Includes bibliographical references. / Polymer electrolyte membrane (PEM) fuel cells are promising clean energy alternatives to non - sustainable fossil fuels. During fuel cell operation, external humidification of reactant gases is typically required in order to increase PEM conductivity for improved performance. However, the use of external humidification is costly and increases system complexity. Recently it has been found that by including a cathode microporous layer (MPL) in the membrane electrode assembly (MEA) , performance under dry conditions (no external humidification) can be significantly improved . However, the precise function of the MPL is not well understood and therefore there is little theoretical basis to optimisation of physical properties. One possible reason for this lack of understanding is the absence of a well-established fabrication, characterization and electrochemical testing methodology for MPL research. In particular, current research places little emphasis on the effect of MEA variance on the uncertainty in MPL electrochemical performance results. In this study a methodology is developed for fabricating, characterizing and testing MPLs to accurately correlate physical properties with in-situ electrochemical performance. MPLs of two significantly different thicknesses (approximately 20 and 50 μm in the thickest regions) were fabricated in - house using a doctor blade method and varying the ink composition. The pore structure and thickness of MPLs were characterized by mercury intrusion porosimetry (MIP), scanning electron microscopy (SEM) and X-ray micro computed tomography (μCT).

Bibliography: leaves 77-81. / The ultimate objective of the study was to investigate the performance of suitably prepared gold catalysts for low temperature water gas shift (LTS) - and more specifically to investigate the performance of these catalysts at temperatures below those traditionally utilised. As opposed to the research undertaken to-date on gold catalysed water gas shift, the reaction was performed, as far as possible, under conditions resembling those found industrially, viz. conditions of temperature, pressure, WHSV and dry feed gas composition typical of those applicable to industrial LTS units. Important to this study was therefore the generation of a comprehensive performance benchmark for the commercially available LTS catalyst, a copper-based material, against which to compare the performance of the gold deposited catalysts. The gold catalysts were prepared by deposition-precipitation, a preparation procedure found to yield nano-sized gold particles, suggested in literature as being crucial for activity, on the metal oxide support. Using this procedure, gold promoted low (copper) and high (iron oxide) temperature shift catalysts and commercial zinc oxide supports were prepared and tested. A wide spectrum of Au particle sizes were prepared on the supports, ranging from approximately 3-500 nm. The gold promoted LTS catalyst was found to exhibit slightly higher activity than the commercially available catalyst at temperatures below the conventional LTS range. It would appear as if gold promotion is advantageous to the industrial catalyst and could impact greatly on LTS catalyst life. Even though substantially less active than the commercial copper catalyst was found, the gold promoted commercial zinc oxide catalyst exhibited significantly higher activity than that previously quoted in literature and better performance than the iron oxide supported catalysts of this study. Consequently, the Au/ZnO system exhibits good potential for further developments in terms of water gas shift conversion.

Despite the benefits of fuel cell technology its advancement to being commercially functional is hindered by a number of crucial factors. These factors are often associated with the lack of appropriate materials or manufacturing routes that would enable the cost of electricity per kWh to compete with existing technology. Whilst most research efforts have been directed towards developing more active catalysts, the amount of catalyst required in the fuel cell can be further reduced by improving the platinum utilisation in the membrane electrode assembly. The platinum utilisation is a strong function of the catalyst layer preparation step and there remains significant scope for optimisation of this step. Whereas significant work has been conducted into the different components of the catalyst ink there is limited work and understanding on the influence of the mixing method of the catalyst ink. This study will focus on the influence of the mixing technique on the catalyst ink properties and on the final fuel cell performance. Specifically, the study will investigate the effect of the three different mixing techniques on (i) catalyst ink quality (ii) the physical properties of the resultant catalyst layer and (iii) the in-situ electrochemical performance of the membrane electrode assembly. A large set of characterisation techniques were chosen to effectively study the step wise processing of the catalyst layer, and fuel cell performance. The results presented here include a comparison of the various mixing techniques and a comprehensive 2 x 2 factorial design into the individual techniques. The results suggest that high energy mixing is required for effective distribution of catalyst layer components, an even catalyst layer topography and a highly functional ionomer network which consequently, enhances performance. The mixing energy referred to involves prolonged mixing time, enhanced mixing intensity or a combination of the two. During bead milling of catalyst inks, high intensity mixing seems to be beneficial however, prolonged mixing time appears to be detrimental to the ionomer film structure. During high shear stirring and ultrasonic homogenisation of catalyst inks, the ink mixture significantly heats up. It has been observed that at higher temperatures, Nafion elongates and the contact with catalyst agglomerates is enhanced. High shear stirring of catalyst inks seems to be most effective at high agitation rates. High mixing energies result in high shear forces and in addition, high mixing temperatures which appear to be beneficial to establishing an effective catalyst/Nafion interface, enhancing the three phase boundary observed during in-situ testing. Ultrasonic homogenisation seems to be more effective at prolonged sonication times. Due to the erosive nature of ultrasonic dispersion, sufficient time is required to establish a well dispersed and distributed catalyst ink. However, the nature of particle size distribution resulting from ultrasonication shows that inks are unstable and is not recommended for high throughput processing. Overall, fuel cell performance is not significantly affected by the mixing step however; mixing does have an observable impact on catalyst layer formulation. Generally, when optimizing membrane electrode assembly fabrication, mixing parameters should be carefully chosen. This goes without saying that parameters need to be effectively studied before foregoing catalyst ink processing.

Bibliography: pages 225-233. / SAP0-34 and mordenite, catalysts with quite different pore structures, are known to be suitable for methanol conversion to light olefins and propene oligomerization to a distillate type product, respectively. In this study, these catalysts were modified in various ways and the effect of these modifications on the activity and selectivity of the above two reactions investigated. SAP0-34, a small-pore silicoaluminophosphate molecular sieve, is highly selective in the formation of ethene and propene from methanol, but deactivates rapidly due to coke formation. This catalyst was synthesized and modified in various ways in an attempt to increase the catalyst lifetime and selectivity to ethene. Mild hydrothermal conditions encountered during deep-bed calcination of SAP0-34 were found to increase the catalyst lifetime. A number of further modifications were made to this deep-bed calcined material in an attempt to increase further the lifetime of this material. These modifications were : (i) Silanization - in order to neutralize the acidity on the external surface of the crystallites and hence reduce pore-mouth blockage by coke species on the crystallite external surface; (ii) Steaming - to investigate the effect of more severe hydrothermal conditions than those encountered under deep-bed-calcination conditions; (iii) Acid site poisoning by ammonia - in an attempt to reduce the rate of coke formation which takes place readily on strong acid sites; (iv) Boron impregnation - in order to reduce the intercrystalline void volume and thereby sterically hinder the formation of bulky coke molecules within the SAP0-34 pores; (v) Acid and caustic treatments - in order to reduce the catalyst acidity and thereby limit the rate of the coke formation reaction.

Gallium/H-ZSM-5 zeolite catalysts have been extensively researched for the aromatization of liquified petroleum gas (LPG). In 1989 BP and UOP collaborated to commission a pilot plant in Grangemouth, Scotland, for the aromatization of propane and butane. This plant, based on a technology called the Cyclar process, used continuous catalyst regeneration (CCR) and a gallium impregnated ZSM-5 zeolite catalyst to achieve yields of ca. 65% aromatics, mainly benzene, toluene and xylenes (BTX) [Guisnet and Gnep, 1992]. As a result of the Fischer-Tropsch process used by SASOL Ltd., South Africa is in an internationally unique position, in that it has a surplus of long chain linear alkenes with carbon numbers in the range C6-C8 . There could be large economic incentives to convert these alkenes into more valuable products, like alcohols or aromatics. Thus the purpose of this project was to determine if gallium/H-ZSM-5 catalysts, similar to those used in the Cyclar process, would be suitable for the aromatization of long chain alkenes. Three methods were investigated for the introduction of gallium into ZSM-5: (i) physical mixing with gallium oxide; (ii) impregnation by incipient wetness with gallium nitrate; (iii) ion-exchange with gallium nitrate. The catalysts were. tested with regard to their catalytic activity for the aromatization of 1-hexene and 1-octene.

Includes bibliographical references. / The interests in an alternative, sustainable power generation method has greatly increased in the past decade due to increases in greenhouse gases and its impact on global climate change. The use of fuel cells as a form of energy generation is extremely promising as it converts chemical potential energy directly to electrical energy, bypassing the Carnot cycle limitations. Various types of fuel cells have been developed, with the proton exchange membrane fuel cell (PEMFC) being most promising for mobile and small-scale stationary uses under transient conditions. The PEMFC uses hydrogen and oxygen to generate electrical energy. While oxygen can be obtained from air, hydrogen does not exist in its elemental form; hence a process train is required to refine fuels (such as fossil fuels and bio-fuels) into pure hydrogen. This has been successfully achieved by large-scale industrial plants. A typical fuel processing train consists of a steam reforming stage converting the fuel into syngas. This is followed by a water-gas shift (WGS) stage to convert carbon monoxide, which is a poison for the platinum catalysts within fuel cells, into carbon dioxide. If the CO concentration required is extremely low, a methanation or preferential oxidation stage can be used subsequent to the WGS stage. This study focuses on the water-gas shift stage of the fuel processing train. Industrial base metal WGS catalysts are not suitable for a miniaturized fuel processing train due to the catalysts being developed for continuous operations, as miniaturized fuel processing trains are expected to operate at transient conditions. A slow and controlled reduction process is also required prior to operation, as well as the pyrophoricity of industrial catalysts after reduction. These can pose safety issues with non-technical personnel in household applications (e.g. CHP). PGM-based catalysts have shown high activities for the water-gas shift reaction in literature, are not pyrophoric and do not require lengthy and sensitive reduction processes prior to operation. The objective of this study was to investigate and compare two base metal catalysts (high temperature (HT) shift Fe₃O₄/Cr₂O₃ and low temperature (LT) shift CuO/ZnO/Al₂O₃ catalyst) with a PGM based counterpart, as well as to investigate whether the catalysts are able to achieve a required 1 vol% CO via the water-gas shift reaction. For these investigations a synthetic feedstock was used, based on typical exit concentrations of a steam methane reformer.

The Low Temperature Fischer-Tropsch (LTFT) synthesis involves the catalytic hydrogenation of carbon monoxide with the aim to produce long-chained hydrocarbons. Commercial cobalt-based catalysts incorporate oxidic supports that are known to negatively affect the reducibility and hinder formation of the active phase. Consequently, reduction promoters such as Pt are introduced to facilitate the reduction of cobalt during catalyst pretreatment. However, synergistic and adverse effects of the promoter have been reported under reaction conditions including a higher site-time yield and higher selectivity towards hydrogenated products. The perspective on the operation of the Pt promoter is polarised between 'Hydrogen spillover', which is a so-called remote-control effect that could otherwise occur in the absence of Pt-Co contact, and 'ligand/electronic effects' that require direct Pt-Co coordination. The objective of this study was to explicate the operation of Pt and Au as promoters of the Co/Al2O3 catalyst by decoupling hydrogen spillover from effects that require direct promotercobalt coordination. The analysis was subdivided into the reduction process and the Fischer- Tropsch reaction, which are the two arenas in which the actions of these promoters have been claimed. The employment of model 'hybrid' catalysts, which are mechanical mixtures of the monometallic constituents of the promoted catalyst, presents a novel way to investigate the role of spillover hydrogen in the Pt-Co and Au-Co catalyst systems. Thus far, no systematic investigation of the hydrogen spillover phenomenon using these catalyst systems during both reduction and under commercially relevant LTFT conditions has been encountered in the published literature. Furthermore, this study serves to contribute to the limited body of literature on the role of Au as a potential promoter for the commercial cobalt-based catalyst.

Density functional theory (DFT) calculations have been performed to gain insight into the role of defects present on the surface of graphene and TiO₂ based supports on supported metal clusters. The clusters considered are a Pt₃₈ cluster and a bimetallic Pt₃₂Pd₆ alloy. The defects considered on graphene based supports are monovacancy defective graphene, OH and COOH functionalised graphene. The defects considered on TiO₂ based supports are a partially reduced TiO₂(110) surface with a surface oxygen bridge vacancy and hydroxylated TiO₂(110) surface with surface OH groups. The defect free graphene and TiO₂ surfaces were also considered. For both the Pt₃₈ and Pt₃₂Pd₆ cluster, and on both defect containing graphene and TiO₂ (except on hydroxylated TiO₂(110) surface) the binding of the clusters is enhanced relative to binding on the defect free supports. Enhanced binding at the defects imply that the clusters are bound strongly to the support and thus less likely to detach from the support material relative to binding on the defect free supports. Therefore, the defects may improve the durability of the catalyst by limiting particle detachment. The electronic properties of the cluster are modified differently depending on the identity of the defect present on the support. On the graphene based supports, OH functionalisation is expected to result in weaker binding energy of adsorbate molecules, whereas COOH functionalisation is expected to result in stronger binding energy of adsorbates for the supported Pt₃₈ cluster. This is due to different shifts in d-band centre of the facets on the cluster supported on these supports. Therefore, it can be expected that the Pt₃₈ cluster supported on OH functionalised graphene will be more tolerant to poison molecules. This is due to reduced binding strength of adsorbates on OH functionalised graphene compared to adsorption on COOH functionalised graphene. For the Pt₃₂Pd₆ cluster the OH and COOH functional groups do not appreciably modify the d-band centre of the facets available to reactants, and thus is expected not to significantly modify the binding strength of adsorbate molecules relative to binding on the free unsupported Pt₃₂Pd₆ cluster. The binding energy of adsorbate molecules on the Pt₃₈ cluster supported on defect containing TiO₂ is expected to be stronger than on the Pt₃₈ cluster supported on defective graphene based supports, due to higher extent of upward shift of the d-band centre of the exposed facets. The enhanced binding energy of adsorbates on the Pt₃₈ cluster supported on TiO₂ supports may be detrimental to catalyst durability and activity. This can be due to strong binding of poison molecules and reaction intermediates which maybe too strongly bound on the surface such that they cannot participate in further reaction steps. Overall it might turn out that the functionalised graphene based supports are better support materials over the TiO₂ based materials for particular reactions. The Nb-doped partially reduced TiO₂(110) surface attaches the Pt₃₂Pd₆ cluster strongly to the support compared to the functionalised graphene supports. Furthermore, the binding energy of adsorbate molecules is expected to be weaker on the Pt₃₂Pd₆ cluster supported on the Nbdoped partially reduced TiO₂(110) surface compared to the functionalised graphene supports. This might be beneficial as poison molecules may be weakly bound to the cluster resulting in high resistance to poisoning which can also have a positive effect on catalyst activity. In addition to enhancing binding of the cluster to the support and affecting the binding energy of adsorbates on the supported clusters, some of the defects can also alter the ordering pattern of Pd and Pt atoms within the Pt₃₂Pd₆ cluster. OH functionalised graphene and Nbdoped partially reduced TiO₂(110) surface result in segregation of Pd towards the clustersupport interface, thereby exposing more Pt atoms at the surface facets of the cluster. The modified arrangement of Pt and Pd atoms may result in modification of the reactivity of the Pt₃₂Pd₆ cluster. The results of this study indicate that the defects can play a vital role in determining the activity and durability of the catalyst. By having the right combination of defects on the support material, the durability and catalytic activity of the catalyst can be fine-tuned simultaneously. This can lead to better design of catalysts.

The Fischer-Tropsch synthesis (FTS) process, better known for its ability to produce synthetic fuel via the hydrogenation of CO, has shown potential to produce valuable chemicals when ammonia is added to the feed. In this work certain aspects of the pathway to the formation of N-containing compounds that form when NH₃ is added during FTS, using mostly iron based catalysts is investigated. In addition, the effect this has on the FTS reaction itself is evaluated. To achieve this goal, both theoretical and experimental techniques are used in this study. The CO adsorption and dissociation reactions are assumed to be important elementary reactions for many proposed FTS pathways. In the theoretical part of this thesis, spin-polarized periodic density functional theory (DFT) calculations are employed to study aspects of the initial stage of the pathway on a model Fe(100) surface. Considering the formation of N-containing hydro- carbons, one would assume that NH₃ initially adsorbs and dissociates on the catalyst surface, which could take place in the presence of CO. The surface chemistry of these adsorbates is well studied both experimentally and theoretically, but their co-existence has not yet been evaluated on model Fe surfaces. Initially a platform is generated by calculating the individual potential energy surfaces (PES) for the decomposition of CO and NH₃ on Fe(100) at a coverage of ϴ = 0.25 ML. These calculations provided the basis for comparing the adsorption and dissoci- ation profiles of CO and NH₃ on the Fe(100) surface via the use of the same computational methodology, and importantly making use of the same exchange correlation functional (RPBE) for both adsorbates. Furthermore, it was desired to evaluate the kinetics and thermodynamics of the NH₃ decomposition on the Fe(100) surface at relevant temperatures and pressures (by combining the DFT results with statistical thermodynamics) to better understand the role of NHₓ surface species involved in the pathway to the formation of the N-containing compounds on a model catalyst surface. The DFT results that are reported for the individual decomposi- tion PES for CO and NH₃ were generally found to be in close agreement with what has been reported in previous DFT studies and deduced experimentally for the relevant adsorption and decomposition pathways. The resulting Gibbs free energies for the PES suggests that NH₂ may be kinetically trapped on the Fe(100) surface at a coverage of ϴ = 0.25 ML and the reaction conditions (T = 523 K and p*NH₃ = 0.2 bar) where NH₃ is co-fed with synthesis gas during FTS. The individual adsorptions of CO and NHₓ (with x = 3, 2, 1, 0) were compared to their coadsorbed states, by calculating the heat of mixing (ΔEmix) and the activation barriers (Eₐ) for CO dissociation in the presence and absence of the NHₓ surface species on the Fe(100) sur- face. Similar to the individual adsorption of NH₃, the 0 K regime inherent to DFT calculations is bridged by calculating the Gibbs free energy of mixing for CO + NH₃ on Fe(100) at higher temperatures. Both repulsive and attractive interaction energies were calculated for the various coadsorbed states (CO + NHₓ on Fe(100)) and similarly some configurations resulted in an energetically favored or unfavored heat of mixing. The activation barrier for CO dissociation was lowered when coadsorbed with NH₃ and NH₂, and raised when coadsorbed with NH and N. With all the coadsorbed structures the CO dissociation reaction became more endothermic. Previous experimental studies have shown a concomitant reduction in oxygenate selectivity with an increase in the selectivity for N-containing compounds, when NH₃ is added during FTS. It is well-known that oxygenates undergo secondary reactions when using iron-based catalysts in FTS. In addition, the catalyst used in aforementioned studies (precipitated Fe/K) are active for the amination reactions of oxygenates. It is therefore hypothesized that some oxygenates pro- duced via the primary FTS pathway are converted to N-containing compounds via a secondary reaction. The experimental part of this thesis is therefore aimed at testing this hypothesis. A base case study included a comparison between a Fe-catalyzed slurry phase FTS reaction and a FTS reaction with all parameters remaining unchanged, except for the addition of 1 vol % NH₃ to the syngas (CO + H₂) feed. The activity (CO and H₂ conversion) data collected did not reveal any appreciable loss in the rate of the FTS reaction when 1 vol % NH₃ was added and steady state was reached (, that is after 48 hours time on stream (TOS)). A slower carburization period was however observed when comparing the CO conversion during the first 24 hours TOS, and further supported by the slow increase in CO₂ selectivity during the same period. The use of two-dimensional gas chromatography (GC × GC-TOF/FID) allowed for the discovery of a formation of a range of secondary and tertiary amines, not reported in previous studies. The expected loss in oxygenate selectivity was observed and further probed by co-feeding 1-octanol with the feed (CO + 2H₂ + 1 vol % NH₃) via a saturator. These results clearly indicated a significant loss in oxygenate formation as a result of secondary conversion to N-containing compounds. Questions regarding the stability of aliphatic nitriles prompted the co-feeding of nonanitrile under similar conditions. The results obtained after co-feeding nonanitrile, sug- gests that nonanitrile is readily converted to secondary and tertiary amines and that the ratios of aliphatic alcohols and nitriles are close to equilibrium. The use of CO₂ as carbon source, the investigation of the product spectrum at higher space velocities and the use of Rh-based catalysts, when NH₃ is added during FTS were included in shorter studies. The combination of these results, adds to the knowledge pool for the case where NH₃ is present in the FTS regime, as a poison or reactant. Additional information regarding the path to the formation of N-containing compounds was obtained via the detailed analysis of the product spectra with two-dimensional gas chromatography and the subsequent co-feeding reactions. The results ob- tained via co-feeding reactions, can be used to devise strategies to increase the selectivity of the desired N-containing compounds.

The preferential oxidation of CO (CO-PROX) has been identified as one route of further reducing the trace amounts of CO (approx. 0.5 - 1 vol%) in the H2-rich reformate gas after the high- and low-temperature water-gas shift reactions. CO-PROX makes use of air to preferentially oxidise CO to CO₂, reducing the CO content to below 10 ppm while minimising the loss of H₂ to H₂O. In this study, a Co₃O₄/γ-Al₂O₃ model catalyst was investigated as a cheaper alternative to the widely used noble metal-based ones. The CO oxidation reaction in the absence of hydrogen has been reported to be crystallite size-dependent when using Co₃O₄ as the catalyst. However, studies looking at the effect of crystallite size during the CO-PROX reaction are very few. Metal-support interactions also play a significant role on the catalyst's performance. Strong metal-support interactions (SMSI) in Co₃O₄/Al2o₃ catalysts give rise to irreducible cobalt aluminate-like species. Under CO oxidation and CO-PROX reaction conditions, such strong interactions in a similar catalyst can have a negative effect on the performance of Co₃O₄ but can keep its chemical phase intact i.e., help prevent the reduction of the Co₃O₄ phase. The catalysts used to investigate these two effects (i.e., crystallite size and metal-support interactions) were synthesised using the reverse micelle technique from which nanoparticles with a narrow size distribution were obtained. Certain properties of the microemulsions prepared were altered to obtain five catalysts with varying Co₃O₄ crystallite sizes averaging between 3.0 and 15.0 nm. Four other catalysts with different metal-support interactions were also synthesised by altering the method for contacting the support with the cobalt precursor. The crystallite size of Co₃O₄ in these four catalysts was kept in the 3.0 - 5.0 nm size range.