Steam reforming of kerosene on Ru/Al2O3 catalyst to yieldhydrogen

Takashi Suzuki*, Hiko-ichi Iwanami, Tomohiro Yoshinari

New Products and Technology Laboratory, Cosmo Research Institute, 1134-2, Gongendo, Satte, Saitama 340-0193, Japan

Abstract

Highly dispersed Ru/Al2O3 catalyst can be obtained by using ruthenium trichloride and aqueous ammonia in thecatalyst preparation. The dispersion of Ru on the catalyst was improved 2 fold, as much as that on the catalyst

prepared by conventional impregnating method. When the ammonia treated catalyst was subjected to the steamreforming of hydrodesulfurized kerosene with 0.1 ppm of sulfur, the kerosene was converted completely intohydrogen enriched gases for several hours at lower S/C ratio (S/C=3.5) at 8008C. When a ceria was doped in thecatalyst system, the sulfur resistance was improved dramatically and as a result the hydrogen production from the

hydrodesulfurized kerosene can be continued after 8000 h by using Ru/CeO2±Al2O3 catalyst with highperformance. # 1999 International Association for Hydrogen Energy. Published by Elsevier Science Ltd. All rightsreserved.

1. Introduction

Recently, production of deeply desulfurized fuels,fuel cell system, hydrogen engine system, CO2 removal

technique, and so on, are important technologies devel-oped to protect the atmospheric environment. It couldbe considered that hydrogen is not only an importantsource in the present industries but also in the new

technologies as mentioned above.Investigations for hydrogen production have been

carried out by several approaches, for example, steam

reforming [1±4], electrochemical [5], photochemical[6,7], biological [8,9] and thermochemical methods [10].Steam reforming is one of the least expensive hydrogen

production methods at the present time [1].

In the present steam reforming process yielding

hydrogen, natural gas (methane) or naphtha (lighterthan C6 hydrocarbons) and steam are introduced ontoa Ni/Al2O3-based catalyst system. Using a natural gas

or a naphtha as a hydrocarbon source, the facility forhydrogen production should be constructed near topipe-lines (natural gas), or near to crude oil re®neries(naphtha), as the location of the hydrogen plant is lim-

ited by the hydrocarbon sources. In contrast to this, ifan easily and safely transportable source, such as kero-sene, can be used for the steam reforming, the hydro-

gen production could be carried out everywhere.Therefore a silent electric generation system, in con-junction with fuel cell, for home, hospital, outdoor use

etc would be realized.However, much less attention has been paid to the

steam reforming of the middle distillate hydrocar-bonsÐheavier than C6 hydrocarbonsÐbecause the

carbon deposition takes place preferentially on the cat-

International Journal of Hydrogen Energy 25 (2000) 119±126

0360-3199/00/$20.00 # 1999 International Association for Hydrogen Energy. Published by Elsevier Science Ltd. All rights reserved.

PII: S0360-3199(99 )00014-2

* Corresponding author. Tel.: +81-480-42-2211; fax: +81-

480-42-3790.

alyst when the middle distillate was streamed on it.Carbon deposition is especially dominant in the sup-plying hydrocarbons including aromatics. In addition,

the sulfur compounds cannot be removed completelyfrom aromatic sulfur compounds, such as 2,6-dimethyl-dibenzothiophene, due to structural hin-

drance [11]. Therefore the middle distillates includingaromatics such as kerosene, should not be used for thesteam reforming systems at the present time.Recently, it is reported by Okada et al., that a ruthe-

nium-based catalyst is very e�ective in preventing thecarbon deposition during the steam reforming of lighthydrocarbons, but that the catalyst system is readily

poisoned by sulfur compounds [12,17,18]. Moreover,they stated that the poisoned ruthenium catalyst bysulfur compounds loses the characteristic to prevent

the carbon deposition [12,17,18].Iwanami et al. demonstrated that a Ni±ZnO catalyst

(CDSC-3 catalyst), which was prepared carefully by

co-precipitation technique, shows highly hydrodesul-furization (HDS) performance for the middle distillatehydrocarbons such as kerosene under mild conditions[13]. If the sulfur resistance can be improved in the

ruthenium-based catalyst, the hydrogen productionfrom kerosene will be realized. Hence, these facts ledus to planning the exploratory process for hydrogen

production by the steam reforming of kerosene asdescribed in Scheme 1.In this work, highly dispersed supported-ruthenium

catalyst was obtained by treating the supported ruthe-nium with aqueous ammonia. The catalyst showedhigh activity for the steam reforming of kerosene and

that sulfur resistance was improved considerably byadding CeO2 to the catalyst. When the CeO2-dopedcatalyst was subjected to the sustained run under feed-ing desulfurized kerosene by CDSC-3 catalyst, hydro-

gen was stably generated for 8000 h (ca 1 year). Thismay be the ®rst case, at least in our country, for thelong term production of hydrogen by steam reforming

of kerosene. It will be realized that the hydrogen pro-duction using kerosene has the potential for commer-cial application.

2. Experimental

2.1. Catalyst preparation

2.1.1. Ru/Al2O3 catalyst (conventional)A N748 alumina (3 mm f � 3 mm (H), cylindrical

form) purchased from Nikki Co. was used as a sup-port. The alumina was calcined at 9008C for 3 h in airand the speci®c surface area was approximately 200 m2/g. The calcined alumina was immersed into an aqueous

solution of ruthenium trichloride, RuCl3�nH2O(Mitsuwa Chemical Co.,) to contain 2.0 wt% of ruthe-nium and the resulting material was dried in an electric

oven maintained at 1058C for 3 h. The specimen wasreduced under hydrogen stream with gas hourly spacevelocity (GHSV)=1000 under 6008C for 3 h prior to

the steam reforming reaction at 8 kg/cm2. The dis-persion of Ru on that catalyst was around 32%. Thecatalyst was denoted as CRI-089.

2.1.2. Ru/Al2O3 catalyst (dispersed)The calcined alumina was immersed into an aqueous

solution of RuCl3�nH2O containing 2.0 wt% of ruthe-

nium (the same as above). It was dried in an oven at1058C for 3 hours. The resulting material was soakedin aqueous ammonia (7 N) for 2 h at 308C with stir-

ring. Then the sample was rinsed by deionized waterto remove excess ammonia and ammonium chloride. Itis deduced that the ruthenium chloride would be chan-

ged into ruthenium hydroxide during soaking in the al-kaline solution. Finally the specimen was dried in anoven overnight at 40±608C to prevent oxidation of the

catalyst. The sample was activated by reduction underthe same conditions as above. The dispersion of Ruwas around 65%. The catalyst was denoted as a CRI-101.

2.1.3. CeO2, Y2O3 and La2O3 doped catalystA catalyst support was prepared by dry mixing of

the alumina powder (Nikki N-748) and one of theoxide powders selected from Y2O3, La2O3 and CeO2.The amount of the additive was set at 20 wt% relative

Scheme 1. Concept of exploratory process for generation of hydrogen by steam reforming of kerosene.

T. Suzuki et al. / International Journal of Hydrogen Energy 25 (2000) 119±126120

to the catalyst weight. The powder mixture waspressed in cylindrical form by using a FK-1 pelletizer(Systems Engineering Co.) with a pressure of 16 ton/

cm2. Finally the support was calcined in air at 9008Cfor 3 h and the speci®c surface was around 160±180 m2/g. Ruthenium was loaded on the support by

the analogous procedure as described in CRI-101 prep-aration, above. The loading rate of ruthenium was thesame as in CRI-089 and CRI-101 preparations, above.

The dispersion of Ru was almost the same as CRI-101.For example, hereinafter, CeO2, Y2O3, and La2O3

doped catalyst was denoted as CRI-101CE, CRI-101Y,CRI-101LA, respectively.

2.2. Catalyst characterization

Speci®c surface area (BET (Braunauer-Emmet-Tailor)) was measured by means of a Belsorp 28instrument (Bel Japan Co.) using nitrogen as a prove

molecule. Dispersion of Ru was calculated by adsorp-tion amount of CO on the catalyst. Specimen (0.5 g)was set in the quartz U-tube and it was reduced withH2 stream (50 ml/min) at 6008C for 3 h. Then the spe-

cimen was then cooled from 6008C to 258C under¯owing He. After the pretreatment, CO pulse (0.2 ml)was streamed to the specimen at 258C. The number of

CO pulse was repeated until the adsorption was satu-rated. The CO used was a highly pure gas having99.99%, purchased from Takachiho Chemical Co. The

adsorption experiment was carried out by using anOhkura 6850 adsorption instrument (Ohkura RikenCo.). The connection diagram of the instrument is

described in Scheme 2. The dispersion of Ru wasdetermined by Eq. 1 which is reported in ref. [14].

Dispersion �%� � �CO�a��=�Ru� � 100 �1�

where [Ru] is amount of loaded Ru atom, [CO(a)] isadsorbed amount of CO molecule.

The loading amount of Ru was analyzed by meansof induced coupled plasma analysis ICP.

2.3. Preparation of the desulfurized kerosene

A JIS-1 (JIS; Japanese Industrial Standard) grade of

kerosene with a sulfur of ca 30±55 ppm was furtherdesulfurized to contain less than 0.1 ppm of the sulfurcontent by using a Cosmo-HDS catalyst (CDSC-3) [6].

The JIS-1 kerosene and hydrogen were co-introducedonto a CDSC-3 catalyst (6 ml (4.8 g)) at 2808C under8 kg/cm2 with volumetric ratio of H2/Oil=200. Thefeeding rate of JIS-1 kerosene was set at 6 ml/h as a

liquid. Further handling procedure of HDS and chemi-cal and physical properties of CDSC-3 was describedelsewhere [13]. Hereinafter, the desulfurized JIS-1 kero-

sene was denoted as HD-kerosene for convenience.

2.4. Steam reforming reaction

The steam reforming of kerosene was performed bymeans of a ®xed bed ¯ow reactor system described in

Scheme 3. 10 ml (9.7±9.8 g) of the reforming catalystwas set in the ®xed bed reactor and the catalyst wasactivated by reduction with H2 as already stated. TheHD-kerosene was fed on the catalyst with steam (S/

C=3.5) at 8 kg/cm2 at 8008C with the feeding rate of10 ml/h. The reaction temperature was set at 8008Cwhich was regulated within258C at the bottom of cat-

alyst bed. Product distribution was determined bymeans of a gas chromatograph (GC9A, Shimadzu Co.)equipped with TCD and FID. The separation column

(3 mm ID � 3 m) used was packed with Unibeads-A(active carbon, 60±80 mesh, GL science Co.), whichwas maintained at 1058C. Conversion of kerosene wascalculated by dividing the sum of carbon in gaseous

products by the carbon in kerosene (feed) as in Eq. 2.

Conversion�%�

� 100� �C� in gaseous product=�C� in feed

� 100� ��C� in CO� �C� in CO2

� �C� in CH4�=�C� in kerosene �2�

where [C] represents the amount of carbon atom.

3. Results and discussion

3.1. Steam reforming of HD-kerosene

In order to con®rm the conventional Ru/Al2O3

Scheme 2. Connection ¯ow diagram of CO pulse adsorption

instrument, where P, pressure gauge; FC, ¯ow controller;

6DV, 6-directions valve; ST, sampling tube (0.2 ml); and V,

stop valve.

T. Suzuki et al. / International Journal of Hydrogen Energy 25 (2000) 119±126 121

(CRI-089) and the dispersed Ru/Al2O3 (CRI-101) cata-

lyst, these catalysts were subjected to the steam reform-

ing of HD-kerosene (sulfur=0.1 ppm, typical

formula=C10H22) at 8008C. Fig. 1 shows the conver-

sion of kerosene in the steam reforming as a function

of the reaction time. The conversion of kerosene was

maintained around 96±99% on both CRI-089 and

CRI-101 by 40 h (initial stage). The composition of

hydrogen in the initial stage of the reaction on CRI-

089 and CRI-101 was 63.6±64.2% and 70.1±70.2%, re-

Scheme 3. Connection ¯ow diagram of apparatus for steam reforming of kerosene, where FC, ¯ow contoroller; F, ¯ow measuring

cylinder; P, feeding pump; PG, pressure gauge; PMV, pressure maintaining valve; S, separator; RFM, rotary ¯ow meter.

Table 1

Product distributions in steam reforming of HD-kerosene on Ru/Alumina and Ni/Alumina catalysts

Catalysta Chemical composition of catalyst/wt% Product distributions/% Conv./%

Ni Ru SiO2 Al2O3 H2 CO CH4 CO2

CRI-089 Ð 2.0 Ð balance 7 h 63.6 14.3 8.1 14.0 98

24 h 64.2 14.7 7.0 14.1 97

CRI-101 Ð 2.0 Ð balance 7 h 70.1 14.1 4.0 11.8 99.5

24 h 70.2 14.0 4.1 11.7 99.5

Commercial 21.0 Ð 16.0 balance 7 h 60.2 17.5 10.6 11.7 99.5

24 h 60.8 17.3 10.5 11.4 72.3

a Catalyst, 10 ml (9.7±9.8 g); reaction temperature, 8008C; LHSV, 1.0 hÿ1, S/C=3.5, pressure, 8.0 kg/cm2, G.

T. Suzuki et al. / International Journal of Hydrogen Energy 25 (2000) 119±126122

spectively as shown in Table 1. This result shows thatthe hydrogen production was preferable in the highly

dispersed catalyst system.In contrast to the supported ruthenium catalyst sys-

tems, when the commercially available Ni-based cata-

lyst for the steam reforming of naphtha was examinedin the reaction with HD-kerosene, the conversion of

kerosene decreased to 72.3% after 24 h and thereforming reaction could not be made continuously byincreasing the pressure which would be caused by car-

bon blocking. It has been accepted that a Ni-based cat-alyst such as Ni±Alumina catalyst shows a resistance

for sulfur poisoning, hence Ni is used for one of theactive metals in the HDS catalyst systems [13,15].Therefore it might be suggested that the deactivation

of the Ni-based catalyst in our experiment was predo-minantly caused by carbon deposition induced by aro-matic compounds in kerosene.

In the case using the Ni-based catalyst, the compo-sition of hydrogen was 60.8%, which was ca 10%

lower than that in the case using CRI-101. Methanewas increased 2.6 fold, as much as that in the caseusing CRI-101, while the composition of CO and CO2

was maintained at 17.3±17.5% and 11.4±11.7%, re-spectively (see Table 1). These observations in the Ni

catalyst imply that a part of the produced methanemight be caused by hydrogenation of surface carbonspecies from kerosene. Thus, the Ru/Al2O3 catalyst is

more suitable to prevent the carbon deposition duringthe hydrogen production from HD-kerosene as well as

that in the steam reforming of light hydrocarbons[12,17,18].

The conversion on CRI-089 decreased and ®nally

reached 90% while that on CRI-101 was 95% at 120 h.

In order to consider on the amount of active site, the

dispersion of Ru was measured on CRI-101 and CRI-089. The dispersion of CRI-089 and that of CRI-101

was 32% and 65%, respectively. In the case of the dis-

persed catalyst (CRI-101), it is conjectured that theconsiderable amount of active site remained through

the steam reforming after 120 h. Okada et al. demon-

strated that the Ru catalyst is readily deactivated by

Fig. 2. Steam reforming of commercial grade kerosene (S=51

ppm) on CeO2, La2O3, MgO, and BaO added Ru/Al2O3 cata-

lyst at 8008C. Reaction conditions were the same as in Fig. 1

except the sulfur content in the feed. Where *, CRI-101CE;

R, CRI-101LA; W, CRI-101Y; and T, CRI-101.

Fig. 1. Steam reforming of desulfurized kerosene (S=0.1

ppm) on CRI-101, CRI-089 and commercial grade Ni based

catalyst at 8008C. Catalyst=10 ml (9.7±9.8 g), S/C=3.5, feed-

ing rate of kerosene=10 ml/h, pressure=8 kg/cm2 where W,

CRI-101; Q, CRI-089; and R commercial grade Ni based cat-

alyst. These catalysts were reduced before use. Reduction was

carried out at 6008C with 8 kg/cm2 for 3 h. Hydrogen was fed

at GHSV=1000 vol/vol/h.

Fig. 3. Adsorption amount of CO after dosing hydrogen sul-

®de on CRI-101 and CRI-101CE at 508C. Specimen; 0.1 g, re-

duction was made at 6008C with GHSV=1000 for 3 h under

atmospheric pressure. Concentration of H2S:100 ppm, where

Q, CRI-101; and W; CRI-101CE.

T. Suzuki et al. / International Journal of Hydrogen Energy 25 (2000) 119±126 123

Fig. 4. (a) Conversion of the HD-kerosene in the reforming on CRI-101CE at 8008C. Reaction conditions were the same as in Fig.

1. (b) Composition of H2, CO, CH4 and CO2 in the HD-kerosene reforming on CRI-101CE at 8008C. Reaction conditions were

the same as in Fig. 1 where w, hydrogen; r, carbon monoxide; *, methane; and P carbon dioxide.

T. Suzuki et al. / International Journal of Hydrogen Energy 25 (2000) 119±126124

the slight amount of sulfur compounds and that thecarbon deposition takes place dominantly on the sul-

furized Ru site [12,17,18]. It is deduced that a part ofactive sites might be deactivated by 0.1 ppm levels ofsulfur compounds or by carbon deposition which

might be induced by sulfur compounds. It is, therefore,very important to improve a resistance for sulfur poi-soning of CRI-101 even in our exploratory process

such as the case using the HD-kerosene.

3.2. Preventing the deactivation by sulfur

In order to elucidate the improvement e�ect of thesulfur resistance on a Ru/Al2O3 catalyst modi®ed by

adding one of the oxides selected from CeO2, La2O3,and Y2O3 in the steam reforming of kerosene, thesecatalysts were subjected to the reaction under feeding a

JIS-1 grade kerosene with 50±52 wt ppm of sulfur. Theconditions of the steam reforming, except for the sulfurcontent in the kerosene, were the same as in the pre-vious section.

As shown in Fig. 2, the conversions on CRI-101,CRI-101CE, CRI-101LA and CRI-101Y in the ®rststage of the reaction were the same around 98±99%.

In the case of CRI-101, the conversion was decreasedto 60% after 25 h whereas that on CRI-101Y, CRI-101LA, and CRI-101-CE catalyst was 66.0%, 80.1%,

and 85.5%, respectively. Hence it is considered thatthese additives might be e�ective to restrain the deacti-vation by sulfur compounds and that CeO2 actedexceptionally as an inhibitor for the deactivation in the

steam reforming of the HD-kerosene.In order to clarify the role of additives, EPMA, and

XPS analysis were done to obtain distribution of a sul-

fur in the depth of the catalyst or electronic state ofRu, unfortunately, signi®cant results could not beobtained. It has been accepted generally that the

adsorption of CO may re¯ect the number of the site,in the form of fully reduced state in the noble-metal-supported catalyst system [14,16]. It is expected that

the adsorption experiment of CO after deactivated bysulfur compounds provides some valuable information.Fig. 3 shows adsorbed amount of CO as a function ofdose of hydrogen sul®de onto CRI-101 and CRI-

101CE. As seen in the ®gure, the adsorbed amount ofCO on CRI-101CE was still higher than that in thecase of CRI-101 after dosing of hydrogen sul®de for

60±90 min. Therefore, it is considered that theadsorbed amount of CO re¯ected the remained activesite after deactivated by H2S, and as a result the ac-

tivity loss was suppressed on CRI-101CE.

3.3. Long life test of CRI-101CE

These results led us to a preliminary long life testfor the steam reforming reaction of HD-kerosene on

the CRI-101CE catalyst. The reaction conditions werethe same as stated in the above section and these were

almost analogous to the conditions our exploratoryplant are now planning. As shown in Fig. 4a and b,the conversion of kerosene and the composition of

hydrogen was almost 98±99% and 65±70%, respect-ively, during the long run test. Here one can point outthat the composition of the products was somewhat

¯uctuated. Water and the HD-kerosene were fed intothe vessel by suction pumps thereby slightly alteringthe feeding ratio during the long life test. Hence the

composition has small latitude as mentioned.Thus the long sustained run of hydrogen production

using HD-kerosene was successfully achieved on theCRI-101CE catalyst (Ru/CeO2±Al2O3). So far the light

hydrocarbonsÐlighter than C6Ðshould be used forthe steam reforming to produce hydrogen, however,this result will throw light on the hydrogen production

which uses safe and easily transportable middle-lightoil such as kerosene.

Acknowledgements

Authors indebted to Mr Osamu Iwamoto for histechnical assistance in the long life test.

References

[1] Saito M. Methanol synthesis by catalytic hydrogenation

of carbon dioxide. Shokubai (Catalysis) 1993;35:485±91.

[2] Kikuchi E, Uemiya S, Matsuda T. Hydrogen production

of methane steam reforming assisted by use of membrane

reactor. Stud Surf Sci Catal 1991;61:509±15.

[3] Steinberg M, Cheng CH. Modern and prospective tech-

nologies for hydrogen production from fossil fuels. Int J

Hydrogen Energy 1989;14:797±820.

[4] Imai J, Tawara K, Iwanami H. Steam reforming cata-

lysts and hydrogen manufacture. Jpn Kokai Tokkyo

Koho JP06031169 Chem Abstr 1994;121:13125.

[5] Compton RG, Cole-Hamilton DJ. The electrochemistry

of [PtH(PEt3)3]+ inverted and ampli®ed cyclic voltam-

metric waves and catalytic hydrogen production at a

mercury electrode. J Chem Soc, Dalton Trans

1995;6:1225±9.

[6] Koenigstein C. Some aspects of photochemical systems

for direct light induced hydrogen production. J

Photochem Photobiol, A 1995;90:141±52.

[7] Willner I, Steinberger B. Solar hydrogen production

through photobiological, photochemical and photoelec-

trochemical assemblies. Int J Hydrogen Energy

1988;13:593±604.

[8] Yamada A, Takano H, Burgess GJ, Matsunaga T.

Enhanced hydrogen production by a marine photosyn-

thetic bacterium, Rhodobacter marinus immobilized onto

light di�using optical ®bers. J Mar Biotechnol 1996;4:23±

7.

T. Suzuki et al. / International Journal of Hydrogen Energy 25 (2000) 119±126 125

[9] Fissler J, Kohring GW, Gi�horn F. Enhanced hydrogen

production from aromatic acid by immobilized cells of

Rhodopseudomon palustris. Appl Microbiol Biotechnol

1995;44:43±6.

[10] Abdel-Aal HK. Coupling of thermochemical cycles with

non-catalytic partial oxidation of heavy residues can

boost the economy of hydrogen production. Int J

Hydrogen Energy 1985;10:333.

[11] Isoda T, Nagao S, Korai Y, Mochida I. Acid assisted

desulfurization of 4,6-dimethyldibenzothiophene and its

reaction network over mixture of Ni-HY zeolite and

CoMo/Al2O3. J Jpn Petrol Inst 1998;41:22±8.

[12] Okada O, Tabata T, Masuda M, Matsui H. Catalyst

deactivation by sulfur poisoning. Shokubai (Catalysis)

1993;35:224±30.

[13] Tawara K, Iwanami H, Nishimura T. Deep desulfuriza-

tion of middle light oils. Jpn Kokai Tokkyo Koho

JP05070780 Chem Abstr 1993;119:76177.

[14] Tsumaki H, Hirofumi S. Measurement of metal dis-

persion in Cl free supported reference noble metal cata-

lysts by CO-pulse method. Nippon Kagaku Kaishi (J

Chem Soc Jpn) 1989;12:1990±8.

[15] Petroleum Re®ning Process (sekiyu seisei process) 1996

Japanese Petroleum Institute (JPI), Saiwai Shobo Publ.

Co., Tokyo.

[16] Narita T, Miura H, Ohira M, Sugiyama K, Matsuda T,

Gonzalez RD. The e�ect of reduction temperature on

the chemisorptive properties of Ru/Al2O3: e�ect of chlor-

ine. Appl Catal 1987;32:185±90.

[17] Okada O, Ipponmatsu M, Masuda M, Takami S. A

study of sulfur poisoning on Ru/Al2O3 methanation cata-

lyst. Nenryokyokaishi (J Jpn Fuel Sci and Technol)

1989;68:39±44.

[18] Okada O, Ipponmatsu M, Masuda M, Sadamori H,

Takimi S. A study of sulfur poisoning on Ru/Al2O3

steam reforming catalyst. Nenryokyokaishi (J Jpn Fuel

Sci and Technol) 1989;68:124±9.

T. Suzuki et al. / International Journal of Hydrogen Energy 25 (2000) 119±126126