This thesis investigates the development of a heuristic based methodology for designing measurement networks with application to the precise accounting of metal flows in mineral beneficiation operations. The term ‘measurement network’ is used to refer to the ‘system of sampling and weight measurement equipment’ from which process measurements are routinely collected. Metal accounting is defined as the estimation of saleable metal in the mine and subsequent process streams over a defined time period. One of the greatest challenges facing metal accounting is ‘uncertainty’ that is caused by random errors, and sometimes gross errors, that obtain in process measurements. While gross errors can be eliminated through correct measurement practices, random errors are an inherent property of measured data and they can only be minimised. Two types of rules for designing measurement networks were considered. The first type of rules referred to as ‘expert heuristics’ consists of (i) Code of Practice Guidelines from the AMIRA P754 Code, and (ii) prevailing accounting practices from the mineral and metallurgical processing industry which were obtained through a questionnaire survey campaign. It was hypothesised that experts in the industry design measurement networks using rules or guidelines that ensure requisite quality in metal accounting. The second set of rules was derived from the symbolic manipulation of the general steady-state linear data reconciliation solution as well as from an intensive numerical study on the variance reduction response of measurements after data reconciliation conducted in this study. These were referred to as ‘mathematical heuristics’ and are based on the general principle of variance reduction through data reconciliation. It was hypothesised that data reconciliation can be used to target variance reduction for selected measurements by exploiting characteristics of entire measurement networks as well as individual measurement characteristics. It was found that experts in the industry minimise metal accounting variance by sampling and weighing key streams with high precision. Terminal streams in general, and Feed and Product streams in particular, were identified as key to metal accounting. The emphasis on the measurement and usage of terminal streams was found to be consistent with the widespread use of the Check In-Check Out method of accounting in the minerals beneficiation industry. Of concern however is the low usage of Tailings stream measurements in metal accounting despite the universal employment of the Check In-Check Out system. It thus appeared that expert design philosophy advocates the precise measurement and utilisation of terminal streams in general to define corporate metal accounts through the Check In-Check Out method of accounting while internal measurements appear to be reserved for the evaluation of internal unit operations. Mathematical heuristics developed in this study illustrate the benefits of precise measurement of internal streams so that terminal streams can experience maximum variance reduction after data reconciliation. However, the design philosophy of concentrating resources on internal streams is contrary to common expert practice where emphasis is on precise measurement of the actual input and output streams of the process. In this case, the metal accounting system will behave as a single node and will not benefit significantly from data reconciliation i.e. a Check In-Check Out type accounting philosophy will be suitable. This thesis concludes by proposing a heuristic design strategy for constructing metal accounting measurement networks depending on quality and governance based imperatives. For accuracy requirements, the use of applicable metrology standards and the Check In-Check Out method of accounting were deemed sufficient for purpose. Adoption of additional tools such as the AMIRA P754 Code of Practice was suggested as necessary in order to achieve governance based requirements that include transparency and credibility of the metal accounting process. Data reconciliation was suggested as a ‘no-cost’ means for primarily improving precision of measured data beyond the capabilities of existing hardware and as a check of the integrity of measured data. Measurement network design was proposed as an additional tool for improving accounting precision by maximising variance reduction of selected measurements after data reconciliation. The use of reconciled data is however conditional on the acceptance of ‘adjusted’ data as valid or legitimate input to metal accounting reporting.

Since the birth of the automobile in 1886, its popularity amongst people has risen dramatically owing to the freedom, comfort, speed, safety and unique designs offered by this mode of transport. 2014 saw approximately 71.15 million units of new vehicle sales globally, showing that private car usage is still on the rise. Rapid degradation of the environment and slumped economic growth can be attributed to the automobilecentric transport system. Raised environmental and social awareness has driven campaigns to promote greener modes of transport instead, such as public and nonmotorised transportation. This has seen the introduction of BRT systems in South African cities however; fully integrated transport systems are yet to be achieved. Thus there is a dire need for a design support tool that is adequately capable of processing built environment characteristics in the development of a BRT feeder network that is fully appreciative of the influence of NMT and the urban fabric, and is thus appropriate to the needs of the community it is trying to serve. This study comprised the application of two Spatial Multi-Criteria based methodologies in which a list of built environment characteristics and public transport demand formed the inputs for the analyses. The analysis produced a composite suitability map for each approach, in which each pixel represented the appropriateness of having a BRT feeder route located in that respective pixel. Routes between O-D pairs identified were solved by carrying out a least cost path assessment based on the mean impedance values along the existing road network. The routes developed were compared to the MyCiTi feeder bus routes using Key Performance Indicators established in this research to determine whether this study was successful in producing an enhanced BRT feeder route planning tool. Apart from one route, the set of feeder routes developed for each approach were exactly the same with the second method producing lower average impedance values per kilometre thus it was deemed stronger. When compared to the MyCiTi feeder routes, similar operational efficiencies were achieved with respect to average travel time, coverage and directness. However; the study methodologies provided a greater level of NMT planning inclusion and consideration of environmental factors. Furthermore; it achieved this in a systematic and transparent manner, providing immensely powerful benefits for transportation planners in the public sector. This study was successful in demonstrating that SMCA combined with the Network Analyst tool in ArcGIS has the ability to enhance the quality and appropriateness of BRT feeder routes, whilst achieving acceptable operational efficiencies. The results could further be improved by incorporating more data on local NMT trends and behaviour. Furthermore; this tool can be applied to solving pedestrian, bicycle and other public transport routing problems.

Proton exchange membrane fuel cell (PEMFC) has been reported as clean and efficient energy technology from conversion of H2. However, one of the main challenges remains the storage and transport of hydrogen. The promising alternative is to produce H2 on site by a reformer using a H2-dense liquid as a fuel, a technology known as fuel processing. Methanol is an attractive source of H2 compared to other fuels as it presents several advantages, i.e. it is obtained sulphur-free, has a high H to C ratio and therefore produces a H2-rich reformate, can be reformed at low temperatures (200 - 300°C) and is a liquid at ambient conditions so that it can be easily handled. Typically, Cu-based catalysts are used for steam reforming of methanol due to their high activity (i.e. H2 production) and high selectivity towards CO2. As CO poisons anodic catalyst of PEMFC, high selectivity towards CO2 is crucial so as to eliminate or at least minimize CO removal load downstream a fuel processor. However, Cubased catalysts are thermally unstable and suffer deactivation due to sintering at high temperatures (> 250°C). Moreover, Cu-based catalysts are pyrophoric and therefore difficult to handle. Recent studies show that PdZn catalysts are very promising as they exhibit comparable activity and selectivity to Cu-based ones. Furthermore, PdZn catalysts are thermally stable in the typically methanol steam reforming temperature range (200 - 300°C). Most literature attributes high CO2 selectivity of PdZn catalysts to formation of PdZn alloy. It is generally agreed that PdZn alloy is formed when PdZn catalysts are reduced in H2 at high temperatures (> 250°C). In this work, a Pd/ZnO catalyst aimed at 2.5 wt% Pd was successfully prepared via incipient wetness impregnation and the duplicate preparation of the catalyst was successful. Both impregnation catalysts were confirmed by ICP-OES to contain similar weight Pd loadings i.e. 2.8 and 2.7 wt%, respectively. The actual Pd loading (ICP-OES) was slightly higher than the target loading (2.5 wt%) due to Pd content of Pd salt underestimated during catalyst preparation. Furthermore, crystallite size distribution, i.e. PdO crystallites on ZnO support, was similar (i.e. 6.7 ± 2.4 nm and 6.3 ± 1.9 nm) for both impregnation catalysts. The TPR analysis of the catalyst showed two peaks, i.e. a narrow peak at 87°C and a broader peak starting at approximately 260°C with a maximum at 348°C and ending around 385°C. The peak which occurred at the lower temperature was due to reduction of PdO to metallic Pd. The H2-TPR analysis of pure ZnO shows that ZnO (in the absence of Pd) did not reduce below 600°C. Therefore the peak which occurred at the higher temperature in the case of the impregnation catalyst was due to reduction of ZnO and this was facilitated by H2 spillover from the metallic Pd. To confirm the difference in selectivity towards CO2 between the ‘only PdO reduced’ and the ‘PdO and ZnO’ reduced catalysts, the impregnation catalyst was reduced at different temperurates (i.e. 120, 180 and 450°C) prior to catalyst performance tests. However, selectivity towards CO2 remained > 99% for all conditions. To not influence the reduction by in-situ reduction due to H2 produced by the methanol steam reforming reaction, reforming was carried out at low temperatures. Consequently, CH3OH conversion was low (< 30%). Since the feed molar steam to carbon ratio was 1.1, sligthly higher than the stoichiometric ratio, the low CH3OH conversions (< 30%) resulted in excess steam (compared to CO and CO2) in the reactor and this condition drove the water-gas shift reaction to high or equilibrium selectivity towards CO2 ( ~ 99%). The co-precipitation preparation method was not successful.i.e. the co-precipitation catalyst was aimed at 3 wt% but was confirmed by ICP-OES to be only 1 wt%. A significant fraction of Pd was lost as it did not precipitate during catalyst preparation. The uncontrollable precipitation of Pd makes the method irreproducible. Hence less focus was paid on the coprecipitation catalyst.

Biodiesel can be used in diesel engines without significant modification of the engine prior to use because it has properties similar to those of petroleum diesel. Biodiesel, however, exhibits lower stability compared to petroleum diesel. Small differences in fuel properties such as component concentration or total acidity can lead to the formation of deposits which can reduce engine performance and increase maintenance requirements and costs. Thermo‐oxidative stressing was performed in two reactor systems in this study. For comparative purposes both sets of experiments were performed at 140oC. The systems used were a quartz crystal microbalance (QCM) in which oxygen was limited and open glass flasks under flowing air (unlimited oxygen). To simplify analysis, diesel model compound systems were used in which full boiling range diesel was replaced with single compounds representing the classes of compounds found in petroleum diesel. The model compounds were n‐hexadecane, tetralin and decalin. Fuel analysis was performed using gas chromatography (GC) with mass spectrometric (MS) and flame ionisation (FID) detection. Further analytical methods included Fourier transform infrared (FTIR) and ultraviolet‐visible (UV‐Vis) spectroscopy as well as electrospray ionisation‐mass spectrometry (ESI‐MS). This study represents the first application of QCM methodology to systems that contain fatty acid methyl esters (FAMEs). FAMEs with different degrees of unsaturation were investigated. The quantities of deposits formed were as follows: methyl linolenate (40.1 μg/cm2) and methyl linoleate (19.2 μg/cm2) when these were blended with diesel model compounds in a 20:80 ratio. These values are significantly higher than those typically reported for middle distillates. Increased polyunsaturation of the FAME led to increased deposit formation. Spectroscopic investigations revealed very small, if any, changes to bulk fuel composition during oxidation in a QCM. GC and ESI‐MS analysis demonstrated the formation of oxygenates but in small concentrations. Nonetheless these were sufficient to initiate the formation of deposits. More severe conditions used in open flask experiments led to greater quantities of, more types of, and more highly oxygenated products as seen by GC and, with respect to the latter, especially by ESI‐MS. FTIR spectroscopy revealed the inclusion of oxygen‐containing functional groups. Increasing deposit precursor molecular mass and polarity led to phase separation in these experiments. Colour and UV‐Vis changes were suggestive of the formation of conjugated systems. Examples of these species were identified by GC‐MS. The QCM was also useful in the exploration of fuel solvency. This study represents, as far as the author is aware, the first application of a QCM in this area. Fuels, with small quantities of biodiesel added, formed greater amounts of deposits than petroleum diesel model compounds or 100% biodiesel. Furthermore these systems consumed oxygen slightly faster. This points to a complex interaction between reactivity (faster oxygen uptake) and fuel polarity. This study has demonstrated the versatility of the QCM methodology for application to fuels beyond petroleum middle distillates. FAME molecules were observed to be highly oxygenated and formed highly oxygenated dimers and a small quantity of trimers/tetramers during oxidation. It was observed that under severe oxidative conditions (flowing oxygen) the types of high molecular species formed differed from those formed during constrained oxygen stressing in a QCM. Existing mechanisms for fuel oxidation were validated and in some cases extended. A possible soluble macromolecular oxidatively reactive species (SMORS), 1,4‐naphthoquinone, was identified from the oxidation of tetralin by GC‐MS and ESI‐MS. Previously unobserved species such as 1,2‐dicarboxylic acids, derived from tetralin and decalin, were reported and used to extend existing mechanistic schemes. The formation of higher molecular weight species that result from the interaction of alkenes (derived) from paraffin cracking with radicals such as the 1‐tetralyl radical are also explained.

Phosphate is an important, non-substitutable nutrient for all life forms and is essential in ensuring universal food security. In the past, waste water treatment works (WWTW) typically installed effluent polishing technologies to eliminate phosphate and lower concentrations of both nitrogen and phosphate to below regulatory levels. More recently, there has been a global shift towards treating waste water as a "water-carried waste", presenting opportunities for both nutrient and energy recovery. South Africa is yet to embrace this shift, as it battles to provide universal access to basic sanitation needs and is faced with massive infrastructure maintenance and upgrading backlogs in the sanitation sector. Mature phosphate recovery technologies that produce high quality struvite for use in food production do exist. However, there is little evidence to indicate that similar phosphate recovery techniques would be economically viable or socially accepted in South Africa. Therefore, this dissertation sets out to investigate the viability of a simpler and cheaper phosphate recovery technology. The dissertation addresses the hypothesis that the South African market is better suited for low quality struvite for use in secondary (non-food) markets and that this would be cheaper than both traditional chemical precipiation (phosphate elimination) methods and highquality struvite production. This dissertation attempts to answer two key questions derived from this hypothesis by means of two separate methodologies. A qualitative methodology explores socio-technical issues to understand the potential of sewage-recovered struvite in the South African markets. This sets out to explore: What space there is for fertilizer production (such as struvite) from human waste in the South African markets? The second research component uses standard engineeirng economic methods, to investigate the potential for centralized recovery of nutrients through the conceptual design and a techno-economic pre-feasibility assessment of two phosphate recovery options at the largest WWIW in the Western Cape. These options are contrasted with a more traditional chemical precipitation process. Expert interviews revealed that although most stakeholders recognized the importance of phosphate recycling in tackling food security and achieving sustainable water and nutrient cycles. However, it is believed by the industry experts that the South African organic market and its consumer may not be ready for fertilizers produced from human waste to be used in food production. Better acceptability could be experienced within the inorganic fertilizer production market, regardless of source, if struvite is proven to be safe and a purification process is identified. More feasible markets could lie within ornamental plant fertilization, commercial fertilizer production and fertilizer use within closed community gardens. Therefore, there is potentially a larger market for lower grade struvite. The techno-economic assessment reveals that the digestate stream at the 200 ML/ day WWfW has the potential to produce -470 kg/d of struvite fertilizer, which only recovers 1-3% of the plant costs in 20 years. Revenue is subject to prices on the South African fertilizer market; and as it stands, the selling price of struvite for both low- and high-grade treatment is significantly lower than the cost of recovering them. Net present costs of R76,2-, R25,4- and RSl ,2 million were calculated for retrofit projects for high-grade struvite, low-grade struvite and chemical precipitation respectively. From this perspective and as hypothesized, low-grade struvite production is the most attractive process option. The establishment costs for chemical precipitation showed to be the most economical, with a CAPEX of R2,5 million, 10 and 30 times less than that of low-grade and high struvite production; which is within the Cape Flats planned budget for a nutrient treatment facility. Although this is the most common treatment technique in South Africa, it is the least sustainable process option resulting in the formation of a toxic by-product that must be disposed of in offsite landfills - an important factor that cannot be overlooked. The high capital costs and unprofitable operations of struvite, production are attributed to the high flowrate to phosphate loading ratio experienced at the CFWWTW. Other WWIWs with a more concentrated waste water profile, may yield better economics. However, unless the value of struvite increases, the cost of running the additional plant will not be recovered. Yet again, production does fall within the cost bracket for struvite production at R8,90/kgP removed. Hence investment may be justified from this angle. If a WWIW is to reduce effluent phosphate loading to within regulated standards, low-grade struvite production has thus been shown to be the most ecologically and economically sustainable option from a life-cycle-costs perspective. From a social stand-point, the experts interviewed believe that the South African food market could resistance fertilizers derived from human waste, hence potentially ruling in favour of low-grade struvite for use in secondary nonfood markets. Although it is a simple process, it is not cheaJ:>; the capital investment is 10 times that of South Africa's more familiar chemical precipitation route. Municipalities will have to consider the lower operating costs, as well as the environmental benefit of producing a useful phosphate fertilizer over the immediate capital costs.

Dense medium separation (DMS) is a method often used to upgrade base metal sulfide (BMS) ores before their main processing stage, with varying results achieved for different ore types. The process makes use of the density differences between the BMS minerals and the lower density silicate/carbonate gangue minerals, using a separating medium of density between the two ore components. The separation is accelerated using a dense medium cyclone (DMC) to form two products: overflow (tailings) and underflow (concentrate). The purpose of DMS is to reject large quantities of gangue upfront, resulting in reduced time, energy and costs associated with processes such as milling and flotation. Preconcentration of ores using physical methods such as DMS is becoming an important consideration as lower grade ores are mined, to increase the feasibility of mining such ores. Two nickel sulfide deposits were chosen as case studies in order to understand differences in DMS efficiency for different ores. The first is the Main Mineralised Zone (MMZ) of the Nkomati Nickel deposit in Mpumalanga, South Africa, which is part of the Uitkomst Complex. The Phoenix deposit is also considered, and forms part of the Tati greenstone belt in eastern Botswana. Both deposits are magmatic Cu-Ni-PGE (platinum group element) deposits with similar sulfide mineralogy and pentlandite as the main nickel host. A process mineralogy approach was used to evaluate samples of both ores, describing the differences in the mineralogical properties within the overflow and underflow of each ore in order to understand the extent to which individual properties affect the separation. A bulk sample of each ore type was subjected to DMS using a pilot plant setup, and the overflow and underflow products further classified into a series of density classes using sinkfloat analysis. These density classes were mineralogically characterised by petrography, quantitative X-ray diffraction, QEMSCAN and electron probe microanalysis, to provide information on differences in bulk mineralogy, mineral textures, mineral chemistry and particle properties between the samples. The nickel contents of both ores were upgraded using DMS and the Nkomati ore experienced a more efficient separation than the Phoenix ore, which is contrary to previous tests on MMZ ore of similar grade. Both the Nkomati and Phoenix ores consisted of primary magmatic minerals such as pyroxene and plagioclase, as well as a variety of secondary silicates formed by alteration of the original mineral assemblages, e.g. amphibole, chlorite and talc. Three sulfide textures were observed: disseminated / bleb-textured, net-textured and massive. Both ores show more than one texture, with the Nkomati ore displaying all three textures and the Phoenix ore mostly consisting of disseminated sulfides with minor massive sulfides. Pentlandite in the disseminated zones dominantly occurs as fine exsolution lamellae in pyrrhotite, with granular pentlandite mostly located within massive sulfide regions. Apart from overall particle density, sulfide texture is the main controlling factor affecting the individual particle recovery by DMS, with massive and net-textured sulfides having larger grain sizes and therefore higher liberation than disseminated sulfides. In addition to the DMS concentration of sulfide minerals, primary and secondary silicate minerals are separated by their density differences, which can affect the recovery of finely disseminated sulfides associated with them. Silicatehosted nickel is another factor that accounts for higher nickel losses to the overflow, observed particularly in the Phoenix ore. Particle size is also an important control on DMS, where particles near the cut-point have a more-or-less equal chance of sinking or floating, and tend to separate on size rather than density. Small particles of less than ~2 mm are also more likely to float, causing even dense, sulfide-rich particles to be lost to the overflow. An evaluation of particle shapes shows that shape separation plays a minor role for the ores studied, and shape differences are most pronounced nearer to the DMS cut-point, where a higher proportion of irregular-shaped and elongated particles have been concentrated to the underflow. The ultimate aim of the characterisation of the DMC products would be to use the information gathered to be able to predict the behaviour of an ore being subjected to DMS, based on its mineralogy.

Africa is developing at a fast pace in the 21st century and this is accompanied by continuing urbanization. The economic growth rate, measured in terms of growth in gross domestic product (GDP), has been of the order of 4-5% p.a. for much of the past decade, well above the global average. With these increases in economic wealth and urbanization, people are becoming more affluent and demanding more consumer goods. This implies not just growth in the volumes of materials entering and leaving cities, but also a qualitative shift in what is increasingly called ‘the metabolism’ of the cities where this consumption is taking place, a trend expressing itself by a shift in provisioning from the traditional market to the supermarket. It stands to reason that with these changes in consumption comes an increase in the quantity and variety of waste generated. Past studies already show under-capacity waste management infrastructure and insufficient investment in urban waste management in Africa. The objectives of this dissertation are to 1) Develop and present models of the food material flow profile for a traditional market and supermarket consumer from production to disposal and develop waste flow profiles for both the traditional market and supermarket consumer; 2) Analyse the effect of change in food provisioning on the waste generated ; and 3) Describe the impact of this effect given the current waste management capacity and structure in African cities. To better inform waste management planning in African cities undergoing such a metabolic transition, metabolic flow models have been developed to describe the acquisition of food from a traditional market vs. a supermarket. Consumer food flows are formulated based on cultural profiling and quantified via the daily nutritional and energy requirements of a healthy person. The associated waste profiles of these food items are then obtained by reading from packaging and processing data sets. Results from the model confirm that there is a change in the waste profile associated with a shift in consumption from the market to the supermarket. The model is first applied to three diets: student, working class and vegetarian. The model shows 1) a shift of 14 to 58% of the organic content from the post-consumer waste to the supply chain where this food is processed and 2) a 3-10 fold increase in the inorganics in the post-consumer waste associated with food packaging used in supermarkets. These results are then interpreted for three case study cities: Kisumu in Kenya, Kitwe in Zambia and Harare in Zimbabwe, showing that this shift is being seen in all these cities. The shift in the organic waste is within the same range across all three cities within each income group. The increase in the inorganics however differs, being higher in Kitwe and Harare compared to Kisumu. This shows that the magnitude of changes in food-related waste generation is affected by the type of food consumed. Results from waste management surveys for the same case study cities show that there is a lack of sufficient capacity to provide adequate service to the people in all three cities. The city councils do not have the resources to collect and dispose of all waste generated in the cities and the private sector only provides services where it makes business sense. There is some effort in all three cities devoted to recycling and to management of organic wastes, but this is also inadequate and likely to be overwhelmed by the continuing metabolic shift. The findings point to the need for waste management becoming more commercially viable, adaptive and inclusive. There is also need for a sociocultural attitude change among the residents at household level who still look down upon waste management.

Proton exchange membrane fuel cells (PEMFCs) are an attractive energy conversion technology, this due to their high theoretical fuel utilization eciencies compared to Carnot engines. However, due to potential losses, the operational eciencies achieved in state-of-the-art PEMFCs are only between 45% and 55%. The slow kinetics of the oxygen reduction reaction (ORR) over a platinum based electrode accounts for ca. 70% of the potential losses. As a result of the sluggish ORR kinetics, high platinum loadings are required. The high cost of platinum has made it crucial to improve the ORR activity and hence reduce platinum loading. The surface-area-specic ORR activity has been reported to decrease with platinum particle size. This places a limitation to the degree to which platinum loading can be reduced by increasing metal dispersion. To understand the origin of this behaviour, experimental studies have measured the ORR activity over dierent single crystalline surfaces and used model nanoparticle shapes to elucidate the overall ORR activity. Theoretical studies use density functional theory (DFT) to investigate the ORR activity on various site-types present on assumed model particle shapes. Thermodynamically, the exposed surface terminations aught to be predominantly Ptf111g and Ptf100g separated by edges and corners. It has been postulated that the overall ORR activity can be calculated as a weighted average of the activity of exposed surface terminations. Using DFT calculations and nanorod models the above postulations are tested for the edge sites between a Pt(111) and Pt(100) surface. A rhombic nanorod model is used due to its computational eciency compared to model nanoparticle clusters which are generally large and computationally expensive models. Furthermore, the use of rhombic nanorod model enables the investigation of the connection and communication between the Pt(111) and Pt(100) facets, this is dicult to investigate with stepped-surface models. It is argued that if, (i) the edge has insubstantial eect on the adsorption strength of adsorbed ORR intermediates as a function of distance from the edge and (ii) the diusion of ORR intermediates between adjacent surface planes is limited, then the above postulation does hold. Using atomic O and O2 it was observed that the edge eect is a local phenomenon with adsorption involving only edge atoms being stronger than on extended surfaces. The adsorption of both atomic O and O2 was weaker on the Pt(111) nanorod terrace sites compared to the adsorption on equivalent sites on extended Pt(111) slabs. These eects were not observed for adsorption on Pt(100) nanorod terrace sites. The diusion of atomic O from the Pt(100) nanorod terrace bridge sites towards the Pt(111) nanorod terrace fcc sites (across the edge) was investigated. It was observed that whilst at lower coverage the diusion of atomic O is limited with an overall hopping frequency of 1.8810􀀀2 s􀀀1 (at T = 85 °C and P = 1 bar), at a higher coverage, the diusion of atomic O across the edge is facile (overall hopping frequency of 1.12105 s􀀀1). At a higher coverage limit and current density of 100 mA/cm2 (an arbitrary current density in the activation polarization region), the hopping frequency of atomic O was determined to be three orders of magnitude higher than the turnover frequency for H2O production. Hence at high coverage, the rst condition (i.e. edge eect is localised) is satised, but the second condition (i.e. limited diusion of atomic O from one facet to the other facet) can not be ruled out. Therefore, the assumption necessary for the evaluation of the overall ORR activity as a weighted average of individual activity, can be questioned.

The work performed in this thesis was aimed at evaluating the influence of lifter face angle on the charge kinematics for a laboratory scale mill. The study involved tracking a single particle representing the ensemble using the Positron Emission Particle Tracking (PEPT) to obtain the location of the particle with time. The particle was radiated with a radionuclide; 68Ga, which has a half-life of 68 minutes. The objectives of the study involved tests with different lifter face angles at different mill speeds and volumetric mill filling. After performing the tests the data was analysed to obtain probability density distributions for each test conditions from key charge descriptors. Charge descriptors such as the Centre of Circulation (CoC), shoulder angle, toe angle, the free surface and also kinematic information such as the velocity profile along a carefully chosen radial line from the centre of the mill that passes through the CoC were obtained. The time averaged velocity data was used when assessing the influence of the lifter face angle on the velocity profile. The results showed notable effects of lifter face angle on charge characteristics. No real definitive trend was observed for the CoC as the lifter face angle was altered at all mill speed and filling conditions. However, the CoC showed an outward shift toward the mill shell with an increase in mill speed but an inward shift toward the mill centre with increase in charge filling degrees. Mill speed is expected to cause a load expansion as the charge approaches centrifugation. The shoulder angle showed a general increase as the lifter angles became steeper for both an increase in mill speed and charge filling degrees. The impact toe angle displayed maxima with increasingly steeper lifter face angle depending on the condition altered. At the higher mill speeds (70 and 85 % of critical speed), the impact toe did not show as much variation with lifter face angle. However, a noticeable difference was observed at the lowest mill speed tested; which was 55 % critical speed. At 20 % charge filling degree, the impact toe increases from 45 to 90 ° lifter. At the higher filling degrees, a maximum was reached from 45 to 75 ° lifter and from 75 to 90 ° a decrease was seen. It is expected that the steeper lifter angles will cause the charge to impact the mill shell as the lifting forces are altered. The velocity profile demonstrated changes with face angle as distances along a radial line in the mill were investigated. As the speed was increased from 55 to 85 % of critical speed, the velocity profile showed minimal changes for the higher mill speed at distances closest to the mill shell, and more perceivable changes in the profiles at 55 % of critical speed for all radial distances investigated.

Polymer electrolyte fuel cells (PEFCs) have wide variety of commercial applications, however due to poor durability and high cost, this technology has currently not reached its commercialization stage. Poor durability is mainly attributed to carbon support corrosion during start-up and shut-down of the fuel cell. Corrosion of the electrocatalyst support materials has numerous adverse effects on the performance of the fuel cell, such as weakening of metal-support interaction which results in Pt detachment, dissolution and sintering. Hence, the electrochemical active surface area is significantly reduced. It is clear that there is an urgent need for more robust, high performance alternative support materials to carbon. In this study, metal nitrides and borides (TiN, ZrN, TiB2, ZrB2 and LaB6) were evaluated as potential support materials, in an attempt to improve the durability and performance of PEFCs. Metal nitrides and borides were first subjected to a stability test by exposing them to potential cycling from 0.0 V vs SHE to 1.5 V vs SHE in 0.1 M HClO4, 3000 times at room temperature. It was found that metal nitrides were more electrochemically stable than metal borides. Metal nitrides were passivated by formation of surface oxides (oxynitrides and metal oxides), whilst metal borides formed a B2O3 layer which is less protective than metal oxides and oxynitrides. Thus metal borides were continuously oxidised over the 3000 cycles; however, the rate of oxidation was much lower compared to Vulcan XC-72 carbon. Using a wet chemical (impregnation) method well dispersed Pt nanoparticles, with a narrow size distribution could be deposited on all support materials except for Zr-based materials. The prepared catalysts were electrochemically characterised and it was found that Pt supported on metal nitrides and borides showed less Pt utilisation and activity towards ORR, compared to Pt/C. This was attributed to the presence of surface oxides, which significantly reduced electron conductivity. Pt/LaB6 showed reasonable activity, although it was still lower than of its counterpart Pt/C. The durability of Pt/LaB6 was evaluated by applying typical load cycling and start stop simulating protocols and it was found to be more durable than Pt/C. From a performance point of view, Pt supported on carbon is still the best but less durable than metal nitrides and borides. However, Pt/LaB6 shows great potential.