Chemical Engineering Journal 238 (2014) 172–177

Contents lists available at SciVerse ScienceDirect

Chemical Engineering Journal

journal homepage: www.elsevier .com/locate /ce j

New catalysts and reactor designs for the hydrogen economy q

1385-8947/$ - see front matter � 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.cej.2013.07.004

q Presented at: XX International conference on Chemical ReactorsCHEMREACTORS-20, 2–6 December 2012, Luxembourg.⇑ Tel.: +1 732 829 7127.

E-mail address: RF2182@Columbia.edu

Robert J. Farrauto ⇑Earth and Environmental Engineering Department, Columbia University, New York, NY 10027, United States

h i g h l i g h t s

� New catalyst and reactor designs forhydrogen generation of fuel cells.� New precious metal catalysts on

monoliths are used for hydrogengeneration.� Precious metals costs are mitigated

by savings due to reduced capitalcost.� Applications are for combined heat

and power.

g r a p h i c a l a b s t r a c t

a r t i c l e i n f o

Article history:Available online 10 July 2013

Keywords:Reforming hydrocarbon fuelsHydrogen generationPrecious metal washcoated catalystsMonolithic structuresHeat exchangers

a b s t r a c t

New catalyst and reactor designs are necessary to meet the expanding use of low temperature hydrogenbased fuel cells. Residential and commercial power generation requires the reformer integrated to thefuel cell must be sufficiently small given that space is often a premium in many of these applications. Pre-cious metal washcoated monolithic structures, similar to those successfully used since 1975 in automo-bile catalytic converters [1], provide high activity per unit reactor volume, low pressure drop and greaterstructural stability than traditional base metal catalysts in packed beds and thus are well suited for dis-tributed hydrogen applications, New precious metal catalysts, with high activity densities, have been for-mulated for hydrogen generation since traditional base metal oxides have much lower activities per unitmass and thus are not sufficiently active when used in limited amounts as washcoats. The higher cost ofprecious metals is mitigated by savings due to reduced system size and existing metal recycling opera-tions. Catalytic fuel processing, of infrastructure fuels (e.g. natural gas and LPG), are being reformed forthe fuel cell to power homes, commercial and residential buildings, schools, hospitals. Such systems mustoperate safely and reliably in the user facilities while unattended. This brief review will illustrate theapplication of automotive washcoating technology to reformers for distributed hydrogen applications.

� 2013 Elsevier B.V. All rights reserved.

1. Introduction

The concept of distributed power allows electricity and heat tobe generated onsite independent of a central power plant. Fuelcells supply heat and electricity without toxic emissions and areprime candidates for the emerging hydrogen economy. Their highefficiency (not bound by heat engine thermodynamic cycles) com-

pared to traditional combustion technologies results in lower CO2

emissions. Since low temperature fuel cells require hydrogen-richgases for the anode, reformers using readily available infrastruc-ture fuels such as natural gas are needed. Designs must be compactgiven the premium of space available in many applications, havelower pressure drop (natural gas is provided to sites at atmo-spheric pressure) and greater structural stability to resist attrition.They must also generate H2 at a rate consistent with varying powerdemands. Large scale hydrogen plants cannot simply be reduced insize to meet the economic, safety, and frequent duty cycle require-ments for fuel cell applications, hydrogen fueling stations, andsmall scale industrial uses such as hydrogenation reactions, gas

Fig. 1. Traditional packed bed of catalysts for high pressure H2 generation withHDS, steam reforming, water gas shift, pressure swing adsorption (PSA) and tail gasoxidation.

Fig. 2. Schematic demonstrating heat transfer resistance and decreasing SR ratetowards the center of the bed.

R.J. Farrauto / Chemical Engineering Journal 238 (2014) 172–177 173

turbine cooling, metal processing, etc. [2–4]. Consequently, there isa need to completely reassess how hydrogen can be made for dis-tributed applications. This review presents some of the technolog-ical advantages of precious metal washcoated monoliths overtraditional base metal pellet (or particulate) catalysts and packedbed reformers for reforming hydrocarbons for the generation ofdistributed hydrogen.

2. Large scale hydrogen generation

Hydrogen is generated in chemical plants by reforming sulfurcontaining hydrocarbon fuels such as natural gas [5]. Since sulfuris a poison to downstream catalysts it is removed by the catalytichydrodesulfurization (HDS) shown as reactions (1) and (2). Reac-tion (1) is catalytic converting sulfur compounds to H2S using aCo, Mo/Al2O3 catalyst at 350 �C, 15 bar at a space velocity (SV)�1000 h�1. Reaction (2) is adsorption of the H2S in a bed of ZnOat 350 �C.

H2 þ R � S! H2Sþ R �H ð1Þ

H2Sþ ZnO! ZnSþH2O ð2Þ

Desulfurization is followed by steam reforming which convertsmethane (the major component in natural gas) to H2 and CO byreaction with steam as shown in reaction (3):

CH4 þH2O$ 3H2 þ CO DH� ¼ 201 kJ=mol ð3Þ

The catalyst is Ni/a-Al2O3 in a packed bed tubular reactor oper-ating at 800 �C, 15–20 bar and a space velocity �3000 h�1.

Carbon monoxide (CO) is then reacted in 2 steps (high and lowtemperature water gas shift) with additional steam to further gen-erate more H2 while reducing the CO content generating CO2 (reac-tion (4)):

COþH2O$ H2 þ CO2 DH� ¼ �44 kJ=mol ð4Þ

The high temperature shift reaction is carried out using an Fe,CrOx catalyst in a packed bed at 350 �C at reformer pressure andSV �5000 h�1. The low temperature shift reaction uses CuO, ZnO,Al2O3 in a packed bed at 180 �C at reformer pressure and a SV�3000 h�1.

The H2 purification step utilizes pressure swing adsorption(PSA) where all gaseous components, except most of the H2, are ad-sorbed in a zeolite. About 90% of the purified H2 is withdrawn atpressure from the chamber with the 10% remains on the zeolite.A reduction in pressure in the PSA chamber desorbs CO2, H2 andresidual CO from the zeolite. This ‘‘tail gas’’ is sent to a combustorwhere the H2 shown in reaction (5) and traces of CO are oxidizedsupplying some of the heat required for the endothermic streamreforming reaction.

H2 þ 1=2O2 ! H2O DH� ¼ �243 kJ=mol ð5Þ

A schematic of the traditional reformer process is shown inFig. 1.

3. Applicability of traditional hydrogen generation fordistributed H2 applications

The HDS process in large hydrogen plants is carried out at highpressure and temperatures [5]. These conditions are too severe forsafe and economic operation (high H2 compression costs) in com-mercial buildings, schools and homes. The natural gas is suppliedto buildings slightly above atmospheric pressure at ambienttemperature.

Steam reforming (SR) of natural gas, catalyzed with Ni pelletsoperating at atmospheric pressure in multi-tubular reactors is

being employed successfully in first generation fuel cells. Howeverissues of pressure drop, heat transfer, system-size, sulfur-contain-ing compounds, such as H2S always present in natural gas, and airsensitivity exist. The NiO catalyst must be activated by reduction tothe metallic state to generate the active sites for SR. One other ma-jor design consideration for SR is heat transfer resistance. Withtubular reactors, containing packed beds of catalyst, heat transferfrom the combustion gas through the outer tube wall throughthe radial catalyst bed limits the rate of reaction. This can beunderstood by reference to Fig. 2. Here we see desulfurized CH4

and steam entering multi-tubular reactors containing packed bedsof catalyst. The rate of steam reforming is highest for the catalyst indirect contact with the externally heated tube wall. Moving radi-ally from the tube wall the catalyst bed is cooler due to the endo-thermic steam reforming reaction and insufficient heat transfer. Tominimize this effect H2 plants operate with small diameter tubularreactors allowing for improved heat transfer to the entire but re-sults in higher pressure drop. These complications can be ad-dressed with a heat exchanger washcoated with a high activitycatalyst providing enhanced heat transfer. These designs will bediscussed later.

Water gas shift catalysts, mainly CuO, ZnO, Al2O3 in pellet form,are being employed in some first generation fuel processors. They

174 R.J. Farrauto / Chemical Engineering Journal 238 (2014) 172–177

operate at low space velocity and therefore occupy almost 50% ofthe reactor volume and create a large pressure drop. Also the activestate for the catalyst requires reduction to a lower oxidation statethat when exposed to air during discharge or an accident can leadto significant self-heating creating potential damage to the entiresystem. Particulates are also prone to attrition.

Pressure swing adsorption (PSA) requires elevated pressures,therefore is not feasible for a system operating on ambient pres-sure natural gas delivered to the building. It can, however, be usedfor stationary applications, such as hydrogen service stations,where compressors are part of the design.

Thermal combustion is commonly used in first generation fuelprocessor systems to provide the heat for the endothermic steamreforming reaction but some undesired emissions are generatedand must be treated.

4. First generation fuel processors for PEM fuel cells

A schematic of the first generation fuel processors used for gen-erating H2 for PEM (Nafion� based) fuel cells is shown in Fig. 3 [6].The HDS high pressure and temperature process (Fig. 1) has beenreplaced with room temperature and atmospheric pressure func-tioning adsorbents (e.g. zeolites, and/or metal exchanged Al2O3

and/or carbons) that remove the organic (mercaptans and thio-phenes) and inorganic sulfur compounds (H2S, COS).

The steam reformer operates with multiple small diametertubular reactors containing a packed bed of Ni based – pellet (par-ticulate) catalyst at atmospheric pressure and about the same tem-perature as a traditional reformer (750–800 �C). The NiO is reducedto its active metallic state and must be free from air to avoid spon-taneous oxidation. At an outlet temperature of about 650 �C a H2

rich stream, containing about 8% CO (diluted in H2 and steamand some CO2), is essentially at equilibrium. It is cooled to 200 �Cand delivered to the WGS reactor. The two WGS shift beds ofFig. 1 (high and low temperature) are replaced by one packedbed of LTS catalyst pellets operating around 200 �C. The WGS cat-alyst must be reduced to its active state and prevented from sub-sequent contact with air to avoid oxidation and a dangerousexotherm. This bed reaches thermodynamic equilibrium leavingunreacted about 2000–5000 ppm of CO. Since CO is a poison forthe Pt anodes of the PEM fuel cell it must be removed to<10 ppm [1]. This is accomplished by using a highly selective oxi-dation catalyst (PROX) downstream. About twice the stoichiome-

Fig. 3. First generation fuel processors integrated to a PEM fuel cell.

tric amount of O2 (from air) for complete CO oxidation is addedto insure removal of CO to safe levels for the anode. The catalystis typically Ru/Al2O3 pellets. It must be understood that the H2 isabout 40–50% of the stream while the CO is less than 1% so selec-tivity towards CO oxidation is critical. The catalyst directs the O2 tooxidize the CO to <10 ppm (reaction (6)). The excess O2 is oxidizedby the H2 (reaction (7)). Some cooling of the exotherm within thebed is necessary to maintain the selectivity.

Yes : COþ 1=2O2 ! CO2 DH� ¼ �283 kJ=mol ð6Þ

No : H2 þ 1=2O2 ! H2O DH� ¼ �243 kJ=mol ð7Þ

The stream is then delivered to the anode of the PEM fuel cell[6] where electrochemical reactions convert H2 and O2 to H2O(reactions (8)–(10)) while generating electricity and heat.

Fuel cell reactions:

Anode : H2 � 2e� ! 2Hþ Eo ¼ 0 Volts ð8Þ

Cathode : 1=2O2 þ 2e� þ 2Hþ ! H2O Eo ¼ 1:23 Volts ð9Þ

Anodeþ Cathode : H2 þ 1=2O2 ! H2O DEo ¼ 1:23 Volt ð10Þ

The fuel cell anode system is designed for 80% utilization of H2.The remaining 20% is oxidized (reaction (5)) in a thermal burner toraise steam for the reforming reaction.

H2 þ 1=2O2 ! H2OðgÞ DH� ¼ �243 kJ=mol ð5Þ

5. Precious metal monolith catalysts

The inspiration for new reactor designs for reformers is anextension of automotive emission control technology first com-mercialized in 1975 in the US [7]. This was the first widespreaduse of monolithic structures as supports for washcoated preciousmetal catalysts. Fig. 4 shows a simple cartoon illustrating the con-cept of a monolith and washcoated catalyst. The washcoat consistsof a high surface area carrier, such as stabilized Al2O3, upon whichprecious metals are dispersed. Precious metals have high activityper gram and thus only small amounts (<1–2 weight%) are neededto provide a thin layer of washcoat (�200 lm) on the walls of themonolith. This allows a final product with high activity per unitvolume.

The main advantage of precious metals over base metals oxidesfor emission control catalysts is their high activity density, allow-ing thin washcoats to be used, and thermal and structural stability.This advantage also exists for reforming reactions allowing forsmaller and more robust systems. The term monolithic structureis used to include all uni-body structures. Monoliths with preciousmetal-containing washcoats have proven to have excellentmechanical and chemical durability, low pressure drop, rapid re-sponse to transient operation and smaller sizes than reactors withtraditional catalyst pellet materials. For these reasons precious me-tal catalysts on monolithic structures were proposed for naturalgas fuel processors integrated to low temperature fuel cells andstand-alone hydrogen generators. These general advantages overbase metal pellets or particulates are summarized in Fig. 5.

6. The next generation of fuel reforming technology for the H2

economy

The steam reforming reaction (3) is a highly endothermic reac-tion and supplying external heat is critical for efficient use of thecatalyst. This was demonstrated using a Pt, Rh washcoated catalyston a heat exchanger [8] in a design for a hydrogen service station.In this design desulfurized natural gas was passed over the Pt, Rh

Fig. 4. Monolith structure, catalyzed washcoat on the wall, catalytic converter in exhaust of a vehicle.

Fig. 5. Washcoated monolith (heat exchanger) structures compared to pellets (particulates).

R.J. Farrauto / Chemical Engineering Journal 238 (2014) 172–177 175

catalyst deposited on the internal walls of a heat exchanger. Theendothermic heat was provided by oxidizing H2 and CH4 over aPt, Pd catalyst deposited on the outer surface of the same heat ex-changer. This allowed heat to be supplied to the reforming reactionthrough the metallic wall essentially eliminating heat transferresistance. This unique design was utilized by Chevron in concertwith Engelhard (now BASF) and Modine for a hydrogen service sta-tion producing 5 kg H2 per day [1,8]. This combination of a highactivity precious metal washcoated catalyst coupled with an en-hanced design for heat transfer permitted a reduction in reactorsize by a factor 10 (10 times greater space velocity, �30,000 h�1)compared to that of a process that utilizes Ni pellets (�2000–3000 h�1). The precious metal catalysts require no pre-reductionand are insensitive to air exposure providing an extra safety mar-gin for consumer use. Additionally no auxiliary equipment is nec-essary for start-up or shut-down as is required for base metalcatalysts due to their air sensitivity. Catalytic oxidation of thePSA tail gas provides some heat through the catalyzed shell fins

for the endothermic steam reforming reaction. The washcoat thick-ness, typically no more than about 200 lm allows greater struc-tural stability since temperature gradients essentially do notexist relative to a 4–6 mm diameter base metal pellet. The reducedweight and volume of the monolithic structure (relative to thepacked bed) and the thin washcoat allows rapid response to tran-sient operation, a requirement for fuel cells with turn down ratios(varying space velocity) of up to 20 to 1 from fuel cells with varyingpower demands. Therefore precious metal washcoated monolithdesigns allow new compact reformer designs with high activityand durability for both stand-alone fuel processors and those inte-grated to stationary fuel cell applications [6,8]. The precious metalscan be recovered and recycled using conventional wet or pyro-metallurgical methods. Cartoons of other designs utilizing wash-coated catalysts on heat exchangers are shown in Figs. 6 and 7.

The WGS reaction (reaction (4)) is slightly exothermic and thusis thermodynamically favored at low temperature where kineticsare rate limiting. In the proposed new designs the Cu-based

176 R.J. Farrauto / Chemical Engineering Journal 238 (2014) 172–177

pelleted catalyst is replaced with a washcoated Pt, Re catalyst [1]on a ceramic monolith allowing a space velocity to exceed about20,000 h�1 at 300 �C (about 2000–3000 h�1 for Cu based LTS cata-lysts). This allows a 10-fold reduction in size of the WGS bed due tothe enhanced activity of the precious metal catalyst verses Cu pel-lets. The Pt, Re catalyst requires no activation and is insensitive toliquid water condensation or air exposure which will deactivatethe Cu- based LTS catalysts. The monolith supported WGS catalystreaches equilibrium of about 1% CO at 300 �C.

Since carbon monoxide adsorbs strongly and poisons preciousmetals below about 180 �C, one of the most important require-ments for successful operation of a Nafion�-based PEM fuel cellis reduction of the CO to <10 ppmv to protect the anode [1]. Witha small amount of injected air over a highly selective preferential

Fig. 6. Cartoon of a steam reforming catalyzed washcoated on a tube wall with heatsupply. Blue circles represent posts to generate turbulence to divert the process gasflow to the wall where the heat is most intense. (For interpretation of the referencesto colour in this figure legend, the reader is referred to the web version of thisarticle.)

Fig. 7. A steam reforming catalyst washcoated on a tubular reactor with endothermic

oxidation catalyst the CO is reduced from up to 10,000 ppmv(1%) to <10 ppmv without oxidizing appreciable amounts of H2

present in the process stream. The PROX reaction generates vary-ing amounts of heat depending on the O2 content which can belower for a more selective catalyst. A Pt, Cu and Fe on Al2O3 wash-coated monolith catalyst achieves CO levels well below 10 ppmwhile operating adiabatically at an inlet of 90–100 C [9]. It is espe-cially designed to limit any reverse water gas shift reaction (re-verse of reaction (4)) that tends to occur at temperature aboveabout 120 �C. The ideal stoichiometry for PROX calls for 1/2 moleof O2 per mole of CO. However due to the large excess of H2 present(60–70%) relative to <1% CO the O2 (in air) added is about twicestoichiometry of the CO to be reduced. Therefore an equivalentamount of H2 is also oxidized leading to an exotherm of 40–50 �C. High selectivity is absolutely essential since the initial H2/CO ratio normally exceeds 100:1, but the final H2/CO ratio can in-crease to 50,000:1 as the reaction proceeds to completion. The cat-alyst operates over a wide range of inlet and outlet temperatures(up to 150 �C) and turn-down ratios of 10:1. It is also possible todeposit the PROX catalyst on a heat exchanger to recover heat.PROX is only used for reformers that are directly integrated tothe low temperature (80 �C) PEM residential fuel cell system. Fuel-ing stations, built for the ‘hydrogen highway,’’ operate at suffi-ciently high pressures that pressure swing adsorption (PSA) isthe technology of choice for H2 purification. PROX is also not nec-essary for high temperature PEM fuel cells, such as those utilizingthe polybenzimidazole (PBI) membrane or phosphoric acid both ofwhich operate above 180 �C since CO does not chemisorb apprecia-bly on the Pt anode at these temperatures [6].

Washcoated monolithic catalysts, containing Pt, Pd combina-tions, are used to oxidize the H2 effluent for the fuel cell or theCO and H2 tail gas from a PSA. The heat is recovered to enhancethe overall energy balance and system efficiency. Catalysts allowmixtures outside of flammability to be oxidized unlike thermalreactions which require the addition of additional fuel (CH4) togenerate a flame (and emissions of CO, HCs and NOx).

Fig. 8 shows a typical conceptual design using monolith cata-lysts for reforming natural gas for either a H2 service station or afuel cell. The sulfur odorant is removed by adsorption in Al2O3

and/or zeolite materials at ambient conditions. Reforming of desul-furized natural gas is conducted over a Pt, Rh catalyst washcoated

heat provided by the catalytic oxidation of H2 using a monolith oxidation catalyst.

Fig. 8. Fuel processors using precious metal-containing washcoats on monolithic structures.

R.J. Farrauto / Chemical Engineering Journal 238 (2014) 172–177 177

onto a heat exchanger. The effluent is cooled via a steam boiler andenters a ceramic monolith washcoated with a Pt, Re catalyst wherewater gas shift is conducted below 300oC. The product from WGScan be further purified with the addition of a small amount of air toa washcoated PROX catalyst consisting of Pt, Cu, and Fe on a cera-mic monolith substrate and delivered to a PEM anode. Alterna-tively the product gas from WGS can be purified via PSA (onlyfor pressurized systems) and used for a hydrogen filling stationfor vehicle applications. The tail gas is combusted catalytically overa washcoat of Pt, Pd on a ceramic monolith providing some of theheat necessary to drive the endothermic steam reformingreactions.

7. Conclusions

This brief review suggests that washcoated precious metal cat-alysts deposited on monolithic structures will play a key role ingenerating hydrogen. The hydrogen economy will includedistributed hydrogen for fuel cells as well as small industrial usessuch as hydrogenations, cooling, materials processing. where on-site generation makes economic sense. This review suggests thereare alternative approaches in both catalysts and reactor designs tomeet these challenges not likely to be met with traditional basemetal pellet catalysts in packed bed reactors.

The major drawback to the use of precious metals is their inher-ent expense. This was a major obstacle in the early days of thecatalytic converter. Automotive catalyst research has resulted inmore efficient use of the precious metals allowing a decrease of50–80% metal in modern converters (2013) than the early units

of the mid-1970s. In parallel precious metal recovery methodshave also advanced. Fuel processing and fuel cell catalysts are ex-pected to follow a similar experience and continuing research willdecrease the amounts needed in both the fuel cell electrodes andreformer. Processes for recycling precious metals are already partof the business loop for automotive, petroleum and chemical appli-cations. Furthermore the reduction in reactor size for monolithicsystems decreases overall material costs. The robustness of pre-cious metals catalysts requiring no activation or any special safetyprecautions due to air exposure translates to less auxiliary equip-ment normally required for base metal oxide catalysts. Advancesin research and optimization of catalysts and reactor designs willreduce the use of precious metals following the same model asthe automobile catalytic converter.

References

[1] R.J. Farrauto, Y. Liu, W. Ruettinger, O. Ilinich, L. Shore, T. Giroux, Catal. Rev. 49(2007) 141–196.

[2] J. Armor, Catal. Lett. 101 (3–4) (2005) 131–135.[3] C. Forsberg, Chem. Eng. Progress. 101 (12) (2005) 20–22.[4] R. Heck, R. Farrauto, S. Gulati, Chem. Eng. J. 82 (2001) 149–156.[5] J.R. Rostrup-Nielsen, In Catalysis, Science, Technology, J.R. Anderson, M.

Boudart (Eds.), Springer-Verlag, New York, 1984.[6] R. Heck, R.J. Farrauto, S. Gulati, Catalytic Air Pollution Control: Commercial

Technology, third ed., Wiley and Sons, Hoboken, NJ, 2009 (Chapter 16).[7] R. Heck, R.J. Farrauto, S. Gulati, Catalytic Air Pollution Control: Commercial

Technology, third ed., Wiley and Sons, Hoboken, NJ, 2009 (Chapter 6–9).[8] K. Nguyen, C. Kraus, B. Balasubramanian, M. Reinke, J. Valensa, in: NHA Annual

Hydrogen Conference March 29-April 1 Washington, DC, 2005.[9] Q. Zhang, L. Shore, R.J. Farrauto, Int. J. Hydrogen 37 (14) (2012) 10874.