Final Report Summary - SUSFUELCAT (Sustainable fuel production by aqueous phase reforming – understanding catalysis and hydrothermal stability of carbon supported noble metals)
Executive Summary:
SusFuelCat aims to boost Europe’s expertise on catalysts for sustainable fuel production, especially hydrogen, by the process of Aqueous Phase Reforming (APR). APR is a technology that allows the catalytic conversion of low value biomass streams into biomass based fuels, with catalysts being the key stone in the process. That’s why SusFuelCat aims to gain fundamental material structure-property relationships for the catalyst using model supports and metal nanoparticles.
To deduce fundamental structure-property relationships the properties of catalysts were varied with respect to their active metal, cluster size, pore size, carbon graphitization. Additionally, in-silico studies for theoretical investigations were carried out. Catalyst activity and selectivity were studied experimentally with real- and model-feedstocks to give feedback on the catalytic performance. The hydrothermal stability of the carbon support materials and later on of the catalysts were studied. These results were used for the model catalyst design in the subsequent iteration loops. Throughout the project the economical viability of the envisaged process was monitored.
The project results showed that proper APR catalysts for renewable hydrogen production could be obtained and that a new platform of model catalysts is available now, which can be used for noble metal on carbon catalyst development in general. The experimental data and simulation allowed deducing further mechanistic insights, which are important for the catalyst and process design. Cost estimations show that APR hydrogen can be economic competitive, if the right feedstock is employed. Thus, it is believed that the developed APR technology can be of importance.
Besides this direct impact on the horizon, already further impact becomes apparent for each partner. The pronounced knowledge on well-controlled colloidal based catalysts, methodologies for immobilization and stabilization agent removal, synthesis of high quality carbonaceous materials, great progress on the simulation of the systems and developed experimental protocols have the potential to be of great us in catalysis and neighbouring disciplines in general.
Project Context and Objectives:
Energy production and supply is one of the most important basics for a high quality of life, industrialization and civilization. Europe’s wealth and leading position in industry depends heavily on energy, for which fossil fuels remain the primary source. In addition to the huge dependency on (mainly non-European) fossil fuel exporting countries, and unstable prices, fossil fuels are a limited resource whose use releases carbon dioxide to the atmosphere, a major contributor to global warming. Hence, economically and environmentally European based sustainable energy production will become ever more pertinent.
For biomass in particular, the energy consumption of the conversion process needs to be as low and the biomass utilization as efficient as possible in order to increase the energy output per square meter of cultivable land. In this project we focus on Aqueous Phase Reforming (APR) of biomass resources. Regarding this the products aimed by APR, viz. hydrogen or, preferably, a combination of hydrogen and alkanes, are of high potential for sustainable fuel production. The sustainably produced hydrogen can be used either as fuel directly or is mandatory in bio-refineries for sustainable liquid fuel production. A major advantage of APR is the use of aqueous or water-soluble feedstocks at reduced temperatures and slightly elevated pressures, enabling processing without energy intensive drying while hydrogen can be in situ generated from biomass by the APR reaction and in parallel from the water content by the Water-Gas-Shift (WGS) reaction. Hence, APR is one of the most promising techniques for sustainable liquid and gaseous fuels production from biomass feedstocks, with catalysts being the key stone in the process.
The versatility of the process, the limited energy consumption and the product flexibility address the crucial challenges of sustainable fuel production from biomass and thus APR can be seen as a major technology to allow rapid industrialization of sustainable fuel production in Europe. In parallel to the interest in sustainable produced hydrogen as fuel, this is also a highly sought commodity, for example to upgrade bio-based mixtures to high quality fuels or chemicals.
A key towards of any successful implementation of APR is catalysis. State-of-the-art catalysts used in lab scale APR have already demonstrated the potential of catalytic APR, but a) are not yet optimized, due to their unknown structure-property relationships, b) require high amounts of costly noble metals (> 5 wt.%) and c) their hydrothermal stability is shown to be limited. For a European APR-based sustainable fuel production to be viable, key objectives of catalyst optimization are a gain in energy efficiency of the process, reduction of total costs of catalyst and independency on material imports for catalyst production. Regarding the latter, the paradigm shifts towards carbon supported catalysts, due to their superior hydrothermal stability. Current research on APR-catalysts is rather unstructured and diversified, hence recommendations to avoid such drawbacks are not clearly presented. In addition, a criterion is to lower the amount of costly noble metal. For all metals a considerable recycling industry is available, so that limited technical uses would not be considered a threat for the existing supplies. However, a considerable use in APR could increase market prices for precious metals. For example, catalysts (mostly automotive applications) already account for one third of annual platinum production. Aqueous phase reforming may be applied in similar smaller units, to allow use at remote locations/close to biomass resources which may mean that total metal inventories are small, at the same time rendering recycling more costly. The use of carbon supports, which are focused on in SusFuelCat, also allows easy and more environmental friendly metal recycling by burning of the carbon support. The high metal content residues can then be easily converted to new catalysts.
The two primary objectives of the project are:
a) To unleash the potential of carbon supported catalysts to establish APR as an energy efficient process to convert different biomass based streams to sustainable fuels and to decrease time to market for a commercial APR catalyst.
b) To generate data to allow the design and scale up of the APR process towards a further demonstration level. The final target is to show the technical and economic viability of the global process.
To achieve this, fundamental material structure-property relationships for the catalyst should be gained. Model supports and metal nanoparticles play a major role for this aim, thus studying how properties of catalysts are varied with respect to their active metal, cluster size, pore size, carbon graphitization will be carried out. These materials form the basis for experimentally catalytic studies, which will be evaluated together with the results from the material characterization and additional in-silico studies. Furthermore, the information obtained shall be used continuously to evaluate the economic viability of the envisaged process.
Thus, the two primary objective can be broken down the following R&D targets:
Nanoparticle synthesis
- Avoiding the undefined influence of the support during nanoparticle deposition leading to broad cluster size distributions through synthesizing well defined colloidal solutions of nanoparticles. For the various metals to study (Ru, Re, Pt, Ni, Pd, Co) synthesis protocols for the polyol or micro-emulsion technique should be developed.
- Studying the influence of upscaling the colloidal synthesis method on the cluster sizes.
- Establishing methodologies for immobilization of these well defined nanoparticles in supports, either through post immobilization or in situ synthesis.
- Establishing methodologies for efficient removal of the capping agents employed in colloidal solutions and its influence on the cluster size
- Comparison of “colloidal” synthesis routes to classical impregnation routes.
Carbon support synthesis
- Production of model porous carbons with inner porosity varying in pore size and graphitization, which are needed as basis for model catalysts to deduce structure-property relationships.
- Production of carbon nanofibers as material with external surface and possibility for superior stability.
- Detailed characterization of materials properties.
- Comparison of model carbon to commercial activated carbons and carbon blacks.
in-silico studies
- individuating the best performing approaches QM, MD, MC or hybrid
- calibration of methods
- realizing metals as clusters or periodic facets
- studying how metal clusters form on carbon supports
- studying sorption processes
- deducing possible reaction pathways
Hydrothermal stability
- Testing hydrothermal stability of different model and commercial carbons (no Pt).
- Testing possibilities to increase hydrothermal of carbon supports.
- Testing hydrothermal stability of catalysts (with Pt).
Catalytic experiments
- Screening of activity and selectivity for different active metals and supports
- Handling studies for different sized supports
- Deduce proper conditions and protocols for APR catalyst testing.
- More detailed studies on feedstock influence (diastereomer influence, C6 sugar alcohols, C5 sugar alcohols, C3 sugar alcohols and acids).
- Detailed study on residence time influence to obtain selectivity vs. degree of conversion
- Obtaining database for reaction network analysis
- Obtaining database for kinetic modeling and flow sheeting
- Carrying out long term experiments to study stability for selected catalysts
- Studying reactivation protocols for deactivated catalysts
Process validation
- Continuously accompanying the development through assessing the catalyst production costs and hydrogen production costs through APR
- Continuously refining the cost calculations
- Providing process insights through chemical reaction engineering simulations to study possibilities how to increase the overall yield.
- Providing process insights through flow sheeting to study possibilities how to increase the economics of the overall process.
Project Results:
Subsequently the main results for the different targets are presented in a concise manner.
Main results on nanoparticle synthesis
Regarding metal model material nanoparticles (NP)s, they were synthesized as colloids by colloidal and the polyol methods and also by microemulsion method with various capping agents and surfactants for six metals. In total more than 400 syntheses were carried out, including a broad screening of reaction systems and reagents and characterization to assess structure. Particle size variations within the following boarders could be obtained: Pt (2.5 – 8.3 nm), Pd (3 – 13 nm), Re (0.7 – 1.4 nm), Ru (1.6 – 4.5 nm), Co (1.8 -18.1 nm) and Ni (2.6 – 9.8 nm). Thereby narrow particle size distributions could be achieved and the sizes are within the interesting range for catalysis. The polyol method was applied for synthesis of Co, Ni and Ru metal NPs. As a stabilization agent polyvinylpyrrolidone (PVP), dodecylamine (DDA), polyethylene glycol (PEG) and 3-mercaptopropionic acid (3-MPA) were employed. TEM, XRD, UV-Vis spectroscopy and XPS were applied to characterize these NPs.
The Pd colloids were prepared using PVP as caping agent and ethanol and methanol as reducing agents. The nanoparticle size was found to decrease with increasing alcohol concentration and PVP dose. The water/AOT/isooctane system was selected for microemulsion synthesis in combination with N2H4 and NaBH4 as reducing agents (Colloids and Surfaces A 2016, 497, 28) (Perez-Coronado et al., 2016). Other systems evaluated for microemulsion synthesis where those based on cetyltrimethylammoniumbromid (CTAB) and Triton X-114 surfactants. The water-to-AOT molar ratio (w0) was the most influencing factor. Smaller nanoparticle size was obtained for low ratios, although higher amount of AOT remained on nanoparticles even after purification. Lower agglomeration and higher hydrophilicity of nanoparticles was observed afterwashing with methanol. Pd nanoparticle size was in a broad range for colloidal (3-13 nm) than for micoremulsion synthesis (6-10 nm).
The preparation of Pt nanoparticles Pt using PVA and PVP-based colloidal synthesis with methanol and NaBH4 showed an influence of synthesis variables similar to that of Pd, although lower tuneability of nanoparticle size (3-6 nm) was achieved . Narrower size distributions were obtained with NaBH4. The microemulsion synthesis with the water/AOT/n-heptane system provided witha a broader range of NPs mean size (1-7 nm). A better tailoring of the properties of the Pt nanoparticles was achieved varying the type of reducing agent, the reducing agent/Pt ratio and water/AOT ratio. The Re colloids were synthesized by simple colloidal (PVP) and microemulsion methods (water/AOT/isooctane) (Bedia et al., 2015) (Colloids and Surfaces A 2015, 469, 202). A strong reducing agent (NaBH4) and hydrogen atmosphere was needed to achieve more complete reduction. Reduction rate was much affected by temperature, although above 60ºC fast reoxidation was observed. The main tool to adjust nanoparticle size was the addition rate of the reducing agent and the PVP to Re molar ratio, however a narrow range of mean size was achieved (0.7–1.4 nm).
Co, Ni and Ru nanoparticles were prepared by reduction of the corresponding chloride precursors in ethylene glycol (10 g/L) (reductant for Ru) using PVP as a stabilizer as well as sodium borohydride as a reductant in the temperature range from 170 to 240oC for Ru NPs, from 23 to 170°C for Ni NPs and from 7 to 140°C for Co NPs. In addition to a typical polyol technique, Ru NPs synthesis using microwaves along with sodium borohydride reduction at room temperature was applied as an effective alternative approach. The effect of gas atmosphere, reduction temperature, the ratio metal/PVP, as well as of metal concentration in the colloidal solution, was investigated to explore regularities of metal NPs formation as well as to develop an approach to the synthesis of metal NPs with a controllable size. Reduction of Co, Ni and Ru precursors was found to be sensitive to air and had to be carried out under an inert atmosphere. A special attention was paid to a possibility of increasing Ru/PVP ratio compared to the classical polyol methods used Ru/PVP = 1/20-1/50 in order to simplify PVP removal after the final deposition of Ru NPs on the catalyst support. Decrease of reduction temperature and Ru/PVP ratio from 1/1 to 1/50 resulted in slight decrease of NPs size from 2.1 to 1.7 nm with a narrower size distribution while no clear effect was observed when the initial Ru concentration increased from 0.001 to 0.1 mol/L (Int. J. Nanotechnol. 2016, 13, 15) The oxidative atmosphere and increase of the reduction temperature was found to increase Ni and Co NPs size favoring NPs growth versus nucleation and clearly indicating different kinetics of NPs formation in the case of base and noble metals. Thus, the size of Ni NPs formed at 140°C in air was 5.4 nm compared to 3.2 nm for that formed in Ar. Co NPs synthesis required only inert atmosphere due to low stability and a fast reoxidation. The average diameter of Ni NPs size increased from 2.6 to 3.2 nm when the reduction temperature increased from 23 to 140°C, when the temperature was increased to 180 C the Ni NPs size increased to 9.8 nm. Similarly, Co NPs size increased from 1.8 to 2.6 nm with the temperature was increased from 7 to 23°C, whereas further changes from 23 to 100°C affected less significantly the NPs size (Int. J. Nanotechnol. 2016, 13, 4).
The microemulsion method was applied for Co and Ni NPs synthesis. As a stabilization agent and emulsion stabilizer dioctyl sulfosuccinate (AOT), cetyltrimethylammoniumbromid (CTAB) and Triton X-114 were employed. Microemulsion approach was studied in detail to establish the operating parameters for size-controllable synthesis of Ni and Co NPs. Effect of the metal precursor nature, the oil phase, the surfactant, the order of microemulsion blending and injecting, as well as water-to-surfactant ratio was explored from the viewpoint of practical synthesis. Colloidal Ni and Co metal NPs were prepared by the reversed microemulsion method using metal chloride and nitrate aqueous solutions (microemulsion I) and hydrazine hydrate as reductant (microemulsion II) in the presence of CTAB or AOT as well as n-octane and n-butanol or n-hexanol as oil phases. Mixing of two microemulsions and heating up to 80оС resulted in formation of series of Ni and Co nanoparticles with mean sizes varied from 2.6 to 3.3 nm and from 1.0 to 5.4 nm, respectively. Utilization of cationic CTAB provided formation of Ni and Co NPs with a narrow size distribution while anionic AOT resulted in formation of metal hydroxide instead of NPs according to UV-Vis. Increase of the heating time and decrease of the water-to-oil ratio gave a rise of NPs size. Probably solvent evaporation involved a drastic NPs aggregation process even if initially reversed micelles were stabilized in the solution as rather small NPs as a consequence of the intermicellar material exchange.
The scaling up of the lab synthesis routes was evaluated with the view of preparing large batches for partners and providing input for the evaluation of the synthesis routes at industrial scale. The first approach to scaling-up was to increase the reaction volume for colloidal synthesis to 500 mL, which was accomplished with excellent reproducibility of nanoparticle size and distribution for Pd, Pt and Re. Additional scaling-up was achieved in the case of the colloidal synthesis of Pt nanoparticles by increasing the concentration of precursor salt and PVP. Next efforts were centered in the reduction of the reaction volume employed to obtain the Pt colloids by increasing the concentration of the stock solutions. Reductions in the reaction volume of around 68% were achieved with only slight variations in the mean NP size (<10%) and in the NP size. The high viscosity of concentrated PVP stock solutions was a limiting factor for additional reduction of reaction volume. Furthermore, scaling up of Ru NPs polyol synthesis was studied both through the preparation of more concentrated Ru colloids or increasing the volume of the synthesized colloid. For Ru colloids with Ru/PVP = 1/5 it was found that 10-fold increase of Ru(3+) concentration from 0.01 to 0.1 M resulted in 1.3 fold increase of NPs size (from 2.1 to 2.8 nm), while at Ru/PVP = 1/10 even 100-fold increase of Ru(3+) concentration from 0.001 to 0.1 M colloid resulted in unproportional 1.5 fold NPs size increase (from 1.7 to 2.6 nm). Furthermore 20-fold increase of Ru colloid volume from 5 to 100 mL gave an increase of the average NPs size from 2.1 to 2.7 nm and from 1.7 to 2.6 nm for Ru/PVP = 1/5 and 1/10, correspondingly. Thus, it was demonstrated that required catalytically active metal NPs with controllable size can be prepared in bench scale quantities with a high metal/PVP ratio suited for further immobilization of prepared colloids on carbon supports. (Catal. Sci. Technol. 2016, 6, 8490).
The stability of Pd, Pt and Re nanoparticles obtained by colloidal synthesis was studied by aging of colloids and monitoring. Re colloids were found to be highly unstable and showed a tendency to reoxidize few hours after synthesis. Pd colloids showed a slight increase of the nanoparticle size 2-3 weeks after preparation, with an increase of about 11-13 % after four months. This tendency was more pronounced for the colloids with nanoparticles of higher size. In the case of Pt no significant increase in nanoparticle size was observed after an ageing time of two months. Colloidal solutions containing Ru, Pt and Pd NPs kept in fridge during 1-2 years were analyzed by TEM. It was shown that only small changes in NPs mean size and particles size distribution occurred during these periods indicating that developed methods of colloidal synthesis provided high nanoparticles stability. Thus, for PVP-capped Ru NPs average particle size dn was changed from 2.3 to 2.4 nm in 2 months, and from 2.3 to 2.7 nm in 24 months, for Pt-MeOH50-PVP10 the average particle size dn practically did not change in 18 months being 3.4 nm, for Pt-NaBH4 10-PVP 100 dn increased from 2.3 to 2.6 nm in 18 months, and for Pd-PdV3 dn changed from 2.0 to 2.1 nm in 20 months.
The immobilization of these well-defined nanoparticles on porous carbons and removal of the stabilization agent was studied in detail. The nanoparticle colloids were impregnated on different supports using water and methanol. The resulting catalysts exhibited poor surface area and low CO chemisorption, due to the blocking effect of the remaining PVP or AOT. Stability tests in water at room temperature showed some leaching of nanoparticles for all the supports (commercial activated carbons, CNFs, CDCs). The solvent used during impregnation and the chemical surface of the support influenced stability. Additional stability was achieved through a novel synthesis method of Pt/C catalysts with size-controlled nanoparticles based on in-situ synthesis of the nanoparticles, i.e. reduction with NaBH4 in the presence of a support and PVP (Catal. Sci. Technol. 2016, 6, 5196). Compared to the conventional ex situ route (colloidal synthesis followed by impregnation), this in situ route also yielded smaller nanoparticles (2.5–3.9 nm) of narrower size distribution. The method was also applicable to Pd/C catalysts synthesis. PVP-stabilized Ru NPs were immobilized on a mesoporous carbon Sibunit, applied as a reference, macroporous carbon nanofibers of platelet structure (CNF-Pl) and micro-/mesoporous TiC carbide-derived carbon (CDC) providing 1.7÷2.9 wt.% Ru/C catalysts with the mean Ru size 2.1÷2.7 nm. Since the presence of PVP on the catalyst surface drastically diminished activity in structure sensitive galactose to galactitol hydrogenation, different PVP removal and support modification methods were tested to elucidate the effect of support hydrophilicity/ hydrophobicity, preliminary support functionalization, additional Ru NPs washing prior to immobilization as well as PVP removal degree on catalytic behavior. Several approaches for PVP removal were applied such as solvothermal (with water and acetic acid aqueous solutions at 220°C) and thermal post-treatment (in air, argon, hydrogen or nitrogen at different temperatures) not resulting in noticeable changes in metal NPs size. For characterization of the carbon supports and synthesized catalysts TEM, XPS, XRD, XRF, water adsorption/desorption experiments and N2 physisorption were applied. Carbon supports were functionalized before Ru NPs immobilization by pretreatment with 5 wt.% HNO3, HNO3 conc., Ar (700oC), air (350oC), H2 (700oC) without visible support microstructure alterations. Textural properties of Ru/C catalysts purified from PVP did not correlate with their catalytic activity. The thermal post-treatment in air at 180°C followed by reduction at 250°C was found to be more effective in the case of Sibunit and CNF-Pl while solvothermal post-treatment in water at 220°C (PN2 25 bar) improved significantly the activity of TiC-CDC based catalysts. Untreated Sibunit and CNF-Pl carbon supports provided higher activity in galactose hydrogenation with Ru/Sibunit exhibiting the best catalytic activity being also the most hydrophilic according to water sorption isotherms. Catalytic activity of untreated micro-/mesoporous Ru/TiC-CDC increased noticeably depending on the support pretreatment in the row: untreated < 5% HNO3 < HNO3 conc. Additional TiC-CDC support functionalization was proposed to be required because of a relatively low amount of oxygen-containing groups on the surface compared to Sibunit and CNF-Pl. Carbon supported PVP-capped Ni and Co NPs were purified from PVP by both chemical washing (water or aqueous solutions of acetic acid) procedure and thermal post-treatment resulted in significant metal leaching, oxidation or sintering and thereby catalysts deactivating (Catal. Sci. Technol. 2016, 6, 8490).
Removal of PVP and AOT blocking the catalyst pores and the active surface was achieved by a novel method (Lemus et al., 2016) based in the activation of the catalysts at APR conditions (water at 200ºC, water+acetic acid at 200ºC). Metal loss was negligible for Pt/C catalysts while PVP was almost completely removed; hence, most of the porosity was recovered and the dispersion measured by CO chemisorption increased from 5 to 34–75%. Water at 200 °C was more effective than diluted acetic acid for the removal of PVP. The removal of the PVP at APR conditions also increased the stability of the catalysts due to improved metal-carbon interaction, showing negligible leaching of the metal phase. Likewise, the Pt nanoparticles did not undergo significant changes either in size or morphology. In the case of Pd this novel method could not be applied successfully, as sintering of the metal nanoparticles occurred and some leaching was observed. Massive leaching was observed in the case of Re based catalysts at APR conditions. The removal of AOT was also achieved by treatment at APR conditions, as evidenced by the recovery of the porosity of the catalysts, but CO chemisorption tests evidenced poisoning/blocking of the metal phase. The studies showed that with specialized posttreatments a high accessibility of the active sites can be combined with a high stability of the immobilized nanoparticles. Without this developed posttreatment active and stable catalysts cannot be obtained.
Furthermore, for the comparison several variations of catalysts were prepared by classical methods and on reference carbon material.
Main results on carbon support synthesis
Carbon supports with high inner porosity were synthesized from the partner FAU by the reactive extraction of carbides (CDC, carbide-derived carbons). Microporous material with pores below 1 nm were obtained from titanium carbide with extraction temperatures below 1000 °C. The specific surface area for this microporous and amorphous carbon is approx. 1600 m2/ g. Two different feeds of TiC were studied with approx. 2.5 μm and 50 μm average particle size. Interestingly, the resulting pore sizes differed slightly. Analysing the results more carefully reveals that the temperature onset for increasing pore size and crystallinity differs. Thus at 1200 °C the resulting pore diameter is 1.68 nm for the 2.5 µm fraction and 1.14 nm for the 50 µm fraction. A possible heat transfer limitation, which could lead to this phenomenon could be excluded. Thus, most probably different amounts of metals impurities (e.g. iron) are responsible for the difference.
To obtain mesoporous carbons it was envisaged to employ Mo2C as precursor. It was demonstrated that highly mesoporous materials with pore sizes starting at 4 nm are obtained when extracting Mo2C at 1200 °C. Nevertheless, these materials are not suitable for use in the experimental testing as the commercial raw material didn’t show the requested particle size. The resulting catalysts would lead to too high pressure drops. To obtain the envisages mesoporous carbons in correct particle size, the extraction temperature above 1200 °C was studied for the first time and for titanium carbide. Importantly mesoporous materials, with a narrow pore size distribution around 3.7 nm and a specific surface area of 450 m2/g resulted from titanium carbide at 1400 °C (Chem. Mater. 2015, 27, 5719). At lower temperatures the pore size can be varied in a bimodal micro- and mesoporous distribution. In parallel with increasing temperature the materials crystallinity changes from amorphous to more graphitic in the new temperature regime for 1200 to 1600 °C, resulting in graphite crystals approx. 20 nm in diameter and above 10 stacks of graphene. The porous carbon inner surface could be functionalized with oxygen by different post treatments. Thereby the amount of oxygen and the change in pore structure increases for the different reactants in the following order: air < sulfuric acid < nitric acid. As a recommendation the sulfuric acid treatment is suitable for amorphous and the nitric acid one for more graphitic carbon. The oxygen content shows a direct influence on the water interaction and with increasing content the hydrophobic character changes to a hydrophilic one. It is noteworthy that the liquid functionalization with sulphuric acid and nitric acid leads to an elevated sulphur or nitrogen concentration, respectively. These results can be either explained with remaining amounts of the corresponding acid in the pores or the formation of sulphur/nitrogen containing groups during the functionalization. As the materials were washed intensively to remove the acids, most likely sulphur and nitrogen containing surface groups form. In summary for all materials suitable post treatment conditions were identified, where some oxygen groups are introduced to break the hydrophobic character, but the specific surface are and pore site distribution is not altered.
Regarding carbon supports with mainly external surface area, the partner FC was successful in establishing a synthesis route for high quality carbon nanofibres (CNF) showing a specific surface area of approx. 125 m2/g. Specifically so called platelet type CNFs were produced. Quality of CNF morphology was determined in four different aspects. First is the selectivity towards platelet type fibers and avoiding screws type ones. As the screws expose no graphene edges, they a supposed to stabilize the Pt to a lower amount. The second aspect of the morphology quality was a distribution of nanofiber diameters in the synthesized CNF-PL when a smaller range of CNF-PL diameters was considered as a sign of better quality. A third aspect is crystallinity or specifically additional amorphous carbon deposits. A forth important aspect is CNF nanomaterial purification from CNF-synthesis catalyst embedded in the carbon support. Within the project the preparation procedure and CNF synthesis catalyst preparation and handling could be optimized to finally satisfy all 4 criteria and produce highly selective platelet type CNFs, being very crystalline and showing a narrow diameter distribution around 150 nm. The purification in non-oxidizing acids was optimized to allow full removal of the metal and metal oxide catalyst used during the synthesis. The CNF can either be produced as dry powder or for improved handling as wet paste of known water content. Optimized material was produced on larger scale.
Main results on in-silico studies
To provide insight with calculations the pathway to carry out simulation studies on the APR was paved. Therefore, exchange-correlation functionals were calibrated and tested for their accuracy for Ni, Ru, Pd, Re, Pt surfaces. It was assured that the structural features and adsorption energy for water monomer can be described accurately.
The co-adsorption of polyols with a water molecule was studied on bare Pd(111), Re(0001) and Ru(0001) surfaces. The adsorption mode which was selected for this investigation can be considered representative of early stages possibly occurring in metal-catalyzed reactions in water. On the bare metals, the adsorption geometry is always characterized by one hydroxyl group interacting atop with the O-H bond parallel to the metallic surface, the residual fragment pointing outward. The three metals show different features with respect to the interactions with the polyols, being the adsorption energy smaller for Pd and quite similar for Re and Ru. In the presence of water co-adsorbed on the metal surfaces, two different configurations were found for the different polyalcohols. The occurrence of a given configuration for a specific polyol seemed to be determined by a balance of competing and cooperative energetic factors, which were related both to the polyol size and to the metal nature. In particular, the investigated polyols are strongly stabilized when adsorbed in the presence of water on Pd, whereas when the co-adsorption takes place on Ru and Re this effect is downsized, very likely due to their larger oxophylicity as detailed in (J. Phys. Chem. C 2015, 119, 17182).
Furthermore, a study on the nucleation of Pt, Pd and mixtures with Ni and Ru on graphene with and without vacancies. Furthermore, the adsorption of polyols (up to C4) on Ni, Ru , Pd and Re could be simulated. In particular, the nucleation of homonuclear (Ni, Pd, Pt, Re) and heteronuclear clusters (Ni-Pd, Re-Pt) on a graphene monovacancy was investigated; this study was focused on the analysis of the magnetic, energetic and structural features of the defective graphene during the trapping of the metal atom and the subsequent growing of the homo and hetero metallic cluster. It emerges that this event is regulated by the different affinity of the metal atoms to both the monovacany and the pristine graphene, the metal bulk cohesive energies and to the relative stability of the nucleation seeds (J. Phys. Chem. C 2016, 120, 12022).
Atomistic insights and detailed reaction mechanism were deduced for the reforming of 1,2-propandiol (1,2-PDO) on small Pt clusters. 1,2-PDO was selected as model substrate for polyfunctional oxygenated molecules present in the biomass based feeds. The reactivity of 1,2-PDO was investigated selecting as reliable model of a Pt catalyst, a cluster composed by thirty atoms opportunely cut out from the fcc platinum bulk. The reaction mechanism for the decomposition of the 1,2-PDO, was devised and fully characterized by means of ab-initio methods. The resulting reaction network formed by C-H, O-H and C-C bond scissions can be found at the project website (http://susfuelcat.eu/files/SusFuelCat_Conference_Project_Posters.pdf; UNIPA poster) and is also under publication by Duca et al.
The energy barriers and stabilities given in the network allow to discuss the main reaction pathways and possible strong adsorbing intermediates, which could poison the catalyst. Comparing the in-silicon and experimental results, in experiments hydroxyacetone is the main product, which agrees with the low activation barriers calculated for the two dehydrogenation reactions leading to this species. Additionally, in the experiments significant amounts of propanal were observed. The presence of this aldehyde suggests that C-O bond cleavage may occur at the early stages of the reaction, before the C-C bond cleavage. Therefore, for selected intermediates with a low degree of dehydrogenation, the activation barriers for the C2-O2 bond cleavage, leading to precursors of propanal, were calculated. According to this calculations, C-O cleavages should not occur at the first stages of the reaction. Moreover, it has also to be considered that various examples in the literature are reported demonstrating that during the reforming on metallic surfaces the C-O bond cleavage is not competitive with other processes.
In a similar manner, the reforming of ethylene glycole (EG) on Pd sub-nanometric catalyst was studied. The deduced reaction network is also given at the same poster as one paragraph prior. Moreover, this system allowed us to explore by means of ab-initio techniques all the reactive channels associated to the ethylene glycol decomposition reaction; this implied the construction of a relatively large grid of possible events that enabled to deepen also the issue concerning the energetic ordering of reactive pathways. The whole reaction network for the decomposition of ethylene-glycol was outlined and, within a graph theory based approach, all the pathways starting from C2 and leading to CO and H2 were analyzed and compared, making possible to draw a comprehensive picture of the competitive routes that enable the production of hydrogen on a sub-nanometric palladium cluster. In detail, the protocol for the reaction network analysis developed can be divided into three steps: i) ab-initio calculation of the activation barriers and pre-sieving of the molecular events, ii) construction of the graph representing the final reaction network, iii) analysis of all the possible pathways.
Main results on hydrothermal stability
Carbide-derived carbons (CDC) and carbon nano fibers (CNF) produced within the project as also commercial activated carbon and carbon black were studied for their stability during hydrothermal treatment. In situ experiments were carried out in water vapor and TGA. Post mortem characterization (XRD, TPO and N2-sorption) was carried out for experiments where material was treated at 240 °C in the liquid phase for 7 days and characterized post mortem. The results show that the SusFuelCat carbons (CDC, CNF) and the carbon black are highly stable against hydrothermal attack, while the commercial activated carbon, showed a pronounced lower stability. Stability against water vapour or oxygen was studied in a temperature programed oxidation, hence with contiguously increasing temperature. Assessing the stability is carried out by comparing the onset temperature for oxidation.
It was shown that the onset temperature of decomposition in oxygen increases with the crystallinity of the materials from 480 °C to 675 °C (Chem. Mater. 2015, 27, 5719), showing the huge influence of the carbon support type on stability. The water vapour studies gave in principle the same trend, but the stability is increased to higher temperatures and above 600 °C. The amorphous activated carbon however show a continuous mass loss also till 600 °C and not a very sharp on-set of decomposition. The stability test which is most close to the APR reaction conditions was a 7 day hydrothermal stress test in an autoclave at 240 °C. Here no in situ data is obtained but post mortem characterization carried out. After this stress test the amorphous activated carbon showed a reduction of its specific surface area (initial 900 m2/g) by 20 %. Most likely surface functional groups were build, which increase the mass of the material. Graphitic materials like CNF or carbon black didn’t show a similar loss in specific surface area, but have in general a lower specific surface area below 200 m2/g from the beginning. Interestingly amorphous carbide-derived carbon with an interestingly high surface above 1600 m2/g didn’t show a reduction of specific surface area. More graphitic CDC also shows no surface area loss, but has interestingly a starting surface area which is comparable to the commercial activated carbon, while being not amorphous. Thus CDCs are a highly interesting catalyst support for reactions carried out under hydrothermal conditions, where stability can be an issue.
Due to the already very good stability, posttreatments of the carbon supports were only studied to a minor amount. In general oxidative methods, which remove the amorphous carbons and leave the more graphitic one behind were studied. Materials studied were the amorphous activated carbon, amorphous CDC and graphitic CNF. Air oxidation at 350 and 425 °C als also a liquid phase oxidation with nitric acid was studied. For the air oxidation at 350 °C no mass loss and thus change of any materials was observed. Air oxidation at 425 °C gave a mass loss of approx. 15 wt.-% for all materials. Despite this removal of less stable carbon the post treated materials didn’t show an increased oxidation resistance, but rather a slightly reduced one. Most likely during the oxidative post treatment slightly additional defects are introduced, which are starting point for later attacks. Thus, no increased, but slightly decreased stability results. Furthermore, it was observed that due to the post treatment, some surface groups were newly introduced. Within the project we demonstrated that Pt/C catalysts with a higher surface acidity favor alkane production in the aqueous phase reforming of xylitol while the ones with lower acidities generate hydrogen with high selectivity and turn-over-frequency (Catal. Sci. Technol. 2014, 4, 387). Thus, these additional surface groups are unwanted and no post treatment of the carbons is recommended.
To study the possible influence of the metal in the final catalyst on the stability, procedures to deposit high amounts of Pt but with comparable particle size needed developed, as it was assumed that this pronounced higher loading would accelerate the stress test. Starting from chloroplatinic acid dissolved in aceton a solution 2 – 3 time more than the pore volume was added to the carbon samples. Intense ultrasound was applied to support a homogeneous infiltration of the solution. After drying a reduction procedure was applied, which either was relatively harsh and started directly at 300 °C with H2 or was more mild and with H2 already during the heating ramp till 300 °C. Loadings ranging from 15 to 20 wt.-% could be achieved at particle sizes ranging for the mild reduction from 2.9 to 4.5 nm. The harsh reduction gave unsatisfactory sizes from 5.1 to 36 nm. The materials were employed in a 7 day hydrothermal stress test at 240 °C. Characterization of the stressed material showed that the Pt loading reduced only slightly e.g. from 18.2 to 17.7 wt-%. This is most likely related to a fast leaching of not completely reduced Pt precursor and not due to leaching of metallic Pt clusters under hydrothermal conditions. Pt particles sizes as also the specific surface area remained unchanged within the stress test. Comparing the onset for oxidation with and without Pt the temperature where decomposition starts lowers by approx. 100 °C as the decomposition is catalyzed by Pt. Nevertheless, still a very high and sufficient stability and onset of oxidation of approx. 400 °C was achieved for the graphitic carbon supports.
Main results on catalytic experiments
To obtain feedback from catalytic experiments three continuous set-ups were reconfigured and set into operation to study APR of i) C5 and C6 polyols in lab scale, ii) APR of C1-C3 oxygenates in lab scale and iii) the long term APR of various feeds in validation scale.
Testing of unsupported colloidal particles faced several technical difficulties, such as a need for a dedicated set-up and an unclear operational process. Unsupported colloidal particles are notorious for their ability to deposit on all metal parts of reactors during catalytic experiments. Not all parts of available batch reactors can be cleaned in an appropriate way after each experiment or alternatively protected. Moreover, separation of unsupported colloidal particles from the liquid phase is also challenging. Thus, neither HPLC, nor TOC analyses can be directly performed due to high concentration of metals in the liquid phase. Another challenge is related to an accurate carbon balance determination since protecting agents PVP and PEG are reactive during APR contributing to generation of carbon containing products and hydrogen. All these complications lead to a high uncertainty in assessing the activity and selectivity of unsupported colloidal particles and finally inability to obtain reliable information. Therefore catalytic data given below were generated using m supported nanoparticles prepared either by classical impregnation or by immobilization of colloids.
After comparison of different metals as catalytically active phase it can be stated that all monometallic non platinum catalysts displayed significantly lower activity (Ni) or deactivated much faster (Ru, Re) than monometallic Pt. Activity of bimetallic Pt-Ni, Pt-Co and Pt-Ru catalysts was practically en par with the monometallic counterpart. The main difference was seen for the Pt-Re catalyst, which had the highest activity level among the tested catalysts. For Ni and Pt-Ni catalysts, leaching of Ni was observed. Selectivity towards hydrogen was the highest with Pt not increasing after addition of other metals. It should be noted that elucidation of catalysts prepared from colloids is not straightforward as their performance depend on the immobilization and capping agent removal procedure. Some catalysts in particular were inactive because of inefficient capping agent removal, while others displayed strong metal leaching as immobilization was not efficient. Finally after careful optimization APR active and stable catalysts have been developed. For these colloidal based materials post-catalysis characterization by TEM showed a slight increase in Pt size, although no metal was lost during the reaction. N2 physysorption also showed that PVP was removed from the support/metal.
No clear trends related to possible mass transfer limitations were seen while studying particle sizes of catalysts varying from < 100 µm to above 1 mm or even extruded pellets. A more detailed characterization revealed that larger sized supports tended to have an egg-shell distribution of the active metal, which probably minimized or even eliminated pore diffusion limitations. The overall picture emerging form the experimental data is that there are indications that an egg-shell distribution is advantageous, even if a solid proof is missing. Studies with unsupported high surface area as well as supported Pt catalysts showed, that carbon-platinum interactions seem to be crucial for the desired activity and selectivity.
An in situ ATR-IR spectroscopy study was carried out for supported Pt catalysts during APR of hydroxyacetone. Carbon as a support could not be employed for the study, but degradation of oxide supports was monitored in-situ. Degradation of alumina under the hydrothermal conditions was obvious. Furthermore, strong adsorption of hydroxyacetone on the surface of zirconia was detected, causing formation of deposits and catalyst deactivation via aldol condensation. Thus, a proper choice of the support could be proven to be highly important to achieve excellent performance of APR catalyst.
Variation of the flow rate, hence, the residence time, showed that the hydrogen yield increases initially with increasing the residence times and thus degree of conversion. Thereafter it reaches a maximum and decreases again. This behavior is a direct indication that hydrogen is consumed in consecutive reactions and thus an optimal degree of conversion should be aimed for.
The influence of chirality of the feed was studied in detail by comparing sorbitol, galactitol and their mixture, showing the same activity and selectivity patterns. Comparison between C5/C6 (xylitol/sorbitol) and C3 feeds (hydroacetone) revealed that the ratio of reactivity in respectively water gas shift and APR is lower for the latter feed.
For xylitol, sorbitol and galactitol structure- property relationships could be deduced for Pt/C. It was proven that the carbon supports are well suited for tuning the hydrogen and alkane selectivity. For high hydrogen selectivity, the carbon support should show a neutral character, whilehigh alkane selectivity requires an acidic function meaning that surface functionalization determines the selectivity. While activated carbons due to their synthesis procedure very often exhibit a non-negligible surface functionalization, resulting in formation of alkanes, other carbons supports such as CNF or CDC displaying excellent purity result therefore in high hydrogen selectivity. Furthermore, indications were found that for bulky polyols the reaction is structure sensitive, hence the turnover frequency of Pt increases with an increase of Pt cluster size. As a consequence of a larger fraction of Pt not available for catalysis when the size is increasing, there is an optimum for the active metal size exhibiting the highest reactivity.
Characterization of as prepared and spent catalysts was carried out. Colloidal Pt and Ru nanoparticles as precursors for immobilization on carbon to obtain heterogeneous catalysts proved to be stable for 1 to 2 years. Ru catalysts prepared by a colloidal route (PVP stabilized) and immobilized on carbon nanofibers or carbide-derived carbon were employed in galactose hydrogenation. High resistance to leaching and sintering under hydrogenation conditions was observed independent of the PVP removal procedure. Ru, Pt, PtNi, Re, PtCo and PtRu on carbon support were applied in aqueous phase reforming. No pronounced sintering or leaching of was detected under APR condition for the noble metals, contrary to Ni and Co. In addition to leaching and sintering, noticeable oxidation was seen for all metals, except Ru.
Based on the knowledge acquired in the project the following catalyst properties were identified for high performance: i) supermicroporous to mesoporous texture (>0.7 nm); ii) non acidic support; iii) sufficiently hydrophilicity of the support; iv) Pt sizes around 2 - 5 nm. Additionally, minimization of mass transfer limitations through application of egg-shell catalysts and a decrease of pressure drop by a proper selection of the particle grain shape and size are beneficial.
Based on the generated experimental knowledge o a recommendation for standardized testing of APR catalysts among different labs was established for both batch and continuous reactors. For the standardized testing, a mixture of 5% sorbitol and 5%mannitol in water should be chosen.
The catalyst should be activated upon heating under H2 at 375°C for a minimum of two hours under atmospheric pressure. Subsequently, the standardized testing would be carried out in the batch reactor at 30 bars pressure using N2, 225°C, for 2h, followed by an analysis of the gas phase by GC and the liquid phase by HPLC and elemental analysis. Note, that for the batch reactor continuous sampling can be critical as the carbon balance is not fully closed for high sampling amounts, due to the gas phase loss. For continuous operation 30 bar N2 with a N2 flow of 0.5Nl/min should be realized at 225°C, sufficient stabilization time prior to data collection, gas analysis by and the liquid phase by HPLC and elemental analysis.
Main results on process validation
In long term studies with Ru/C catalyst for polyol production large quantities of glucose and sucrose were converted to technical sorbitol (both 50 wt-% in water). At a degree of conversion of 96 % a yield of 92 % could be realized and in total more than 40 kg of feed were produced. No major catalyst deactivation was observed for a total >400 h time on stream. Nevertheless, some solid formation was observed, which can lead to a pressure increase. Probably small amounts of sugar monomers are converted to char like products by side reactions. Depending on the feed crystallization can occur and be avoided by slight heating of all process lines and during storage by diluting the product to 29% wt-% in water.
Long term studies on the APR of model compounds and technical sorbitol were carried out in combination with a temperature variation from 220 to 250°. Normally no induction period was observed in these studies. Opposite results were obtained for a Pt/C catalyst on a special carbon support with very low loading (0.5 wt.-%) and reduction perfomed within the reactor rather than externally prior to the reaction. Here continuous activation was also seen as continuous changes of CO2 to CO ratio, representing the ability to perform water-gas shift reaction, even after 150 hours of time-on-stream. Thus, changes of the active sites seem to take place during this long induction period. Typically, such changes in catalytic behavior are associated with solid-phase transformations of the active phase, for example reduction or oxidation. Note, that the catalyst was pre-reduced at a high enough temperature to ensure complete reduction in line with TPR data, even if the reduction was carried out directly in the reactor. TEM data confirmed almost no changes in the cluster size of Pt, thereby excluding sintering. Finally, within the project it could not be unequivocally concluded in which cases the catalysts would show a long induction period. At the same time such induction period is not needed to achieve an active and selective catalyst.
The long term APR studies were also employed to deduce possible deactivation of the catalysts. In this sense a clear difference for model and technical feedstock was observed. For Pt/C on activated carbon and with model feedstock only a slight deactivation was observed and the degree of conversion reduced from 66 to 60 % after 250 h time on stream. On the contrary employing strong catalyst deactivation was obvious for the technical feedstock. Most likely such deactivation is related to blocking the active sites with reactants/intermediate products of these more concentrated feedstocks (30 to 50 %), while the model feeds were only in 10 % concentration. It is expected that dilution of the technical feed can minimize deactivation. Reactivation of the catalysts through extraction/flushing was studied. Flushing with water is not sufficient to restore the activity. Reactivation procedure with an organic solvent (acetone) allowed removal of organic residues blocking the catalyst and regaining catalytic activity after reduction in hydrogen.
Four kinetic models for xylitol and sorbitol APR, which serve different purposes, were developed. One mechanistic model accounting for consumption of hydrogen in consecutive side reactions was used chemical reaction engineering simulations allowing to deduce hydrogen selectivity profiles. Modelling of mass transfer showed that with ca. 1 mm particle diameter, pore diffusion limitations might occur lowering activity. Pore diffusion limitation has a minor influence on the resulting hydrogen selectivity. Variation of the WHSV showed that hydrogen selectivity shows a maximum with the residence time. Beyond this maximum the selectivity drops only slightly, therefore losses of hydrogen are small at too high residence time. Analysis of the reactor hydrodynamics (residence time behavior) showed that pronounced back-mixing leads to a loss of the maximum in hydrogen selectivity vs. residence time. Thus ca. 20 % of hydrogen selectivity can be lost. Most interestingly, simulating a continuous hydrogen removal along the reactor through a membrane, showed that the total amount of hydrogen obtained can be increased by a factor of 3, as a result of efficient suppression of side reactions involving hydrogen. A sufficiently high hydrogen removal through the membrane is needed to obtain this significant increase in hydrogen yield.
While the calculations above used a kinetic model capable of accurately describe hydrogen selectivity, flow sheeting accounting especially for separation needs a detailed description of all intermediates. Thus, a model based on a detailed reaction network involving the key intermediates, excluding however detailed kinetic information, was employed in a so-called stoichiometric reactor option. The yield of the products and side products was set by coefficients, according to the experimental results. Within the framework of this model, heat and mass balances calculations as well as costs estimations were carried out for a 500 kg/h hydrogen production plant operation with sorbitol syrup as a feedstock in APR of sorbitol using Aspen HYSYS software. For reactor modelling a complex reaction network was taken into account along with phase equilibrium simulations which determined to have a significant impact on the total process heat due to possibility of water evaporation. Improved economy could be achieved e.g. through incoperating a medium pressure steam generator after a high temperature separation, despite a low temperature separation without steam generation. The total costs of hydrogen were estimated as 9.4 $/kg, where the feedstock costs take the major contribution of 86%. Thus, the cost of hydrogen from APR can be in an economic range if the feedstock prices drops below 150 $/kg.
Based on the results of the project and expertise of the industrial partners an assessment of the catalysts and process was carried out. From an economic point of view, the metal supported catalysts can be sorted in two groups. i) Expensive noble metals (Pt, Pd, Rh) where the metal price dictates the final catalyst price. Here recycling of the metal is essential to lower the price. ii) Cheaper metals (Ru, Re, Ni), where also process and support costs can play a significant role. Pt shows by far the best performance and selectivity towards hydrogen, thus high costs cannot be prevented. Thus, also carbon is due to the recycling possibility by far the best option, as the Pt recovery rate is very high and thus the Pt price can be dropped pronouncedly through the recycling. For the special catalysts developed in SusFuelCat additional costs could stem from the newly developed colloidal preparation route and for the costly carbon supports (carbide-derived carbons, carbon nanofibres). Thus, it is advised to employ the new colloidal preparation route especially for expensive metals of the first group, where the production process costs are less relevant. The colloidal route, nevertheless, can also be employed predominantly to guide catalyst development, while the final catalyst would be produced by more economically viable classical impregnation procedures
Nanomaterials (carbon support and catalytically active metals) in general can be of potential risk, thus full health and safety studies might be necessary in the future. To minimize the costs of employing more expensive advanced carbon supports, the strategy was developed to coat cheap and readily available carbon pellets with these advanced carbon materials. Additional benefits of these coated carbon pelletized catalysts are related to a decrease of pressure drop, which would be high otherwise with the small sized carbide-derived carbon or carbon nanofiber grains. The additional costs for the preparation were estimated to be €150-175/kg. With such approach affordable prices of Pt catalysts can be achieved especially if significant advantages of these special supports in terms of higher selectivity arise.
Potential Impact:
1. Potential impact of SusFuelCat
The consortium partners of SusFuelCat found proper APR catalysts for renewable hydrogen production. Important steps were carried out during the project towards this aim. Important indications for structure activity relationships were deduced. New possibilities to tune active and stable metal catalysts were developed and can be of high value also for other catalytic problems. The experimental data and simulation allowed deducing further mechanistic insights, which will be important for the catalyst and process design. First cost estimations show that APR hydrogen can be economically competitive. Thus it is believed that APR technology can be of importance for sustainable hydrogen production in Europe.
Besides this direct impact on the horizon, already further impact becomes apparent for each partner. The pronounced knowledge on well-controlled colloidal based catalysts, methodologies for immobilization and stabilization agent removal, synthesis of high quality carbonaceous materials, great progress on the simulation of the systems, and developed experimental protocols have the potential to be of great use in catalysis and neighboring disciplines in general.
2. Socio-economic and societal impact of SusFuelCat
Especially for the three industrial partners in the consortium a direct economic impact can be expected. The two SMEs within the consortium are BTG Biomass Technology Group and FutureCarbon. BTG is specialized in process development for conversion of biomass into biofuels and bio-energy via flash-pyrolysis. FutureCarbon is specialized in development, production and refinement of carbon nanomaterials and carbon nanomaterial based products. Both SMEs profit directly from the results obtained.
The established industrial partner is Johnson Matthey, which is a specialty chemicals company, focusing on catalysis, precious metals, fine chemicals and process technology, with extensive knowledge and experience of carbon supported noble metal catalysis. With the combination of innovative SMEs and an established global industrial partner it is expected that a transfer of the project results can become true.
i. health and well-being
The use of fossil fuels in road transport, aircraft and industry applications releases different greenhouse gases and particles into air and environment. Hydrogen, as produced in SusFuelCat, instead serves as an alternative green fuel. Therefore, investigations focusing on effective production of alternative fuels like hydrogen are a possibility to contribute to a healthier environment, especially in cities and industrialized regions.
ii. employment
For all three participating companies BTG, FutureCarbon, and Johnson Matthey, the work at SusFuelCat was important to find new products (catalysts) and processes that can be commercially exploited in order to maintain or increase their number of employees.
iii. training and career development
Several young scientists worked at SusFuelCat. They learned about chemistry, process engineering, built-up and run of experiments, writing papers, presenting the results at conferences, and even about teaching young students. Finally with incorporating their findings to their dissertations they got trained for going ahead in their career.
All seven participating universities from Erlangen, Darmstadt (Germany), Enschede (Netherlands), Madrid (Spain), Turku (Finland), Palermo (Italy) and Novosibirsk (Russia) joined in carefully training the young scientist.
3. Main dissemination activities
Together with publications SusFuelCat had 126 dissemination activities, of which the most notable are described below:
i. Publications
In total there have been 13 peer review publications; among them two have been published open access via OpenAIRE.
Exemplary three publications are discussed. The article “Preparation of carbide-derived carbon supported platinum catalysts” in Catalysis Today, Vol. 249, (2015) page 30 disseminates important results on novel nanoporous carbons and their used in catalysis. The publication “Controlled synthesis of PVP-based carbon-supported Ru nanoparticles: synthesis approaches, characterization, capping agent removal and catalytic behavior” in Catalysis Science and Technology, Vol. 6 (2016) page 8490 disseminates and summarizes important findings on the synthesis of well controlled nanoparticles as also the important removal of stabilizing agent for the application in catalysis. In the publication “Aqueous-phase reforming of xylitol over Pt/C and Pt/TiC-CDC catalysts: catalyst characterization and catalytic performance” in Catalysis Science and Technology, Vol. 4 (2014) page 387 disseminates structure-activity relationships identified for the APR reaction.
ii. Conferences and accompanying activities
The consortium participated on several conferences and presented the project to scientists, industry and to the wider public. There have been 48 oral presentations to a scientific audience, which have been accompanied by 41 posters presented at these conferences. Besides the scientists, also the wider public has been delivered information by publishing twelve press releases, one oral presentation, and three articles to popular press and audience. This has been completed with setting up a project website, which was constantly updated during the project to keep the stakeholders informed.
As proposed in the DoW, SusFuelCat partners held a scientific conference subsequent to the final project meeting in Enschede, Netherlands organised by BTG Biomass Technology Group and University of Twente and held at the campus of the latter. The scientific partners and BayFOR further contributed substantially to the CarboCat conference in Strasbourg, France. At both conferences, the consortium had the possibility to exchange ideas with external experts or to find applicants for project results.
SusFuelCat partners visited two large conferences, EURONANOFORUM and Industrial Technologies, which are both sponsored and co-organised by the EU Commission. Here, they presented the project to a wider audience, as to participants from other EU projects, stakeholders from industry and academia. As for all the other events, posters, flyers, roll-ups, and gadgets were presented at booths. Through in-depth bilateral discussions more than 30 valuable contacts were acquired.
iii. Clustering with other EU projects (@Bastian: please mention your experiences ...)
In 2014 SusFuelCat was invited by the EU Commission to take part in the cross-cutting EU Clusters on “Catalysis” and "Engineering & Upscaling". The SusFuelCat coordinator was actively involved within the European Cluster on Catalysis (http://www.catalysiscluster.eu/). The subgroup “Advanced analytical approaches in catalysis, in situ and in operando studies” was headed by the coordinator and input for the roadmap on catalysis was given. On the Engineering & Upscaling activities the coordinator gave input through a survey.
4. Exploitation of results
i. Key exploitable results
Key exploitable result No 1
Short Title: High pressure FTIR-ATR
Full Title: Method for carrying out ATR studies at elevated pressure. This allows for in situ studies not only in this reaction
Lead Partner: UT, partners involved: UT, UNIPA
Brief description: Equipment and knowhow are available at UT to carry out in situ FTIR-ATR studies at elevated pressures. Together with the simulation competency of UNIPA complex spectra can be evaluated.
Key exploitable result No 2
Short Title: Simulation-package
Full Title: A good package of computational tools developed
Lead Partner: UNIPA , partners involved: UNIPA.
Brief description: Simulation code for adsorption and reaction studies in aqueous phases were developed. Also code to study metal cluster development on graphene is available. The code can be used for studies and further refinement at UNIPA.
Key exploitable result No 3
Short Title: CNF-PT-upscale
Full Title: Up-scaling the production of CNF-PT nanomaterial without loosing the high quality
Lead Partner: FC, partners involved: FC, FAU, and JM (exploitation partner)
Brief description: Steps for scale-up of CNF-PL synthesis from lab scale to pilot plant scale. 1st step: More CNF-PL from concrete quantity of synthesis catalyst, achieved by process optimization (e.g. gas flow turbulences); scaling factor ~1.3) 2nd step: Planning and dimensioning: Single plug-flow reactor --> bundled tube reactor (scaling factor ~6)
Key exploitable result No 4
Short Title: CDC-High-T
Full Title: Mesoporous Graphite as stable, highly porous, electrical and thermal conductive material
Lead Partner: TUDA, partners involved: TUDA
Brief description: Mesoporous Graphite can be produced, which is interesting as chemically very stable and highly electrical and thermal conductive material; The knowhow is used in follow up research projects.
Key exploitable result No 5
Short Title: CDC-Synt
Full Title: Synthesis procedures and strategies for carbide-derived carbons
Lead Partner: TUDA, partners involved: TUDA
Brief description: Synthesis procedures and strategies for production of carbide-derived carbons; The knowhow is used in follow up research projects.
Key exploitable result No 6
Short Title: Colloidal-NP
Full Title: Synthesis procedures and up-scaling for well designed colloidal particles
Lead Partner: BIC, UAM; partners involved: BIC, UAM;
Brief description: Identification of different synthesis conditions leading to nanoparticles of different metals (Pt, Pd, Re, Ru, Ni, Co) with controlled size and a broad range of sizes. Re and Ru NPs are difficult to obtain. Conditions for the reduction and for the growth of large NPs have been identified. The nanoparticle colloids prepared by different methods (PVP colloids, polyol, microemulsion) provide different types of interactions with carbon supports. The size control is a key for a better control of NPs performance. Colloidal synthesis can be scaled up, is reproducible and NPs are stable (Pd, Pt). The concentration of reagents can be increased to reduce the reaction volume and make scale up easier.
Key exploitable result No 7
Short Title: Immo
Full Title: Immobilization procedure for colloidal nanoparticles by simulated APR conditions
Lead Partner: UAM, BIC, partners involved: UAM, BIC, FAU, FC, AAU, UT
Brief description: Identification of synthesis conditions for nanoparticles and catalysts that promote interactions between nanoparticles and carbon supports. Enhanced stability and activity of the catalysts prepared following the procedure proposed. The procedure can be applied to other metals and other supports not studied in this project. The distribution of the nanoparticles within the carbon porous structure can be controlled.
Key exploitable result No 8
Short Title: Mini-APR
Full Title: Up-scaling the lab APR process. Probably also making mini-APR-reactor of lab size or a little big bigger commercially available, which could be used for feedstock screening at interested partners.
Lead Partner: BTG, partners involved: BTG, AAU, UT
Brief description: Upscaling of APR/LPR process for converting liquid biomass feeds, including pyrolytic sugars into platform chemicals and fuels (alkanes, hydrogen).
Key exploitable result No 9
Short Title: PVP-Removal
Full Title: PVP-Removal – Method for effective removal of PVP stabilizer from nanoparticles, not suffering of dramatic particle size increase or metal loss.
Lead Partner: UAM, BIC, partners involved: AAU, BIC, JM, UAM, UT
Brief description: Currently known APR catalysts are based on metal oxide supports. The SusFuelCat work has shown applicability of carbon supports. These enable easier recycling and higher recovery of the precious metals used for the catalysis. The concentration of the catalyst during the removal and removal time can be increased to reduce volume and make scale up easier.
List of Websites:
www.susfuelcat.eu
SusFuelCat aims to boost Europe’s expertise on catalysts for sustainable fuel production, especially hydrogen, by the process of Aqueous Phase Reforming (APR). APR is a technology that allows the catalytic conversion of low value biomass streams into biomass based fuels, with catalysts being the key stone in the process. That’s why SusFuelCat aims to gain fundamental material structure-property relationships for the catalyst using model supports and metal nanoparticles.
To deduce fundamental structure-property relationships the properties of catalysts were varied with respect to their active metal, cluster size, pore size, carbon graphitization. Additionally, in-silico studies for theoretical investigations were carried out. Catalyst activity and selectivity were studied experimentally with real- and model-feedstocks to give feedback on the catalytic performance. The hydrothermal stability of the carbon support materials and later on of the catalysts were studied. These results were used for the model catalyst design in the subsequent iteration loops. Throughout the project the economical viability of the envisaged process was monitored.
The project results showed that proper APR catalysts for renewable hydrogen production could be obtained and that a new platform of model catalysts is available now, which can be used for noble metal on carbon catalyst development in general. The experimental data and simulation allowed deducing further mechanistic insights, which are important for the catalyst and process design. Cost estimations show that APR hydrogen can be economic competitive, if the right feedstock is employed. Thus, it is believed that the developed APR technology can be of importance.
Besides this direct impact on the horizon, already further impact becomes apparent for each partner. The pronounced knowledge on well-controlled colloidal based catalysts, methodologies for immobilization and stabilization agent removal, synthesis of high quality carbonaceous materials, great progress on the simulation of the systems and developed experimental protocols have the potential to be of great us in catalysis and neighbouring disciplines in general.
Project Context and Objectives:
Energy production and supply is one of the most important basics for a high quality of life, industrialization and civilization. Europe’s wealth and leading position in industry depends heavily on energy, for which fossil fuels remain the primary source. In addition to the huge dependency on (mainly non-European) fossil fuel exporting countries, and unstable prices, fossil fuels are a limited resource whose use releases carbon dioxide to the atmosphere, a major contributor to global warming. Hence, economically and environmentally European based sustainable energy production will become ever more pertinent.
For biomass in particular, the energy consumption of the conversion process needs to be as low and the biomass utilization as efficient as possible in order to increase the energy output per square meter of cultivable land. In this project we focus on Aqueous Phase Reforming (APR) of biomass resources. Regarding this the products aimed by APR, viz. hydrogen or, preferably, a combination of hydrogen and alkanes, are of high potential for sustainable fuel production. The sustainably produced hydrogen can be used either as fuel directly or is mandatory in bio-refineries for sustainable liquid fuel production. A major advantage of APR is the use of aqueous or water-soluble feedstocks at reduced temperatures and slightly elevated pressures, enabling processing without energy intensive drying while hydrogen can be in situ generated from biomass by the APR reaction and in parallel from the water content by the Water-Gas-Shift (WGS) reaction. Hence, APR is one of the most promising techniques for sustainable liquid and gaseous fuels production from biomass feedstocks, with catalysts being the key stone in the process.
The versatility of the process, the limited energy consumption and the product flexibility address the crucial challenges of sustainable fuel production from biomass and thus APR can be seen as a major technology to allow rapid industrialization of sustainable fuel production in Europe. In parallel to the interest in sustainable produced hydrogen as fuel, this is also a highly sought commodity, for example to upgrade bio-based mixtures to high quality fuels or chemicals.
A key towards of any successful implementation of APR is catalysis. State-of-the-art catalysts used in lab scale APR have already demonstrated the potential of catalytic APR, but a) are not yet optimized, due to their unknown structure-property relationships, b) require high amounts of costly noble metals (> 5 wt.%) and c) their hydrothermal stability is shown to be limited. For a European APR-based sustainable fuel production to be viable, key objectives of catalyst optimization are a gain in energy efficiency of the process, reduction of total costs of catalyst and independency on material imports for catalyst production. Regarding the latter, the paradigm shifts towards carbon supported catalysts, due to their superior hydrothermal stability. Current research on APR-catalysts is rather unstructured and diversified, hence recommendations to avoid such drawbacks are not clearly presented. In addition, a criterion is to lower the amount of costly noble metal. For all metals a considerable recycling industry is available, so that limited technical uses would not be considered a threat for the existing supplies. However, a considerable use in APR could increase market prices for precious metals. For example, catalysts (mostly automotive applications) already account for one third of annual platinum production. Aqueous phase reforming may be applied in similar smaller units, to allow use at remote locations/close to biomass resources which may mean that total metal inventories are small, at the same time rendering recycling more costly. The use of carbon supports, which are focused on in SusFuelCat, also allows easy and more environmental friendly metal recycling by burning of the carbon support. The high metal content residues can then be easily converted to new catalysts.
The two primary objectives of the project are:
a) To unleash the potential of carbon supported catalysts to establish APR as an energy efficient process to convert different biomass based streams to sustainable fuels and to decrease time to market for a commercial APR catalyst.
b) To generate data to allow the design and scale up of the APR process towards a further demonstration level. The final target is to show the technical and economic viability of the global process.
To achieve this, fundamental material structure-property relationships for the catalyst should be gained. Model supports and metal nanoparticles play a major role for this aim, thus studying how properties of catalysts are varied with respect to their active metal, cluster size, pore size, carbon graphitization will be carried out. These materials form the basis for experimentally catalytic studies, which will be evaluated together with the results from the material characterization and additional in-silico studies. Furthermore, the information obtained shall be used continuously to evaluate the economic viability of the envisaged process.
Thus, the two primary objective can be broken down the following R&D targets:
Nanoparticle synthesis
- Avoiding the undefined influence of the support during nanoparticle deposition leading to broad cluster size distributions through synthesizing well defined colloidal solutions of nanoparticles. For the various metals to study (Ru, Re, Pt, Ni, Pd, Co) synthesis protocols for the polyol or micro-emulsion technique should be developed.
- Studying the influence of upscaling the colloidal synthesis method on the cluster sizes.
- Establishing methodologies for immobilization of these well defined nanoparticles in supports, either through post immobilization or in situ synthesis.
- Establishing methodologies for efficient removal of the capping agents employed in colloidal solutions and its influence on the cluster size
- Comparison of “colloidal” synthesis routes to classical impregnation routes.
Carbon support synthesis
- Production of model porous carbons with inner porosity varying in pore size and graphitization, which are needed as basis for model catalysts to deduce structure-property relationships.
- Production of carbon nanofibers as material with external surface and possibility for superior stability.
- Detailed characterization of materials properties.
- Comparison of model carbon to commercial activated carbons and carbon blacks.
in-silico studies
- individuating the best performing approaches QM, MD, MC or hybrid
- calibration of methods
- realizing metals as clusters or periodic facets
- studying how metal clusters form on carbon supports
- studying sorption processes
- deducing possible reaction pathways
Hydrothermal stability
- Testing hydrothermal stability of different model and commercial carbons (no Pt).
- Testing possibilities to increase hydrothermal of carbon supports.
- Testing hydrothermal stability of catalysts (with Pt).
Catalytic experiments
- Screening of activity and selectivity for different active metals and supports
- Handling studies for different sized supports
- Deduce proper conditions and protocols for APR catalyst testing.
- More detailed studies on feedstock influence (diastereomer influence, C6 sugar alcohols, C5 sugar alcohols, C3 sugar alcohols and acids).
- Detailed study on residence time influence to obtain selectivity vs. degree of conversion
- Obtaining database for reaction network analysis
- Obtaining database for kinetic modeling and flow sheeting
- Carrying out long term experiments to study stability for selected catalysts
- Studying reactivation protocols for deactivated catalysts
Process validation
- Continuously accompanying the development through assessing the catalyst production costs and hydrogen production costs through APR
- Continuously refining the cost calculations
- Providing process insights through chemical reaction engineering simulations to study possibilities how to increase the overall yield.
- Providing process insights through flow sheeting to study possibilities how to increase the economics of the overall process.
Project Results:
Subsequently the main results for the different targets are presented in a concise manner.
Main results on nanoparticle synthesis
Regarding metal model material nanoparticles (NP)s, they were synthesized as colloids by colloidal and the polyol methods and also by microemulsion method with various capping agents and surfactants for six metals. In total more than 400 syntheses were carried out, including a broad screening of reaction systems and reagents and characterization to assess structure. Particle size variations within the following boarders could be obtained: Pt (2.5 – 8.3 nm), Pd (3 – 13 nm), Re (0.7 – 1.4 nm), Ru (1.6 – 4.5 nm), Co (1.8 -18.1 nm) and Ni (2.6 – 9.8 nm). Thereby narrow particle size distributions could be achieved and the sizes are within the interesting range for catalysis. The polyol method was applied for synthesis of Co, Ni and Ru metal NPs. As a stabilization agent polyvinylpyrrolidone (PVP), dodecylamine (DDA), polyethylene glycol (PEG) and 3-mercaptopropionic acid (3-MPA) were employed. TEM, XRD, UV-Vis spectroscopy and XPS were applied to characterize these NPs.
The Pd colloids were prepared using PVP as caping agent and ethanol and methanol as reducing agents. The nanoparticle size was found to decrease with increasing alcohol concentration and PVP dose. The water/AOT/isooctane system was selected for microemulsion synthesis in combination with N2H4 and NaBH4 as reducing agents (Colloids and Surfaces A 2016, 497, 28) (Perez-Coronado et al., 2016). Other systems evaluated for microemulsion synthesis where those based on cetyltrimethylammoniumbromid (CTAB) and Triton X-114 surfactants. The water-to-AOT molar ratio (w0) was the most influencing factor. Smaller nanoparticle size was obtained for low ratios, although higher amount of AOT remained on nanoparticles even after purification. Lower agglomeration and higher hydrophilicity of nanoparticles was observed afterwashing with methanol. Pd nanoparticle size was in a broad range for colloidal (3-13 nm) than for micoremulsion synthesis (6-10 nm).
The preparation of Pt nanoparticles Pt using PVA and PVP-based colloidal synthesis with methanol and NaBH4 showed an influence of synthesis variables similar to that of Pd, although lower tuneability of nanoparticle size (3-6 nm) was achieved . Narrower size distributions were obtained with NaBH4. The microemulsion synthesis with the water/AOT/n-heptane system provided witha a broader range of NPs mean size (1-7 nm). A better tailoring of the properties of the Pt nanoparticles was achieved varying the type of reducing agent, the reducing agent/Pt ratio and water/AOT ratio. The Re colloids were synthesized by simple colloidal (PVP) and microemulsion methods (water/AOT/isooctane) (Bedia et al., 2015) (Colloids and Surfaces A 2015, 469, 202). A strong reducing agent (NaBH4) and hydrogen atmosphere was needed to achieve more complete reduction. Reduction rate was much affected by temperature, although above 60ºC fast reoxidation was observed. The main tool to adjust nanoparticle size was the addition rate of the reducing agent and the PVP to Re molar ratio, however a narrow range of mean size was achieved (0.7–1.4 nm).
Co, Ni and Ru nanoparticles were prepared by reduction of the corresponding chloride precursors in ethylene glycol (10 g/L) (reductant for Ru) using PVP as a stabilizer as well as sodium borohydride as a reductant in the temperature range from 170 to 240oC for Ru NPs, from 23 to 170°C for Ni NPs and from 7 to 140°C for Co NPs. In addition to a typical polyol technique, Ru NPs synthesis using microwaves along with sodium borohydride reduction at room temperature was applied as an effective alternative approach. The effect of gas atmosphere, reduction temperature, the ratio metal/PVP, as well as of metal concentration in the colloidal solution, was investigated to explore regularities of metal NPs formation as well as to develop an approach to the synthesis of metal NPs with a controllable size. Reduction of Co, Ni and Ru precursors was found to be sensitive to air and had to be carried out under an inert atmosphere. A special attention was paid to a possibility of increasing Ru/PVP ratio compared to the classical polyol methods used Ru/PVP = 1/20-1/50 in order to simplify PVP removal after the final deposition of Ru NPs on the catalyst support. Decrease of reduction temperature and Ru/PVP ratio from 1/1 to 1/50 resulted in slight decrease of NPs size from 2.1 to 1.7 nm with a narrower size distribution while no clear effect was observed when the initial Ru concentration increased from 0.001 to 0.1 mol/L (Int. J. Nanotechnol. 2016, 13, 15) The oxidative atmosphere and increase of the reduction temperature was found to increase Ni and Co NPs size favoring NPs growth versus nucleation and clearly indicating different kinetics of NPs formation in the case of base and noble metals. Thus, the size of Ni NPs formed at 140°C in air was 5.4 nm compared to 3.2 nm for that formed in Ar. Co NPs synthesis required only inert atmosphere due to low stability and a fast reoxidation. The average diameter of Ni NPs size increased from 2.6 to 3.2 nm when the reduction temperature increased from 23 to 140°C, when the temperature was increased to 180 C the Ni NPs size increased to 9.8 nm. Similarly, Co NPs size increased from 1.8 to 2.6 nm with the temperature was increased from 7 to 23°C, whereas further changes from 23 to 100°C affected less significantly the NPs size (Int. J. Nanotechnol. 2016, 13, 4).
The microemulsion method was applied for Co and Ni NPs synthesis. As a stabilization agent and emulsion stabilizer dioctyl sulfosuccinate (AOT), cetyltrimethylammoniumbromid (CTAB) and Triton X-114 were employed. Microemulsion approach was studied in detail to establish the operating parameters for size-controllable synthesis of Ni and Co NPs. Effect of the metal precursor nature, the oil phase, the surfactant, the order of microemulsion blending and injecting, as well as water-to-surfactant ratio was explored from the viewpoint of practical synthesis. Colloidal Ni and Co metal NPs were prepared by the reversed microemulsion method using metal chloride and nitrate aqueous solutions (microemulsion I) and hydrazine hydrate as reductant (microemulsion II) in the presence of CTAB or AOT as well as n-octane and n-butanol or n-hexanol as oil phases. Mixing of two microemulsions and heating up to 80оС resulted in formation of series of Ni and Co nanoparticles with mean sizes varied from 2.6 to 3.3 nm and from 1.0 to 5.4 nm, respectively. Utilization of cationic CTAB provided formation of Ni and Co NPs with a narrow size distribution while anionic AOT resulted in formation of metal hydroxide instead of NPs according to UV-Vis. Increase of the heating time and decrease of the water-to-oil ratio gave a rise of NPs size. Probably solvent evaporation involved a drastic NPs aggregation process even if initially reversed micelles were stabilized in the solution as rather small NPs as a consequence of the intermicellar material exchange.
The scaling up of the lab synthesis routes was evaluated with the view of preparing large batches for partners and providing input for the evaluation of the synthesis routes at industrial scale. The first approach to scaling-up was to increase the reaction volume for colloidal synthesis to 500 mL, which was accomplished with excellent reproducibility of nanoparticle size and distribution for Pd, Pt and Re. Additional scaling-up was achieved in the case of the colloidal synthesis of Pt nanoparticles by increasing the concentration of precursor salt and PVP. Next efforts were centered in the reduction of the reaction volume employed to obtain the Pt colloids by increasing the concentration of the stock solutions. Reductions in the reaction volume of around 68% were achieved with only slight variations in the mean NP size (<10%) and in the NP size. The high viscosity of concentrated PVP stock solutions was a limiting factor for additional reduction of reaction volume. Furthermore, scaling up of Ru NPs polyol synthesis was studied both through the preparation of more concentrated Ru colloids or increasing the volume of the synthesized colloid. For Ru colloids with Ru/PVP = 1/5 it was found that 10-fold increase of Ru(3+) concentration from 0.01 to 0.1 M resulted in 1.3 fold increase of NPs size (from 2.1 to 2.8 nm), while at Ru/PVP = 1/10 even 100-fold increase of Ru(3+) concentration from 0.001 to 0.1 M colloid resulted in unproportional 1.5 fold NPs size increase (from 1.7 to 2.6 nm). Furthermore 20-fold increase of Ru colloid volume from 5 to 100 mL gave an increase of the average NPs size from 2.1 to 2.7 nm and from 1.7 to 2.6 nm for Ru/PVP = 1/5 and 1/10, correspondingly. Thus, it was demonstrated that required catalytically active metal NPs with controllable size can be prepared in bench scale quantities with a high metal/PVP ratio suited for further immobilization of prepared colloids on carbon supports. (Catal. Sci. Technol. 2016, 6, 8490).
The stability of Pd, Pt and Re nanoparticles obtained by colloidal synthesis was studied by aging of colloids and monitoring. Re colloids were found to be highly unstable and showed a tendency to reoxidize few hours after synthesis. Pd colloids showed a slight increase of the nanoparticle size 2-3 weeks after preparation, with an increase of about 11-13 % after four months. This tendency was more pronounced for the colloids with nanoparticles of higher size. In the case of Pt no significant increase in nanoparticle size was observed after an ageing time of two months. Colloidal solutions containing Ru, Pt and Pd NPs kept in fridge during 1-2 years were analyzed by TEM. It was shown that only small changes in NPs mean size and particles size distribution occurred during these periods indicating that developed methods of colloidal synthesis provided high nanoparticles stability. Thus, for PVP-capped Ru NPs average particle size dn was changed from 2.3 to 2.4 nm in 2 months, and from 2.3 to 2.7 nm in 24 months, for Pt-MeOH50-PVP10 the average particle size dn practically did not change in 18 months being 3.4 nm, for Pt-NaBH4 10-PVP 100 dn increased from 2.3 to 2.6 nm in 18 months, and for Pd-PdV3 dn changed from 2.0 to 2.1 nm in 20 months.
The immobilization of these well-defined nanoparticles on porous carbons and removal of the stabilization agent was studied in detail. The nanoparticle colloids were impregnated on different supports using water and methanol. The resulting catalysts exhibited poor surface area and low CO chemisorption, due to the blocking effect of the remaining PVP or AOT. Stability tests in water at room temperature showed some leaching of nanoparticles for all the supports (commercial activated carbons, CNFs, CDCs). The solvent used during impregnation and the chemical surface of the support influenced stability. Additional stability was achieved through a novel synthesis method of Pt/C catalysts with size-controlled nanoparticles based on in-situ synthesis of the nanoparticles, i.e. reduction with NaBH4 in the presence of a support and PVP (Catal. Sci. Technol. 2016, 6, 5196). Compared to the conventional ex situ route (colloidal synthesis followed by impregnation), this in situ route also yielded smaller nanoparticles (2.5–3.9 nm) of narrower size distribution. The method was also applicable to Pd/C catalysts synthesis. PVP-stabilized Ru NPs were immobilized on a mesoporous carbon Sibunit, applied as a reference, macroporous carbon nanofibers of platelet structure (CNF-Pl) and micro-/mesoporous TiC carbide-derived carbon (CDC) providing 1.7÷2.9 wt.% Ru/C catalysts with the mean Ru size 2.1÷2.7 nm. Since the presence of PVP on the catalyst surface drastically diminished activity in structure sensitive galactose to galactitol hydrogenation, different PVP removal and support modification methods were tested to elucidate the effect of support hydrophilicity/ hydrophobicity, preliminary support functionalization, additional Ru NPs washing prior to immobilization as well as PVP removal degree on catalytic behavior. Several approaches for PVP removal were applied such as solvothermal (with water and acetic acid aqueous solutions at 220°C) and thermal post-treatment (in air, argon, hydrogen or nitrogen at different temperatures) not resulting in noticeable changes in metal NPs size. For characterization of the carbon supports and synthesized catalysts TEM, XPS, XRD, XRF, water adsorption/desorption experiments and N2 physisorption were applied. Carbon supports were functionalized before Ru NPs immobilization by pretreatment with 5 wt.% HNO3, HNO3 conc., Ar (700oC), air (350oC), H2 (700oC) without visible support microstructure alterations. Textural properties of Ru/C catalysts purified from PVP did not correlate with their catalytic activity. The thermal post-treatment in air at 180°C followed by reduction at 250°C was found to be more effective in the case of Sibunit and CNF-Pl while solvothermal post-treatment in water at 220°C (PN2 25 bar) improved significantly the activity of TiC-CDC based catalysts. Untreated Sibunit and CNF-Pl carbon supports provided higher activity in galactose hydrogenation with Ru/Sibunit exhibiting the best catalytic activity being also the most hydrophilic according to water sorption isotherms. Catalytic activity of untreated micro-/mesoporous Ru/TiC-CDC increased noticeably depending on the support pretreatment in the row: untreated < 5% HNO3 < HNO3 conc. Additional TiC-CDC support functionalization was proposed to be required because of a relatively low amount of oxygen-containing groups on the surface compared to Sibunit and CNF-Pl. Carbon supported PVP-capped Ni and Co NPs were purified from PVP by both chemical washing (water or aqueous solutions of acetic acid) procedure and thermal post-treatment resulted in significant metal leaching, oxidation or sintering and thereby catalysts deactivating (Catal. Sci. Technol. 2016, 6, 8490).
Removal of PVP and AOT blocking the catalyst pores and the active surface was achieved by a novel method (Lemus et al., 2016) based in the activation of the catalysts at APR conditions (water at 200ºC, water+acetic acid at 200ºC). Metal loss was negligible for Pt/C catalysts while PVP was almost completely removed; hence, most of the porosity was recovered and the dispersion measured by CO chemisorption increased from 5 to 34–75%. Water at 200 °C was more effective than diluted acetic acid for the removal of PVP. The removal of the PVP at APR conditions also increased the stability of the catalysts due to improved metal-carbon interaction, showing negligible leaching of the metal phase. Likewise, the Pt nanoparticles did not undergo significant changes either in size or morphology. In the case of Pd this novel method could not be applied successfully, as sintering of the metal nanoparticles occurred and some leaching was observed. Massive leaching was observed in the case of Re based catalysts at APR conditions. The removal of AOT was also achieved by treatment at APR conditions, as evidenced by the recovery of the porosity of the catalysts, but CO chemisorption tests evidenced poisoning/blocking of the metal phase. The studies showed that with specialized posttreatments a high accessibility of the active sites can be combined with a high stability of the immobilized nanoparticles. Without this developed posttreatment active and stable catalysts cannot be obtained.
Furthermore, for the comparison several variations of catalysts were prepared by classical methods and on reference carbon material.
Main results on carbon support synthesis
Carbon supports with high inner porosity were synthesized from the partner FAU by the reactive extraction of carbides (CDC, carbide-derived carbons). Microporous material with pores below 1 nm were obtained from titanium carbide with extraction temperatures below 1000 °C. The specific surface area for this microporous and amorphous carbon is approx. 1600 m2/ g. Two different feeds of TiC were studied with approx. 2.5 μm and 50 μm average particle size. Interestingly, the resulting pore sizes differed slightly. Analysing the results more carefully reveals that the temperature onset for increasing pore size and crystallinity differs. Thus at 1200 °C the resulting pore diameter is 1.68 nm for the 2.5 µm fraction and 1.14 nm for the 50 µm fraction. A possible heat transfer limitation, which could lead to this phenomenon could be excluded. Thus, most probably different amounts of metals impurities (e.g. iron) are responsible for the difference.
To obtain mesoporous carbons it was envisaged to employ Mo2C as precursor. It was demonstrated that highly mesoporous materials with pore sizes starting at 4 nm are obtained when extracting Mo2C at 1200 °C. Nevertheless, these materials are not suitable for use in the experimental testing as the commercial raw material didn’t show the requested particle size. The resulting catalysts would lead to too high pressure drops. To obtain the envisages mesoporous carbons in correct particle size, the extraction temperature above 1200 °C was studied for the first time and for titanium carbide. Importantly mesoporous materials, with a narrow pore size distribution around 3.7 nm and a specific surface area of 450 m2/g resulted from titanium carbide at 1400 °C (Chem. Mater. 2015, 27, 5719). At lower temperatures the pore size can be varied in a bimodal micro- and mesoporous distribution. In parallel with increasing temperature the materials crystallinity changes from amorphous to more graphitic in the new temperature regime for 1200 to 1600 °C, resulting in graphite crystals approx. 20 nm in diameter and above 10 stacks of graphene. The porous carbon inner surface could be functionalized with oxygen by different post treatments. Thereby the amount of oxygen and the change in pore structure increases for the different reactants in the following order: air < sulfuric acid < nitric acid. As a recommendation the sulfuric acid treatment is suitable for amorphous and the nitric acid one for more graphitic carbon. The oxygen content shows a direct influence on the water interaction and with increasing content the hydrophobic character changes to a hydrophilic one. It is noteworthy that the liquid functionalization with sulphuric acid and nitric acid leads to an elevated sulphur or nitrogen concentration, respectively. These results can be either explained with remaining amounts of the corresponding acid in the pores or the formation of sulphur/nitrogen containing groups during the functionalization. As the materials were washed intensively to remove the acids, most likely sulphur and nitrogen containing surface groups form. In summary for all materials suitable post treatment conditions were identified, where some oxygen groups are introduced to break the hydrophobic character, but the specific surface are and pore site distribution is not altered.
Regarding carbon supports with mainly external surface area, the partner FC was successful in establishing a synthesis route for high quality carbon nanofibres (CNF) showing a specific surface area of approx. 125 m2/g. Specifically so called platelet type CNFs were produced. Quality of CNF morphology was determined in four different aspects. First is the selectivity towards platelet type fibers and avoiding screws type ones. As the screws expose no graphene edges, they a supposed to stabilize the Pt to a lower amount. The second aspect of the morphology quality was a distribution of nanofiber diameters in the synthesized CNF-PL when a smaller range of CNF-PL diameters was considered as a sign of better quality. A third aspect is crystallinity or specifically additional amorphous carbon deposits. A forth important aspect is CNF nanomaterial purification from CNF-synthesis catalyst embedded in the carbon support. Within the project the preparation procedure and CNF synthesis catalyst preparation and handling could be optimized to finally satisfy all 4 criteria and produce highly selective platelet type CNFs, being very crystalline and showing a narrow diameter distribution around 150 nm. The purification in non-oxidizing acids was optimized to allow full removal of the metal and metal oxide catalyst used during the synthesis. The CNF can either be produced as dry powder or for improved handling as wet paste of known water content. Optimized material was produced on larger scale.
Main results on in-silico studies
To provide insight with calculations the pathway to carry out simulation studies on the APR was paved. Therefore, exchange-correlation functionals were calibrated and tested for their accuracy for Ni, Ru, Pd, Re, Pt surfaces. It was assured that the structural features and adsorption energy for water monomer can be described accurately.
The co-adsorption of polyols with a water molecule was studied on bare Pd(111), Re(0001) and Ru(0001) surfaces. The adsorption mode which was selected for this investigation can be considered representative of early stages possibly occurring in metal-catalyzed reactions in water. On the bare metals, the adsorption geometry is always characterized by one hydroxyl group interacting atop with the O-H bond parallel to the metallic surface, the residual fragment pointing outward. The three metals show different features with respect to the interactions with the polyols, being the adsorption energy smaller for Pd and quite similar for Re and Ru. In the presence of water co-adsorbed on the metal surfaces, two different configurations were found for the different polyalcohols. The occurrence of a given configuration for a specific polyol seemed to be determined by a balance of competing and cooperative energetic factors, which were related both to the polyol size and to the metal nature. In particular, the investigated polyols are strongly stabilized when adsorbed in the presence of water on Pd, whereas when the co-adsorption takes place on Ru and Re this effect is downsized, very likely due to their larger oxophylicity as detailed in (J. Phys. Chem. C 2015, 119, 17182).
Furthermore, a study on the nucleation of Pt, Pd and mixtures with Ni and Ru on graphene with and without vacancies. Furthermore, the adsorption of polyols (up to C4) on Ni, Ru , Pd and Re could be simulated. In particular, the nucleation of homonuclear (Ni, Pd, Pt, Re) and heteronuclear clusters (Ni-Pd, Re-Pt) on a graphene monovacancy was investigated; this study was focused on the analysis of the magnetic, energetic and structural features of the defective graphene during the trapping of the metal atom and the subsequent growing of the homo and hetero metallic cluster. It emerges that this event is regulated by the different affinity of the metal atoms to both the monovacany and the pristine graphene, the metal bulk cohesive energies and to the relative stability of the nucleation seeds (J. Phys. Chem. C 2016, 120, 12022).
Atomistic insights and detailed reaction mechanism were deduced for the reforming of 1,2-propandiol (1,2-PDO) on small Pt clusters. 1,2-PDO was selected as model substrate for polyfunctional oxygenated molecules present in the biomass based feeds. The reactivity of 1,2-PDO was investigated selecting as reliable model of a Pt catalyst, a cluster composed by thirty atoms opportunely cut out from the fcc platinum bulk. The reaction mechanism for the decomposition of the 1,2-PDO, was devised and fully characterized by means of ab-initio methods. The resulting reaction network formed by C-H, O-H and C-C bond scissions can be found at the project website (http://susfuelcat.eu/files/SusFuelCat_Conference_Project_Posters.pdf; UNIPA poster) and is also under publication by Duca et al.
The energy barriers and stabilities given in the network allow to discuss the main reaction pathways and possible strong adsorbing intermediates, which could poison the catalyst. Comparing the in-silicon and experimental results, in experiments hydroxyacetone is the main product, which agrees with the low activation barriers calculated for the two dehydrogenation reactions leading to this species. Additionally, in the experiments significant amounts of propanal were observed. The presence of this aldehyde suggests that C-O bond cleavage may occur at the early stages of the reaction, before the C-C bond cleavage. Therefore, for selected intermediates with a low degree of dehydrogenation, the activation barriers for the C2-O2 bond cleavage, leading to precursors of propanal, were calculated. According to this calculations, C-O cleavages should not occur at the first stages of the reaction. Moreover, it has also to be considered that various examples in the literature are reported demonstrating that during the reforming on metallic surfaces the C-O bond cleavage is not competitive with other processes.
In a similar manner, the reforming of ethylene glycole (EG) on Pd sub-nanometric catalyst was studied. The deduced reaction network is also given at the same poster as one paragraph prior. Moreover, this system allowed us to explore by means of ab-initio techniques all the reactive channels associated to the ethylene glycol decomposition reaction; this implied the construction of a relatively large grid of possible events that enabled to deepen also the issue concerning the energetic ordering of reactive pathways. The whole reaction network for the decomposition of ethylene-glycol was outlined and, within a graph theory based approach, all the pathways starting from C2 and leading to CO and H2 were analyzed and compared, making possible to draw a comprehensive picture of the competitive routes that enable the production of hydrogen on a sub-nanometric palladium cluster. In detail, the protocol for the reaction network analysis developed can be divided into three steps: i) ab-initio calculation of the activation barriers and pre-sieving of the molecular events, ii) construction of the graph representing the final reaction network, iii) analysis of all the possible pathways.
Main results on hydrothermal stability
Carbide-derived carbons (CDC) and carbon nano fibers (CNF) produced within the project as also commercial activated carbon and carbon black were studied for their stability during hydrothermal treatment. In situ experiments were carried out in water vapor and TGA. Post mortem characterization (XRD, TPO and N2-sorption) was carried out for experiments where material was treated at 240 °C in the liquid phase for 7 days and characterized post mortem. The results show that the SusFuelCat carbons (CDC, CNF) and the carbon black are highly stable against hydrothermal attack, while the commercial activated carbon, showed a pronounced lower stability. Stability against water vapour or oxygen was studied in a temperature programed oxidation, hence with contiguously increasing temperature. Assessing the stability is carried out by comparing the onset temperature for oxidation.
It was shown that the onset temperature of decomposition in oxygen increases with the crystallinity of the materials from 480 °C to 675 °C (Chem. Mater. 2015, 27, 5719), showing the huge influence of the carbon support type on stability. The water vapour studies gave in principle the same trend, but the stability is increased to higher temperatures and above 600 °C. The amorphous activated carbon however show a continuous mass loss also till 600 °C and not a very sharp on-set of decomposition. The stability test which is most close to the APR reaction conditions was a 7 day hydrothermal stress test in an autoclave at 240 °C. Here no in situ data is obtained but post mortem characterization carried out. After this stress test the amorphous activated carbon showed a reduction of its specific surface area (initial 900 m2/g) by 20 %. Most likely surface functional groups were build, which increase the mass of the material. Graphitic materials like CNF or carbon black didn’t show a similar loss in specific surface area, but have in general a lower specific surface area below 200 m2/g from the beginning. Interestingly amorphous carbide-derived carbon with an interestingly high surface above 1600 m2/g didn’t show a reduction of specific surface area. More graphitic CDC also shows no surface area loss, but has interestingly a starting surface area which is comparable to the commercial activated carbon, while being not amorphous. Thus CDCs are a highly interesting catalyst support for reactions carried out under hydrothermal conditions, where stability can be an issue.
Due to the already very good stability, posttreatments of the carbon supports were only studied to a minor amount. In general oxidative methods, which remove the amorphous carbons and leave the more graphitic one behind were studied. Materials studied were the amorphous activated carbon, amorphous CDC and graphitic CNF. Air oxidation at 350 and 425 °C als also a liquid phase oxidation with nitric acid was studied. For the air oxidation at 350 °C no mass loss and thus change of any materials was observed. Air oxidation at 425 °C gave a mass loss of approx. 15 wt.-% for all materials. Despite this removal of less stable carbon the post treated materials didn’t show an increased oxidation resistance, but rather a slightly reduced one. Most likely during the oxidative post treatment slightly additional defects are introduced, which are starting point for later attacks. Thus, no increased, but slightly decreased stability results. Furthermore, it was observed that due to the post treatment, some surface groups were newly introduced. Within the project we demonstrated that Pt/C catalysts with a higher surface acidity favor alkane production in the aqueous phase reforming of xylitol while the ones with lower acidities generate hydrogen with high selectivity and turn-over-frequency (Catal. Sci. Technol. 2014, 4, 387). Thus, these additional surface groups are unwanted and no post treatment of the carbons is recommended.
To study the possible influence of the metal in the final catalyst on the stability, procedures to deposit high amounts of Pt but with comparable particle size needed developed, as it was assumed that this pronounced higher loading would accelerate the stress test. Starting from chloroplatinic acid dissolved in aceton a solution 2 – 3 time more than the pore volume was added to the carbon samples. Intense ultrasound was applied to support a homogeneous infiltration of the solution. After drying a reduction procedure was applied, which either was relatively harsh and started directly at 300 °C with H2 or was more mild and with H2 already during the heating ramp till 300 °C. Loadings ranging from 15 to 20 wt.-% could be achieved at particle sizes ranging for the mild reduction from 2.9 to 4.5 nm. The harsh reduction gave unsatisfactory sizes from 5.1 to 36 nm. The materials were employed in a 7 day hydrothermal stress test at 240 °C. Characterization of the stressed material showed that the Pt loading reduced only slightly e.g. from 18.2 to 17.7 wt-%. This is most likely related to a fast leaching of not completely reduced Pt precursor and not due to leaching of metallic Pt clusters under hydrothermal conditions. Pt particles sizes as also the specific surface area remained unchanged within the stress test. Comparing the onset for oxidation with and without Pt the temperature where decomposition starts lowers by approx. 100 °C as the decomposition is catalyzed by Pt. Nevertheless, still a very high and sufficient stability and onset of oxidation of approx. 400 °C was achieved for the graphitic carbon supports.
Main results on catalytic experiments
To obtain feedback from catalytic experiments three continuous set-ups were reconfigured and set into operation to study APR of i) C5 and C6 polyols in lab scale, ii) APR of C1-C3 oxygenates in lab scale and iii) the long term APR of various feeds in validation scale.
Testing of unsupported colloidal particles faced several technical difficulties, such as a need for a dedicated set-up and an unclear operational process. Unsupported colloidal particles are notorious for their ability to deposit on all metal parts of reactors during catalytic experiments. Not all parts of available batch reactors can be cleaned in an appropriate way after each experiment or alternatively protected. Moreover, separation of unsupported colloidal particles from the liquid phase is also challenging. Thus, neither HPLC, nor TOC analyses can be directly performed due to high concentration of metals in the liquid phase. Another challenge is related to an accurate carbon balance determination since protecting agents PVP and PEG are reactive during APR contributing to generation of carbon containing products and hydrogen. All these complications lead to a high uncertainty in assessing the activity and selectivity of unsupported colloidal particles and finally inability to obtain reliable information. Therefore catalytic data given below were generated using m supported nanoparticles prepared either by classical impregnation or by immobilization of colloids.
After comparison of different metals as catalytically active phase it can be stated that all monometallic non platinum catalysts displayed significantly lower activity (Ni) or deactivated much faster (Ru, Re) than monometallic Pt. Activity of bimetallic Pt-Ni, Pt-Co and Pt-Ru catalysts was practically en par with the monometallic counterpart. The main difference was seen for the Pt-Re catalyst, which had the highest activity level among the tested catalysts. For Ni and Pt-Ni catalysts, leaching of Ni was observed. Selectivity towards hydrogen was the highest with Pt not increasing after addition of other metals. It should be noted that elucidation of catalysts prepared from colloids is not straightforward as their performance depend on the immobilization and capping agent removal procedure. Some catalysts in particular were inactive because of inefficient capping agent removal, while others displayed strong metal leaching as immobilization was not efficient. Finally after careful optimization APR active and stable catalysts have been developed. For these colloidal based materials post-catalysis characterization by TEM showed a slight increase in Pt size, although no metal was lost during the reaction. N2 physysorption also showed that PVP was removed from the support/metal.
No clear trends related to possible mass transfer limitations were seen while studying particle sizes of catalysts varying from < 100 µm to above 1 mm or even extruded pellets. A more detailed characterization revealed that larger sized supports tended to have an egg-shell distribution of the active metal, which probably minimized or even eliminated pore diffusion limitations. The overall picture emerging form the experimental data is that there are indications that an egg-shell distribution is advantageous, even if a solid proof is missing. Studies with unsupported high surface area as well as supported Pt catalysts showed, that carbon-platinum interactions seem to be crucial for the desired activity and selectivity.
An in situ ATR-IR spectroscopy study was carried out for supported Pt catalysts during APR of hydroxyacetone. Carbon as a support could not be employed for the study, but degradation of oxide supports was monitored in-situ. Degradation of alumina under the hydrothermal conditions was obvious. Furthermore, strong adsorption of hydroxyacetone on the surface of zirconia was detected, causing formation of deposits and catalyst deactivation via aldol condensation. Thus, a proper choice of the support could be proven to be highly important to achieve excellent performance of APR catalyst.
Variation of the flow rate, hence, the residence time, showed that the hydrogen yield increases initially with increasing the residence times and thus degree of conversion. Thereafter it reaches a maximum and decreases again. This behavior is a direct indication that hydrogen is consumed in consecutive reactions and thus an optimal degree of conversion should be aimed for.
The influence of chirality of the feed was studied in detail by comparing sorbitol, galactitol and their mixture, showing the same activity and selectivity patterns. Comparison between C5/C6 (xylitol/sorbitol) and C3 feeds (hydroacetone) revealed that the ratio of reactivity in respectively water gas shift and APR is lower for the latter feed.
For xylitol, sorbitol and galactitol structure- property relationships could be deduced for Pt/C. It was proven that the carbon supports are well suited for tuning the hydrogen and alkane selectivity. For high hydrogen selectivity, the carbon support should show a neutral character, whilehigh alkane selectivity requires an acidic function meaning that surface functionalization determines the selectivity. While activated carbons due to their synthesis procedure very often exhibit a non-negligible surface functionalization, resulting in formation of alkanes, other carbons supports such as CNF or CDC displaying excellent purity result therefore in high hydrogen selectivity. Furthermore, indications were found that for bulky polyols the reaction is structure sensitive, hence the turnover frequency of Pt increases with an increase of Pt cluster size. As a consequence of a larger fraction of Pt not available for catalysis when the size is increasing, there is an optimum for the active metal size exhibiting the highest reactivity.
Characterization of as prepared and spent catalysts was carried out. Colloidal Pt and Ru nanoparticles as precursors for immobilization on carbon to obtain heterogeneous catalysts proved to be stable for 1 to 2 years. Ru catalysts prepared by a colloidal route (PVP stabilized) and immobilized on carbon nanofibers or carbide-derived carbon were employed in galactose hydrogenation. High resistance to leaching and sintering under hydrogenation conditions was observed independent of the PVP removal procedure. Ru, Pt, PtNi, Re, PtCo and PtRu on carbon support were applied in aqueous phase reforming. No pronounced sintering or leaching of was detected under APR condition for the noble metals, contrary to Ni and Co. In addition to leaching and sintering, noticeable oxidation was seen for all metals, except Ru.
Based on the knowledge acquired in the project the following catalyst properties were identified for high performance: i) supermicroporous to mesoporous texture (>0.7 nm); ii) non acidic support; iii) sufficiently hydrophilicity of the support; iv) Pt sizes around 2 - 5 nm. Additionally, minimization of mass transfer limitations through application of egg-shell catalysts and a decrease of pressure drop by a proper selection of the particle grain shape and size are beneficial.
Based on the generated experimental knowledge o a recommendation for standardized testing of APR catalysts among different labs was established for both batch and continuous reactors. For the standardized testing, a mixture of 5% sorbitol and 5%mannitol in water should be chosen.
The catalyst should be activated upon heating under H2 at 375°C for a minimum of two hours under atmospheric pressure. Subsequently, the standardized testing would be carried out in the batch reactor at 30 bars pressure using N2, 225°C, for 2h, followed by an analysis of the gas phase by GC and the liquid phase by HPLC and elemental analysis. Note, that for the batch reactor continuous sampling can be critical as the carbon balance is not fully closed for high sampling amounts, due to the gas phase loss. For continuous operation 30 bar N2 with a N2 flow of 0.5Nl/min should be realized at 225°C, sufficient stabilization time prior to data collection, gas analysis by and the liquid phase by HPLC and elemental analysis.
Main results on process validation
In long term studies with Ru/C catalyst for polyol production large quantities of glucose and sucrose were converted to technical sorbitol (both 50 wt-% in water). At a degree of conversion of 96 % a yield of 92 % could be realized and in total more than 40 kg of feed were produced. No major catalyst deactivation was observed for a total >400 h time on stream. Nevertheless, some solid formation was observed, which can lead to a pressure increase. Probably small amounts of sugar monomers are converted to char like products by side reactions. Depending on the feed crystallization can occur and be avoided by slight heating of all process lines and during storage by diluting the product to 29% wt-% in water.
Long term studies on the APR of model compounds and technical sorbitol were carried out in combination with a temperature variation from 220 to 250°. Normally no induction period was observed in these studies. Opposite results were obtained for a Pt/C catalyst on a special carbon support with very low loading (0.5 wt.-%) and reduction perfomed within the reactor rather than externally prior to the reaction. Here continuous activation was also seen as continuous changes of CO2 to CO ratio, representing the ability to perform water-gas shift reaction, even after 150 hours of time-on-stream. Thus, changes of the active sites seem to take place during this long induction period. Typically, such changes in catalytic behavior are associated with solid-phase transformations of the active phase, for example reduction or oxidation. Note, that the catalyst was pre-reduced at a high enough temperature to ensure complete reduction in line with TPR data, even if the reduction was carried out directly in the reactor. TEM data confirmed almost no changes in the cluster size of Pt, thereby excluding sintering. Finally, within the project it could not be unequivocally concluded in which cases the catalysts would show a long induction period. At the same time such induction period is not needed to achieve an active and selective catalyst.
The long term APR studies were also employed to deduce possible deactivation of the catalysts. In this sense a clear difference for model and technical feedstock was observed. For Pt/C on activated carbon and with model feedstock only a slight deactivation was observed and the degree of conversion reduced from 66 to 60 % after 250 h time on stream. On the contrary employing strong catalyst deactivation was obvious for the technical feedstock. Most likely such deactivation is related to blocking the active sites with reactants/intermediate products of these more concentrated feedstocks (30 to 50 %), while the model feeds were only in 10 % concentration. It is expected that dilution of the technical feed can minimize deactivation. Reactivation of the catalysts through extraction/flushing was studied. Flushing with water is not sufficient to restore the activity. Reactivation procedure with an organic solvent (acetone) allowed removal of organic residues blocking the catalyst and regaining catalytic activity after reduction in hydrogen.
Four kinetic models for xylitol and sorbitol APR, which serve different purposes, were developed. One mechanistic model accounting for consumption of hydrogen in consecutive side reactions was used chemical reaction engineering simulations allowing to deduce hydrogen selectivity profiles. Modelling of mass transfer showed that with ca. 1 mm particle diameter, pore diffusion limitations might occur lowering activity. Pore diffusion limitation has a minor influence on the resulting hydrogen selectivity. Variation of the WHSV showed that hydrogen selectivity shows a maximum with the residence time. Beyond this maximum the selectivity drops only slightly, therefore losses of hydrogen are small at too high residence time. Analysis of the reactor hydrodynamics (residence time behavior) showed that pronounced back-mixing leads to a loss of the maximum in hydrogen selectivity vs. residence time. Thus ca. 20 % of hydrogen selectivity can be lost. Most interestingly, simulating a continuous hydrogen removal along the reactor through a membrane, showed that the total amount of hydrogen obtained can be increased by a factor of 3, as a result of efficient suppression of side reactions involving hydrogen. A sufficiently high hydrogen removal through the membrane is needed to obtain this significant increase in hydrogen yield.
While the calculations above used a kinetic model capable of accurately describe hydrogen selectivity, flow sheeting accounting especially for separation needs a detailed description of all intermediates. Thus, a model based on a detailed reaction network involving the key intermediates, excluding however detailed kinetic information, was employed in a so-called stoichiometric reactor option. The yield of the products and side products was set by coefficients, according to the experimental results. Within the framework of this model, heat and mass balances calculations as well as costs estimations were carried out for a 500 kg/h hydrogen production plant operation with sorbitol syrup as a feedstock in APR of sorbitol using Aspen HYSYS software. For reactor modelling a complex reaction network was taken into account along with phase equilibrium simulations which determined to have a significant impact on the total process heat due to possibility of water evaporation. Improved economy could be achieved e.g. through incoperating a medium pressure steam generator after a high temperature separation, despite a low temperature separation without steam generation. The total costs of hydrogen were estimated as 9.4 $/kg, where the feedstock costs take the major contribution of 86%. Thus, the cost of hydrogen from APR can be in an economic range if the feedstock prices drops below 150 $/kg.
Based on the results of the project and expertise of the industrial partners an assessment of the catalysts and process was carried out. From an economic point of view, the metal supported catalysts can be sorted in two groups. i) Expensive noble metals (Pt, Pd, Rh) where the metal price dictates the final catalyst price. Here recycling of the metal is essential to lower the price. ii) Cheaper metals (Ru, Re, Ni), where also process and support costs can play a significant role. Pt shows by far the best performance and selectivity towards hydrogen, thus high costs cannot be prevented. Thus, also carbon is due to the recycling possibility by far the best option, as the Pt recovery rate is very high and thus the Pt price can be dropped pronouncedly through the recycling. For the special catalysts developed in SusFuelCat additional costs could stem from the newly developed colloidal preparation route and for the costly carbon supports (carbide-derived carbons, carbon nanofibres). Thus, it is advised to employ the new colloidal preparation route especially for expensive metals of the first group, where the production process costs are less relevant. The colloidal route, nevertheless, can also be employed predominantly to guide catalyst development, while the final catalyst would be produced by more economically viable classical impregnation procedures
Nanomaterials (carbon support and catalytically active metals) in general can be of potential risk, thus full health and safety studies might be necessary in the future. To minimize the costs of employing more expensive advanced carbon supports, the strategy was developed to coat cheap and readily available carbon pellets with these advanced carbon materials. Additional benefits of these coated carbon pelletized catalysts are related to a decrease of pressure drop, which would be high otherwise with the small sized carbide-derived carbon or carbon nanofiber grains. The additional costs for the preparation were estimated to be €150-175/kg. With such approach affordable prices of Pt catalysts can be achieved especially if significant advantages of these special supports in terms of higher selectivity arise.
Potential Impact:
1. Potential impact of SusFuelCat
The consortium partners of SusFuelCat found proper APR catalysts for renewable hydrogen production. Important steps were carried out during the project towards this aim. Important indications for structure activity relationships were deduced. New possibilities to tune active and stable metal catalysts were developed and can be of high value also for other catalytic problems. The experimental data and simulation allowed deducing further mechanistic insights, which will be important for the catalyst and process design. First cost estimations show that APR hydrogen can be economically competitive. Thus it is believed that APR technology can be of importance for sustainable hydrogen production in Europe.
Besides this direct impact on the horizon, already further impact becomes apparent for each partner. The pronounced knowledge on well-controlled colloidal based catalysts, methodologies for immobilization and stabilization agent removal, synthesis of high quality carbonaceous materials, great progress on the simulation of the systems, and developed experimental protocols have the potential to be of great use in catalysis and neighboring disciplines in general.
2. Socio-economic and societal impact of SusFuelCat
Especially for the three industrial partners in the consortium a direct economic impact can be expected. The two SMEs within the consortium are BTG Biomass Technology Group and FutureCarbon. BTG is specialized in process development for conversion of biomass into biofuels and bio-energy via flash-pyrolysis. FutureCarbon is specialized in development, production and refinement of carbon nanomaterials and carbon nanomaterial based products. Both SMEs profit directly from the results obtained.
The established industrial partner is Johnson Matthey, which is a specialty chemicals company, focusing on catalysis, precious metals, fine chemicals and process technology, with extensive knowledge and experience of carbon supported noble metal catalysis. With the combination of innovative SMEs and an established global industrial partner it is expected that a transfer of the project results can become true.
i. health and well-being
The use of fossil fuels in road transport, aircraft and industry applications releases different greenhouse gases and particles into air and environment. Hydrogen, as produced in SusFuelCat, instead serves as an alternative green fuel. Therefore, investigations focusing on effective production of alternative fuels like hydrogen are a possibility to contribute to a healthier environment, especially in cities and industrialized regions.
ii. employment
For all three participating companies BTG, FutureCarbon, and Johnson Matthey, the work at SusFuelCat was important to find new products (catalysts) and processes that can be commercially exploited in order to maintain or increase their number of employees.
iii. training and career development
Several young scientists worked at SusFuelCat. They learned about chemistry, process engineering, built-up and run of experiments, writing papers, presenting the results at conferences, and even about teaching young students. Finally with incorporating their findings to their dissertations they got trained for going ahead in their career.
All seven participating universities from Erlangen, Darmstadt (Germany), Enschede (Netherlands), Madrid (Spain), Turku (Finland), Palermo (Italy) and Novosibirsk (Russia) joined in carefully training the young scientist.
3. Main dissemination activities
Together with publications SusFuelCat had 126 dissemination activities, of which the most notable are described below:
i. Publications
In total there have been 13 peer review publications; among them two have been published open access via OpenAIRE.
Exemplary three publications are discussed. The article “Preparation of carbide-derived carbon supported platinum catalysts” in Catalysis Today, Vol. 249, (2015) page 30 disseminates important results on novel nanoporous carbons and their used in catalysis. The publication “Controlled synthesis of PVP-based carbon-supported Ru nanoparticles: synthesis approaches, characterization, capping agent removal and catalytic behavior” in Catalysis Science and Technology, Vol. 6 (2016) page 8490 disseminates and summarizes important findings on the synthesis of well controlled nanoparticles as also the important removal of stabilizing agent for the application in catalysis. In the publication “Aqueous-phase reforming of xylitol over Pt/C and Pt/TiC-CDC catalysts: catalyst characterization and catalytic performance” in Catalysis Science and Technology, Vol. 4 (2014) page 387 disseminates structure-activity relationships identified for the APR reaction.
ii. Conferences and accompanying activities
The consortium participated on several conferences and presented the project to scientists, industry and to the wider public. There have been 48 oral presentations to a scientific audience, which have been accompanied by 41 posters presented at these conferences. Besides the scientists, also the wider public has been delivered information by publishing twelve press releases, one oral presentation, and three articles to popular press and audience. This has been completed with setting up a project website, which was constantly updated during the project to keep the stakeholders informed.
As proposed in the DoW, SusFuelCat partners held a scientific conference subsequent to the final project meeting in Enschede, Netherlands organised by BTG Biomass Technology Group and University of Twente and held at the campus of the latter. The scientific partners and BayFOR further contributed substantially to the CarboCat conference in Strasbourg, France. At both conferences, the consortium had the possibility to exchange ideas with external experts or to find applicants for project results.
SusFuelCat partners visited two large conferences, EURONANOFORUM and Industrial Technologies, which are both sponsored and co-organised by the EU Commission. Here, they presented the project to a wider audience, as to participants from other EU projects, stakeholders from industry and academia. As for all the other events, posters, flyers, roll-ups, and gadgets were presented at booths. Through in-depth bilateral discussions more than 30 valuable contacts were acquired.
iii. Clustering with other EU projects (@Bastian: please mention your experiences ...)
In 2014 SusFuelCat was invited by the EU Commission to take part in the cross-cutting EU Clusters on “Catalysis” and "Engineering & Upscaling". The SusFuelCat coordinator was actively involved within the European Cluster on Catalysis (http://www.catalysiscluster.eu/). The subgroup “Advanced analytical approaches in catalysis, in situ and in operando studies” was headed by the coordinator and input for the roadmap on catalysis was given. On the Engineering & Upscaling activities the coordinator gave input through a survey.
4. Exploitation of results
i. Key exploitable results
Key exploitable result No 1
Short Title: High pressure FTIR-ATR
Full Title: Method for carrying out ATR studies at elevated pressure. This allows for in situ studies not only in this reaction
Lead Partner: UT, partners involved: UT, UNIPA
Brief description: Equipment and knowhow are available at UT to carry out in situ FTIR-ATR studies at elevated pressures. Together with the simulation competency of UNIPA complex spectra can be evaluated.
Key exploitable result No 2
Short Title: Simulation-package
Full Title: A good package of computational tools developed
Lead Partner: UNIPA , partners involved: UNIPA.
Brief description: Simulation code for adsorption and reaction studies in aqueous phases were developed. Also code to study metal cluster development on graphene is available. The code can be used for studies and further refinement at UNIPA.
Key exploitable result No 3
Short Title: CNF-PT-upscale
Full Title: Up-scaling the production of CNF-PT nanomaterial without loosing the high quality
Lead Partner: FC, partners involved: FC, FAU, and JM (exploitation partner)
Brief description: Steps for scale-up of CNF-PL synthesis from lab scale to pilot plant scale. 1st step: More CNF-PL from concrete quantity of synthesis catalyst, achieved by process optimization (e.g. gas flow turbulences); scaling factor ~1.3) 2nd step: Planning and dimensioning: Single plug-flow reactor --> bundled tube reactor (scaling factor ~6)
Key exploitable result No 4
Short Title: CDC-High-T
Full Title: Mesoporous Graphite as stable, highly porous, electrical and thermal conductive material
Lead Partner: TUDA, partners involved: TUDA
Brief description: Mesoporous Graphite can be produced, which is interesting as chemically very stable and highly electrical and thermal conductive material; The knowhow is used in follow up research projects.
Key exploitable result No 5
Short Title: CDC-Synt
Full Title: Synthesis procedures and strategies for carbide-derived carbons
Lead Partner: TUDA, partners involved: TUDA
Brief description: Synthesis procedures and strategies for production of carbide-derived carbons; The knowhow is used in follow up research projects.
Key exploitable result No 6
Short Title: Colloidal-NP
Full Title: Synthesis procedures and up-scaling for well designed colloidal particles
Lead Partner: BIC, UAM; partners involved: BIC, UAM;
Brief description: Identification of different synthesis conditions leading to nanoparticles of different metals (Pt, Pd, Re, Ru, Ni, Co) with controlled size and a broad range of sizes. Re and Ru NPs are difficult to obtain. Conditions for the reduction and for the growth of large NPs have been identified. The nanoparticle colloids prepared by different methods (PVP colloids, polyol, microemulsion) provide different types of interactions with carbon supports. The size control is a key for a better control of NPs performance. Colloidal synthesis can be scaled up, is reproducible and NPs are stable (Pd, Pt). The concentration of reagents can be increased to reduce the reaction volume and make scale up easier.
Key exploitable result No 7
Short Title: Immo
Full Title: Immobilization procedure for colloidal nanoparticles by simulated APR conditions
Lead Partner: UAM, BIC, partners involved: UAM, BIC, FAU, FC, AAU, UT
Brief description: Identification of synthesis conditions for nanoparticles and catalysts that promote interactions between nanoparticles and carbon supports. Enhanced stability and activity of the catalysts prepared following the procedure proposed. The procedure can be applied to other metals and other supports not studied in this project. The distribution of the nanoparticles within the carbon porous structure can be controlled.
Key exploitable result No 8
Short Title: Mini-APR
Full Title: Up-scaling the lab APR process. Probably also making mini-APR-reactor of lab size or a little big bigger commercially available, which could be used for feedstock screening at interested partners.
Lead Partner: BTG, partners involved: BTG, AAU, UT
Brief description: Upscaling of APR/LPR process for converting liquid biomass feeds, including pyrolytic sugars into platform chemicals and fuels (alkanes, hydrogen).
Key exploitable result No 9
Short Title: PVP-Removal
Full Title: PVP-Removal – Method for effective removal of PVP stabilizer from nanoparticles, not suffering of dramatic particle size increase or metal loss.
Lead Partner: UAM, BIC, partners involved: AAU, BIC, JM, UAM, UT
Brief description: Currently known APR catalysts are based on metal oxide supports. The SusFuelCat work has shown applicability of carbon supports. These enable easier recycling and higher recovery of the precious metals used for the catalysis. The concentration of the catalyst during the removal and removal time can be increased to reduce volume and make scale up easier.
List of Websites:
www.susfuelcat.eu