of Ni is well resolved from those of La2O3, it allows to estimate properly the Ni par-ticle size using the XLBA method (X-ray line broadening analysis) By employing Scherrer’s equation, it is estimated that the average Ni particle size present on La2O3 support
is of the order of 330 QA
The major crystalline phase of the working Ni=La2O3
catalyst was also studied by XRD The catalyst which had
Trang 8Table 3
Various parameters of Ni=-Al 2 O 3 and Ni=La2O 3 catalysts
1100 d
3240 e
a Crystalline phase was determined by XRD measurements.
b Since no Ni crystalline phase was detected by XRD in the Ni=-Al 2 O 3 catalyst, the Ni particle size is not estimated due to uncertainty
in the shape of the Ni particles.
c The Ni particle size was derived from XRD results.
d The Ni particle size was derived from the uptake of H 2 chemisorption assuming that H=Nisurface= 1.
e The Ni particle size was derived from the uptake of CO chemisorption assuming that CO=Nisurface= 1.
been exposed to the reaction mixture at 750◦C was quickly
quenched to room temperature and transferred to the XRD
apparatus It was found that the catalyst experiences a
profound change in its bulk phase, following exposure to
the CH4=CO2 mixture at 750◦C While the Ni and La2O3
phases which existed in the fresh Ni=La2O3 catalyst
disap-pear, La2O2CO3 phases are formed, following more than
half hour of reaction time The formation of La2O2CO3
phase should be the result of the reaction between La2O3
and the CO2gaseous reactant However, the occurrence of
this reaction should be accompanied by a process which
brings about the disappearance of the Ni crystalline phase
4.2 H2 andCO chemisorption
The uptake of H2 at room temperature is used to
deter-mine the dispersion of nickel on the support, assuming that
each surface metal atom chemisorbs one hydrogen atom, i.e
H=Nisurface= 1 It is found that the H2uptake of Ni=-Al2O3
and Ni=La2O3 are rather low, only amounting to ca 0.99
and 0:33 cm3=g, respectively (Table3) These correspond to
Ni dispersion of ca 3.0% and 1.0%, respectively Since no
metallic Ni particles are observed by XRD in the Ni=-Al2O3
catalyst, the apparent low nickel dispersion on the high
sur-face area -Al2O3 carrier should be largely due to the
for-mation of NiAl2O4, which is not capable of chemisorbing
hydrogen at room temperature The relatively higher H2
up-take on the Ni=Al2O3, as compared to the Ni=La2O3, may be
due to high dispersion of the remaining metallic Ni particles
(most of nickel is in the form of NiAl2O4) which could not
be detected by XRD The unusually low nickel dispersion
on La2O3 appears, at least partially, to be due to formation
of large nickel particles on the relatively low surface area
(¡ 5 m2=g) carrier, as revealed by the XRD study (Table
3) However, as described above, the Ni particle size based
on the XRD results is of the order of 330 QA which is still
much smaller than the one (ca 1000–1100 QA) derived from
H2chemisorption (1.0% dispersion)
CO chemisorption at room temperature was studied
by measuring the CO responses upon passing 1.1% CO through the catalyst It was estimated that the CO uptake
on the Ni=La2O3 and Ni=-Al2O3 catalysts amount to ca 0.22 and 1:97 ml=gcat, respectively (Table 3) Assuming that each surface Ni atom chemisorbs one CO molecule, i.e CO=Nisurface= 1, the number of surface Ni atoms on the Ni=-Al2O3 derived from CO uptake amounts to ca
5:5 × 1019atoms=gcat, which is close to the value derived from the H2uptake (5:6 × 1019atom=gcator 0:99 cm3=gcat) Previous studies [41,42] have shown that a reliable estima-tion of Ni particle size on Al2O3could be obtained for cata-lysts containing more than 3 wt% metal For the case of the Ni=La2O3catalyst, the CO uptake only amounts to ca 10%
of the respective one on the Ni=Al2O3 catalyst The Ni par-ticle of the Ni=La2O3catalyst, derived from the CO uptake,
is about 3–10 times larger than that derived from XRD and
H2chemisorption (Table3) Apparently, CO chemisorption
on the Ni=La2O3catalyst is signiDcantly suppressed 4.3 Temperature-programmeddesorption experiments TPD proDles of H2 from the Ni=La2O3 and Ni=-Al2O3
catalysts were obtained following H2 adsorption at 25◦C and 400◦C The TPD proDles of H2 from Ni=La2O3 and Ni=-Al2O3 catalyst are shown in Figs.7(a) and (b), re-spectively Two desorption peaks at ca 120◦C and 280◦C are observed from the Ni=La2O3 catalyst which has ad-sorbed H2at 25◦C As adsorption temperature is raised from
25◦C to 400◦C, the quantity of desorbed H2 increases sig-niDcantly (Fig 7(a)), which might imply that H2 adsorp-tion on the Ni=La2O3catalyst is partly an activated process The major desorption peak from the Ni=La2O3 is shifted from ca 120◦C to 165◦C, and a new peak at ca 200–
220◦C appears, as the adsorption temperature is raised from
25◦C to 400◦C It seems that hydrogen originating from adsorption at higher temperature, tends to desorb at higher temperatures
Trang 925 125 225 325 425 525
H2
a b
600 ppm
H2
a b
800 ppm
(a)
(b)
Fig 7 (a) TPD proDles of H 2 obtained over a 17 wt% Ni=La2O 3
after adsorption (a) at 25◦C and (b) at 400◦C = 28◦C=min.
(b) TPD proDles of H 2 obtained over a 17 wt% Ni=-Al 2 O 3 after
adsorption (a) at 25◦C and (b) at 400◦C = 23◦C=min.
The H2-TPD proDle from the Ni=-Al2O3catalyst are very
di5erent from those from the Ni=La2O3catalyst (Fig.7(b))
The quantity of hydrogen desorbed from the Ni=-Al2O3is
found to be about 2.5–3 times that of the Ni=La2O3catalyst
At least Dve discernible peaks at ca 120◦C, 220◦C, 320◦C,
440◦C and 520◦C can be distinguished on the Ni=-Al2O3
catalyst after H2 chemisorption at 25◦C While the Drst
three peaks at 120◦C, 220◦C and 320◦C may correspond
to the respective three peaks on the Ni=La2O3 (Fig.7(a)),
the two peaks, at 440◦C and 520◦C, are absent from the Ni=La2O3catalyst These two peaks correspond to strongly bound H species, probably the hydride species or the hy-drogen species in the subsurface layers of the metal catalyst [43] The population of hydrogen species under these two peaks accounts for about 15–20% of all hydrogen species adsorbed While the major hydrogen species desorb at ca
120◦C from the Ni=La2O3, they remain on the Ni=-Al2O3
surface at temperatures higher than 200◦C
The general characteristics revealed by H2-TPD experi-ments are: (1) a larger amount of hydrogen is desorbed from the Ni catalysts which have been exposed to hydrogen at higher temperature It seems that H2 adsorption on the Ni catalysts is partly an activated process; (2) the H–Ni bond
on the Ni=-Al2O3 appears to be stronger than that on the Ni=La2O3, suggesting that there might exist a certain kind of interaction between Ni and La2O3which leads to weakening
of H–Ni bond; and (3) the quantity of hydrogen desorbed from the Ni=-Al2O3catalyst is about 2.5–3.0 times that of the Ni=La2O3 catalyst This is in harmony with the results obtained by isothermal H2chemisorption at 25◦C (Table3)
5 On the unique stabilitycharacteristics of the Ni=La2O3catalyst
One of the major problems encountered in the process
of reforming of methane with carbon dioxide to synthe-sis gas over Ni-based catalysts is rapid carbon deposi-tion, which leads to blocking of active sites and decrease
of activity However, in contrast to other nickel-based catalysts (e.g Ni=-Al2O3) which exhibit continuous deactivation with time on stream, essentially no deactiva-tion was observed over the Ni catalyst supported on La2O3 Moreover, the reaction rate over the Ni=La2O3 catalyst increases with increasing time on stream during the initial several hours of reaction This leads to the suggestion that the La2O3 support plays a key role, a5ecting the kinetic behaviour of the Ni=La2O3 catalyst
In the present Ni=La2O3 catalyst, Ni dispersion is very low Based on the results of XRD (Table3), the average
Ni particle size of a 17 wt% Ni=La2O3 catalyst is of the order of 330 QA Results of H2and CO chemisorption give a mean Ni particle size of ca 1100 and 3200 QA, respectively Although di5erent techniques may result in di5erent metal particle sizes, the signiDcant di5erence (3–10 times) cannot
be simply attributed to uncertainties of the techniques It could be argued that part of the Ni surface is not accessible
to H2 and CO adsorption, thus leading to artiDcially small
Ni dispersion The Ni surface could be covered by a species originating during the preparation of the catalyst or during the pretreatment of the catalyst, and could be related to the support material, La2O3
To explain the unique stability of the Ni=La2O3catalyst,
it is proposed that a portion of the Ni surface is decorated
by lanthanum species originating from the La2O3 support
Trang 10The lanthanum species which are decorating the Ni
crys-tallites may interact with metallic Ni to form a new type
of surface compound or synergetic sites at the interfacial
area which are active and stable towards the reaction of
CH4=CO2to synthesis gas The unusual suppression of CO
and H2chemisorption of large Ni particles on the Ni=La2O3
catalyst can thus be attributed to blocking of Ni sites by the
lanthanum species
The nature of the lanthanum species which are
decorat-ing the Ni crystallites is revealed by the XRD results which
show that while the Ni and La2O3phases, which existed in
the fresh Ni=La2O3catalyst, disappear, La2O2CO3phase is
formed, after more than half an hour of reaction time
Oxy-gen species from the La2O2CO3 at the interface with Ni
surface participate, to a signiDcant extent, in formation of
CO and CO2with interaction of CH4=O2mixture,
presum-ably via fast exchange between gaseous O2and the oxygen
species from La2O2CO3 It may be reasoned that under
reac-tion condireac-tions the La2O2CO3, which is formed by reaction
between La2O3 and CO2, also participates in formation of
product CO On the other hand, it is known that CH4 only
weakly adsorbed on La2O3 [44,45] while it easily cracks
on metallic Ni at high temperatures [44–47] Thus, it may
be proposed that under CH4=CO2 reaction conditions, CH4
mainly cracks on the Ni crystallites to form H2and surface
carbon species, while CO2preferably adsorbs on the La2O3
support or the lanthanum species which are decorating the
Ni crystallites in the form of La2O2CO3 At high
tempera-tures, the oxygen species of the La2O2CO3may participate
in reactions with the surface carbon species on the
neigh-bouring Ni sites, to form CO Due to the existence of such
synergetic sites which consist of Ni and La elements, the
carbon species formed on the Ni sites are favourably
re-moved by the oxygen species originating from La2O2CO3,
thus o5ering an active and stable performance
Based on the mechanism described above, it is easy to
interpret the observation that signiDcant amounts of carbon
are deposited on the Ni=La2O3 catalyst, presumably on the
Ni crystallites, while the catalyst does not exhibit any
sig-niDcant deactivation This can be attributed to the fact that
the catalytic reaction is occurring at the Ni–La2O3
interfa-cial area which is not signiDcantly a5ected by carbon
depo-sition on the surface of Ni crystallites (as long as no excess
carbon is accumulated, blocking totally the surface of the Ni
crystallites) The fact that the reaction rate is increased
dur-ing the initial hours of time on stream could be explained
by a slow process of establishment of the ‘equilibrium’
con-centration of the La2O2CO3as well as other surface carbon
species on the Ni crystallites
Thus the Ni=La2O3catalyst provides a new reaction
path-way occurring at the Ni=La2O3interface It is proposed that
while CH4cracks on Ni crystallites, CO2favourably adsorbs
on the La2O3 support, in the form of La2O2CO3 The
re-action between oxygen species, originating from the La2O3
support, and carbon species, formed upon cracking of CH4
on Ni crystallites, gives active and stable catalytic
perfor-mance for carbon dioxide reforming of methane to synthesis gas, in spite of signiDcant carbon deposition on the surface
of Ni crystallites
In order to test and verify the proposed model, TEM in-vestigation of the Ni=La2O3catalyst was performed [37]: (i) after reduction, (ii) after reaction under di5erential condi-tions, (iii) after reaction under integral conditions and (iv) after regeneration by calcination:
(i) On the reduced sample, rather large and faceted Ni particles were observed (50–100 nm), which give an aver-age Ni dispersion around 1%, in agreement with the volu-metric data Each Ni particle is decorated by a continuous layer of around 2 nm in thickness as can be seen on Fig.8 EDX analyses were carried out on the overlayer (a), on the core of the particle (b), and on the support (c), as reported
in Table4 The overlayer which decorates the Ni particles was un-ambiguously found to contain a signiDcant amount of lan-thanum atoms, with the relative amount of La increasing when decreasing the EDX spot size Indeed, the smallest
EDX spot size (5 × 5 nm) being at least 2 times larger than
the overlayer (2 nm in thickness), a part of the nickel par-ticle is necessarily included in the analysed area which ex-plains the large contribution of Ni in the analysis (Table4)
As seen in Fig.8A(a), lattice planes of the overlayer are visualized and the crystallographic parameters of this layer, though not easy to measure, are consistent with a lanthanum carbonate structure
(ii) After 5-min exposure on the reforming stream under di5erential conditions (conversion about 5%), the decora-tion of nickel particles is still observed, along with some carbon deposits: veils and hollow Dlaments, characteristic
of the forms of coke observed on nickel catalysts for CO2
reforming
(iii) After 20 h exposure to the reforming stream, either under di5erential or integral conditions, most of the Ni parti-cles and lanthana grains appear to be completely surrounded
by carbon, which hinders any precise analysis of bulk and surface particles composition However, the particles present the same average size as the ones of the freshly reduced sample, which discards any signiDcant sintering e5ect un-der reaction conditions It is also observed that some nickel particles have been extracted from the lanthana support by growing Dlaments, as depicted in Fig.8B The lanthanum element is no more detected by EDX on the border of the extracted particles
(iv) After regeneration by calcination at 750◦C, the cat-alyst presents a strongly sintered aspect with particles ag-glomerated together with an average size of around 300–
400 nm (Fig.8C) On this micrograph, EDX analysis re-veals zones of pure nickel (a) and zones of mixed lanthanum and nickel composition with two dominant La/Ni atomic ra-tios around 2 (b) and 1 (c) They probably correspond to the local formation of nickel lanthanum oxides LaNiO3and
La2NiO4as identiDed by XRD Zones of pure lanthana are observed on other area of similar aspect