Catalysis: Role and Challenges for a Sustainable Energy

Abstract

After introducing the role of catalysis as pillar of chemical industry to reach the sustainability of the society, the contribution discusses catalysis for energy as one of the critical areas of development to respond to societal needs and which further demonstrate the link between catalysis, innovation and sustainability. In particular, some aspects of development of catalysts for the photo- and electro-driven conversion of carbon dioxide and water are analyzed. In PEM fuel cells, advanced anode electrocatalysts with improved performances and reduced sensitivity to poisoning by CO could be developed by proper understanding of the role of nanostructure, three phase boundary and metal particle-carbon substrate interaction. It is shown also that similar materials could be used for addressing the societal challenge of converting back CO₂ to fuels. New results on the electrocatalytic conversion of carbon dioxide to liquid fuels (isopropanol, in particular) are reported and it is evidenced how in perspective, by combining these electrodes to a nanostructured titania photoanode, it is possible to realize photoelectrocatalytic devices which have the potential to capture CO₂ and convert it to liquid fuels (long-chain alcohols and hydrocarbons) using solar energy and water in “artificial energy trees”.

1 Introduction

Catalysis in an enabling technology to promote sustainability, environment, energy, health and quality of life [1]. In addition to promote feasibility, eco-efficiency and economics of over 90% of the chemical processes, catalysis is a critical element for the sustainability in many fields out of chemical industry, from mobility (catalytic converters in vehicles, but also cleaner fuels, better lubrificants, etc.) to a wide range of emission clean-up technologies (DeNOx for power plants, etc.).

Catalysis has been often considered only a tool in chemical industry, e.g. like a necessary component to develop a process, but not as an enabling factor and a pillar for the chemical industry. There are several examples which instead demonstrate that innovation in chemical industry and the creation of new sustainable processes derive from the development in catalysis [2]. The discovery of TS-1 has opened the possibility of new large-scale processes using H₂O₂ as oxidant, like the synthesis of propylene oxide [3], caprolactam and phenol; the first two are already at a commercial scale (DOW and SUMIMOTO processes, respectively) and good prospects exist also for the third one. These new processes allow to reduce process complexity and risks with respect to the old routes, decrease energy consumption and waste emissions, and create new market opportunities. In other words, they fulfill perfectly the E³ requirements for sustainability: energy, economy and environment. The discovery of TS-1, i.e. that isolated Ti ions in a MFI silica framework are able to activate H₂O₂ generating peroxo-titanate species highly selective in epoxidation and hydroxylation reactions, opened the door to the use of H₂O₂ as clean and selective oxidant in a larger range of other applications in the field of specialty and fine chemicals, starting from the synthesis of diphenols, the first commercial process for TS-1. At the same time, the discovery has stimulated a large research effort, at both the industrial and academic side, on the investigation of H₂O₂ reactions catalyzed by zeo-type materials containing isolated transition metal centers [4]. A range of new synthesis possibilities, some of them already at a commercial scale, derived from this field of investigation [5]. The increasing demand of H₂O₂ has intensified the research on the direct synthesis of H₂O₂ from H₂/O₂ [6, 7], particularly suited for small/medium productions and which reduces the environmental impact of the actual commercial process (alkylanthraquinone route). This new possibility opens new perspectives for a larger use of H₂O₂ in various processes out from the chemical industry field, from pulp/paper industry to technologies for wastewater treatment. In all these cases, the use of H₂O₂ allows a significant decrease of the environmental impact and a better use of the resources, e.g. a further step towards a sustainable society. This is an example of the virtuous cycle introduced from the discovery of a new catalyst. Development of a sustainable chemistry and society requires concerted actions in many aspects, from materials to reaction engineering, but catalysis plays a special role. There are more possible examples, but already this one evidences how catalysis is not a tool, but a pillar of chemical industry to reach the sustainability of the society.

1.1 Catalysis and Sustainable Chemistry

Research on catalysis has fast changed over the years to respond to this increasing role as core component for a sustainable chemistry. Research was often based on a trial-and-error approach in the past, but the increasing availability of advanced methodologies for understanding catalysis and a controlled preparation on the molecular, nano-, meso- and micro-scale level have consolidated catalysis as a fundamental science bridging basic and applied research over the last two decades. The Nobel prize to prof. G. Ertl in 2007 is a further demonstration of this concept. Catalysis is the science of complexity, being necessary to combine knowledge from molecular level to reactor level and which thus span in a wide time and space range. Understanding catalysis thus requires to integrate an exceptional wide range of experimental techniques which takes a long time to be developed. However, catalysis science grows now over solid bases and can address the new societal challenges of twenty-first century, first of all the problem of sustainability.

The increasing concerns for the too fast depletion of natural resources and global issues (from climate change to pollution) have pushed the need to accelerate the transition to a sustainable society. Realizing a more sustainable chemistry is a critical part of this process, because chemistry is pervasive to all societal objectives, from the availability of improved materials which can reduce the use of natural resources (for example, light materials for transport which allow to reduce energy consumption) to novel solutions to improve health (for example, nanomaterials for target drug delivery). A sustainable society requires to maintain the eco-balance in a world with increasing number of peoples accessing to resources, from energy to other feedstocks. Development of new solutions to these demanding problems requires an intense and innovative effort. Catalysis, being critical for the largest majority of chemical processes, is thus one of the critical elements to achieve these objectives.

The European Technology Platform on Sustainable Chemistry (ETP SusChem), promoted by Cefic (European Chemical Industry Council) and EuropaBio (European Association for Bioindustries), and related documents (Vision, Strategic Research Agenda and Implementation Action Plan) [8] evidences how reaching sustainability objectives requires a progress in catalysis. For example, the future F³-Factory (Future, Fast, Flexible), the visionary idea for the future of chemical production, has catalysis as one of the pillars. Future sustainable F³ chemicals production will combine a much broader range of production scales with interlinking technologies and logistics. An important strategy to meet the F³ challenge is process intensification. This is a strategic and interdisciplinary approach employing different tools (such as micro reaction technology and modularization) to improve processes holistically. However, a core technology to achieve process intensification is a new design of catalysts (for example, to develop catalysts suitable for microreactor technology), evidencing how advances in catalysis are strictly related to innovation and sustainability in chemical production.

1.2 Catalysis for Energy

The progressive revolution in chemical and fuel areas deriving from the socio-political pressure of new uses in biomass resources [9] is another factor which determines the need to intensify the research on new catalyst solutions. Handle biomass is far more complex than oil and convert biomass to a concentrated and easy transportable form of energy is more difficult than for the equivalent based on oil derivatives. Biomass utilization requires therefore an intense research effort. There are many process steps which require the development of novel catalysts, starting from the need of efficient and stable solid catalysts for vegetable oil transesterification, to solid catalysts for cracking, hydrolysis or selective depolymerization of (hemi)cellulose and lignin, and to new enzymes for fermentation to products different from ethanol. The passage from the first to the second generation of biofuels requires an intensified research on new catalysts.

More in general, catalysis for energy is one of the critical areas of development to respond to societal needs which further demonstrate the link between catalysis, innovation and sustainability. A book on this topic, resulting from a specific workshop organized in 2006 by the EU Network of Excellence Integrated Design of Catalytic Nanomaterials for a Sustainable Energy and Production (IDECAT), will soon appear [10]. An extensive report of a similar workshop organized by the US Department of Energy (DoE) [11] has been also published. There are large coincidences between the conclusions, challenges in catalysis and priority areas which were identified. In particular, three priority research directions for advanced catalysis science for energy applications were evidenced.

• 1. Advanced catalysts for the conversion of heavy fossil energy feedstocks.

• 2. Advanced catalysts for conversion of biologically derived feedstocks and specifically the deconstruction and catalytic conversion to fuels of lignocellulosic biomass.

• 3. Advanced catalysts for the photo- and electro-driven conversion of carbon dioxide and water.

The depletion of light, sweet crude oil has caused increasing use of heavy oils and other heavy feedstocks. The complicated nature of the molecules in these feedstocks, as well as their high heteroatom contents, requires catalysts and processing routes entirely different from those used in today’s petroleum refineries. The need to use biomass in a not competitive way to food requires to develop new innovative catalytic routes, starting from the controlled biomass deconstruction to the upgrading/valorization of the obtained products. Both areas are very challenging and open also new perspectives in terms of needs and opportunities for catalysis. These areas are already covered in other manuscripts of this special issue dedicated to the conference “Catalysis for Society” (Krakow, Poland, May 2008) and thus the focus of this contribution will be to discuss some relevant aspects of the third great challenge indicated in the cited US DoE report, e.g. the development of advanced catalysts for the photo- and electro-driven conversion of carbon dioxide and water. The aim is to provide an outlook on the possibilities and problems to overcome, but not to provide a systematic review of the state-of-the-art on this topic.

Catalytic conversion of carbon dioxide to liquid fuels facilitated by the input of solar or electrical energy presents an immense opportunity for new sources of energy. Furthermore, the catalytic generation of hydrogen from water could provide a carbon-free source of hydrogen for fuel and for processing of fossil and biomass feedstocks. These reactions still represent grand-challenges for a sustainable energy, but there are several opportunities for catalysis to really allow a breakthrough in this area.

2 Catalysis and Fuel Cells

Catalysis for fuel cells is another relevant area which needs an intensified effort of research. There are many relevant challenges for catalysis (increase of the efficiency of the chemical to electrical energy conversion and the stability of operations, reduce costs of electrocatalysts) which are necessary to contribute in making a step forward in the application of fuel cells out of niche areas. This objective requires also to develop efficient fuel cells fuelled directly with non-toxic liquid chemicals (ethanol, in particular, but also other chemicals such as ethylene glycol can be used). Together with the improvement in other fuel cell components (membranes, in particular), ethanol direct fuel cells require to develop new more active and stable electrocatalysts.

Different types of fuel cells are possible which can broadly classified according to the type of membrane used and temperature of operations. However, there are two main classes of fuel cells which are expected to be applied on a large scale: (i) proton-exchange membrane fuel cells (PEM-FC), showing a low weight, fast start-up and high power density to volume ratio which make them preferable for applications such as light-duty vehicles (either full or hybrid FC vehicles) or (ii) solid-oxide membrane fuel cells (SOFC) best suited for stationary applications. The first uses proton conducting membranes (Nafion^®, for example) and operates at around 80 °C. The second uses O²⁻ conducting membranes (Y-stabilized zirconia, for example) and operates in the 800–1000 °C temperature range. The first type should operate with H₂ as fuel. Hydrocarbons should be reformed on site in a pre-processor to generate H₂. Methanol or ethanol could be fed directly, but still a number of problems exists (alcohol cross-over, formation of oxygenated, alcohol conversion, low power density, etc.) for their use, even if already some pre-commercial examples of use in portable devices are in use.

SOFC in principle could operate directly with different fuels, although stability is still an issue. There are also still considerable problems of materials engineering. SOFC have an higher efficiency than PEMFC, due to waste heat which can be captured for increased efficiency. However, capital cost is typically much higher than for PEMFC, and for this reason SOFC are preferable for the production of stationary power, where efficiency is of high importance relative to capital cost, while PEMFC are preferable in automotive applications. We focus attention here only on PEMFC as an example of the role of catalysis in this field, but catalysis plays a relevant role also in the development of SOFC, even though the more critical aspects are related to materials characteristics.

It should be also briefly commented that fuel cells should be considered as an element of a broader area where catalysis is used in combination to electrons to perform selective reactions. For example, it is possible to feed waste streams from agro-food production to an electrochemical device essentially analogous to a fuel cell to produce at the same time electrical energy and chemicals [1]. This approach is interesting in SMEs for using wastewater or byproduct solutions derived from agro-food production. A limiting factor is the need to develop new nanostructured electrocatalysts, because conventional fuel cell electrodes have limited effectiveness and are mainly tailored for total oxidation. The challenge is to design new electrocatalysts which do not break the C–C bond, present a high activity in order to make the process industrially feasible and are stable in the strong basic medium required for the use of anion-exchange membranes.

2.1 The issue of Energy Efficiency

An important factor in PEM fuel cells is the efficiency of energy conversion. To clarify this concept and why the introduction of fuel cells in cars is a significant contribution towards sustainability it should be remembered that about one-third of fossil fuel consumption (and thus also of CO₂ emissions) is related to transport. Actual vehicles engine efficiency is low. Only about 13–16% of gasoline energy content is effectively used to move the car (Tank-to-Wheel Life Cycle Analysis), due to the low Lower Heating Value (LHV) efficiency of the chemical → thermal → mechanical conversion (Carnot Cycle). This efficiency is theoretically two times higher in FCs (chemical → electrical → mechanical) due to the very efficient electrical to mechanical conversion, but practically the total efficiency is limited from the still poor chemical to electrical conversion (around 50% in PEMFC). Increase this efficiency is thus a critical factor to improve the overall efficiency in using energy in transport, with a potential enormous impact on sustainability of society in terms of both better use of resources and reduced CO₂ emissions.

A fundamental graph to understand this problem is presented in Fig. 1 which evidences a typical cell voltage versus current density curve in PEM fuel cells. The graph also evidences that three major factors contribute to the polarization curves and thus determines the loss in efficiency: in the first region at low current density kinetic control determines the behavior and the shape of the polarization curve is governed by the activation overpotential, which arises from the kinetics of charge transfer reactions across interfaces. It represents the magnitude of activation energies when reactions propagate at the rate required by the current and depends on the type of reactions and catalyst materials, electrode microstructure, reactant activities, electrolyte material, temperature and current density. The activation overpotential for hydrogen oxidation is negligible compared to the overpotential associated with oxygen reduction reaction (ORR) at the cathode.

Fig. 1

Cell potential and power density versus current density in a PEM fuel cell operating at room temperature, with the indication of the main factors controlling the performances in the different indicated regions

After the rapid drop due to the activation overpotential, fuel cell polarization exhibits a pseudolinear behavior as the current density increases. In this region, the shape of the curve is governed by the Ohmic overpotential or IR-loss, which arises from the resistance of fuel cell components. Bulk materials and interfaces between components display an intrinsic resistance to electron flow, and electrolyte materials offer resistance to transport of ions, which carry the current in the electrolyte phase. The largest single contribution to cell resistance is usually the electrolyte resistance to ionic current. However, the slope depends also on other factors which are typically not considered. As discussed later, the characteristics of the metal-carbon support interaction also determine in a significant manner the performances, evidencing that the problem of attributing the loss of performances to only Ohmic resistance, e.g. to electrolyte resistance is an oversimplification.

At large current density, the shape of the curve is determined by the mass transfer overpotential. When current density reaches a certain level, sluggishness of the mass transfer processes starts to limit the supply of reactants to the electrodes, and cell voltage begins to decrease rapidly with increasing current demand. Mass transfer overpotential can be viewed as an activation energy required to drive mass transfer at the rate needed to support the current.

The fuel cell should operate at the highest power density (maximum in the curve indicated with triangles in Fig. 1) and the difference in this condition between cell voltage and open circuit cell voltage is essentially related to the loss of thermal energy. Decreasing the slope of the curve will thus not only increase the maximum power density of the fuel cell, e.g. increase of the cell performance, but also increase the efficiency of chemical to electrical energy conversion.

It is usually considered that the only possibility to improve these aspects is associated to (i) a decrease of the overpotential of ORR, e.g. at the cathode site where O₂ reduction occurs, and (ii) an optimization of membrane performances (electrolyte resistance to ionic current), besides to an improvement in fuel cell engineering (optimization of mass transfer). For catalysis, according to this common interpretation, the only possible contribution is associated to the development of new cathodes for oxygen reduction. However, as shown below, this is not correct and an improvement of anode design and a modification of metal-support interactions on the anode electrocatalyst have a marked influence on the performances of the FC. There are thus many additional parameters determining the performances of FCs which are usually not considered from who studies them from the traditional electrochemical engineering view. Also in this case, a catalytic approach to this field opens new perspectives not only of improving the performances, but also of understanding the mechanisms determining the behavior of FCs.

2.2 Characteristics of Electrodes and the Three-Phase Boundary

Commercial electrodes for fuel cells are prepared typically using as substrate a tissue of carbon macrofibers (carbon cloth—CC) which has the function of electron transport and moreover allows a homogeneous dispersion of the electrocatalyst and a good diffusion of the gases. On this CC, the electrocatalysts is deposited on the side at the contact with the proton-conducting membrane (Nafion^® 112, for example). For the anode side usually Pt(20 wt%) deposited on carbon black (for example, Vulcan XC-72 carbon black) is used as electrocatalyst. The typical size of noble metal particles is around 2 nm. For the cathode side (O₂ reduction) Pt–Ru bimetallic electrocatalysts deposited also on carbon black are used. On the other side of the CC, e.g. that at the contact with the gas phase, an hydrophobic porous layer is created (for example, by deposition of a Teflon solution), because it is necessary to avoid the dehydration of the proton-conducting membrane. This structure is indicated as gas diffusion electrode (GDE). The anode and cathode GDE are then hot pressed with the Nafion to realize the final composite known as membrane-electrode assembly (MEA) to be used in the PEMFC.

Let us now focus on the anode side, even if similar considerations are valid also for the cathode. The effectiveness of the reaction requires to realize an efficient three-phase contact between electrons, protons and the diffusing reactant (H₂). At the same time, a problem in PEMFC is the stability, because metal particles tend easily to sinter during operations with a consequent loss of performances. Both aspects can be improved introducing an hierarchically organized structure for the substrate (CC), e.g. by growing carbon nanofibres (CNF) or nanotubes (CNT) over the carbon macrofibres (CC) [12]. Figure 2 shows an examples of this type of materials prepared by deposition of Co-Fe/SBA-15 catalyst over carbon cloth and the use of these catalysts to grow CNT by propane CVD method. It may be observed that the CNT cover uniformly the carbon macrofibres allowing to improve the surface area of carbon and avoiding the need to use the carbon black as support for Pt. The use of hierarchically organized structures provides a better 3D geometry which favors the contact with the Nafion and improves the fast transport of electrons. A similar CNT-CC structure could be prepared also by depositing directly the Fe-Co particles over the CC, but their inclusion into the mesopores of SBA-15 allows the synthesis of CNT with smaller and controlled diameter (around 8 nm), as shown in the inset of Fig. 2 [12].

Fig. 2

Left image: electron microscopy images of (Co,Fe)/SBA-15 deposited over carbon cloth; in the inset higher resolution image of SBA-15 crystallites. Right image: carbon nanotubes covering a carbon macrofibre; they were obtained by propane CVD of the sample on the left; in the inset a higher resolution image of one of the carbon nanotubes growing from the SBA-15 crystallites

2.3 Nanostructured Carbon Electrodes

Research on nanostructured carbon materials have been relevant over the last years and several advances in this area have been reviewed by Lee et al. [13]. The formation of an improved 3D nano-architecture for PEM fuel cell electrodes would allow a better three-phase contact (reactant, proton and electron) which in turn determine the performances of the fuel cell. This concept has been extensively investigated for example by 3M company which developed NanoStructured Thin Film Catalysts (NSTFC) based on a layer of oriented crystalline organic whiskers on which the noble metal particles are deposited [14]. This layer is at the interface between the Nafion membrane and a conventional gas diffusion layer (GDL) and allows a much better three phase contact (gas diffusion of reactants or products and transport of protons and electrons) with consequent increase in the fuel cell performances.

The NSTFC-type electrodes, however, require a relatively complex procedure of preparation and therefore the direct growth of carbon nanotubes or nanofibres over conventional carbon-cloth materials used to prepare the gas diffusion layer in PEM electrodes could be an interesting alternative. The hierarchically organized carbon structure which derives from the growth of carbon nanotubes (CNT) or nanofibres (CNF) over the macrostructured carbon cloth (CC) would offer the possibility not only of a better dispersion of the Pt noble metal, but also to optimize the cited three phase boundary. However, the problems are more complex, as shown in Fig. 3 which compares the performances of a lab-scale PEM fuel cell in which the anode (H₂ oxidation side) is either a commercial conventional electrode (based on 20 wt% Pt on carbon black which is then deposited on a carbon-cloth GDE) or carbon-cloth on which CNT or CNF were secondarily grown and then Pt was deposited by impregnation in an amount similar to that of the commercial catalyst. Note that the tests reported in Fig. 3 were made at room temperature. At 80 °C, temperature of typical PEM fuel cell operations, the effects observed were similar, but the difference between the polarization curves are less evident.

Fig. 3

Comparison of the performances at r.t. of the following anode in a PEM fuel cell using a commercial Pt-Ru/E-Tek GDM on the cathode side: (a) E-TEK commercial Pt/C electrode (20 wt%), (b) Pt/CNF-CC and (c) Pt/CNT-CC. In the inset a simplified model of the different graphene sheets in CNF and CNT, and model of oxygenated species which may form on carbon

Up to about 40 mA/cm² of current density the performances of both Pt/CNT-CC and Pt/CNF-CC anode were superior to those of the commercial electrode, but above this threshold a significant worsening of the behavior of Pt/CNT-CC is observed, while the Pt/CNF-CC still maintains better than the commercial one. Two observations derive from these experiments. The first is that the overall performances of the PEM fuel cell could be significantly affected from the anode characteristics, while it is usually considered in literature that only the cathode side (O₂ reduction) would be important. The second is the role of specific nanostructure of carbon.

2.4 Carbon NanoFibres (CNF) and NanoTubes (CNT)

Carbon nanofibres have the typical Chinese hut nanostructure, e.g. the graphitic planes have an orientation of around 45° with respect to the growth direction of the nanofibres. The carbon nanotubes have instead the graphitic layers parallel with respect to the direction of growing. This is a simplified ideal nanostructure, because many defects are present, but even this simple model evidences that on CNF the Pt particles are probably located at the sites where these graphitic layers terminate on the surface, while on CNT should be located on a flat graphitic surface (basal plane). Defects or surface terminating graphitic layer are easier oxidized than a “flat” graphitic layer and thus these sites can give rise to various types of surface oxygen-containing sites (carbonyl anhydrides, lactone, quinone, carboxylic acids, phenol, cyclic peroxides) which play a role not only in the stabilization of metal particles, but also in modification of the hydrophilicity of carbon, a very critical factor to enhance the good contact with Nafion and thus guarantee an efficient proton transport. We may observe in Fig. 3, in fact, that Pt/CNT-CC shows a similar behavior to Pt-CNF-CC in the low current density region, while it is significantly worse at higher current density, where the mass transfer becomes the most determining factor.

How proof this concept? Characterization of these materials is difficult, because it is necessary to analyze the full GDE assembled to Nafion and it is not possible to quantify the effectiveness of the three phase contact. However, it is possible to investigate what happens when defects are specifically created in CNT. There are many possible treatments to induce defects in carbon nanotubes, but usually various other characteristic changes occur at the same time and it is thus difficult to single out the specific effect. Therefore we chose a method which induces only a local modification at the metal particle-carbon support interface. In fact, when the sample after impregnation is treated with microwaves, the local overheating of the metal nanoparticles induces a local modification of the carbon nanostructure around the metal particles. More in detail, the method involves H₂PtCl₆ reduction with ethylene glycol in the presence of KOH and microwaves (600 W, 20 s + 10), while the conventional impregnation method is based on the incipient wetness impregnation of a H₂PtCl₆ in C₂H₅OH/H₂O (1:1 v/v) solution [15]. The CNF-CC is instead prepared starting from commercial carbon cloth (CC) which is impregnated with Ni and then used to growth the carbon nanofibers (CNF) by chemical vapour deposition (ethane + H₂ in a 1:1 ratio at 630 °C) [16, 17]. Transmission electron microscope images (Fig. 4) seems to confirm the presence of a local deformation of the graphitic structure around metal particles which is not present in the samples prepared in the absence of microwave treatment [18]. It is difficult to have conclusive indications from these images, but the indication that microwaves induce a change in the interaction between metal particles and the carbon substrate is supported by cyclic voltammetry tests [18].

Fig. 4

TEM image of a Pt particle on a carbon nanotubes in an electrode prepared in the presence of microwave (see also text)

It is worth noting to analyze the effect on the behavior of the fuel cell using an anode prepared with the microwave treated CNF (Fig. 5). With respect to the Pt/CNT-CC sample prepared by the conventional method, an improvement of the performances particularly at higher current density is observed, i.e. an improvement of the efficiency in chemical to electrical energy conversion. The sample is significantly better than the commercial reference sample and better also than the Pt/CNF-CC sample reported in Fig. 3. These results clearly indicate that the 1D nanostructure of the carbon is less relevant than the specific metal nanoparticle–carbon interaction, which in turn depends on a large number of factors occurring during the preparation of these materials. This observation also explains the large discrepancies which may be found on the topic in literature.

Fig. 5

Comparison of the performances at r.t. of the following anode in a PEM fuel cell using a commercial Pt-Ru/E-Tek GDM on the cathode side: (a) E-TEK commercial Pt/C electrode (20 wt%), (b) Pt/CNT-CC and (c) Pt_microwave/CNT-CC. In the inset XRD pattern of E-Tek and Pt/CNT-CC samples

The inset in Fig. 5 compares the X-ray diffraction pattern of the Pt/CNT-CC sample (prepared by microwave) and that of the commercial reference electrode. Note that the amount of Pt in both cases is similar (0.5 mg Pt/g sample) [17]. The broader lines of Pt particles in the commercial reference sample (from E-Tek company) indicates smaller mean dimensions of them with respect to the Pt/CNT-CC sample. This is confirmed by TEM analysis which shows a mean diameter of about 3 nm for Pt/CNT-CC (see Fig. 4), while about 2 nm for the commercial reference sample. This indicates that the superior performances of the first sample are not due to a better dispersion of Pt and smaller particle diameter, but to the difference in the metal particle carbon substrate interaction, a factor usually neglected or little considered in the investigation of fuel cells. A second relevant factor is that the creation of O-containing sites on the surface of carbon changes the hydrophobic characteristics and probably allows either a better contact with Nafion or a better diffusion of protons, both favoring an improvement of the three phase contact and in turn enhanced properties. We may note, in fact, that data of Fig. 5 evidence either a reduction of slope in the Ohmic region, as also an improvement in the mass-transport controlled region.

2.5 The Role of Defects in CNT

A different and elegant method to demonstrate the role of defects in carbon nanotubes is to induce these defects by a mechanical treatment such as ball milling [18]. Figure 6 reports some TEM images of the carbon nanotubes before and after 150 min of ball milling and of the nature of Pt nanoparticles deposited on it. Ball milling induces a visible fracture and rupture of the carbon nanotubes, as well as their opening. Before the treatment the CNT are curved, twisted together and most ends are closed. The external diameter was ranging from 40 to 80 nm and the mean length was over 100 μm. After 150 min of ball milling, CNT are efficiently shortened and also opened; they result straight and not entangled. The mean length was 10 µm. Deposition of Pt over the untreated CNT result in samples in which decoration by Pt metal particles is not very homogeneous. The diameter of Pt nanoparticles ranges between 2 and 15 nm, and several of them shows an elongation along the axis parallel to the surface of CNT, as shown in Fig. 6. Deposition of Pt with the same modality (wet impregnation, as described above) leads instead to a more homogeneous distribution and size of the Pt particle particles. Pt particle size ranges between 1 and 4 nm, with few larger particles of 5.5–6.5 nm, mainly on the outer surface, almost always round-shaped. In the untreated sample, besides small (~2 nm) round-shaped particles, elongated, faceted particles 9–16 nm long and 7–11 nm large were often found. Few big round-shaped particles of up to 17 nm in diameter were also observed. It may be also noted that in the case the CNT treated by ball milling a preferential exposition of Pt(200) crystalline planes could be noted, even if HRTEM analysis alone could be not conclusive on these aspects. However, cyclic voltammetry tests [18], in confirmation of the previous indication, show that the Pt(200) crystalline plane is predominantly exposed on Pt/CNT(ball milled), while Pt(111) on Pt/CNT(untreated) and both differ from what observed for the commercial reference sample.

Fig. 6

Electron micrographs of carbon nanotubes before and after ball milling, and of the Pt metal particles deposited on these carbon nanotubes

In terms of performances in a PEM fuel cell this ball milling treatment has a significant effect (Fig. 7). Note that differently from results shown in Fig. 3, where the CNF or CNT were grown by CVD method over the CC substrate, in this case CNT were prepared separately. After impregnation of Pt, the sample was deposited over the CC substrate following a modality very close to that used to deposit the Pt/carbon black catalyst over the CC substrate. Mean diameter and other characteristics of the CNT is also different from those reported in Fig. 3 and thus the results are not directly comparable. The CNT are instead the same as those reported in Fig. 7, apart from the ball milling treatment, and thus the results are comparable.

Fig. 7

Comparison of the performances at r.t. of the following anode in a PEM fuel cell using a commercial Pt-Ru/E-Tek GDM on the cathode side: a E-TEK commercial Pt/C electrode (20 wt%), b Pt/CNT-CC (untreated) and c Pt/CNT-CC (ball milling). Tests were made feeding pure hydrogen (left graph) or hydrogen + 50 ppm CO (right graph). In the latter, the behavior of a E-Tek Pt-Ru (1:1) commercial sample is also reported. The amount of noble metal in all samples is 0.5 mg/cm²

Using pure H₂ feed (Fig. 7a) both Pt/CNT samples show a significant better behavior that the reference commercial sample, and particularly the sample in which the CNT were modified by ball milling before the addition of Pt. The maximum power density increases from about 6 mW/cm² for reference E-Tek anode to about 17 mW/cm² for Pt/CNT (untreated) and 28 mW/cm² for Pt/CNT (ball milled). Tests reported in Fig. 7a were made at room temperature, because the effect of diffusion limitations are enhanced at this lower temperature with respect to typical operation temperatures of about 75–80 °C. Therefore, it is possible to better evidence the influence of the nature of carbon on this critical aspect for the optimization of the performances of fuels cells. In fact, a higher power density in a lab scale FC corresponds, for a defined power, to a lower number of stacks necessary in a FC for final applications, and thus a lower cost. While the use of CNT instead of carbon black improves the maximum possible power density, the creation of defects in the CNT by ball milling nearly doubles the maximum power density. This is reasonably associated to a better proton transport over the functionalizated (defective) carbon substrate. Note that all the samples have a similar Pt loading of about 0.5 mg/cm² and as shown in Fig. 6 the mean dimensions of Pt particles is higher than in the reference commercial samples where the mean Pt particle diameter is of about 2 nm. Therefore, the effect observed in Fig. 7a cannot be associated to a higher Pt dispersion, the common interpretation used to explain different performances in fuel cell electrodes. In these tests, the macro substrate (carbon cloth) is also similar as well as the method of preparation of the GDE. In conclusion, the samples are similar in terms of macroscopic electrode characteristics, but different in terms of nanoscale environment around the Pt particles. This is the different critical factor which determines the overall performances and which is often not taken into consideration is studying fuel cells.

2.6 Nanostructure and Sensitivity to Poisoning by CO

Due to room temperature operations, the dominant factor determining the performances in samples of Fig. 7 is the three phase contact and mass transfer, but it is not possible to analyze whether the presence of defects on the carbon substrate could induce also a modification in the nature of the metal particle–carbon interaction. We should introduce as efficiency the ratio between ideal power density, e.g. the current density estimated for ideal open circuit voltage at the same value of current density of the maximum in power density, and maximum power density for a given sample. This efficiency increases slightly from about 25% to 31% in going from a reference E-Tek sample to the Pt/CNT (ball milled) sample. This could be explained by the fact that, in the experimental reaction conditions, the specific nature of the interaction between the metal particle and carbon substrate, which determines the efficiency of chemical to electrical energy conversion, has only a minor role in determining the overall performances.

However, this interaction plays a relevant role in determining the sensitivity of the anode to poisoning by CO. Usually, Pt–Ru bimetallic electrodes are utilized to reduce this sensitivity, because even few ppm of CO could drastically lower the maximum power density in the case of bare Pt. Typically, the concentration of CO should be maintained below 5–10 ppm in the H₂ feed. The role of Ru is essentially related to an assisted oxidation of CO strongly chemisorbed on Pt sites which block the Pt sites for H₂ dissociation. The promoted oxidation of CO could be in principle also associated to oxidized sites of the carbon substrate, even if there are no specific studies in literature on this aspect.

Reported Fig. 7b is the behavior of reference E-Tek Pt commercial samples and as sake of comparison also of commercial E-Tek bimetallic sample (Pt/Ru = 1:1) using a high CO concentration (50 ppm) in the hydrogen feed. With this high CO concentration the E-Tek Pt commercial sample shows an almost half maximum power density with respect to the case of a pure H₂ feed. Using an E-Tek commercial Pt:Ru sample, instead only a reduction of about 10–20% in the maximum power density is observed, which still remains low (about 6 mW/cm²). Using the Pt/CNT (untreated) sample the maximum power density is also about half of that observed for the sample feeding pure H₂. On the contrary, for Pt/CNT (ball milled) the reduction is about 10–15%, e.g. similar to that for Pt:Ru bimetallic sample. The creation of defects on the carbon substrate, and thus of oxidized sites such as those summarized in Fig. 3, has an analogous effect to that of Pt:Ru bimetallic particles, taking into account that the power density shown by the cell is about 5 times higher. This evidences the importance of a proper design of these aspects, e.g. of the understanding of the catalytic aspects related to the development of PEM fuel cells, and in turn of electrocatalysis for reaching a critical societal objective such as the development of more powerful and less costly devices for clean and mobile electrical energy production devices.

3 Carbon Dioxide Conversion to Fuels

Another society great challenge indicated in the cited DoE report “Catalysis for Energy” [11] is to find new solutions to solve the issue of carbon dioxide and particularly to develop novel approaches to convert carbon dioxide to liquid fuels (alcohols, hydrocarbons, etc.) [19, 20]. In fact, liquid fuels offer still a better way to store and transport energy with respect to H₂ or electrical energy. In addition, the use of H₂ as energy vector, for example, requires very large investments in the infrastructure of H₂ distribution.

However, the use of fossil fuels requests to find a solution to the issue of carbon dioxide recycle. Sequestration is an option which is under active investigation and first large scale applications are under construction. However, this solution is adopted essentially for large scale power plants and needs that suitable storage sites could be individuated near to the power plant. Only about 15–20% of the sites have these characteristics and thus still exists the need to find alternative solutions which ideally can convert carbon dioxide back to fuels.

There are many possible options to convert carbon dioxide to fuels [19, 20]:

• Reverse Water Gas Shift (RWGS) coupled to hydrogenation to form methanol or dimethyl ether (DME)

• Direct hydrogenation to methane or light alkanes or alkenes, or to oxygenates such as (DME), ethanol, and formic acid

• Reaction with hydrocarbons (dry, mixed or tri-reforming of methane)

• Biological conversion by algae and their use to form bio-diesel, -methane or -oils

• Photochemical (catalytic) direct conversion of CO₂ to formic acid, methanol or methane

• Photothermal, eventually assisted by a redox cycle, dissociation of CO₂ to CO and O₂, eventually coupled to analogous processes involving water to form syn gas which may then transformed to methanol, DME or Fischer–Tropsch hydrocarbons

• Photoelectrochemical (catalytic) conversion of CO₂, either to methane or methanol, or to longer carbon-chain hydrocarbons and alcohols.

The first routes are the most advanced, but the key issue is how to supply the H₂ and energy needed for the reactions. Biological route is based on an accelerated photosynthesis followed by biomass transformation, but there are several technical problems which still hinder this possibility. In a longer term perspective, the use of solar energy to convert CO₂ back to fuels is the preferable route, but there are many problems which have to be solved.

A first key question is the nature of the products which form. Let us take the example of direct photochemical (catalytic) reduction of CO₂ in water using doped-semiconductor particles. This should be the ideal solution, but the problem is that the result of the reaction is a high diluted aqueous solution of methanol, formic acid or oxalic acid. Similarly, the various attempts to perform a photoelectrochemical reduction of CO₂ in liquid solution resulted in the formation of CO and formic acid, or at high negative potentials of methane and low amounts of higher hydrocarbons.

The challenge is to develop a conceptually simple device which integrate from one side a photoanode which is able to dissociate water to protons and electrons which can be used on the electrocathode to reduce CO₂ to liquid fuels. We proposed a photoelectrocatalytic (PEC) device with these characteristics and which operates in gas phase to allow an easier collection of the products of reaction [21, 22]. A first relevant step to go in this direction is to demonstrate the possibility of electrocatalytic reduction of CO₂ to longer carbon-chain hydrocarbons or alcohols [23].

3.1 Electrocatalytic Reduction of CO₂

Several studies have been made on the electrochemical reduction of carbon dioxide over the last 20 years [24–27]. Despite the fact that electrochemical generation of chemical products is a mature technology and already practiced on enormous scales in the production of chlorine and aluminum and in water electrolysis, efforts on the electrochemical reduction of CO₂ have been generally not going beyond the laboratory scale and the use of metal electrodes.

Electrochemical utilization of CO₂ has been studied for many years. Two classes of reactions could be identified, depending whether the conversion of CO₂ is studied in aqueous or non-aqueous solutions. A primary reaction product obtained by electrolysis of aqueous CO₂ solutions is formic acid. The formic acid formation was reported for the first time in the year 1870 and competes with hydrogen evolution in the reduction of CO₂. A problem in the utilization of CO₂ in aqueous solution has to do with its low solubility in water at standard temperature and pressure. This means that at the surface of the electrode only very small amounts of CO₂ are available for the reaction to proceed. For aqueous solutions, in order to speed-up the reaction process, the pressure must be increased. The CO₂ concentration increases from 0.033 to 1.17 mol/L, when the pressure is increased to 60 bar.

Numerous studies have been carried out on the electrochemical reduction of CO₂ under high pressure on various electrodes in an aqueous electrolyte. A reference paper on this topic was published by Hara et al. [28]. Even if this paper is relatively old, no further relevant progresses have been made later. Productivity increases substantially at 30 bar of CO₂ with respect to that at 1 bar of CO₂. However, the total cathodic current barely increased with increasing CO₂ pressure. In terms of products, CO, H₂ and formic acid are mainly observed. Traces of methane and C2 hydrocarbons are observed in very low amounts (below few percent). In addition, the observed efficiencies were recorded for only limited period of times (few minutes), due to fast deactivation.

For non-aqueous electrochemical reduction of CO₂, solvents are employed which exhibit high solubility of CO₂. Dimethyl-formamide can contain up to twenty times more CO₂ than corresponding amounts of aqueous solutions, such as KHCO₃, potassium formate and water. Carbon dioxide in propylene carbonate is eight times more soluble, and in methanol five times more soluble. However, a basic problem by using organic solvents exists. Increased CO₂ solubility enables operation under large current densities, but low electrolytic conductivity leads to high ohmic losses. For this reason, methanol is often used to balance these two aspects.

Moreover a further typical problem, especially when copper cathodes are used, is that, in order to maximize hydrocarbon formation, high current densities are necessary, which show in these conditions a fast deactivation [29].

The best results in terms of hydrocarbon formation were reported by Hori et al. [29] using immobilized CuCl on Cu-mesh electrodes. A Faradic efficiency of about 70% for C₂H₄ was obtained, even if in the presence of a fast deactivation. In addition, it should be noted that corrosive media (high pressure, metal halides) are used. Metal halides are necessary to promote surface concentration of CO₂ at the electrode. To note also, that Cu is the only metal which gives appreciable amounts of C₂ hydrocarbons.

The main aspects which determine the performances are the following:

• Gas evolution in electrochemical cells. It has a strong impact on its performance due to three main reasons: (i) gas bubbles reduce electrolyte conductivity and increase ohmic resistance; (ii) bubbles that adhere to an electrode block the surface and reduce the area available for reaction; (iii) convection, local heat and mass transfer are increased as the bubbles rise. Kaneco et al. [30] have shown the importance in obtaining a high efficiency in the electrochemical conversion of CO₂ to methane:

  $$ {\text{CO}}_{ 2} + 8 {\text{H}}^{ + } + 8 {\text{e}}^{ - } \to {\text{CH}}_{ 4} + 2 {\text{H}}_{ 2} {\text{O}} $$

• pH and reaction temperature. They influence the performances, but Salimon and Kalaji [31] showed that in the 2.5–9.2 pH range and 0–80 °C temperature range the effect is related to the solubility of CO₂. Note, however, that on increasing the temperature decreases the relative selectivity to C1 products and increases that to C2 products.

• Porosity of electrodes. This is not only a way to increase the electrode surface area (and thus activity), but offers new possibilities to tune the performances. Nanotube composite electrodes were recently used to study the electrochemical reduction of CO₂ to methanol [32]. Current efficiencies up to 60.5% were reported. The starting electrode was Pt which, after mechanical polishing, was soaked in a slurry of Al₂O₃ and pretreated by electrolyzing in H₂SO₄ solution. This electrode was then placed in a mixed solution containing TiO₂ nanotubes on which RuO₂ had been preloaded. The solvent was left alone to air dry resulting in a nanotube (NT)–nanoparticle (NP) film applied to the platinum electrode surface.

The reaction mechanism on Cu electrodes involves as initial stage the formation of a carbon dioxide anion radical CO ^●−₂ which explains why metal halides are necessary in stabilizing this intermediate avoiding the formation of the formate ion which desorbs from the surface. The reduction of this intermediate, or alternatively of that formed by reaction of CO₂ with the carbon dioxide anion radical forms adsorbed carbon monoxide (CO) as key intermediate. Chemisorbed carbon monoxide can react with protons and electrons (in the presence hydroxide anions) to give water and chemisorbed methylene (:CH₂) which may either be further hydrogenated to CH₄ or react with another methylene intermediate following a Fischer–Tropsch like chain growth mechanism. Surface science studies on Cu(100) crystals suggest the possible alternative that two vicinal chemisorbed CO molecules react together with simultaneous breaking of respective C–O bonds to form a chemisorbed radical-anion C2 species by one electron transfer [25]. This species is then the precursor for C2 hydrocarbons.

Shibata et al. [33] very recently showed that alkanes and alkenes up to C6 hydrocarbons can be obtained in CO₂ electroreduction at room temperature and atmospheric pressure by application of a commercially available Cu-electrode, provided pretreatment by electropolishing is avoided. The product distribution follows the Schultz–Flory distribution and, depending on the applied potential, the chain growth probability (α) ranges from 0.23 to 0.31, values lower than those obtained in Fischer–Tropsch synthesis over heterogeneous Co- or Fe-based catalysts. When the same electrode material was pre-treated by electro-polishing it behaved like a pure Cu electrode: mainly methane and ethane were observed. It was suggested that the oxygen coverage of the electrodes is a function of the surface crystallinity. Surface oxygen was proposed to be an important factor in controlling the selectivity in the CO₂ activation.

3.2 Gas-phase Electrocatalytic Reduction of CO₂ over Nanostructured Carbon Electrodes

The previous short review on the state-of-the-art in electrochemical CO₂ reduction evidenced that electron conversion efficiencies of more than 50% can be obtained, even if at the expenses of very high overpotentials (~1.5 V). In only few studies methane or methanol have been observed as the primary products of the reduction; the overall yields are low in the absence of high overpotentials and the electrode surface is susceptible to poisoning.

However, all these studies have been carried out in solution using traditional-type electrodes. Using instead a gas-phase approach and nanocarbon based electrocatalysts, e.g. similar to those used in PEM fuel cells, it is possible to form long-chain hydrocarbons and alcohols up to C9–C10 and in particular isopropanol, as shown in a more detail later [23]. Productivities are still limited (20% energy efficiency), but the change of approach opens is a breakthrough for the development of new solutions for an effective recycle of CO₂ to fuels [20]. The electrocatalyst is based on metal particles inside nanopores of conductive carbon or carbon nanotubes, as shown in the conceptual model reported in Fig. 8. With this type of electrocatalysts it is possible to synthesize isopropanol, a 19 electron reaction, by reduction of CO₂.

Fig. 8

Schematic representation of the electrocatalytic conversion of CO₂ to isopropanol in carbon nanotubes based electrocatalysts

An example of the results which are obtained is shown in Fig. 9 which reports the formation of the main products obtained at 60 °C during the electrocatalytic reduction of CO₂ using a Fe-Co/CNT based GDM. In addition to these products, minor amounts of alkanes and aromatics up to C9–C10 products were observed. Therefore, it is possible to synthesize isopropanol by reduction of CO₂ even though Faradaic efficiency (limited in these tests from a still large amount of products which remain chemisorbed on the electrocatalyst) should be improved as well as stability.

Fig. 9

Product distribution at 60 °C in three consecutive runs of Fe-Co/CNT electrocatalysts for CO₂ reduction

3.3 Use of Solar Energy to Drive the Electrocatalytic Reduction of CO₂

The electrochemical reduction of CO₂ to fuels is sustainable only when the electrical energy and possibly the protons can derive from solar energy, e.g. when this step on the cathode side is coupled to a photoanodic catalyst able to split the water with solar energy to produce electrons and protons. In this way it is possible to develop “artificial energy trees” which can capture CO₂ by the atmosphere, for example by solid adsorbents which heated, by solar energy itself, can produce the stream of CO₂ which is used in a photoelectrocatalytic (PEC) device to produce fuels using the electrons and protons produced by solar water dissociation at the photoanode side. This vision is schematically illustrated in Fig. 10.

Fig. 10

Vision of the integration of a photoelectrocatalytic (PEC) device in artificial energy trees for the conversion of CO₂ back to fuels

In the photoelectrocatalytic (PEC) device [21–23] the general assembly is similar to a PEM fuel cell and the photoanode is separated from the electrocathode by a proton-conducting membrane (Nafion^® 117, for example). The PEC device is characterized by (i) a gas diffusion membrane (GDM) electrode for CO₂ reduction which operates in the gas phase, and not in liquid solution as those described in the previous section, avoiding thus the problems of solubility of CO₂ and (ii) an electrode based on metal particles supported over a conductive carbon support, e.g. a very different configuration with respect to the planar-type electrode used in photoelectrochemical devices described before.

This configuration has various advantages:

• There is no need to absorb CO₂ in a liquid and thus the problems related to CO₂ solubility are eliminated;

• The reaction products back diffusing from the electrode are collected directly from the gas phase, eliminating thus the problems and costs related to concentration of the liquid phase;

• The design scheme is close to that of PEM fuel cells and therefore all the developments in fuels cells can be used for a better design and to lower the costs also of the PEC device;

• The GDM electrode allows to use the concept of confinement, e.g. catalytic reaction in nanometric size environments—carbon nanotubes, for example—to promote the formation of long-chain alcohols and hydrocarbons with respect to C1–C2 products.

For the photocatalyst, a nanostructured TiO₂ thin film is used [34, 35]. It is characterized by an ordered array of TiO₂ nanotubes, in order to realize a nanostructure which has the following characteristics:

• Allows a fast collection and transport of the electrons generated during water splitting;

• Allows a fast transport of protons to the protonic membrane which is in contact with the photocatalyst on one side and the electrocatalyst on the other side;

• Has a high surface area (to increase activity) and reduced number of grain boundaries which favors hole-electron recombination;

• Optimize the light harvesting.

These materials are produced by anodic oxidation of Ti foils. There are still several issues to be solved, and in particular how to promote the activity in the visible light region, a key aspect to promote quantum efficiency by using the full solar light range instead that only of the UV part (around 4% of solar light).

There are thus still several issues to be solved before the implementation of the concept vision shown in Fig. 10 and may be observed that alternative solutions also exist to convert carbon dioxide to fuels, but this example evidences how the progresses on development of photocatalysts for water splitting on one side and of electrocatalysts on the other side, e.g. on two of the main new areas of development of catalysis, could be joint to find novel solutions for the societal challenge of reducing greenhouse gases.

4 Conclusions

Catalysis plays a key role for a sustainable society and is an enabling technology to promote sustainability, environment, energy, health and quality of life. We have limited here discussion on catalysis for energy and more specifically on some aspects related to the design of electro-catalysts for PEM fuel cells and photoelectrocatalysts for CO₂ reduction to fuels. The aim was not to provide a specific analysis of the state-of-the-art, even if some general issues in this area were discussed, but instead to provide a personal view of how catalysis plays a significant role in these areas to promote performances and open new outlooks as well. At the same time the investigation of these novel areas of catalysis with respect to the traditional fields opens new perspectives for catalysis itself. The role of defects in carbon nanotubes to promote the performances of supported metal particles is an aspect relevant not only for PEM fuels cells, but also for the development of novel catalysts. At the same time, a better understanding of these aspects would be a key factor to solve either some challenges in this area, for example for direct ethanol fuel cells, or to use fuel cells as novel reactors for electro-assisted chemical syntheses. A novel design of PEM fuel cell electrocatalysts would be necessary to develop reversible fuel cells, which can be either used in the normal mode or to convert electrical energy to chemical energy, e.g. as an alternative to batteries for storing energy. A still significant effort is necessary to improve the efficiency of chemical to electrical energy conversion minimizing energy loss as thermal energy. All these objectives require the development of improved electrocatalysts and a better understanding of the complex chemistry which occurs.

In conclusion, research on catalysis for energy and particularly in the area of novel photo- and electro-catalysts opens new perspectives either to solve societal challenges and to put catalysis even more at the core of future technologies for a sustainable society.