lable at ScienceDirect

Energy 90 (2015) 748e758

Contents lists avai

Energy

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

Development of catalysts for hydrogen production through theintegration of steam reforming of methane and high temperaturewater gas shift

Trevor L. LeValley a, Anthony R. Richard a, Maohong Fan a, b, *

a Department of Chemical and Petroleum Engineering, University of Wyoming, Laramie, WY 82071, USAb School of Energy Resources, University of Wyoming, Laramie, WY 82071, USA

a r t i c l e i n f o

Article history:Received 16 April 2015Received in revised form21 July 2015Accepted 24 July 2015Available online 15 August 2015

Keywords:Hydrogen energyLow temperature steam reforming ofmethaneWater gas shiftReduction in carbon emissionsNoble metal replacement

* Corresponding author. Department of ChemicalUniversity of Wyoming, Laramie, WY 82071, USA. Tel

E-mail address: mfan@uwyo.edu (M. Fan).

http://dx.doi.org/10.1016/j.energy.2015.07.1060360-5442/© 2015 Elsevier Ltd. All rights reserved.

a b s t r a c t

There is great concern about the increasing demand for energy with respect to carbon emissions, buthydrogen (H2), a clean fuel, could help alleviate this concern. The replacement of fossil fuels with H2 iscost prohibitive, but integration of SRM (steam reforming of methane) and WGS (water gas shift) couldgreatly decrease production costs. A composite catalyst of nickel, cerium, zirconium, and zinc wasdesigned to provide activity in both the SRM and WGS reactions. The catalysts were characterized by avariety of techniques including BET (Brunauer Emmett Teller), TEM (transmission electron microscopy),SEM (scanning electron microscopy), TGA (thermogravimetric analysis), and XRD (X-ray diffraction). Itwas found that the addition of zinc decreased the surface area, and therefore activity of the SRM reaction,although it increased WGS activity as observed by improved carbon dioxide selectivity and H2 produc-tion. Zinc also increased resistance to carbon deposition. Additionally, aging of precipitates duringcatalyst synthesis improved stability. A Ce/Zr/Zn catalyst doped with 10% Ni and aged for 2 h was foundto have a final conversion of nearly 20% at 650 �C, and high CO2 selectivity around 55%. This catalyst is animportant step in the emerging field of low temperature SRM, a field that could lead to a reduction incarbon emissions.

© 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Hydrogen (H2) is becoming an important clean fuel, and due tothe increasing concerns over carbon emissions [1e4] researchfocused on its production has also increased [5e7]. Until sustain-able renewable methods are developed, providing a more viable,efficient, and reliable hydrogen product, SRM (steam reforming ofmethane) and WGS (water gas shift) reactions based on fossil fuelswill remain the major H2 production technology [8e11]. Thesereactions are important for an emerging hydrogen economy andthe implementation and distribution of hydrogen to consumers. Intypical hydrogen production, the first step is SRM performed in ahigh temperature reactor (~1000 �C), where methane (CH4) andsteam are used to produce H2 and carbonmonoxide (CO), a mixturecalled syngas [12] Syngas is valuable as a feedstock for

and Petroleum Engineering,.: þ1 307 766 5633.

FischereTropsch synthesis and the production of many chemicalsincluding ammonia and methanol [13,14]. The SRM reaction [15] isshown as R1.

CH4 þ H2O4COþ 3H2 (R1)

While many feedstocks can be used for reforming, methane ispreferred for hydrogen production as it has high hydrogen to car-bon ratio and lower byproduct formation compared to otherstarting materials.

In separate reactors, the subsequent WGS steps take placeduring which CO and steam are converted to produce additionalhydrogen and CO2, as shown as R2. This WGS step is used to in-crease the H2 to CO ratio for the production of hydrogen [16].

COþ H2O4H2 þ CO2 (R2)

The complete conversion of CO to CO2 at high temperatures isimpossible without product removal due to equilibrium constraints[17]. This necessitates that the WGS reaction be done in two re-actors, a HTSR (high temperature shift reactor) and a LTSR (low

Table 1Calculated compositions of catalysts in weight percentage.

Catalyst Ce (wt%) Zr (wt%) Zn (wt%) Ni (wt%)

0Zn0Ni 80 20 0 010Zn0Ni 72.7 18.2 9.1 020Zn0Ni 66.7 16.7 16.7 030Zn0Ni 61.5 15.4 23.1 040Zn0Ni 57.1 14.3 28.6 0

0Zn5Ni 76.2 19.0 0 4.810Zn5Ni 69.3 17.3 8.7 4.820Zn5Ni 63.5 15.9 15.9 4.830Zn5Ni 58.6 14.7 22.0 4.840Zn5Ni 54.4 13.6 27.2 4.8

0Zn10Ni 72.7 18.2 0 9.110Zn10Ni 66.1 16.5 8.3 9.120Zn10Ni 60.6 15.2 15.2 9.130Zn10Ni 55.9 14.0 21.0 9.140Zn10Ni 51.9 13.0 26.0 9.1

0Zn15Ni 69.6 17.4 0 13.010Zn15Ni 63.2 15.8 7.9 13.020Zn15Ni 58.0 14.5 14.5 13.030Zn15Ni 53.5 13.4 20.1 13.040Zn15Ni 49.7 12.4 24.8 13.0

T.L. LeValley et al. / Energy 90 (2015) 748e758 749

temperature shift reactor). The HTSR converts CO equivalent to theequilibrium, followed by the LTSR which often converts theremaining CO [18]. Due to the use of syngas in the aforementionedindustries, the formation of CO2 in steam reforming has beenselected against [19]. However, for the production of hydrogen, thisshould be encouraged to allow for smaller WGS reactors therebylowering the overall cost.

In an effort to lower hydrogen production costs, integration ofSRM and WGS using a composite catalyst is an attractive choice.When both SRM and WGS proceed simultaneously, the overall re-action becomes R3.

CH4 þ 2H2O44H2 þ CO2 (R3)

Reaction conditions for the integrated reactor need to satisfy therequirements for both SRM and WGS. At higher pressures SRM ishighly suppressed, but conversion linearly increases with temper-ature [20]. Traditional SRM temperatures would be too high toallow for more than a small conversion of CO throughWGS, thoughnew catalysts capable of low temperature SRM have been found[21,22]. Under these low temperature conditions, conversion of COto CO2 through WGS would also be possible, thus a compositecatalyst for both SRM and WGS with these characteristics could beused for hydrogen production. If high CO conversion could beachieved in the integrated reactor, then the use of a single down-stream low temperatureWGS reactor could become a viable option.This would dramatically decrease the cost of the production ofhydrogen from fossil fuels. The combination of WGS and SRM hasbeen attempted [22e27], although no catalyst has been found withhigh activity for both reactions, high stability, efficient operation atlower temperatures, and a low steam to carbon ratio (S/C).

Ce, Zr, Zn, and Ni were the four elements chosen to compose thecatalyst for this work. Ce was chosen as it shows promise in manyapplications as a catalyst support and promoter [14], partiallyattributed to its high OSC (oxygen storage capacity) which is inte-gral to the SRM reaction [28]. Zr was chosen as it increases the OSCof Ce supports [28] and has been shown to increase stability of Nicatalysts [29]. Zn has a high propensity to increase the CO2 selec-tivity, and Ni is used as an inexpensive alternative to noble metalssince it has fair activity for both reactions. All four elements havebeen proven to have activity in both SRM andWGS [30]. Due to thiscrossover in activity, a composite catalyst including these four el-ements was designed to convert a large percentage of CH4 and tohave high selectivity towards CO2. Ni is susceptible to carbondeposition [31,32], which is often one of the largest causes ofdeactivation in both SRM andWGS catalysis. It has been shown thatnickel crystal size greatly effects carbon deposition with smallercrystals beingmore resistant [33]. Due to nickel's affinity for coking,the desired catalyst must also reduce the activity in two cokingreactions, CO disproportionation and CH4 decomposition, shown asR4 and R5 respectively [34].

2CO4C þ CO2 (R4)

CH44C þ 2H2 (R5)

The catalysts evaluated in this work are prepared with Ce:Zr at a4 to 1 ratio as ratios in the range of three to four times the Celoading compared to Zr have been shown to have higher activity forSRM; however this ratio is not optimized for this reaction system[28,35,36]. Zn is loaded between 0 and 40 wt% of the total Ce/Zrmixture to increase theWGS activity. Finally, nickel is doped as 5,10or 15 wt% of the metal, and while the best methane reforming isfound on Ni/Ce/Zr catalysts with 15% Ni, higher CO2 production isfound at lower nickel concentrations [37]. This research wasdesigned to progress the field of integrated SRM and WGS, and the

development of new catalysts is the most important step towardsthe realization of this goal.

2. Experiments

2.1. Catalyst preparation

A series of catalysts were prepared by co-precipitation inpredetermined ratios of zirconium (IV) oxynitrate hydrate (99%,SigmaeAldrich), zinc nitrate hexahydrate (98%, SigmaeAldrich),and ammonium cerium (IV) nitrate, (99%, Fluka). For co-precipitation of the mixed nitrates, 400 ml of deionized (DI)water was stirred at 600 RPM and heated to 60 �C where it waskept until the nitrates dissolved. The pH was then raised to 9.0using liquid ammonium hydroxide (28.97%, Fisher Scientific), tocreate the desired precipitate. After filtering and washing with DIwater to remove excess ions, the resulting precipitate wasallowed to dry overnight and subsequently calcined at 500 �C indry air for 1 h. The catalyst was then sieved to obtain particles nolarger than 125 mm in diameter (sieve No. 120). Nickel (II) nitrate(99%, SigmaeAldrich) was deposited by incipient wetnessimpregnation at 5, 10, or 15 wt% of the total metal content, fol-lowed by additional drying and 1 h calcination in dry air at500 �C. All catalysts were prepared with the desirable ratio of 4 to1 for Ce to Zr. Due to this unchanging ratio, the catalysts will bedenoted only by their Zn and Ni content to delineate betweenthem. For example, the catalyst containing no Zn, precipitatedwith 80% Ce, 20% Zr, and 0% Zn, and doped with 15 wt% Ni isdenoted as 0Zn15Ni. The catalyst 20Zn10Ni was used to investi-gate the effect of precipitate aging for times of 0, 1, and 2 h. Agedcatalysts are prepared using precipitation conditions previouslydescribed with the addition of aging in the mother liquor at 60 �Cand a pH of 9.0 for the desired time prior to filtering and washing.Aged catalysts are prefixed with 1A or 2A to denote an aging timeof one or two h, respectively. Calculated weight percentages forall catalysts are shown in Table 1.

2.2. Catalyst characterization

Elemental analysis was performed with a PerkinElmer ICP-OES(Inductively Coupled Plasma Optical Emission Spectrometer) to

T.L. LeValley et al. / Energy 90 (2015) 748e758750

determine loading percentages of nickel. The samples were dilutedfor measurement after being dissolved in concentrated hydro-chloric acid for 48 h. Multi-point BET (Brunauer Emmett Teller)surface area, pore volume, and pore radius were measured usingadsorption isotherms of N2 porosimetry. Hydrogen chemisorptionwas performed to determine nickel dispersion, average nickelparticle size, active metal surface area (SAm), and to calculate TOF(turn over frequency). Both chemisorption and physisorption wereperformed on a Quantrachrome instruments autosorb iQ. XRD (X-ray diffraction) studies were conducted on a Rigaku smartlab X-raydiffractometer with monochromatic Cu Ka radiation, operating atconditions of 40 kV and 40 mA. 2q was recorded from 20� to 80�

with a step size of 0.02� per second. XRD peak identification wascompleted using literature defined peaks, and the lower limit of Niparticle size was determined using the Scherrer equation. TGA(Thermogravimetric analysis) was conducted on an SDTQ600. TEM(Transmission electron microscopy) using a 200 kV FEI Tecnai G2F20 and SEM (scanning electron microscopy) using a Quanta 450FEG operated at 20 kV were used to determine morphology of theaged catalysts.

2.3. Catalyst evaluation

SRM activity experiments were carried out at ambient pressurein a tubular quartz reactor of 3.57 mm inside diameter. The reactorwas heated by a Thermo Scientific Lindberg Blue M electric furnaceand temperature was monitored using an internal thermocouple.The packed bed reactor was loaded with 100 mg of catalyst mixedwith 300 mg of sand ±0.002 g. Catalysts were reduced to the activephase by heating to 400 �C while a H2 to N2 mixture at a ratio of1:7.5 flows for 2 h. During reactivity experiments, the ratio of CH4,water, and N2 was 1:2:7.5, giving a S/C of 2. The S/C ratio of 2 waschosen to allow for the complete conversion of CH4, followed by thecomplete conversion of CO. DI water was injected by a New EraPump Systems Inc. syringe pump through a coil heated to 180 �C toproduce steam. A desiccator filled with Drierite, a mixture of cal-cium sulfate and cobalt sulfate, was used to remove unreactedwater. Catalytic activity was measured using an Inficon 3000 MicroGC while the temperature was ramped from 500 �C to 850 �C. AnMKS Cirrus 1000 mass spectrometer was used to determine steadystate. All gas flows were controlled by Porter Instruments flowcontrollers, and the reaction conditions mentioned yield a spacevelocity of approximately 4500 h�1. The reaction setup is shown inFig. 1.

Fig. 1. Experimental setup use

2.4. Calculation methods

Calculations were performed using the effluent gas streams ofH2, CH4, CO, and CO2 measured via GC. An internal standard wasused to adjust the concentrations of all effluent gases. Conversion, X(%), is measured as a sign of catalyst activity, with a conversion of100% showing complete conversion. The equation used to calculateX is E1:

X ¼ CH4;in � CH4;out

CH4;in$100% (E1)

Selectivity (YCO2) was calculated to determine the percentage of

carbonmonoxide that was converted to CO2, and the equation usedis shown in E2:

YCO2¼ CO2;out

COout þ CO2;out$100% (E2)

Activity (aH2) was calculated to determine the amount of

hydrogen produced per gram of catalyst per hour using E3:

aH2¼ H2;out

gcat$hr(E3)

where H2,out is the moles of hydrogen leaving the reactor, and gcat isthe amount of catalyst used. TOF is used as a measure of activity onan active metal site basis, and is calculated using E4:

TOF ¼ aH2$SNi$NA

SAm(E4)

where SNi is the nickel surface area per atom and SAm is the activemetal surface area determined by chemisorption tests, and NA isAvogadro's number.

3. Results and discussion

3.1. Characteristics

Catalyst surface area was calculated through analysis of BETadsorption and desorption isotherms, which showed hysteresisleading to some level of mesoporous structure. As shown in Table 2,the surface area typically decreases with both increasing Zn and Niconcentration; however, at 15% Ni an increase in surface area is

d for catalyst evaluation.

Table 2Catalyst characteristics [surface area (SBET); pore volume (Pv); pore radius (Pr);elemental nickel analysis (Ni); Ni particle size: Scherrer equation (dNiO); used Niparticle size: Scherrer equation (UdNiO)].

Catalyst SBET (m2/gcat) Pv (cm3/g) Pr (Å) Ni (%) dNiO (nm) UdNiO (nm)

0Zn0Ni 138.186 0.031 15.661 e e e

10Zn0Ni 106.485 0.032 14.812 e e e

20Zn0Ni 76.537 0.024 15.656 e e e

30Zn0Ni 61.988 0.024 14.846 e e e

40Zn0Ni 54.315 0.042 19.615 e e e

0Zn5Ni 106.590 0.039 15.668 4.6 e e

10Zn5Ni 92.173 0.033 15.663 6.4 e e

20Zn5Ni 71.310 0.060 15.668 5.5 e e

30Zn5Ni 51.640 0.039 18.508 5.1 e e

40Zn5Ni 50.562 0.050 19.620 7.6 e e

0Zn10Ni 83.408 0.031 16.563 11.1 27.9 109.810Zn10Ni 76.719 0.031 15.669 12.2 3.1 74.820Zn10Ni 73.135 0.033 14.838 12.8 22.4 38.130Zn10Ni 70.853 0.035 15.671 11.8 26.2 53.840Zn10Ni 46.789 0.043 15.681 13.9 24.6 38.1

0Zn15Ni 85.339 0.035 16.596 11.2 32.2 101.810Zn15Ni 81.188 0.059 15.638 15.6 30.9 183.720Zn15Ni 87.354 0.058 15.676 13.1 34.2 54.430Zn15Ni 55.074 0.035 15.639 16.2 37.0 120.440Zn15Ni 58.624 0.051 15.661 14.5 33.7 289.0

Fig. 2. XRD spectra for fresh catalyst. [(a) undoped; (b) 5% Ni doped; (c) 10% Ni doped;(d) 15% Ni doped. *: NiO, #: CeO2, ,: ZrO2. Radiation source: Cu Ka; voltage: 40 kV;amperage: 40 mA; step size: 0.02�/sec].

T.L. LeValley et al. / Energy 90 (2015) 748e758 751

observed along with an increase in pore volume which may be dueto decreased pore filling from deposition of Ni in catalyst pores.Surface area for the un-doped support differed greatly with Znpercentage, ranging from 138.186 m2/gcat at 0% Zn to 54.315 m2/gcatat 40% Zn. The surface areas shown here eclipse surface area foundby Roh et al. of 40 m2/g for Ni on a Ce/Zr support [23]. For all cat-alysts tested, the measured pore volume varied from 0.024 to0.06 cm3/g, with an average of 0.039 cm3/g. The measured poreradius varied from 14.812 to 19.615 with an average of 16.167 Å.Results of the ICP-OES analysis for Ni content in samples are shownin Table 2. As the oxides could not be dissolved with a single acid,sampling of only the supernatant was undertaken.

Following X-ray powder diffraction tests, the Scherrer equationwas used to estimate the lower limit of Ni particle size which wascalculated as approximately 25 nm for the 10% Ni catalysts and35 nm for the 15% Ni catalysts. The values were estimated assuminga shape factor of 0.94 and no peak widening due to instrumentalerror. These results can be used for comparative analyses todemonstrate that particle sizes are fairly consistent. Roh et al. foundan average particle size of 112 nm for a Ni Ce/Zr catalyst preparedthrough the sol gel method, which is much larger than particle sizesestimated here, and this is supported by the higher surface areavalues for catalysts in this work [23]. The 5% Ni doped speciesproduced very small Ni peaks on the XRD spectra, with only thelargest peak able to be consistently identified. The lack of multiplepeaks for calculations introduces increased error in the Scherrerequation, and therefore these data are not reported. Used catalystswere also analyzed with the Scherrer equation, often on peaks ofelemental Ni rather than NiO, and these peaks show an increase inNi particle size due to sintering. The most common method pro-posed for particle sintering is migration and coalescence [38].

Evaluation of the catalysts by XRD reveals the phases and crystalplanes of the incorporated compounds. Un-doped catalysts inFig. 2a show phases of CeO2 (111) at 28.6�, CeO2 (2 0 0) at 33�, CeO2(2 2 0) at 47.3�, CeO2 (3 1 1) at 56.7� and CeO2 (4 0 0) at 70� [39].These five peaks of CeO2 are also seen in all other XRD spectra.Fig. 2bed features XRD spectra of catalysts with increasing Ni%.With an increase in Ni% there is an increase in NiO peak intensityfor NiO (111) at 37.3�, NiO (2 0 0) at 43.4�, NiO (2 2 0) at 62.8�, andNiO (3 11) at 75.4�. An increase in NiO peak intensity was also seen

with increasing Zn%, particularly the peak at 43.4� where Zn (11 0)is known to have a peak. There were no metallic Ni peaks observedin the XRD spectra of the fresh catalysts. Metallic Ni is found at 44.5,51.8, and 76.5� for Ni (111), Ni (2 0 0), Ni (2 2 0) respectively [27,40].Discrete ZnO peaks are not observed at 31.8, 34.3, and 36.5�. Zr wasobserved as having peaks in correlationwith Ce, with only the peakat 59.6� Zr (2 2 2) able to be distinguished [41].

3.2. Effects of various factors on the performance of catalysts

3.2.1. Composition and temperatureThe effect of temperature, Ni concentration, and Zn concentra-

tion on catalyst reactivity was studied. Temperature was evaluatedin the relevant range for the integration of SRM and WGS of550 �Ce750 �C. The catalyst 0Zn15Ni showed the highest CH4conversion over the temperature range chosen which is shown inFig. 3 along with equilibrium values for SRM from Hegarty et al.[42]. Conversion of CH4 decreases with Zn% which results from thereduction in surface area observed in physisorption tests. Thehighest CH4 conversion was reached at 750 �C with near 80%conversion using 0Zn15Ni. Additionally, it was found that loweringNi concentration leads to higher selectivity as shown in Fig. 4,which is likely due to the presence of excess water in the reactionsystem at lower CH4 conversion. Selectivity increases with Zn%, anddecreases with temperature, which is less apparent for the catalystswith higher Ni% due to a decrease in water available for the WGSreaction. As Fig. 5 shows, the overall highest activity was achievedwith 0Zn15Ni at 750 �C, but it can also be seen that temperature hasa greater effect on activity than the Zn concentration.

WGS shift equilibrium values from Haryanto et al. [43] showconversions of CO ranging from 69% at 500 �C to 54% at 750 �C witha S/C ratio of 1, and from 92% at 500 �C to 76% at 750 �C with a S/Cratio of 3. This demonstrates that the S/C ratio greatly affects the

Fig. 3. CH4 conversion vs. temperature [(a) 5% Ni; (b) 10% Ni; (c) 15% Ni. Catalyst mass: 100 mg; space velocity: 4000 h�1; S/C ratio: 2; reduction time: 2 h. Equilibrium valuesadapted from Ref. [42]].

Fig. 4. CO2 selectivity vs. temperature [(a) 5% Ni; (b) 10% Ni; (c) 15% Ni. Catalyst mass: 100 mg; space velocity: 4000 h�1; S/C ratio: 2; reduction time: 2 h].

T.L. LeValley et al. / Energy 90 (2015) 748e758752

conversion of CO [43]. The equilibrium values were derived using aminimization of Gibbs free energy over supported Ni catalysts witha nonstoichiometric approach under different partial pressures ofeach analysis gas [44]. The WGS conversions obtained with40ZnXNi in this work range from 60% at 500 �C to almost zero at750 �C.

Matsumura and Nakamori tested a 20% Ni on ZrO2 catalystwhich showed comparable conversion around 25% at 500 �C [27].Studies on 10% Ni catalysts have also shown comparable resultswith CH4 conversions of 29.2% for 15Ce85Zr at 600 �C, and 33.8% foralumina at 700 �C [45]. Typically, Ni on alumina catalysts show highactivity for high temperatures, but below 700 �C they lose activity

Fig. 5. Surface plot of H2 activity vs. zinc and temperature [(a) 5% Ni; (b) 10% Ni; (c) 15% Ni. Reaction conditions: catalyst mass: 100 mg; space velocity: 4000 h�1; S/C ratio: 2;reduction time: 2 h].

Fig. 6. (a) Aged catalyst conversion and (b) selectivity vs. temperature [Catalyst mass: 100 mg; space velocity: 4000 h�1; S/C ratio: 2; reduction time: 2 h].

T.L. LeValley et al. / Energy 90 (2015) 748e758 753

Table 3Characteristics of aged catalyst [surface area (SBET); pore volume (Pv); pore radius (Pr); nickel dispersion (DNi); average nickel particle size (dNi); active metal surface area (SAm);turn over frequency (TOF)].

Catalyst SBET (m2/gcat) Pv (cm3/g) Pr (Å) DNi (%) dNi (nm) SAm (m2/gcat) TOF$10�3 (H2/Ni h)

20Zn10Ni 73.135 0.033 14.838 3.472 29.1 2.314 4.231A20Zn10Ni 64.231 0.083 19.172 2.573 39.3 1.714 5.402A20Zn10Ni 70.802 0.031 16.970 2.653 38.1 1.768 6.02

T.L. LeValley et al. / Energy 90 (2015) 748e758754

quickly; however, the conversion for 10% Ni catalysts in this workapproaches 40% at 600 �C. Higher conversion of 52% at 550 �C witha S/C ratio of 3 was achieved with a 17% Ni alumina spinel catalystby Jim�enez-Gonz�alez et al. [24]. This is above the conversion shownin this work, which is approximately 35% for the catalysts withadded zinc. This may be due to increased Ni loading as the activemetal surface area of 30 m2/gcat is much higher than that of thecatalysts examined in this study.

Kusakabe et al. used noblemetal catalysts of 5 wt% Pt, Ru, and Rhto achieve 35.5, 43.2, and 52.4 percent conversion at 600 �Crespectively [45]. The Rh catalyst outperformed the catalysts re-ported here due to the propensity of Rh to decompose methane,while the others show similar or lower conversion. Conversion ofCH4 close to 40% at 550 �C has been found for both 10% Ni Ce/Zr/Laand 1% Rh on Ce/Zr/La catalysts, with an S/C ratio of 3 [46]. Thecatalysts studied in this work show very similar activity at thistemperature. At 500 �C, 1% Rh on 25Ce75ZrO2 achieved 31% CH4conversion which was higher than other precious metals doped at

Fig. 7. TEM images of fresh catalyst [(a) 20Zn

1% and tested at 500 �C, including Pt and Pd. These experimentswere done with an S/C ratio of 2 and had higher space velocitiesthan those examined in this work [26].

3.2.2. Catalyst aging timeAging of precipitates during catalyst synthesis has been shown

to improve stability [47], and the applicability of this method on thecatalysts in this work has been investigated using the catalyst20Zn10Ni. Temperature ramping reactivity experiments for zero,one, and 2 h aged catalysts showed similar conversion. However,the selectivity of the un-aged catalyst was slightly greater than thatof the aged catalysts. These results are shown in Fig. 6a,b.

Measurements using physisorption and chemisorption werecompleted on the aged catalysts to determine if any propertieschange with aging time. Surface area was shown to decreaseslightly after aging for both the 1 h and 2 h case. Dispersion alsodecreased slightly from 3.4% to approximately 2.5%, as well as SAmwhich decreased from 2.3 to 1.7 m2/gcat. The nickel particle size

10Ni; (b) 1A20Zn10Ni; (c) 2A20Zn10Ni].

Fig. 8. XRD patterns for catalysts used in aging comparison (Radiation source: Cu Ka;voltage: 40 kV; amperage: 40 mA; step size: 0.02�/sec; *: NiO; #: CeO2).

T.L. LeValley et al. / Energy 90 (2015) 748e758 755

was shown to increase slightly with aging from 29 nm to 39 nm.Results of these tests are shown in Table 3. Aged catalysts seem tobe more resistant to sintering than the un-aged catalysts.Chemisorption experiments performed on used 1A20Zn10Nishowed an increase in particle size to 123.7 nm, proving that

Fig. 9. SEM images of catalysts with differing aging times [(a) 20

particle sintering is a major cause of catalyst deactivation. Addi-tionally, surface area was shown to decrease after the reactivityexperiments. TOF was calculated at 550 �C using the activity, SAm,and the Ni surface area per atom. This gives TOF in molecules ofH2 per active metal nickel site per hour (H2/Ni$h). The TOFincreased with aging time from 4.2$10�3 H2/Ni$h for 0 h aging to6.0$10�3 H2/Ni$h for 2 h aging.

Improved TOF shown by the aged catalysts are comparable toNi/Al catalysts examined by Nieva et al. at temperatures of500e600 �C, but with higher initial CO2 selectivity [31]. In bothcases, increased CO2 selectivity was observed with decreasedcatalyst activity. The aged catalysts also showed less carbon depo-sition, and it is proposed that this is due to a decrease in the COdisproportionation reaction as the catalysts produced by Nievaet al. catalyzed this reaction prevalently.

TEM images were obtained to examine support size differencesof the aged and un-aged 20Zn10Ni. As support precipitate agingwas the only difference in preparation method, it is expected thatonly the support size would vary with aging. As shown in Fig. 7, asaging time increases more defined mixed oxide support particlesare seen.

Evaluation of the aged catalysts by XRD was performed toelucidate if any changes occurred to the crystal structure duringaging of the precipitates. Shown in Fig. 8 are the XRD spectra of 0, 1,and 2 h aged 20Zn10Ni catalysts. There is little perceptible differ-ence between the spectra except a decreased peak intensity of Niwith aging time. The same crystal phases are present which is notsurprising since the preparation method for these three catalysts isconsistent except for aging time.

Zn10Ni; (b) 1A20Zn10Ni; (c,d) 2A20Zn10Ni. Voltage: 20 kV].

T.L. LeValley et al. / Energy 90 (2015) 748e758756

SEM was completed to determine catalyst morphology, and theresulting experiments show no perceptible change between the 0,1, and 2 h aged samples. Small particles are seen on very largesupports, with support particle sizes ranging from very small tolarger than the 10 micron particles seen in Fig. 9. In the images, thelight particles are nickel oxide and the large dark particles arecerium, zirconium, and zinc mixed oxides produced through co-precipitation. The smaller particles seen in Fig. 9d are of the 2 haged sample and show similar nickel particle sizes to those shownin the TEM images (Fig. 7). It is postulated that much of the activitycomes from the small support particles with close interaction to theNi, while the very large particles give very little activity. SEM im-ages of similar catalysts from Izquierdo et al. show smaller supportand particle sizes, providing greater activity [48].

3.3. Stability

Catalyst stability was evaluated over 48 h at 650 �C and a S/Cratio of 2. It was found that the CH4 conversion decreased rapidlyover the first 12 h and in most cases approached zero within 48 h,but can be improved with precipitate aging. Un-aged 20Zn10Ni lostnearly all of its activity by 30 h, while aged catalysts showed steadyconversion over 48 h. The 2 h aged catalyst decreased to a lowerconversion than the 1 h aged catalyst in its initial deactivation, butwas subsequently more stable. As seen in Table 3, the 1 h agedcatalyst has a significantly larger pore radius and pore volumewhich may reduce pore blocking resulting from carbon depositionor Ni sintering, which can lead to better stability. Very few long-term stability experiments have been performed on low tempera-ture steam reforming catalysts due to their high activity for cata-lyzing carbon deposition through R4 or R5. The momentarydecrease in activity observed in the 2 h aged catalyst near hour 28 is

Fig. 10. Stability of CH4 conversion. [(a) Varying Ni% with constant Zn%; (b) varying Zn% with constant Ni%; (c) varying aging time. Catalyst mass: 100 mg; reaction tem-perature: 650 �C; space velocity: 4000 h�1; S/C ratio: 2].

due to a small amount of water condensation occurring before thereactor which temporarily decreased the available water for thereactions, but soon returned to its steady state value aftercondensation stopped. Matsumura et al. found that during the firsthour activity increased, which is consistent with the findings pre-sented in this work [27]. The conversion over time at 650 �C isshown in Fig. 10.

Increases in CO2 selectivity with time due to deactivation wereobserved with selectivity often approaching 100 percent, largelydue to the high water to CO ratio. The aged catalysts had lower CO2selectivity than any other catalyst, again due to the decrease in H2Oavailable for the WGS reaction. The more stable aged catalysts alsoshowed rapid deactivation over the first 12 h; however, after thisinitial deactivation CO2 selectivity increased and thereforeincreased WGS activity is seen, maintaining hydrogen activity over48 h in the 2A20Zn10Ni case. Results are shown in Fig. 11.

TGA experiments were performed on the catalysts after 48 h onstream to determine the total percentage of deposited carbon. Theresults of these experiments are shown in Fig. 12. As very littledeposited carbon on the catalysts containing Zn is observed, thissuggests sintering of the Ni particles to be the leading cause ofdeactivation. Due to larger Ni particle sizes resulting from thepreparation method and excess steam at low conversion, Ni sin-tering is a large concern even at low temperature. Excess steam isknown to be a significant cause of Ni particle sintering [38]. Smallerparticle sizes with increased dispersion could reduce sinteringproblems. The catalysts prepared with low Zn% showed smallamounts of carbon deposition. 0Zn10Ni was the only catalyst toshow significant carbon deposition of nearly 7 wt%. This is mostlikely due to the large particle size created during preparation and alack of Zn to prevent carbon buildup.

Fig. 11. Stability of CO2 selectivity [(a) varying Ni% with constant Zn%; (b) varying Zn%with constant Ni%; (c) varying aging time. Catalyst mass: 100 mg; reaction tempera-ture: 650 �C; space velocity: 4000 h�1; S/C ratio: 2].

Fig. 12. TGA plots of reference weight vs. temperature for used catalysts after 48 hstability experiments. [(a) Varying Ni% with constant Zn%; (b) varying Zn% with con-stant Ni%; (c) varying aging time. Heating rate: 10�/min].

T.L. LeValley et al. / Energy 90 (2015) 748e758 757

Evaluation of the used catalysts with XRD was completed tocheck for phase changes. Used XRD spectra show reduction of NiOpeaks. Fig. 13a shows that 5% Ni was completely reduced except inthe 40% Zn catalyst where a NiO peak is observed. In Fig. 13b NiOpeaks are present for the catalysts with 10% Ni and higher Zn per-centages. Only the spectra for 10% and 0% Zn show reduced Nispecies. As shown in Fig. 13c, all of the Ni species were reducedcompletely, showing metallic Ni peaks at 44.5, 51.8, and 76.5�

[27,40]. At lower Zn percentages, complete reduction of the Ni is

Fig. 13. XRD spectra for used catalysts [(a) 5% doped Ni; (b) 10% doped Ni; (c) 15%doped Ni. *: NiO, #: CeO2, D: Ni, ,: ZrO2. Radiation source: Cu Ka; voltage: 40 kV;amperage: 40 mA; step size: 0.02�/sec].

found, while at higher Zn ratios, some or all of the NiO remainsunreduced by hydrogen. This could be due to many factors in thehigh Zn catalysts. Zn could act as a barrier to the connection be-tween Ce and Ni, not allowing oxygen transfer to take place. Thiscould result in oxygen remaining with the NiO and not moving intothe Ce lattice replenishing the oxygen that was donated to the re-action. Zn could also act as an oxygen donator allowing Ni to remainin the oxide phase.

4. Conclusion

Integration of steam reforming of methane and water gas shiftusing composite catalysts offers advantages over conventionalhydrogen production, including decreased production costs. For theCe/Zr/Zn/Ni catalysts tested, an increase in Ni content providedhigher activity and stability, while increased Zn improved CO2selectivity and decreased carbon deposition. Increasing Zn contentwas also found to decrease CH4 conversion. Stability was improvedthrough precipitate aging, and by the decreased carbon depositiondue to the addition of Zn. Aging did not have an effect on the initialCH4 conversion, although it did decrease initial CO2 selectivityslightly. After 30 h almost all activity was lost for the un-agedcatalyst 20Zn10Ni, whereas the 2 h aged 2A20Zn10Ni came tosteady state at just below 20% CH4 conversion after 48 h. Thehighest activity observed was 0.10 mol of hydrogen per gram ofcatalyst per hour for the 0Zn15Ni catalyst.

The integration of SRM and WGS can decrease the cost of theproduction of hydrogen from fossil fuels, providing a much neededbridge between renewable energy methods of the future andtraditional methods of today.

Acknowledgments

The authors would like to thank National Science FoundationGK-12 project # 0841298 and the University of Wyoming forfunding this work.

References

[1] Mei Z, Li Y, Fan M, Argyle MD, Tang J. The effects of bimetallic CoeRu nano-particles on Co/RuO2/Al2O3 catalysts for the water gas shift and methanation.Int J Hydrogen Energy 2014;39(27):14808e16.

[2] Kong Y, Jiang G, Fan M, Shen X, Cui S, Russell AG. A new aerogel based CO2adsorbent developed using a simple sol-gel method along with supercriticaldrying. Chem Commun 2014;50(81):12158e61.

[3] Irani M, Fan M, Ismail H, Tuwati A, Dutcher B, Russell AG. Modified nano-sepiolite as an inexpensive support of tetraethylenepentamine for CO2 sorp-tion. Nano Energy 2015;11:235e46.

[4] Chicco G, Stephenson PM. Effectiveness of setting cumulative carbon dioxideemissions reduction targets. Energy 2012;42(1):19e31.

[5] Kothari R, Buddhi D, Sawhney RL. Sources and technology for hydrogenproduction: a review. Int J Glob Energy Issues 2004;21(1e2):154e78.

[6] Peng X, He C, Fan X, Liu Q, Zhang J, Wang H. Photovoltaic devices in hydrogenproduction. Int J Hydrogen Energy 2014;39(26):14166e71.

[7] Mei Z, Li Y, Fan M, Zhao L, Zhao J. Effect of the interactions between Pt speciesand ceria on Pt/ceria catalysts for water gas shift: the XPS studies. Chem Eng J2015;259:293e302.

[8] Song C. Fuel processing for low-temperature and high-temperature fuel cells:challenges, and opportunities for sustainable development in the 21st cen-tury. Catal Today 2002;77(1e2):17e49.

[9] Navarro RM, Pe~na MA, Fierro JLG. Hydrogen production reactions from carbonfeedstocks: fossil fuels and biomass. Chem Rev 2007;107(10):3952e91.

[10] Bartels JR, Pate MB, Olson NK. An economic survey of hydrogen productionfrom conventional and alternative energy sources. Int J Hydrogen Energy2010;35(16):8371e84.

[11] Boyano A, Blanco-Marigorta AM, Morosuk T, Tsatsaronis G. Exergoenvir-onmental analysis of a steam methane reforming process for hydrogen pro-duction. Energy 2011;36(4):2202e14.

[12] Dirksen HA, Riesz CH. Equilibrium in the steam reforming of natural gas. IndEng Chem 1953;45(7):1562e5.

[13] Rostrup-Nielsen JR. Syngas in perspective. Catal Today 2002;71(3e4):243e7.[14] Gao P, Li F, Zhao N, Xiao F, Wei W, Zhong L, et al. Influence of modifier (Mn, La,

Ce, Zr and Y) on the performance of Cu/Zn/Al catalysts via hydrotalcite-like

T.L. LeValley et al. / Energy 90 (2015) 748e758758

precursors for CO2 hydrogenation to methanol. Appl Catal A General2013;468:442e52.

[15] Moon D. Hydrogen production by catalytic reforming of liquid hydrocarbons.Catal Surv Asia 2011;15(1):25e36.

[16] Huang T-J, Huang M-C. Effect of Ni content on hydrogen production via steamreforming of methane over Ni/GDC catalysts. Chem Eng J 2008;145(1):149e53.

[17] Rhodes C, Hutchings GJ, Ward AM. Water-gas shift reaction: finding themechanistic boundary. Catal Today 1995;23(1):43e58.

[18] Chen W-H, Hsieh T-C, Jiang TL. An experimental study on carbon monoxideconversion and hydrogen generation from water gas shift reaction. EnergyConvers Manag 2008;49(10):2801e8.

[19] Kodama T, Shimizu T, Satoh T, Shimizu KI. Stepwise production of CO-richsyngas and hydrogen via methane reforming by a WO3-redox catalyst. En-ergy 2003;28(11):1055e68.

[20] Liu Z-W, Jun K-W, Roh H-S, Park S-E. Hydrogen production for fuel cellsthrough methane reforming at low temperatures. J Power Sources2002;111(2):283e7.

[21] Chin Y-H, King DL, Roh H-S, Wang Y, Heald SM. Structure and reactivity in-vestigations on supported bimetallic AuNi catalysts used for hydrocarbonsteam reforming. J Catal 2006;244(2):153e62.

[22] Dan M, Mihet M, Biris A, Marginean P, Almasan V, Borodi G, et al. Supportednickel catalysts for low temperature methane steam reforming: comparisonbetween metal additives and support modification. React Kinet Mech Catal2012;105(1):173e93.

[23] Roh H-S, Jun K-W, Dong W-S, Chang J-S, Park S-E, Joe Y-I. Highly active andstable Ni/CeeZrO2 catalyst for H2 production from methane. J Mol Catal AChem 2002;181(1e2):137e42.

[24] Jim�enez-Gonz�alez C, Boukha Z, de Rivas B, Gonz�alez-Velasco JR, Guti�errez-Ortiz JI, L�opez-Fonseca R. Behavior of coprecipitated NiAl2O4/Al2O3 catalystsfor low-temperature methane steam reforming. Energy Fuels 2014;28(11):7109e21.

[25] Soria MA, Mateos-Pedrero C, Marín P, Ord�o~nez S, Guerrero-Ruiz A,Rodríguez-Ramos I. Kinetic analysis of the Ru/SiO2-catalyzed low tem-perature methane steam reforming. Appl Catal A General 2012;413e414:366e74.

[26] Sekine Y, Haraguchi M, Matsukata M, Kikuchi E. Low temperature steamreforming of methane over metal catalyst supported on CexZr1�xO2 in anelectric field. Catal Today 2011;171(1):116e25.

[27] Matsumura Y, Nakamori T. Steam reforming of methane over nickel cata-lysts at low reaction temperature. Appl Catal A General 2004;258(1):107e14.

[28] Laosiripojana N, Assabumrungrat S. Methane steam reforming over Ni/Ce-ZrO2 catalyst: Influences of Ce-ZrO2 support on reactivity, resistance towardcarbon formation, and intrinsic reaction kinetics. Appl Catal A General2005;290(1e2):200e11.

[29] Jing QS, Zheng XM. Combined catalytic partial oxidation and CO2 reforming ofmethane over ZrO2-modified Ni/SiO2 catalysts using fluidized-bed reactor.Energy 2006;31(12):2184e92.

[30] LeValley TL, Richard AR, Fan M. The progress in water gas shift and steamreforming hydrogen production technologies e a review. Int J Hydrogen En-ergy 2014;39(30):16983e7000.

[31] Nieva MA, Villaverde MM, Monz�on A, Garetto TF, Marchi AJ. Steam-methanereforming at low temperature on nickel-based catalysts. Chem Eng J2014;235:158e66.

[32] Halabi MH, de Croon MHJM, van der Schaaf J, Cobden PD, Schouten JC. Lowtemperature catalytic methane steam reforming over ceriaezirconia sup-ported rhodium. Appl Catal A General 2010;389(1e2):68e79.

[33] Christensen KO, Chen D, Lødeng R, Holmen A. Effect of supports and Ni crystalsize on carbon formation and sintering during steam methane reforming.Appl Catal A General 2006;314(1):9e22.

[34] Rostrup-Nielsen JR, Sehested J. Whisker carbon revisited. Stud Surf Sci Catal2001;139:1e12.

[35] Roh HS, Eum IH, Jeong DW. Low temperature steam reforming of methaneover Ni-Ce(1�x)Zr(x)O2 catalysts under severe conditions. Renew Energy2012;42:212e6.

[36] Hori CE, Permana H, Ng KYS, Brenner A, More K, Rahmoeller KM, et al.Thermal stability of oxygen storage properties in a mixed CeO2-ZrO2 system.Appl Catal B Environ 1998;16(2):105e17.

[37] DongWS, Roh HS, Jun KW, Park SE, Oh YS. Methane reforming over Ni/Ce-ZrO2catalysts: effect of nickel content. Appl Catal A General 2002;226(1e2):63e72.

[38] Sehested J, Gelten JAP, Remediakis IN, Bengaard H, Nørskov JK. Sintering ofnickel steam-reforming catalysts: effects of temperature and steam andhydrogen pressures. J Catal 2004;223(2):432e43.

[39] Kümmerle EA, Heger G. The structures of CeCe2O3þd, Ce7O12, and Ce11O20.J Solid State Chem 1999;147(2):485e500.

[40] Arcotumapathy V, Vo D-VN, Chesterfield D, Tin CT, Siahvashi A, Lucien FP,et al. Catalyst design for methane steam reforming. Appl Catal A General2014;479:87e102.

[41] Parvulescu VI, Bonnemann H, Parvulescu V, Endruschat U, Rufinska A,Lehmann CW, et al. Preparation and characterisation of mesoporous zirco-nium oxide. Appl Catal A General 2001;214(2):273e87.

[42] Hegarty MES, O'Connor AM, Ross JRH. Syngas production from natural gasusing ZrO2-supported metals. Catal Today 1998;42(3):225e32.

[43] Haryanto A, Fernando SD, To SDF, Steele PH, Pordesimo L, Adhikari S.Hydrogen production through the water�gas shift reaction: thermodynamicequilibrium versus experimental results over supported Ni catalysts. EnergyFuels 2009;23(6):3097e102.

[44] Adhikari S, Fernando S, Haryanto AA. Comparative thermodynamic andexperimental analysis on hydrogen production by steam reforming of glyc-erin. Energy Fuels 2007;21(4):2306e10.

[45] Kusakabe K, Sotowa K-I, Eda T, Iwamoto Y. Methane steam reforming overCeeZrO2-supported noble metal catalysts at low temperature. Fuel ProcessTechnol 2004;86(3):319e26.

[46] Angeli SD, Pilitsis FG, Lemonidou AA. Methane steam reforming at low tem-perature: effect of light alkanes' presence on coke formation. Catal Today2015;242(Part A):119e28.

[47] Kniep BL, Girgsdies F, Ressler T. Effect of precipitate aging on the micro-structural characteristics of Cu/ZnO catalysts for methanol steam reforming.J Catal 2005;236(1):34e44.

[48] Izquierdo U, Barrio VL, Cambra JF, Requies J, Güemez MB, Arias PL, et al.Hydrogen production from methane and natural gas steam reforming inconventional and microreactor reaction systems. Int J Hydrogen Energy2012;37(8):7026e33.