Effect of acid (CH3COOH, H2SO4 and H3PO4) and basic (KOH and NaOH) impurities on glycerol valorisation by aqueous phase reforming
This work analyses and compares, under the same operating conditions (220 ºC and 44 bar with a Ni-La/Al2O3 catalyst), the effects of some of the most common acid (CH3COOH, H2SO4, H3PO4) and basic (KOH and NaOH) biodiesel-derived impurities on the aqueous phase reforming (APR) of a 30 wt.% glycerol solution. The statistical analysis of the results revealed that the impurities did not greatly influence the initial reforming results. Conversely, they significantly influenced catalyst deactivation, which resulted in different evolutions over time for the glycerol conversion, liquid production and the composition of the gas and liquid phases. Significant decreases over time in the glycerol conversion and liquid production were detected, the severity of decay being as follows: H3PO4 (KOH=NaOH)>H2SO4 (KOH<NaOH)>CH3COOH (KOH<NaOH). The characterisation of the spent catalyst and the liquid phases revealed that poisoning/fouling, and catalyst active phase or support modification (leaching and crystalline phases alteration) were the major deactivation mechanisms. The proportions of metals (K or Na) deposited on the catalysts with the different acids was as follows: H3PO4 > H2SO4 > CH3COOH. In addition, S and P were also deposited on the catalyst and boehmite and other new crystalline phases were detected in the spent catalyst after the APR reaction.
Keywords: crude glycerol, aqueous phase reforming, catalyst deactivation, acid impurities and basic impurities.
Highlights (maximum 85 characters including spaces)
- Effect of several biodiesel-derived acids and bases on glycerol APR
- Acids & bases do not affect the initial APR results but deactivate the catalyst
- CH3COOH and H3PO4 show the smallest and highest deactivation, respectively
- K, Na, P and S deposition on the catalyst influences catalyst deactivation
- New crystalline phases appeared in the catalyst and contributed to its deactivation
Worldwide new strategies and policies to mitigate global climate change have led researches to seek new processes, renewable materials and more sustainable strategies to replace the current petroleum based energy industry towards a greener and more environmentally friendly energy market. Under this new energetic picture, crude glycerol is on particular interest because of its ample availability as a biodiesel by-product (on average, 1 kg of crude glycerol is yielded with the production of 10 kg of biodiesel); thus converting this feedstock into a cheap resource for which new processes, policies and strategies need to be developed .
Given this scenario, different valorisation routes such as gasification, steam reforming, aqueous phase reforming and supercritical reforming have been recently addressed for the production of value-added chemicals and energy from glycerol [2, 3]. Among them, Aqueous Phase Reforming (APR), a low temperature, moderate pressure, catalytic process is a promising route for the valorisation of crude glycerol for the production of value-added gaseous and liquid chemicals. The gas phase is made of H2, CO, CO2 and CH4, while the liquid phase consists of an aqueous solution containing carboxylic acids, ketones, esters, alcohols, aldehydes, paraffins along with other oxygenated compounds. The yields and compositions of the gas and liquid phases depend on the nature of the feed, the catalyst type and the operating conditions [4-7], allowing the customisation of this valorisation route to suit the different needs of the market.
There are an extensively number of works in the literature dealing with the APR of pure glycerol addressing the effect of the catalyst type and the operating conditions. Noble metal catalysts based on Pt [8-13], Ni [8, 9, 13-17], Pt-Ni, Cu, Co or Ru [7, 8, 13, 15, 18] supported on Al2O3, ZrO2, MgO, SiO2, CeO2, or carbon [4, 19] and altered, in some cases, with promoters such as La, Ce, Mg and Zr have been synthesised, characterised and tested. In addition, different parametric analyses have been conducted with reagent grade glycerol [11, 16, 18, 20, 21]. These studies provide valuable information on the APR process for pure glycerol for both batch and flow reactors. However, the impurities accompanying the crude glycerol obtained from the biodiesel industry can significantly alter the results of the APR of glycerol, modifying the selectivity and causing the deactivation of the catalyst. Therefore, for the development and scale up of this technology is essential to know the effects of these impurities on the process.
Crude glycerol consists not only of glycerol but also of many other troublesome impurities such as alcohols, soaps, catalysts, salts and non-glycerol organic matter . To overcome this issue, a first cost effective purification of the crude glycerol was addressed in other works dealing with crude glycerol [22, 23]. This method is fully described in the work of Manosak et al. . Briefly, it consists of the physical separation of the FAMES and the elimination of the soaps by an initial acidification, normally with acetic, sulphuric or phosphoric acid, and a subsequent liquid-liquid extraction with a polar solvent such as methanol . As a result, a glycerol solution with 85-90% purity is obtained. This purified glycerol still contains some of the acid used in the neutralization, part of the catalyst employed in the biodiesel production (usually KOH or NaOH) as well as the alcohol (normally CH3OH) used during the transesterification reaction and/or in the purification step.
The works dealing with the APR of crude glycerol are extremely scarce [4, 7, 9, 24] and few of them analyse the effect of the impurities found in the feedstock on the process. Lehnert and Claus  reported the APR of pure and crude glycerol at 250 ºC and 20 bar Ar using various Pt catalysts. The presence of NaCl in the glycerol solution was responsible for the lower selectivity to H2 and the higher deactivation of the catalyst occurred with crude glycerol. King et al.  analysed the APR of 10 wt.% glycerol/KOH solution using various Pt/C and Re/C catalysts. The addition of 0.1 wt.% of KOH to the solution until a pH of 12 was reached resulted in an increase in the glycerol conversion and the H2 production. Boga et al.  conducted APR experiments with pure and crude glycerol (6.85 wt.% glycerol, 1.62 wt.% soaps, 1.55 wt.% methanol and 0.07 wt.% esters). The use of this crude glycerol solution resulted in a dramatic depletion in the conversion compared to pure glycerol. The results obtained with different synthetic mixtures revealed that fatty acids sodium salts inhibited H2 formation and exerted a much more pronounced negative influence than NaOH. Remón et al.  analysed the effect of the presence of different amounts of acetic acid, methanol and potassium hydroxide in a 30 wt.% glycerol solution during the APR at 220 ºC and 44 bar using a Ni-La/Al2O3 catalyst. It was found that methanol decreased the glycerol conversion while acetic acid and potassium hydroxide decreased and increased the gas production, respectively. Potassium hydroxide promoted H2 production due to the greater gas formation and the lower H2 consumption. As regards liquid production, methanol increased the proportion of monohydric alcohols, while potassium hydroside did not greatly vary the liquid product distribution. Acetic acid promoted dehydration reactions, which led to a decrease in the proportion of monohydric alcohols and an increase in the relative amount of C3-ketones in the liquid product.
These publications provide valuable information about the presence of some of the most common impurities found crude and refined glycerol. However, since different feedstocks, catalysts, reactors and operating conditions were used, the comparison between the effects of the different impurities in the process is unreliable. Given this scenario, this work analyses and compares under the same operating conditions (220 ºC and 44 bar using a Ni-La/Al2O3 catalyst) the effects of some of the most common acid and basic impurities found in crude and purified glycerol: CH3COOH, H2SO4, H3PO4, KOH and NaOH. These operating conditions were used in our previous work , thus allowing a deeper insight into the APR of crude glycerol to be gained. The effects of these impurities have been analysed on the glycerol conversion, the product distribution in carbon basis and the compositions of the gas and liquid phases. In addition, the spent catalyst was exhaustively characterised by several techniques, which allowed a relationship to be established between the impurities and both the APR results and catalyst deactivation. Given the fact that the effect of the presence in crude glycerol of all these acid and basic impurities under the same operating conditions has never been addressed before, and considering the limited number of publications dealing with crude glycerol along with the scarce number of publication analysing catalyst deactivation in APR, this work represents a new investigation not only for a better insight into the APR of crude glycerol to be gained, but also to understand and prevent the most common catalyst deactivation mechanisms during this thermochemical process.
2.1 Experimental system
The experiments were conducted in a continuous pressurised fixed bed reactor for 3 hours using a Ni-La/Al2O3 catalyst with the following characteristics: 28% (relative atomic percentage) of Ni expressed as Ni/(Ni+Al+La), an atomic La/Al ratio of 0.035 and a BET surface area of 187 m2/g. The experimental facility is a “microactivity unit” designed and built by PID (Process Integral Development Eng Tech, Spain). It consists of a stainless steel pressurised fixed bed reactor (9 mm of inner diameter) heated up by means of an electric furnace. The system pressure, measured with a pressure gauge located at the exit of the reactor, is reached with the aid of an automatic micrometric valve operated by a rotor. The aqueous solutions containing the glycerol and impurities are fed into the reactor from its bottom part by using a high performance liquid chromatography (HPLC) pump. The liquids (unreacted reactants and reaction products) together with the gases produced during the reaction leave the reactor from its upper part, pass through the micrometric valve and arrive at the condensation system. This consists of a set of three condensers where the liquids are separated from the gas mixture (N2 used as an internal standard and the gases produced) at intervals of 1 h to analyse the evolution over time of the liquid phase. A micro gas chromatograph (Micro GC) equipped with thermal conductivity detectors (TCD) was used for the online analysis of the gas phase. The liquid samples were collected and analysed offline with a gas chromatograph equipped with Flame Ionization (FID), and Mass Spectrometry (MS) detectors. More detailed information about the experimental unit and the catalyst properties can be found in our previous communications [22, 25, 26].
2.2 Experimental plan, response variables and data analysis
The influence of the presence of some common glycerol biodiesel derived acid (CH3COOH, H2SO4 and H3PO4) and basic (KOH and NaOH) impurities has been investigated during the aqueous phase reforming of a water solution containing 30 wt.% glycerol and 2.5 wt.% methanol. Table 1 shows the pH of the solutions and the concentrations of the acids (CH3COOH, H2SO4, H3PO4) and bases (KOH and NaOH) used to prepare them. The concentration of methanol was chosen having regard to the glycerol/methanol ratio found in biodiesel-derived refined glycerol solutions [1, 27] and previously used in our previous communication , while the concentrations of the acid and basic impurities were selected to have the same concentration of H+ and OH– (same pH) in all the solutions as those reported for CH3COOH .
The aqueous phase reforming experiments were conducted at 44 bar and 220 ºC with a Ni-La/Al2O3 catalyst using a spatial time defined as the mass of catalyst/mass flow rate of glycerol (W/mglycerol) ratio of 25 g catalyst min/g glycerol and a liquid flow rate of 1 mL/min . The response variables studied were the global glycerol conversion (X gly), the carbon converted to different products: carbon conversion to gas, liquids and solid (CC gas, CC liq and CC sol) as well as the composition of the gas (N2 and H2O free, vol.%) and liquid (relative chromatographic area free of water and un-reacted glycerol, %). More information about the response variables and the analytical methods used for their calculation can be found in our previous communications [22, 25]. In addition, Inductive Coupled Plasma Optical Emission Spectrometry (ICP-OES) was used to measure the amount of metals leached from the catalyst to the liquid phase.
The experimental results for each response variable are divided into three intervals, which correspond to the average value of each variable obtained during the first, second and third hour of the experiment. All the experiments were conducted at least in duplicate and the influence of each impurity in the process have been analysed using a one-way analysis of variance (one-way ANOVA) with 95 % confidence. The results of the ANOVA analyses are provided as p-values. P-values lower than 0.05 indicate that at least two values are significantly different. Furthermore, the Fisher’s least significant difference (LSD) test was used to compare pairs of data, i.e. either to analyse the effect of the reaction time between the same experiment and the effect of the impurities between different experiments. The LSD test results are presented graphically by means of LSD bars. Significant differences (with 95% confidence) between any pair of data can be ensured when their LSD bars do not overlap.
2.3 Characterisation of the catalyst after reaction
The spent catalyst was characterised by elemental (CHNS) and thermo gravimetric (TG) analyses, X-Ray diffraction (XRD) and Inductive Coupled Plasma Optical Emission Spectrometry (ICP-OES). TG analyses were conducted under a N2 atmosphere from 25 ºC to 600 ºC at a heating rate of 10 ºC/min, monitoring the weight loss of the samples. XRD patterns of the original and the spent catalysts were obtained with a D-Max Rigaku diffractometer equipped with a CuK 1.2 at a tube voltage of 40 kV and current of 80 mA using continuous-scan mode with steps of 0.03◦/s at Bragg’s angles (2) ranging from 5◦ to 85◦. The phases present in the samples were defined with reference to the JCPDS-International Centre for Diffraction Data 2000 database.
2.4 Glycerol aqueous phase reforming reaction network
The aqueous phase reforming of glycerol comprises the formation of gases and liquid products. Three possible parallel routes, shown in Figure 1, explain the formation of liquids products: glycerol dehydration to 1-hydroxypropan-2-one (A) [4, 5, 18, 28-31] and/or to 3-hydroxypropanal (B) [5, 29-31] and/or glycerol dehydrogenation to 2,3-dihydroxypropanal (C) [4, 5, 18, 28-31]. This mechanism has been thoroughly discussed in our previous communications [22, 25]. The gas phase is made up of a mixture of H2, CO2, CO and CH4. The thermal decomposition and/or reforming reactions of the glycerol and all the liquid intermediates (Eq.1) as well as by all the decarbonylation reactions are responsible for the formation of H2 and CO while the water gas shift reaction (Eq.2) and methanation reactions (Eq.3-4) accounts for the presence of CO2 and CH4 in the gas phase [4, 5, 18, 28-31].
CnHmOk + (n-k) H2O n CO + (n+m/2 –k) H2 (Eq.1)
CO + H2O CO2 + H2 (Eq.2)
CO + 3 H2 CH4 + H2O (Eq.3)
CO2 + 4 H2 CH4 + 2 H2O (Eq.4)
3.1 Glycerol conversion and C product distribution (CC gas, CC liq and CC sol)
Figure 2 shows the amount of carbon converted into gas and liquid products with respect to the total amount of carbon in the feed: carbon converted to gas and liquid (CC gas, CC liq) along with the global conversion of glycerol (X gly). Statistically significant differences between the results obtained in the experiments were found for the CC gas, CC liq and X gly (p-values < 0.001) with variations between 5-19%, 10-55% and 8-72%, respectively. The CC sol was lower than 5% in all the cases and the effect of the impurities on the CC sol was not significant (p-value > 0.05). The effects of the type of acid and base for these variables have been analysed comparing the results obtained during the first hour of reaction (initial values) and the evolution of these variables over time to analyse a possible catalyst deactivation and/or changes in the reaction pathway.
The statistical analysis of the results reveals that the type of acid and base does not significantly influence the initial global results; i.e. the values obtained for the CC gas, CC liq and X gly during the first hour of experiment with the different acids (CH3COOH, H2SO4 or H3PO4) and bases (KOH or NaOH) are non statistically different (p-value >0.05). These results are the consequence of having used the same pH to prepare the solutions  and suggest that the presence of different amounts of S, P, Na or K in the solution does not exert a significant influence on the initial global results during the APR of glycerol at the experimental condition tested in this work.
Conversely, the impurities exert a significant influence on the evolution over time of these variables. Specifically, significant decreases occur for the X gly and the CC liq; the variations observed for the CC gas being less marked. These developments suggest a possible catalyst deactivation, which does not influence gas production in some cases. Specifically, the variations observed for the CC gas depends on the type of acid. More precisely, while not significant variations occur with CH3COOH or H2SO4 (either accompanied by KOH or NaOH), a sharp decrease in the CC gas over time takes place when H3PO4 is present in the solution regardless of the base. For this latter acid, the same decay over time in the CC gas takes place with KOH and NaOH; thus suggesting that in this case, H3PO4 is the responsible for the decrease observed in gas production. The evolution over time for the CC liq shows significant decreases in some cases. While not significant variations occur with CH3COOH in the solution (either with KOH or NaOH), a moderate and a sharp decay occurs with H2SO4 and H3PO4, respectively. For H3PO4, a sharper decrease takes place in the presence of KOH than with NaOH. This suggests a synergetic acid-base interaction that might lead to an increase in the deactivation of the catalyst. The glycerol conversion shows significant decreases over time in most of the experiments. Both the type of acid and base influence the evolution over time of the X gly. Regardless of the type of base (KOH or NaOH), the decreases for the X gly are as follows: CH3COOH (KOH < NaOH) < H2SO4 (KOH < NaOH) < H3PO4 (KOH = NaOH).
3.2 Gas composition
Figure 3 shows the gas composition obtained with the different acids and bases considered in this work. The analysis of the results reveals a statistically significant influence (p-values < 0.05) of the impurities on the proportions of H2, CO2, CO and CH4 in the gas. The greatest variations occur for the relative amounts of H2 (29-44 vol.%) and CH4 (15-22 vol.%); the variations occurring for CO2 (38-46 vol.%) and CO (1-3 vol.%) being less marked and irrelevant from a practical point of view. In addition, the statistical analysis reveals that the impurities exert different effects on the initial composition of the gas phase. Specifically, the relative amounts of H2 and CH4 are greater affected by the presence of the different types of acids and bases than the proportions of CO and CO2. For these two later gases, although the variations observed are statistically significant, they are not practically important.
A multivariate analysis by means of Spearman’s test was carried out for the relative amounts of H2, CO2, CO and CH4 in the gas. This test revealed a statistically significant relationship between the proportions of H2 and CO2 (p-value = 0.0001; R2 = 0.91) and the proportions of H2 and CH4 (p-value = 0.0001; R2 = 0.94). This indicates that the impurities considered exert a significant influence on the methanation reaction (CO2 + 4 H2 CH4 + 2 H2O) and the WGS (CO + H2O CO2 + H2) reactions. These variations depend on the type of acid and base considered. The highest proportion of H2 in the gas along with the lower concentration of CH4 is obtained with H3PO4, either in the presence of KOH or NaOH. A lower proportion of H2 together with a higher relative amount of CH4 in the gas is produced with H2SO4 and NaOH in the glycerol solution.
The evolution of the gas over time shows small variations for the composition of the gas. This indicates that catalyst deactivation does not greatly affect gas selectivity, i.e. lower amounts of all the gases are produced but the composition of the gas is not altered. This development was reported in our previous work analysing the influence of CH3OH, CH3COOH and KOH during the APR of a 30 wt.% glycerol solution with the same catalyst and at the same operating conditions . The greatest variation in the gas composition occurs for H3PO4. For this acid, regardless of the type of base used, an initial decay followed by a posterior increase in the proportion of H2 in the gas takes place. This development occurs along with opposite variations in the amounts of both CO2 and CH4. The increases in the H2 content in the gas from the second to the third hour could be accounted for by a lesser spread of hydrogenation reactions due to the progressive deactivation of the catalyst; thus increasing the proportion of H2 in the gas [22, 25].
3.3 Liquid composition
Figure 4 summarises the relative amount of the different families of compounds present in the liquid for the experiments conducted. The liquid phase is made up of a mixture of aldehydes, carboxylic acids, alcohols and ketones together with unreacted glycerol and water. Acetaldehyde and acetic acid are the most abundant compounds for the aldehydes and carboxylic acids, respectively. The alcohols mostly comprise monohydric (methanol and ethanol) and polyhydric (propane-1,2-diol, ethane-1,2-diol and butane-2,3-diol) alcohols. The ketones include C3-ketones such as propan-2-one (acetone) and 1-hydroxy-propan-2-one and C4-ketones such as 3-hydroxy-butan-2-one. The presence of these compounds in the condensates is consistent with the pathway proposed in Figure 1 [22, 25], and those proposed by several authors who have studied the APR of glycerol [4, 5, 18, 28-31].
The statistical analysis reveals that the impurities exert different effects on the composition of the liquid phase. Specifically, the initial amounts (1st h) for the different families of compounds are not statistically affected by the presence of the impurities in the glycerol solution (p-value > 0.05). The initial composition of liquid phase shifts as follows: aldehydes (0.84-1%), carboxylic acids (0-1.5%), monohydric alcohols (21-23%), polyhydric alcohols (59-61%), C3-ketones (11-14%) and C4-ketones (2-4%). Conversely, the impurities exert a significant influence on the composition of the liquid phase over time. Significant variations take place for the proportions of alcohols and ketones; the variations observed for the relative amounts of aldehydes and carboxylic acids being less important from a practical point of view due to the small proportion of these two families in the liquids.
These variations over time depend on the type of acid. While significant variations do not take place in the presence of CH3COOH, minor and great changes in the proportions of alcohols and ketones occur with H2SO4 and H3PO4, respectively. Among them, the greatest variations take place for the proportions of polyhydric alcohols and ketones in the presence of H3PO4. Specifically, decreases over time happen for the proportions of polyhydric alcohols and C4-ketones in the liquid, while C3-ketones show increases over time. These evolutions depend on the type of base and more pronounced variations occur with KOH than with NaOH, except for C3-ketones. The changes observed in the presence of H2SO4 are less marked; a small depletion occurs in the proportion of polyhydric alcohols along with an increase and a reduction in the relative amounts of C3-ketones and C4-ketones, respectively. In this case, these developments do not depend on the type of base and the same variations are observed with either KOH or NaOH. The variations in the proportions of polyhydric alcohols and C3-ketones account for the decrease in the proportion of propane-1,2-diol and the increase in the relative amount of 1-hydroxypropan-2-one, respectively. These developments indicate that catalyst deactivation hinders hydrogenation reactions in route A [22, 25]. This latter is believed to be a consequence of the deactivation of the metal active centres in the catalyst.
3.4 Characterisation of the spent catalysts
To gain a better insight into the effects of the impurities on the process (initial values for the response variables and their evolution over time), the liquid phase produced and spent catalysts were thoroughly characterised by different analytical techniques. This characterisation aims to elucidate the causes of the different evolutions over time for the response variables considered. These could be a consequence of catalyst deactivation by coking, poisoning/fouling, and/or catalyst active phase or support modification (leaching and phase transformation). Specifically, elemental (CHNS) and thermo gravimetric (TG) analyses, X-Ray diffraction (XRD) were used for the characterisation of the spent catalysts while Inductive Coupled Plasma Optical Emission Spectrometry (ICP) was used for both the spent catalyst and the liquid phase.
As regards leaching, the ICP analysis of the liquid phases revealed the presence of Ni, Al and La in the liquids. Table 2 lists the relative amounts (%) of Ni, Al and La leached from the catalyst to the liquid phase during the APR reactions with respect to the original amount of these metals initially loaded in the catalyst. The proportion of metals leached from the catalyst is relatively low, especially for the amounts of Ni and Al, the amount of La being slightly higher. However, very interestingly, the statistical analysis of the results revealed that the type of acid and base do not exert a significant influence (p-value > 0.05) on leaching under the operating conditions used in this work. This is believed to be a consequence of having used the same pH to prepare the solutions employed in the experiments. These results also suggest that leaching is not responsible for the different reforming results observed in this work. This development is in agreement with our previous study  and with the results reported by other authors [32, 33] in which it was found that leaching depends on the pH of the solution.
With respect to coking and poisoning, Table 3 shows the amounts of C, K, Na, S and P deposited on the spent catalysts calculated from the elemental and ICP analyses of the used catalyst. The amount of C deposited on the catalyst surface was lower than 0.28 mg C/g catalyst g of C for all the experiments, which indicates that coke formation is minimal under the operating conditions tested in this work. In addition, although statistically significant variations are detected for the amount of C deposited on the catalysts, they are not important from a practical point of view. This suggests that catalyst deactivation by coking is not the main process responsible for the decreases observed some of the response variables over time. Similar results were reported in previous works in which this catalyst was also used for the APR of different of different organic feedstocks (crude glycerol, lactose and cheese whey) [22, 25, 26, 34]. In addition, it should be borne in mind that part of the C deposited on the catalyst can be accounted for by the formation of LaCO3OH (XRD analysis) and therefore, the amount of C quantified in the spent catalyst might include both coke and LaCO3OH formation.
Conversely, significant variations were detected in the amounts of K, Na, S and P deposited on the catalysts. The amounts of K and Na deposited on the catalyst statistically depend on the type of acid employed during the experiments and the following trend is observed: H3PO4 > H2SO4 > CH3COOH. This trend coincides with the decay observed for the carbon conversion to liquid, the glycerol conversion and the variations observed in composition of the liquid phase, which might suggest that the deposition of K and Na contribute (at least partially) to the catalyst deactivation observed in this work.
As regards S and P, a greater amount of these compounds is deposited in the experiments conducted with KOH than with NaOH. This development could explain the more pronounced decay in the carbon conversion to liquid and glycerol conversion observed for the former than for the latter base. However, the drops displayed are not proportional to the amount of S and P deposited and enormous differences are not observed. This could be the consequence of the positive effect of K on the activity of the catalyst, which can compensate for the negative effect of S and P. This is in agreement with the work of King et al. , who also reported the positive effect of K in the APR of glycerol and suggested that K could ascribe the catalytic activity making the active metal phase of the catalyst more electron-deficient, thus increasing its Lewis acidity. In addition, in a previous work using this catalyst , we also reported the positive effect of K during the APR of glycerol.
Regarding phase transformation, Figures 5 and 6 shows the characterisation results of the spent catalysts by TG and XRD, respectively. The TG analysis displays four decomposition steps (25-166 ºC; 166-247 ºC; 247-362 ºC and 362-600 ºC), which correspond to the decomposition of boehmite into alumina [35-37] involving: (I) the loss of physisorbed water, (II) the loss of chemisorbed water, (III) the conversion of boehmite into transition alumina, and (IV) the dehydration of transition alumina (loss of residual hydroxyl groups). The experimental temperature intervals for the decomposition of the used catalysts are fairly similar to those reported for pure boehmite and those reported in a previous work with the same catalyst under similar operating conditions ; the small differences being the consequence of having incorporated different metals on the catalyst structure .
From the experimental results it can be observed that the type of acid and base exert a significant influence on boehmite formation according to the TG analyses. In particular, boehmite formation in the presence of KOH is as follows: CH3COOH = H2SO4 > H3PO4; while the following trend is observed when NaOH is used: CH3COOH = H3PO4 > H2SO4, respectively. Very interestingly, similar boehmite formation occurs for CH3COOH regardless of the base used, whilst for H2SO4 and H3PO4 boehmite formation depends on the type of base. While a greater boehmite formation takes place with KOH than with NaOH for H2SO4, the opposite trend is observed for H3PO4; i.e. a greater boehmite formation with NaOH than with KOH. This development suggests a significant acid-base interaction, which can affect boehmite formation under the operating conditions used in this work. This interaction might be the consequence of the formation of different water-soluble S and P complexes with K than with Na. In all the cases, the highest boehmite formation occurs with CH3COOH regardless of the type of base. However, this development is not in good agreement with the less pronounced variations over time observed in the glycerol conversion and carbon converted to gas and liquid products for the experiments conducted with CH3COOH than for those when either H2SO4 or H3PO4 were used. This suggests that boehmite formation can partially contribute to catalyst deactivation, but it cannot explain the entire experimental trends observed under the experimental conditions tested in this work; thus indicating the existence of other deactivation mechanisms, especially when H2SO4 or H3PO4 are in the glycerol solution.
Figure 6 shows the XRD patterns of the reduced fresh catalyst and the spent catalyst obtained in each experiment. In the reduced fresh Ni-La/Al catalyst only Ni phases (Ni and NiAl2O4) are detected, suggesting a good dispersion of La on the catalyst structure, while Ni, NiAl2O4, and boehmite (AlO(OH)) are crystalline phases detected in all the spent catalysts. This finding confirms that under the operating conditions of APR, the alumina of the support can be transformed in boehmite [13, 20, 38], as described above. In addition, other crystalline phases are present in some of the used catalysts depending on the type of acid and base used in the experiments. For the experiments conducted with CH3COOH, apart from the common phases detected, the presence of LaCO3OH in the used catalyst can be confirmed regardless of the base (KOH or NaOH) used. Moreover, very similar XRD patters can be observed for the two spent catalysts, and phases of K or Na are not detected. This is consistent with the low amounts of these two elements deposited on both spent catalysts as described above. In addition, the absence of substantial differences for the amount of C deposited on the catalysts together with the greater LaCO3OH formation for the experiments conducted with CH3COOH suggest a lower coke deposition in the catalyst when feeding this impurity in comparison with that produced in the presence of H2SO4 or H3PO4. This development might contribute to explain the lower deactivation of the catalyst with CH3COOH than with H2SO4 or H3PO4.
Conversely, a higher number of new crystalline phases, such as K10La2(SO4)8 , (Na2O) 0.33 NaAlSiO4 , Na3PO4 and LaPO4, are detected for the spent catalysts used in the experiments conducted with H2SO4 and H3PO4; thus showing a greater modification of the crystalline structure of the catalyst than that observed for CH3COOH during the APR experiments. This development might also account for the lower catalyst deactivation occurred with CH3COOH than that observed with either H2SO4 or H3PO4 and it also suggests that boehmite formation is not main responsible for the catalyst deactivation observed in this work. In addition, for H2SO4 and H3PO4, the modification of the crystalline phases of the catalyst depends on the type of base used, i.e. KOH and NaOH; accounting for a more severe phase modification to take place with the later than with the former base. This also suggests that the incorporation of K on the catalyst occurs without forming new crystalline phases. Conversely, the presence of Na in the glycerol solution favours catalyst deactivation, aiding the deposition of S and P on the catalyst structure and creating new crystalline phases, thus promoting catalyst deactivation. Very interestingly, the drops observed in the evolutions over time for the carbon conversion to liquid and the glycerol conversion between the experiments conducted with KOH and NaOH for both H2SO4 and H3PO4 are not greatly different as they could have been expected according to the great difference in the amounts of K and Na deposited. This might indicate that the modification of the crystalline structure of the catalyst exerts a greater effect on catalyst deactivation than poisoning and/or fouling. Moreover, it should be borne in mind that K deposition can also promote the catalytic activity of the catalyst; thus exerting a balancing effect on the catalytic activity.
This work analyses and compares, under the same operating conditions (220 ºC and 44 bar with a Ni-La/Al2O3 catalyst and a W/mglycerol ratio of 25 g glycerol min/g catalyst), the effects of some of the most common acid (CH3COOH, H2SO4, H3PO4) and basic (KOH and NaOH) biodiesel-derived impurities on the aqueous phase reforming (APR) of a 30 wt.% glycerol solution. The most relevant conclusions are summarised as follows.
1. The impurities considered in this work do not significantly influence the APR of glycerol (before catalyst deactivation) under the operating conditions tested. Specifically, not statistically significant differences were found for the initial glycerol conversion, carbon conversion and the composition of the liquid phase produced in the experiments with the different acid and bases tested.
2. The impurities exerted a significant influence on catalyst deactivation, resulting in different evolutions over time for the CC gas, CC liq and X gly and the composition of liquid phase; the catalyst deactivation being as follows: H3PO4 (KOH = NaOH) > H2SO4 (KOH < NaOH) > CH3COOH (KOH < NaOH). The characterisation of the spent catalyst and the liquid phase revealed that poisoning/fouling, and catalyst active phase or support modification (leaching and phase transformation) were the major responsible for the deactivation of the catalyst.
3. The proportion of metals leached from the catalyst (Ni, Al and La) was relatively low and the type of acid and base employed did not affect the amounts leached. Conversely, significant variations were detected for the amounts of K, Na, S and P deposited on the catalysts. In addition, the structure of the catalyst was altered during the APR reaction: the alumina of the support was transformed into boehmite and new crystalline phases appeared.
4. The most severe catalyst deactivation took place in the presence of H3PO4 and KOH, which accounts for the greater deposition of P and K on the spent catalyst. Conversely, the lowest catalyst deactivation occurred in the presence of CH3COOH and KOH coinciding with the lowest K deposition and a high boehmite formation.