The Central Role of Catalysis in a Future Energy ...

11 downloads 65 Views 750KB Size Report
The switch to reformulated gasoline (RFG) ... gasoline-fueled cars fitted with auto-exhaust .... interests in the Department of General Energy ...... is to couple the reformer to a spark-ignition in- ..... should recall that evidence for mechanistic.
Trends in Physical Chemistry, Research Trends, Trivandrum, India. vol. 5, 1995, pp. 91-156

The Central Role of Catalysis in a Future Energy Cycle based on Renewable Hydrogen and Carbon Dioxide as Reactive Liquefier

J.G. Highfield* Energy Storage Section, Paul Scherrer Institute, 5232 Villigen PSI Switzerland



Current address: Heterogeneous Catalysis. Institute of Chemical & Engineering Sciences, 1 Pesek Rd.. Jurong Island, SINGAPORE 627833 E-mail: [email protected]

Abstract This article presents the author's viewpoint on a renewable energy scheme with good potential to support future sustainable development. This cycle is based on simple alcohols as efficient and convenient mid- to long-term (seasonal) hydrogen storage forms. In the synthesis, carbon dioxide trapped from concentrated sources or, ideally, from the atmosphere is proposed as a reactive “liquefier” for hydrogen produced from clean, renewable primary sources. The hydrogen fuel may be released on demand by co-reaction of alcohols and water, i.e., the reverse of the synthesis process. In the case of methanol, the relevant technology is close to technical maturity and offers early prospects for commercial implementation in the impending transition away from dependence on fossil fuel resources. Both methanol and ethanol will be important in the longerterm in view of their good economic outlook as renewable fuels from biomass gasification/ synthesis and fermentation, respectively. The role of catalysis and catalytic reaction engineering in alternative industrial synthesis routes and in alcohol fuel processing will be vital to the ultimate success of the proposed energy scheme. Described here are the recent advances of major significance in carbon dioxide hydrogenation to liquid alcohol fuels, and the reverse process, i.e., steam-reforming of alcohols, reported in the field of thermal heterogeneous catalysis at both the fundamental and technical level. A short overview of the present state-of-the-art in CO 2 recovery technologies is also included because of their physicochemical bases, and their relevance to the energy cycle under consideration. The more distant prospects for electrocatalytic synthesis of alcohols from CO2, a recently revitalized field, will be briefly summarized.

Contents

page

1. Introduction

4

1.a Beyond fossil fuels to sustainable energy: The challenge 4 1.b Hydrogen energy and its storage

5

1.c Simple alcohols as hydrogen storage media: The role of carbon dioxide

6

1.d The scope of this review article

10

2. Thermodynamic considerations

11

3. Trends in carbon oxides hydrogenation 13 3a.

Industrial and Technical Aspects

13

3a.1 Methanol

13

3a.2 Ethanol (and higher alcohols)

15

3a.2A: Direct Routes

15

3a.2B: Indirect Routes

16

3b.

18

Fundamental Advances

3b. 1 Methanol 3b.1.1 CO2 activation: surface science view 18 3b.1.2 Methanol synthesis & water-gas shift 18 3b.1.3 In-situ studies & surface species

18

3b.1.4 Mechanism:

Role of formate and/or formyl?

19

3b.1.5 The CO2 / CO synergy

21

3b.1.6 Kinetics: active role of support?

22

3b.1.7 Kinetic parameters /rate-determining-

step(s)

23

3b.1.8 Synthesis route “post-formate”?

29

3b.1.9 Summary

30

3b. 2

Ethanol (& higher alcohols)

3b.2.1 General prospects

31

5.

Recent advances in CO2 recovery technologies: A brief overview

48

5a.

Trends in chemical absorption

48

5b.

Perm-selective & microporous hollow-fibre membranes

48

5c.

Adsorption on porous solids

49

5d.

CO2 recovery from the atmosphere

49

6.

Prospects for alcohols synthesis via electrocatalytic reduction of CO2 50

31

3b.2.2 Alkali-modified methanol synthesis

catalysts

32

3b.2.3 Modified Fischer-Tropsch catalysts

34

3b.2.4 Supported Rhodium catalysts

35

3b.2.5 Strategy for catalyst selection and

screening

37

4. Steam-reforming of simple alcohols

38

4a. Industrial and Technical Aspects

38

4a.1

Background

38

6a.1 Background

50

4a.2

Steam-reforming of methanol

38

6a.2 Recent advances

51

6b.1 Current Density Enhancement

51

6b.2 Reduction of overvoltage & selectivity to alcohols

52

7.

Concluding remarks

53

Acknowledgments

53

Glossary

54

Bibliography

55

4a.2.1 Advantages of pre-reforming

4a.3

Methanol for on-board storage of H2

39

4a.3.1 Reformer/ICE link-up

39

4a.3.2 Reformer/Fuel cell link-up

40

4b.

42

4b.1

Fundamental advances Methanol S-R

42

4b.1.1 Current views on the mechanism:

role of methyl formate?

42

4b.1.2 Dehydrogenation vs. S-R

43

4b.1.3 Chemistry of methyl formate

44

4b.1.4 Implications for the RDS and CO-free operation

44

4b.2

Ethanol S-R

44

4b.2.1 Rationale for study

44

4b.2.2 Early progress

45

4b.2.3 Summary

47

-41. Introduction 1.a

Beyond fossil fuels to sustainable energy: The challenge

Since the United Nations Conference on Environment and Development in 1992 in Rio de Janeiro (“Earth Summit”), it is now almost universally acknowledged that even the present quality of life, as experienced in the nations of the Developed World, can only be assured in the Third Millenium if government policy at an international level is guided by a policy of sustainable development. Nowhere is this more urgent than for energy supply and usage [1]. A difficult transition must be made away from our overwhelming dependence on fossil fuels, with their manifold polluting effects, towards an energy supply system which is ubiquitous, i.e., domestically viable, inexhaustible, and whose usage is environmentally benign. Since the “oil crisis” of 1973, when public awareness of the limited and unpredictable nature of conventional fuel supply first took hold, the emphasis has shifted justifiably to the more immediate fear of irreversible ecological consequences of usage on both a local and global level. For example, the growing health risk due to pollutants like CO, NOx, ozone, particulates, etc., from transportation fuel combustion in congested urban locations is now acute. In the broader perspective, anthropogenic emissions of SO2 from coal-fired power stations have created “acid rain" with its pernicious effects on forests and freshwater lakes. Once considered innocuous by comparison, emissions of greenhouse gases (GHGs) like methane and carbon dioxide, which are virtually proven to have caused the average global temperature to rise measurably in just a few decades [2-3], figure as a longer-term threat. The consequences of such micro- and macro- perturbations to the ecosystem are difficult to predict. Nevertheless, the “carrying capacity” or resilience of the environment is obviously limited such that the risk of ecological catastrophe, even on a global scale, is an inevitable danger with a continuing policy of “business as usual”.

Appropriate legislation to combat such problems is now growing at the national level and is likely to become more stringent in the foreseeable future. In the USA, emissions in the transportation sector are of great concern. Oxygenate/gasoline blends are believed to result in lower CO emissions in the winter months, whilst reducing volatile organic compounds or VOC emissions (ozone/smog precursors) in the summer in many cities [4]. The switch to reformulated gasoline (RFG) is one major program mandated by the Environmental Protection Agency (EPA) to compl y with the Clean Air Act Amendment of 1990, and is actively encouraged by the government using economic incentives to fuel suppliers in the form of waivers of excise duty on the additives. Since 1995, RFG is required to contain at least 2% by weight of oxygen, 15% of which should be derived from “renewable” sources, viz., fermentation ethanol [5] and its derivative ethyl tertiary-butyl ether (ETBE). Many auto manufacturers [6] now offer “flexible-fuel vehicles” (FFVs) which can run on alternative blends called M85 or E85 (15% gasoline in methanol or ethanol). The present U.S. administration has initiated a joint government/industry venture to make a “supercar” with unprecedented fuel efficiency and emissions profile [7]. The State of California mandate to auto manufacturers is to bring onto the market a significant number of so-called “low emission”, “ultra-low emission” and “zero emission” vehicles (LEV, ULEV, ZEV) by 1998 [8]. Current gasoline-fueled cars fitted with auto-exhaust catalysts do not yet meet the exacting standards of even the first of these categories (LEV), whilst only electric cars qualify as ZEVs under the present definition. In Europe, activities have been directed more against global warming, i.e., the GHG emissions problem in general. The lead has been taken by all the Scandinavian countries, Denmark and the Netherlands, in imposing an energy or carbon tax, albeit at a modest rate [9]. Economic analysis has shown that this need not lead to industrial disincentives and inflationary effects by allowing suitable relief

-5 on other taxes [10]. All signatories to the Nations Framework Convention on Climate Change (166 countries as of May 1994) are committed to stabilization, if not reduction, of GHG emissions from a base level (at ca.1990) within 10-15 years. What is not yet clear is how determined individual governments will be in achieving this goal. Measures such as those itemized above, encouraging mainly the rational and cleaner use of conventional fuels, are both timely and commendable. However it will be the task of scientists from academia and industry working in collaboration, and with far-sighted government support, to address the difficult prospect of bringing to technical maturity alternative schemes to satisfy an ever-increasing world energy demand and with ecological sustainability. The relevant utilities must also be phased in at a rate sufficient to smooth the transition to a renewable energy economy and be made available on a scale ultimately large enough to supplant fossil fuel resources altogether. A probable time-scale for this transition may be as little as 50 years, with the downturn in fossil fuel supply starting a mere 15 years from today, so there is much to accomplish in the meantime! 1.b

Hydrogen energy and its storage

The general consensus is that any alternative energy system capable of fulfilling the criteria for sustainable development will be based on hydrogen and electricity as the main carriers. The former will be produced from clean, renewable primary energy sources, e.g., solar-, wind-, or hydro-powered water electrolysis [11-14]. A chemical fuel, preferably in liquid form, has the primary advantage compared to electricity of indefinite storage time with virtually no losses, and is a vital component of any viable energy economy. It can be transferred over long distances, distributed with existing pipeline infrastructure, and has the high energy density required for practical onboard storage (with good range) as transportation fuel. Indeed, these are major logistical reasons as to why oil and gasoline ever came to dominate the energy economy and, by the same token, why they are proving so hard to replace!

Hydrogen is the cleanest-burning fuel in existence, producing essentially water and small levels of NO x by-product from nitrogen in the air. Even the latter can be controlled by exploiting the wide flammability range of hydrogen (4-75 vol%) in a conventional combustion process (engine) by running “lean”. With catalytic heaters or fuel cells, both under intensive development, NOx emissions become vestigial or non-existent. Although proponents of a future hydrogen energy economy have their fair share of “purists” [13,15], there is unanimous agreement that a major technical barrier to its widespread implementation (apart from economic factors such as production costs from renewables [16]) is the difficulty of storing hydrogen conveniently and efficiently. Despite having the highest mass-specific energy density (142 kJ. g-1) of all fuels, making it potentially an excellent aviation fuel, its volumetric energy density either as compressed gas or in liquefied form at -253°C (∼10 MJ.1-1) is still less than one third that of gasoline. This would be especially problematic in ground transportation usage, making handling and onboard storage somewhat impractical or even uneconomic, due to refueling (evaporative) losses in the case of the liquid. Various methods of storage other than pressurization and liquefaction are being considered, including physical entrapment, e.g., cryogenic adsorption on activated carbon [17] or in glass microspheres [18] at high pressure; and chemically-bound forms such as inorganic metal hydrides. The latter has been by far the most popular approach to date, as seen by consulting the relevant sections of Hydrogen Energy Progress IX and X [17-18]. Nevertheless, a system which offers realistic storage densities (>5 wt% as H) together with ease of recycling and long term stability, i.e., resistance to contamination and irreversible physical changes, is not yet forthcoming. Alternative chemical storage forms, which have received less attention thus far are organic hydrides or hydrocarbons. As all hydrocarbons, by definition, contain incipient hydrogen, there exists the possibility to remove selectively the H component by dehydrogenation (for clean

6 fuel applications), and to recycle the carrier. The principal difference between inorganic and organic hydride storage is that the latter requires a catalyst for H-cycling at an acceptable rate for technical viability. Advantages over inorganic systems are the fluidity of the carrier (invariably liquid) and the higher energy density of the stored H. Current research interests in the Department of General Energy Technology at PSI include H-storage in toluene (methylcyclohexane, MCH, in hydrogenated form), and in simple alcohols [19]. This review deals with the latter. 1.c

Simple alcohols as hydrogen storage media: The role of carbon dioxide

Simple monohydric alcohols such as methanol and ethanol occupy a uniquely valuable position in renewable energy schemes as they can be produced from biomass and cellulosic urban wastes by established process technology, viz., gasification/synthesis and fermentation, respectively [20]. With the proviso that the biomass from which they are derived is grown sustainably, the combustion of these fuels does not contribute to a net increase in atmospheric carbon dioxide, i.e., they are CO2neutral in the long-term. Remarkable progress has been made recently in biomass conversion technology, notably at the National Renewable Energy Laboratory (NREL, Colorado, U.S.A). Ethanol can now be produced from cellulose via the simultaneous saccharification and fermentation (SSF) process at close to the price for methanol ($0.28 1-1). With further development a realistic short-term (5 year) forecast price for both alcohols ($0.20 l - 1 ) would be competitive with gasoline derived from crude oil at $25 per barrel. In view of the imminent price breakthrough expected, there is no doubt that simple alcohols will play a vital role in the transition to a renewables-based economy and possibly beyond. Early signs of their emerging role are clear, attracting serious industrial interest, along

with their ether derivatives MTBE and ETBE, as oxygenated blending agents for mitigation of gasoline emissions and octane improvement. The industrial Higher Alcohol Synthesis (HAS) of the Institut Français du Petrole (IFP) and the Octamix process of Lurgi, both based on catalytic conversion of syngas, have recently been commercialized for this market [21-22]. However, neither strictly would qualify for tax waivers on the basis of renewability unless the syngas feedstock was derived from, e.g., gasification of biomass. Neat alcohols are excellent fuels per se, as illustrated by the Brazilian “Proalcool” experience of the last two decades with ethanol produced from fermentation of domestic sugar [23]. Both alcohols have higher octane numbers and wider combustion envelopes than gasoline. This translates, with only minor engine adjustment, into better performance (higher compression ratios) and fuel economy by running “lean” [24]. However, despite many favourable properties, alcohols are still not ideal fuels when compared to H2 because although their overall emissions inventory is much lower than that of gasoline, straight combustion also leads to potentiallyharmful levels of toxic aldehydes [6], organic acids and other smog-enhancing derivatives [25]. Even their value as oxygenate blending agents for gasoline has been questioned recently [26]. The concept of alcohols as H-storage media is certainly not new. It was already implicit in explorations into the potential benefits of advanced engine design for on-board “pre-reforming” of methanol to a H-rich gaseous fuel mixture, notably by workers at Nissan Motor Co. in the early 80's [27-28]. Two major advantages were demonstrated over the direct combustion mode; viz., further dramatic reduction in emissions (obviating the need for a tailpipe exhaust catalyst), and a significant thermal efficiency improvement. The latter was achieved because the endothermic pre-reforming step was driven by recycled exhaust heat.

7

Unfortunately, the advanced engine research at Nissan was suspended shortly thereafter in favour of FFV development, in part due to the limited durability of the far from optimum reformer catalyst utilized [29]. Pre-reforming has since been taken up again and the obtainable benefits verified by, inter alia, Swedish researchers [30]. On a larger scale, conversion of methanol for on-site H2 supply (by catalytic steam-reforming, S-R), in modest demand industrial applications is now commercial reality [31]. Although it has been estimated that alcohols from biomass could be produced on a scale sufficient to meet the present U.S. consumption of gasoline equivalent (425 billion l.p.a.) using domestic resources alone [20], the extensive cultivation of the “energy crops” necessary in this scenario will be difficult, if not impossible, to sustain in the longer-term without careful planning. Potential threats to successful agricultural management of bioenergy plantations, such as pestilence and disease (to which monocultures are especially susceptible), soil nutrient depletion, etc., will always be present and will require adequate countermeasures [32]. In view of this logistical uncertainty, the development of a complementary industrial process for renewable fuels production would serve as vital diversification in helping to secure a future energy economy. In effect, this requires substitution of conventional feedstocks (coal and natural gas), as the H2 source for syngas production, by renewable primary energies. The ideal process should parallel the natural one, i.e., photosynthesis, but make more suitable end-products, and the term “renewable petrochemistry” has been coined to convey the idea [19]. Nature's own elegant solution to the energy conversion and storage problem is to use the visible component of sunlight to create H (but as bound, reducing intermediates rather than as free gas) from water, which goes on to build carbohydrates (biomass) by “fixation” of ambient carbon dioxide. The CO2 is ultimately released again to the atmosphere during the gradual process of vegetative decay. Thus, the role of CO2 as recyclable substrate has been around for

quite some time! It is only in the last few decades that Man is realizing the benefits of recycling in general and, only in recent years, of recycling CO2 in particular. Part of the reason for the retarded growth of awareness in the potential of CO2 as feedstock for fuels and chemicals is because it has traditionally been thought of, along with H2O, as a stable end-product of many chemical processes. However, interest in CO2 as a potentially cheap and ubiquitous carbon source for industrial processing has risen continuously since the early 80's, spurred on inevitably by growing environmental concerns. The first indications that CO 2 might have advantages over CO in syngas conversion to methanol also appeared at this time [33-35]. After confirmation, on the basis of kinetics and isotope-tracing, by ICI and the Evanston group, working independently, that CO2 is converted faster than CO under typical industrial synthesis conditions, it was clear that this constituted a major advance in mechanistic understanding [36-39]. Conventional wisdom up to this time was that the beneficial role of CO2, which is always present as a minor component of syngas, was limited to keeping a fraction of the metallic Cu in the catalyst as CuI, the species believed responsible for activity [40]. Indeed, CO2 levels were typically kept low (< 5 vol%) to avoid inhibition of CO conversion in these early catalyst formulations. The overwhelming evidence since of the pivotal role of CO2 in methanol synthesis is no longer contested [41]. An important consequence of this pioneering work is that all commercial Cu/ZnO/Al 2 O 3 catalyst formulations are now optimized to work with high CO2 levels in syngas [42]. In view of the advances outlined above, it is now realistic to consider energy schemes operating in a global context which couple the advantages of renewable hydrogen fuel with the excellent storage properties of alcohols synthesized from recycled carbon dioxide. A working cycle based on methanol, involving both synthesis and steam-reforming in a single unit (possibly even using a single catalyst!), is

Fig.1 Idealized energy scheme based on hydrogen storage in simple alcohols close to demonstration [43]. Extension of the concept to ethanol in the longer-term would further diversify the scheme and bring additional benefits. Ethanol is the major biofuel alcohol at the present time (methanol is made from non-renewable CH 4 ) and has the preferred handling characteristics, which may be a decisive factor in the issue over consequences of their usage in the public sector on an unprecedented scale. The idealized scheme is shown in Fig.1 above. A primary feature of this scheme is that it is CO2 -neutral in the long-term. Thus, if ever implemented on a large-scale, it would allow the natural buffering capacity of the earth, primarily vegetation and the oceans, to recover the “surplus” CO2 in the atmosphere. As with the principle of symmetry often expressed in Nature herself, a solution lies potentially at the root cause of any problem; in this case the global energy (fuel) cycle. To be fullyeffective, such a strategy to deal with the CO2 problem should include a massive program of afforestation and energy-crop cultivation. It is

estimated that up to half of the excess CO 2 , 1-2 gigatons out of 3.5 gigatons (as C) annually [44], could be sequestered in this way [45-46]. The economic burden of such a long-range strategy will inevitably fall on governments. What is needed is a “global alliance” which puts high priority on this issue. Impetus for the major industrial reorientation required in the plan could be provided by government guarantees to buy the product (alcohol fuel), initially for a “global reserve”, at a baseline price which offsets industry's risk of oversupply and commodity devaluation, which would follow in a free market economy. The incentives for government are potential alleviation of global warming and the first step in securing an energy infrastructure for sustainable development in a single strategy. In an ideal world, the money would be forthcoming from a major diversion of funds from existing priorities, viz., defence budgets, still the greatest drain on world progress and potential quality of life. We owe it to future generations.

9

The most suitable primary energy sources, i.e., those with the greatest potential to supply clean and unlimited power for H2 production in the near- to mid-term future are hydroelectric power and solar/photovoltaic electricity. The first is established technology which presently accounts for about 20% of the world consumption of electricity [47]. It is an attractive, ubiquitous, and economic source with the property of continuity, and would therefore be of great value in any energy scheme based on renewables, which are generally intermittent resources on a diurnal and/or seasonal basis. Its low off-peak cost (Can.$0.02 kWh-1, Sept Isles, Quebec) provides the economic basis for the proposed 100 MW EuroQuebec Hydro-Hydrogen Pilot Project, in which H2 is liquefied and transported by tanker to Europe for electricity and heat cogeneration, vehicle/aviation propulsion, and “Hythane” production for industrial and domestic use [48]. The technical status of the project (Phase 111,0, to 1997; H2 utilization and advanced storage technology), looks promising [15]. The forecasted economic global capacity of hydro-electric power is large enough (10,00 0-15, 000 TW h per annum or 40 exajoules) to satisfy, in terms of H2 energy, ∼12 % of the current world consumption in oil equivalent, i.e, ∼300 exajoules. Most of the untapped resources are situated in the developing countries, whose present output is limited to only ∼5% of their estimated potential [49]. However, the construction of large new plants is difficult to gain government approval for, especially in developing countries, due to valid public welfare arguments, such as environmental degradation of the immediate locality, and the social upheaval (population displacement, etc.) which often occurs during and after the building phase. Such conflicts can only be dealt with effectively by inviting public participation sufficiently early in the planning stage, and holding constructive talks towards cooperative regional development. In the longer-term, hydroelectricity will be superseded by solar electricity as the dominant renewable energy source for H2 production. Although this will begin with installations (solar-thermal & solar-photovoltaic) based in

land areas of high (direct) insolation, i.e., deserts, the photovoltaic (PV) flat-plate modular based systems are more versatile because they can also operate in diffuse lighting, thus providing better continuity and freedom for site location. It has been estimated that the current global consumption of fossil fuel energy could be replaced by H2 produced from PV electrolysis installations (with present-day efficiencies), which would occupy only 5% of the global desert area, or 1% of the global land area [14]. Unfortunately, PV-based systems are currently the most expensive option for solar hydrogen production based on established Si cell technology. In 1991, PV electricity cost ∼U.S.$ 0.20 kWh-1, which is an order of magnitude higher than the price of typical offpeak hydro-power. With state-of-the-art electrolysers operating at 70% efficiency, this leads to a H2 cost of $70 GJ-1 [14]. However, major technical advances and cost-savings in PV module manufacture [50], e.g., the use of amorphous Si [51], and thin films such as CuInSe, [52], are expected to lead to a H2 price as low as $10 GJ-1 within the next decade. The dye-sensitized TiO2 cell looks particularly encouraging due to its superior efficiency in diffuse lighting compared to conventional Si cells [53]. As a comparison, gasoline retails currently at ∼$3 GJ-1, but this is based on an artificially-low oil price and with no internalization of environmental costs, which may constitute the submerged fraction of the proverbial “cost iceberg” for conventional fuel use [15]. A recent economic analysis [54] has reaffirmed the basic notion that renewable hydrogen will only get a significant market “toehold” against fossil fuels in the immediate future by government incentives in the form of tax waivers on renewables and disincentives to conventional fuel use, e.g., CO2-emission quotas, internalized costing of damage to public health and environment caused by pollutants, etc. Looking beyond the currently gloomy political and economic outlook vis-à-vis H 2 energy research [55], the technical groundwork for alternative energies must be established before any transition, mediated by governments

- 10 or PPPs, is even conceivable. In the opinion of this author, the scheme under consideration involves a synergistic coupling of the main strength of conventional energies, viz., the availability of liquid fuels as convenient, high-density energy storage forms, with the environmental neutrality of hydrogen fuel usage. Systems based on this hybrid concept may effectively bridge the “technology gap”, which hinges on the storage problem, and thereby accelerate the establishment of a hydrogen energy economy for future generations to build on. The ramifications of a renewable energy storage system based substantially on recycled atmospheric CO2 (a truly universal and potentially cheap “liquefier”), especially for energy self-sufficiency in the developing world, are enormous provided that efficient trapping methods can be developed. This aspect will be considered later. 1.d

The scope of this review article

The primary subject matter of this review will be the wealth of recent progress in the field of catalytic CO2 hydrogenation and steam reforming (S-R) of alcohols, both at the fundamental and technical level. Such developments and their prospects are viewed as crucial groundwork upon which the future viability of the proposed renewable energy storage cycle will hinge. The rapid growth of interest in CO2 chemistry in general is attested to by the rising frequency of relevant international conferences and the availability of some excellent reference sources, notably that of Halmann [56]. Other major source materials, apart from the general literature, include the proceedings of several recent conferences and symposia specializing in CO2 chemistry, hydrogen energy, and alcohol fuels: viz., CO2 Chemistry: Environmental Issues, based on a workshop in Hemavan, Sweden,1993; CO 2 Fixation and Efficient Utilization of Energy, Tokyo, 1993; 3rd International Conference on CO2 Utilization, Oklahoma, 1995; 10th World Hydrogen Energy Conference, Cocoa Beach, 1994; and the 10th International Symposium on Alcohol Fuels, Colorado Springs, 1993.

This incomplete list of specialist meetings is given here not only because of their topicality and the quality of the articles contained therein, but also because many of these sources are by their very nature of limited access to the general reader, who may be acquainted thereby with some important details. Starting from the thermodynamics of CO2 hydrogenation (steam-reforming is formally the reverse process), advances primarily in heterogeneous catalysis are reported. Emphasis will be placed on synthesis work claiming kinetic control, as reflected in high selectivity to the most desirable products, i.e., alcohols, and the influence of fundamental variables, e.g., catalyst surface composition, structure, etc., reported ideally at a mechanistic level of understanding. Some technical aspects, such as reaction engineering to overcome equilibrium limitations in the synthesis, and special operating constraints in on board steamreforming of methanol for fuel cell (electric) vehicular propulsion, will also be considered. Industrial progress in syngas conversion to alcohols and other liquid fuels will be summarized because of the close link in chemical feedstock composition with a hypothetical “renewable feedstock”, due to the ready interconversion of CO and CO2 via the water-gas shift (WGS) equilibrium in situ. Methods now practiced, or currently in development, to improve syngas conversion per pass by overcoming equilibrium limitations will be outlined. This constitutes the greatest obstacle in methanol synthesis and will be potentially more severe starting from CO2. In view of its link with the main theme, recycling CO2 ultimately implies the development of efficient methods for its pre-concentration (trapping) from the atmosphere. Although these are admittedly non-catalytic processes, their physicochemical bases justify a short digression into the field and its outlook. Whilst heterogeneous catalysis is already welladvanced at the techno-economic level, the electrocatalytic synthesis of alcohols from CO2

will also be mentioned in view of exciting recent advances. Only a full review of the field would strictly do it justice but this is beyond the scope of the present article. 2. Thermodynamic Considerations The reaction central to the proposed scheme, linear monohydric alcohol synthesis via CO2 hydrogenation (the H2 storage step), can be expressed in the generalized form as:n CO2+ 3n H2 ↔ C n H 2 n + 1 O H + (2n-1) H2O where n =1 refers to methanol, n = 2, ethanol, etc. The complementary process of steamreforming (S-R) of alcohols (the H 2 release step) is simply the reverse of synthesis, so it will be convenient to consider both together. The reaction above parallels the conventional synthesis from syngas:n CO + 2n H2 ↔ CnH2n+1OH + (n-1) H2O but contains an extra term, as revealed by subtraction:n CO2 + n H2 ↔ n CO + n H2O which corresponds formally to the reverse water-gas shift (WGS) reaction. As this is endothermic (∆H° ≈ +40 kJ.mol -1 [57]), alcohols synthesis from CO2 is less favourable than from CO under standard conditions. This parallel also means that, just as for syngas conversion [41], there are many alternative reactions of CO 2 hydrogenation which are more favourable than alcohols synthesis, so that the process must run under kinetic control, i.e., the catalyst must be selective for alcohol products. Likely side-reactions include:1.

Methanation CO2+ 4H2 ↔ CH4 + 2H2O

2. Fischer-Tropsch

a. saturated hydrocarbons n CO2 + (3n+1) H2 ↔ CnH2n+2 + 2n H2O b. unsaturated hydrocarbons n CO2 + 3n H2 ↔ CnH2n + 2n H2O

and reactions to various oxygenated products, e.g., ethers, aldehydes, ketones and carboxylic acids [41]. Some representative plots of free energies of reaction over a range of temperature (at 1 bar) are shown in Fig. 2 overleaf. These plots are based on standard (JANAF) thermochemical tables [58]. It can be seen that CO 2 hydrogenation is normally favoured at low temperatures where the (negative) enthalpy term, ∆H, dominates in the general expression for free energy:∆GT,p = ∆HT,p − T∆ST,p

..1.

which must be around zero, or preferably negative, for the reaction to proceed to any great extent. As a class, the exothermicity is associated mainly with the formation of excess water per mole CO2 converted, which is maximized for saturated hydrocarbons. The case of methanol synthesis is marginal (∆H 0 ≈ −50 kJ.mol-l), in which the lower enthalpy content of the oxygenate itself (relative to its fully-hydrogenated C1 analogue, CH4), can be considered as a partial driving force along with water formation. Ethanol synthesis is an intermediate case, as may be better appreciated by consideration of the collective change (decrease) in bond energy and the relative contributions to stability from the molecular species involved [59]. As the entropy term in eqn. 1, T∆S, works against ∆H (∆S is invariably negative, i.e., disfavouring reaction, due to the volume reduction), there is a small increase in free energy (∆G ≈ +3.7 kJ.mol -l ) for methanol synthesis at close to standard conditions (see Fig. 2, T = 300 K). Purely on this basis, synthesis processes which operate at low or ambient temperature are the most viable, i.e., the least likely to suffer from equilibrium limitations. It should be appreciated that small positive ∆Gs do not necessarily rule out process viability. Indeed, the equilibrium yield of methanol at 300 K and 1 bar is almost workable (25%) due to the Law of Mass Action, which relates conversion to the square root of K, the equilibrium constant. In general terms, this factor acts as a compensating effect on yield, increasing rapidly with the molecularity of the reaction in question.

Fig. 2 Free energy changes for selected reactions in CO2 hydrogenation at 1 bar The effect of pressure as an operational variable is given by the relation:∆GT,p = ∆GT,p0 + RT ln (p0/p)∆n

..2.

where the exponent, ∆n, is the difference between the molecularities of reactants and products (∆n = +2 for methanol synthesis). For positive values of ∆n (volume shrinkage), the second term in equation 2 is negative, resulting in more favourable thermodynamics. This is the mathematical basis for the well-known Le Chatelier principle, i.e., the shift in equilibrium tends to relieve any imposed constraint. The exponential dependence on pressure can have major effects on conversion. Thus, application of only modest pressure (20 bar) in methanol synthesis can increase the potential yield to >99% at 300 K. Unfortunately, the best commercial heterogeneous catalysts to date for syngas conversion, based on supported Cu, work at 500 K or higher due to kinetic limitations, i.e., activation energy barriers to reaction [41]. To maintain a workable convers-

ion of CO2 at this temperature (say >40% per pass, resulting in >90% yield after 5 passes with external recycle and product removal) would require pressures approaching 100 bar. Alternatives to raising pressure in order to overcome equilibrium limitations will be discussed later. Ethanol synthesis is clearly favoured over methanol, with a “crossover” temperature (∆GT,0 ≈ 0) of ∼470 K. This can be increased to 600 K if necessary, to allow for catalyst activity limitations, by operating at 20 bar. The chief difficulty in ethanol synthesis from syngas is lack of selectivity at present, but starting from pure CO 2 as C-source may have advantages in this respect (vide infra). The corollary is that the reverse process, i.e., steam-reforming (S-R), is easier for methanol than for ethanol. Nevertheless, from the thermodynamic viewpoint, both alcohols can be reformed at temperatures low enough (T < 750 K) for compatibility with normal heatexchanger systems based on recycled exhaust heat from a H2 combustion engine. The resulting

- 13 increase in exergy ( energy available for useful work) is both substantial and readily exploitable (vide infra). From the foregoing considerations, it can be concluded that simple alcohols are ideal compounds for a cycle of H2 storage and release. By judicious selection of operating conditions, alcohols synthesis and steam-reforming are viable in a temperature range (450-750K) which appears within the scope of advances expected in heterogeneous catalyst development (better activity, stability, & selectivity) and reaction engineering in CO2 hydrogenation and alcohol fuels processing. What follows is a review of the present stateof-the-art in this branch of applied catalysis and its future prospects. 3. Trends in carbon oxides hydrogenation 3a. Industrial & technical aspects 3a.1 Methanol Recent trends in syngas conversion at the industrial level have been reported by Mills [41]. The main commercial objective at present is the synthesis of “clean” liquid hydrocarbon and oxygenate fuels, as driven by environmental factors, e.g., the absence of sulphur and aromatics (benzene), and the value of oxygenates as blending agents in gasoline to reduce CO, ozone, etc., for compliance with ever more stringent emissions control legislation (see also p.4). In the latter case, the main additive is MTBE, manufactured from methanol and isobutene, because methanol itself is not sufficiently fungible (miscible with gasoline) and also has a relatively high vapour pressure. The risk of violating legal thresholds for evaporative emissions from blends is consequently higher, thus restricting its usefulness. On the other hand, commercial growth in the MTBE market has been phenomenal, increasing roughly fourfold in the last 8 years, to 400,000 barrels per day. This growing outlet for methanol as a commodity chemical, which now accounts for 30% of production, is arguably the main cause of present undersupply and the economic driving force

for rapid expansion of the industry [60-62]. Industrial methanol synthesis from syngas is a classic example of a process operating under kinetic control, with a selectivity of >99% to the least-favoured product in thermodynamic terms. The main catalyst component, metallic Cu, almost invariably supported on ZnO (with lesser amounts of Al2O3, present as structural promoter) is considered responsible for this remarkable selectivity because of its moderate “hydrogenating power” and ineffectiveness in CO dissociation. Even the recent insight gained on the mechanistic importance of CO2 (see p.7) has apparently not led to radical changes in catalyst formulation. The “CO2tolerance” of present catalysts is claimed to be associated rather with greater resistance to sintering of the CuO precursor, or the supporting oxide, by excess steam during initial activation and/or on-stream. The most successful development reported has been the incorporation of oxides with better hydrothermal stability, e.g., MgO in the ICI 51-7 catalyst, and similar matrix stabilizers in Süd-Chemie's C79-5GL and -6GL types, which are claimed to produce a higher initial Cu crystallite dispersion, and greater stability on-stream [63]. How this could overcome the suppressing effect of higher levels of CO 2 observed in earlier forms remains a matter of speculation, but it is likely that certain promoters have been found which act more directly on the Cu, possibly influencing its average oxidation state in-situ. The immediate incentives to the industry of developing catalysts which accommodate higher CO2 levels in syngas are to anticipate carbon taxation and its effects on technology reorientation [42], and future shifts in syngas composition supplied from, e.g., gasification of coal, integration of off-gas recycling from other industrial processes [64], and the growth of biomass as potential feedstock for liquid fuels. Processing of the biomass by gasification, pyrolysis, etc., will be done in tomorrow’s “biorefinery”. This branch of applied catalysis will soon become important for technoeconomic reasons [65].

- 14 A promising technical advance in catalyst preparation concerns the Raney method, which was first reported in the mid-twenties. Massive, skeletal metal catalysts such as Ni and Co, made by selective leaching of the Al component from the respective M/Al binary alloy precursors, are best-known for niche applications in liquid-phase or lowtemperature processes in which the perceived risk of sintering of these self-supporting forms is low. What is not yet generally appreciated is that they offer broader scope as practical catalysts even at elevated temperatures. For example, the value of Raney Ni in methanation has been demonstrated, with working lifetimes >1000 h at temperatures up to 700 K [66]. In recent years, Australian scientists adopted the method to produce ZnOpromoted Raney Cu catalysts for methanol synthesis. Patented in Germany in 1993 [67], they appear to offer advantages in terms of activity and selectivity relative to traditional formulations, though the question of long-term stability, arguably the major factor for industrial suitability, is still open. Their high activity is thought to derive from a unique pore-size-distribution, which can be “tailored” in the leaching stage, leading to a superior effectiveness factor in pellet form. Additional features include better product selectivity, ease of pelletization, and, most interesting of all, evidence for more selective action on the CO2 component of syngas [68]. Extension of the “Raney approach” to bimetallic skeletal catalyst preparation is described in a later section. Industrial activities have focused more on reaction engineering and especially ways to overcome equilibrium limitations. Modern commercial operation of methanol synthesis is restricted to only 25 vol% per pass due to thermodynamic factors, resulting in dependence on costly recycle systems. Potentially more economical ways of increasing process efficiency include selective removal of methanol in situ by, e.g., adsorption [69], solvation, condensation, and co-production of dimethyl ether [41]. The last two approaches, developed by Haldor-Topsoe in Denmark, hold great promise for “renewable petrochemistry" based on a CO2/H2 feedstock. The “condens-

ing product” approach requires that the reactor temperature be lower than the dew point value of the product at the working pressure in the exit stage of the reactor, as given by the Antoine relation. In the case of methanol from a typical syngas feedstock (CO/CO 2 ≈ 10), P ≈100 bar, T 200 °C and P > 20 bar.

3a.2 Ethanol (and higher alcohols) 3a.2A: Direct routes There is no current industrial process for the selective production of ethanol directly from syngas. Historically, alcohols were first encountered as minor products of the FischerTropsch (FT) process under certain conditions. Higher Alcohol Synthesis (HAS), as now practiced, is aimed at a mixture of alcohols with optimum properties in terms of octane number, vapour pressure, fungibility, etc., to serve as blending agents for gasoline. A typical alcohols range i s t hrough C 1 − C 6 , but with wide variation in the carbon number /isomer distribution, depending on the manufacturer. The market continues to be dominated by MTBE, but this could change if current fears of its potentially harmful side-effects are justified [76]. Companies or conglomerates involved in HAS are Snamprogetti/Enichem/ Haldor-Topsoe (SEHT), Institut Français du Petrole/Idemitsu Kosan (IFP/IK), Lurgi/SüdChemie, and Dow Chemicals [41]. All processes operate in a temperature range from 520650 K (at 100 bar), give low CO conversion per pass (∼20%), but offer > 90% selectivity to alcohols. Unfortunately, all except the lastmentioned lead to methanol as the predominant product (∼60 wt%). This is perhaps not surprising as the main catalyst component is alkalidoped, oxide-supported Cu, modified either by inclusion of Cr (SEHT), Co or Ni (IFP/IK), or just K+ (Lurgi/Süd-Chemie). The Dow catalyst is the only exception, being based on MoS2 (with Co9S8 and K+ promoters). It is also the most promising because it is the only one which allows methanol recycle to zero, raising ethanol selectivity to >60%. It is thought that the Co promotes homologation of methanol in this case [77]. The role of alkali promoters (Li + , K + , Rb + , Cs + ) is to inhibit methanation and, to some extent, also methanol synthesis. Their value is well illustrated in the case of the IFP/IK catalyst based on Cu/Ni, as opposed to Cu/Co. Whereas Co is known to promote chain growth and finds ubiquitous use in FT processes, Ni is typically

- 16 the metal of choice for methanation. Nevertheless, K+ promotion allows substitution of Co by the cheaper Ni with no increase in methane production, the major side-reaction in HAS [78]. Although supported Rh is known to selectively catalyze ethanol synthesis under certain conditions [79], and extensive interest has been shown in promoted Rh by the Japanese [80], the economic outlook for this precious metal is not encouraging as there is considerable over-demand for its use in exhaust catalytic converters. Unlike the case for methanol synthesis, evidence for a major role of CO 2 in Higher Alcohol Synthesis has not yet been clearly established although it is known that the process is more effective, at least over Cu/Co, with a high proportion of CO2 as C-source in syngas [81]. 3a.2B: Indirect routes Considerable interest is being shown at a technical level into the possibilities of indirect routes to ethanol. The main reason is that the direct (HAS) process from syngas nearly always produces excess methanol. Otherwise, the product distribution parallels that observed in FT synthesis, in which chain growth conforms to the predictions of Anderson-SchultzFlory polymerization kinetics [82]. The main advantage of an indirect route is the potentially greater selectivity achievable by comparison. Plausible routes for ethanol synthesis from a hypothetical renewable H2/CO2 feedstock are shown overleaf (Fig. 3). There is now clear evidence that the synthesis of methanol from pure CO2 as C-source (step 1 in Fig. 3) is now feasible, at least on a laboratory-scale. Recent flow reactor studies by workers at the ABB Corporate Research Centre in Switzerland have confirmed that respectable methanol yields (5-10% per pass) are obtainable at space velocities of industrial relevance over a range of supported Cu catalysts (from commercial and academic sources) under mild reaction conditions, e.g., T ∼500 K. P < 30 bar, and syngas compositions ranging from H2/CO2 = 3-10 [83]. In fact, calculations

indicate that these yields are remarkably close to equilibrium values, emphasising the excellent quality of existing catalyst formulations. Plans are now imminent to explore some of the methods to overcome equilibrium limitations, as described above (see p.14). A promising multi-step route from methanol to ethanol may start with dehydration to DME (step 2a). As already stated, this is an effective way to improve the economics of syngas conversion to oxygenates and is technically close to maturity. The second step (2b), isomerization of DME to ethanol:CH3OCH3 → CH3CH2O

..8.

is perfectly feasible (∆H° = −56 kJ.mol-1), and under intensive study [41, 84]. Although DME is considered rather inert as ethers go, it is known to be an active intermediate in the industrial (Mobil) process for conversion of methanol to gasoline (MTG) over an acidic (ZSM-5) zeolite, in which it is believed to be degraded to carbonium and, subsequently, carbene species which promote chain growth [85]. Chemical intuition would suggest that catalysts which are selective for isomerization will have basic or superbasic properties. Abstraction of a proton from a methyl group of the ether would leave a carbanion intermediate which may readily undergo a 1,2-rearrangement to ethoxy [86]. Once again, CH4 may constitute the least desirable side-product from the reaction:CH3OCH3 → CH4 + HCHO

...9.

though this is almost thermoneutral, and so lessfavoured energetically than isomerization. Carbonylation of methanol (step 3a) leads to methyl acetate (MA) via esterification of acetic acid intermediate:CH3OH + CO → CH3COOH

..10.

CH3OH

CH3COOH → CH3COOCH3 + H2O ..11. Hydrogenolysis of MA (step 3b) produces ethanol and methanol as co-products:CH3COOCH3 + 2H2 → C2H5OH + CH3OH .12.

Fig. 3. Schemes for the selective synthesis of ethanol from “renewable” svngas While being a valuable and safer fuel mixture than methanol per se, its composition may be further enriched in ethanol (even to 100%), depending on the degree of methanol recycling to step 3a. Concerning toxicity, it is known that the presence of ethanol delays or prevents intoxification of the human body by methanol because the former is selectively metabolized by alcohol dehydrogenase. Indeed, orally administered ethanol is one standard procedure to treat patients with methanol poisoning [87]. This processing option was originally proposed by the Halcon SD Group [88], but has been further developed by researchers at the Korean Institute of Science and Technology (KIST), who are motivated by the economic potential of converting low-grade carbonaceous resources (coal, biomass, etc.) to liquid fuels [89]. KIST has developed its own (noble-metal) catalyst for the vapourphase carbonylation step whilst hydrogenolysis is facile over commercial Cu/ZnO. Plant investment and operating costs are potentially lower due to the relatively mild conditions required for the reaction sequence above (1012). The economics of ethanol thus produced

are claimed to be competitive with those for direct fermentation of biomass. A marginal alternative is the so-called MTO (methanol-to-olefins) process developed at Mobil, currently demonstrated on the 100 barrels-per-day scale. This is a variant of the MTG process, based on HZSM-5 zeolite or aluminosilicophosphate catalysts with zeoliterelated structures like SAPO-17 and SAPO-34. Selectivity to olefins (typically C2-C4) is optimized (to >90%) by decreasing methanol partial pressure, increasing temperature, or changing the SiO2/Al2O3 ratio [90]. However, selectivity to ethylene remains too low (∼60%) to be of commercial relevance. If this can be improved appreciably in the near future, C2H4 can then be easily hydrated to ethanol (step 4b) by established technology. 3b. Fundamental Advances 3b.1 Methanol 3b.1.1 CO2 activation: surface science view As in metal-complex chemistry [91], the key initial step in activating the rather stable,

- 18 -

linear CO2 molecule via adsorption on a metal surface may be to encourage coordination through the C-atom, which has Lewis acid character, i.e., it is electrophilic. Molecular orbital (MO) theory predicts that both the first excited, neutral, state and anionic radical forms, CO2δ− (δ = 0-1), should adopt a bent configuration to minimize the repulsive effect caused by electron occupancy of the next available (2πu) orbital, which has strong anti-bonding character. In many cases, e.g., the carboxylate moiety of organic species and coordination complexes, a predicted C-O-C bond angle of about 135° is actually observed. From recent surface science investigations, conclusive evidence for the CO2δ− ad-species (based on its vibrational spectrum by highresolution electron energy loss spectroscopy, HREELS) has been obtained over various transition metal single- crystal surfaces, including Ni, Fe, and Re [92], Cu [93], Rh [94], Pd [95], and Ru [96]. However, apart from the first three metals listed, in which surface steps and “roughness” are believed responsible, all others require the presence of alkali metal (AM) promoters to act as electron donors. The CO2δ− intermediate, which forms already at sub-ambient temperature, is highly reactive. It has been shown to dissociate spontaneously below 450 K to yield adsorbed CO (and O), either directly or via formate, HCOO δ− , which is readily formed over Ni by co-adsorption of H2 and CO2 [92]. Coadsorption of CO and CH 3 I (as the source of methyl radicals) produces acetate, CH3COOδ−, though it is not yet clear if this occurs directly from CO 2 or via CO. In any event, the evidence for C-C bond formation is strong. This point will be taken up again later. 3b.1.2 Methanol synthesis & water-gas shift An appreciation of the fundamental concepts outlined above is clearl y important in a general sense. However, the relevance of direct activation of CO 2 , i.e., that which may occur in the absence of H but requiring alkali promoters, for the mechanism of methanol syn-

thesis from CO2 is questionable. In actuality, alkali metal doping (AMD) of conventional (Cu) methanol synthesis catalysts is a recognized method to suppress methanol formation in favour of higher alcohols and oxygenates [41, 97], although promotion has been claimed for syngas containing no CO2 [98]. Thus, the role of AMD in promoting CO2 activation (and decomposition to CO) over Cu [93] and related metals probably has most relevance to the mechanism of higher alcohol formation, in which CO plays a major role [97]. It should also be appreciated that the now well-documented effects of AMD on a range of adsorbents and adsorbates [98-100], including those of relevance here (H2, CO, and CO2), are complex and difficult to unravel in-situ. Since the recognition that CO2 is converted to methanol faster than CO, at least over the industrial Cu/ZnO/Al203 catalyst, considerable attention has been devoted to the unravelling of the mechanism of methanol synthesis over Cu (both supported and unsupported), and promoted group VIII metals like Pd, which also show significant activity [101]. However, universal agreement on the mechanism has not yet been reached even over supported Cu. This i s p a rt l y b e c a us e t h e be h avi ou r o f t h e industrial formulation may not be truly representative (other catalysts like Cu/CeO2, derived from alloys, appear to act preferentially on CO [102]), and also because of the mechanistic complexity itself, in particular the rapid interconversion of CO and CO2 in situ due to the WGS reaction. 3b.1.3 In-situ studies and surface intermediates In line with the general trend in experimental catalysis research, techniques now favoured are those which offer some possibility of insitu observation of the catalyst surface under reaction conditions [103]. Prominent among these is Fourier-Transform Infrared (FTIR) spectroscopy, both in transmission mode and its popular recent variant Diffuse-Reflectance Infrared Fourier-Transform (DRIFT) spectroscopy, which is clearly fulfilling its early

- 19 -

promise [104]. FTIR offers excellent in-situ capability, good resolution, sensitivity, versatility and convenience. It is also inexpensive and so available to most laboratories. In this sense, it can be considered the spectroscopic “workhorse” of the modern practitioner of catalysis and surface chemistry. A wide variety of sorbed intermediates have been observed during carbon oxides conversion to methanol over supported Cu by transmission FTIR or DRIFT spectroscopies under realistic conditions. What is not yet clear is ho w m an y o f th ese speci es are involved in the mechanism leading specifically to methanol product, and which of these are kinetically important; viz. carbonate (CO32−), carboxylate (CO2−), bicarbonate (HCO3−), formate (HCOO−), formaldehyde (HCHO), and rnethoxy (CH3O−), on both Cu and the oxide support, e.g., ZnO, ZrO 2 , TiO 2 , SiO 2 , etc. [105-112]. The detection of such an extensive range of intermediates may be due to the fact that methanol synthesis in industrial conditions is essentially under thermodynamic control, and quite evenly balanced as modest yields are typical. Thus, assuming the principle of microscopic reversibility, this also justifies the inclusion of results from studies of methanol decomposition [112], showing evidence for dioxomethylene or H2CO2 (DOM). Further evidence for DOM (and other polymerized forms of HCHO) comes from direct exposure of Cu/ SiO2, as well as various basic and amphoteric oxides [113-115], to formaldehyde, from which methanol is ultimately produced. 3b.1.4 Mechanism: roles of formate/formyl ? The above list appears to represent a progressive sequence of plausible intermediates (with increasing degree of hydrogenation) en route to methanol. On this basis alone, the question arises as to how to exclude any single observable species from the likely mechanistic pathway!? This is contrary to general experience in mechanistic work, which is usually fraught with problems of invoking hypothetical intermediates which must then be hunted down

by good experimentation. A recent example is chemical trapping of formyl, HCO, postulated as an intermediate in methanol synthesis from CO/H2 over supported Pd, in which the oxidized metal (Pdn+ ) is believed to play a stabilizing role [97,100]. On the other hand, direct IR spectroscopic evidence for formyl on Cu, even from CO/H2, is still rather tenuous [116] and has not been substantiated by the more recent wealth of studies reported on here. For all their obvious advantages, in-situ studies are by definition holistic, providing an often complex picture of the catalyst surface mainly during steady-state operation. Only by transient methods, in which the rate of disappearance of suspected intermediates may be monitored and correlated (qualitatively and preferably quantitatively) with the appearance of other intermediates and ultimately the desired product, can the kinetically-relevant pathway(s) be isolated and verified [see, e.g., refs.117a-d]. Such an approach is still somewhat neglected in surface chemical research, and many of the works cited above are no exception. These limitations aside, there is now strong circumstantial evidence from data accumulated over many years up to today that formate on Cu is a pivotal intermediate in both methanol synthesis:CO2 + 3 H2 → CH3OH + H2O

..13.

and the reverse WGS reaction:CO2 + H2 → CO + H2O

..14.

In each case, Cu is the most-active metal known [118-119], such that these reactions are often considered together [106, 118,120-121]. It has been found that optimization of Cu/ZnO activity for the (assumed) creation of formate accelerates both reactions (13 and 14), although reverse WGS is typically two orders of magnitude faster than methanol synthesis in-situ [106,122]. This rapid interconversion of CO2 and CO under synthesis

- 20 -

CO2 initial rate

200 150

100

(a)

− o− RC

−• − R CD −◊ − CATX

50

CO initial rate

0 60 50 40 30 20 10

Ratio of initial rates (rCO2/rCO)

conditons, which complicates the overall chemistry, is another reason for the “late realization” of the potential of CO2 as C-feedstock mentioned earlier. Invoking formate as primary intermediate in methanol synthesis provides a simple explanation for the observed (selective) conversion of CO2 as compared to CO; formate is known to be made more easily from CO2 /H2 than from CO/H2 [121,123]. It also provides a rationale for the otherwise paradoxical fact that the addition of CO to a CO2 /H2 syngas mix increases the selectivity to methanol, as recently observed over Cu/SiO2 [106]. Since the overall process is demonstrably in quasi-equilibrium, the addition of CO creates a higher steady-state level of formate (by the WGS process) from water, which might otherwise back react with methanol. This is a typical manifestation of the Le Chatelier principle. What is not yet certain is whether there is a direct, but slower, parallel route to methanol from CO over Cu via, e.g., formyl, HCO, as appears to be the case for Pd (vide ultra), and possibly Rh also. In the latter case, the absence of isotopic scrambling of the C-O bond in methanol (starting from labelled CO) is strong evidence that a symmetrical carboxylate intermediate, such as formate, is not involved [124]. In any event, the existence of a faster route to methanol from CO over Cu via WGS & formate is strongly indicated by the catalytic effect of small amounts of added steam in CO/H 2 [125]. A similar promoting effect of water (or OH− from initial CO2 conversion) on methanol synthesis from the CO component in H2/CO2/CO mixtures has been observed recently over Raney Cu, ZnO-promoted Raney Cu, and co-precipitated Cu/ZnO/ Al2O3, by the Australian group of Wainwright et al. [68,126] as shown in Fig. 4. Its secondary effect of reducing the commonly observed induction period in CO conversion provides an alternative view (in terms of a mechanism via formate) to that propounded earlier by the ICI group, viz., the autocatalytic effect of a slowly accumulating adlayer of O over the Cu [127].

Fig. 4 Initial conversion rate [mol.m - 2 h - 1 x 10 5 ] for (a) CO 2 , (b) CO, and (c) their ratios; vs. CO2/CO composition. RC = Raney Cu, RCD = ZnO-doped Raney Cu, CATX = Cu/ZnO/Al2O3 (with permission from ref. [126])

The positive effect of AM promoters, such as K+ and Cs + , on methanol synthesis from CO2 free syngas [98] is probably linked to direct activation of CO on Cu, evidence for which has been reported [93].

CO2/CO = 0.25

CO2/CO = 0.25

CO2/CO = 0.25

CO2/CO = 0.50

CO2/CO = 0.50

CO2/CO = 0.50

CO2/CO = 1.15

CO2/CO = 1.15

CO2/CO = 1.15

CO2/CO = 2.0

CO2/CO = 2.0

CO2/CO = 2.0

Fig. 5 Relative conversion of syngas constituents, CO (o), and CO2 (•), from mixtures of various stoichiometry in methanol synthesis over (a) Raney Cu, (b) ZnO-promoted Raney Cu, and (c) commercial (coprecipitated) Cu/ZnO/Al2O3 catalysts. Reproduced with permission from ref. [126].

3b.1.5 The CO2/CO synergy A synergistic effect between CO2 and CO in methanol synthesis has been recognized for a long time, but was interpreted earlier as due to the action of CO2 in maintaining the optimum balance between oxidation states of Cu (Cu0 and CuI, the latter possibly stabilized by adsorbed O), to which the activity for CO conversion was attributed [40, 127]. Although the potential importance of Cu+ or Cuδ+ is still a

matter of debate, the most recent evidence casts serious doubts on the O adlayer theory. From DRIFTS experiments in which any Oad was presumed to exist as surface carbonate in a CO2/H2 environment the O coverage under genuine synthesis conditions was inferred to be only 1-2% of a monolayer [105]. On the basis of independent studies, exploiting step changes and pulse additions of CO2 and CO in

- 22 syngas mixtures during methanol synthesis, Danish workers concluded likewise [128]. In view of this, the most plausible explanation of the observed synergy now lies in the combined kinetic and thermodynamic effect of steam on CO2, and CO on CO2, respectively. To summarize in equation form:CO2 + 3 H2 → CH3OH + H2O

..15.

CO + H2O → HCOOH

..16.

HCOOH + 2 H2 → CH3OH + H2O

..17.

CO2 + CO + 5 H2 → 2CH3OH + H2O

...18.

The balance (eqn. 18) is equivalent to the sum of the two formal expressions for methanol synthesis from CO2 and CO. The stoichiometry number for this reactant composition (given by [nH2 − nCO2]/[nCO + nCO2]) is equal to 2.0, which also fits with the working “rule of thumb” in industrial practice of maintaining a value in the range 2.03−2.10 to optimize the global synthesis rate from various pre-reformed feedstocks, which are trending to CO2 -rich [63-64]. In the work of Sizgek et al.[126], methanol yields from syngas mixtures, in which the H 2 /(CO2+CO) ratio was kept fixed at ∼6, the best results were obtained at a CO2/CO ratio from 0.5-1.0 (see Fig.5). This is fairly consistent with the origin of synergy as proposed above. 3b.1.6 Kinetics: active role of support ? Although the above picture does not exclude an important catalytic function attributable to oxidized forms of Cu, recent studies of unsupported single crystal and polycrystalline samples, in which the surfaces were kept scrupulously free of contaminants, found little evidence for Cu+ and specific activities were similar to industrial Cu/ZnO catalysts[128130]. In other words, the free Cu metal surface is believed to be directly responsible for methanol synthesis activity, which suggests that the function of the support is of a minor, or passive, nature. Consistent with this argument, many reports claim that

the activit y is simply proportional to the total number of surface Cu atoms, i.e., it is a structure-insensitive reaction [118]. However, the question of the surface cleanliness of Cu in the working catalyst is as yet unanswered. What is known is that site-specific synthesis rates over unsupported Cu can be several orders of magnitude lower than over supported Cu [132], an effect recently attributed to surface contamination by C [131]. In view of such extreme sensitivity of methanol synthesis to Cu surface cleanliness, this is at least partial justification for invoking a more versatile role for the support, either in terms of selective trapping of contaminants to help maintain a clean metal surface, or as an auxilliary pool of reactive intermediates to compensate for an inevitable level of contamination under process conditions. Certainly, the evidence for an active role for the support is also very strong. It may well provide an explanation for the apparent ease of formation of formate on Cu in the absence of AM promoters, i.e., without prior activation of CO2 (e.g. via a metaladsorbed CO2δ− species). It is well known that formate (in bi-dentate and mono-dentate coordination) is readily-formed from CO2/H2 over those pure oxides (basic or amphoteric) which are also the best supports for Cu in terms of methanol synthesis activity, i.e., ZnO [133-138] and ZrO2 [107,136-139]. Elsewhere, evidence from transient methods that formate, initially on TiO2, serves as an active supply of (Ru) metaladsorbed CO (via the reverse WGS reaction) and subsequent methanation, suggests that migration of formate to the metal or the metal /support interface is facile [140]. Corroborative evidence for transport of ad-species between the metal and support is that of ZnO in exerting “action at a distance” in increasing significantly methanol synthesis activity over Cu/SiO2 [141]. This was later attributed to its role as an active reservoir of H atoms “spiltover” from the Cu [142]. The clear superiority of such oxides, compared to clean metals, in activating CO 2 is understandable. The adsorption heats are higher by virtue of the amphotericity of the oxide; the Lewis acid sites (cations) attracting the basic O atoms of CO2 and the contiguous surface O anions acting

- 23 -

0

0.035

absorbance

0.030 0.025 0.020 0.015 0.010 0.005

The site-specific rate or turnover frequency (TOF) for methanol synthesis (see Fig. 7) also passes through a maximum but slightly later (∼520 K), confirming its qualitative link to formate.

TOF x 105 (s−1) Methanol synthesis

5.05,0

1.4 1.2

 TPR @ 1 K per min • Isothermal

4.0

1.0

3.0

0.8 0.6

2.0

TOF x 103 (s−1) Reverse WGS

on the electrophilic C atom (see p.18). By the same token, their propensity to activate H2 by heterolytic dissociation, to H+ and H− ions, is well documented. The hydride species on surface Zn ions has been identified by IR spectroscopy [143]. Thus, a low energy pathway to formate is created via simple association of H− with the electrophilic C atom in adsorbed CO 2 . Although it would appear that formate can form directly over pure Cu, an alternative mode of provision, via such an “auxiliary reservoir”, may become kinetically significant under conditions which would otherwise lead to (its) depletion at the active site(s). This may be due to an insufficient rate of formation directly on the Cu component, or to increasing “losses” via competitive reaction(s). As already noted, predominance of the reverse WGS reaction (over methanol synthesis) from formate is the obvious “sink”, at least under differential conditions. Consistent with this viewpoint, Clarke and Bell [106] have monitored the steadystate surface coverage of formate (θ HCOO ) and other species in the reaction of CO2 with H2 over Cu/SiO2, at increasing temperature and fixed pressure (∼7 bar). This reaches a maximum around 410 K whilst at typical synthesis conditions (T >500 K) it is lower by roughly an order of magnitude, as shown in Fig. 6 below.

0.4 1.0

0.2

300

350

400

450

500

550

600

T (K)

Fig. 7 TOFs for methanol synthesis and reverse WGS reaction over 7% Cu/SiO2 under the same conditions as in Fig. 6. Reproduced with permission from ref. [106].

Curiously, the appearance of a maximum in rate must also mean that an apparent activation energy (Eapp) measured in a temperature range straddling Tmax would be around zero or even negative! If confirmed, such anomalous behaviour is arguably the strongest evidence yet for the pivotal role of formate in both methanol synthesis and reverse WGS reactions. This is currently under investigation in our ABB laboratory. In contrast, the TOF for CO production shows “normal” behaviour, at least insofar as it increases continuously up to the limit explored (∼600 K). This is to be expected as there is effectively no significant competition (for formate) from the synthesis pathway. 3b.1.7 Kinetic parameters/rate-determining-step

0.000 300

350

400

450

500

550

600

Fig. 6. Temperature dependence of ad-species f orm e d o n 7 % Cu/ S i O 2 i n H 2 / CO 2 / A r (3.3/1/1) stream at 7.2 bar, as monitored by FTIR spectroscopy [b: bidentate, m: monodentate form]. Reproduced with permission from ref. [106].

Evaluation of global kinetic parameters like apparent activation energies (Eapp), rate dependence on reactant partial pressures, isotope effects, etc., and comparison with data for (ideally) isolated sub-reactions implicated in the mechanism, is a popular means to identify the rate-determining-step (RDS), assuming that

- 24 -

only one critical transition-state is involved. Unfortunately, any attempt to verify the RDS in methanol synthesis, which requires quantitative evaluation, faces severe practical obstacles. As the foregoing evidence shows, even Eapp values from CO2/H2 determined in the normal temperature regime may have almost no validity due to the progressive drop in θ HCOO with rising t emperat ure. This is, in turn, due to growing dominance of the reverse WGS reaction, as reflected in the observed selectivity shift towards CO [106]. Eapp values for the latter will also be suspect for the same reason, except that the cause of the depletion in surface formate must in this case lie in supply limitation. In other words, the rates of supply and consumption of formate in reverse WGS appear roughly balanced at 410 K. Beyond this point, the RDS becomes progressively associated with the supply step and manifests predominantly its kinetic behaviour. An inevitable result is that the Arrhenius plot will yield an E a p p value whi ch is a weighted average of the two “pure” exponents, i.e., those for supply and consumption. A practical consequence of this effect, which is of general validity in kinetic studies, is that the measured Eapp will approach that of the least temperaturesensitive step in the mechanism, i.e., that with the lower Eapp, evidently supply in this case. Recognition of such effects helps to explain the wide range of Eapp reported for the reverse WGS reaction, as exemplified by two recent values of 64 kJ.mo1−1 on Cu/SiO2 [106] and 135 kJ.mol−1 on polycrystalline Cu. The latter value is derived from the work of Yoshihara et al. [131], whose Arrhenius plot is reproduced in Fig. 8. Most values fall typically in the range 70-110 kJ.mo1−1[144], but these are probably more representative of the supply step, as discussed above. The exceptionally high value for polycrystalline Cu can be rationalized within the above framework because the estimated surface coverage of formate in this case was close to a full monolayer, or unity. Under such conditions, consumption appears to be rate-limiting so that the measured Eapp is closer to the intrinsic value for formate decomp-

osition to CO. Whether such a high value is representative of the same process on supported Cu, i.e., the normal shift catalyst, is not yet clear. Temperature / K 550

500

450

CO

methanol

Fig. 8 Arrhenius plots of TOF for methanol synthesis and reverse WGS reaction over polycrystalline Cu foil at 4.67 bar H2 and 0.41 bar CO 2 . Reproduced with permission from ref. [131].

There is the added complication that both the reverse and forward WGS reactions are known to be structure-sensitive [118,131]. The 110 face of Cu is significantly more active than the 111 face (supposed to predominate in polycrystalline forms), and has a much lower Eapp, 70-75 kJ.mo1−1, in reverse shift [145]. In CO2 methanation over Ru/TiO2, Eapp for formate decomposition (to CO) was close to that of the consecutive process of CO hydrogenation. This was concluded because the steady-state coverage of CO on Ru (θ CO) was almost constant over a wide temperature range, showing the fortuitous operation of balanced supply/consumption kinetics. A simple model was used to show that the rate constant (and thus Eapp ) for methanation was that corresponding to the step of formate conversion to CO. A value of 79 kJ.mol −1 was measured, and an estimated value of 83 kJ.mol−1 (consistent with the literature) for hydrogenation of CO on

- 25 -

Ru was extracted from the global value and the small drop in θCO with increasing temperature [117d]. A provisional Eapp value for formate decomposition to CO over oxide-supported Cu can be estimated from the above value for Ru/TiO 2 (∼80 kJ.mol −1 ) with adjustment for the different heat of adsorption (∆Had ) of CO on the respective metals. [The support oxides are here assumed to be substantially hydrated under typical experimental conditions so that the H2O co-product is either weakly bound or readily desorbed. In any case, its contribution to Eapp can be neglected to a first approximation]. Published values of ∆H ad for CO on Al 2 O 3 -supported metals [146] indicate that it is ∼40 kJ.mol −1 more exothermic on Ru than Cu. On this basis, Eapp for formate decomposition on supported Cu can be set in the range 110-130 kJ.mol−l. As expected, this is significantly higher than the average of measu red v alues (∼90 ± 10 kJ . mol − l ) but app roaches that of Yoshihara et al. [131] for unsupported Cu, which is believed to be quite accurate (vide ultra).

from CO2/H2 over Cu/SiO2 has been shown to be ∼20 times faster than from CO/H2, even at equal partial pressures of the respective C sources [106]. Limited evidence for the buffer effect comes from the same work, which has shown the beneficial effect of CO (in CO2/H2 ) on the synthesis rate and on θHCOO in steadystate. The latter increases by roughly 40% relative but its band shape, which is Tdependent, and position (Tmax) remain essentially unchanged. The synthesis rate enhancement (measured at the maximum, T ∼550 K) scales quite well with θ HCOO . This partial improvement is probably due to the reaction of CO with adventitious ' traces of H 2 O in the gas stream and/or surface hydroxyl groups likely to be present after pre-reduction in H2. Unfortunately, Eapp for synthesis in the presence of CO was lower than in its absence, viz., 71 and 88 kJ.mol−1, respectively, just the opposite of the predicted trend. Nevertheless, the inclusion of H2O vapour in the reactant stream, along with CO, is the definitive condition for experiments to test the hypothesis.

Concerning methanol synthesis from CO2, the above line of reasoning leads inexorably to the paradoxical conclusion that the most reliable values of Eapp will be those obtained from studies in which competition from the reverse WGS reaction is effectively “decoupled”. This can best be achieved under differential conditions by starting with a supply mixture of CO2/CO/H2/H2O in which the stoichiometry matches that of a CO2/H2 mixture that has been pre-equilibrated via WGS. As the latter is insensitive to operating conditions, the composition can be approximated by equilibrium values at the temperature of typical industrial synthesis, or ∼530K [63]. This is only 1-2 vol% each of CO and H 2 O in CO2/H2 [57] but it may be crucial. Only under these conditions can the “buffer” effect on θHCOO be fully exploited, and it has not yet been investigated to the knowledge of this author. The formate coverage should obviously be monitored in-situ and estimated, e.g., by DRIFTS. There seems to be little danger of interference by any direct route to methanol from CO because the synthesis rate

Most literature values of Eapp for methanol synthesis over Cu relate inevitably to simulated industrial syngas mixtures (CO2/CO/H2), such that they may, fortuitously, also be more reflective of CO2 conversion than might otherwise be the case. They tend to lie in the range 70 −110 kJ.mol−1 [106, 147-148], and are in fact numerically close to those for reverse WGS. However, the RDS is here associated with (a step in) the composite process of formate hydrogenolysis, rather than formate supply. In view of the reported increase in selectivity to CO with temperature [106], this sets an upper bound on Eapp for formate conversion (to methanol) of ∼120 kJ.mo1 −1 , i.e., it must be lower than the best estimate for formate degradation to CO (vide ultra). The best literature values have been obtained from direct monitoring of formate “removal” in H2 (∼6 bar) at very low temperatures (T < 370K) over an unsupported Cu (100) single crystal [147,149], and methanol synthesis over polycrystalline Cu at near saturation coverage by formate [131]. These are 82 and 77 kJ. mo1-1, respective-

- 26 -

ly, the latter being derived from the lower plot in Fig. 8, as shown above. Comparative data for supported Cu catalysts, originating from both industry and our research laboratory, will be available shortly. In light of the dubious value of using apparent activation energies as the primary basis for mechanism elucidation in methanol synthesis from CO2 (at least when determined by routine kinetic methodology with its simplistic assumptions), it is imperative that the RDS be characterized as fully as possible in other respects, e.g., orders of reaction, isotopic tracing, reactive trapping, etc. Among these, the kinetic isotope effect, KIE, is often overlooked in mechanistic studies, which is surprising as it is one of the few convenient experimental methods to derive information directly on the involvement of H transfer in the crucial transition state of the RDS in hydrogenation or dehydrogenation. On the other hand, isotopic labelling to trace the microscopic details of bond breaking/making in the mechanism has enjoyed much greater popularity, although these methods are complementary and can easily be assimilated into a single experiment. The KIE in CO 2 reduction has been explored briefly by Clarke and Bell [106]. An H/D isotope effect is indeed seen in both reverse WGS and methanol synthesis. The suppression of rate in D2 is greater for the former (k H /k D ≈ 3.5) than for methanol synthesis (kH/kD ≈ 1.5). In the reverse WGS, this is clearly a primary effect involving H(D) transfer in the plane of the reaction coordinate. The lower zero-point vibrational energy of the heavier D atom leads to a higher activation energy for the transition state and, thus, slower reaction under otherwise equivalent conditions [150]. There are various mechanistic steps which involve H transfer in both reactions, but an isotope effect can only be manifested if such a transfer forms part of the RDS. Herein lies the intrinsic value of such an approach to mechanism elucidation. In the reverse WGS reaction, this primary KIE is believed to be associated with the supply of formate for reasons already given. As such, it is related either to H−H (D−D) bond cleavage, or

C−H(D) bond formation at the CO2 carbon to yield formate. In the first case, even allowing for the fact that H2 dissociation on Cu is known to be an activated process [151], simple calculation shows that the rate of supply of H(D) atoms to the surface under typical pressure conditions (10 3 −10 4 atoms site−1 s−1 [105]), is still at least three orders of magnitude greater than the TOF for reverse WGS. A reliable TOF value in the literature is ∼5 site−1 s−1 for polycrystalline Cu at 573 K and 1.4 bar [131]. The only alternative is a transition state involving H transfer to CO2 with suitable geometry, i.e., formate production. It is difficult to link the KIE with formate decomposition as there is no evidence to support the notion that adsorbed hydrogen, Had, assists the process. In fact, the evidence suggests just the opposite, i.e., inhibition by Had on supported Cu, Ru and Rh. The first can be gleaned from ref. [106], in which the TOF drops with total pressure, roughly as rate ∝ Ptot−0.8. As the reactant mixture consists mainly of H2, while CO2 is only weakly adsorbed on the predominant lowindex planes of Cu [152], it appears reasonable to attribute the suppressing effect to Had. More convincing data in support of this conclusion, obtained from studies on Ru/TiO 2 and Rh/Al 2 O 3 , is available in the literature [140,153]. This makes sense if the driving force for C−O bond cleavage in formate (and the simultaneous creation of the high energy formyl radical co-product HCO•) is the formation of M−Oad. This process is more exothermic than formation of the corresponding M−OH a d specie. For Cu−O a d , the free energy of formation is estimated to be as high as −240 kJ.mol−1 [127], or almost twice the value for formation of bulk Cu2O. It is presumed that the intrinsically unstable HCOad species will then quickly decompose to CO and Had, whilst Oad is rapidly hydrogenated by the latter to H2O [154]. If C−H bond cleavage in formate precedes that of C−O bond rupture, this constitutes merely the back reaction to H ad and CO 2 , such that formate serves no useful role. The corollary of this view is that the H in formate maintains the carboxylate group in the “activated” (bent) geometry, thence

- 27 -

favouring C-O bond-cleavage, at least in the absence of alkali ion promoters (see p. 18). This is a somewhat controversial issue as some authors [127,154] claim that CO2 spontaneously dissociates to COad and Oad over clean Cu at a rate similar to that measured for reverse WGS, i.e., a dissociative or regenerative mechanism is operating rather than an associative mechanism involving formate intermediate [155-156]. However, it is difficult to incorporate the KIE data of Clarke and Bell [106] satisfactorily into the former scheme. Rationalization of a smaller isotope effect for the route to the more hydrogenated product (methanol) is no easy task as it runs counter to intuition. In this case, although H(D) influences the rate it is not a dominant factor. The value is more typical of a regular secondary effect. Such indirect effects are usually attributed to differences in inductive power of the substituent involved, e.g., D is considered to be “electron-donating” relative to H. Thus, one interpretation is that the DCOOad species is more stable than HCOOad, such that there is a lower probability that C−O bond cleavage (to give CO) will precede a second C−H bond formation step to give dioxomethylene (DOM, H2COOδ−ad). DOM can subsequently react with a neighbour (vide infra) or undergo C−O bond cleavage to yield a new product, formaldehyde (or a related species), implicated in methanol synthesis. The overall effect on the synthesis rate is difficult to predict except in a qualitative sense. An increase in selectivity towards methanol, nominally an inverse secondary effect of D (in DCOO), would partially counteract a stronger primary (suppressive) effect on rate by the second D transfer to form DOM, if this last is actually the RDS. The global result would then be a “moderated” primary effect. The implied greater steadystate level of formate in the deuterated state would be supporting evidence for this conjecture and, thus worthy of investigation. The preceding analysis, and its associated speculation, would be on firmer ground if the

effects of partial pressures of reactants CO2 and H2 on measured rates over Cu are at least qualitatively consistent with those expected from likely candidates for the RDS. Unfortunately, there is as yet a remarkable paucity of data, especially in the case of CO2, and much has to be gleaned from work which provides little in explicit form. Rate expressions for methanol synthesis over the industrial Cu/ZnO/ Al2O3 catalyst are invariably for CO-rich syngas, where a semi-empirical factor is typically included to allow for the enhancement in rate due to the CO2 component. For example, the following rate expression:r = kPCO0.2−0.6PH20.7ΦCO2

..19

was reported by Andrew [157], where P CO is the partial pressure in CO, etc., and ΦCO2 was not explicitly determined. Many kinetic equations show a rate dependence on the product PCO.PH22 [40]. This high order in hydrogen is difficult to explain. In rate expressions for CO hydrogenation to hydrocarbons, e.g., methanation, Fischer-Tropsch, etc., over 1st row group VIII metals, the exponent in PH2 is generally around unity despite severe adsorptive competition from CO [158]. In CO 2 methanation, this kinetic order is even lower (∼0.5), which is expected for a RDS involving H2 dissociation [140, 153]. In CO 2 conversion to methanol, a high-order dependence on total pressure (x in P tot x ) was recently reported by Danish workers from studies over unsupported Cu(100). This was subsequently used in an attempt to identify the RDS by micro-kinetic modeling [130,147]. For pressures ranging from 1 to 4 bar, x was found to be 2.36. This was compared to various model expressions predicting order exponents of < 2, 2.5, 2, and 3, for hydrogenation of formate, dioxomethylene (DOM), formaldehyde, and methoxy, respectively. The RDS was attributed to DOM hydrogenation on the basis of best fit. Although a previous spectroscopic search for the dioxomethylene intermediate was inconclusive [159], more recent FTIR studies of methanol decomposition or direct exposure to formaldehyde have since verified its existence,

- 28 -

as noted earlier (see p.19). Unfortunately, the WGS equilibrium was omitted from the modeling to avoid over-complexity. Regarding the exponents in pressure for the individual reactants, i.e., x and y in the equation:r = kPH2 x.PCO2y

..20.

the method of continuous variations was used, which yields only a “best ratio” for x/(x + y), and not absolute values. The observed maximum in rate near x/(x+y) ≈ 0.6 was taken to support their model as it fits with the predicted exponents x=1.5, y=1. However, the maximum is so broad that all that can be safely concluded is that x and y are both positive, and x/y lies in the range 1.5− 2.5. Thus, the order dependence in H2 pressure is significantly higher than that in CO2 pressure. Setting the value (x + y) = 2.36 (the measured dependence on P tot ) leads to a range of x = 1.4−1.8, and y = 1.0−0.5, which is hardly incisive. Nevertheless, supporting evidence for an order in P H2 as high as 2 has been inferred by Yoshihara et al. [131]. In view of the similarity in catalysts under study (unsupported Cu), they compared their higher rate (at higher PH2) with that obtained by Rasmussen et al. [130,160]. In contrast, kinetic orders in P H2 from studies of supported Cu are much lower. A value of ∼1 over Cu/SiO2 may be gleaned from the work of Clarke and Bell [106], and ∼0.7 over Cu/ZrO2, from the thesis of Froelich [161]. These studies were both conducted with a reaction stoichiometry H2/CO2 = 3, but in the latter case over a wider and higher pressure range. Such a major difference in behavior between pure- and supported-Cu implies, once again, a role for the support in terms of auxiliary supply of Had or formate. That this may not be reflected in the overall synthesis rate has been discussed already. Similar conclusions have been reached by others in the case of Cu/ZnO [105,142, 162]. For Cu alone, low θH is to be expected on the basis of closely similar Eapp values for both adsorption and desorption, viz., ∼55 kJ.mo1−1 [154]. As increasing PH2 will tend to suppress desorption, and assuming a strong (positive) rate dependence on θH, then a high order in PH2

is the likely result. If formate hydrogenolysis is the RDS in methanol synthesis, the rate expression should be essentially independent of PCO2, i.e., zero-order in CO2 pressure. This is certainly the case for CO2 methanation over Ru/ TiO2, in which the conversion to CO is rapid, and the RDS is hydrogenation of the latter [140]. Unfortunately, order exponents in PH2 and PCO2 for methanol synthesis, measured independently, have not yet been reported to the knowledge of this author. In view of their importance in mechanistic work, this is an urgent problem to be resolved. In regard to the reverse WGS reaction, the situation is hardly much better. The most explicit data available comes from studies on Cu (110), but these were made at sub-atmospheric pressures, and at temperatures well above the range for methanol synthesis [154]. Thus, extrapolation to predict the likely behaviour of reverse WGS occurring as a parallel reaction under synthesis conditions is precarious. Nonetheless, it is generally known that orders in both P H2 and P CO2 can vary over a wide range, depending on the reaction stoichiometry. At progressively lower CO2 levels (but remaining super-stoichiometric), the order in PH2 rises from 0−2, while the order in P CO2 itself rises from 0−0.6. However, under conditions most relevant to both methanol synthesis and reverse WGS, i.e., with H2/CO2 = 1-3, the rate can be expressed provisionally as:r = kPH20.5 P CO20

..21.

This bears a striking resemblance to the rate expression for CO2 methanation. A positive half order in PH2 is consistent with formate supply as RDS, passing through a LangmuirHinshelwood transition-state involving H transfer to CO2ad as discussed above. A zero-order dependence on PCO2 is rather surprising in view of its weak adsorption on Cu per se. Ernst et al. [154] reconcile the data, which is inconsistent with their preferred redox mechanism, by invoking an adsorbate-induced surface reconstruction on the basis of an observed discontinuity in the rate dependence (on CO2 pressure) when PCO2 and PH2 become comparably low.

- 29 -

Supporting evidence for a similar rate dependence on PH2 over Cu/ZnO, again obtained at low pressure, is forthcoming from the work of Stirling et al. [122]. Inspection of their data reveals an order in PH2 in the range 0.3−0.5, assuming to a first approximation no contribution by the (small) simultaneous change in PCO2 in their experiments. Perhaps the most useful data is that of Clarke and Bell [106] on Cu/SiO2, as described above. This was obtained in a more realistic pressure range up to 7.2 bar. The overall negative (−0.8) order dependence on Ptot is plausible as it can be interpreted partially on the basis of inhibition of the rate of formate decompos i t i o n b y H 2 , fo r wh i c h t h er e i s s om e evidence, at least over other supported metals (see p.26). A possible inhibitory effect by CO2 cannot be discounted either, and the entire problem needs definitive investigation. In summary, increasing reaction pressure has a complex but, on balance, beneficial effect on methanol synthesis from CO2 from both the kinetic and thermodynamic perspective. 3b.1.8 Methanol synthesis route “post-formate”? Despite general agreement on the importance of CO2 and formate in the early stage of the mechanism of carbon oxides hydrogenation to methanol over Cu, the relevance of the more hydrogenated species observed by FTIR, e.g., HCHO, H 2 CO 2 , and H 3 CO, is not yet clear. Recent calculations by the Danish group [147] indicate that the largest energy barrier in a synthesis route which proceeds via formate, dioxomethylene (DOM), formaldehyde, and methoxy, is DOM hydrogenolysis. This lends support to their argument that this is the most likely RDS in methanol synthesis from CO 2 . However, it should be noted that their work precedes the most recent evidence [106], suggesting instead that formate may be the pivotal intermediate in both reverse WGS and methanol synthesis. As regards the mechanism of methanol synthesis “post-formate”, there is substantial evidence that formaldehyde is rapidly convert-

ed to formate and methoxy over a wide range of oxides, via DOM [114-115]. The key step in the mechanism is considered to involve a hydride transfer between two adjacent DOM species, in a surface analogue of the Cannizzaro (disproportionation) reaction, to produce formate and methoxy groups. This occurs already at ambient temperature, and methoxy is subsequently hydrolyzed at 370 K to yield gaseous methanol. The higher activity observed over basic oxides like MgO and ThO2 is in accordance with the known catalytic effect of alkali (OH−) in the more familiar homogeneous process. Nevertheless, the occurrence of this reaction over various amphoteric, and even acidic, oxides like ZrO2, TiO2, Al2O3 and SiO2, illustrates that it is remarkably facile. The involvement of surface O in the mechanism was proven by isotopic labeling. Roughly equal amounts of CH318OH and CH316OH were produced from exposure of an 18 O−labeled ThO 2 surface to HCH 16 O. This suggests not only that formaldehyde and DOM are possible intermediates in the synthesis mechanism, but also that the latter ad-species is symmetrical, with chemically equivalent O atoms. Evidence that such a facile reaction is catalyzed by metals, albeit perhaps less effectively, has also been reported [163]. If DOM hydrogenation was indeed ratecontrolling in methanol synthesis, it is hard to reconcile the absence of DOM in the FTIR spectrum of the catalyst surface under industrial process conditions, and the strong presence of formate [105], an almost universal feature in FTIR studies reported to date. If DOM was present in the highest concentration of any intermediate in the mechanism, as required by kinetic considerations, it would be expected to dominate the spectrum. From studies of the transformation of pre-formed DOM over oxides upon warming in vacuo, the extinction coefficient of its main band envelope, centered at ∼1150 cm−1, appears similar to those of the intense and characteristic features of formate, which develop (along with those of methoxy) at its expense, at 1580 and 1370 cm−1 [114-115]. This is not surprising as

-30-

both sets of bands are associated largely with stretching vibrations of the carboxylate moiety common to each species. From the foregoing discussion, it is clear that oxide supports have the potential for cooperative involvement in several steps in the mechanism of methanol synthesis from CO2, with the proviso that transport of intermediate atomic and molecular species between the support and the metal (and vice-versa) is not a limiting factor. Their role takes on special signifcance in light of the known sensitivity of unsupported Cu to contamination, as might be more representative of working conditions. However, none of the “active” properties of the support so far considered appear to relate to the RDS, for which the best candidate is formate hydrogenation to a DOM-like ad-species. The mechanism favoured by this author is summarized below:CO2 → CO2∗

.22.

H2 → 2H∗

..23.

CO2∗ + H∗ → HCOO∗ ∗

HCOO + H



..24

RDS

→ H2COO∗

..25.

CANNIZZARO

2 H 2 C OO ∗ → HC OO ∗ + H 3 C O ∗ + O ∗ . 26. RDS

[HCOO∗ + H∗ → H2COO∗

..27]

H3CO∗ + H∗ → CH3OH

..28.

O∗ + 2H∗ → H2O

..29.

CO2 + 3H2 → CH3OH + H2O ..30. ∗ where denotes adsorbed species on Cu and/or the oxide support. Formal charges are omitted to avoid undue speculation at this stage, though partial negative charge on all carboxylate ad-species and O∗ is most probable. The former is implicit from geometric considerations (see p.18), while the last is inevitable due to the known electrophilicity of O. From

recent cluster modeling, the formal charge on O adsorbed on Cu is estimated to be around minus one, (Cu δ+ ) n −O − , essentially an ionic bond but with a non-negligible covalent contribution [164]. In a general sense, the charge on formate is likely to be greater for species adsorbed on the more ionic oxide support compared to the metal. Although step 26 is written as a direct reaction between two DOM species, it is also conceivable that HCHO could form as a distinct entity in a side reaction. Nevertheless, its existence may only be fleeting as it would be expected to be immediately hydrated even by traces of ambient wat er vapour, t o meth ylene gl ycol, CH2(OH)2, or the glycolate anion, which are both implicated in the Cannizzaro reaction. Only the simplest process is formally represented here. 3b.1.9 Summary The best catalyst for CO2 hydrogenation to methanol is supported Cu. The metal is primarily responsible for the activity, but the support may play an active auxiliary role in the process under industrial conditions, when contaminants may moderate the efficacy of Cu per se. A mechanistic view in terms of a pivotal formate intermediate, and its hydrogenation to dioxomethylene as the rate-determining-step, fits well with the bulk of kinetic and surface spectroscopic evidence reported to date. In addition, the well-known CO 2 /CO synergy displayed in conventional syngas conversion can be readily accommodated into the scheme. CO2 conversion to methanol is up to two orders of magnitude faster than CO when compared as sole C sources at low conversion. The likelihood of competition (for formate) between methanol synthesis and the reverse WGS reaction is strong, but this should be confirmed in view of its practical implications for methanol synthesis from pure CO 2 , e.g., the need for CO recycling, steam addition, etc., under process conditions. Crucial future experiments in the field have been identified.

- 31 -

3b.2 Ethanol (and higher alcohols) 3b.2.1 General prospects

The synthesis of ethanol from CO2/H2 is only now being explored at the fundamental level. As such, this section is inevitably confined to preliminary evaluation of the prospects for “renewable syngas” conversion. While there are a variety of interesting routes to ethanol worthy of exploration and one exciting new report involving reductive coupling of CO2 to ethylene (vide infra), emphasis is placed on the direct route, i.e., synthesis achievable in a “one pot” operation without isolation of intermediate products. This is partly due to lack of space, and also to limit undue speculation. In any case, by focusing here primarily on the selectivity issue, the most serious problem in ethanol synthesis, the assessment which follows is of general validity. In addition, any direct route proven to be viable would ultimately render all others superfluous in the technical sense for economic reasons. There is just cause for optimism regarding CO2 conversion to ethanol, both on the basis of favourable thermodynamics (see pp. 11-13) as compared to methanol, and strong circumstantial evidence of active involvement of CO2 in the synthesis mechanism from normal syngas. Due to the rapid establishment of the WGS equilibrium over many catalysts, and certainly those including Cu, a CO2/CO/H2/H2O feed will be produced in-situ. Insofar as it can be termed a “CO2-rich” syngas, the behaviour of the latter may reflect to a significant degree that of a renewable CO2/H2 feed. So the present wealth of kinetic and mechanistic data for syngas conversion to higher alcohols is still valuable, albeit with compositional effects to accommodate by a certain amount of “extrapolation” from proven trends in syngas behaviour. For obvious reasons, fundamental studies have focused primarily on catalysts developed for industrial Higher Alcohol Synthesis (HAS). These are all oxide-supported systems, e.g., alkali (K+)−promoted Cu, Cu/Co(Ni) and MoS 2 /Co 9 S 8 . Metallic Rh is also being explored although it is doubtful whether this precious

metal will be used industrially in the foreseeable future (see p.16). The field has been extensively reviewed in recent years [41,79,8182,97,165]. Only the last two catalyst types show good selectivity to ethanol (C2). Modified Cu yields mainly methanol and iso-butanol, whereas Cu/Co gives a cut of C1-C6 alcohols with a product distribution resembling that observed for the corresponding hydrocarbons in Fischer−Tropsch Synthesis (FTS). In this case, the implication of a mechanistic parallel between HAS and FTS is clear [101], and the deliberate inclusion and control of “FT-functionality” in the catalyst is a major development goal. This is perhaps best exemplified by the IFP catalyst comprising Cu and Co (or Ni) as major components, in which their roles appear separable into those of CO insertion/hydrogenation (“alcoholization”) and chain growth, respectively. This is a blatant oversimplification but serves as a useful practical rule. In catalysts more selective for ethanol, it is not yet clear if their special properties are associated with a moderated “FT-quality”, or whether indeed the mechanisms are different. In the Mo-based (Dow) formulation, there is some evidence that methanol may be an intermediate [77]. If correct, the reaction may be represented as:CH3OH + [-CH 2 -] → C H 3 C H 2 O H

..31.

which formall y constitutes homologation, although this is an unlikely mechanistic step. However, the mode of action of this catalyst is the least understood to date. There is evidence that methanol, or its derivatives, may also be involved in ethanol synthesis over modified Cu, but the implications for a route from CO2 are ambivalent because of possible inhibition by water, the co-product of methanol synthesis (vide infra). Viewed in the simplistic terms described above, one advantage of CO2 -rich syngas in the selective synthesis of ethanol, or at least simple alcohol mixtures, may accrue from the fact that chain growth in CO2 hydrogenation is severely restricted relative to CO. For example, high selectivity has been observed in CO2 methanation and attributed to the low surface

- 32 -

coverage of the metal with CO intermediate, θ CO , which minimizes competition between C−C bond formation and the RDS, hydrogenation in this case [117c-d,140]. Thus, by judicious control of operating parameters (and its observable effect on θCO), an optimum condition may be reached to drive the synthesis towards short chain-length products like ethanol. 3b.2.2 Alkali-doped methanol synthesis catalysts Work on alkali-doped Cu is particularly valuable because it is the simplest formulation active in HAS, at least for selective synthesis of C1-C4 alcohols. As more is known about methanol synthesis and the role of CO2 in this reaction, valuable clues regarding the prospects for ethanol synthesis may be gleaned from behaviour trends in HAS and any demonstrable inter-relationship with methanol synthesis. Alkali metal ions (ami) are ubiquitous as promoters in HAS, although their role is evidently complex and multi-functional. As mentioned earlier (see p.18), surface science investigations have shown that ami-promoted Cu leads to activation of CO2 even in the absence of hydrogen. Furthermore, the C−C bond forming reaction between CO2 δ− and a methyl donor to give acetate (CH3COOδ−) over Ni is facile [92-93]. This is highly significant because an acetate intermediate is believed to be pivotal in HAS over Cu/Co [97] and Rh [79]. It is also implicated in the reverse process, i.e., ethanol steam- reforming, as observed in this Laboratory (vide infra). This suggests a parallel with methanol since the analogous carboxylate (formate) also plays a key role in the synthesis. In recent literature, the most comprehensive study of an ami-modified methanol catalyst, Cu(K+)/ZnO/Cr2O3, is that of Calverley and Smith [166]. A promoting effect of K+ was found for both methanol synthesis and HAS, but only when CO2 was absent from the syngas feed. This confirmed earlier findings, principally by the Klier group, for unsupported Cu

& Cu/ZnO/Al2O3[167-169]. For normal syngas (H2/CO2/CO), both reactions were actually suppressed. This parallel behaviour suggests that alkali promotion is linked rather to CO activation, and that CO2 plays a major role in both methanol synthesis and HAS. It is known that CO dissociation is facilitated by alkali on many metals [99-100]. The effect on Cu is to create new adsorption sites, possibly via surface reconstruction, which weaken substantially (activate) the C−O bond for hydrogenation [93]. Inhibition of methanol synthesis clearly suggests interference in the mechanism involving the primary C source, CO2. Evidence reported for stabilization of formate, e.g., reduced reactivity in H2 induced by K+-doping (of Cu/SiO2 ) is consistent with this view [106]. The observed parallel (inhibition) in HAS suggests that a similar influence is also exerted on acetate in ethanol synthesis. This is further compounded by the known deleterious effect of alkali on H2 activation. In a recent review by Kiskinova [170], it has been pointed out that alkali has a dramatic (high-order negative exponential) poisoning effect on the sticking coefficient of H2, and thence θH in steady-state, over many metals. It may even scavenge atomic H (after H2 dissociation) by formation of stable (A+−H−) hydride-like species. A classic example is inhibition by K+ of CO methanation over Ni, in favour of higher hydrocarbons, despite simultaneously promoting CO dissociation, which is evidently not rate-determining [171]. Carbon oxides hydrogenation reactions are normally positive order in P H2 , but zero-, and negative order in PCO2 and P CO , respectively. In other words, CO is self-poisoning due to its preferential adsorption at the metal surface, often to near-saturation coverage, i.e., θCO ≈ 1. Consistent with this view, it is known from industrial practice, e.g., in the FT process [158], that any catalyst modification to improve selectivity toward longer chain products (presumably favoured by higher θCO and, thence, θC) reduces the overall rate of CO conversion while producing more unsaturated products [158]. This is the trade-off for conferring several pro-

- 33 -

cess benefits, including one primary function of alkali in the FT process, viz., to impart greater stability on-stream via control of coking. The same trend is seen in HAS over alkalimodified Cu. The rate invariably drops as the selectivity shifts towards higher alcohols at the expense of methanol. HAS is actually one of the slowest industrial processes in syngas conversion [81]. Here again, there may be an advantage in using CO2 -rich, or possibly even CO-free, syngas feed. If the selectivity to short-chain products like ethanol can be steered by restricting θCO to a suitable value, then the conversion rate may also increase to a more workable level. Some elements of the HAS mechanism were also explored in the work of Calverley & Smith [166]. They studied the effect of C source ratio over supported Cu, both alkali-doped and undoped. All catalysts showed a broad maximum in rate at CO 2 /CO ≈ 0.1, where the undoped sample was twice as active as one doped with 0.5 wt% K2CO3. This clearly shows that alkali is undesirable in HAS over supported Cu under CO2-rich conditions. However, the kinetic and mechanistic significance of t hi s ap p aren t o p t i m um in com pos i t i on i s not so clear because the syngas used was very lean in hydrogen at (CO2+CO)/H2 = 2. In contrast, the ratio used in industrial HAS is from 0.4 to 0.5, i.e., close to that of the nominal reaction stoichiometry. The effect of added methanol was generally to inhibit HAS, whereas ethanol in syngas produced an increased yield of propanol and higher (C3+) alcohols. This is consistent with earlier kinetic modeling [172], suggesting that ethanol is synthesized slowly and easily converted to higher (typically C4) alcohol products. In more recent studies over similar catalysts, Boz et al. [173] confirmed that butanol levels reach a maximum, apparently at the expense of ethanol, at a doping level of ∼0.5 wt% K2O. A maximum in the yield of aldehydes and a continuous drop in methanol with increasing contact time (and CO conversion), was

taken as evidence for an aldol condensation mechanism. In contrast, addition of Li+ [167] and Cs+ [174] improved the selectivity to ethanol. As these promoters are at the extremes of the basicity range in the alkali series (K+ being an intermediate case), the selectivity effect appears unsystematic and complex. In the last cited work on Cu(Cs+)/ ZnO, the use of labeled 13CH3OH proved that methanol is the source of both carbons in ethanol. The most likely mechanism of C−C bond formation is via formaldehyde, a known product of methanol decomposition over Cu. From stepaddition experiments in syngas, HCHO yielded ethanol preferentially whereas CH3OH gave more iso-butanol, normally the main product in HAS [175]. The identity of the co-reactant (with HCHO) is not yet clear but is believed by Nunan et al. [174] to be Cs + −stabilized formyl (HCO), in which the C has sufficient donor character to bind with the more electropositive C in HCHO. Methyl iodide is used as a trap for formyl species on a similar basis, producing the stable acetaldehyde molecule [176]. This would also explain the positive effect of alkali on alcohols synthesis from CO 2 free syngas. The alkali stabilizes formyl, the most likely intermediate from CO in the absence of water. [This is consistent with the earlier view (see pp. 21-22) that the CO2/CO synergy is associated with the promoting effect of water vapour on CO during methanol synthesis]. That the HAS mechanism does not seem to involve FT-type chain growth (via carbide intermediates) is a potential advantage in terms of achievable selectivity. However, if formyl and formaldehyde are important, the implications for selective synthesis of ethanol from CO2-rich syngas, at least over alkali-doped Cu, are unclear. The previously cited report [166], noting that the CO2 content in syngas must be kept low, was also reaffirmed in this work [173]. In summary, the main strength of modified Cu catalysts is in their selectivity to methanol and iso-butanol. This mixture is ideal for conversion to MTBE by dehydration and rearrangement over acid catalysts.

- 34 -

3b.2.3 Modified Fischer-Tropsch catalysts For alcohols synthesis, the Cu/Co catalyst manufactured by IFP is the nearest analogue of a conventional FTS catalyst, both in terms of composition and behaviour [41,97,165]. The close resemblance of the product distribution to that of the corresponding hydrocarbons from syngas conversion over the FT-metal, Co, is strong testimony to this view. Yet, there is an unquestionable role of Cu because the yield of alcohols over Co alone is typically much lower [158]. The synergy between the two components implies cooperative functioning at a microscopic level. The recommended preparatory route for the catalyst underlines this point, with great emphasis being placed on maintaining homogeneity throughout the work-up stage [81,165]. The distinction between this catalyst type and modified (Cu) methanol synthesis catalysts is absolutely clear. For example, extremely small amounts of Co (< 1 atom %) already suppress methanol synthesis activity by a factor of ∼5 [177]. Cu and Co generally comprise up to 30% of the catalyst, on a total metals basis, with an optimum Cu/Co ratio being ≈ 1.5 [81]. Insofar as the selectivity to alcohols can be interpreted in terms of an additive effect of Cu and Co, this suggests that their individual character is at least partially retained in the mixed state. Evidence from electron microscopy for the existence of separate nanometre-sized crystallites of Cu and Co, appears to support this view [178]. Nevertheless, the synergy might better be explained as due to bimetallic clusters and/or alloy formation. Evidence for the former comes from FTIR studies showing that the vibrational frequency of CO (presumed to be adsorbed only on the Co component) is shifted down by ∼40 cm−1 in the presence of Cu [179]. This is interpreted in terms of electron transfer from Cu to Co, which only seems feasible if there is mixing on an atomic scale. The question of Cu/Co alloy formation and its relevance, if any, to HAS is important but not yet resolved. This unfortunate situation may be attributed, in part, to the intrinsic difficulty of making the alloy by conventional

methods of catalyst preparation and, even if formed, of maintaining its integrity during activation and in the catalytic reaction environment. Bulk alloys are known to be metastable and undergo phase separation above ∼500°C [180-181]. This point will be taken up again later. The prevailing wisdom regarding the HAS mechanism over supported Cu/Co is based on the coupling of two distinct processes, viz., chain- growth via FT chemistry, and alcohol formation, in which chain termination by CO insertion is necessary [97]. Chain growth is represented schematically as below:CO2→CO→ C → [-CH2-] → CH3 → CH4 31. -[CH2]- + -[CH2]- → -[CH2−CH2]-

..32.

in which the probability of chain growth, via coupling of methylene groups, is a function of θCH2, which is affected by the balance in supply and consumption kinetics [101]. If hydrogenation dominates, methanation is the likely end result (scheme 31). If hydrogenation is slow, chain growth can proceed (scheme 32). If the rates are similar, this will lead to higher hydrocarbons, e.g., ethane:CH3- + -CH2- → C2H5-

..33.

C2H5- + H → C2H6

..34.

in which scheme 34 constitutes chain-termination. Selectivity in FTS is regulated mainly by the identity of the metal and the effect on it of any support, although there is a considerable influence of operating parameters (P,T, feed composition, space velocity, etc.), which can be used to moderate the process. However, the result is often a range of products whose carbon number and distribution can be quite accurately predicted on the basis of a probabilistic model of polymerization with simple assumptions; the so-called Anderson-Schultz-Flory (ASF) rule. Unfortunately, this is of such general validity that it leaves FTS extremely limited in its potential for selective synthesis of any single product of carbon number greater

- 35 -

than unity. Nevertheless, modest progress has recently been made in this direction with a Co/MnO catalyst, which gives non-ASF selectivities to C2−C4 alkenes [97]. The second major process needed for higher alcohol synthesis (in the context of FTS) is CO insertion into the growing chain. It is evident that this constitutes an alternative chain termination step in FTS as the alcohol products, if formed at all, are normally of the linear monohydroxylic type [158]. Thus, the analogous reactions to schemes 33 and 34 involving CO would be formation of the acetyl species, CH3CO, followed by stepwise hydrogenation to acetaldehyde and ethanol. A reaction between methylene and CO to produce ketene, H2CCO, is also conceivable. Enhancement of ethanol synthesis by addition (to normal syngas) of methylene donors like CH2C12, as compared to CH3Br, militates in favour of a carbene intermediate, RCH=M, where R = H, CH 3 ,.., and M is the metal surface [182183]. This is consistent with the view that CO insertion is most probable when the free valencies in the CHx fragment are low enough to permit mobility, i.e., when x ≥ 2 [184]. From chemical trapping or CO2 yields in thermal desorption, the Strasbourg group [185] has been able to demonstrate, over a wide range of supported Cu/Co catalysts, a strong correlation of the surface coverage by formate with methanol synthesis activity, and by acetate with HAS. Although acetate is believed by Kiennemann et al. to be a mere “spectator” in HAS, it can be taken as an index of acetyl, the suspected intermediate. This raises the likelihood of plausible routes to ethanol from CO2. The primary function of alkali in the Cu/Co system is to suppress the production of CH4, which is a dominant product of syngas conversion over Co. This is shown dramatically in the Idemitsu-Kosan catalyst [165], in which K+ −doping allows Co to be substituted by cheaper Ni. The latter is the superior methanation catalyst and the preferred choice in FTS for this reason [158]. Studies into the potential of Cu/Co for HAS from CO2/H2 have just started, likely

prompted by the implication that CO 2 may be an important source of carbon over these catalysts. The industrial process, at least in earlier operations [81], was based on syngas with similar levels of each C source. Kieffer and Udron [186] have compared Cu and Cu/Co, supported on La2 Zr2O7, in alcohols synthesis from CO/H2 and CO2/H2 syngas mixtures. Cu/La2 Zr2O7 gave expectedly good yields of methanol from both feedstocks. Indeed, it was originally developed for this purpose [187]. Addition of Co resulted in predominant formation of hydrocarbons, especially CH 4 from the CO2 -based syngas. Small amounts of C 2 + alcohols, more from CO than CO2, were also produced. Although these results do not lend support for a major role of CO2 in HAS, it is fair to point out that these catalysts are far from optimization as yet. Furthermore, they were also not alkali-doped, a virtual necessity to suppress FT characteristics (hydrocarbon production) normally dominant in the Co component. It should also be borne in mind that this research topic is complex and that experimental work is at an early stage. 3b.2.4 Supported rhodium catalysts Rhodium, when suitably modified, seems to have a unique combination of qualities which favour the selective synthesis of ethanol, at least from normal syngas. In view of this fact, it merits consideration apart from the other catalyst types. Its special properties, and their link with ethanol synthesis, have been the subject of an excellent review by Bowker [79]. A clean Rh surface chemisorbs CO but dissociation proceeds only slowly due to a high activation energy. The thermodynamics are such that CO dissociation is only mildly exothermic, in balance with C and O atom recombination and molecular desorption. Thus, it is extremely difficult to build up a contiguous carbon network on the surface. On the other hand, it has moderate activity in H 2 dissociation. Hydrogenation of the active C leads to methane quite selectively, although Rh is a mediocre methanation catalyst per se due to its low propensity for CO dissociation [158].

- 36 This balance of qualities would be expected to facilitate acetyl or ketene formation, i.e., the view of an isolated CHx fragment surrounded by adsorbed CO and H, an environment that should facilitate CO insertion in competition with the hydrogenation. The role of the support is generally seen as providing access for the unstable acetyl (or ketenyl) intermediate to surface O at the interfacial region, thereby creating acetate [79]. This “stabilized” acetyl is less likely to undergo C−C bond cleavage and decomposition back to CH3 and CO, which has been shown to occur at a high rate on Rh [188]. Evidence for a direct role of the support in promoting oxygenate synthesis over Rh is clearer than for methanol synthesis over Cu insofar as the latter has virtually no methanation activity to suppress. Ichikawa [189] has reported that Rh cluster-derived catalysts on amphoteric or slightly acidic oxides like La 2 O 3 , ZrO 2 , TiO2 , etc., give good selectivity to ethanol and C 2 oxygenates. In contrast, basic oxides like CaO, MgO, and ZnO yield methanol, while conventional supports like Al2O3 and SiO2 favour methane. Such an extraordinary range of behaviour is probably due to the high dispersion of the Rh typically obtainable via cluster precursors, and the consequent increase in metal/support interfacial area. This would be expected to give full expression to cooperative effects of this kind if they exist. The promise inherent in the “cluster” approach to preparing supported metal catalysts with well-controlled properties has been documented recently by Ichikawa [190]. Its value in CO2 hydrogenation, and especially control of selectivity, has just been reported [191]. Iron carbonyl clusters, Fe3(CO)12, supported on ZSM -5 zeolite displays high activity already at 260°C, and good selectivity to light alkenes, especially ethylene. In the best case, the molar selectivity reached 92%, but with traces of propylene and the balance CH4. Although this is strictly an indirect route to ethanol from CO2, still requiring hydration of the alkene, this is standard industrial practice. At the present time, this appears to represent a potentially excellent

alternative route to ethanol, and should be investigated further, along with the direct option under consideration here. Promoters like vanadia [192] and ceria [193] appear to have a multiplicity of roles, but the one most favoured by Bowker is akin to the support effect, i.e., the provision of surface O to stabilize acetyl. In view of the importance of acetate, the mechanistic proposal of Bowker leaves room for the involvement of CO 2 , which could yield acetate directly via insertion into M−CH 3 . Koerts and van Santen [194] have recently used isotope tracing to elucidate the effects of Vn+ promotion (ex. NH4VO3) on Rh/SiO 2 at a Rh/V ratio of 3. Vanadium increases the activity roughly by a factor of 4, yet the rate of CO insertion is little affected, demonstrating that the latter is fast and not the RDS. It also improves the selectivity to ethanol by promoting hydrogenation of the acetaldehyde intermediate and suppressing its hydrogenolysis to CH 4 . The promotion of overall activity (CO conversion to ethanol) was again attributed to stabilization of C2 oxygenate intermediates. Preliminary exploration of the potential of supported Rh for ethanol synthesis from CO2/ H 2 has just been reported by Arakawa et al. [195]. They examined the effect of Rh loading, and the addition of a wide range of metal salts, on activity and selectivity of Rh/SiO2 in syngas conversion at 50 bar in the temperature range 200-260 °C. On 5 wt% Rh, the reaction was found to be almost exclusively methanation, whereas on 1% Rh, the product was mainly CO with significant amounts of methanol and ethanol (up to 20% on a carbon basis). This result shows, once again, the wide range of behaviour open to Rh. The dominance of reverse WGS and alcohols synthesis was attributed to the higher dispersion of the metal at this low loading. The mean particle diameter (2 nm vs. 3-4 nm at 5% Rh) indicated structure sensitivity. The conversion of CO2 was limited to ∼1 %. The effect of promoters on 5% Rh under the same conditions was dramatic. Conversions were raised by typically an order of magnitude, and selectivities to alcohols were in

- 37 -

in some cases better than those over highlydispersed Rh. The best selectivity to ethanol (∼40%) was found for Li + addition, while a doubl y-promoted sample (Rh/Li/Fe = 1/1/1) raised conversion to ∼14%, maintained selectivity to ethanol (35%), raised that of methanol (to >20%) at the expense of CH4 (< 10%), with the balance being CO (∼30%). A correlation was made between alcohols selectivity and a major increase in Rh-adsorbed CO in the “bridge” position, as identified by in-situ FTIR. The authors speculate that the rate of CO insertion may be improved by Li+. addition. On this basis, the early prospects for promoted Rh in simple alcohols synthesis from CO2 look quite good, at least on the laboratory scale. 3b.2.5 Strategy for catalyst selection & screening In view of the foregoing evidence that acetyl and/or acetate intermediates are important in HAS in general, and the evident parallels with methanol synthesis, a set of mechanistic criteria can be drawn up for the optimization of ethanol synthesis from CO2/H2. Catalysts which appear to offer the best prospects, i.e., those which satisfy most or all of the above constraints can then be identified for preliminary screening, and ideally under optimum reaction conditions. As it is a major aim to find materials upon which a technical process might ultimately be established, the use of Rh per se is ruled out on the basis of its poor economic outlook. On the other hand, a clear factor in this strategy is to incorporate or “mimic” the desirable properties of Rh by judicious tailoring of cheaper metals, in either single- or binary-form. Starting from CO2/H2, the major steps leading to ethanol can be schematized as below:1. Creation of CO from CO2 (reverse WGS) 2. Hydrogenation of CO to CH2 or CH3 (FTS) 3. Insertion of CO and/or CO2 into M−CH3 4. O-stabilization of CH3CO (ex. step 3)

5. Hydrogenolysis of CH3COO (RDS ?) As an example, the most promising single metal to explore next is ruthenium. First, it is adjacent to rhodium in the Periodic Table. If modified to an electron-rich form, e.g., by AM doping (which is known to lower the work function of Ru considerably [196]), it may take on some catalytic properties of Rh. It also shows a wide range of behaviour in FT chemistry, having the highest methanation activity of any metal at low pressures of CO- or CO2based syngas, whilst promoting “infinite” chain growth in the polymethylene synthesis at high pressures [158]. It clearly satisfies criterion 1 in the clean state on a TiO 2 support [140]. The hydrogenation function (2) needs to be moderated, and CO/CO 2 insertion (3) to be promoted, both of which may be achievable by control of metal dispersion (CO dissociation on Ru is structure-sensitive, being faster on small particles), AM doping [196], and/or reaction engineering [Criteria 4 and 5 are the most speculative and difficult to evaluate, and are not considered further here.] Its cheaper 1st-row counterpart, Fe, is worthy of study, especially in view of its promise for selective synthesis of ethylene from CO in supported cluster form [191]. The binary Cu/Co and Cu/Ni systems should be further explored in terms of the role of alloying and AM-doping. In the case of Cu/ Co, it has just been claimed that AM-doping promotes alloy formation, i.e., a synergy operates, and that the alloy is mainly responsible for ethanol synthesis activity from CO2 [197]. Mo appears to fulfill a similar role to Cu in promoting reverse WGS (1) but requires a second component to confer hydrogenation (2) and CO activitation and insertion (3) functionalities; though this is too-simplistic a picture because earlier tests showed that commercial Co/Mo hydrodesulphurization catalysts were disappointing in HAS, at least in normal syngas [77]. However, in view of the viability of the little understood Dow catalyst (based on Mo), and a recent report that the binary system Mo/Ir on SiO2 is active and modestly selective for C1−

- 38 -

C3 alcohols from CO2 [198], Mo-containing catalysts should be investigated in more detail. It is an interesting fact that most success in the electrocatalytic reduction of CO2 has come from work with electrodes composed of many of the metals proposed for study here by conventional thermal catalysis, e.g., Cu, oxidized Cu, Ru, Mo, etc. [266]. Evidence for extensive parallels in the electrochemical and thermochemical catalysis of CO2 hydrogenation are considered briefly in Section 6 (p.50). 4. Steam-reforming of simple alcohols 4a. Industrial and technical aspects 4a.1 Background Catalytic Steam-Reforming (S-R) of hydrocarbons is a well-established industrial process for the production of syngas, the feedstock for some of the world's major commodity chemicals, e.g., methanol. It is currently also the cheapest source of hydrogen for ammonia synthesis and hydrotreating in the petrochemicals field [199]. The trend in the last few decades in industrial S-R has been away from coal and naphtha resources towards cheaper (and cleaner) natural gas, which is primarily methane. The field has been well-documented by Rostrup-Nielsen [200-201]. The main reaction:CH4 + H2O

→ CO + 3 H2

..35.

is strongly-endothermic (∆H = +206 kJ.mol−1), and operation above 1000 K is necessary to achieve good yields. The use of a catalyst, normally supported Ni, is also obligatory because CH 4 is otherwise stable even under these conditions. Despite being energy-intensive, the process economics are quite respectable. For example, when coupled to methanol synthesis, the overall thermal efficiency is ∼65 %. However, the reforming stage alone constitutes more than half of the plant capital outlay. Mainly for this reason, there is growing interest in direct routes from CH4 to value-added chemicals and alternative feedstocks. Major goals are

partial oxidation to methanol and oxidative dehydrogenation (“CH4 coupling”) to lower alkenes [202]. Interest is also being shown in CO 2 -reforming of methane [203] and the feasibility of using renewable energy, e.g., solar high-temperature heat, to drive this process. Catalysts based on Ni, modified to reduce coking, are already promising alternatives to Rh in this reaction [203-204]. 4a.2 Steam-reforming of methanol Although little commercial interest has been shown in this process to date, unlike methanol synthesis, it nevertheless has a major role to play in a future renewable energy economy as a convenient and efficient form of H2-storage (see section 1.c). Niche applications based on this concept are now emerging. For example, storage of H 2 in methanol for on-site generation is cheaper than water electrolysis or bringing in H2 (derived from CH4 S-R), at least in modest-demand applications, i.e., at supply rates ≤ 300 Nm 3 h−1 . Industrial units have been on the market since the late 1980s [31]. On a smaller scale, Johnson-Matthey offer skidmounted units as a supply of ultra-pure H2 based on Pd/Ag membrane “clean-up” technology. This is suitable for small technical applications, usually in remote areas where the logistics of supply disfavour alternatives. Methanol S-R proceeds rapidly over supported Cu above ∼200 °C and industrial catalysts are now available. The close resemblance of the steam-reforming and synthesis catalysts is not surprising because S-R is simply the reverse of synthesis from CO2, as below:CH3OH + H2O → CO2 + 3 H2

..36.

The reaction is endothermic due to the production of H 2 , with ∆H 3 0 0 K = +131 kJ. mol − 1 . However, the overall thermodynamics are just favourable already at room temperature, with ∆G300K = −3.7 kJ.mol −1 . [The free energy plots in Fig. 2 (see p.12) are validated for steam-reforming reactions simply by reversing the sign on ∆G.]

- 39 -

4a.2.1 Advantages of pre-reforming An endothermic reaction which proceeds at low temperature is a fortuitous situation as it allows the very real prospect of driving S-R of methanol (and, in principle, higher alcohols), using recycled waste heat from combustion of the H2 fuel, e.g., in a modified Otto-cycle engine for vehicular propulsion. Thus, a substantial increase in the thermal efficiency of the overall process is feasible but little-exploited to date, as was also pointed out by Mills [41]. Indeed, this principle has recently been shown to be of general validity by analysis of process exergy, i.e., the energy available for useful work. Hijikata [205] has concluded that all solid and liquid hydrocarbons should be reformed to H 2 , driven ideally by lowgrade heat, i.e., “waste” heat recovered from the combustion process. While direct combustion of some fuel is inevitable to establish start-up, incorporation of such a “thermal upgrading” cycle would lead to major energy savings in all power utilities based on conventional fuels. The concept is now being explored in respect of vehicular propulsion [206-207]. Methanol S-R is a special case in view of its favourable thermodynamics in the relevant temperature range, and the virtual absence of harmful side reactions, including a low probability of coke formation. Compared to direct combustion, other benefits of pre-reforming accrue from the clean-burning quality of H2 fuel. This leads to dramatic reduction in pollutant emissions, and increased flexibility of operation, e.g., the option to run the combustion process “lean”, thereby further improving efficiency and reducing NO x emissions. 4a.3 Methanol for on-board storage of H2 There are two major technical possibilities in the transportation sector based on pre-reformed methanol as the source for H2 fuel. The first is to couple the reformer to a spark-ignition internal combustion engine (ICE) adapted to run on H2. The second is to link it with a H2− powered fuel cell to provide electricity to the drive system. Each has its own merits and limitations, and it is too early to tell which may

ultimately dominate in a future market for H2-fueled vehicular propulsion. The ICE-linked system clearly has an economic advantage at present. Nevertheless, there are good prospects for both in respect of energy density (fuel tank volume), system weight, driving range, storage efficiency, and ease of refueling, when compared with the other main alternatives, i.e., pressurized H2 gas, liquid H2, or storage in hydrides. The fuel cell-based system may ultimately be favoured if its intrinsic efficiency advantage is not compromised in the link-up, and if emissions legislation continues to get tougher in the future. 4a.3.1 Reformer/ICE link-up In the early 80's, workers at Nissan explored the potential of a methanol reformer/dissociator linked to a modified internal combustion engine (ICE) for on-board application [27-28]. In stationary tests, brake thermal efficiencies were 42%, 33%, and 30%, for reformate, methanol and gasoline, respectively. This efficiency gain (∼30% relative) actually exceeded the value expected on the basis of reforming alone (∼16%) due to the lean combustion mode possible with H2-rich reformate. In spite of power limitations at wide-open throttle (a problem for gas-fueled ICEs in general), the maximum speed for a typical passenger car running on reformate-only was estimated to be a respectable 100 km/h. At higher speeds, direct supply of methanol would be necessary as a power booster. Emissions of CO, hydrocarbons and NO x , without an exhaust catalyst fitted, were below standards in force for “conventional vehicles” (CV) in California in 1991 [8]. This work was later suspended in favour of flexible-fuel vehicle (FFV) development, partly due to the uncertainty in methanol supply at that time [29]. The main technical problem was reformer catalyst deterioration, believed to be caused by exposure to high-temperature transient cycling under simulated driving conditions. Thus, the main goal in the development of superior catalysts for this application appears to be longevity, which is primarily a function

- 40 -

of sintering-resistance of both the active metal and the support. The Nissan group screened several metals, including Cu, Zn, Cr, Pt, Pd and Rh; in every case on an Al 2 O 3 support, which is not ideal as it has partial acidic character, promoting dehydration. Dimethyl ether (DME) formation was indeed observed. They finally selected a commercial Pt/Al 2 O 3 sample for scale-up testing, mainly on the basis of better durability over the working temperature range of 300-650°C, as governed by their exhaust gas recycle/heat exchanger configuration [27]. However, it is not completely clear how much of the problem was related to sintering and/or to possible catalyst “fouling”, as the fuel was technical methanol, at best containing only small levels of adventitious water. Under such conditions, there is a slight danger of coke deposition, e.g., via cracking of methyl formate, DME, or CO dissociation. The ideal catalyst would consist of a cheap, high-melting metal with good dehydrogenation properties but low activity for CO dissociation, supported on a basic oxide with excellent hydrothermal stability. It should be readily activated in-situ, i.e., easily reduced in methanol/steam vapour mixtures. The traditional metal, Cu, supported in pure form is clearly unsuitable as it is known to sinter rapidly at elevated temperatures. Furthermore, unlike in a fuel cell (vide infra), the excellent WGS activity of Cu is almost irrelevant here as there is no poisoning issue for the ICE, CO has similar (clean) combustion properties to H 2 , and creation of a CO/H2 mixture from methanol is almost as endothermic as S-R, so not compromising the thermal efficiency advantage. This allows reasonable flexibility in the consideration of alternative metals and operating conditions. Catalysts consisting of Ni or Fe, supported on neutral or basic oxides with excellent hydrothermal stability like ZrO 2 or “stabilized” MgO [208], seem promising candidates, and are under exploration in our Laboratory. Good results have just been reported elsewhere for the binary Ni/Cu system on SiO2/MgO [209].

Following on from Nissan, equally encouraging results, in terms of thermal efficiency and emissions (low NOx), favouring on-board prereforming, have been reported recently by Swedish workers, albeit derived from tests on simulated “reformate” gas mixtures supplied directly to the engine [30]. 4a.3.2 Reformer/fuel cell link-up Unlike batteries, fuel cells transform the chemical energy from an externally supplied fuel and air into electricity, in which the problem of recharging is eliminated. In a mobile application, the fuel cell has immediate operational advantages over a battery in terms of stability (lifetime), energy density (range), and ease of refueling. The latter is the equivalent of adding gasoline to an ICE-powered vehicle. Although it is a flexible and ubiquitous electrical (DC) power source [210-211], its most severe limitation at present is that it can only work efficiently on hydrogen fuel at the power levels relevant for vehicular propulsion (> 20 kW) or higher, e.g., as stationary modules for stand-alone or auxiliary grid power supply up to 10 MW. This means that to exploit the excellent storage properties of conventional liquid fuels, an extra fuel processing or pre-reforming unit must be incorporated into the system. Since methanol reforming is technically the easiest process for deriving H2 from an organic liquid fuel, considerable interest is being shown in this combination. Although a wide variety of fuel cell types are under development, the only one attracting serious attention for mobile applications [212213] is the so-called solid polymer electrolyte (SPE-) or polymer electrolyte membrane fuel cell (PEM-FC), due mainly to the high power densities achievable, viz., ∼500 W kg−1. Unlike the alkaline type, the PEM cell (see Fig. 9, below) tolerates CO2. It also operates at low temperature (80-95°C), which gives it one of the highest intrinsic efficiencies of any fuel cell, at around 75% of the theoretical maximum:ηmax= ∆G/∆H = 1−T∆S/∆H

..37.

- 41 -

MEMBRANE

H2O

ELECTROCATALYST

O2

H2

CATHODE

ANODE −

½ O2 +2H + 2e → H2O +

H2 → 2H+ + 2e−

Fig. 9. Schematic layout of a PEM fuel cell

with 50-60% efficiency attainable at practical power densities [211]. This is substantially higher than that of the best ICE and constitutes a major potential advantage of the PEM-based fuel cell system in propulsion units. Another feature is the excellent emissions profile, with NOx levels being virtually nonexistent. The economic outlook for market penetration by PEM fuel cells is uncertain. Membranes are commercially available but prohibitively expensive (with existing PEM technology at ∼$500 kW−1) for use on the scale envisaged [214] as compared with the mature ICE (∼$30 kW−1). However, researchers at this Institute [215] and elsewhere are making good progress in the development of cheaper and better membrane materials. The price of Pt for the electrodes (∼$35 kW − 1 ) is not a major issue because the present loading is quite low (∼0.8 mg cm−2), and likely to be further reduced in the near future (to < 0.4 mg cm−2 ) without seriously jeopardizing efficiency. Due to a number of unresolved problems in the reformer/fuel cell link-up (vide infra), research into the longer-term prospects for a “direct methanol” oxidation fuel cell also continues apace [216]. Recent progress has

been reported in membrane selectivity (prevention of methanol permeation) and improved poisoning resistance of the electrocatalysts [217-218]. Unfortunately, the achievable output of this direct cell are severely restricted by the mechanistic complexity of methanol oxidation, with the kinetics lying roughly two orders of magnitude below H 2 at the present stateof-the-art. Since the early '80s, the Kingston (Canada) group have been investigating methanol S-R as a source of H2 for potential fuel cell driven propulsion [219-222], and have now advanced to the stage of defining the performance criteria required of the reformer for link-up to a fuel cell of 60 kW nominal power [223]. This has lead to collaboration with Ballard Power Systems in Canada, and serious competitive interest from the U.S. government (the Jet Propulsion Laboratory) and several major U.S. companies, including General Motors [224]. In Europe, Dornier are active in reformer development as part of a Daimler-Benz/ Ballard joint venture. KFA Jülich (Germany) have a similar project in collaboration with the Danish catalyst company Haldor-Topsoe [224]. It is widely recognized that the major problem associated with the reformer/PEM fuel cell linkup is poisoning of the Pt electrocatalyst at the anode by even traces of CO (< 20 ppm) in the H2 /CO2 fuel stream exiting the reformer. The CO is believed to derive from the following reaction schemes:“cracking” S-R

CH3OH → C O + 2 H2

CH3OH + H2O → C O 2 + 3 H 2

..38. ..39.

RWGS CO2+ H2 → CO + H2O ..40. in which there are two likely processes for forming CO; either directly from “cracking” (dehydrogenation) of methanol (38), or from CO2 (ex S-R) via reverse WGS (39 + 40). Circumstantial evidence suggests that the level of CO is controlled by the WGS equilibrium. This predicts 2-3 vol% CO, which is also found in typical reformer conditions, i.e., at high methanol conversion in the temperature range 250−300 °C.

-42 -

This implies that the CO problem is unavoidable and demands recourse to on-board “CO cleanup” technology [212, 219]. It also adds justification to the viewpoint of sceptics that the system needs to be, in effect, a mobile “chemical refinery”, making the task hopelessly impracticable. However, the above mechanism has recently been questioned on the basis of new evidence that methanol SR does not involve cracking (38) at normal temperatures, and that methanol itself dramatically inhibits reverse WGS, so that CO levels under certain conditions may be vestigial or even negligible (vide infra). Hence, a practical solution may be to run the reformer in excess methanol, and/or at partial conversion with recycle. Various solutions to the CO contamination problem are under consideration. Selective oxidation over an intermediate bed of supported Pt [226] has been proposed, but the weight of this unit is likely to be greater than that of the reformer itself. Amphlett et al [227] prefer direct injection of air into the fuel stream fed to the anode, though there are concerns over possible acceleration of membrane deterioration and explosion hazards. The Jülich group are exploring selective methanation, and polymer and Pd/Ag (H 2 -selective) membrane purifiers. They are also developing CO-resistant electrodes, in which Pt/Ru alloys look promising as alternatives to pure Pt [212]. A minor disadvantage of the reformer/fuel cell link-up is that, unlike the ICE case, the reformer cannot be driven entirely by waste heat from the fuel cell because this is too low a quality (T < 100°C) to effect S-R over existing catalysts. On the other hand, it can be used for pre-heating and partial evaporation of the liquid fuel fed to the reformer, which constitutes up to 30% of total system heat demand. In terms of auxiliary heat supply, the design and testing of a catalytic burner, running on unconverted H2 recycled from the fuel cell and sacrificial methanol has been reported [228]. However, this is likely to result in non-negligible emissions and to reduce the impact of the highly efficient PEM

fuel cell component on system performance, although to what extent is not clear. Summary Both of the above approaches involving onboard pre-reforming of methanol are promising and likely to be intensified in the near future, spurred on by such initiatives as the US government/industry joint venture towards a prototype vehicle having unprecedented fuel efficiency and ultra-low emissions profile [7]. 4b. Fundamental Advances 4b.1 Methanol S-R 4b.1.1 Current views on the mechanism: The role of methyl formate ? The most interesting advance in methanol S-R at the fundamental level is the recent work of Jiang et al. [229] on an industrial Cu/ZnO/ Al2 O3 catalyst. They have questioned the popular conception of the S-R mechanism, involving a sequential cracking/WGS process (vide ultra, steps 38 & 40), proposing instead that methyl formate, HCOOCH3, is a crucial intermediate according to the following reaction sequence:2 CH3OH → HCOOCH3 + 2 H2

..41.

HCOOCH3 + H2O → CH3OH + HCOOH ..42. HCOOH → CO2 + H2

..43.

This is based on the observations that methyl formate (MF) is the major intermediate in methanol decomposition in the absence of steam, MF hydrolysis (step 42) is fast, and the WGS reaction is suppressed in the presence of methanol. As written, this scheme allows for no CO formation in principle, although it should be appreciated that formic acid decomposition (step 43) selectively to CO2 proceeds only on metals, with a greater possibility of dehydration (CO formation) on the oxide support [230]. In view of the last possibility, it appears that studies over Raney Cu are worthy of investigation. Nevertheless, these

-43 -

authors claim that CO was undetectable in their S-R experiments below 250°C, although the analytical sensitivity in the GC method used was not stated, unfortunately [231]. If MF is indeed important, this implies that the principle of microscopic reversibility (PMR) may not hold between methanol synthesis from CO2 (see sequence 22-29) and methanol S-R, i.e., the mechanisms proceed through different intermediates and/or transition-states. This is a surprising idea, especially in view of the compositional similarity of the catalysts for both processes. The rate expression under differential conditions, obtained by the same group, has an activation energy, Eapp, ≈105 kJ.mol −1 , and orders in methanol and steam concentration of 0.26 and 0.03, respectively [231]. As this E a p p is at the high end of the range reported for the synthesis (see p.25), it suggests that the RDS in both may also be different. A similar conclusion has been reached by the Kingston group, which has been given cause to reconsider the problem [232]. However, at least one common feature is the likely involvement of formic acid or formate. Since a previous in-situ DRIFTS study of a Cu/SiO2, catalyst surface during methanol decomposition did not find evidence for adsorbed MF per se [112], more studies of this kind seem necessary in order to resolve the question definitively. An added incentive for such work lies in what it may presage for the ultimate viability of a reformer/fuel cell mobile system, as discussed above.

Cu loading, precursor calcination, and activation (pre-reduction in H2 and equilibration in the reactant stream), were examined and compared. It is claimed that the activity is related to the average oxidation state of the metal, probably involving both Cu° and Cu+, and that the last is a strong function of pre-treatment and conditioning in-situ. This suggests that S-R does have some features in common with the synthesis. Three reaction temperatures were used (170, 200, 250 °C) and methanol was fed at high space velocity (whsv ≈ 20 h-1). The undoped samples had similar acti vities in both reactions in terms of methanol conversion, which ranged from 10−80 %, but H 2 production was significantly higher in S-R [235]. Differences in selectivity were also observed, depending on the reaction temperature. In pure methanol at 170 °C, the major C-containing product was MF (∼35 mol%) with traces of CO, while the balance was H2. At 250°C, CO grew at the expense of MF, more H 2 was produced, and CO 2 , CH 4 , and DME were detected in small amounts (≤ 5%). In a methanol/steam (1/1) mixture, the product distribution was generally close to the S-R reaction stoichiometry (39), consisting mainly of H2 and CO2 with traces (< 2%) of CO. The suppression of CH4 and DME formation was attributed to moderation of the acidic properties of the Al2O3 support by water adsorption. However, at 170°C, MF was again observed and the level of CO was negligible.

Fu rt h er evi d ence agai nst the cracking/ WGS sequence in favour of the involvement of methyl formate has been reported from studies over Pd/ZnO [233]. Here the ZnO support, either directly or via its influence on the metal, is believed to promote an alternative route to H2 and CO2.

Independent work on methanol decomposition by Cheng [237] has just appeared, confirming the appearance of MF at low temperature and low conversion. It was also shown that methanol synthesis catalysts based on Cu/ZnO are inferior in the decomposition reaction. The most active and stable catalysts comprised Cu /Cr2O3, doped with Ba, Mn, and Si oxides.

4b.1.2 Dehydrogenation vs. S-R

4b.1.3 Chemistry of methyl formate (MF)

In an extensive series of investigations, Idem and Bakhshi have explored preparational and dopant effects on coprecipitated Cu/Al2O3 catalysts for methanol dehydrogenation and S-R [234−236]. The effects on activity/selectivity of

The above evidence suggests that methyl formate is indeed an intermediate in both methanol dehydrogenation and S-R. The chemistry of methyl formate (MF) has been reviewed [238]. It can be formed in various ways

- 44 -

from C1 building blocks, but the most likely route in this environment is either from a secondary reaction between methanol and its initial dehydrogenation product, formaldehyde, as below:CH3O H → H2C O + H 2

. . 44.

H 2C O + C H 3 O H → HCOOCH3 + H 2 ..45. or a Tischenko type reaction between two formaldehyde molecules:2 CH 3 OH → 2 H 2 CO + 2 H 2 2 H2CO → HCOOCH3

..46. ..47.

In either case, this adds up to step 41, the first equation in the mechanism proposed by Jiang et al. [229]. The reaction is well known to proceed over Cu [238] although it is quite endothermic (∆H° = +100 kJ.mol−1). Of these possibilities, the first sequence is considered more likely on the basis that MF is only seen as a product from a methanol-rich feed. In the absence of steam, MF decomposes to CO at typical reaction temperatures, probably via decarbonylation [237]:HCOOCH3 → CH3OH + CO

..48.

which can even occur below 200°C, at least over alkali-doped active carbon [239]. Complete decomposition to syngas:HCOOCH3 → 2 CO + 2 H2

..49.

is considered unlikely in view of its highly endothermic character (∆H° ≈ +160 kJ.mol−1), and the literature reports that it proceeds only above 300°C [238]. The hydrolysis of MF (reaction 42) is an autocatalytic homogeneous process conducted below 100°C and in excess water to drive the equilibrium towards formic acid [240]. 4b.1.4 Implications for the RDS and CO-free operation As both ester hydrolysis and selective decomposition of formic acid to CO2 (over metals) are

facile, this suggests that the likely RDS is a step in the formation of MF, i.e., reaction 44 or 45, neither being implicated in the RDS for synthesis. Jiang et al. [229] have convincingly shown that methanol dehydrogenation (MF formation) is indeed the RDS in methanol S-R, with an almost identical activation energy, Eapp ≈ 103 kJ.mol−1 (vs. 105 kJ.mol−1 for S-R). If the mechanism proceeds only through MF, then there is some prospect of developing a catalyst which will allow CO-free operation in S-R. The key is to drive MF formation at a temperature low enough (≤ 170 °C) to allow its subsequent hydrolysis, to the exclusion of decarbonylation. Promotion of methanol S-R over Cu/Al 2 O3 by doping with cationic redox ions such as Mnn+ and Crn+ has been observed, and attributed to their effect on the balance of oxidation states of Cu in-situ [236]. Although the increase in conversion at 170°C was quite modest (∼50% relative), this result is sufficiently promising to encourage research in similar directions. It should be noted that these same metal ions are beneficial in methanol decomposition [237]. This is again consistent with the view that the RDS in methanol S-R does not involve steam per se. 4b.2 Ethanol S-R 4b.2.1 Rationale for study Although thermodynamically more restricted than methanol S-R, exploratory work on ethanol S-R has begun in the last few years. There are several good reasons to develop catalysts for this rather endothermic process:C2H5OH + 3 H2O → 2 CO2 + 6 H2

..50.

wh e re ∆H ° = +1 75 kJ .m ol − 1 ( fo r reactants in the vapour state). It is favoured thermodynamically at modest temperature (T ∆G=0 ≈ 210°C, see Fig.2). This would allow a thermal efficiency gain (via waste heat-driven reforming) of roughly double that for methanol, viz., 32% vs. 16%. In fact, this would raise the effective volumetric energy density of pure ethanol, in terms of hydrogen content (∼30 MJ. 1−1), close to gasoline (∼33 MJ.l−1) as C + H. In

-45 addition, ethanol is the major biofuel at present. It can now be produced quite economically, and on a significant scale, from fermentation of biomass and cellulosic wastes. If ethanol S-R is feasible over suitable catalysts, there are good prospects to verify the technical viability of an efficient, clean and renewable energy cycle based on simple alcohols. Furthermore, as it is likely that the first demonstration units will be of the stationary type, the CO2-rich off-gases from both fermentation and reforming can be scrubbed economically for recycling, e.g., as co-feed for methanol synthesis. Thus, CO2 neutrality can also be featured in such an integrated (closed-loop) operation, based on a complementarity between the two alcohols. Finally, elucidation of the mechanism of ethanol S-R may also shed valuable light on the corresponding process in methanol. The role of formaldehyde in particular is difficult to verify due to its extreme reactivity, rarely being identified in the product stream. Since ethanol and methanol are adjacent homologues in the series of primary monohydric alcohols, and share common features in many reactions, there is no good reason to exclude S-R. The reader should recall that evidence for mechanistic parallels in the reverse process, alcohols synthesis, is already quite strong (see Section 3b.2). 4b.2.2 Early progress Thermodynamic analysis of ethanol S-R by Luengo et al. [242] stresses the need to work under conditions which inhibit coke formation, while at the same time minimizing methane formation. Although a clean-burning fuel, CH4 is an undesirable product because it is difficult to reform, thereby constituting a loss in thermal upgrading. In preliminary catalytic investigations from the same group [241], ethanol S-R was explored from 300−550°C over a Ni/Cu/Cr/αAl2O3 catalyst. Water/ethanol molar ratios from 0.4−2.0, and space velocities from 2.5−15 h − 1 were examined. Hydrogen yields were not measured directly but inferred from material balance. Good conversion (> 80%) was obtained above 300 °C at high space velocities (w h s v ≥ 1 0 h − 1 ) .

However, the selectivity towards Ccontaining products was not ideal, yielding CH4 > CO ∼ CO 2 . Investigating ethanol S-R as potential H2 supply for fuel cells, Matsuda's group in Japan have examined Pt, Rh and Ni, supported on MgO, ZnO, and their physical mixtures [243244]. From tests over the range 200−500 °C, the main products were in the order H2 > CO > CH4 > CO2. Platinum was the most active metal, and Ni the most selective (for H2 & CO) of the three, whilst Rh produced more CO2. This was attributed to its superior WGS properties. However, in all cases deactivation (coking) caused significant activity loss over 10-12 h, although this is not surprising because the steam content used (H2O/EtOH ≈ 1) was well below the stoichiometric value. Activities in this Laboratory have focused on the development of catalysts for ethanol S-R, testing in continuous-flow reactors on the micro- and laboratory scale, and initial studies of the mechanism using DRIFT spectroscopy [245-246]. The most promising catalysts to date consist of nanoparticulate Cu, Co, and/or Ni, acting in synergy, supported on basic oxides like MgO and ZnO. ThermogravimetryFTIR, adapted to serve as a microreactor, has already proven invaluable in the characterization of precursors, monitoring of the activation process (typically reduction in H2), and rapid screening of catalysts. A sample suspended in a flowing gas stream is subjected to controlled heating to induce thermal decomposition. The progress of the reaction is followed continuously via sample weight loss and its correlation with evolved gases, identified and estimated by FTIR. If the activated form of the catalyst is then exposed to reactant gases, real-time monitoring of transient behaviour is possible, and correlation with sample condition, e.g., deactivation vs. weight gain due to coking, etc. Ethanol conversion over Cu/Co(Ni)/MgO was complete by ∼350°C at space velocities of technical relevance (whsv = 2−5 h −1 ). The major problem was deactivation. This is shown in a TG-FTIR

85

∆W ≈ 0.7%

80 40

100

160

220

280 min

TG balance / % starting we8ight

90

95

Concenration / A. U.

∆W ≈ 8%

TG balance / % starting weight

Concentration / A. U.

100

7575 40

100

160

(a)

220

280 min

(b)

Fig. 10 Accelerated coking of Cu:Co:Mg = 1:1:3 in TG-FTIR. ( a) undoped (b) 5% Cs+ dopant Figure reproduced from ref. [246].

30%Cu/30%Co/MgO

Log [RMgO/R]

experiment carried out deliberately at substoichiometric steam level to exacerbate coking (see Fig 10a). The progressive drop in C 1 products and increase in unconverted ethanol with time on-stream is typical, and correlates with weight gain due to coking, as confirmed after the run by in-situ temperatureprogrammed oxidation of the coke to CO 2 . Fig. 10a also shows the intermediacy of acetaldehyde and acetic acid, which grow initially, and then decay, as the coke level approaches a limit, at ∼8 wt% in this case. Doping with alkali, and Cs+ ion in particular, has the dual benefit of dramatically reducing coking rates as well as undesirable CH4 production, as shown in Fig. 10b. These effects are attributed to the superbasicity of the catalyst surface. This suppresses direct decarbonylation of the first intermediate, acetaldehyde (the source of CH4), binds acetate to the surface more aggressively, and promotes gasification of carbon and/or coke precursors, probably via activation of steam [247]. The source of coke has been shown to be acetic acid decomposition, which is known to deposit atomic C [248]. The pivotal role of acetic acid in the mechanism has been shown by DRIFTS [246].

30% Co/MgO

30% Cu/MgO

4000

3400

2800

2200

1600 1450 1345

1000 cm−1

Fig. 11 DRIFT spectra showing surface acetate species on catalysts during ethanol S-R. Figure reproduced from ref. [246].

Strong bands of adsorbed acetate are already seen over the working catalyst and model supports in-situ below 200°C (see Fig. 11). The doublet at ∼1600 and 1450 cm−1 is characteristic of the carboxylate moiety, while the weaker feature at 1345 cm−1 is attributed to the symmetric bend of the methyl group. The origin of the acetate species was found to be the Cannizzaro disproportionation reaction between

-47 -

CH3CHO:H2O = 2:1 (in He)

Log [RMgO/R]

50°C/1hour

50°C/20min

50°C/5min

4000

3400

2800

2200

1600

1000 cm−1

Fig. 12 DRIFT spectra of acetate species formed on MgO by exposure to CH3CHO/H2 mixtures at 50°C. Figure reproduced from ref. [244].

acetaldehyde and water vapour, which is facile over the basic support (see Fig. 12). These observations lead to the following mechanistic proposal for ethanol S-R:2 C2H5OH → 2 CH3CHO + 2 H2

..51.

2CH3CHO + H2O → CH3COOH + C2H5OH 52. RDS

CH3COOH + 2 H2O → 2 CO2 + 4 H2

..53.

C2H5OH + 3 H2O → 2 CO2 + 6 H2

S-R

The confirmation of acetate as pivotal intermediate in ethanol S-R is highly significant as it is also linked with ethanol synthesis (see p.28). This not only reinforces the case for the acetate route in both, but also suggests that microscopic reversibility holds, which is not obvious in the case of methanol with the possible intervention of methyl formate in S-R. Neither has any evidence been found for the involvement of ethyl acetate, the analogous ester from ethanol. It can hardly be coincidental that the best ethanol S-R catalyst developed in this Laboratory bears a striking resemblance to one major (IFP) industrial catalyst for the synthesis.

Just as in the synthesis, a major unresolved question is whether Cu/Co alloy is beneficial, or even necessary, for good performance in ethanol S-R. The results of preliminary studies in this Laboratory suggest that it is indeed important. Because conventional preparations are not ideal to optimize formation of this metastable alloy (see p. 34), a novel alternative route from ternary Cu/Co/Al alloys and quasicrystalline precursors [249] is being explored. Selective leaching of Al creates a Raney (skeletal) bimetallic structure in which evidence is seen in X-ray diffraction for bulk Cu/Co alloy. Such catalysts show remarkable insensitivity to deposited coke, unlike coprecipitated forms, and their activity actually grows on-stream, soon surpassing that of the latter in accelerated stability testing [250]. This distinctive feature has been tentatively linked with surface alloy, but still requires corroboration by surface analysis methods, e.g., gas chemisorption, X-ray photoelectron spectroscopy, etc., which is currently in progress. These catalysts will also be explored in higher alcohol synthesis in the near future. 4b.2.3 Summary Although good progress has been made in recent years, knowledge of the fundamental chemistry of alcohols S-R is still somewhat limited when compared to related industrial processes, such as methane S-R or methanol synthesis. However, in view of the imminent prospects for technical applications of simple alcohols as H2-storage forms, there is a growing need for advances in the field. Methanol S-R is kinetically and thermodynamically easier than ethanol S-R and, thus, closer to technical maturity. Work on ethanol is still at the fundamental stage but offers, as incentives for technical development, easier handling, greater thermal efficiency, and its wider availability as a genuinely renewable energy source for demonstration purposes, etc.

- 48 5. Recent advances in CO2 recovery technologies: A brief overview Trapping of CO2 for conventional uses, e.g., enhanced oil recovery, urea production, refrigeration, beverage carbonation, etc., has traditionally been practiced with scrubbing, or “gassweetening” technologies linked to major industrial plants for natural gas processing, ammonia, hydrogen, and ethylene oxide production. Ethanol plants, based on biomass processing and coupled fermentation, are becoming increasingly important sources [251]. Here the CO2 concentration in the off-gases is sufficiently high (>20 vol%), from the techno-economic viewpoint to justify the use of simple methods, such as absorption into chemical solvents like alkanolamines, and subsequent recovery by mild heating and/or pressure let-down. In recent years, environmental concern over atmospheric build-up of CO 2 is providing a new impetus to develop better ways to sequester this “greenhouse” gas, much of which originates from fossil-fuel power plants [252253]. As these utilities vent CO2 in relatively dilute form (≤ 10 %), ever more efficient technologies are needed to deal with the trapping problem. Remarkable, bordering on grotesque, artificial schemes for disposal of the CO2 recovered are under consideration, including pumping in liquefied form into the deep ocean, or storage in insulated solid balls 400m in diameter, dotted liberall y around the landscape! [254-255]. However, when compared against natural, albeit slower, alternatives like afforestation, these artificial methods of sequestration and their related high cost (much of which is linked to the disposal step), appears to make them a realistic option only in a “worstcase scenario”. Once the CO2 is pre-concentrated, it makes more sense to use it as a feedstock for production of fuels, or even better, chemicals with high “added value”, if commercial outlets on a large scale can be found. 5a. Trends in chemical absorption Modified, or “sterically hindered”, amines have been developed which offer several advantages

over conventional alkanolamines for CO2 recovery from power-plant flue gases [256257]. A normal amine like monoethanolamine (MEA, HOCH2CH2NH2) forms a stable carbamate with CO2 as below:2 R−NH2 + CO2 → R−NH3+ + RNHCO2− whereas an amine with a bulky alkyl (R) group like t-butyl, (CH3)3C−, destabilizes the carbamate, instead promoting the reaction:R−NH2 + CO2 + H2O → R−NH3+ + HCO3− in which the CO2/amine stoichiometry increases from 0.5 to unity, thereby doubling the trapping capacity. It also leads to lower energy demand for (amine) solution regeneration, although set against these benefits is a slower absorption process. These sterically hindered amines, known under the Exxon trademark FLEXSORB  , are already used in over 30 plants worldwide for selective removal of CO2 and H2S [258]. 5b. Perm-selective and microporous hollow-fibre membranes The general trend in industrial gas separation and processing is towards the use of membranes of both polymeric and metal alloy materials. They are becoming popular due to their amenability for use in continuous operations. Polymer membranes also find extensive application in the treatment of complex liquids and vapours, e.g., in medical and filtration devices, where thermal stability is not an important consideration. The task of CO2 recovery (from mostly N2) is another case where they appear to be useful, although selectivities (to CO 2 permeation) of conventional gas absorption membranes are still orders of magnitude below those required to be commercially interesting, i.e., several hundredfold [258]. Development of “facilitated transport” membranes may be a way forward. In this type, a carrier molecule with high diffusivity chemically binds CO2, conveys it through the membrane, and releases it again on the low-pressure side [259]. However, the

-49 -

most promising option is to use hydrophobic microporous, hollow-fibre membranes. These offer easy access to the gas stream of a very large area of polar liquid absorbent, technology which is already used for, e.g., efficient removal of SO2 from flue gases. The main operational advantages over a counter-current, packed column, arrangement are low pressure drop, compactness, elimination of hydrodynamic problems, and easy independent control of gas and liquid flows. The fi rst benefit al one wo ul d resu lt i n substantial savings in pumping costs, currently the main operating expense in conventional absorption units. For CO 2 recovery, preliminary economic assessment indicates a saving of roughly 30% on the equipment capital outlay alone. The predicted effect on the generating efficiency of a hypothetical coal-fired power plant with CO2 removal based on this concept is to raise it by over 1% relative, or to restrict the loss against the same plant without CO2 removal to ∼25% [258]. Further improvements can be expected in hollow-fibre membrane modules in terms of greater durability. This would enable operation at higher pressure, thereby driving the absorption process and improving trapping efficiency substantially in single-pass operation. 5c. Adsorption on porous solids Selective adsorption of CO2 onto porous solids is another technique of interest, stimulated by the prospect of exploiting pressure-swing adsorption (PSA) as an economic way to separate simple gas mixtures. Synthetic zeolites appear quite promising in this context [260]. The basis of the separation of CO2 from N2 is the attractive interaction between the large quadrupole moment of the CO2 molecule with the electric field gradient (EFG) in the supercages of the zeolite. This is the highest possible in the A type (Si/Al = 1), though still quite high in the Y type (Si/A1 = 2.4). Separation factors > 400 were found for NaY around ambient temperature, and nearly 20 for Silicalite-1, which has minimal EFG. From isosteric adsorption studies and data-fitting to a Langmuir-Freundlich isotherm, extremely high CO2 capacities are

indicated, lying in the range 3.5−9.5 mmol.g−1. Due to the high heats of adsorption involved, uptakes close to these limits were achieved already at low pressures (∼1 kPa). Although the A and Y zeolites investigated were judged unsuitable for PSA applications per se (their interaction with CO2 is actually too strong!), the data illustrates well the promise inherent in this approach. Rapid advances may be expected in view of the wide range of compositions (Si/A1 ratios) available, and the relative ease with which adsorptive properties of zeolites are tailored by cations. 5d. CO2 recovery from the atmosphere The ultimate strategy in CO2 trapping is to consider processing of the atmosphere itself. Closure of the “open” cycle, as shown schematically in Fig. 1 (p. 8), would decouple CO2 collection from concentrated sources such as those described, and allow direct integration with utilities for renewable H 2 production, e.g., hydro- or solar-powered electrolysis, together with on-site methanol synthesis. In such a scenario, domestic sustainable energy in the complementary forms of electricity and liquid (alcohol) fuels, efficiently produced, stored and distributed, would become a reality for many nations, and especially those in the Developing World. Although the economical recovery of CO 2 from its atmospheric level (0.035%) appears at first sight a daunting task, several research groups are giving serious consideration to it [261-264], and the viability of some aspects of process design have already been demonstrated on the laboratory scale. At this Institute, Stucki et al. [261-262] favour a CO2 trapping device based on hollow-fibre porous membrane bundles containing KOH as the liquid absorber, as shown below in Fig. 13. The advantage of this almost passive arrangement lies in its minimal energy expenditure. A 1M KOH solution has an equilibrium absorption capacity up to 0.55M in CO2 (as K2 CO3 ). This is fed gravitationally from the top of the module, while natural convection (wind) typical

- 50 -

absorption liquid

a

b

seal

absorption liquid

microporous membrane

air flow

frame 240 µm

hollow fibre

Fig.13 Scheme of a). typical porous hollow-fibre membrane, and b). modular assembly for atmospheric CO2 recovery. Reproduced with permission from ref. [262].

of most coastal regions is sufficient, even in cross-flow, to saturate the KOH in a single liquid pass. From experimental mass transfer coefficients, a module of 1 m2 geometric area, containing a membrane surface area of 18 m2 (or 23,000 m total fibre length), is estimated to reach saturation with a liquid flow of 2 cm s−1 and a wind velocity of just 3 m s−1. A single module would retrieve ∼4 kg CO2 per hour in continuous-flow operation. The CO2 can subsequently be desorbed, together with a stoichiometric amount of H2, during electrochemical acidification of the aqueous K 2 CO 3 in a cation-exchange membrane cell. However, complementary H2 must be supplied from water electrolysis, driven in tandem, to achieve a syngas stoichiometry for methanol synthesis. In this coupled scheme, the energy input is estimated to be < 1.4 MJ.mol−1, corresponding to an efficiency of >60% based on the higher heating value of the syngas (3H2+CO2) produced. A similar coupled system, based on fandriven (conventional) atmospheric CO2 recovery together with electrodialytic regeneration and solid-oxide electrolysis technology for syngas generation, is under development by scientists in Stuttgart [263-264]. A more speculative alternative for atmospheric CO2 pre-concentration, under exploration

at NREL, is via electrochemical pumping based on redox active CO2 carriers like quinones. As with all electrochemical methods (fuel cells etc.), one potential advantage is the avoidance of the Carnot cycle, which ultimately limits the efficiency of all processes involving thermal energy transformations. The major problem here is to find carriers with a suitable combination of properties, i.e., electrochemical stability, wide ranging CO2 binding constants in the reduced (high-affinity) and oxidized (low-affinity) forms, and probably most important of all, oxygen tolerance [265]. 6. Prospects for alcohols synthesis by electrocatalytic reduction of CO2 6.a Background This field strictly warrants a full review which is beyond the scope of the present article. The intention here is rather to present a few highlights of recent advances which have rejuvenated research interest and raised the prospects of converting CO2 efficiently and selectively to simple alcohols, possibly on a technical scale in the longer term. Another intention in this Section, from a fundamental perspective, is to illustrate the complementarity between electrocatalytic and thermal catalytic processes and the growing perception that the underlying mechanism(s), at least in the case of CO2 hydrogenation, in both may have more parallels than realized heretofore. The electrocatalytic route to alcohols from C O 2 h as t wo i nt ri n s i c a dv ant a ge s o v er conventional heterogeneous catalysis. First, it operates at temperatures close to ambient where the thermodynamics for alcohols synthesis, at least from CO2 and H2, are favourable (see Fig. 2, p. 12). Second, it offers the prospect of a “one-pot” synthesis because the energy carrier needed to drive the reduction process, hydrogen, is generated in-situ. It also has the operational convenience that the intrinsic kinetic limitations (activation energy barriers) to reaction are overcome by increasing the electrical bias for reduction, i.e., simply applying an

- 51 -

overvoltage. However, set against these benefits is the difficulty of forestalling H2 evolution by promoting useful H/CO2 chemistry at the electrode surface, at least in aqueous electrolytes. This is a serious impediment to technical advance in the field, though the use of Pd membranes to decouple surface H concentration (θH) from the applied potential seems a promising area for research [266]. In addition, the sluggishness of diffusional kinetics in liquid media are often compensated in practice by further increasing the overvoltage, which is a waste of high-quality energy that should be utilized in more effective ways. Thermodynamic, kinetic, and product selectivity considerations in the electroreduction of CO2 have been reviewed [267-268]. Just as in heterogeneous catalysis, a complex multi-electron transfer process has the best chance to proceed efficiently via a sequence of relatively simple elementary steps, involving stable or metastable intermediates serving as “stepping stones” en route to the desired end-products. For example, the 6 e− reduction of CO2 to methanol is equivalent to a series of three 2 e− reduction steps as shown below:−0.20V

+0.034V

+0.232V

CO2(g )→ HCOOH(aq) → HCHO(aq) → CH3OH(aq) )

in which the standard redox potentials, E° (in volts at 25°C, pH = 0, vs. NHE), are also shown in superscript for each step. Such a pathway, for which there is some evidence, bears a striking resemblance to the favoured mechanism in the heterogeneously catalyzed synthesis (see p. 30). Indeed, anodized Cu electrodes are among the most active for the electrochemical production of methanol from CO 2 [269]. Another key to success lies in steering the synthesis chemistry through intermediates with low energy barriers, thereby minimizing the overvoltages required (vide infra). For example, routes to methanol via formate or carboxyl species [268] would be far more favourable energetically than via the solvated carboxylate radical anion, CO 2 • − (E° = −1.65V vs. NHE). Such reasoning is very similar to that

applied in heterogeneous catalysis. Until recently, attempts to reduce CO2 electrochemically were conducted primarily on metals (serving as both electrode and electrocatalyst) immersed in aqueous solutions where the substrate, typically in the form of CO32− or HCO3− ion, was extremely dilute. Products usually observed were CO, H 2 , HCOOH, HCHO, CH 4 , etc. In the case of formic acid synthesis over Cu, Zn, Pb, and Na amalgam electrodes, Faradaic efficiencies approached 100% [270] but the current densities were invariably too low (< 10 mA cm−2) to be of commercial interest, lying one or two orders of magnitude below that of a mature technology like water electrolysis. However, a flurry of publications, many originating from Japanese groups, and reports in several books and conference proceedings [56,271−272], have claimed remarkable progress in the last few years. Dramatic improvements in current densities, reduction of overvoltages, and good selectivities to alcohol products have now been achieved. 6.b Recent advances 6b.1 Current density enhancement The main breakthrough in improving the rate of CO2 electroreduction was inspired by the realization that the poor performance of earlier experimental designs was linked to a very large extent with the low rate of supply of CO 2 to the electrode surface, i.e., reactions were mass transfer-limited or operating in the diffusion-controlled kinetic regime. While this problem is conceptually obvious, experimental solutions are far from trivial. This has been dependent on advances in the development of the gas diffusion electrode (GDE), and of cells capable of working under high pressure, viz., “electrochemical autoclaves”. The GDE is a special device, originally developed for fuel cell technology, which optimizes the interfacial contact area between a gaseous reactant, liquid electrolyte, and solid electrode. Its implementation was recognized as a landmark advance in practical CO2 electroreduction by Halmann [273].

- 52 -

A number of research papers from Japanese groups have appeared, demonstrating the advantages of using high pressures of CO2, the GDE, or a combination of both [274−278]. Under pressures of up to 60 bar, current densities in the range 150−350 mA cm-2 have been obtained at Cu electrodes in aqueous KHCO3 electrolyte, in which the equilibrium CO 2 solubility is ∼1M, as compared to 30 mM at l bar [275]. High Faradaic efficiencies were measured (60−90%) and the reduction products were mainly CH4, H2, and HCOOH. The selectivity was a sensitive function of operating parameters like CO2 pressure, current density, and stirring speed. Another strategy is to avoid aqueous electrolytes altogether. The practical advantages are twofold. First, the solubility of CO2 is much higher in aprotic or non-aqueous solvents. Second, restriction of H2 evolution and better control of θH (and product selectivity) at the electrode surface may result. Saeki et al. [276, 278] have exploited this approach by working in completely miscible CO 2 /methanol mixtures, in which the conductivity and H-donor properties were adjusted by inclusion of low levels of water. Remarkable current densities (>400 mA cm -2 ) were measured, similar to those in commercial water electrolysis. The Faradaic efficiency at a Cu electrode was ∼85 %, m ost o f whi ch was as so ci at ed wit h t he production of methyl formate (MF) and CO. The ester was presumed to derive from a reaction between methanol and formic acid, a likely intermediate in CO2 reduction. A major feature in both the aqueous and “non-aqueous” works cited was the marked suppression of H2 evolution by the high pressures of CO2 employed. This may be attributed to adsorptive competition by intermediate CO. In experiments where a GDE was utilized under high CO2 pressure, Hara et al. [277] achieved an exceptional current density of 900 mA cm-2 at 30 bar on a Pt electrode, but with a lower, albeit still respectable, Faradaic efficiency of 46% for CO2 reduction. Unfortunately, the main product was CH4. Elsewhere, Kudo et al.

[274] explored Ni electrodes in aqueous medium. In this case, there was little effect of CO2 pressure on the current densities, which remained disappointingly low (< 10 rnA cm-2). However, their study was highly significant in terms of the products, which consisted of a range of saturated hydrocarbons conforming to an Anderson-Schultz-Flory distribution typical of Fischer-Tropsch chemistry in thermal catalysis. In other words, there was a promotional effect of electrical bias on the Ni at ambient temperature, sufficient to stimulate catalytic behaviour which ordinarily requires temperatures over 200°C. The authors attributed this tentatively to the negative polarization of Ni and its possible activation of CO desorption. Carbon monoxide is a known poison in FT chemistry over many metals. 6b.2 Reduction of overvoltage and selectivity to alcohols The reversible potential to drive an uphill free energy change in an electrochemical reaction under given conditions is given by:∆GT,P = −n F ET,P where E is potential, n is the number of electrons transferred, and F is the Faraday constant. However, the voltage necessary in practice is always greater, and the difference is termed the overvoltage. This is the electrochemical kinetic analogy of the activation energy in thermal catalysis, and electrocatalysts are needed to mitigate this value, thereby improving process efficiency. Alcohols synthesis from CO2 is only slightly uphill with respect to H2 evolution in aqueous medium, e.g., for ethanol:2CO2 + 9H2O + 12e− → C2H5OH + 12OH− where E = −0.33 V (vs. NHE) at pH 7 [279]. However, the cathode overvoltage for CO2 reduction in the absence of an electrocatalyst is from −1.5 to −2.0 V. This limits process efficiency to below 60% even before allowance for Faradaic losses and overvoltages at the O2−evolving anode.

-53-

Most electrocatalysis research has focused on anchored metal complexes and macrocycles as electron transfer mediators [280], e.g., Ni and Co porphryns and cyclams (cyclam = 1,4,8,11tetraazatetradecane), which were originally developed for homogeneous catalysis. However, the overvoltages are still quite high (1 V) and CO is almost the only product. In a significant recent advance, Ogura et al. [281-283] have achieved both substantial reduction of overvoltages (to −0.6V), and unusual selectivities to simple alcohols, and ethanol in particular. They utilized a Pt-supported dual-film (polyaniline or polymethylpyrrole/Prussian blue) electrode carrying immobilized complexes of Fe(II) and Co(II). However, the Faradaic efficiencies were rather low at below 15%. The most encouraging results in terms of alcohols selectivity and efficiency in CO2 electroreduction have come from the Eltron group in Colorado. Previous (Japanese) work on Cu electrodes had reported good selectivities to ethylene and ethanol at high Faradaic efficiencies (>60%) in certain aqueous electrolytes [284]. With this hindsight, and drawing on evidence from thermal catal ysis that alcohols synthesis activity may be linked to mixed oxidation states of the metals, Cook et al. [285286] investigated Cu perovskites (A1.8B0.2CuO4, where A = La, Pr, Gd, and B = Sr, Th, ) as semiconductor electrodes in a Teflon-bonded carbon GDE with 0.5M KOH as electrolyte. Total Faradaic efficiencies of ∼40% were measured for ethanol (30%) and n-propanol (10%), at current densities of 180 mA cm-2. Non-Cu perovskites were found to be almost inactive under the same conditions, confirming that Cu ions promote alcohols synthesis in CO2 electroreduction, possibly in synergy with metallic Cu°, small levels of which may be generated in-situ. The obvious follow-up would be to explore the effect of high pressures of CO2, but no report has appeared at the time of writing.

7. Concluding remarks In this review, the author has presented and discussed various aspects of physical chemistry and related technology t hat lie at t he heart of a potential solution to the energ y problem . A l ong-t erm renewabl e alt ernative to conventional, non-sustainable, energy use and its environmental consequences has been outlined. An energy scheme based on efficient storage of renewable hydrogen in simple alcohols, co-synthesized from recycled carbon dioxide, is in harmony with Nature and appears to offer realistic prospects for sustainable world development into the Third Millenium, provided that the technical groundwork has been laid in time. The role of heterogeneous catalysis applied on an industrial scale is pivotal in this respect. As such, both the fundamental and technical developments in the field are relevant, and this is reflected in the balance of the text. The realization that carbon dioxide has already been used, unwittingly, as an intermediate chemical feedstock since the inception of industrial syngas conversion is of both scientific and psychological importance. In the first, very practical, sense it means that the technological basis for “renewable petrochemistry” is already in place. Finally, the idea that what was once considered Man's greatest “waste product” can actually be put to valuable use carries a vital message of renewal and hope for the future. Of course, Nature herself makes no such arbitrary distinctions. Acknowledgments The author would like to express his deepest thanks to a number of colleagues at this Institute and E.T.H. Zurich. For his painstaking reading of the manuscript and invaluable scientific advice, the author is indebted first and foremost to Dr. Samuel Stucki. Valuable contributions from Drs. Gunther Scherer, Ruud Struis, Manfred Kobel, Marc Garland, Thomas Schucan, and Werner Hediger are gratefully acknowledged. For assistance with the literature

- 54 -

search and figure preparation, the author is grateful to Messrs. Alain Bill and Erol Uenala, respectively. Finally, the author is grateful to Professor Alexander Wokaun for Institute approval to publish this article.

GDE

Gas diffusion electrode

GHG

Greenhouse gas

HAS

Higher alcohol synthesis

ICE

Internal combustion engine

KIE

Kinetic isotope effect

MTG

Methanol to gasoline (Mobil)

Private Companies/Research Institutes

NHE

Normal hydrogen electrode

ABB Asea Brown Boveri Corporate Research Centre, Dättwil, Switzerland.

PEM(FC) Polymer electrolyte membrane (Fuel cell) PSA Pressure-swing absorption

EPFL Ecole Polytechnique Fédérale de Lausanne, Ecublens, Switzerland.

PV

Photovoltaic

ETHZ Eidgenossische Technische Hochschule, Zürich, Switzerland. IFP Institut Francais du Petrole, Rueil Malmaison, France.

RDS

Rate-determining step

RFG

Reformulated gasoline

S-R

Steam-reforming

KFA

TG

Thermogravimetry

TOF TPO

Turnover frequency (surfacespecific reaction rate) Temperature-programmed oxidation

WGS

Water-gas shift (reaction)

WHSV

Weight-hourly space velocity

Glossary

KFA Forschungszentrum, Jülich, Deutschland

KIST Korean Institute of Science and Technology, Seoul, Korea. NREL National Renewable Energy Laboratory, Golden, Colorado, U.S.A. PSI

Paul Scherrer Institut, Villigen, Switzerland.

Chemical terms and symbols Cn +

Products with carbon number ≥ n

Technical terms

Eapp

Apparent activation energy

AM(D)

alkali metal (doping)

DME

Dimethyl ether, CH3OCH3

ami

alkali metal ion(s)

DOM

Dioxomethylene, CH2O2δ−

ASF

Anderson-Schultz-Flory (product distribution)

ETBE

Ethyl t-butyl ether, C2H5OtC4H9

MA

Methyl acetate, CH3COOCH3

DRIFT(S)

Diffuse-Reflectance Infrared Fourier-Transform (spectroscopy)

MEA

Monoethanolamine, HOCH2CH2NH2

EFG

Electric field gradient

MF

Methyl formate, HCOOCH3

FFV

Flexible-fuel vehicle

FTIR

Fourier-transform Infrared

FT(S)

Fischer-Tropsch (synthesis)

MTBE Methyl t-butyl ether, CH3OtC4H9 NOx

Nitrogen oxides (NO, N2O, NO2)

θn

Normalized surface coverage by adsorbate species n

- 55 Bibliography 1. T. B. Johansson et al., preface to Renewable Energy: Sources for Fuels and Electricity, eds. T. B. Johansson, H. Kelly, A. K. N. Reddy and R. H. Williams, Island Press, Washington D.C., 1993. 2. R. A. Kerr, Science, 268, 1567, 16 June 1995. 3. C. D. Keeling, T. P. Whorf, M. Wahlen and J.van der Plicht, Nature, 375, 22 June 1995. 4. T. Reichhardt, Environ. Sci. Tech., 29, 36A, (1995). 5. E. V. Anderson, Chem. Eng. News, Nov.2, 1992, p.7. 6. See, e.g., Proceedings of the 9th and 10th International Symposia on Alcohol Fuels, ISAF IX (Florence, Italy,1991) and ISAF X (Colorado Springs, U.S.A.,1993). 7. D. L. Inman, Chem. Eng. News, Aug.1, 1994, p.8. 8. T. Y. Chang, R. H. Hammerle, S. M. Japar and I. T. Salmeen, Environ. Sci. Tech., 25, 1190 (1991).

IAHE (eds., T. N. Veziroglu, C. Derive and J. Pottier), 1992. 19. S. Stucki and T. Schucan, VDI Berichte *1129, 175 (1994). 20. C. E. Wyman, R.L. Bain, N. D. Hinman and D. J. Stevens, in Renewable Energy: Sources for Fuels and Electricity, eds.T.B.Johansson, H.Kelly, A. K. N. Reddy and R. H. Williams, Island Press, Washington D.C., 1993, Chap. 21, p. 865. 21. P.Courty, P. Chaumette, C. Raimbault and P. Travers, Rev. Inst. Francais du Petrole, 45, 561 (1990). 22. B. Hohlein, D. Mausbeck, E. Supp and P. König, Proc. 9th Int. Symp. Alcohol Fuels, USAF IX), Florence, Italy, 1991, vol. 1, p.43. 23. L.A.B. Cortez, S.M.G. Guerra and R.A. Lamparelli, Proc. 9th Int. Symp. Alcohol Fuels, Florence, Italy, 1991, vol. 3, p.1027. 24. M. D. Jackson and C. B. Moyer, in KirkOthmer Encyclopaedia of Chemical Technology, 4th edn., Vol.1, Wiley-Interscience, 1991, p.826.

9. Climate Change Policy Initiatives: 1994 Update, OECD/IEA 1994.

25. P. L. Hanst and E. R. Stephens, Spectrosc. Internat., 2 (Jan. 1990), 44.

10. P. Ekins, Energy Policy, 22, 571 (1994).

26. D. Illman, Chem. Eng. News, April 12 1993, p.28.

11. C.-J. Winter and J. Nitsch, eds., Hydrogen as an Energy Carrier, Springer, Berlin, 1988. 12. J. O'M. Bockris, Energy: The Solar-Hydrogen Alternative, Halstead Press, New York, 13. L. O. Williams, in „Therapy for the Earth", special issue of Appl. Energy, 47, 97 (1994). 14. J. M. Ogden and J. Nitsch, „Solar Hydrogen", Chap. 22 of 'Renewable Energy: Sources for Fuels and Electricity, eds. T. B. Johansson, H. Kelly, A. K. N. Reddy and R.H. Williams, Island Press, Washington D.C., 1993, p. 925.

15. J. Gretz, B. Drolet, D. Kluyskens, F. Sandmann, and O. Ullmann, Int. J. Hydrogen Energy,19, 169 (1994). 16. K. Hassmann and H.-M. Kühne, Int. J. Hydrogen Energy, 18, 635 (1993). 17. S. Hynek, W. Fuller, J.Bentley and J. McCullough, Hydrogen Energy Progress X, eds., D. L. Block and T. N. Veziroglu, IAHE, 1994, Proc. 10th World Hydrogen Energy Conf., Cocoa Beach, Florida, June 1994, Vol.2, p. 985, 18. B. Duret and A. Saudin, Hydrogen Energy Progress IX, Proc. 9th World Hydrogen Energy Conf., Paris, France, June 1992, vol.1., p. 137,

27. T. Inagaki, T. Hirota and Z. Ueno, Proc. 3rd. Int. Symp. Alcohol Fuels Technology, Washington D.C., U.S.A., Apr. 1980, Paper III-47. 28. T. Hirota, .I.S.A.E. Rev. #4, 7 (1981). 29. T. Hirota, T. Ohta and H. Mitsumoto, Proc. 9th. Int. Symp. Alcohol Fuels, Florence, Italy, 1991, Vol. 1., p. 587. 30. L. Pettersson and K. SjostrOm, Combust. Sci. Tech., 71, 129 (1990). #

31. J. von Linde, Gas Wärme Int., 38, 9, 1989 (Caloric GmbH, D-82166, Grafelfing bei München, Germany). 32. D. O. Hall, F. Rosillo-Calle, R. H. Williams and J. Woods, in Renewable Energy: Sources for Fuels and Electricity, eds.T.B.Johansson, H.Kelly, A. K. N. Reddy and R. H. Williams, Island Press, Washington D.C., 1993, Chap. 14, p.593. 33. A. Y. Rozovskii, Kinet. Katal, 21, 97 (1980). 34. R. Bardet, J. Thivolle-Cazat and Y. Trambouze, J. Chim. Phys.,78, 135 (1981). 35. B. Denise and R. P. Sneeden, CHEMTECH, Feb. 1982, p.108.

- 56 36. G. C. Chinchen, P. J. Denny, D. G. Barker, G. D. Short, M. S. Spencer, K. C. Waugh and D. A. Whan, ACS Symp. Ser., 29, 178 (1984). 37. G. Liu, D.Wilcox, M.Garland and H.H.Kung, J. Catal, 90, 139 (1984); and 96, 251 (1985). 38. G. C. Chinchen, K. C. Waugh and D. A. Whan, Appl. Catal., 25, 101 (1986). 39. G. C. Chinchen, P. J. Denny, J. R. Jennings, M. S. Spencer and K. C. Waugh, Appl. Catal., 36, 1 (1988). 40. K. Klier, Adv. Catal., 31, 243 (1982). 41. G. A. Mills, Fuel, 73, 1243 (1994). 42. H. Göhna and P. König, CHEMTECH, June 1994, p. 36. 43. M. Specht, private communication. 44. D. F. Spencer, Ann. Rev. Energy Environ., 16, 1991, p.259. 45. J. T. Houghton, G. J. Jenkins and J. J. Ephraums, eds., Climate Change, The IPCC Scientific Assessment, Intergovernmental Panel on Climate Change, Cambridge University Press, U.K., 1990. 46. P.W.F. Riemer, I. C. Webster, W. G. Ormerod and H. Audus, Fuel, 73, 1151 (1994). 47. T. B. Johansson, H. Kelly, A.K.N Reddy and R. H. Williams, eds., in Renewable Energy: Sources for Fuels and Electricity, Island Press, Washington D.C., 1993, Chap.1, p. 1. 48. J. Gretz, J. P. Baselt, O. Ullmann and H.Wendt, Int. J. Hydrogen Energy, 15, 419 (1990). 49. J. R. Moreira and A. D. Poole, in Renewable Energy: Sources for Fuels and Electricity, eds. T.B.Johansson, H.Kelly, A. K. N. Reddy and R.H.Williams, Island Press, Washington D.C. 1993, Chap. 2, p.73. 50. W. H. Bloss, F. Pfisterer, M. Schubert and T. Walter, Progr. Photovoltaics, 3, 3 (1995). 51. D. E. Carlson and S. Wagner, in Renewable Energy: Sources for Fuels and Electricity, eds. T.B.Johansson, H.Kelly, A. K. N. Reddy and R.H.Williams, Island Press, Washington D.C. 1993, Chap. 9, p.403. 52. K. Zweibel and A. M. Barnett, in Renewable Energy: Sources for Fuels and Electricity,eds. T.B.Johansson, H.Kelly, A. K. N. Reddy and R.H.Williams, Island Press, Washington D.C. 1993, Chap. 10, p.437. 53. M. Graetzel, Plat. Met. Rev., 38, 151 (1994). 54. K. Hassmann and H. -M. Kühne, Int. J. Hydrogen Energy, 18, 635 (1993).

55. W. Lepkowski, Chem. Eng. News, March 6th 1995, p.22. 56. M.M. Halmann, Chemical Fixation of Carbon Dioxide, CRC Press, Boca Raton, 1993. 57. D. S. Newsome, Catal. Rev.-Sci. Eng., 21, 275 (1980). 58. See, e.g., I. Barin, Thermochemical Data of Pure Substances, VCH, Weinheim, 1989. 59. See, e.g., R. T. Sanderson, Chemical Bonds and Bond Energy, 2nd. edn., Academic Press, New York, 1976. 60. G. Peaff, Chem. Eng. News, Oct. 24 1994, p.13. 61. J. R. Crocco, Proc. 1994 World Methanol Conference, Geneva, Paper I-1. 62. Methanol Production and Use, eds., W.-H. Cheng and H. H. Kung, Marcel Dekker, New York, 1994. 63. “Methanol's New Horizons”, Nitrogen, NovDec. 1994, p.46. 64. K. Mansfield, Proc. 1994 World Methanol Conference, Geneva, Paper X-1. 65. A. V. Bridgwater, Appl. Catal. A: Gen., 116, 5 (1994). 66. R. D. Kelley, G. A. Candela, T. E. Madey, D. E. Newbury and R. R. Schehl, J. Catal., 80, 235 (1983). 67. H. E. Curry-Hyde, M. S. Wainwright and D. J. Young, German Pat. #DE38745371Z (28 Jan. 1993). 68. M. S. Wainwright and D. L. Trimm, Catal. Today, 23, 29 (1995). 69. A. Y. Kruglov, Chem. Eng. Sci., 49, 4699 (1994). 70. J. B. Hansen and F. Joensen, Natural Gas Conversion I, Elsevier, Amsterdam, 1991, p.457. 71. V. L. Halloin and S. J. Wajc, Chem. Eng. Sci., 49, 4691 (1994). 72. J. Topp-Jørgensen and J. R. Rostrup-Nielsen, Oil and Gas J., 84, 68 (1986). 73. A. M. Rouhi, Chem. Eng. News, May 29, 1995, p.37. 74. H. H. Kung and K. Smith, in Methanol Production and Use, eds., W.-H. Cheng and H. H. Kung, Marcel Dekker, New York, 1994, Chap.5, p. 200 ff. 75. R.P.W.J. Struis, S. Stucki and M. Wiedorn, J. Membrane Sci., in press.

- 57 76. J. A. Graham, T. R. Gerrity and J. M. Davis, Proc.10th. Int. Symp. Alcohol Fuels (ISAF X), Colorado Springs, 1993, Addendum, p.163. 77. C. B. Murchison, M. M. Conway, R. R. Stevens and G. J. Quarderer, Proc. 9th. Int. Congr. Catal., eds., M. J. Phillips and M. Ternan, Chemistry Institute of Canada, 1988, p.626. 78. S.Uchiyama, Y.Ohbayashi, T.Hayasaka and N. Kawata, Appl. Catal., 42, 143 (1988). 79. M. Bowker, Catal. Today, 15, 77 (1992). 80. A. Muramatsu, T. Tatsumi and H. Tominaga, J. Phys. Chem., 96, 1334 (1992). 81. X. Xiaoding, E. B. M. Doesburg and J. J. F. Scholten, Catal. Today, 2, 125 (1987). 82. E. K. Poels and V. Ponec, Catalysis, Spec. Per. Rep., Royal Society of Chemistry,Vol. 6, p.196, 1983. 83. A. Bill, J. G. Highfield, B. Eliasson and E. Killer, Book of Abstracts, 3rd. Int. Conf. Carbon Dioxide Utilization (ICCDU III), Oklahoma, Apr.30 - May 4, 1995. 84. J. B. Hansen, Haldor-Topsøe, private communication. 85. C.D. Chang, Catal. Rev.-Sci. Eng., 25, 1(1983). 86. K. Tanabe, Catalysis by Acids and Bases, eds. B. Imelik et al. Elsevier, Amsterdam, 1985, p. l. 87. B.Elvers, S.Hawkins, G.Schultz, eds., Ullmann's Encyclopaedia of Industrial Chemistry, vol.A16, 5th edn., VCH,Weinheim, 1990, p. 465. 88. B. Juran and R. V. Porcelli, Hydrocarbon Processing, Oct. 1985, p.85. 89. S.-J. Uhm, S.-H. Han, K.-D. Jung and O.-S. Joo, Proc.10th. Int. Symp. Alcohol Fuels (ISAF X), Colorado Springs, 1993, Vol. 2, p. 763. 90. C. D. Chang, in Methanol Production and Use, eds., W.-H. Cheng and H. H. Kung, Marcel Dekker, New York, 1994, chap. 4, p.159 ff. 91. See, e.g., M. Aresta and G. Ford, eds., Carbon Dioxide as a Source of Carbon, Reidel, Dordrecht, 1987. 92. J. Wambach and H.-J. Freund, in Carbon Dioxide Chemistry:Environmental Issues, eds. J. Paul and C.-M. Pradier, Royal Society of Chemistry, 1994, p. 31. 93. A. F. Carley, M. W. Roberts and A. J. Strutt, Catal. Lett., 29, 169 (1994). 94. F. Solymosi, in Carbon Dioxide Chemistry: Environmental Issues, eds., J. Paul and C.-M Pradier, Royal Society Chemistry, 1994, p. 44.

95. F. Solymosi and L. J. Bugyi, J.C.S. Faraday Trans. I, 83, 2015 (1987). 96. F. M. Hoffmann, M. D. Weisel and J. A. K. Paul, in Carbon Dioxide Chemistry: Environmental Issues, eds., J. Paul and C.-M.Pradier, Royal Society of Chemistry, 1994, p. 55. 97. J. P. Hindermann, G. J. Hutchings and A. Kiennemann, Catal. Rev.-Sci. Eng., 35, 1 (1993). 98. G. R. Scheffer and T. S. King, J. Catal.,116, 488 (1989). 99. H. P. Bonzel, A. M. Bradshaw, and G. Ertl, eds., Physics and Chemistry of Alkali Metal Adsorption, Elsevier, Amsterdam, 1989. 100. M. P. Kiskinova, Poisoning and Promotion in Catalysis based on Surface Science Concepts and Experiments, Elsevier, Amsterdam, 1992. 101. G. v.d. Lee and V. Ponec, Catal. Rev.-Sci. Eng., 29, 183 (1987). 102. J. R. Jennings, R. M. Lambert, R. M. Nix, G. Owen and D. G. Parker, Appl. Catal., 50, 157 (1989). 103. See, e.g., R. Burch, ed., “In-Situ Methods in Catalysis”, Catal. Today, 9, 1991. 104. I. M. Hamadeh, D. King and P. R. Griffiths, J. Catal., 88, 264 (1984). 105. S. Bailey, G. F. Froment, J. W. Snoeck andK. C. Waugh, Catal. Lett., 30, 99 (1995). 106. D. B. Clarke and A. T. Bell, J. Catal., 154, 314 (1995). 107. A. Wokaun, J. Weigel, M. Kilo and A. Baiker, Fresenius J. Anal. Chem, 349, 71 (1994). 108. T. Tagawa, N. Nomura, M. Shimakage and S. Goto, Res. Chem. Intermed., 21, 193 (1995). 109. G. J. Millar, C. H. Rochester and K. C. Waugh, Catal. Lett., 14, 289 (1992). 110. S. Fujita, M. Usui, E. Ohara and N. Takezawa, Catal. Lett., 13, 349 (1992). 111. G. J. Millar, C. H. Rochester, S. Bailey and K. C. Waugh, J. C. S. Faraday Trans., 88, 2085 (1992). 112. D. D. Clarke, D.-K. Lee, M. J. Sandoval and A. T. Bell, J. Catal., 150, 81 (1994). 113. G. J. Millar, C. H. Rochester and K. C. Waugh, J. Catal., 155, 52 (1995). 114. J.-C. Lavalley, J. Lamotte, G. Busca and V. Lorenzelli, J. C. S. Chem. Commun., 1006 (1985).

- 58 115. G Busca, J. Lamotte, J.-C. Lavalley and V. Lorenzelli, J. Am. Chem. Soc., 109, 5197 (1987). 116. J. Saussey, J. C. Lavalley, J. Lamotte and T. Rais, J. C. S. Chem. Comm., 278 (1982). 117. a: A. T. Bell and L. L. Hegedus, eds., Catalysis under Transient Conditions, ACS Symp. Ser. 178, 1982; b: J. Happel, Isotopic Assessment of Heterogeneous Catalysis, Academic Press, 1986, c: J. G. Highfield, M. R. Prairie and A. Renken, Catal. Today, 9, 39 (1991); d: M. R. Prairie, J. G. Highfield and A. Renken, Chem. Eng. Sci., 46, 113 (1991).

134. G. J. Millar, C.H. Rochester and K.C.Waugh, J.C.S. Faraday Trans., 88, 2257 (1992). 135. S. G. Neophytides, A. J. Marchi and G. F. Froment, Appl. Catal., A86, 45 (1992). 136. Y. Nitta, 0. Suwata, Y. Ikeda, Y. Okamoto and T. Imanaka, Catal. Lett., 26, 345 (1994).

118. G. C. Chinchen and M. S. Spencer, Catal. Today 10, 293 (1991). 119. C. Rhodes, G. J. Hutchings and A. M. Ward, Catal. Today 23, 43 (1995).

137. T. Fujitani, M. Saito, Y. Kanai, T. Kakumoto, T. Watanabe, J. Nakamura and T. Uchijima, Catal. Lett., 25, 271 (1994). 138. T. Tagawa, N. Nomura, M. Shimakage and S. Goto, Res. Chem. Intermed., 21, 193 (1995). 139. M. -Y. He and J. G. Ekerdt, J. Catal., 87, 381 (1984). 140. M. R. Prairie, A. Renken, J. G. Highfield, K. R. Thampi and M. Graetzel, J. Catal., 129, 130 (1991).

120. J. F. Edwards and G. L. Schrader, J. Phys. Chem., 88, 5620 (1984).

141. R. Burch and R. J. Chappell, Appl. Catal., 45, 131 (1988).

121. G. Ghiotti and F. Boccuzzi, Catal. Rev.-Sci. Eng., 29, 151 (1987).

142. R. Burch, S. E. Golunski and M. S. Spencer, Catal. Lett., 5, 55 (1990).

122.D. Stirling, F. S. Stone and M. S. Spencer, New Frontiers in Catalysis, eds., L. Guczi et al., Elsevier, 1993, p. 1507. 123. M. E. Ekley, J. R. Jennings and M. S. Spencer, J. Catal., 118, 483 (1989).

143. F. Boccuzzi, E. BoreIlo, A. Zecchina, A. Bossi and M. Camia, J. Catal., 51, 150 (1978).

124. A. Takeuchi and J. R. Katzer, J. Phys. Chem., 85, 937 (1981).

146. M. A. Vannice, J. Catal., 50, 228 (1977).

125. G. A. Vedage, R. Pitchai, R. G. Herman and K. Klier, Proc. 8th. Int. Congr. Catalysis, Verlag-Chemie, Berlin, 1984, Vol.2, p. 47. 126. G. D. Sizgek, H. E. Curry-Hyde and M. S. Wainwright, Appl. Catal. A: Gen. 115, 15 (1994). 127. G. C. Chinchen, M. S. Spencer, K. C. Waugh and D. A. Whan, J. C. S. Faraday Trans. I, 83, 2193 (1987). 128. M. Muller, E. Tornquist, L. P. Nielsen, B. S. Clausen and H. Topsøe, Catal. Lett., 25, 1 (1994). 129. J. M. Campbell, J. Nakamura and C. T.Campbell, J. Catal.,136, 24 (1992). 130. P. B. Rasmussen, M. Kazuta and I. Chorkendorff, Surf. Sci., 318, 267 (1994). 131. J. Yoshihara, S. C. Parker, A. Schafer and C. T. Campbell, Catal. Lett., 31, 313 (1995). 132. J. Szanyi and D. W. Goodman, Catal. Lett., 10, 383 (1991). 133. M. Bowker, H. Houghton and K. C. Waugh, J. C. S. Faraday Trans. I, 77, 3023 (1981).

144. M. S. Spencer, Catal. Lett., 32, 9 (1995). 145. J. Nakamura, J. M. Campbell and C.T. Campbell, 1 C. S. Faraday Trans., 86, 2725 (1990). 147. P.A. Taylor, P.B. Rasmussen and I. Chorkendorff, J.C.S. Faraday Trans., 91, 1267 (1995). 148. K. Klier, V. Chatikavanij, R. G. Herman and G. W. Simmons, J. Catal., 74, 343 (1982). 149. I. Chorkendorff, P. A. Taylor and P. B. Rasmussen, J. Vac. Sci. Technol., A10, 2277 (1992). 150. E. K. Thornton and E. R. Thornton, in Isotope Effects in Chemical Reactions, eds., C. J.Collins and N. S. Bowman, ACS Monograph 167, van Nostrand Rheinhold, New York, 1970, Chap. 4, p.213. 151. S. Holloway, in Interaction of Atoms and Molecules with Solid Surfaces, eds., V. Bortolani, N. H. March and M. P. Tosi, Plenum, New York, 1990, Chap. 17, p. 575 ff. 152. S. S.-B. Fu, Ph.D. Thesis, Univ. California, Berkeley, 1991. 153. F. Solymosi and A. Erdohelyi, J. Catal., 91, 327 (1985). 154. K.-H. Ernst, C. T. Campbell and G. Moretti, J. Catal., 134, 66 (1992).

- 59 155. T. van Herwijnen and W.A. de Jong, J. Catal., 63, 83 (1980). 156. T. van Herwijnen, R. T. Guczalski and W. A. de Jong, J. Catal., 63, 94 (1980). 157. S. P. S. Andrew, Post Congr. Symp., 7th. Int. Congr. Catal., Osaka, July 1980, Paper 12. 158. See, e.g., R. B. Anderson, The Fischer-Tropsch Synthesis, Academic Press, New York, 1984, Chap. 4, p. 100. 159. B. A. Sexton, A. E. Hughes and N. R. Avery, Surf. Sci., 155, 366 (1985). 160. P. B. Rasmussen, P. M. Holmblad, T. Askgaard, C. V. Ovesen, P. Stoltze, J. K. Nørskov and I. Chorkendorff, Catal. Lett., 26, 393 (1994). 161. Ch. Froelich, Ph.D. Dissertation #10433 , E.T.H. Zürich, 1993. 162. M. Bowker, R. A. Hadden, H. Houghton, J. N. K. Hyland and K. C. Waugh, J. Catal., 109, 263 (1988). 163. J. F. Walker, Formaldehyde, ACS Monograph Series, Rheinhold, New York, 3rd ed., 1964, p.214. 164. J. M. Ricart, J. Torras, F. Illas and J. Rubio, Surf . Sci., 307-309, 107 (1994). 165. Ph. Courty, P. Chaumette, C. Raimbault and Ph. Travers, Rev. Inst. Français du Petrole, 45, #4, Jul.-Aug., 1990, p.561. 166. E. M. Calverley and K. J. Smith, J. Catal., 130, 616 (1991). 167. G. A. Vedage, P. B. Himelfarb, G. W. Simmons and K. Klier, ACS Symp. Ser., 279, Washington D.C., 1985, p.295. 168. J. C. Nunan, K. Klier, C.-W. Young, P. B. Himelfarb, and R. G. Herman, J.C.S. Chem. Commun., 193 (1986). 169. G. R. Scheffer and T. S. King, J. Catal.,115, 376 (1989); and 116, 488 (1989). 170. M. P. Kiskinova, Poisoning and Promotion in Catalysis based on Surface Science Concepts and Experiments, Elsevier, Amsterdam, 1992, Chap.6, p.252ff.

174. J. G. Nunan, C. E. Bodgan, R. G. Herman and K. Klier, Catal. Lett., 2, 49 (1989). 175. A. Kiennemann, R. Kieffer, H. Idriss, P. Chaumette and D. Durand, Ind. Eng. Chem. Prod. Res. Dev., 30, 1130 (1991). 176. A. Deluzarche, J. P. Hinderrnann and A. Kiennemann, J. Mol. Catal., 31, 225 (1985). 177. D. J. Elliot and F. Penella, J. Catal., 102, 464 (1986). 178. Ph. Courty, D. Durand, E. Freund and A. Sugier, J. Mol. Catal., 17, 241 (1982). 179. R. M. Baillard-Letournel, A. J. Gomez Cobo, C. Mirodatos, M. Primet and J. A. Dalmon, Catal. Lett., 2, 149 (1989). 180. See, e.g., T. B. Massalski, ed., Binary Phase Alloy Diagrams, 2nd ed., ASM International, Ohio, 1990, Vol. 1, p. 758. 181. E. Kneller, J. Appl. Phys., 33, 1355 (1962). 182. T. L. Favre, G. v.d. Lee and V. Ponec, J.C.S. Chem. Commun., 230 (1985). 183. A. Kiennemann, C. Diagne, J. P. Hindermann, P. Chaumette and Ph. Courty, Appl. Catal., 53, 197 (1989). 184. J. Hackenbruch, W. Keim, M. Roper and H. Strutz, J. Mol. Catal., 26, 129 (1984). 185. A. Kiennemann, S. Boujana, C. Diagne, Ph. Courty and P. Chaumette, Natural Gas Conversion, Elsevier, Amsterdam, 1991, p. 243. 186. R. Kieffer and L. Udron, Book of Abstracts, 3rd. Int. Conf. Carbon Dioxide Utilization (ICCDU III), Oklahoma, Apr. 30-May 4, 1995 187. W. Chu, R. Kieffer, A. Kiennemann and J. P. Hindermann, Appl. Catal. A: Gen. 121, 95 (1995). 188. J. L.Davis and M. A. Barteau, Surf. Sci., 187(2), 387 (1987). 189. M. Ichikawa, CHEMTECH, 12, 674 (1982). 190. M. Ichikawa, Adv. Catal., 38, 283 (1992). 191. Y. Huang, X. Meng, Z. Dang, S. Weng and C. Zhang, J.C.S. Chem. Commun., 1025 (1995). 192. G. v.d. Lee, A. Bastein, J. v.d. Boogert, B. Schuller, H. Luo and V. Ponec, J.C.S. Faraday Trans. I, 83, 2103 (1987).

171. D. W. Goodman, in Physics and Chemistry of Alkali Metal Adsorption, eds., H. P. Bonzel, A. M. Bradshaw and G. Ertl, Elsevier, Amsterdam, 1989, p.293.

193. A. Kiennemann, R. Breault and J. P. Hindermann, J.C.S. Faraday Trans. I, 83, 2119 (1987).

172. K. J. Smith and R. B. Anderson, J. Catal., 85, 428 (1984).

194. T. Koerts and R. A. van Santen, J. Catal., 134, 13 (1992).

173. I. Boz, M. Sahibzada and I. S. Metcalfe, Ind. Eng. Chem. Res., 33, 2021 (1994).

195. H. Arakawa, H. Kusama, K. Sayama and K. Okabe, Book of Abstracts, 3rd.Int.Conf. Carbon

- 60 Dioxide Utilization (ICCDU III), Oklahoma, Apr.30-May 4, 1995. 196. J. Paul, R. A. dePaola and F.M. Hoffmann, in Physics and Chemistry of Alkali Metal Adsorption, eds., H. P. Bonzel, A.M. Bradshaw, G. Ertl, Elsevier, Amsterdam, 1989, p. 213.

211. See, e.g., A. J. Appleby and F. R. Foulkes, Fuel Cell Handbook, Van Nostrand Rheinhold, New York, 1989. 212. V. M. Schmidt, P. Bröckerhoff, B. Hohlein, R. Menzer and U. Stimming, J. Power Sources, 49, 299 (1994).

197. C. Wang, S. Li, W. Huang and L. Yin, Book of Abstracts,3rd.Int.Conf. Carbon Dioxide Utilization (ICCDU III), Oklahoma, Apr.30May 4, 1995.

213. R. E. Billings and M. Sanchez, Int. J. Hydrogen Energy, 20, 521 (1995).

198. M. Kishida, K. Yamada, H. Nagata and K. Wakabayashi, Chem. Lett., 555 (1994).

215. F. N. Büchi, B. Gupta, J. Halim, O. Haas and G. G. Scherer, Proc. Symp. Electrode Mater. Process Energy Conversion and Storage, S. Srinivasan, D. MacDonald and A. Khandkar, eds., Publ. Electrochem. Soc., 1994, p. 220 216. See, e.g., J. Garche, W. Vielstich and M. Waidhas, in Angewandte Chemie: 1st. Ulmer Elektrochemische Tage, Universitätsverlag, Ulm, Germany, 1994, p. 178. 217. B. Rauhe, F. R. McLarnon and E. J. Cairns, J. Electrochem. Soc., 142, 1073 (1995).

199. See, e.g., G. C. Bond, Heterogeneous Catalysis: Principles and Applications, O.C.S., Clarendon Press, Oxford, 2nd.ed., 1987, Chap.8, p. 85. 200. J. R. Rostrup-Nielsen, Steam Reforming Catalysts, Danish Technical Press, Copenhagen, 1975. 201. J. R. Rostrup-Nielsen, in Catalysis: Science and Technology, eds., J. R. Anderson and M. Boudart, Springer-Verlag, Berlin, 1984, Vol.5, Chap.1. 202. See, e.g., Natural Gas Conversion II, H. E. Curry-Hyde and R. F. Howe, eds., Studies in Surface Science and Catalysis, Vol. 81, Elsevier, Amsterdam, 1994. 203. J. R. Rostrup-Nielsen, in Natural Gas Conversion II, H. E. Curry-Hyde and R. Howe, eds., Elsevier, Amsterdam, 1994, p. 25. 204. Z. Zhang and X. E. Verykios, J.C.S. Chem. Commun., 71 (1995). 205. K. Hijikata, Proc. Int. Symp. CO2 Fixation & Efficient Utilization of Energy, Tokyo, 1993, RCCRU, Tokyo Institute of Technology, p. 7. 206. Y. Jamal and M. L. Wyszinski, Int. J. Hydrogen Energy, 19, 557 (1994). 207. M. L. Wyszinski, T. Wagner, M. R. Jones. H. West and Y. Jamal., Hydrogen Energy Progress X, D. L. Block and T. N. Veziroglu, eds., IAHE, 1994, Vol. 3, p. 1847. 208. H. Schaper, J. J. Berg-Slot and W.H. J. Stork, Appl. Catal., 54, 79 (1989). 209. O. G. Percival, D. N. Eley and A. L. Chaffee, Natural Gas Conversion II, H. E. CurryHyde, R. F. Howe, eds., Studies in Surface Science and Catalysis, Vol. 81, Elsevier, Amsterdam, 1994, p. 405. 210. K. V. Kordesch and G. R. Simader, Chem. Rev., 95, 191 (1995).

214. See, e.g., Proc. 3rd Grove Fuel Cell Symp., London, 1993, in J. Power Sources, 49 (1994).

218. C. Pu, W. Huang, K. L. Ley and E. S. Smotkin, J. Electrochem. Soc., 142, L119 (1995). 219. J. C. Amphlett, M. J. Evans, R. A. Jones, R. F. Mann and R. D. Weir, Can. J. Chem. Eng., 59, 720 (1981). 220. J. C. Amphlett, M. J. Evans, R. F. Mann, R. D. Weir, Can. J. Chem. Eng., 63, 605 (1985). 221. J. C. Amphlett, R. F. Mann and R. D. Weir, Can. J. Chem. Eng., 66, 950 (1988). 222. J. C. Amphlett, K.A.M. Creber, J. M. Davis, R. F. Mann, B. A. Peppley and D. M. Stokes, Hydrogen Energy Progress IX, IAHE, 1992, Vol.3, p. 1541. 223. J. C. Amphlett, K. A. M. Creber, J. M. Davis, R. F. Mann, B. A. Peppley and D. M. Stokes, Int. J. Hydrogen Energy, 19, 131 (1994). 224. J. Haggin,Chem. Eng. News, July 24, 1995, p. 8 225. B. Hohlein, G. Colsman, R. Menzer, P. Brockerhoff, E. Riedel and J. Bøgild-Hansen, Hydrogen Energy Progress X, D. L. Block and T. N. Veziroglu, eds., IAHE, 1994, Vol.3, p.1461. 226. R. Lemons, J. Power Sources, 29, 251 (1990). 227. J. C. Amphlett, R. F. Mann, B. A. Peppley and D. M. Stokes, Proc. 26th Intersociety Energy Conversion Eng. Conf., Boston, 1991, p.642. 228. P. Brockerhoff and B. Emonts, Proc. 10th Int. Symp. Alcohol Fuels, Colorado Springs, 1993, vol.2, p.1045.

-61229. C. J. Jiang, D. L. Trimm, M. S. Wainwright and N. W. Cant, Appl. Catal. A: Gen. 97, 145 (1993). 230. See, e.g., O. V. Krylov, Catalysis by NonMetals: Rules for Catalyst Selection, Academic Press, New York, 1970. 231. C. J. Jiang, D. L. Trimm, M. S. Wainwright and N. W. Cant, Appl. Catal., 93, 245 (1993). 232. J. C. Amphlett, R. F. Mann and B. A. Peppley, Natural .Gas Conversion II, H. E. Curry- Hyde and R.F. Howe, eds. Studies in Surface Science and Catalysis, Vol. 81, Elsevier, Amsterdam, 1994, p. 409. 233. N.Iwasa, S.Kudo, H.Takahashi, S.Masuda and N.Takezawa, Catal. Lett., 19, 211 (1993). 234. R. O. Idem and N. N. Bakhshi, Ind. Eng. Chem. Res., 33, 2047 (1994). 235. R. O. Idem and N. N. Bakhshi, Ind. Eng. Chem. Res., 33, 2056 (1994). 236. R. O. Idem and N. N. Bakhshi, Ind. Eng. Chem. Res., 34, 1548 (1994). 237. W. H. Cheng, Appl. Catal. A: Gen., 130, 13 (1995). 238. J. S. Lee, J. C. Kim and Y. G. Kim, Appl. Catal., 57, 1 (1990). 239. T. lkarashi, Chem. Econ. Eng. Rev., 12, 31 (1980). 240. Hydrocarb. Process., 58, 176 (1979). 241. C. A. Luengo, G. Ciampi, M. O. Cencig, C. Steckelberg and M. A. Laborde, Int. J. Hydrogen Energy, 17, 677 (1992). 242. E. Y. Garcia and M. A. Laborde, Int. J. Hydrogen Energy, 16, 307 (1991). 243. M. Morita, A. Sato and Y. Matsuda, Nippon Kagaku Kaishi, 164 (1993). 244. M. Ishikawa, M. Morita, T. Kojima and Y. Matsuda, Prog. Batteries & Battery Mater., 13, 397 (1994). 245. J. G. Highfield, F. Geiger, E. Uenala, V. Shklover and Th. H. Schucan, Proc. 10th Int. Symp. Alcohol Fuels, Colorado Springs, Nov. 1993, Vol. 1, p. 238. 246. J. G. Highfield, F. Geiger, E. Uenala and Th. H. Schucan, Hydrogen Energy Progress X, eds., D. L. Block and T. N.Veziroglu, IAHE., 1994, vol. 2, p. 1039. 247. R. Meijer, Ph.D. Thesis, Univ. Amsterdam, 1992.

248. G. R. Schoofs and J. Benziger, Surf. Sci., 143, 359 (1984). 249. See, e.g., W. Steurer, Mater. Sci. Forum, 150-151, 15 (1994). 250. J. G. Highfield, E. Uenala, F. Geiger, W. Steurer, V. Shklover and B. Zhang, unpublished results. 251. S. J. Barer and K. M. Stern, in Catalytic Activation of Carbon Dioxide, ed., W.M. Ayers, ACS Symp. Ser., 363, 1988, Chap. 1, p. 1. 252. See, e.g., B. Eliasson, in Carbon Dioxide Chemistry: Environmental Issues, eds., J. Paul and C.-M. Pradier, Royal Society of Chemistry, 1994, p. 5. 253. J. H. Edwards, Catal. Today 23, 59 (1995). 254. F. Pearce, “The High Cost of Carbon Dioxide”, New Scientist, 17 July 1993, p. 26. 255. J. Rose, “Disposing of CO2 from Fossil-Fueled Power Plants”, Environ. Sci. Technol., 27, 1282 (1993). 256. G. Sartori, W. S. Ho, W. A. Thaler, G. R. Chludzinski and J. C. Wilbur, in Carbon Dioxide Chemistry: Environmental Issues, eds., J. Paul and C.-M. Pradier, Royal Society of Chemistry, 1994, p. 205. 257. T. Suda, M. Fujii, T. Miura, S. Shimojo, M. Iijima and S. Mitsuoka, in Carbon Dioxide Chemistry: Environmental Issues, eds., J. Paul and C.-M. Pradier, Royal Society of Chemistry, 1994, p. 222. 258. P. H. M. Feron, in Carbon Dioxide Chemistry: Environmental Issues, eds., J. Paul and C.-M. Pradier, Royal Society of Chemistry, 1994, p. 236. 259. J. J. Pellegrino et al, in Gas Separation Technology, eds., E. F. Vansant and R. Dewolfs, Elsevier, Amsterdam, 1990, p. 445. 260. L. V. C. Rees and J. A. Hampson, in Carbon Dioxide Chemistry: Environmental Issues, eds., I. Paul and C.-M.Pradier, Royal Society of Chemistry, 1994, p. 250. 261. S. Stucki, D. Bosch, A. Schuler and M. Constantinescu, Hydrogen Energy Progress IX, eds.; T. N. Veziroglu, C. Derive and J. Pottier, IAHE, 1992, Vol. 1, p. 309. 262. S. Stucki, A. Schuler and M. Constantinescu, Int. J. Hydrogen Energy, 20, 653 (1995). 263. M. Specht, A. Bandi, K. Schaber and T. Weimer, Proc. Int. Symp. CO2 Fixation & Effi-

- 62 cient Utilization of Energy, RCCRU, Tokyo, 1993, p. 165. 264. R. Frobose, Bild der Wissenschaft #5, 1995, 82. 265. D. L. Dubois, A. Miedaner, W. Bell and J. C. Smart, in Electrochemical and Electrocatalytic Reactions of Carbon Dioxide, eds., B. P. Sullivan, K. Krist and H. E. Guard, Elsevier, Amsterdam, 1993, Chap. 4, p. 94. 266. H. Yoshitake, K. Takahashi and K. Ota, J.C.S. Faraday Trans., 90, 155 (1994). 267. F. R. Keene, in Electrochemical and Electrocatalytic Reactions of Carbon Dioxide, eds., B. P. Sullivan, K. Krist and H. E. Guard, Elsevier, Amsterdam, 1993, Chap. 1, p. 1. 268. K. W. Frese, in Electrochemical and Electrocatalytic Reactions of Carbon Dioxide, eds., B. P. Sullivan, K. Krist and H. E. Guard, Elsevier, Amsterdam, 1993, Chap. 6, p. 145. 269. K. W. Frese, J. Electrochem. Soc., 138, 3338 (1991). 270. M.M. Halmann, Chemical Fixation of Carbon Dioxide, CRC Press, Boca Raton, 1993, Chap. 7, p.68 ff. 271. Electrochemical and Electrocatalytic Reactions of Carbon Dioxide, eds., B. P. Sullivan, K. Krist and H. E. Guard, Elsevier, Amsterdam, 1993. 272. Proc. Symp. Environ. Aspects Electrochem. Photoelectrochem., eds., M. Tomkiewicz, R. Haynes, H. Yoneyama and Y. Hori, Electrochem. Soc. Inc., Pennington, NJ, 1993. 273. M.M. Halmann, Chemical Fixation of Carbon Dioxide, CRC Press, Boca Raton, 1993, Chap. 7, pp. 85-87. 274. A. Kudo, S. Nakagawa, A. Tsuneto and T. Sakata, J. Electrochem. Soc., 140, 1541 (1993). 275. K. Ham, A. Tsuneto, A. Kudo and T. Sakata, J. Electrochem. Soc., 141, 2097 (1994). 276. T. Saeki, K. Hashimoto, Y. Noguchi, K. Omata and A. Fujishima, J. Electrochem. Soc., 141, L130 (1994).

277. K. Hara, A. Kudo, T. Sakata and M. Watanabe, J. Electrochem. Soc., 142, L57 (1995). 278. T. Saeki, K. Hashimoto, A. Fujishima, N. Kimura and K. Omata, J. Phys. Chem., 99, 8440 (1995). 279. Y. Hori, in Proc. Symp. Environ. Aspects. Electrochem. Photoelectrochem., Electrochem. Soc. Inc., Pennington, NJ, 1993, p.1. 280. M.M. Halmann, Chemical Fixation of Carbon Dioxide, CRC Press, Boca Raton, 1993, Chap. 7, pp. 82-85. 281. K. Ogura, K. Mine, J. Yam, and H. Sugihara, J.C.S. Chem. Commun., 20 (1993). 282. K. Ogura, K. Mine, J. Yano, and H. Sugihara, in Proc. Symp. Environ. Aspects. Electrochem. Photoelectrochem., Electrochem. Soc. Inc., Pennington, NJ, 1993, p. 66. 283. K. Ogura, H. Sugihara, J. Yano and M. Higasa, J. Electrochem. Soc.,141, 419 (1994). 284. Y. Hori, A. Murata, R. Takahashi and S. Suzuki, J.C.S. Chem. Commun., 17 (1988). 285. A. F. Sammells and R. L. Cook, in Electrochemical and Electrocatalytic Reactions of Carbon Dioxide, Elsevier, Amsterdam, 1993, Chap. 7, p. 217. 286. M. Schwartz, R. L. Cook, V. M. Kehoe, R. C. MacDuff, J. Patel and A. F. Sammells, J. Electrochem. Soc., 140, 614 (1993).