8/16/2019 CO y Al2O3

1/21

Kinetic studies of the methanation of CO over a Ni/ γ -Al2 O3 catalystusing a batch reactor

Jin Yang Lim a , James McGregor b , Andrew J. Sederman a , John S. Dennis a ,n

a Department of Chemical Engineering and Biotechnology, University of Cambridge, Cambridge CB2 3RA, United Kingdomb Department of Chemical and Biological Engineering, The University of Shef eld, Sir Robert Had eld Building, Portobello Street, Shef eld S1 3JD, UnitedKingdom

H I G H L I G H T S

CO methanation investigated in a spinning-basket reactor operated in batch. Rate and selectivity determined for 12 wt% Ni/ γ Al2 O3 catalyst , 443 – 473 K to 16 bar. Marked increase in rate after depletion of CO. Selectivity for CH 4 markedly lower than in CO 2 methanation. DRIFTS showed carbonyl species on Ni clusters and formate groups on different Ni sites.

a r t i c l e i n f o

Article history:Received 28 October 2015Received in revised form10 January 2016Accepted 2 February 2016Available online 8 February 2016

Keywords:Methanation of CObatch reactor experimentsNi/gamma aluminaCatalystReaction kineticsModelling

a b s t r a c t

The methanation of CO was investigated in a gradientless, spinning-basket reactor at temperatures 443 –

473 K and pressures up to 16 bar. The reactor was operated in batch and the composition of its contentswas determined periodically. Temperature programmed studies and DRIFTS analysis were performed togain an understanding of the nature of the surface of the catalyst. In all experiments, the reaction initiallyproceeded with a constant rate period. This was followed by a marked increase in the rate of production

of CH4 after the depletion of CO, attributed to the hydrogenation of remaining carbonyl groups on thesurface as well as the hydrogenolysis of long-chained paraf ns in the reactor. The selectivity for CH 4 wasfound to be signi cantly lower than that observed in CO 2 methanation, consistent with the low H 2 to COratio on the surface of the catalyst. Temperature-programmed studies and DRIFTS studies of the spentcatalyst identi ed two main types of carbonaceous species on the surface of the catalyst, with the resultsbeing consistent with the presence of (i) carbonyl species on nickel clusters and (ii) formate groups onnickel sites which have a stronger interaction with the alumina support. The former were found to bereactive at the temperatures studied. Finally, the rate of methanation was found to be insensitive to H 2 O.This was attributed to the strong af nity of the nickel catalyst for CO, which saturates the surface of thecatalyst leaving little opportunity for the adsorption of H 2 O. Two models were derived assuming that therate-limiting steps were either (i) the adsorption of H 2 on the catalyst, or (ii) the reaction of gaseous H 2with adsorbed CO. The strong adsorption of CO on the surface of the catalyst, evident from variousexperimental observations, is consistent with both mechanisms.

& 2016 Elsevier Ltd. All rights reserved.

1. Introduction

The methanation of CO, viz. the production of CH 4 from thereaction between CO and H 2 , is important in satisfying theincreasing demand for synthetic natural gas (SNG):

COþ 3H 2 - CH4 þ H2 O: ðR1Þ

A number of transition metals have been shown to be active incatalysing Reaction (1) , including Ni, Ru, Rh, Pt and Co ( Vannice,1975 ; Khodakov et al., 2007 ; Tada and Kikuchi, 2015 ). Nickel-basedcatalysts remain promising candidates for methanation reactionsin industrial processes because of their high activity and low cost.The reaction of CO and H 2 over a nickel-based catalyst is lessselective towards methane than is CO 2 methanation ( Weatherbeeand Bartholomew, 1982 ). Side reactions include the water – gasshift reaction (R2), the Boudouard reaction (R3) and

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/ces

Chemical Engineering Science

http://dx.doi.org/10.1016/j.ces.2016.02.0010009-2509/ & 2016 Elsevier Ltd. All rights reserved.

n Corresponding author.

Chemical Engineering Science 146 (2016) 316 – 336

http://www.sciencedirect.com/science/journal/00092509http://www.elsevier.com/locate/ceshttp://dx.doi.org/10.1016/j.ces.2016.02.001http://dx.doi.org/10.1016/j.ces.2016.02.001http://dx.doi.org/10.1016/j.ces.2016.02.001http://dx.doi.org/10.1016/j.ces.2016.02.001http://dx.doi.org/10.1016/j.ces.2016.02.001http://crossmark.crossref.org/dialog/?doi=10.1016/j.ces.2016.02.001&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.ces.2016.02.001&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.ces.2016.02.001&domain=pdfhttp://dx.doi.org/10.1016/j.ces.2016.02.001http://dx.doi.org/10.1016/j.ces.2016.02.001http://dx.doi.org/10.1016/j.ces.2016.02.001http://www.elsevier.com/locate/ceshttp://www.sciencedirect.com/science/journal/00092509

8/16/2019 CO y Al2O3

2/21

polymerisation to higher hydrocarbons (R4):COð g Þþ H2 Oð g Þ⇌ CO2ð g Þþ H2ð g Þ ðR2Þ

2COð g Þ⇌ CðsÞþ CO2ð g Þ ðR3Þ

COþ 2H 2 - CH2ð Þ þ H2 O ðR4Þ

Two main mechanisms, illustrated in Fig. 1, have been proposedin the literature for the methanation of CO:

(i) The rst involves adsorbed CO dissociating to adsorbed carbonand oxygen, which are hydrogenated to CH 4 and H 2 O (e.g.Vanho and Harriott, 1980 ; Klose and Baerns, 1984 ; Sughrueand Barholomew, 1982 ; Tada and Kikuchi, 2015 ).

(ii) The second mechanism involves an oxygenated species, such

as a COH x complex. Fig. 1 (ii) shows an example of the reactionscheme proposed by Vannice (1975) , where an enol inter-mediate was proposed.

Araki and Ponec (1976) studied the rate of hydrogenation of surface carbon species by admitting a known volume of 13 CO to aclean Ni lm followed by the introduction of a reaction mixture of H2 and 12 CO2 . They found that 13 CH4 was produced immediatelyafter the mixture of H 2 and 12 CO2 was introduced whilst 12 CH4 wasproduced only after an induction period. The increase in theamount of 12 CH4 produced was accompanied by the formation of 12 CO2 ; however, no 13 CO2 was detected. This suggests that adsor-bed carbon does not recombine with oxygen to form CO or CO 2molecules. Also, CO dissociates into surface carbon and oxygen

from the observations that the Boudouard reaction is active on

nickel catalysts above 250 °

C (Tøttrup, 1976 ; Nielsen and Trimm,1977 ; Gardner and Bartholomew, 1981 ). While many agree thatmechanism (i) is a reasonable description, different kinetic rateexpressions have been proposed based on different assumptions of the rate-limiting steps and the species present ( Vanho and Har-riott, 1980 ; Klose and Baerns, 1984 ; Sughrue and Barholomew,1982 ; Xu and Froment, 1989 ), discussed later in this paper.

Vannice (1975) , Van Herwijnen et al. (1973) and Huang andRichardson (1978) proposed rate expressions consistent withmechanism (ii). Further evidence for the presence of an oxyge-nated intermediate can be found in infra-red (IR) studies. Sanchez-Escribano et al. (2007) studied the nature of the surface speciespresent on a supported 10 wt% Ni/Al 2 O3 catalyst, under operandoconditions in an IR cell. A mixture of CO and H 2 was passed over

the catalyst bed and the FT-IR spectra of the adsorbed species wereobtained at different reaction temperatures. Signi cant IR bandsfor formate and carbonate groups were observed, indicating thepresence of oxygenated species on the surface. They also foundthat methane was produced after a catalyst sample pre-treatedwith methanol was heated in H 2 , suggesting that oxygenatedspecies formed from the adsorption of methanol following a cer-tain reaction to form CH 4 . Similar IR bands were observed by Fujitaet al. (1993) , who noted that the amounts of linear carbonyl spe-cies were much larger in CO methanation compared to CO 2methanation and suggested that the presence of a considerableamount of carbonyl species on the surface inhibits the rate of COmethanation. Using density functional theory calculations (DFT)on different model surfaces of nickel, Andersson et al . (2008)

showed that the dissociation of CO via a COH complex has a

CO (g)

H2 (g)

Mechanism (ii) :

Catalyst surface

H C

H O HCO

Catalyst surface

HH

Formation of enol complex

H

Catalyst surface

H2O CH 2H

Catalyst surface

H2O CH 4H

Fast hydrogenation

H2O (g)

CH 4 (g)

Formation of CH 4 and H 2O, followed by desorption

CO (g)

H2 (g)

CH 4

Catalyst surface

HH H2OH2OH2O (g)

CH4 (g)

Mechanism (i) :

CH OH

Catalyst surface

HHC O

Catalyst surface

HH

Dissociative adsorption of H 2 and CO Hydrogenation of surface C and O

Formation of CH 4 and H 2O, followed by desorption

Fig. 1. Schematic diagram of reaction mechanisms for CO methanation (i) via the dissociative adsorption of CO, (ii) via an enol intermediate.

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 317

8/16/2019 CO y Al2O3

3/21

slightly lower activation energy compared with the direct dis-sociation of CO to adsorbed C and O. They also showed that attemperatures higher than 850 K, the rate-limiting step became thehydrogenation of surface carbon, consistent with the observationsby Sughrue and Barholomew (1982) .

Further investigation is necessary to improve understanding of CO methanation. The aim of this paper was to obtain a repre-sentative kinetic expression valid over a wide range of partial

pressures of CO and H 2 . A way of validating rate expressions is toexamine their applicability over a wide range of partial pressuresof reactants and products, conveniently achieved by conductingthe reaction in a batch reactor ( Lim et al., 2015 ). Here, we havestudied the kinetics of the methanation of CO 2 in a gradientless,spinning-basket reactor operating in batch.

2. Experimental

2.1. Materials, apparatus and method

The catalyst, reduced 12 wt% Ni/ γ -Al2 O3 , was synthesised asdescribed in detail by Lim et al . (2015) , who also gave details of the

apparatus and technique employed. Information on the char-acterisation of the catalyst is given in Supplementary informationSection 1 . The total available surface area and the pore size dis-tribution was measured by gas adsorption analysis and deter-mined using the BET and BJH models. Temperature programmedstudies were performed by off-gas analysis in a CATLAB micro-reactor and in a thermogravimetric analyser (TGA) in order todetermine the different oxidation states within the samples. TheTGA was also used to con rm the mass fraction of Ni in the fullyoxidised sample and determine the fraction of metallic Ni whichcould be oxidised at room temperature. Powder X-ray diffraction(XRD) was used to determine the different crystalline phaseswithin the solid samples. The dispersion of surface Ni on thesample was measured by pulse H 2 chemisorption experiments,

which were performed in the CATLAB apparatus.Reaction studies were undertaken in a Carberry, spinning

basket reactor to study the behaviour of the catalyst in the pre-sence of CO, H 2 O or both. Depending on the experiments, gas fromcylinders containing (i) 24 vol% CO, 4 vol% Ar, H 2 balance ( 7 2 vol%precision, BOC), (ii) pure H 2 (99.99 vol% purity, Air Liquide),(iii) CO2 (99.99 vol% purity, Air Liquide), (iv) CO (BOC), (v) CH 4(99.5 vol% purity, Air Liquid), (vi) N 2 (99.998 vol% purity, AirLiquide) and (v) Ar (99.998 vol% purity, Air Liquide) was used topressurise the reactor.

In a typical experiment, the basket in the reactor was rstloaded with a known amount of catalyst and packed with non-porous, inert glass beads (1.4 mm diam.), such that about 5.0 g of catalyst pellets were mixed with an equal mass of glass beads in

the basket. The reactor was sealed, evacuated using a vacuumpump and then heated to 250 ° C. The catalyst was then subjectedto a ow through the reactor of 100 ml/min (at room temperatureand pressure) of H 2 with stirring at 1.7 Hz for 12 h at 1 bar. Fol-lowing reduction in H 2 , the reactor was evacuated again and theinternal temperature of the reactor brought to the desired reactiontemperature. The rate of the reaction of interest was studied inbatch by bringing the reactor to a desired initial pressure andcomposition, using gas supplied from cylinders connected to thereactor. A protocol for introducing gas was developed, describedby Lim et al. (2015) , which resulted in the nal pressure of thereactor being consistently achieved with a precision of 7 0.1 bar. Astirrer speed of 9.2 Hz was always used. The entire process of bringing the reactor to the desired pressure and starting the stirrer

after the introduction of the gases typically took 10–

15 s.

After the reactive gases were introduced into the reactor, thechanges in the composition of its contents were measured overtime. This was performed by taking volumes of 4 7 0.2 ml (atatmospheric temperature and pressure) from the sampling lineleading from the reactor using a gas-tight sampling syringe. Priorto the removal of the sample by the syringe, the gaseous contentsof the sampling lines were evacuated by the vacuum pump. Gasfrom the reactor was then allowed into the sampling line. The gas

collected was evacuated once again before the actual sample wastaken. This procedure ensured that the composition of the sampleof gas obtained from the reactor was representative of the con-tents of the bulk phase of the reactor volume. Only about 6 – 10samples were taken for each experiment in order to minimise theerrors incurred from the removal of gaseous contents from thereactor. The composition of the sample was analysed using off-linegas chromatography (Agilent 7890 GC Extended Re nery GasAnalysis) by passing the sample in the syringe through the sam-pling loop in the gas-chromatograph. The sampling loop in the gaschromatograph was evacuated using a vacuum pump before thegaseous contents of the syringe were introduced.

The composition of the gas given by off-line gas chromato-graphy is only equal to the bulk phase of the reactor if all species in

the gas phase are above their dew point at room temperature andpressure. This was not the case for most reactions performedbecause water was involved as a product or a reactant. Since theanalysis by gas chromatography provided a water-free composi-tion of the gas, the partial pressures of different species in the gasphase of the reactor were determined by using argon as aninternal standard, such that

pi ¼ xi xAr

pAr ; 0 ð1Þ

where p i is the partial pressure of species i, xi is the mole fractionof species i in the syringe, xAr is the mole fraction of Ar in thesyringe and pAr ; 0 is the partial pressure of argon at the start of thereaction. In most experiments, gas cylinders (of different mixtures

of H2 , CO2 and CO) contained 4% Ar. Hence, pAr ; 0 could be easilydetermined by measuring the total pressure of the reactor andmultiplying with the known composition of the cylinder.

2.2. Parameters affecting the measurement of kinetics

2.2.1. Control experimentsThe collective catalytic activity of the supporting Al 2 O3 , the

interior surface of the reactor and the nickel oxides present in thecatalyst was investigated for the methanation of CO. Thus, 5.0 g of the support (3 mm dia. SA-62125 alumina spheres, Saint-Gobain)was mixed with 5.0 g of non-porous glass beads (1.4 mm diam.).The Carberry reactor was evacuated and then 7.2 bar of H 2 , 2.4 barCO and 0.4 bar Ar was admitted to the evacuated reactor. The

composition of the reactor was measured periodically by takingsmall samples of the gas from the reactor volume analysed byof ine gas chromatography as described above. At both 293 and463 K, no signi cant decreases in pCO and pH2 were observed,indicating that no reaction of CO and H 2 had occurred. Tinyincreases in the partial pressure of methane were observed whenthe experiment was performed at 463 K with only the support, i.e.the alumina spheres and when the experiment was performedwith the fully-oxidised catalyst Ni/Al 2 O3 . However, the maximumamount of methane observed was 1.5 10 4 bar after 6361 s,signi cantly smaller than that observed when a similar mass of active catalyst was used. It was therefore concluded that the oxi-dised form of the catalyst was inactive in the catalysis of COmethanation and that the support material used in the synthesis

of the Ni/Al 2 O3 catalyst, the interior surface of the reactor and the

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 318

8/16/2019 CO y Al2O3

4/21

nickel oxides present in the catalyst could collectively be taken as

inert compared to the reduced nickel catalyst.

2.2.2. Catalyst deactivationReplicate experiments were performed using the same batch

of catalyst for ve repeated runs. Fig. 2 shows the variation of thepartial pressures of CO, H 2 , CH4 and CO 2 as a function of time for

ve consecutive, replicate experiments at 463 K with pCO; 0 , theinitial partial pressure of CO, of 2.4 bar and pH2 ; 0 , the initialpartial pressure of H 2 , 7.2 bar. The pro les of the partial pres-sures of CO, H 2 , CH4 and CO 2 for different runs are almostidentical, illustrating that there was negligible deactivation for atleast ve experiments, corresponding to a total time on streamof 4 10 4 s. The insigni cant rate of deactivation and good

reproducibility of the replicate experiments meant that COmethanation experiments could be performed on the same batchof Ni/Al 2 O3 catalyst. It also shows that loss of nickel by the for-mation of nickel carbonyl, or by the migration of Ni by thiscompound, was not signi cant. However, the same batch of catalyst was used for no more than 10 experiments before beingreplaced by a fresh batch. CO 2 methanation, with an initialcondition of pCO2 ; 0 ¼ 2 : 4bar and pH2 ; 0 ¼ 7 : 2bar at 463 K, was per-formed as a control for the nal experiment on a given batch of catalyst. CO 2 methanation was used as the reference conditionbecause the experiments were of shorter duration than thoseusing CO. Also, Ni-based catalysts undergo less deactivationduring the methanation of CO 2 methanation compared to that of

CO (Chang et al, .1997 ; Zhang et al., 2013 ).

2.2.3. Effect of heat and mass transfer external to the catalyst

particleThe effect of heat and mass transfer from the bulk gaseous

phase of the reactor to the external surface of the catalyst pelletwas studied by performing CO methanation reactions, with aninitial condition of pCO ¼ 2 : 4bar and pH2 ¼ 7 : 2bar at 463 K, at dif-ferent impeller speeds. At the extremes, the initial rate of pro-duction of CH 4 when the impeller was stationary was about 14%faster than at 9.3 Hz. There was very little difference in thebehaviour of the reaction for spinning speeds higher than 7.1 Hz.

The methanation of CO is exothermic with Δ H 3298K ¼ 206kJmol 1 so that the decrease in the observed rate of reaction couldhave been attributable to the enhanced dissipation of heat withincreased impeller speeds. To investigate whether heat transfer issigni cant during typical experimental conditions, the difference

between the bulk temperature and the surface of the particle, Δ T p,was estimated by a pseudo-steady heat balance on the catalystpellet:

r 0Δ H 3T 1 ρcat4π R p3

3 !¼ h 4π R p2 Δ T p ; ð2Þwhere r 0 is the speci c rate of reaction, Δ H 3T 1 is the heat of reac-tion, T 1 is the temperature of the bulk phase, ρcat is the density of the catalyst, R0 is the radius of the pellet and h is the heat transfercoef cient. The initial rate of consumption of CO at the referenceinitial condition pCO; 0 ¼ 2 : 4 bar, pH2 ; 0 ¼ 2 : 4 bar and T ¼ 463 K wasdetermined experimentally to be 9.9 10 4 mol kg 1 s 1 . Valuesof the transport properties for the gas were based on a mixture of 75 vol% H2 and 25 vol% CO at 463 K at 10 bar. For an isolated,

spherical catalyst, diameter 3 mm, and an assumed super cial

Fig. 2. Partial pressure of (a) CO, (b) CH 4 , (c) H 2 and (d) CO 2 with time for 5 consecutive runs in batch for the same catalyst: p CO; 0 ¼ 2 : 4bar, pH2 ; 0 ¼ 7 : 2bar, T ¼ 463 K, 5 gcatalyst.

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 319

8/16/2019 CO y Al2O3

5/21

velocity of 0.9 m s 1 , Nu ¼ 30 and the heat transfer coef cient was1080 W m 2 K 1 . Using Eq. (2) , Δ T p, 0.4 K. However, for a

stationary particle, Δ T p was 1.3 K, suggesting a very slight effect

from heating effects. However, the empirical evidence showedthere to be negligible differences in the response of the reaction inthe batch reactor for impeller speeds 4 7.1 Hz, so all experimentswere performed using an impeller speed of 9.5 Hz. Thus, it couldbe assumed that there were negligible effects of heat and masstransfer from the bulk volume of the reactor to the surface of thecatalyst.

2.2.4. Effect of intra-particle heat and mass transfer Using a rate of consumption of CO during methanation at 463 K

of 9.9 10 4 mol kg 1 s 1 and an average pore diameter of 8.9 nm, the Weisz-Prater number ( Weisz and Prater, 1954 )

N WP ¼r 0 ρcat RP

2

C sDef f ; ð3Þ

was 0.008, much smaller than the estimated threshold of 0.3 atwhich mass transfer limitations are expected to become sig-ni cant. Here, C s was the concentration of CO in mol m 3 . Giventhe small pore diameter, the effective diffusivity, Deff , was taken tobe the product of the Knudsen diffusivity of CO and ( ε /τ 2 ), withε ¼ 0.60 and τ 2 assumed to be 3. Here, ε was determined from thecumulative pore volume of the Al 2 O3 support, accounting for poresranging from 17 to 300 nm in diameter, of 0.55 cm 3 g 1 . The group(ε /τ 2 ) is appropriate for use with the model of Young and Todd(2005) to model diffusion within the particle of catalyst.

To estimate the temperature difference between the centre andthe surface of the pellet, the model of reaction and diffusion in a

catalyst pellet of Lim and Dennis (2012) was solved for the extremecase of constant kinetics of 9.9 10 4 mol CO kg 1 s 1, with condi-tions at the surface of the pellet taken to be pCO ¼ 2 : 4 bar and pH2 ¼ 7 : 2 bar. The effective thermal conductivity of the pellet wastaken as the thermal conductivity of Al 2 O3 , 9 W m 1 K 1 . At463 K, the estimated temperature difference between the centreand the surface of the catalyst pellet was only about 0.1 K for apellet with a diameter of 3 mm, indicating that intraparticletemperature gradients are negligible. This conclusion was sub-stantiated experimentally by measuring the change in partialpressure of H 2 and CH 4 as a function of time for two differentparticle sizes at the reference initial condition of pCO2 ; 0 ¼ 2 : 4 barand pH2 ; 0 ¼ 7 : 2 bar at 463 K. The behaviour of the 3 mm dia. pelletwas equal to that of the 1.7 – 2.3 mm pellets, suggesting that intra-

particle mass transfer limitations were small.

2.2.5. Effect of total pressureThe effect of total pressure was investigated by comparing the

experimental results methanation for different initial partialpressures of the inert gas N 2 , rst admitted to the evacuated vesselbefore the CO and H 2 were introduced at 463 K. There was noobservable effect of total pressure on the CO methanation reactionfor partial pressures of N 2 up to 6 bar.

3. Results

3.1. General features

This section reports the transient pro les of the partial pressuresof various gaseous species in the batch reactor at the reference initialcondition, i.e. pCO; 0 ¼ 2 : 4 bar, pH2 ; 0 ¼ 7 : 2 bar at T ¼ 463 K, which hasbeen illustrated in Fig. 2. Fig. 2(a) and (c) shows that for t o 4000 s,the decreases in pCO and pH2 were approximately linear with time.The corresponding increase in pCH4 was also found to be largelylinear, as seen in Fig. 2(b). At t E 4500 s, pCO was depleted. After thistime, a marked increase in the rate of formation of methane,accompanied by a decrease in the rate of reaction of H 2 , wasobserved. The variation of pCO2 with time, Fig. 2(d), featured an initialrapid increase in pCO2 with time for t o 4000 s. However, an extre-mely sharp decrease in pCO2 was observed at approximately 4500 s,coinciding with the depletion of CO in the gas phase.

The origin of the CO 2 is attributable to the water – gas shift reac-tion. Other results, Lim et al. (2016) , indicated that the rate of thewater – gas shift reaction, in an atmosphere with high pCO, was foundto increase with pH2 O and, with the exception of very low pCO,viz. o 0.1 bar, pCO had no effect on the on the rate. Fig. 2(d) shows thatfor an initial condition of pCO; 0 ¼ 2 : 4 bar and pH2 ; 0 ¼ 7 : 2 bar, the rateof production of CO 2 increased with time. This is the result of theproduction from the CO methanation of H 2 O, which accumulatesbecause of the batch operation. As pH2 O increased with time, the rateof formation of CO 2 increased correspondingly. This provides evi-dence that the water – gas shift reaction was responsible for the for-mation of CO 2 in the reactor and not the Boudouard reaction, whereexperiments ( Lim et al., 2016 ) gave a rate of production of CO 2decreasing over time, contrary to Fig. 2(d).

The carbon balance for the batch reaction is illustrated in Fig. 3(a).Here, the total amount of carbon in the gas phase in equivalent partialpressure, pC, was calculated as

pC ¼ pCOþ pCO2 þ X6

i ¼ 1N i pHC; i ð4Þ

where pHC; i is the partial pressure of hydrocarbon i, and N i is thecarbon number of hydrocarbon i. It should be noted that the sum-mation includes all paraf ns and ole ns detected by the gas chro-matograph. The variation of pC with time, plotted in Fig. 3, representsthe carbon balance of the CO methanation reaction in the batchreactor. It is clear that as the reaction proceeded, pC decreased almostlinearly with time, suggesting the formation of carbon species whichcould not be detected in the gas phase. When pCO¼ 0, at t E 4500 s, pCwas observed to recover from about 1.8 bar to 2.1 bar. The overallcarbon balance at the end of each experiment was approximately 87%of the original inventory, i.e. 2.4 bar of CO. The increase of pC is mainlycaused the marked increase in pCH4 after the depletion of pCO, as seenin Fig. 2(b). This is further supported by Fig. 4(a) and (b), where nosigni cant increase in the partial pressures of C 2 H6 and C 3H8 werefound after t E 4500 s. Since the carbon balance of the reaction couldnot be fully addressed by all the species in the gas phase, the selec-

tivity of the CO methanation reaction is best expressed as a fractional

Fig. 3. The variation of p C with time for ve replicate experiments. In all experi-ments, pCO; 0 ¼ 2 : 4 bar, pH2 ; 0 ¼ 7 : 2 bar, T ¼ 463 K and m cat ¼ 5.0 g.

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 320

8/16/2019 CO y Al2O3

6/21

conversion of CO to species i, F i ; such that

F i ¼ pHC; i

pCO; 0 pCO þ pCO2 ð5Þ

where pCO; 0 is the initial partial pressure of CO. Of course, the valuesof F i, pHC; i, pCO and pCO2 varied with time as the reaction proceededand the variation of the F i for different hydrocarbons is illustrated inFig. 5(a). In this paper, the term ‘selectivity ’ refers to the qualitativedescription of the product distribution while ‘ fractional conversion ’ isa quantitative value as de ned by Eq. (5). It is noted that before

t E 4500 s, there was a small increase in the fractional conversiontowards CH 4 , F CH4 , from 0.23 to 0.27. However, a large increase inF CH4 was observed after t E 4500 s, as expected from the earlierobservations. The fractional conversions for other paraf ns remainedlargely constant over the course of the experiment. Fig. 5(b) showsthe fractional conversion of different paraf ns at the end of eachexperiment for the ve replicate experiments. The distribution of theproducts appeared to decrease markedly with the carbon number of the paraf n. It should be noted that only very small quantities of ole ns were detected (not shown), with partial pressures severalorders of magnitudes smaller than the paraf n with the same carbonnumber.

The transition from the constant rate period before pCO ¼ 0 tothe period after this time is a key feature in many of the experi-

mental results illustrated in the following Sections. For

convenience, the period before P CO ¼ 0 will be called period I andtimes thereafter period II.

3.2. Effect of pH2

The effect of pH2 on the rate of CO methanation was studied byrst raising the pressure of the evacuated reactor to a desired pres-

sure with pure H 2 before a mixture of 24 vol% CO, 72 vol% H 2 and4 vol% Ar was introduced. Fig. 6 illustrates changes in the partialpressures of CO, H 2 , CH4 and CO 2 over time for different initial partialpressures of H 2 . Fig. 6(a) shows that increasing the partial pressure of H2 increased the rate of consumption of CO. In fact, the rate of con-sumption of CO increased from 9.4 10 4 mol CO kg 1 s 1 with pH2 ; 0 ¼ 7.2 bar to 1.34 10

3 mol CO kg 1 s 1 when pH2 ; 0 ¼ 13.2 bar,an increase of 30%. Since the rate of consumption of CO was faster,the transition from period I to II also occurred earlier for hig-her partial pressures of H 2 , re ected in the pro les of CO and CH 4 inFig. 6(a) and (c). The maximum in pCO2 with time also occurredearlier, consistent with the observation that the consumption of CO 2occurred only when pCO had fallen to zero.

The rate of production of CH 4 was signi cantly greater than theincrease in the rate of consumption of CO with higher partial pres-sure of H 2 . The initial rate of production of CH 4 when P H 2 ; 0 ¼ 7.2 barwas 2.3 10 4 mol CH4 kg 1 s 1 increasing by a factor of 2.3 to

5.2 104

mol CH4 kg1

s1

when P H 2 ; 0 ¼ 13.2 bar, signi cantly

Fig. 4. The partial pressure of (a) C 2 H6 and (b) C 3 H8 with time for 5 consecutive runs in batch for the same catalyst. In all experiments, p CO; 0 ¼ 2 : 4 bar, p H2 ; 0 ¼ 7 : 2 bar,T ¼ 463 K and m cat ¼ 5.0 g.

Fig. 5. (a) The variation of the fractional conversion of different paraf ns with time for Run 1. (b) The fraction conversion of paraf ns of different carbon number at the end of each experiment. In all experiments, pCO; 0 ¼ 2 : 4 bar, pH2 ; 0 ¼ 7 : 2 bar, T ¼ 463 K and m cat ¼ 5.0 g.

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 321

8/16/2019 CO y Al2O3

7/21

higher than the increase in the rate of consumption of CO as noted

earlier. The marked increase in the rate of production of CH 4 is theresult of (i) the inherent increase in the rate of reaction and (ii) theincrease in the selectivity towards methane, both a result of thehigher ratio of partial pressure of H 2 to CO.

The effect of increasing pH2 on the selectivity towards CH 4 can beobserved in Fig. 7(a), where the fractional conversion increased sig-ni cantly with pH2 ; 0 . The increase in selectivity towards CH 4 wasprevalent throughout the length of the experiment, viz. for higher pH2 ; 0 , the fractional conversion for CH 4 was higher in both period Iand II. Fig. 7(b) – (d) illustrate the fractional conversions of ethane,propane and n-butane over time for different pH2 ; 0 . There was nosigni cant change in the selectivity towards the higher hydrocarbonsfor the range of pH2 explored, with only a small decrease in theselectivity of propane and n-butane for higher pH2 ; 0 . Unlike the case

with CH 4 , the fractional conversions of higher hydrocarbonsremained largely constant for the duration of the reaction and nosigni cant changes were observed after the depletion of pCO. Theoverall carbon balance, represented by pC, appeared to be closer tothe initial value of pC ¼ 2.4 bar with higher partial pressures of H 2 asillustrated in Fig. 8 and consistent with the observation that moreCH4 was produced. This also suggests that with higher pH2 , there wasa smaller rate of formation of liquid hydrocarbons or surface carbo-naceous species which led to more gaseous hydrocarbons.

3.3. Effect of pCO

Here, the pressure of the reactor was raised to a desired pres-sure with pure CO before the introduction of the mixture of CO

and H 2 in order to study the effects of variations in pCO. Fig. 9

shows that the rate of reaction, calculated from the rate of con-

sumption of pH2 or the rate of production of pCH4 , decreased as theinitial partial pressure of CO, pCO; 0 , increased. The initial rate of production of CH 4 decreased from 2.3 10 4 mol CH4 kg 1 s 1 at pCO; 0 ¼ 2.4 bar to 1.6 10

4 mol CH4 kg 1 s 1 when pCO; 0 ¼ 5.8 bar.When pCO; 0 Z 4.0 bar, the stoichiometry of the reaction resulted inthe limiting species changing from CO to H 2 . No transitions fromperiod I to II were observed in experiments when pCO; 0 Z 4 : 0 bar,because in those experiments pCO was never depleted andremained in excess in the reactor after the depletion of pH2 . Noabrupt changes in the rate of increase of pCH4 were observed inFig. 9(c) when CO was in excess. It was also noted that, for pCO; 0 Z 4 : 0 bar, pCO2 continued to increase for the duration of thereaction, consistent with the observation that the consumption of CO2 only occurs after CO has been depleted.

Fig. 10 illustrates the selectivity for CH 4 , C2 H6 , C3H8 and C 4 H10 fordifferent initial partial pressures of CO at 463 K. Fig.10 (a) shows thatfor pCO; 0 ¼ 4.0 and 5.8 bar, the selectivity of CH 4 remained largelyconstant at about 0.22 7 0.01 over the course of the reaction. Thiswas signi cantly lower than the fractional conversion of CH 4 when pCO; 0 ¼ 2.4 bar. The selectivities for the higher hydrocarbons were alsofound to be smaller when pCO; 0 was increased. However, in all cases,the product distribution of the reaction when pCO; 0 ¼ 4.0 bar wasidentical to that when pCO; 0 ¼ 5.8 bar. As expected, the total amountof carbon in the gas phase, pC, decreased monotonically over time for pCO; 0 ¼ 4.0 and 5.8 bar (results given in Supplementary information ,Section 2 ), suggesting a build up of liquid hydrocarbons in the reactoror carbonaceous species on the catalyst. No recovery of pC was

observed when CO was in stoichiometric excess.

Fig. 6. The variation of the partial pressure of (a) H 2 , (b) CO, (c) CH 4 and (d) CO 2 over time for different initial partial pressures of H 2 . pCO,0 ¼ 2.4 bar, T ¼ 463 K and mcat ¼ 5.0 g.

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 322

8/16/2019 CO y Al2O3

8/21

3.4. Effect of pCH4 and p CO2

The effect of CH 4 on the rate and selectivity was explored bymeasuring the variation of partial pressure of H 2 , CO, CH4 and CO 2over time for different initial partial pressures of CH 4 , from 0 to6 bar, at 463 K. The results, described in Supplementary informa-tion Section 3 , indicated there was no signi cant effect of pCH4 on

the rate of reaction and also that the product distribution was not

affected by the introduction of additional pCH4 . Thus, CH 4 behaved

as a spectator molecule.To study the effect of CO 2 , the reactor was rst raised to the

desired pressure with pure CO 2 before the mixture of CO and H 2 wasintroduced. The initial rates of reactions were identical for all differ-ent initial partial pressures of CO 2, pCO2 ; 0 considered (0, 3 and 6 bar)as shown by the overlapping points in Fig. 11 (a) – (c) for t o 4000 s. Inall experiments, the transition from period I to II occurred atapproximately the same time, t ¼ 5000 s. However, it is noted that pCH4 differed after the transition from period I to II, i.e. at t 4 5000 s.At t ¼ 8000 s, pCH4 increased from 1.1 bar when pCO2 ; 0 ¼ 0 to 1.2 barfor pCO2 ; 0 ¼ 3 bar. These observations suggest that H 2 reacted with COduring period I, and that following the depletion of pCO, theremaining H 2 reacted with CO 2 , resulting in the overall increase in pCH4 at the end of the experiment. This is also consistent with the

decrease of pCO2 at t 4 5000 s, as shown in Fig. 11(d). The productdistribution of the reaction was also not affected by the presence of CO2 (results given in Supplementary information Section 3 ). Themarked increase in pCH4 during period II is therefore attributed to theproduction of additional CH 4 from CO 2 methanation.

3.5. Effect of pH 2O

The effect of pH2 O was studied by rst performing CO 2 metha-nation from a known initial pressure of CO 2 and H 2 in order toform CH 4 and H 2 O, followed by the introduction of 2.4 bar of COand 7.2 bar of H 2 . According to the stoichiometry of CO 2 metha-nation, the mixture of CO 2 and H 2 yielded, at the end of thereaction, a mixture of mainly CH 4 and H 2 O with some excess CO 2 .

It is important to note that no signi cant amounts of H 2 remained

Fig. 7. Fractional conversion of (a) CH 4 , (b) C2 H6 , (c) C3 H8 and (d) C 4 H10 function of the conversion of CO, X CO , for different initial partial pressures of H 2 . pH2 ; 0 ¼ 7 : 2 bar,T ¼ 463 K and m cat ¼ 5.0 g.

Fig. 8. Variation of equivalent partial pressure of total carbon in the gas phase, pC,versus time for different initial partial pressures of H 2 . pCO; 0 ¼ 2.4 bar, T ¼ 463 K andm cat ¼ 5.0 g.

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 323

8/16/2019 CO y Al2O3

9/21

at the end of the CO 2 methanation reaction, before the introduc-tion of CO and H 2 . This is because pH2 has been shown, above, tohave a signi cant in uence on the rate and selectivity of metha-nation. The time, t , at which the mixture of CO and H 2 wasintroduced into the reactor was taken as t ¼ 0 s. Since it has alreadybeen established that pCO2 and pCH4 have no effect on the rate andthe selectivity of the reaction in the presence of CO, any effectsobserved could be attributed to the presence of H 2 O. The effect of pH2 O was explored for temperatures 453

– 473 K. For the rangeexplored up to pH2 O ¼ 1.44 bar, there were no effects on the rate of reaction and also the addition of pH2 O had very little effect on theproduct distribution of the reaction (results shown in Supple-mentary information Section 4 ).

3.6. Effect of temperature

The effect of temperatures from 443 to 473 K on the rate of COmethanation reaction was studied using the reference initialcondition of pCO; 0 ¼ 2 : 4 bar and pH2 ; 0 ¼ 7 : 2 bar. The variations of the partial pressures of H 2 and CO with time are illustrated inFig. 12 . In general, the rate of reaction doubled for every increase intemperature of 10 K. This can be observed from the pro le of pH2and pCO in Fig. 12(a) and (b), where the time taken for pCO to bedepleted decreased by approximately 50% for every increase of 10 K. The rate of production of CH 4 also followed a similar trend.The transition from period I to II, noted by a marked increase in pCH4 , was found to occur at all temperatures explored, as illu-

strated in Fig. 13(a). This was also accompanied by the initial

accumulation of CO 2 , followed by a fast consumption after thetransition to period II when pCO ¼ 0, as shown in Fig. 13(b).

It is interesting to note that the selectivity of the reaction wasnot affected by temperature. Results, shown in the Supplementaryinformation Section 5 , indicated that the fractional conversion toCH4 remained approximately constant at 0.26 within period I andincreased to about 0.44 after the transition to period II, i.e. afterthe depletion of CO. The fractional conversions to other hydro-carbons were also found to be unchanged for the temperaturerange explored. Therefore, it can be concluded that the selectivityof the reaction was not a strong function of temperature, fortemperatures 443 – 473 K, and that temperature only has an effecton the overall rate of reaction.

3.7. Temperature-programmed studies

Following CO methanation in the Carberry reactor, the spentcatalyst was kept in a sealed glass jar after its removal from thereactor, after the nal reaction – the methanation of CO 2 – hadbeen conducted. During the transfer from the reactor, the nickelon the catalyst would have been partially oxidised by atmo-spheric oxygen and some of the weakly adsorbed species wouldhave left the surface of the catalyst. The spent catalyst wassubsequently subjected to temperature-programmed studiesusing the CATLAB apparatus. In an experiment, 50 mg of spentcatalyst was placed in the quartz, tubular reactor and thensubjected to a temperature programme whilst gases of various,

xed composition were passed through the packed bed of cata-

lyst in the reactor. In all studies, the catalyst was subjected to a

Fig. 9. The variation of the partial pressure of (a) H 2 , (b) CO, (c) CH 4 and (d) CO 2 over time for different initial partial pressures of CO. pH2 ; 0 ¼ 7.2 bar, T ¼ 463 K and mcat ¼ 5.0 g.

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 324

8/16/2019 CO y Al2O3

10/21

owrate of the chosen gas mixture of 40 ml/min (at room tem-perature and pressure) at 120 ° C for 1 h before the temperaturewas increased at a linear rate of 10 ° C/min. Different gases wereused during the temperature ramp: Fig. 14 shows the tempera-ture programmed desorption (TPD) with a ow of 40 ml/min (atroom temperature and pressure) of He, Fig. 15 the temperature-programmed hydrogenation (TPH) in a ow of 40 ml/min (atroom temperature and pressure) of a mixture of 5 vol% H 2 in Heand Fig. 16 the temperature programmed oxidation (TPO) with a

ow of 5 ml/min (at room temperature and pressure) of O 2 and35 ml/min (at room temperature and pressure) of He.

The TPD results show the evolution of H 2 , CO2 , CO and CH 4 intwo main regions; a sharp, narrow peak at 200 ° C and a broad

peak ranging from 300 to 500 °

C. The evolution of water, illu-strated in Fig. 14 (b) shows two broad peaks, probably the resultof some moisture on the catalyst as well as the possible reduc-tion of nickel oxide by adsorbed H 2 or hydrocarbons. The evo-lution of carbon dioxide shows some interaction between theadsorbed carbon species with nickel oxide. It also demonstratesthe presence of carbonaceous species on the surface of the cat-alyst during CO methanation. The TPH shows the evolution of CO2 , CO and CH 4 at about 200 ° C, illustrated in Fig. 15 . Broadpeaks for CO 2 and CH 4 were also observed at a higher tem-perature of 280 – 500 ° C. However, no evolution of CO wasobserved within this temperature range. Compared to the TPD,the broad peaks for CO 2 and CH 4 at the higher temperaturesappeared to start at a lower temperature in the hydrogenation

experiment. This suggests that the carbonaceous material on the

surface of the catalyst was more easily removed by reacting withH2 than in an atmosphere of helium. It is also noted that thepro le of H 2 O evolution shown in Fig. 15(b) is similar to that of the temperature-programmed reduction of the passivated cata-lyst, which is unsurprising given that some of the catalyst wouldhave been partially oxidised after its removal form the Carberryreactor.

In the TPO in Fig. 16, CO2 and CO were detected over a widerange of temperatures, ranging from 120 to 500 ° C. No signi cantCH4 was observed over the duration of the experiment, as seen byits complete absence in both Fig. 16(a) and (b), suggesting thatmost adsorbed carbonaceous species were oxidised directly toCO2 . Since the pro les of CO 2 and CO followed the same pattern, itis likely that the pro le of CO is the result of the mass spectro-meter detecting the CO fragment which originated from CO 2 . Thepro le of H 2 O also followed a similar pattern, suggesting thecombustion of hydrocarbons to form CO 2 and H 2 O. Thesetemperature-programmed studies show the presence of two mainforms of carbon on the surface of the catalyst following COmethanation: one which could be reduced and hydrogenated at200 – 300 ° C and another in the region of 300 – 500 ° C.

The import of these observations is discussed in Section 5 ,below, after the modelling has been introduced and discussed.

3.8. DRIFTS measurements

The surface of the catalyst was probed using in-situ DRIFTS inwhich 50 mg of fresh, passivated 12 wt% Ni/ γ -Al 2 O3 catalyst was

packed as a differential bed in the reactor chamber and

Fig. 10. Fractional conversion of (a) CH 4 , (b) C2 H6 , (c) C3 H8 and (d) C 4 H10 vs. partial pressure of CO consumed, for different initial partial pressures of CO. p H2 ; 0 ¼ 7.2 bar,T ¼ 463 K, m cat ¼ 5.0 g.

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 325

8/16/2019 CO y Al2O3

11/21

supported by a wire mesh. The catalyst was reduced at 450 ° C for2 h under 100 ml/min (at room temperature and pressure) of H 2 .Following reduction, a mixture of 90 ml/min (at room tempera-ture and pressure) of H 2 and 30 ml/min (at room temperatureand pressure) of CO was introduced into the reaction chamberand passed through the differential bed. In-situ IR spectra were

obtained at temperatures from 463–

723 K. Fig. 17 illustrates the

main features of the IR spectrum obtained at 463 K at steady-state during CO methanation at atmospheric pressure. It is notedthat the peaks at 2180 and 2120 cm 1 in Fig. 17(a) are the result of the CO in the gas phase, which was con rmed with the reference IR spectrum of carbon monoxide. The presence of fringes around thisregion was attributed to the rotation of gas phase CO molecules. The

peak at 2070 cm1

is the result of linear carbonyl groups on the

Fig.11. Variation of partial pressure of (a) H 2 , (b) CO, (c) CH 4 and (d) CO 2 over time for different initial partial pressures of CO 2 . pCO; 0 ¼ 2 : 4 bar, pH2 ; 0 ¼ 7 : 2 bar, T ¼ 463 K andm cat ¼ 5.0 g.

Fig. 12. The partial pressure of (a) H 2 and (b) CO with time for different reaction temperatures. In all experiments, pCO; 0 ¼ 2 : 4 bar, pH2 ; 0 ¼ 7 : 2 bar and m cat ¼ 5.0 g.

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 326

8/16/2019 CO y Al2O3

12/21

surface of the catalyst. The large peaks at 1600, 1390, 1380 and1330 cm 1 in Fig. 17(a) can be attributed to formate groups. Thepeak at 2900 cm 1 was also attributed to the presence of formate

groups, as shown in Fig. 17(b).

Following CO methanation at 463 K, the stream of CO was turnedoff and only H 2 was passed through the catalyst bed. Spectrum (ii) inFig. 17 was obtained after 40 min in a ow of H 2 of 90 ml/min (at

room temperature and pressure). The amount of carbonyl species,

Fig. 13. The partial pressure of (a) CH 4 and (b) CO 2 with time for different reaction temperatures. In all experiments, pCO; 0 ¼ 2 : 4 bar, pH2 ; 0 ¼ 7 : 2 bar and m cat ¼ 5.0 g.

Fig. 14. Temperature-programmed desorption using CATLAB on a spent Ni/ γ -Al2 O3 catalyst after CO methanation in the Carberry reactor. A temperature ramp of 10 ° C/minwas used. Signals of (a) hydrogen, carbon dioxide and carbon monoxide and (b) methane and water were plotted against temperature. The ow rate of He was 40 ml/min (atroom temperature and pressure).

Fig. 15. Temperature-programmed hydrogenation using CATLAB on a spent Ni/ γ -Al2 O3 catalyst after CO methanation in the Carberry reactor. A temperature ramp of 10 ° C/min wasused. Signals of (a) hydrogen, carbon dioxide and carbon monoxide and (b) methane and water were plotted against temperature. A ow rate of 40 ml/min (at room temperature andpressure) of a gas mixture of 5 vol% H 2 in He was passed through the catalyst bed.

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 327

8/16/2019 CO y Al2O3

13/21

Fig. 16. Temperature-programmed oxidation using CATLAB on a spent Ni/ γ -Al2 O3 catalyst after CO methanation in the Carberry reactor. A temperature ramp of 10 ° C/minwas used. Signals of (a) hydrogen, carbon dioxide and carbon monoxide and (b) methane and water were plotted against temperature. A ow rate of 5 ml/min (at roomtemperature and pressure) of O 2 and 35 ml/min of He (at room temperature and pressure) was passed through the catalyst bed. N.B. methane was below the limit of detection in (b).

Fig.17. Infra-red spectra of adsorbed species in the range of (a) 1200 – 2400 cm 1 and (b) 2600 – 3200 cm 1 formed on reduced 12 wt% Ni/Al 2 O3 in a ow of (i) 90 ml/min (atroom temperature and pressure) of H 2 and 30 ml/min (at room temperature and pressure) of CO and (ii) in 100 ml/min (at room temperature and pressure) of H 2 only, takenafter 40 min following the introduction of H 2 . The DRIFTS spectrum of the catalyst under He at 463 K was used as the background.

0 1000 2000 3000 40000

2

4

6

8

10

12

14

Time / s

P a r t

i a l p r e s s u r e

/ b a r

7.2 bar H2

9.2 bar H2

10.5 bar H2

12.5 bar H2

0 1000 2000 3000 4000 50000

0.2

0.4

0.6

0.8

1

Time / s

P a r t

i a l p r e s s u r e

/ b a r

7.2 bar H2

9.2 bar H2

10.5 bar H2

12.5 bar H2

Fig. 18. Comparison between the modelling results and the experimental results for different initial partial pressures of H 2 . (a) Shows the partial pressure of H 2 with timeand (b) the partial pressure of CH 4 with time. T ¼ 463 K, pCO: 0 ¼ 2 : 4bar and m cat ¼ 5 : 0g. Solid lines are the predictions of Model I. The symbols represent experimental resultswith different initial partial pressures of H 2 .

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 328

8/16/2019 CO y Al2O3

14/21

measured by the peak at 2070 cm 1 , decreased signi cantly. How-ever, a signi cant quantity of carbonyl species were still present after40 min in a ow of H 2 , indicating a strong adsorption of CO on thesurface of nickel. The intensity of the absorbance bands of the for-mate groups remained unchanged at the end of this period. This canbe compared with CO 2 methanation (results shown in Supplemen-tary information Section 6 ) where the carbonyl groups disappearedcompletely after 40 min in H 2 whilst the formate groups experienced

only a small decrease in intensity.

4. Theory

4.1. Reactor model

The batch reactor was modelled using the following set of ordinary differential equations, allowing for the water – gas shiftreaction and the fact that the selectivity of the reaction towardsCH4 was only about 25% of the consumed CO:

dp COdt

¼ mcat RT

V reactor 10 5 r 01 r

02 ð6Þ

dp H2dt

¼ mcat RT

V reactor 10 5 α r 01 þ r

02 ð7Þ

dp CH4dt

¼ mcat RT

V reactor 10 5 β r 01 ð8Þ

dp H2 Odt

¼ mcat RT

V reactor 10 5 r 01 r

02 ð9Þ

dp CO2dt

¼ mcat RT

V reactor 10 5r 02 ; ð10 Þ

Here r 01 is the rate of CO methanation, r 02 is the rate of the

water – gas shift reaction, α is the consumption ratio of H 2 to CO inreaction r 01 and β is the fractional conversion of CO of CH 4 . Thevalue of α is an average of the stoichiometric ratio of H 2 to CObased on the relative rates of Reactions (1) and (4) . In the fore-going, the value of β was found to be constant for temperatures453 – 473 K, but to be a strong function of the ratio of H 2 to CO. Thefractional conversion of CO to CH 4 was also found to be largelyconstant during period I, viz. before the depletion of CO. A linearcorrelation for β was estimated as a function of the ratio of theinitial partial pressure of H 2 to CO. The line of best t had a gra-dient of 0.044 7 0.008, intercepting the ordinate at 0.16 7 0.03,where the uncertainty represents a 95% con dence interval. Acorrelation coef cient of 0.91 was obtained. This correlation wasused to estimate the value of β in the solution of the aboveequations. It can also be assumed that Reactions (1) and (4) werethe two main reactions during the reaction. Therefore, the con-sumption ratio of pH2 to pCO, α , was estimated using

α ¼ 3 β þ 2 1 β : ð11 ÞEqs. (6) – (11) were solved using the initial conditions of theexperiments, i.e.

for t ¼ 0 pi ¼ pi : 0 ð12 Þ

where pi is the partial pressure of component i in bar and p i,0 isthe initial partial pressure of species i. Given suitable rateexpressions for r 01 and r

02 , the equations were solved using the

MATLAB solver ode45 to give the variation of the partial pressuresof CO, H2 , CH4 , H2 O and CO 2 with time, for comparison directly

with the experimental measurements.

4.2. Kinetic modelling

This Section investigates different expressions for the rate of COmethanation, represented by r 01 in the above equations, using aLangmuir – Hinshelwood approach. The active sites for the reactionwere assumed to be identical and their distribution uniformthroughout the catalyst pellets. Extra-particle and intra-particletransport effects were taken to be absent. Four kinetic models are

listed in Table 2 . For Models I–

III, it was assumed that COmethanation proceeded via the dissociative adsorption of CO witha sequence of steps:

COþ θ ⇌k1

k 1θ CO ðR5Þ

H2 þ 2θ ⇌k2

k 22θ H ðR6Þ

θ CO þ θ ⇌k3

k 3θ C þ θ O ðR7Þ

θ C þ θ H ⇌k4

k 4θ CH þ θ ðR8Þ

where ki and k i are the forward and reverse rates of reaction of the speci ed elementary step. Further steps include hydrogenationof the θ C and the subsequent desorption of θ CH4 to form CH 4 in thegas phase. This involves a succession of steps of the form:

θ C þ θ H⇌ θ CH þ θ θ CH þ θ H⇌ θ CH2 þ θ θ CH þ θ H⇌ θ CH3 þ θ θ CH3 þ θ H⇌ θ CH4 þ θ

Table 1Kinetic rate expressions based on different assumptions of the rate limiting stepand the most abundant surface species. p i is the partial pressure of component i.

Model Rate expression Rate-limiting step Most abundant sur-face species

I aI pH21 þ b I pCO

2 (13)Adsorption of H 2 CO

II aII pCO1 þ bII ffiffiffiffiffiffiffiffi pH2p (14)

Adsorption of CO H

III a III pCO1 þ bIII pCO þ c III ffiffiffiffiffiffiffiffi pH2p

2 (15)Dissociation of CO CO and H

IV aIV pH2 pCO1 þ b IV pCO

2 (16)Reaction of H 2 withadsorbed CO

CO

Table 2Kinetic rate expressions proposed by different studies for CO 2 methanation.

Rate expression Reference

r 0 ¼kpCO pH2

1 þ K CO pCO þ K H2 pH2 2 (17)

Vanho and Harriott (1980)

r 0 ¼ kpCO

1 þ K CO pCO 2 (18)

Van Herwijnen et al. (1973)(1980)

r 0 ¼kpH2

k1 1 þ K H20 : 5 pH2 0 : 5 þ K CO pCO

2

þ k2 1 þ K CO pCO

2 (19)

Sughrue and Bartholomew(1982)

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 329

8/16/2019 CO y Al2O3

15/21

θ CH4 ⇌ CH4ð g Þþ θ :

Three rate expressions, Eqs. (13) – (15) were derived based ondifferent rate limiting steps and the most abundant surface spe-cies, given in Table 2 . Model IV assumed an Eley – Rideal mechan-ism, where the rate-limiting step was taken as the reactionbetween gaseous H 2 and adsorbed CO, i.e.

H2 þ θ CO þ θ ⟹k5

2θ CH þ θ OH ðR9ÞIt is reasonable to assume that the surface of the catalyst is satu-

rated with adsorbed CO species in order for such a scenario to befeasible. Hence Eq. (16) in Table 1 . Three rate expressions for COmethanation, proposed by different investigators, are given in Table 2 .

4.3. Model discrimination

It is obvious that not all the expressions given in Table 1 agreewith the experimental results. Eq. (14) predicts the rate of reactionto increase with pCO and decrease with pH2 , contrary to theexperimental measurements, where the rate of reaction was found

to increase with pH2 and decrease with pCO. The same argumentcould be applied to Eq. (15) . Therefore, Eqs. (14) and (15) were notconsidered for further study. The rate expression proposed by VanHerwijnen et al. (1973) predicted that the rate of reaction woulddecrease with pCO at high values of pCO. However, they did notaccount for the effect of pH2 . Sughrue and Bartholomew (1982)proposed Eq. (19) with the assumption that the rate-limiting stepof the reaction was different at different temperatures. At tem-

peratures below 525 K, Eq. (19) collapses to Eq. (13) . Since all theexperiments performed in this study were below 500 K, onlyEq. (13) will be considered. There is some similarity between Eqs.(17) and (16) , discussed later. Vanho and Harriott (1980) derivedEq. (17) based on a Langmuir – Hinshelwood model with reactionbetween adsorbed carbon monoxide and hydrogen atoms.

Since the two most plausible rate expressions are Models I andIV, further comparison of these was performed by substituting therate expressions into the model of the reactor, and comparing theagreement between theory and experiment at different conditions.To do this, the parameters for each model were estimated based ona least-squares minimisation. Thus, the agreement between modeland experiment was studied by comparing the solution of the

0 1000 2000 3000 40000

1

2

3

4

5

6

7

8

Time / s

P a r t

i a l p r e s s u r e

/ b a r

2.4 bar CO

5.0 bar CO

7.8 bar CO

0 1000 2000 3000 40000

1

2

3

4

5

6

7

8

Time / s

P a r t

i a l p r e s s u r e

/ b a r

7.2 bar H2

9.2 bar H2

10.5 bar H2

0 1000 2000 3000 40000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Time / s

P a r t

i a l p r e s s u r e

/ b a r

2.4 bar CO

5.0 bar CO

7.8 bar CO

0 1000 2000 30000

0.005

0.01

0.015

0.02

0.025

0.03

0.035

Time / s

P a r t

i a l p r e s s u r e

/ b a r

2.4 bar CO

5.0 bar CO

7.8 bar CO

Fig.19. Comparison between the modelling results and the experimental results for different initial partial pressures of CO. The partial pressure of (a) H 2 , (b) CO, (c) CH 4 and(d) CO 2 are plotted against time. T ¼ 473 K, p CO: 0 ¼ 2 : 4 bar and m cat ¼ 5 : 0 g. The solid lines are the predictions of Model I. The symbols represent experimental results for

different initial partial pressure of CO.

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 330

8/16/2019 CO y Al2O3

16/21

system of ODEs with the measured temporal variation of the partialpressures of the various species. In the minimisation, the difference,di t ð Þ, between these values was compared for each iteration at timet with the experimental measurements for pCO2 , pH2 and pCH4 :

di t ð Þ ¼ pi; model t ð Þ pi ; exp t ð Þ ð20 Þ

where pi; model t ð Þ is the partial pressure of species i determined bythe solution of the ODEs and pi ; exp t ð Þ is the partial pressure of species i measured experimentally. The sum of all the squares of each component was evaluated at a given time, t , such that

D ¼ Xi d i t ð Þð Þ2

: ð21 Þ

Values of the parameters in the model were obtained byminimising D using the MATLAB routine lsqnonlin. Parameterswere estimated using only the kinetic measurements in period I of the experiments, viz. for pCO 4 0 : 1bar. During this process, rateexpressions for Models I and IV were substituted into r 01 of Eqs. (6) to (10) while a rate expression for the water – gas shiftreaction ( Lim et al., 2016 ) was substituted into r 02 , i.e.

r 0

2 ¼

aWGS pH2 O pCO1 þ bWGS pCO

: ð22 Þ

In general, the solution of Eqs. (6) to (10) was insensitive tovalues of r 02 because the maximum fractional conversion of CO toCO2 is only 0.02 – 0.05. Therefore, a WGS was determined empiri-cally at each temperature before the least-squares minimisationwas performed. Values of aWGS about 20% of those obtained by Limet al. (2016) were found to agree with the experimental mea-surements of pCO2 with time.

4.3.1. Model I Model I was rst proposed by Sughrue and Bartholomew

(1982) for CO methanation for temperatures below 525 K. Theyperformed continuous experiments on monolithic catalysts usingan internal recycle reactor. Since their experiments were per-formed in a continuous reactor, the batch method developed hereoffered an opportunity to compare techniques. Fig. 18 illustratesthe result using Model I for CO methanation and the experimental

results obtained at 453 K for different initial partial pressuresof H2 .

Fig. 19 compares the modelling and experimental results at473 K for different initial partial pressures of CO. In general, themodel was capable of predicting a fairly constant rate of reactionduring period I of the reaction. The kinetic parameters for Model I,a I and bI, at different temperatures are tabulated in Table 3 . It isnoted that bI represents a measure of the equilibrium constantbetween adsorbed and gaseous CO. In general, the predictedvalues of bI were found to be approximately constant over therange 443 – 473 K. However, the values obtained in Table 3 werevery much larger than those reported by other studies. For atemperature of 463 K, the value of bI was 22 bar

1 , based on theextrapolation of the measurements by Sughrue and Bartholomew

(1982) and Huang and Richardson (1978) gave bI ¼ 74 bar 1 .However, the reported values of bI by these researchers werebased on experiments at temperatures 4 523 K, signi cantlyhigher than those in the present work.

4.3.2. Model IV The agreement between the experimental measurements and

the modelling results, using Model IV for the rate of CO metha-nation, is illustrated in Figs. 20 and 21. It is clear that there isgood agreement between theory and experiment. Model IV pre-dicts a deceleration of the rate of reaction towards the end of period I of the reaction, so that the variation of pH2 with timeappeared to level off at t 4 2000 s in Fig. 20 (a). A similar beha-viour was found in Fig. 20(b), which illustrates the variation of

pCH4 with time. The values of bIV , which also represents a measureof the equilibrium constant between adsorbed and gaseous CO,are given in Table 4 . It was found that, as with Model I, there wasvery little variation in the values of bIV obtained from the least-squares minimisation at different temperatures. While the values

Table 3Values of the kinetic constants from the least-squares t of Model I with theexperimental results.

Temperature/K a I/mol s 1 kg 1 bI/bar 1

443 (3.4 7 0.2) 10 6 0.16 7 0.05453 (4.2 7 0.2) 10 6 0.05 7 0.02463 (9.8 7 0.5) 10 6 0.05 7 0.01473 (2.2 7 0.1) 10 5 0.07 7 0.01

0 1000 2000 3000 40000

2

4

6

8

10

12

14

Time / s

P a r t

i a l p r e s s u r e

/ b a r

7.2 bar H2

9.2 bar H2

10.5 bar H2

12.5 bar H2

0 1000 2000 3000 40000

0.2

0.4

0.6

0.8

1

Time / s

P a r t

i a l p r e s s u r e

/ b a r

7.2 bar H2

9.2 bar H2

10.5 bar H2

12.5 bar H2

Fig. 20. Comparison between the modelling results and the experimental results for different initial partial pressures of CO. (a) Shows the partial pressure of H 2 with timeand (b) the partial pressure of CH 4 with time. T ¼ 463 K, pCO: 0 ¼ 2 : 4 bar and m cat ¼ 5 : 0 g. The solid lines are the predictions of Model IV. The symbols represent experimental

results for different initial partial pressure of CO.

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 331

8/16/2019 CO y Al2O3

17/21

of bIV were an order of magnitude larger than those obtained forbI, they were still signi cantly smaller than those reported in theliterature.

5. Discussion

Fig. 2(a) and (b) clearly shows a transition in the behaviour of the reaction when pCO approached zero. When signi cant pCO waspresent in the gas phase, the rate of reaction was found to be

almost constant, as observed by the linear increase in pCH4 with

time. When pCO approached zero, the rate of production of pCH4was found to increase signi cantly, as illustrated by the large risein pCH4 . This behaviour was not found in the partial pressures of higher hydrocarbons, such as C 2 H6 and C 3 H8 , and their partialpressures were found to be approximately unchanged after theamount of CO present fell to zero. Also, the total equivalentamount of carbon in the gas phase decreased steadily over timebefore the depletion of CO, suggesting the accumulation of carbon-containing species during this period which could not beaccounted for by the sum of all the carbon-containing species inthe gas phase. The nature of these species is speculated to beeither liquid hydrocarbons in the pores of the catalyst or carbo-naceous species on the surface of catalyst. It should be noted thatCO adsorbed on the surface of the catalyst could also contribute tothe consumption of CO from the gas phase. The recovery of thetotal carbon in the gas phase after the depletion of pCO, i.e. duringperiod II of the batch reaction, is probably attributable to theincrease in the overall quantity of pCH4 produced. The conversionof the CO 2 accumulated during period I to CH 4 was insuf cient toaccount for the total amount of pCH4 produced in period II sinceonly a very small amount of pCO2 was produced during period I.

This suggests either the hydrogenation of carbonaceous species on

0 1000 2000 3000 40001

2

3

4

5

6

7

8

Time / s

P a r t

i a l p r e s s u r e

/ b a r

2.4 bar CO

5.0 bar CO

7.8 bar CO

0 1000 2000 3000 40000

1

2

3

4

5

6

7

8

Time / s

P a r t

i a l p r e s s u r e

/ b a r

2.4 bar CO

5.0 bar CO7.8 bar CO

0 1000 2000 3000 40000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Time / s

P a r t

i a l p r e s s u r e

/ b a r

2.4 bar CO

5.0 bar CO

7.8 bar CO

0 1000 2000 30000

0.005

0.01

0.015

0.02

0.025

0.03

0.035

Time / s

P a r t

i a l p r e s s u r e

/ b a r

2.4 bar CO5.0 bar CO

7.8 bar CO

Fig. 21. Comparison between the modelling results and the experimental results for different initial partial pressures of CO. The partial pressure of (a) H 2 , (b) CO, (c) CH 4 and(d) CO 2 are plotted against time. T ¼ 473 K, pH2 : 0 ¼ 7 : 2 bar and m cat ¼ 5 : 0 g. The solid lines are the modelling the predictions of Model IV. The symbols represent experimentalresults for different initial partial pressure of CO.

Table 4Kinetic constants from the least-squares t of Model IV with the experimentalresults.

Temperature/K a IV /mol s 1 kg 1 bIV /bar 1

443 (8.0 7 0.2) 10 6 0.93 7 0.03453 (1.3 7 0.1) 10 5 0.90 7 0.10

463 (3.2 7 0.2) 105

0.87 7 0.04473 (8.0 7 0.3) 10 5 0.96 7 0.02

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 332

8/16/2019 CO y Al2O3

18/21

the surface of the catalyst to form CH 4 or the production of CH 4 viahydrogenolysis of the higher hydrocarbons to form CH 4 , i.e. thesplitting of higher hydrocarbons to light hydrocarbons. Ni catalystsare known to perform hydrogenolysis on longer-chain paraf ns(Kikuchi and Morita, 1969 ). In fact, θ-Al2 O3 is active in thehydrogenolysis of cis-2-pentenenitrile ( McGregor et al., 2010 ).

The values of pH2 and pCO had a signi cant in uence on the rateand selectivity of the reaction. When pH2 was increased, the rate of reaction increased and the selectivity towards CH 4 was greater.However, the selectivity towards other light paraf ns remainedrelatively unchanged. The increased selectivity of the reactiontowards CH 4 led to a higher proportion of carbon contained in thespecies in the gas-phase as illustrated in Fig. 8. Conversely, when p

CO was increased, both the rate of reaction and the CH 4 selectivity

decreased. Over the range 443 – 473 K, the fractional conversion of CO to CH4 was found to range from 0.22 to 0.30. In general, thiswas found to be slightly lower than the values reported in theliterature of about 0.4 – 0.7% (Zhang et al., 2013 ; Fujita and Take-zawa, 1997 ; Vance and Bartholomew, 1983 ). However, it is notedthat the values of pCO explored in this study were signi cantlylarger than those used in investigations by other researchers.Therefore it is reasonable that the fractional conversion of CH 4 wasslightly lower.

5.1. Comparison with CO 2 hydrogenation

The methanation of CO 2 over the 12 wt% Ni/ γ -Al2 O3 catalyst in

a batch reaction has been investigated by Lim et al . (2015) . When asimilar partial pressure of CO 2 and H 2 was introduced into thereactor in the presence of CO, the reaction was found to proceedvia CO methanation instead of CO 2 methanation. This means thatCO2 behaves as a spectator molecule in the bulk gas phase and isnot involved as a reactant in any reaction until the depletion of COin the gas phase. This is consistent with Zhang et al . (2013) , whostudied CO methanation from 548 – 633 K and over a total pressurerange of 1 – 5 bar in a continuous, xed-bed reactor. They found noeffect of CO 2 on the rate and selectivity of the reaction when0.6 bar of CO 2 was co-fed with a mixture of 1.2 bar H 2 and 0.6 barCO. This observation suggests that nickel has a much strongeraf nity for CO compared to CO 2 and in the presence of CO, thesurface of the catalyst preferentially adsorbs CO over CO 2 . This has

been con rmed by Inui et al. (1978), who found that the amount of

CO adsorbed was nearly six times the amount of CO 2 adsorbed ona 5 wt% Ni/SiO 2 at 295 K.

Fig. 22 compares the rates of production of CH 4 and the rates of consumption of CO and CO 2 , in CO and CO 2 methanation reactionsrespectively, in the batch reactor. At 463 K, the initial rate of consumption of CO 2 , for an initial condition of pH2 ; 0 ¼ 7 : 2 bar and pCO2 ¼ 2 : 4 bar, was the same as the rate of consumption of COwhen pH2 ; 0 ¼ 7 : 2 bar and pCO ¼ 2 : 4 bar. However, the rates of pro-duction of CH 4 with time were very different. The CH 4 selectivityin CO2 methanation was found to exceed 99.5%, which is sig-ni cantly larger than that observed in CO methanation, with avalue of only about 25%. When CO 2 methanation was performed,the rate of reaction decreased over time as a result of the decreasein the partial pressures of the reactants and no marked increasesin the rate production of CH 4 were found on depletion of the CO 2 ,in contrast to the case with CO methanation during the transitionfrom period I to II. The strong adsorption of CO on the surface of the catalyst suggests that the ratio of hydrogen to carbon on thesurface is much lower than in CO 2 methanation, explaining thereduced selectivity for CH 4 in CO methanation.

The variation of the rate with temperature of CO methanationwith temperature was found to be more sensitive than that of CO 2methanation, as observed in Fig. 22(b). For a 10 K rise in tem-perature, the increase in the rate of consumption of CO, in COmethanation, was more than the increase in the rate of con-sumption of CO 2 in CO2 methanation. This implies that theapparent activation energy of CO methanation is different to thatof CO2 methanation, suggesting that the rate-limiting step is dif-

ferent in each case. This is consistent with the postulate that therate-limiting step in CO 2 methanation is the dissociation of adsorbed CO while that in CO methanation is either the adsorptionof H 2 or the reaction between gaseous H 2 and adsorbed CO. Havingdifferent rate-limiting steps for these processes is not unreason-able given that the surface of the catalyst is expected to be sig-ni cantly different under different reaction conditions.

The rate of CO 2 methanation decreased in the presence of pH2 O(Lim et al., 2015 ), but the results in the present paper show nosuch in uence of H 2 O in CO methanation. The literature suggests,however, that H 2 O inhibits the rate of CO methanation ( Zhanget al ., 2013 ; Sughrue and Bartholomew, 1982 ). This con ict inobservations is probably explained by the experiments in thispaper being performed at temperatures signi cantly lower than

those used in studies where inhibition occurred. It is likely that the

Fig. 22. Partial pressure of (a) CH 4 and (b) CO or CO 2 with time at different temperatures for CO methanation and CO 2 methanation. The solid symbols represent the partialpressures of CH 4 and CO for initial conditions of pH2 ; 0 ¼ 7 : 2bar and p CO ¼ 2 : 4bar. Open symbols represent the partial pressure of CH 4 and CO 2 for initial conditions of pH2 ; 0 ¼ 7 : 2 bar and p CO2 ¼ 2 : 4 bar. The mass of catalyst m cat ¼ 5.0 g for all experiments.

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 333

8/16/2019 CO y Al2O3

19/21

low temperature, together with the strong adsorption of CO, led tolittle opportunity for the competitive adsorption of H 2 O on thesurface of the catalyst, which has been proposed as the primarymethod of inhibition of the rate in CO 2 methanation.

5.2. Nature of catalyst surface

Evidence for the presence of carbonaceous species on the sur-face of the catalyst or heavy hydrocarbons in the pores of thecatalyst can be found in the measurements from temperature-programmed studies of the spent catalyst. That signi cant quan-tities of carbon-containing species were available on the surface of the spent catalyst is evident from the evolution of CO 2 in Fig. 14and CH 4 in Fig. 15. The temperature-programmed studies alsosuggest two main forms of carbon based on the presence of tworanges of temperature in which CH 4 and CO 2 were evolved duringthe temperature programme. Kester and Falconer (1984) madesimilar observations and suggested that the CO was adsorbed ontwo different sites, i.e. one where nickel atoms are bonded to othernickel atoms, the other where there is a stronger interaction

between nickel and the supporting material, i.e. Al2 O3 . This is

consistent with the temperature-programmed reduction studies of the passivated catalyst, where two main regions of reduction of nickel were found. The presence of multiple active sites would alsosuggest that different processes are active at different tempera-ture, which would lead to different kinetic expressions as sug-gested by Sughrue and Bartholomew (1982) .

It is challenging to decide whether the presence of carbon-containing species on the surface of the catalyst is the result of adsorbed CO, liquid hydrocarbons or carbonaceous species on thesurface from the existing evidence. Studies by Tottrop (1976) andGardner and Bartholomew (1981) have shown that the rate of formation of carbon on the surface of the catalyst decreases in thepresence of H 2 and H 2 O, both of which were available in abun-dance for most of the duration of the batch reaction. By extra-polating the rate of carbon deposition observed by Gardner andBartholomew (1981) , where the mass of the catalyst was mon-itored under a ow of H 2 and CO, there is negligible formation of carbon at temperatures below 473 K. The origin of the carbondetected in the temperature-programmed studies is probably acombination of small quantities of liquid hydrocarbons in the

pores and adsorbed CO.

0 2000 4000 6000 8000 10000 120000

0.5

1

1.5

2

2.5

Time / s

P a r t

i a l p r e s s u r e

/ b a r

0 2000 4000 6000 8000 10000 120000

1

2

3

4

5

6

7

8

Time / s

P a r t

i a l p r e s s u r e

/ b a r

0 2000 4000 6000 8000 10000 120000

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

Time / s

P a r t

i a l p r e s s u r e

/ b a r

0 2000 4000 6000 8000 10000 120000

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

Time / s

P a r t

i a l p r e s s u r e

/ b a r

Additional CO andH2 introduced att = 3600 s

Additional CO andH2 introduced att = 3600 s

Additional CO andH2 introduced att = 3600 s

Additional CO andH2 introduced att = 3600 s

Additional CO andH2 introduced att = 3600 s

Fig. 23. Comparison of the experimental results and the modelling results of Models I and IV. The partial pressures of (a) CO, (b) H 2 , (c) CH4 and (d) CO 2 were compared. Att ¼ 0 s, p CO; 0 ¼ 2 : 4 bar and p H2 ; 0 ¼ 7 : 2 bar. Additional CO and H 2 were introduced at 3600 s. For all experiments, T ¼ 463 K and m cat ¼ 5.0 g. The solid black line representsmodel I and the red line represents Model IV.

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 334

8/16/2019 CO y Al2O3

20/21

The nature of the adsorbed CO was explored using DRIFTS.Signi cant amounts of bridge and linear carbonyl groups, asshown by the persistent absorbance bands at 2180 and 2120 cm 1

in Fig. 17(a), are present on the surface of the catalyst. Thedecrease in the intensity of the carbonyl groups when the catalystwas subjected to pure H 2 at 463 K suggests that the hydrogenationof carbonyl groups was responsible for the evolution of CH 4 in theTPR pro le of the spent catalyst at about 473 K. Furthermore, the

intensities of bands assigned to the formate groups remainedunchanged during the same period, suggesting that a highertemperature is required to hydrogenate these species. The evolu-tion of additional CO 2 in the TPD and CH 4 in the TPR pro les is ingood agreement with this observation. The decrease in theintensity of the carbonyl peaks after CO methanation was found tobe slower than that after CO 2 methanation, which is expectedgiven that the surface of the catalyst in CO methanation has beenfound to be much more heavily covered by carbonyl speciescompared to the case with CO 2 methanation. The IR spectra attemperature above 623 K were also found to be markedly differentto those at lower temperatures, suggesting a possible change inthe surface of the catalyst as the temperature was raised above623 K.

5.3. Model comparison

It is dif cult to discriminate between the two models based onthe computer tting of the models to the experimental resultsbecause both models gave reasonable agreement.

Model I assumes that the rate-limiting step is the adsorption of H2 on the surface of the catalyst and that the surface of the catalystis mainly adsorbed CO. This agrees with Sughrue and Barholomew(1982) , who also observed that the reaction order for [H 2 ] wasunity and that of [CO] was 1 at 475 K. However, the ranges of pCOand pH2 explored in Sughrue and Barholomew's (1982) studies( pCO from 0.05 – 0.12 bar and pH2 from 0.12

– 0.30 bar) were very lowcompared to those explored in this paper. However, it is inter-

esting to note that the value of bI was very much smaller than thatreported by Sughrue and Barholomew (1982) .

Model IV was based on an Eley – Rideal mechanism. For thismechanism to be true, the surface of the catalyst must be satu-rated with CO. The evidence for this has already been discussed.However, it should be noted that a similar rate expression wasderived by Vanho and Harriott (1980) , as given in Eq. (17) . If theadsorption constant of H 2 is small, then Eq. (17) converges to Eq.(16) . The derivation of Eq. (17) was based on a rate-limiting stepinvolving the hydrogenation of adsorbed CO by two adsorbed H. Itis impossible to discriminate between the two underlyingmechanisms for the reaction of CO and H 2 , highlighting the lim-itations of kinetic studies where different mechanisms could leadto the same rate expression. Nevertheless, it can be established

that the strong adsorption of CO on the surface of the catalystcompetes with H 2 for available active sites. It is also possible thatH2 adsorbs on different nickel sites to CO as a result of hetero-geneity in the surface of the catalyst, as suggested by Anderssonet al. (2008) . The adsorbed hydrogen on different sites wouldreplace the position of gaseous CO in the derivation of the Eley –

Rideal mechanism to obtain the same expression. There alsoremains the possibility of the dissociated hydrogen migrating tothe Al 2 O3 support, i.e. hydrogen-spillover, which occurs on nickelcatalysts supported on Al 2 O3 (Kramer and Andre, 1979 ; Gardes etal., 1974 ). The reaction between H on Al 2 O3 and CO on the nickel isalso consistent with Model IV. Furthermore, the ability of Al 2 O3 tocatalyse the hydrogenolysis of longer-chain hydrocarbons(McGregor et al., 2010 ) also suggests the presence of active sites on

the supporting material.

Fig. 23 shows the variation of the partial pressure of CO, H 2 , CH4and CO 2 as a function of time for an additional batch experimentwith an initial condition of pCO; 0 ¼ 2 : 4 bar and pH2 ; 0 ¼ 7 : 2 bar at463 K. At t ¼ 3600 s, additional H 2 and CO were introduced in theratio of 3:1. In general, both models were found to give goodagreement with the experimental results, but both deviate fromthe experimental measurements at low pCO. These models werederived on the assumption that the surface of the catalyst remains

approximately constant over the length of the reaction. Since therelative quantities of adsorbed species on the surface of the cata-lyst is expected to change signi cantly as CO depletes, it is likelythat values of the kinetic parameters also change during thisperiod, which could be a limitation of the analysis. A change in thereaction mechanism as a result of the change in the amount of gaseous CO remains a possibility, suggesting that different rateexpressions are valid for different pCO. It can also be seen that thetransition from period I to II was delayed when additional H 2 andCO was introduced at 3600 s. This is evident from the extension of the increase of pCH4 at an approximately constant rate until theeventual depletion of CO when a marked increase in pCH4 wasobserved. This experiment con rmed the need for CO to bedepleted before carbon-containing species on the surface could behydrogenated or adsorbed liquid hydrocarbons could undergohydrogenolysis.

In both models, the increase in the rate of production of CO 2was successfully reproduced by the rate expression for the water –

gas shift reaction, Eq. (22) , with the increase in pH2 O leading to anincrease in the rate of reaction. Measuring the effect of pH2 on therate of the water – gas shift reaction independent of CO methana-tion is dif cult because the introduction of H 2 to an initial mixtureof H2 O and CO would lead to the consumption of CO by both thewater – gas shift and CO methanation.

6. Conclusions

The conclusions were as follows:

The rate of reaction of the CO methanation reaction and theselectivity of CH 4 was found to increase with pH2 and decreasewith pCO.

Two main regimes were identi ed over the length of the reac-tion in batch. In all experiments, the reaction initially proceededwith a constant rate period. This was followed by a markedincrease in the rate of production of CH 4 after the depletion of pCO. This increase in CH 4 was attributed to the hydrogenation of remaining carbonyl groups on the surface as well as thehydrogenolysis of long-chained paraf ns in the reactor.

The selectivity for CH 4 was found to be signi cantly lower thanthat observed in CO 2 methanation. This was found to be con-sistent with the low H 2 to CO ratio on the surface of the catalyst.

Temperature-programmed studies performed on the spentcatalyst identi ed two main types of carbonaceous species onthe surface of the catalyst. These ndings were correlated withthe observations in the DRIFTS studies in order to postulate thepresence of (i) carbonyl species on nickel clusters and (ii)formate groups on nickel sites which have a stronger interactionwith the alumina support. The former was found to be reactiveat the temperatures explored in this paper.

The rate of CO methanation was insensitive to the presence of H2 O. This was attributed to the strong af nity of the nickelcatalyst for CO, which saturates the surface of the catalystleaving little opportunity for the adsorption of H 2 O.

The kinetic measurements were compared against the model-ling predictions from several rate expressions. Eqs. (13) and (16)

were both found to provide reasonable agreement with the

J.Y. Lim et al. / Chemical Engineering Science 146 (2016) 316 – 336 335

8/16/2019 CO y Al2O3

21/21

experimental results. The derivation of Eq. (13) assumes theadsorption of H 2 to be the rate-limiting step while the rate-limiting step of Eq. (16) is the reaction of gaseous H 2 withadsorbed CO. The strong adsorption of CO on the surface of thecatalyst, evident from various experimental observations, isconsistent with both mechanisms. The agreement of thesemodels was found to be less good towards lower pCO, suggestingthat the kinetic parameters or the rate mechanism could be a

function of the CO coverage on the surface. However, it remainsdif cult to discriminate between the validity of Eqs. (13) and(16) with the existing experimental evidence.

The production of CO 2 during the batch reaction of CO metha-nation was the result of the water – gas shift reaction.

Using a batch reactor, rather than a continuous one, for thestudy of CO methanation has certain advantages. The ability toobserve intermediate products, such as CO 2 , over the length of batch reaction has demonstrated the strong af nity of the nickelcatalyst towards CO. Furthermore, the validity of different rateexpressions, as well as the corresponding kinetic constants, wasreadily examined over a wide range of partial pressures using thebatch reactor. The progress of CO 2 methanation and CO metha-nation in the batch reactor was signi cantly different, as a result of different rate-limiting steps and a signi cantly higher quantity of carbonyl groups on the surface of the catalyst in the case of COmethanation.

Acknowledgements

JYL was funded by the Cambridge International ScholarshipScheme. The Cambridge Philosophical Society, the LundgrenResearch Award and Corpus Christi College are also gratefullythanked for contr