Hydrogen production from bio-fuels using precious metal catalysts

Fuel cell systems with integrated autothermal reforming unit require active and robust catalysts for H 2 production. Thus, an experimental screening of catalysts for autothermal reforming of commercial biodiesel fuel was performed. Catalysts consisted of a monolithic cordierite substrate, an oxide support (  -Al 2 O 3 ) and Pt, Ru, Ni, PtRh and PtRu as active phase. Experiments were run by widely varying the O 2 /C and H 2 O/C molar ratios at different gas hourly space velocities. Fresh and aged catalysts were characterized by temperature programmed methods and thermogravimetry to find correlations with catalytic activity and stability.


Introduction
According to the current policies scenario, by 2030 the global CO 2 emissions will be more than 50% higher than today.In addition, the world population is supposed to grow to more than 9 billion of people in 2050.As a consequence, the energy demand strongly increases, especially driven by China and India.The Institute of Energy and Climate Research (IEK-3) at Forschungszentrum Jülich has been developing a two-step approach to contribute to facing these environmental problems.On the one hand, efficient fuel cell systems based on the autothermal reforming of diesel fuels have been developed, which can be used as auxiliary power units for efficient on-board power supply in aircraft or long-haul trucks, for example.On the other hand, a renewable Bio-To-Liquid (BTL) fuel is used as feedstock for these systems.

Biodiesel fuel synthesized by hydro-treating of triglycerides
Transesterification is one of the most important routes for the production of bio-fuels [1][2][3].During this process, triglycerides from vegetable oil or animal and plant fats are converted with alcohol molecules (methanol or ethanol) yielding mono-alkyl esters and glycerol.The mono-alky ester molecules then make up the bio-diesel.Another option for the production of bio-diesel is hydro-treating.This process can be run in existing petroleum refineries [4].Figure 1 shows a simplified scheme for the conversion of triglycerides contained in Jatropha oil, algae oil or plant and animal fats to straight-chain and branched alkanes [4].In the first two steps of this reaction scheme, the triglyceride molecule is hydrogenated and split into a number of intermediates which can be monoglycerides, diglycerides and carboxylic acids.These intermediates are then converted into alkanes on three different routes: decarboxylation, decarbonylation and a combination of hydrogenation with dehydration.Reaction equations ( 1)-( 3) show exemplarily the different routes for alkane production starting from carboxylic acids.Decarboxylation involves converting a carboxylic acid into a methyl group and CO 2 as shown in equation (1).CO is produced by decarbonylation where the carboxylic carbon is reduced by hydrogen to yield a methyl group, CO and water appear as side products, as shown by equation (2).During the combination of hydrogenation with dehydration the carboxylic acid reacts with hydrogen to produce an alkane and water as shown by the third equation [4].
For the investigations in this work, so-called NExBTL diesel fuel produced by the Finnish company Neste was used as fuel for the autothermal reforming reactor.NExBTL diesel fuel has a sulfur mass proportion of less than 1 ppm and a fraction of aromatics far lower than 1 vol%.
Figure 1 Simplified scheme for the conversion of triglycerides contained in in Jatropha oil, algae oil or plant and animal fats to straight-chain and branched alkanes [4]

Catalyst preparation
The catalysts were synthesized in mass fractions related to the cordierite substrates, whereby for the oxide support and the active metal phase approx.10 wt.% and 1 wt.%, respectively, were obtained.
Each value referred to the total mass of the catalyst.The synthesis of the support -Al 2 O 3 was identical to that described in previous work [5] except for the temperature, at which the impregnated cordierite substrate was dried after impregnation.This temperature was decreased from 110 °C to 80 °C in order to reach a more homogeneous coating of the support on the substrate.For the synthesis of the

Results and discussion
In the above mentioned previous work by our group [5], it was demonstrated that among the   showing no sign of deactivation.However, it can be concluded from the first three sequences that both catalysts, Ru/-Al 2 O 3 and PtRu/-Al 2 O 3 , were also not suitable for efficient autothermal reforming of BTL diesel fuel.The observed H 2 concentrations were simply too low compared to those given in the literature.This would lead to poor H 2 -productivity of the reforming step within a fuel cell system and thus also significantly reduce the overall efficiency of the whole fuel cell system.Also during the second and third sequence the H 2 concentrations of the bimetallic catalyst PtRh/-Al 2 O 3 were highest among the investigated samples and varied between 34 and 36 vol% at most favorable H 2 O/C molar ratios between 2.3 and 1.9.Under these reaction conditions these values were even higher than those of the reference catalyst Rh/-Al 2 O 3 , which was considered suitable for autothermal reforming of diesel fuels in previous work [5].So, from the first three sequences with widely varying H 2 O/C molar ratios it can be concluded that the bimetallic catalyst PtRh/-Al 2 O 3 from this work is promising for the autothermal reforming of BTL diesel fuel.It showed a high initial catalytic activity and no detectable signs of deactivation.2 vol%.In Figure 3 on the right the H 2 concentrations of the bimetallic sample PtRu/-Al 2 O 3 and the monometallic catalyst Ru/-Al 2 O 3 were very similar.They started at approx.20 vol% and decreased to 14 vol% at an O 2 /C molar ratio of 0.40.Obviously, a significant deactivation of these samples occurred under the rough conditions of the fourth sequence.As can be seen in Figure 3 in the middle and on the left, the deactivation of the catalyst Ru/-Al 2 O 3 proceeded.It revealed H 2 concentration gradients between 20 and 12 vol% at a GHSV of 41,000 h -1 and between 14 and 12 vol% at a GHSV of 27,000 h -1 , respectively.The H 2 concentrations of the sample Pt/-Al 2 O 3 remained at their very low level in the range of 3 to 2 vol%.In both figures (middle and left), a slight deactivation of the bimetallic catalyst PtRh/-Al 2 O 3 could be observed.H 2 concentrations slightly decreased from the range of 34 to 33 vol% to the level of 32 vol% and 31 to 32 vol%, while the concentrations of the reference catalyst remained constant.Nevertheless, catalyst PtRh/-Al 2 O 3 confirmed the conclusion drawn from the results of the first three sequences that it is a promising candidate for the autothermal reforming of BTL diesel fuel with high initial catalytic activity and only marginal signs of deactivation under rough reaction conditions.
In the following, the reducibility of the catalysts from this work will be analyzed with the aid of measurements from temperature-programmed reduction.Figure 4  and 900 °C, which can attributed to the formation of solid-state solutions of nickel-aluminum-oxide [6].
This signal has a comparatively high intensity, which points at a high reducibility of the fresh sample.
The finding stands in sharp contrast to the very low catalytic activity of Ni/-Al 2 O 3 for the autothermal reforming of NExBTL diesel.The profile b of the catalyst Pt/-Al 2 O 3 does not show any reduction signal at all.This means that this sample does not possess any reducibility, which can be detected by temperature programmed reduction, and thus that it is only marginally able to catalyze the autothermal reforming of NExBTL diesel fuel by adsorbing and releasing electrons.This result is well consistent with the comparatively low initial catalytic activity of Pt/-Al 2 O 3 .In the literature [7], the reduction of platinum oxide is reported at a temperature of approx.250 °C.In the case of the catalyst Ru/Al 2 O 3 , the TPR profile c shows a weak shoulder at 105 °C and a more pronounced peak at 145 °C.In the literature [8], the weak shoulder is assigned to the conversion of RuO 3 to RuO 2 , while the main peak is explained by the reduction of RuO 2 to elementary Ru.In the profile e of the fresh catalyst PtRu/-Al 2 O 3 very similar signals for the consumption of H 2 were detected at temperatures of 103 and 139 °C, respectively.As in the case of catalyst Ru/-Al 2 O 3 , they can be explained by the conversion of RuO 3 to RuO 2 and the reduction of RuO 2 to elementary Ru [9].No interaction between Ru and Pt particles possibly forming a new joint phase was detected.This is in accordance with results from the literature [10].The overall intensity of both peaks of PtRu/-Al 2 O 3 in profile e is slightly lower than that of the monometallic sample Ru/-Al 2 O 3 in profile c.That means that the addition of Pt to the Ru-based catalyst lowered the reducibility in the fresh state.This graduation is in good accordance with the initial H 2 concentrations of these both catalysts showing initial values of 31 vol% (Ru/-Al 2 O 3 ) and 27 vol% (PtRu/-Al 2 O 3 ), respectively.The largest consumption of H 2 during the TPR experiments with the fresh catalysts was observed in the case of sample PtRh/-Al 2 O 3 .In profile d two very pronounced signals at temperatures of 83 and 150 °C were identified, which can be attributed to the reduction of a threedimensional Rh i O x phase or a dispersed two-dimensional Rh i O x phase on -Al 2 O 3 [11].The overall intensity of these signals is even significantly higher than that of the signals of the monometallic catalyst Rh/-Al 2 O 3 (cf.profile f).Also in this case no interaction between the two metals Pt and Rh forming a joint phase could be detected in profile d.The observed outstanding reducibility of the fresh sample PtRh/-Al 2 O 3 is in good accordance with its very high initial catalytic activity with H 2 concentrations in the range of 36 to 35 vol%.Figure 4  difference is that for the aged catalyst the peak at 98 °C is less pronounced than it was in the case of the fresh catalyst PtRh/-Al 2 O 3 .For the catalyst Ru/-Al 2 O 3 a significant decrease in the overall intensity of the weak shoulder and the main peak could be detected after the sample was run in the evaluation pattern indicating a loss in reducibility.This is in good agreement with the observed deactivation of this catalyst in the course of the evaluation pattern.However with all other catalyst, the overall intensities of fresh and aged samples were comparable.The rough reaction conditions applied during the evaluation pattern did not significantly lower the reducibility of the active phases of these catalysts.This is in contradiction to the deactivation phenomena, which were observed with the   [12] found in their work a combustion temperature of coke on Rh/-Al 2 O 3 of 600 °C.Therefore, it can be derived that the observed mass losses are due to the burn-off of carbon-containing components from the surface of the catalysts which were formed during experiments on autothermal reforming.Forzatti and Lietti [13] describe how the carbon-containing deposits diminish the catalytic activity, sometimes strongly, since they accumulate on the active catalyst centers so that the latter are no longer accessible for the reactants in autothermal reforming.For the catalysts Pt/-Al 2 O 3 and PtRu/-Al 2 O 3 from this work, the detected carbonaceous deposits can explain the observed loss in catalytic activity of these samples during the evaluation pattern.The thermogravimetric investigation of the catalysts Ru/-Al 2 O 3 and PtRh/-Al 2 O 3 showed only minor carbon-containing deposits on the catalysts' surfaces.Therefore, correlations between these findings and the catalytic activities of the catalysts cannot be derived.

Conclusions
Of the samples tested, PtRh/-Al 2 O 3 was found to be the most active and robust catalyst for the autothermal reforming of NExBTL diesel fuel.H 2 concentrations were in the range of 35 -36 vol% and decreased only slightly when the evaluation pattern was applied.In contrast, Pt/-Al 2 O 3 showed the strongest deactivation.Ni/-Al 2 O 3 was not active at all.As described above in detail, decreased reducibilities of the active phases and coke depositions on the catalyst surfaces can be adduced to explain the deactivation phenomena observed with the catalysts from this work.

2 E3S
Figure 2 shows H 2 concentrations of catalysts Pt/-Al 2 O 3 , Ru/-Al 2 O 3 , Ni/-Al 2 O 3 , PtRh/-Al 2 O 3 and PtRu/-Al 2 O 3 as a function of the H 2 O/C molar ratio at an O 2 /C molar ratio of 0.47 and different GHSVs.NExBTL diesel fuel was used.For means of comparison, the corresponding trend of the H 2

Figure 2 H
Figure 2 H 2 concentration on all six catalysts as a function of the H 2 O/C molar ratio, O 2 /C molar ratio = 0.47, at different mean GHSVs, NExBTL diesel fuel

Figure 2
Figure 2 in the middle and on the left show the results of the second and third experimental sequence with respect to H 2 concentrations of the five investigated catalysts and the reference sample.The catalysts were not replaced between the first and second or between the second and third sequence.This means that the series of sequences run with each catalyst can be considered an accelerated aging experiment.It can be seen that during sequences 2 and 3 the H 2 concentrations of the catalyst Pt/-Al 2 O 3 further decreased to values between 9 vol% and only 3 vol%.An additional deactivation of this catalyst for autothermal reforming occurred.As a consequence of the deactivation step from the first sequence, the H 2 concentrations of the monometallic catalyst Ru/-Al 2 O 3 were always lower than those of the bimetallic sample PtRu/-Al 2 O 3 during the second and third sequence.At a GHSV of 44,000 h -1 , they were almost constant at 22 -23 vol% and thus in the same order magnitude as after

Figure 3 H
Figure 3 H 2 concentration on all six catalysts as a function of the O 2 /C molar ratio, H 2 O/C molar ratio = 1.9, at different mean GHSVs, NExBTL diesel fuel on the left shows the profiles from the temperature programmed reduction of the fresh catalysts Ni/-Al 2 O 3 , Pt/-Al 2 O 3 , Ru/-Al 2 O 3 , PtRh/-Al 2 O 3 and PtRu/-Al 2 O 3 , before they were subjected to the evaluation pattern.Again for comparison reasons, the corresponding TPR profile of the reference catalyst Rh/-Al 2 O 3 is added.The profile a of fresh catalyst Ni/-Al 2 O 3 shows a very broad signal in the temperature range between 300 on the right shows the profiles from the temperature programmed reduction of the aged catalysts Ni/-Al 2 O 3 , Pt/-Al 2 O 3 , Ru/-Al 2 O 3 , PtRh/-Al 2 O 3 and PtRu/-Al 2 O 3 , after they were subjected to the evaluation pattern as described above.A comparison of the figures on the left and on the right makes obvious that in any case the profiles of the fresh and aged catalysts are very similar with respect to the position of the respective peaks.Profile a of the aged catalyst Ni/-Al 2 O 3 again shows the broad signal between 300 and 900 °C.Again, no H 2 was consumed by the catalyst Pt/-Al 2 O 3 .In the profiles c and e of the catalysts Ru/-Al 2 O 3 and PtRu/-Al 2 O 3 the weak shoulder at approx.130 °C and the more pronounced peaks at 160 and 145 °C, respectively, can be rediscovered.Again as in the case of the fresh samples, the overall intensity of the two peaks of the aged sample PtRh/-Al 2 O 3 (98 and 144 °C) is by far highest.The only