Gasification of Glycerol Over Ni/γ-Al2O3 For Hydrogen Production: Tailoring Catalytic Properties to Control Deactivation

Abstract The effects of catalyst loading, calcination and reaction temperatures on the structural properties and catalytic behavior of Ni/γ-Al2O3 catalyst system in relation to steam reforming of glycerol and catalyst deactivation were investigated. The results showed that catalyst loading, reaction and calcination temperatures had a profound influence on the structure and catalytic activity in glycerol conversion. Use of high calcination temperature (900-1000 °C) led to phase transformation of the active Ni/Al2O3 to less active spinel specie NiAl2O4 that resulted in a successive change of texture and color. The particle size growth and phase change at this temperature were responsible for the catalyst deactivation and low performance especially among the catalyst calcined at high temperatures. Conversely, at low reaction temperatures, catalyst surfaces were marred by carbon deposition. Whilethe polymeric carbon deposited at metal-support interface was associated with low reaction temperatures, high reaction temperatures were characterized predominantly by both amorphous carbon deposited on the active metal surface and polymeric or graphitic carbon deposited at metal-support interface respectively. Calcination temperature showed no significant influence on the location and type of coke deposited on the catalyst surface. Hence, catalyst loading, calcination and reaction temperatures could be tailored to enhance structural and catalytic properties and guard against catalyst deactivation.


Introduction
The demand for energy will inevitably be on the rise considering the exponential growth in world population and rapid industrialization. Fossil fuels (largely petroleum) which are the most predominant and major energy resource in the world are depleting rapidly and are unevenly distributed around the globe; they are responsible for environmental issues affecting the world today. Such issues include atmospheric pollution, global warming, climate change, toxicity, non-biodegradability, and energy insecurity [1]. The Solid Oxide Fuel Cell (SOFC) produces electricity efficiently through the conversion of chemical energy which uses hydrogen and oxygen without compromising the environment. Development of SOFCs has recorded a technical peak and it is evident that hydrogen driven systems could replace fossil fuel-based combustion systems in transportation, stationary and distributed power systems in the near future [2]. Problems that include a lack of sustainable hydrogen resources, efficiency, affordability and reliable fuel processing systems as well as issues pertaining to hydrogen storage have continued to hinder its full commercialization [3,4].
Therefore, there is a need to explore more reliable and sustainable energy sources for energy diversification. Indicators revealed that biomass, if properly utilized, could maintain a high level of energy production for future needs. This is not only due to its inherent potential as an energy reserve, but also due to its renewability, sustainability, biodegradable nature, even distribution and other environmental benefits [5]. The success recorded in the transesterification of vegetable oil to biodiesel has led to growth in the biodiesel industry which has left an excess of the by-product glycerol; this in turn exceeds its demand and is being considered a waste product [6]. By-product glycerol lacks good fuel properties to burn in petrol or diesel engines and it does not have the required purity to be useful for conventional uses of glycerol in cosmetics, pharmaceutical, food industries or synthesis of other chemicals. More so, its disposal renege the environment. Interestingly, glycerol is biodegradable and an intense energy resource [7]. Because SOFCs operate at high temperatures and have flexibility regarding fuel choice, extensive research is focusing on using different processes for hydrogen production from biomass materials. These include liquid glycerol for the direct utilization of SOFCs or with little processing by way of reforming. Such reforming processes include steam reforming, partial oxidation and auto thermal reforming [8,9].
Glycerol has demonstrated great potential for hydrogen production using steam reforming as seen in several theoretical equations as follows: Ni/γ-Al 2 O 3 is a popular and promising catalyst system for steam reforming essentially due to nickel's propensity for C-C, C-H and H-O bond cleavages, high selectivity for syngas, cheap and ready availability. Gamma alumina is a good material widely used as a catalyst support or adsorbent due to its mesoporosity, ordered sponge-like pore structure, thermal stability and large surface area that could reduce coke formation and enhance diffusivity of reacting species and resulting products [10,11]. As such, a lot is reported on the use of Ni-based catalyst for steam reforming of biomass materials [12,13]. The major setback in the use of Ni based catalyst is deactivation due to carbon deposition and sintering at high temperatures. Some side reaction that could lead to coking includes but not limited to: 2CO ↔ CO 2 + C -171. 5

Catalyst Characterisations
Micrometrics TriStar II model was used to determine the Brunauer-Emmett-Teller (BET) surface area, Barret-Joyner-Halenda (BJH) pore volume and size using nitrogen adsorption/desorption at 77.35K. Each sample was degassed at 120 o C for 3hrs before the analysis. Crystallographic studies were carried out using Pan-Analytical Empyrean X-Ray Diffractometer operating on reflection mode using Cu-Kα 1 radiation at λ = 1.5406 Å and 10-90 2θ angle range for 1 hour.
Dispersion and morphology of Ni particles on the surface of the alumina was investigated using Field Emission Scanning Electron Microscopy (FESEM) JEOL JSM6700F micrographs using the Backscatter Method. The surface of each sample was coated with gold using Quorum Technologies Q150R instrument to enhance conductivity prior to the introduction into the SEM .
Used catalysts were analyzed for carbon deposition by temperature programmed oxidation (TPO) using NETZSCH STA 449C trigravimetric analyzer equipped with Thermostar mass spectrometer. After putting a known weight of the used catalyst into the TGA instrument, the system was flushed with argon to dispose of any other gas in the system. Oxidation of the surface carbon was done using O 2 gas and increased gradually 900 o C. Gases generated were directed to the mass spectrometer for detection.

Catalyst Testing
Steam reforming of pure glycerol was carried out in a fixed quartz tube (10mm outer diameter (OD), 8mm inner diameter (ID) and 25cm long) at 300, 500, 600, 700 and 800 o C at atmospheric pressure (standard pressure). The glycerol/water mixture was supplied using a syringe Harvard apparatus 22 infusion pump at the flow rate of 0.019ml/min to a stainless steel pipe wrapped with a heating tape at 250 o C for vaporization and gas mixing. The vaporized reactant mixture was conveyed into the reactor by a carrier gas (Helium) at a flow rate of 40 ml/ min and using H 2 O/C =3. Gaseous products were analyzed using GC equipped with TCD (HP 6890 series) and a mass spectrometer residual gas analyzer.

Catalyst Characterization
Results of surface area analysis in Table-1a show that dehydration due to calcination at 700 o C reduced the surface area of the fresh γ-Al 2 O 3 from 76.64 m 2 g -1 to 74.46 m 2 g -1 and resultant phase transformation from hydroxide to an active form of γ-Al 2 O 3 as corroborated by the XRD pattern in Figure 2b. In Table 1a, the addition of active component NiO particles on the surface of the dehydrated alumina by wet impregnation progressively reduced the surface area of the γ-Al 2 O 3 support as the loading increased. This indicates that active components have covered specific proportions of the surface of γ-Al 2 O 3 support resulting in the dispersion of the active phase on the γ-Al 2 O 3 support as evident in the SEM micrograph in Figure 1 below. Table 1b revealed that the surface area also progressively reduced as the calcination temperature increased from 300-1000 o C due to particle size growth and agglomeration at high temperatures. There is a progressive increase in pore volume with an increase in the catalyst loading in Table  1 attributable to the piling up or lumping of the catalyst , thereby making the pore volume larger. The pore volume is reduced in Table 2 and is constricted due to the influence of temperature.
The diffraction peaks for both NiO and active γ-Al 2 O 3 were observed in the samples XRD pattern as indicated below in Figure 2a. The two most noticeable peaks at 46 o , 68 o and a wider one at 37 o confirms γ-Al 2 O 3 and their presence at all calcination temperatures as seen in Figure  3 signifying stability of γ-Al 2 O 3.
The increase in the intensity of the alumina peak at 37 o with an increase in nickel loading (from 5-30%) shows an overlap of the nickel peak with the alumina peak which depicts a possible formation of the spinel NiAl 2 O 4 structure. The peak due to the spinel NiAl 2 O 4 becomes more evident at 1000 o C in Figure 3 due to its temperature dependency. More so, in the same figure, the loss of NiO at 45 o , 78 o and 80 o at high temperatures of 900-1000 o C leaves only peaks corresponding to NiAl 2 O 4 has further support phase transformed to NiAl 2 O 4 at high temperature.
In Figure 3, the XRD of Ni/Al 2 O 3 catalyst shows the effect of calcination temperatures intensity of the Al 2 O 3 peaks as they become sharper at high temperature of 900 o C and increase in intensity at 1000 o C indicating crystallinity  [11]. Observed increases in peak intensity with an increase in nickel loading could be caused by the lumping or piling up of nickel particles as confirmed in the result of crystallite size calculation using Debye-Scherrer's equation below in Table 2.

Catalyst Activity in relation to Catalyst Loading
To evaluate performance of the catalysts containing different Ni loading, a reforming test was carried out at 500 o C using catalysts prepared at 500 o C calcination temperature, but containing different catalyst loading at the weight hourly space velocity (WHSV) of 28 h -1 . The results of the catalyst test based on 100mg of the catalyst amount in the reactor revealed that catalyst activity taken as glycerol conversion to gaseous products and product distribution increased with catalyst loading from 5wt% to 10wt% and decreased at 30wt% as shown in Figure 4a. As the loading increased to 30wt%, the drop in conversions could be attributed to the reduced surface area and particle size growth due to nickel particles piling up as exhibited in Table 1a and Table 2. This may have reduced the surface active site accessibility leading to a decrease in surface reaction. Therefore, 10wt% catalyst loading represents the optimum catalyst (NiO) loading for a maximum catalyst conversion in this work. The product distribution shows H 2 , CO 2 , CO, and CH 4 as the only gaseous products detected during the steam reforming of glycerol at such temperatures. H 2 and CO 2 were the predominant products with high readings of CO. The product distribution suggests catalytic decomposition due to nickel's propensity for glycerol C-C cleavages which could be responsible for the high CO content. Other reactions that contribute to the observed performance include a water gas shift reaction (WGSR) as shown in the equations above. Furthermore, hydrogen production per mole glycerol increased with catalyst loading. The hydrogen was about 60% of the equilibrium composition with 10wt% catalyst which increased to 66% with 30wt% catalyst loading. The turnover frequency which expresses the number of molecules that react per surface area of the catalyst shows reasonable value corroborating the usefulness of the catalyst system under these conditions for hydrogen and syngas production.

Correlation between Reaction Temperature and the Catalytic Behaviour.
The influence of the reaction temperature was investigated using 10wt% Ni/Al 2 O 3 calcined at 500 o C which gave the highest conversion of glycerol in the screening test seen in Figure 4. The test was carried out at different reaction temperatures between 500-800 o C using the S/C ratio of 3 and WHSV of 28 h -1 . The graph below in Figure 5a  illustrates product distribution changes with temperature.
The results based on 100mg catalyst amount in the reactor show that reaction kinetics influenced by temperature play an important role in the product distribution. This is attributable to enhanced rapid glycerol decomposition due to kinetics. H 2 and CO 2 yield was high at 500 o C which may be due to a water gas shift reaction (WGSR) since is slightly exothermic (Eqn3). There was a simultaneous decrease in CO and CH 4 at the same temperature (500 o C) because CO was used in the WGSR and CH 4 decomposition to yield more H 2 in a slightly exothermic reaction (Eqn8). H 2 and CO 2 yield dropped at 600 o C with a corresponding stabilization of CO and CH 4 formation or yield. This suggests, as the temperature increased, a reverse water gas shift reaction which is slightly endothermic and which favors a CO formation by consuming CO 2 and H 2 . The drop in H 2 and CO 2 at 600 o C is attributable to a reversible self-poisoning deactivation effect which occurs most frequently at 550-600 o C [15]. The observed increase in CO with temperature is also an indicator of catalyst deactivation. At high temperatures, highly endothermic steam reforming and glycerol decomposition favored a rise in H 2 and CO 2 selectivity, while CO and CH 4 decreased. The CO rise from 600 o C-700 o C could be attributed to methane reforming which is a highly endothermic reaction. These findings relate to the thermodynamic observations [16,17]. Figure 5b shows how a change in reaction temperature influences H 2 selectivity and how that compares with glycerol conversion at different temperature in reverse trend. Glycerol conversion decreased at 600 o C from 500 o C but increased steadily to 800 o C. Conversely, hydrogen selectivity decreased progressively as temperatures rose to 800 o C as evident in Figure 5b. This could be due to the increased glycerol decomposition and the reforming process as both are highly endothermic processes (Eqn1&2). Carbon deposition also decreased with temperature as seen in section 3.3.1 which might have favored a glycerol conversion. It is evident that temperature enhances rapid glycerol decomposition and glycerol conversion; however, hydrogen production and selectivity did not show a noticeable increase at high temperatures. 500 o C represents the optimum reforming temperature with the 10wt% Ni/Al 2 O 3 catalyst system in this work.

Influence of calcination temperature on the steam reforming
Both reaction temperatures and calcination temperatures of the catalyst has profound influence on the selectivity of the desired products. The test was carried out under the same reaction conditions as stated in section 3.2.2. Figure  6 shows that the catalyst prepared at a low temperature had a better selectivity for hydrogen than the one prepared at high calcination temperatures. The catalyst prepared at 300 o C showed a steady increase in hydrogen selectivity with an increase in reforming temperatures. In contrast, the catalyst prepared at 500 o C showed a steady decrease in hydrogen selectivity as the reaction temperature increased. This is attributable to the change in surface morphology and chemistry with increase in reaction temperature. It is apparent that H 2 and CO 2 yield is better among catalysts calcined at low temperatures, which suggests they are good for WGSR. This may be due to their better surface area, smaller particle size and good catalyst-support interaction. Relatively, due to sintering and agglomeration, those catalysts prepared at high temperatures-900-100 o C favor reverse water 9 (a) (b) Figure 5 (a), Change in reformate composition with temperature (b), Comparison between hydrogen selectivity and glycerol conversion and influence of reaction temperature gas reaction, hence reduced H 2 and CO 2 and increased CO and CH 4 formation. Consequently, low calcination temperatures (300-500 o C) favor maximum hydrogen yield and selectivity.

Carbon deposition
Post catalysts test investigation on the XRD pattern of the used catalyst samples (Ni/γ-Al 2 O 3 ) in Figure 7 shows a peak at 26.2 2θ which remains absent in the fresh Ni/γ-Al 2 O 3 . The peak corresponds to carbon which suggests carbon deposition. A similar observation was made by Newnham et al [18]. More so, peaks due to Ni metal are prominent at 52.1 2θ and 76.6 2θ in the XRD pattern of the used catalyst as a result of reduction of the catalyst before use compared to the fresh sample pattern with peaks due to NiO only. Temperature programmed oxidation (TPO) analysis on the used samples showed one or more CO 2 peaks that corresponded to the temperature at which coke oxidation occurred. Different types of coke such as graphitic, polymeric or amorphous carbon are oxidized at different temperatures. The TPO profile could be used to characterize the coke deposited. Moreover, depending upon the position of the major CO 2 peak, the point or particular place on the catalyst surface where the bulk of the carbon is deposited, could be determined. Low temperature peaks correspond to amorphous coke deposited on the Ni active phase or metal centres; medium temperature peaks correspond to polymeric or graphitic coke deposited at metal-support interface while high temperature CO 2 peaks signify carbon or coke deposited on the support [19]. Figure 8 shows the TPO profiles obtained from oxidation of used catalyst calcined at 500 o C; but they ran at different temperatures between 500-800 o C for the steam reforming. It is evident that the extent or amount and type of coke deposited was affected by changing the reaction temperature. At intermediate reaction temperatures between 500 and 600 o C,a major CO 2 peak was detected at 500-520 o C. This indicated that the bulk of the coke was polymeric or graphitic and deposited at the metal-support interface. The small CO 2 peak detected at low temperature shows that oxidation starts at a low temperature. Table 3 summarizes the peak positions as seen in Figure 8. Conversely, high temperature reforming between700-800 o C is characterized by low, intermediate and high temperature peaks with the dominant peaks at low temperatures indicating that most of the carbon was deposited on the active phase (Ni) and is an amorphous type usually easy to remove and could be regenerated by oxidation. The extra peaks at intermediate and high temperatures suggest a presence of polymeric or graphitic carbon as well as bulk carbon which are difficult to remove and therefore, oxidize at higher temperatures.

Influence of reaction temperature on coke deposited
The TGA profile showed a decrease in carbon deposition with an increase in reforming temperatures Figure 6 Influence of calcination and reaction temperature on hydrogen selectivity  which proved that carbon deposition was indeed limited by temperature as suggested in Eqn7-9 above. The equations show that the reactions leading to carbon deposition are exothermic and therefore favored at low temperatures and hindered at high temperatures. At a reaction temperature of 800 o C, carbon deposition was almost completely hindered which aligns with thermodynamic observations [16,17]. The catalyst used at a reaction temperature of 600 o C gave the highest carbon deposition which is attributable to the occurrence of side reactions due to the self-poisoning effect highlighted above which led to catalyst deactivation. Generally, the carbon deposition was about 16% and the highest at 600 o C. The acidic Al 2 O 3 support surface attracted a large amount of glycerol to its surface leading to the formation of carbonaceous products that could lead to deactivation. Figure 8 shows how calcination temperature influences the quantity and type of coke deposited. The catalysts were prepared at different temperatures of 300 o C, 500 o C, 900 o C and 1000 o C respectively and tested for steam reforming at a constant temperature of 500 o C. TPO on the used catalyst was analyzed, and the results are shown in Figure 8. The major CO 2 peak is a medium temperature type, although the catalyst calcined at lower temperatures show some lower temperature peaks. The carbon is more graphitic with the catalyst calcined at 500 o C due to the intense surface reaction it attracted. Others gave polymeric carbon, hence the CO 2 peak occurred at a slightly lower temperature. The lower temperature catalysts had more carbon deposition compared to the high temperature catalysts in reference tothe TGA profile. This might be due to their large surface area and higher glycerol adsorption; therefore intense surface reactions could leave behind more carbon deposition. Table 4 summarises the peak positions as per Figure 9.

Conclusion
Ni/γ-Al 2 O 3 catalyst was successfully synthesized containing different Ni loading by wet impregnation and the catalytic performance in steam reforming of glycerol was evaluated. H 2 yield and selectivity increased with an increase in the catalyst loading up to 30wt%, while the glycerol conversion increased to 10wt% and     dropped at 30%. Hence, 10wt% nickel catalyst loading is the optimum catalyst loading for maximum glycerol conversion. Both reaction and calcination temperature have profound influence on steam reforming of glycerol. Lower reforming temperature-500 o C favor water gas shift reaction to yield more H 2 and CO 2, while high reforming temperatures between 700-800 o C favour reverse water gas shift reaction, hence the domination of CO and CH 4 . Phase transformation of the catalyst to spinel specie NiAl 2 O 4 occurred at higher temperatures between 900-1000 o C. The NiAl 2 O 4 formation, agglomeration and carbon deposition were identified as the main agents of deactivation. Reforming and calcination temperatures affect both extent and the type of carbon that was deposited. At low reforming temperatures of 500-600 o C, most of the carbon was polymeric carbon and deposited at metal-support interface. High reaction temperatures were predominantly characterized by amorphous carbon depicted by the low temperature CO 2 peak deposited on the active phase (Ni). There is also little polymeric or graphitic carbon at the metal-support interface shown by the medium temperature CO 2 peak as well as bulk carbon represented by the higher temperature CO 2 peak deposited on the support at higher temperature. The amount of carbon deposited decreased with an increase in reaction temperature as well as calcination temperature. Calcination temperature did not have an influence on the type of carbon except for the carbon being more graphitic as the active catalyst calcined at 500 o C. Hence, catalyst loading, calcination and reaction temperatures could be tailored to enhance structural and catalytic properties as well as suppress the carbon deposition to mitigate catalyst deactivation.