Final Report Summary - MACADEMIA (MOFs as Catalysts and Adsorbents: Discovery and Engineering of Materials for Industrial Applications)
Executive Summary:
A major challenge facing European industry involves the development of more specific, energy saving processes with less environmental impact. The recent development of Metal Organic Frameworks (MOFs) may prove a major milestone in achieving these goals. The MACADEMIA project is an extension to the FP6 project DeSANNS, which highlighted some MOF materials for CO2 capture and storage. It expanded and continued this work on a much larger scale, with additional industrial applications: TOTAL’s focus is on bringing applications of MOFs to key market sectors - gas separation and storage, and liquid phase separations. The TOTAL-led consortium, with Total Petrochemicals and Refining-Marketing as industrial partners, 11 academic partners from across EU, one leading South Korean partner, among world leaders in their particular domain of MOF science, contributed enthusiastically to the project, with the constant and dedicated support of the management partner.
MACADEMIA aimed at producing new MOFs and optimizing those already of promising interest, characterizing MOFs using specialised techniques, testing MOFs using a three-tiered process, using predictive modelling and demonstrating the use of MOFs in key industrial processes. More specifically, the main targets of the project were:
- separation processes in gas/vapour phase (propene/propane, acid gases separation, CO2 and hydrogen purification);
- liquid phase separations (xylene separations, recovery of N- and/or S-compounds from hydrocarbons);
- catalysis for specialty chemicals (Lewis-acid MOFs as catalysts for epoxide polymerization, redox-active MOFs as catalysts for hydrocarbon autoxidation).
While most of the deliverables contributed to expand considerably the knowledge on synthesis, characterization and industrial applications of MOFs, some limitations were encountered in some specific domains, such as:
- Catalysis: in epoxide polymerization, the activity of theMIL-101(Cr)nano MOF in consecutive polymerization runs drops significantly. Cyclooctane autoxidation has a moderate conversion rate, 25 g cyclooctane per g of catalyst: this number needs to be increased several orders of magnitude in order to become a competitive process. However, this is a minimum productivity number, since it would be possible recover, reactivate and reuse the catalyst.
-Purification of fuels by S- and N-adsorption on the Cu3(BTC)2 MOF : All the precautions have been taken to perform the evaluation of the performances of the shaped MOF supplied by KRICT (glove-box, repeatability of the tests,…). However in the standard operating conditions traditionally used at Total Research&Technology Feluy (Belgium) to qualify the sorbent batches, the performances of Cu3(BTC)2 were not conclusive.
Overall, 97 scientific, peer reviewed articles were published in high impact journals in total over the three years of the project. Three partners, FEUP, TOTAL, and KRICT filed jointly a patent for a process aiming at recovery of nitrogen or recovery of nitrogen and propylene using MIL-100(Fe) shaped material. The patent “Procédé cyclique de production d'azote de haute pureté et éventuellement d'hydrocarbure de haute pureté à partir d'une charge contenant de l'azote et un hydrocarbure.”, having as inventors F. Luck, C. Leroi, A. Fernandez, A. M. Ribeiro, J. C. Santos, A. E. Rodrigues, J-S. Chang, U- H. Lee, Y-K. Hwang, J-P. Bonne, Patent Application PCT/FR2012/051623, was filed on July 9, 2012.
Project Context and Objectives:
A major challenge facing European industry involves the development of more specific, energy saving processes with less environmental impact. The recent development of Metal Organic Frameworks (MOFs) may prove a major milestone in achieving these goals. The MACADEMIA project is an extension to the FP6 project DeSANNS, which highlighted some MOF materials for CO2 capture and storage. It allowed expanding MOF synthesis on a much larger scale, as well as multiple applications. TOTAL’s focus is on bringing applications of MOFs to key market sectors -gas separation and storage, and liquid separation. The TOTAL-led consortium, with Total Petrochemicals and Refining-Marketing as industrial partners, 11 academic partners from across EU, one leading South Korean partner, among world leaders in their particular domain of MOF science, contributed to the project, with a dedicated management partner.
The main objectives of the MACADEMIA project are:
- production of new MOFs and optimisation of those already of promising interest;
- characterization of MOFs using specialised techniques,
- use of predictive modelling for gas- and liquid phase applications, with Grand Canonical Monte-Carlo and Moleculr Dynamics tools,
- testing MOFs in a wide range of applications, such as gas and vapour recovery and separation (propene /propane, acid gases separation, CO2 and H2 purification), liquid phase adsorption and separation (xylene separations, recovery of N- and/or S-compounds from hydrocarbons), catalysis for specialty chemicals (Lewis-acid MOFs as catalysts for epoxide polymerization, redox-active MOFs as catalysts for hydrocarbon autoxidation).
Project Results:
WP1 - MOF discovery and synthesis optimisation
The discovery and synthesis optimisation of new MOFs is a tedious, time- and starting material consuming process. Therefore, in addition to the trials and errors solvothermal synthesis, high-throughput methods were employed in this step at CAU and CNRS-ILV and only minute amounts of materials were obtained. Based on these results first scale up reactions to the gram scale were carried out and selected materials were further synthesized in even larger amounts at CNRS-ILV (5-40 g) and then KRICT (0.1-2 kg).
Since the main application focus of MOFs was in the fields of separation and catalysis, stable MOFs were targeted and hence tri- and tetravalent metal ions (Fe3+, Al3+, Cr3+, Ti4+, Zr4+) were mainly used. These lead mostly to the formation of microcrystalline reaction products which hampers the structure determination of the MOFs.
WP 1.1. Synthesis optimization, scale-up and shaping of selected MOFs
Based on the screening of the properties of known MOFs, the synthesis optimization of series of selected functionalised solids (-NH2, -NO2, -Br, -CO2H, -2CO2H…) based on the robust metal terepthalates UiO-66(Zr) and MIL-125(Ti) or the hydrothermally stable iron polycarboxylates MIL-127(Fe) and MIL-100(Fe), was investigated. Ambient pressure and dynamic conditions were selected when possible. As high as possible space time yields, cheap and safe precursors as well as green conditions (water, alcohols) were privileged for each solid. In order to gain understanding prior to the synthesis optimisation work, in situ energy dispersive crystallization studies were carried out in parallel at the Synchrotron for some of the selected solids. Finally, the scale-up from the multigram scale to the kg scale was performed further optimizing the synthesis conditions (STY, temperature, concentrations of precursors, type of metal precursors, etc.) while new shaping methods were developed, leading to very little loss of performances compared with the powdered materials, in order to cope with the weaker mechanical stability of MOFs compared to inorganic porous solids.
WP 1.2. Synthesis of functionalized MOFs
Within the project functionalized derivatives of the framework types MIL-53(Al, Fe),MIL-88(Fe), MIL-125(Ti), CAU-1(Al), CAU-10,1UiO-66(Zr) and Cr-MIL-101 were synthesized in characterized in detail. In addition to the use of functionalized terephthalic and isopthalic acid, post-synthetic modification reactions were employed to introduce new functional groups. An important example is the nitration of Cr-MIL-101 using nitric and sulfuric acid and the subsequent reduction with SnCl2. yielding the Cr-MIL-101-NH2 which can be used in PSM reactions.1j The first keto-functionalized Al-MOF (CAU-8) was obtained using benzophenonedicarboxylic acid (H2BPDC).
WP 1.3 New MOFs based on mixed-linker or mixed-metal systems
Mixed-linker MOFs were investigated in order to tune the sorption properties and to study synergestic effects. First, in order to develop new porous architectures, di and tri-topic polycarboxylate linkers (1,4 BDC and BTB) were used to produce new micro- and meso-porous iron(III) MOFs of the MIL-142 (∅ = 7-12 Å)and MIL-143 (∅ = 12-25 Å) structure types. Noteworthy, functionalized terepthalates could also be used to produce these architectures. An alternative consisted in mixing linkers within the same structure type. For instance, starting from CAU-10 ([Al(OH)(1,3-BDC)] (1,3-BDC = isophthalalte) mixed linker compounds CAU-10-H/SO3H, CAU-10-H/Br and CAU-10-NH2/NO2 were synthesized and characterized. Through the amount of -SO3H groups the hydrophilicity of the pores can be tuned. Cr-MIL-101 derivatives were synthesized in order to make use of its stability in possible post-synthetic modification reactions. Thus, terephthalic acid containing various functional groups (F, Cl, Br, OH, NO2, NH2, CH3, COOH) were tested and synergistic effects were observed when TA-COOH was employed.
WP 1.4. Exploratory syntheses
Exploratory syntheses using Al salts employing the high-throughput method led to a number of new Al-based MOFs: CAU-1, CAU-3, CAU-4, CAU-6, CAU-8, CAU-10, CAU-13. Using non-aqueous reaction conditions unprecedented inorganic building units such as octameric, dodecameric and polymeric (based on the well known Al13-ion) were observed in CAU-1, -3 and -6 respectively. Variation of the linker geometry from linear (1,4-H2BDC) to V-shaped (1,3-H2BDC) opened the way to a new series of compounds (CAU-10). Extension of the linker (benzophenonedicarboxylic acid H2BPDC) resulted in yet a new MOF (CAU-8) which contains small pores (∅ = 4Å) in addition to one-dimensional channels (∅ = 8Å). Employing microwave assisted heating the manganese analogue of Cr-MIL-100 could be synthesized and the extension of the linker from 1,4-H2BDC to the naphthalene derivative 2,6-H2NDC allowed us to synthesize the Cr-MIL-101-NDC derivative which contains pores with a diameter of up to Å.
Using iron(III) salts, one could obtain a new porous iron(III) biphenyldicarboxylate MO, denoted MIL-126(Fe), that bear an interwoven MIL-88D architecture. The interpenetration ruled out any flexibility of the structure leading to a highly microporous rigid MOF (∅ = 4-10 Å) possessing a high amount of Lewis acid sites.
In the case of Zr, one could isolate and characterize a whole new series of micoporous rigid Zr4+ dicarboxylate MOFs, denoted MIL-140’s or ZrO[O2C-R-CO2] (R=benzyl, napthyl, biphenyl or dichloro-azobenzene).2c Noteworthy, the inorganic secondary building unit is comprised of a Zr oxide chain while the connection through the dicarboxylate anions leads to a 1D triangular pore system (∅ = 4-10 Å). Remarkably, despite their higher density and lower surface areas, the stability (hydrothermal, mechanical) of this series is much higher than those of their UiO-66 polymorphs while a significant amount of Lewis sites was also evidenced.
NB: note that series of powdered samples of new porous Al, Ti or Zr MOFs were still under structural investigation at the end of the project and will not be described here. WP2 Detailed structural and surface characterisation
The studies within WP2 allowed the full characterisations of many MOF materials. The results highlighted here are important:
- for materials optimisation by the characterisation of their chemical sites and evolution
- for gas processing applications by the study of gas diffusivities and the study of the adsorption sites
- for liquid phase separations
- for catalysis applications by the study of the reaction steps involved.
JUK
Infrared characterisation has been applied to ascertain surface properties of defined target samples: acidity, redox and exchange properties have been analysed on CuBTC samples by JUK using CO, CO2, pyridine, acetonitrile, thiophene and H2O together with D2O as probe molecules. Specific water interactions were remarked, being different depending on the sample preparation protocol proving that for most cases thermal treatment (and not hydrothermal treatment) was responsible for reported instability of MOF materials.
For most cases delivered MOF samples were investigated in terms of their optimal activation conditions, thermal stability, traces of impurities, hydrophilicity and affinity towards specific compounds.
Hybrid nanomaterials were obtained by post-synthesis combination of the two components: tungsten heteropolyacid (HPW) and Basolite™ F 300 used as a support showing the influence of the location of the active sites on the internal versus external surfaces of the crystals in hybrid catalysts on selectivity in ethanol conversion. Such influence is very important for the reaction with the use of bulky substrates/products, its understanding may be helpful for the overcoming of the diffusion limitation in nanoporous materials, like zeolites or MOFs.
As the highlights of the work, carried out during the MACADEMIA grant, several achievements can be named, most probably applicable to the majority of MOFs:
1. possibility of formation of anhydride in the case of MOFs bearing not saturated –COOH functionalities. Such transformation enabled increase of the sorption capacity and subsequent post-synthesis treatment leading to valuable catalysts (including NHPI: N-Hydroxyphthalimide, a catalyst of oxidation synthesis reactions);
2. reversible/irreversible storage of NO (and NO2) on amine-functionalized MOFs, which reversibility is dependent on the fact whether primary and secondary amines are introduced into the linker molecules.
CNRS-LCS
In situ and operando IR spectroscopy have been used as a technique particularly adapted for the investigation of porous compounds. A large amount of the MOFs studied in the project have been characterised at CNRS-LCS. This allowed notably the optimisation of the activation process for the selected solids before their upscaling. Chemical properties were evidenced on each material, which allowed to finely tune their properties throughout the project.
Besides monitoring from both the qualitative and quantitative point of view the physico-chemical properties of solids, these tools have been effective for the description of the properties of selected MOFs under duty.
In view of the propane/propene separation, the study on the MIL-100(Fe) sample, revealed that from a gas flow, both gases are adsorbed on the solid. An increase in the activation temperature resulted in a greater amount of propene adsorbed onto the solid, while the amount of propane remained constant .
Using quantitative IR spectroscopy of the NO and CO probe molecules, we demonstrated that the thermal treatment at increasing temperatures leads to the formation of Fe2+ sites having a higher affinity for the unsaturated C=C bond of propene than Fe3+. Interesting insights in the formation of the reduced iron sites and their role in the C3 separation were obtained by the selective poisoning of Fe2+ entities by strong NO adsorption. This knowledge of the structure–activity relationship gives precious indications to optimise the sample. In other words, an iron MOF with a larger number of FeII CUSs is required in order to increase the separation effect for an alkane/unsaturated hydrocarbon mixture at low partial pressure.
A possible reactive transformation of the ligand was shown to occur by the formation of diazonium species on functionalized MIL-125(Ti)-NH2. This could allow alternative post-synthesis ligand modification which can favour materials modification. On this solid, the formation of diazonium species was observed, verified and explained through a mechanism via the reaction of nitrosation of the amine by nitrous acid, giving a N-nitroso compound as intermediate, which tautomerizes to a diazohydroxide, successively protonated and dehydrated. The main reaction intermediate was identified as being NO2, issued by the dismutation of NO and producing nitrous acid via a reaction with adsorbed water or OH groups.
A catalytic transformation of methanol was demonstrated on MOFs. Methanol was found to adsorb dissociatively on different MOFs and to react into dimethyl ether and/or CO2. On MIL-100(Fe), these phenomena could be associated with the oxidation states of iron, while on the other non-reducible samples they would be related to the acid/base properties of the materials, favouring the dehydration/dehydrogenation steps of methanol and dimethylether into methyl formate, successively decomposing into CO2.
UWAR
In situ diffraction of guest adsorption UWAR has tested the structural modifications induced by liquid phase sorption of various organic guests by the host MOF MIL-53(Fe). This has allowed the swelling of the structure to be tracked in real time and the kinetics of uptake of various guest molecules to be compared directly. In the case of N/S heterocycles, molecules that must be separated from crude petroleum to avoid poisoning of processing catalysts, the solvent not only influences pore opening but is also a competing guest. Successful uptake by MIL-53(Fe) requires that the adsorbate is primarily a good hydrogen bond acceptor; additionally, based on UV-visible spectroscopy, a charge-transfer interaction between the S atoms of benzothiophene and the aromatic rings in the MOF pore wall is proposed. Since there is not a strong guest-metal interaction this system may be useful as a recyclable host for N/S adsorption since the guest molecules can be released readily after uptake.
Characterisation of Mixed-Metal MOFs Compared to mixed-ligand systems, mixed-metal-organic frameworks where the structure is preserved upon partial metal substitution are relatively rare in the literature, but should provide a subtle way of tuning properties such as magnetism or catalysis. We have prepared and characterised novel mixed-metal iron-vanadium analogues of the 1,4-benzenedicarboxylate (BDC) metal-organic framework MIL- 53. The identity of the materials was confirmed using high resolution powder X-ray diffraction and the oxidation states of iron and vanadium in all samples are verified using XANES spectroscopy at the metal K-edges, and for iron confirmed by 57Fe Mössbauer spectrometry. Infrared and Raman spectroscopies as a function of temperature allowed the conditions for removal of extra-framework species to be identified, to form single-phase MIL-53-type materials, (Fe,V)III(BDC)(OH,F). The iron-rich phase shows structural flexibility that is distinct from either the pure Fe material or the pure V material, with a thermally induced pore opening upon heating that is reversible upon cooling.
CNRS-IRCE
Knowledge of gas dynamics in MOFs is important for the design of gas processes.
By combining two experimental techniques operating at different time scales with molecular simulations, we characterized the peculiar dynamics of benzene confined in MIL-47(V): quasi-elastic neutron scattering (QENS) probes diffusion at short time scales (10-13 – 10-8 s), and simulations yield diffusivities in agreement with QENS but show in addition less frequent motions which can only be observed by 2H NMR. The picture we obtain leads to a molecule which is diffusing along the 1D channels, while randomly rotating by 90° jumps, performing thus a corkscrew motion. Benzene is expected to travel along a minimum energy pathway involving jump sequences between these regions.
Two diffusion regimes were found for xylenes isomers in the same MOF: a low temperature regime where the xylene molecules are close to the MIL-47(V) pore walls with a high activation energy barrier for diffusion, and a high temperature regime (above 300 K) where the xylenes are mainly located in the center of the channels associated with a lower activation energy. It has been further shown that the diffusivity is only slightly affected on the ns timescale when one compares xylenes in single component and in mixtures. However, large changes are found by 2H NMR (time scale > 10-7 s) for reorientation motions in mixtures. For example, para-xylene is much more mobile in a 50/50 mixture with meta-xylene. This indicates that packing effect and guest-guest interactions are crucial on long time scales and will influence separation processes.
WP3 Molecular modelling
The accurate computational description of interaction energies of molecules with MOFs containing coordinatively unsaturated transition metal sites (cus) represents a significant challenge for contemporary quantum chemistry. Commonly used force fields or methods based on density functional theory often fail to provide even qualitatively correct results. As an important achievement within the MACADEMIA project we consider the DFT/CC method developed for the accurate description of adsorption in MOFs. The method is based on the pair- wise representability of the DFT error E → ECCSD(T) EDFT which is calculated as the difference between the interaction energy obtained at the accurate reference level method, coupled clusters CCSD(T) in our case, and particular DFT functional. The DFT/CC method has been applied for the description of adsorbate/adsorbent systems (including H2O, CO, CO2, CH4, C3H6, C3H8 adsorption on CuBTC). The agreement was demonstrated for experimental and calculated adsorption heats at low coverage (experimental data obtained in collaboration with CNRS-MADIREL). Based on the excellent agreement between the experiment and calculations the mechanisms of adsorption have been proposed. In addition, DFT/CC method has served as a benchmark for fitting of new highly accurate force fields in collaboration with UEDIN and CNRS- ICGM.
In complement to this, high-throughput screening approach based on Monte Carlo simulations has been developed to not only predict the optimal existing MOF candidate for xylene separations but also to design a novel MOF with outstanding performance for the selective adsorption of acid gases (CO2, H2S) which was further successfully synthesized in the frame of MACADEMIA.
Indeed, studying xylene separation in a large number of MOFs we discovered a few simple rules that allowed for quick and efficient screening of MOFs and could guide the experimentalist searching for promising structures. Simply determining the pore diameter using geometrical methods, we found that MOFs selective to p-xylenes should ideally have one dimensional pores with pore sizes in the range between 4.2 Å and 5.5 Å. Electrostatic interactions which could be introduced to a framework e.g. by choosing an appropriate linker or post-synthetic modification further enhance this selectivity. Our simulations which showed good agreement with available experimental results also evidenced that sometimes very subtle effects can have a significant impact on the selectivity. For example in MIL-125 in the octahedral cages, separation of the xylene molecules happens on the basis of differences in packing and interaction with the pore walls, whereas the smaller tetrahedral cages are capable of separating molecules by molecular sieving.
Regarding the selective capture of acid gases, the first step consisted of considering a collection of recently synthesized MOFs (more than 100 candidates), which can a priori lead to a breakthrough in terms of separation performance for CO2/N2 and H2S/CH4 mixtures. Regarding this sampled series of materials, our grand canonical Monte Carlo (GCMC) simulations evidenced that the UiO-66(Zr) solid shows the highest CO2/N2 and H2S/CH4 selectivities when its terepthalate ligands are grafted by one – COOH group. This conclusion prompted us to further imagine a hypothetical UiO-66(Zr) material containing a second grafted carboxylic function in order to verify whether its separation performance is even higher. This modeling effort was based on a computational assisted structure determination strategy (force field based and DFT geometry optimizations) to propose a plausible structure candidate for this MOF. Starting with the parent UiO-66(Zr), the crystallographic structure of the material was first built using a ligand replacement strategy proposed in our previous studies. This newly designed solid labeled as UiO-66(Zr)-(COOH)2 exhibited simulated CO2/N2 and H2S/CH4 selectivities reaching ~90 and ~70 respectively, values significantly higher than that for the commonly used zeolite 13X. These excellent predicted separation performances further initiated a synthesis effort that was deployed to prepare a la carte this novel material. UiO-66(Zr)-(COOH)2 was indeed successfully synthesized by means of an environmentally friendly synthetic route with the involvement of an aqueous media. The simulated performances were confirmed by real co-adsorption measurements performed by the FPMS partner. Molecular Dynamics simulations further predicted that despite a relatively bulky environment of the cages due to the presence of the two carboxylic functions that at first sight would suggest a very slow diffusion process or even the absence of any possible mobility, the resulting diffusivity for the CO2 species is finally very similar to the values previously reported in the conventional faujasite 13X for the same range of temperature. This was confirmed by the quasi-elastic neutron scattering measurements performed by CNRS-IRCE. Indeed, such a conclusion emphasizes that the kinetics will not be a drawback for the use of such a novel material in physisorption-based processes.
Throughout this project, we have seen that even MOFs that are relatively rigid (i.e. do not show large structural changes such as breathing) often show inherent flexibility in their linkers. However, taking this flexibility into account in molecular simulation through flexible force fields is computationally expensive not least as such force fields have so far only been developed for a limited number of MOFs. Whereas the frameworks for MOFs can be treated as rigid when simulating equilibrium adsorption, framework flexibility has to be considered carefully in Molecular Dynamics simulations when studying the diffusion of larger molecules in order to get a realistic description of the movement of the molecules. For example, in MIL-47 framework flexibility only plays a significant role at lower temperatures (at higher temperatures the predicted diffusivities in a rigid and a flexible framework are similar). In contrast, in UiO-66 either no movement (xylenes) or limited motions (carbon dioxide, methane) of molecules from pore to pore is observed when treating the framework rigidly, which would mean that experimentally no diffusion should be measured which is clearly not the case when the partner CNRS-IRCE performed quasi-elastic neutron scattering measurements. When framework flexibility is incorporated in the simulations, the size and shape of the window is altered through the rotation and distortion of the window-defining BDC linkers. This distortion is in response to guest-framework interactions and is sufficient to allow adsorbate molecules to move between pores. Thus our work has shown that the flexibility of MOFs must be considered in cases where the dominant diameter (i.e. the smallest pore or window diameter) and kinetic diameter of guest molecules are similar. Additionally, in those cases where specific interaction sites are of importance, the impact of framework flexibility should be considered. This is case even for those MOFs which show no significant flexibility or structural changes in experiments.
Process simulation is a tool that can successfully be applied for a better understanding of all the phenomena involved in a given process without the need of doing unnecessary, time consuming and sometimes expensive experiments, otherwise necessary for achieving the same level of expertise.
The knowledge gained through the process simulation is extremely important for guidance in the experimental runs. The accuracy of the simulator depends on the mathematical model used. The complexity of the model should be large enough to account for all the main mechanisms involved in the process while, at the same time, keeping the computational effort necessary to simulate as low as possible. The process simulation should then account for all the phenomena at the particle scale (equilibrium and kinetics of adsorption), all the phenomena at the bed scale (axial dispersion, pressure drop and heat exchanges among the particle, fluid, columns and its surroundings), and all the aspects of the unit such as number of layers, number of columns, valve equations, compressors, vacuum pumps, storage tanks, cycles and all the flows in and out of the columns. From the process simulation it is then possible to predict the unit’s productivity, the product purity and recovery, as well as the energy consumption.
Therefore, particle scale and bed scale adsorption models were presented in Deliverables D3.22 (gas phase) and D3.25 (liquid phase). These models were subsequently used in the process simulation of PSA and SMB units, their sizing, and optimization of such units (D3.26 D3.27 D3.30 D3.31 D3.32 and D3.33). The separation of propane/propylene mixtures in pilot scale units was studied by simulation. Two technologies were considered: pressure swing adsorption (PSA) and simulated moving bed (SMB). The best adsorbent was considered to be the CuBTC spheres provided by KRICT. In the pressure swing adsorption process two units in series were required to obtain polymer grade propylene (purity > 99.5%) and high purity propane (> 96.5%). In the first unit, a cycle with five steps was used (adsorption, rinse, blowdown, purge, countercurrent pressurization) for the production of a high purity propylene stream. For a feed stream with 75% of propylene in propane, a 99.5% purity propylene stream and a 39% propane stream are obtained in the first unit. The propane enriched stream obtained in the first unit is separated in the second unit for which a five step cycle was designed. A high purity propane stream (96.7%) is produced in the second unit and the propylene enriched stream is recycled to the feed of the first unit. The performance of the global PSA unit was estimated to be:
Propylene purity/recovery: 99.5%/98.9%, Propylene productivity: 0.37 mol/kg/h, Propane purity/recovery: 96.7%/98.6%, Propane productivity: 0.12 mol/kg/day, and Power requirements: 116.8 Wh/molC3H6. In the simulated moving bed process, isobutane was used as desorbent. Using a 1-3-3-1 configuration, the separation region was reported for purity values in the extract outlet over 99.5 % and raffinate outlet over 96.5 % considering operating pressures of 1.5 bar and 2.5 bar. The performance of the SMB process is calculated for an operation point selected within the separation region obtained for 1.5 bar. The performance of the process for operation point selected within the separation region for 1.5 bar was: Propylene purity/recovery: 99.9%/99.8%, Propylene productivity: 0.97 mol/kg/h, Propane purity/recovery: 99.4%/99.8%, Propane productivity: 0.32 mol/kg/day, and Desorbent consumption: 1.973 NL/min. For the nitrogen/propylene separation, an industrial-scale PSA column was designed, several PSA cycles and processes based on these cycles were proposed for recovery of only nitrogen (Case 1) and for recovery of both components, i.e. nitrogen and propylene (Case 2). The effect of operating parameters on the process performance parameters were examined based on simulations. For the nitrogen recovery, the best result was obtained with the 5-step (pressurization, adsorption, cocurrent depressurization, blowdown, and purge), 2-column PSA process. From this cycle a nitrogen product stream of 95.4 % purity was obtained with a recovery of 85.2 %. The adsorbent productivity was 6.0 mol N2/kg adsorbent/h and power consumption was 156 W/kg N2 produced/h. Among the PSA processes proposed for recovery of both components, the one which is extended from the 6-step cycle to a 4-column process exhibited better performance than those (3-column) based on the 5-step cycle. Nitrogen with 96.2 % purity and propylene with 97.6 % purity were obtained from the 6-step 4-column process proposed for recovery of both components. The nitrogen and propylene recoveries were determined as 98.4 % and 91.0 %, respectively. The nitrogen and propylene productivities of the process were estimated as 4.61 and 1.83 mol product/kg adsorbent/h and the power consumption as 383 W/kg nitrogen produced/h. In deliverables D3.27 D3.31 and D3.33 the separation of xylene isomers mixture was studied in different materials (MIL-53(Al) – Basolite A100, UiO_66(Zr) and MIL_125(Ti)_NH2), by simulated moving bed (SMB) technology. Although MIL-125(Ti)-NH2 is a para-xylene selective material, all
studied material still present poor behavior in for this separation. However, some new ideas can be explored using MOF based adsorptive processes. The reverse shape selective of UiO-66(Zr) brought the possibility of finding materials that can produce para-xylene as the light component, and highly ortho-selective materials like MIL-53(Al) open the possibility of new hybrid processes combining SMB and crystallization.
To simulate the performances of different adsorbents for natural gas purification (H2S/CH4), syngas purification (H2S/CO2) and nitrogen-propene separation, four macroscopic thermodynamic models were developed: the Ideal Adsorbed Solution Theory (IAST), the Real Adsorbed Solution Theory (RAST) and the Vacancy Solution Theory with the activity coefficients of Flory-Huggins (VST-FH) and with the coefficients of Wilson (VST-Wilson). These models were able to predict co- adsorption isotherms and selectivities for a series of MOFs (MIL-47(V), MIL-101(Cr), MIL-125(Ti), MIL-125(Ti)-NH2, UiO-66(Zr) and UiO-66(Zr)-NH2) with respect to the gas mixtures mentioned above.
While all macroscopic models converged toward similar single component adsorption isotherms, they led to some deviation for the predicted adsorbed amounts. Significant differences have been pointed out for the predicted selectivities, in the curve shapes and in the calculated adsorbed amounts. Moreover, the IAS model seems to be inefficient to simulate the presence of a minimum in the selectivity profile with the loading. This conclusion clearly emphasizes that it is of great importance to perform real co-adsorption for an accurate estimation of the selectivity of MOFs containing specific adsorption sites. Furthermore, these real co-adsorption measurements allowed us to use improved models such as (i) the Wilson activity coefficient model, which takes into account the binary interactions, or (ii) the RAST model which both require mixture adsorption data to simulate the adsorption behaviour of mixtures.
WP4 Applications of MOF materials in gas adsorption/separation
High-throughput analysis of materials
Hundreds of samples have been prepared in the MACADEMIA project. Many of these have never left the synthesis labs as they were deemed not of sufficient quality. In some cases, we were not sure whether a given sample was of interest or whether it was optimised. To screen these, as well as optimised materials, a unique high throughput high pressure adsorption apparatus was developed to measure single gas isotherms to up to 40 bars. With less than 100mg of sample it is now possible to evaluate whether the porosity is optimal, its cycling behaviour (reversibility/regenerability), effect of adsorption temperature and to predict mixture properties using IAST. One problem which arises from the results obtained in this work package is how to compare these results with those obtained with the reference materials (zeolite 13X and Takeda 5A active carbon). This was eventually carried out with the development of an ‘Adsorbent Performance Indicator’ (API). This indicator was then used in other workpackages to select the most appropriate materials for upscale and analysis at stage 2.
Separation of propene from propane
Propane/propylene separation is commonly carried out by low-temperature distillation at high pressures; however, this is an energy-intensive distillation. Adsorption is a separation technology with a great potential for reducing energy consumption compared to the conventional distillation. Separation by adsorption could be carried out either by a cyclic pressure swing adsorption (PSA) process or by means of simulated moving bed (SMB) technology. The primary requirement for any adsorption process is to find an adsorbent with sufficiently high selectivity, capacity and life. A wide variety of adsorbents have been studied to carry out propane/propylene separation: silica gel, activated carbons, carbon molecular sieves, zeolites, chemical adsorbents with complexing metals (i.e. copper or silver ions), and most recently, MOFs. Although promising, the huge number of MOFs turns difficult the identification of individual candidates for specific applications.
Two generations of shaped MOFs have been evaluated as potential adsorbents to carry out propane/propylene separation by adsorption. Isobutane is proposed as a possible desorbent for Simulated Moving Bed (SMB) operation. In order to evaluate these materials, propane, propylene and isobutane adsorption isotherms were obtained by means of a gravimetric method, and single, binary, pseudobinary and pseudoternary breakthrough curves were obtained in a fixed-bed adsorption unit. Seven different shaped samples were evaluated, the 1st generation comprised Cu-BTC tablets (provided by BASF), MIL-100(Fe) (provided by KRICT) and Mg-Formate (BASF). Within these samples the best performing one was Cu-BTC tables, which present high selectivity towards propylene, and good capacities far higher than the capacities presented by 13X zeolite. Within the 2nd generation we could find the following samples: Cu-BTC tablets (KRICT), Cu-BTC spheres (KRICT), UiO-66(Zr) (KRICT), and MIL-125(Ti)_NH2 (KRICT). The UiO-66(Zr) and MIL-125(Ti)_NH2 presented poor selectivity between propane and propylene. However, the Cu-BTC spheres provided by KRICT we very selective, as selective as the CU-BTC material provided by BASF, but with far higher capacity. This material presented very low capacity loss when compared to the Cu-BTC powder material. Furthermore, it was demonstrated that the production of polymer grade propylene (purity above 99.5 %) is possible by PSA using CuBTC from BASF with a recovery of 28.5 % and productivity of 0.8 mol kg-1 h-1. Additionally propane was also obtained with a high purity (99.5%) and a recovery of 99.5 %. For CU-BTC from KRICT (spheres), the cycle carried out produced propylene with purity of 99.8 % (above polymer grade specifications), with a productivity of 0.6 mol kg-1 h-1. However the propylene recovery was low: 15 %. If part of the raffinate streams produced during the feed and rinse steps are to be recycled for their use in the pressurization and purge steps, the recovery can be higher.
Nitrogen recovery from light hydrocarbons
The separation process for the nitrogen purge gas recovery from light hydrocarbons is an important part of the polypropylene production. The operation conditions for this process were provided as the feed composition of 30% C3H6 and 70% N2, 70 °C and 1 bar by TOTAL. The process aims to recover nitrogen with purity higher than 95 mol%.
During manufacture of polypropylene, nitrogen gas is used in a polymer-degassing step to remove unreacted monomer, solvents, and additives from the polypropylene. The off-gas stream is processed to recover a portion of the hydrocarbons, but the remaining nitrogen and hydrocarbons are disposed of through combustion either in a flare or by diluting a fuel gas stream. The N2 recovery process enables reuse of the substantial raw material, reduces the polypropylene manufacturing cost and results in a more ecofriendly polypropylene manufacturing system. In this task N2 recovery from the N2/C3H6 mixtures using one of the second generation MOFs, MIL-100(Fe) granulates supplied by KRICT, was investigated by means of adsorption dynamic experiments using fixed bed. The single and binary breakthrough curves of nitrogen and propylene were obtained at 343 K and 1.3 bar. This was followed by the validation, with lab-scale PSA experiments, of the mathematical model, previously employed in the optimization of the industrial-scale PSA processes for the recovery of nitrogen and propylene from the propylene production purge. Three PSA cycles, one for recovery of only nitrogen (Case 1) and two for recovery of both nitrogen and propylene (Case 2), were proposed and carried out experimentally on a pilot-scale single-column PSA unit available at LSRE. The temperature, pressure, and concentration histories acquired during the experimental runs of the cycles were compared with the results of simulations employing the mathematical model. By simulation of an industrial scale process it was possible to design a cycle where a nitrogen product stream of 95.4 % purity was obtained with a recovery of 85.2 %. The adsorbent productivity was 6.0 mol N2/kg adsorbent/h and power consumption was 156 W/kg N2 produced/h. Among the PSA processes proposed for recovery of both components, the one which is extended from the 6-step cycle to a 4-column process exhibited better performance than those (3-column) based on the 5-step cycle. Nitrogen with 96.2 % purity and propylene with 97.6 % purity were obtained from the 6-step 4-column process proposed for recovery of both components. The nitrogen and propylene recoveries were determined as 98.4 % and 91.0 %, respectively. The nitrogen and propylene productivities of the process were estimated as 4,61 and 1.83 mol product/kg adsorbent/h and the power consumption as 383 W/kg nitrogen produced/h. Regarding to the most interesting applications (Biogas and Natural Gas purification, CTL and CTO processes), the measurement of the mixture adsorption equilibrium was performed for H2S/CH4,k,. Nevertheless, as far as purification of adsorbable gases is concerned, it appears necessary to obtain experimental adsorption data at low concentrations for the several components of the mixture (acid gases, sulfur compounds and/or carbon dioxide compounds in natural gas, in air or impurities in gases for semi-conductor applications). After performing pure adsorption isotherms measurements, 10 MOFs, regenerable and stable under H2S atmosphere, have been identified. The MIL-100(Al) presents the biggest adsorption capacities but the CH4 adsorption is too high for the H2S/CH4 separation process. MIL-125(Ti)-NH2 presents the biggest H2S slope at zero coverage with a good adsorption capacity. The UiO-66-BTEC has lower adsorption capacities than most of the other studied adsorbents but the lowest ones correspond to CH4 Co-adsorption measurements were performed on the most promising adsorbents. The H2S/CH4 selectivities on shaped MIL-125(Ti)-NH2 are good: 22 and 35, at 5 bar and 50 bar respectively, and the adsorption capacities are correct. It is a very promising material for H2S removal from natural gas or biogas. The best H2S/CH4 selectivities are obtained on shaped UiO-66(Zr)-BTEC: more than 40 at the all range on H2S concentration (0-10%) at 5 and 50 bar, but unfortunately the adsorption capacities are lower. It is also very promising materials for H2S removal from natural gas or biogas. It is important to note that the result of the mixture adsorption equilibrium predictions shows how it is necessary, for selective systems with different kind of adsorption sites as the main interesting systems on MOFs, to realize mixture measurements and not to only use macroscopic models to provide selectivity.
C6 / aliphatics separations
We could design and build a set up allowing for the study of the co-adsorption as well as the separation of benzene from hydrocarbons. This set-up also allows for measuring transient partial pressure of each component of the mixture being adsorbed on a MOF. Furthermore the enthalpy of adsorption can also be directly measured. The most significant result is the co-adsorption of benzene and n-hexane onto MIL-101(Cr) at ambient conditions. We could show that benzene is preferentially adsorbed as compared to n-hexane, even in the very early stages of the adsorption process. It has been shown that once benzene is adsorbed, n-hexane can also adsorb in the same material, likely on different surface sites.
WP5 Application of MOF materials in liquid phase adsorption/separation
Xylene separation
Because of their close boiling points, xylene isomer separation is a challenging process. Industrially, these isomers are separated using a KBaY zeolite in a simulated moving bed (SMB) process. The most important isomer is p-xylene which is used in large quantities for polyester and polyethylene terephthalate production. During the project, MOFs with high separation factors were discovered and their application in an SMB process evaluated.
To complete task 5.2 with success, an iterative process was developed between the different partners involved in this task, CNRS-ILV: synthesis of new structures, KRICT: synthesis scale up and material shaping, KUL: small scale screening, and FEUP: large scale screening and performance testing. As result of this process in the end of the MACADEMIA project a p-xylene selective material was found, characterized and tested in pilot-scale SMB unit. Furthermore, within this task many new structures were tested by KUL and FEUP, such as MIL-53(Al), MIL-47(V), MOF-14, UiO-66(Zr), and MIL-125(Ti)_NH2, just to name the most important ones.
From the screening of the newly developed MOFs, one class of MOFs with identical topologies was found to be highly para-selective adsorbents. MIL-125(Ti) ([Ti8O8(OH)4(BDC)6]), MIL-125(Ti)-NH2 ([Ti8O8(OH)4(BDC-NH2)6]), and CAU-1(Al)-NH2 ([Al8(OH)4(OCH3)8(BDC-NH2)6]) (BDC = 1,4-benzenedicarboxylate) are built up from octahedral cages and tetrahedral cages. These cages both exhibit the para-selectivity. In the octahedral cages, the molecules can be separated on the basis of differences in packing and interaction with the pore walls; in the smaller tetrahedral cages, the xylene isomers are separated by molecular sieving.
Shaped MIL-125(Ti)-NH2 was further tested in breakthrough experiments for selective adsorption and separation of xylene isomers, in liquid phase and using n-heptane as eluent. This material is para-selective, allowing the separation of p-xylene from its isomers in ternary or quaternary mixtures.
In another approach, UiO-66(Zr) was chosen to be the 2nd generation MOF, due to its interesting feature of reverse shape selectivity, which could allow recovering p-xylene as the light product (raffinate), besides the obvious possible separation of o-xylene in the heavy product (extract), in a SMB unit. Powder and tablets were tested for selective adsorption and separation of xylene isomers, in liquid phase and using n-heptane as eluent. In the liquid phase, this material exhibited a reverse shape selectivity. The selectivities of 1.6 and 2.3 were determined from the breakthrough experiments, for o-xylene over m-xylene and over p-xylene at 313 K, respectively, in the presence of n-heptane.
Nitrogen/Sulfur Adsorption
As regulation dictates that the S-level in fuels needs to be decreased below 10 ppm, catalytic hydrodesulphurization becomes increasingly difficult. In principle, adsorptive separation could be a solution, but currently available adsorbents, such as zeolites, have insufficient capacity and stability to realistically meet the industrial requirements. Because of their high adsorption capacities and metal content that can be used to selectively adsorb N or S-containing compounds, MOFs were investigated for this purification.
During the project, several structures were identified with superior affinity and capacity for S-and N-containing compounds. CPO-27 and [Cu3(BTC)2] were found to have the highest adsorption capacities for the selective adsorption of the S-heterocyclics, reaching weight% uptakes of elemental sulfur of up to 6 wt%.
Early in the project, MIL-100(Fe) was found to not only selectively adsorb S-heterocyclics, but also to have a very high affinity for N-heterocyclics. The adsorption of these compounds is crucial because they can poison the catalyst used in the hydrodesulphurization. Using a representative raw fuel feed containing 950ppmw S and 750 ppmw N, it was demonstrated that a column filled with MIL-100(Fe) could effectively adsorb all S and nitrogen containing compounds.
In a follow-up study, the influence of the metal ion in MIL-100(Al3+,Cr3+,Fe3+,V3+) on the adsorptive removal of N/S-heterocyclic molecules from fuels was investigated. The results show a clear influence of the different metals (Al, Fe, Cr, V) on the affinity for the heterocyclic compounds, on the integral adsorption enthalpies, and on the uptake capacities. Among several factors, the availability of coordinatively unsaturated sites and the presence of basic sites next to the coordinative vacancies are important factors contributing to the observed affinity differences for N-heterocyclic compounds.
MIL-100(V) was found to be the best adsorbent for the deep and selective removal of nitrogen contaminants out of fuel feeds.
WP6 Applications of MOF materials in catalysis
Lewis Acid catalyzed reactions over MOFs
The possibility to synthesize MOFs with any desired metal, large metal content, large pore size and surface area make them more advantageous than such conventional solid catalysts as zeolites or metal oxides for liquid phase reactions. Another important feature of MOFs consists in the periodicity and regularity of their active sites. Since that, MOFs may be considered as catalysts with single-sites in contrast to metal oxides and zeolites with the random distribution of the first and second coordination sphere around a metal. In this contribution, we aimed to compare the catalytic behavior of selected MOFs (Cu3(BTC)2, Fe(BTC), MIL-53(Al), MIL-100(Cr/Fe), ZIF-8) and some zeolites (TS-1, Beta, USY) in Prins reaction of β-pinene with paraformaldehyde, and Knoevenagel condensation of aldehydes with active methylene compounds.
In Prins condensation of β-pinene and paraformaldehyde to produce nopol, activity of MOFs increased with increasing of the accessibility of Lewis acid centers in the order ZIF-8 < MIL-53(Al) < Fe(BTC) < MIL-100(Cr) < MIL-100(Fe). In contrast to MOFs, the presence of relatively strong Brønsted acid sites within zeolites Beta and USY resulted in the formation of side products of β-pinene isomerisation, decreasing the selectivity of these catalysts. The yield of nopol formed over MIL-100(Fe) increased with the increasing of the relative polarity of the used solvent in the order: dodecane < p-xylene < cyclohexanone < acetonitrile. The high activity, as well as the preservation of the framework and the active centers of MIL-100(Fe) makes possible the reuse of this catalyst without significant loss of the activity at least in 3 catalytic cycles.
Cu3(BTC)2 and Fe(BTC) appear to be more active and selective catalysts for the Knoevenagel condensation than zeolites Beta and USY due to the absence of consecutive reactions of the primary condensation products and favorable diffusion of reactants and products providing by mild acidity of the MOFs and their larger pore size, respectively. Interestingly, while yields of target products over zeolites and MOFs were comparable in condensation with ethyl acetoacetate and methyl cyanoacetate, unexpectedly high activity of Cu3(BTC)2 was observed for benzaldehyde condensation with malononitrile. In the last case aldehyde totally transformed to corresponding unsaturated nitrile after 1 h at relatively low reaction temperature as for acid-catalyzed Knoevenagel condensation (60 °C). Under the same conditions, reaction on the zeolite does not proceed at all. By using computational methods, it was established that the dual coordination of malononitrile molecule to both active sites of Cu3(BTC)2 takes place. It was calculated, that such adsorption complex is characterized by interaction energy that is twice as large as the one for single adsorption site. Presumably, Knoevenagel reaction catalyzed by Cu3(BTC)2 proceeds through the transfer of proton from malononitrile to carboxylic group of framework (formation of defect Brønsted acid site) that can be mediated by the benzaldehyde molecule. Interaction of benzaldehyde with Brønsted site almost entirely compensates the energy loss due to the formation of deprotonated malononitrile and defect acid center. After the final reaction step, dehydration, the original paddlewheel structure is fully recovered.
Consequently, it was shown that the preferences of MOFs over traditionally used zeolites in acid-catalyzed reactions are caused not only by the textural or acid characteristics, but mainly by the regular organisation of the active centers in the framework. In some cases (as it was established for Knoevenagel reaction) the last feature may result in changing of reaction mechanism over MOFs in comparison to zeolites, which provides the higher catalytic activity of metal-organic framework materials. Thus the deep experimental and theoretical investigation of the structure of active centers and their modification during catalytic process seems to be a key not only for the purposeful development of the effective catalysts but also for the understanding of the fundamental perspectives of MOFs in heterogeneous catalysis.
WP7 Scale-up of MOF synthesis, shaping and pilot applications
FEUP
Pilot-scale tests for determination of long-term integrity, separation performances and catalytic activity/selectivity using shaped MOFs were conducted at TOTAL. However, these tests can be time consuming and sometimes expensive experiments, if run without the knowledge gained through process simulation. Therefore, in this work package FEUP partner played an important role with the development of an accurate of the simulator based on a detailed adsorption fixed-bed mathematical model. For instance, in the nitrogen/propylene separation case, a PSA cycle was proposed for recovery of nitrogen from a purge stream of the polypropylene production, with a product purity of 98 %, recovery of about 83 %, and productivity of 1.42 mol kgads-1 h-1 (see Table 2). From this collaboration three partners (FEUP, TOTAL, and KRICT) filed a patent for a process, aiming the recovery of nitrogen or recovery of nitrogen and propylene using MIL-100(Fe) shaped material. The patent is titled “Procédé cyclique de production d'azote de haute pureté et éventuellement d'hydrocarbure de haute pureté à partir d'une charge contenant de l'azote et un hydrocarbure.”, having as inventors F. Luck, C. Leroi, A. Fernandez, A. M. Ribeiro, J. C. Santos, A. E. Rodrigues, J-S. Chang, U- H. Lee, Y-K. Hwang, J-P. Bonne, Patent Application PCT/FR2012/051623, filled on July 9, 2012.
TOTAL
A gas phase pilot was set up successfully for separations like propane/propylene, nitrogen/propylene, syngas purification and natural gas treatment. For each separation, one or several MOFs have been tested. If the propane/propylene separation wasn’t suitable by adsorption with MOF, the nitrogen/propylene separation with MiL-100(Fe) gave good results which were patented.
It is still too early to get an idea of the MOF performances for syngas purification and natural gas treatment. In the first case, the results were very close to those obtained with activated carbon used as the reference. Even with a process optimization, activated carbon might be as good as MOFs, or even better. For natural gas treatment, no reference was available but results with MOF seemed pretty interesting. However, more work is required to estimate the real performances of MOFs in different applications. It is however likely that there is room for adsorption and separations with MOFs in this application.
Potential Impact:
Starting from a strong knowledge based on structure-performance relationships of MOFs, the primary result of this project will be to speed up time-to-market for the development of metal- organic frameworks materials, their processing and applications. In terms of industrial use of MOF materials, for certain applications such as gas sorption/separation, the potential is certain whereas in other fields (liquids, catalysis), additional R&D is required for major key breakthroughs. Gas separation & purification, separation of liquid mixtures and catalysis are three potential high-volume applications, and essential to sustainable, resource-economical chemical production. Propene/propane separation is relevant for propylene production (annual worldwide production 69 million tons), for production of polypropylene. Propane/propylene mixtures from a conventional FCC are separated by distillation in a C3 splitter in quite a complicated set up which can have up to 200 plateaus. In the project, this separation was performed using Pressure Swing Adsorption and Simulated Moving Bed technologies. In the EU, production of a petrochemical intermediate such as propene faces fierce competition with large petrochemical platforms especially in the Middle East and Asia, which have access to locally cheaper feedstocks. Any technical/economical advantage would be most welcome to allow such energy-intensive activities to remain based in Europe.
Both acid gases H2S and CO2 have been chosen by Total with respect to the Coal-to-Liquids (CTL) and Coal-to-Olefins (CTO) processes, which will have to be developed commercially in the next decades. Various solvents can be used to recover the H2S and CO2, such as Selexol and Rectisol. These processes are high capital expenditure intensive and require low temperatures. A sorption based process could prove an interesting alternative; especially as the raw syngas pressure is around 30 bar: the most promising adsorbent for the purification of CH4 containing CO2 and H2S is the MIL-125(Ti)-NH2. Moreover, MIL-125(Ti)-NH2 presents the best kinetic properties, an important advantage allowing better regeneration and improved performances in a PSA process.
Furthermore, with increasing environmental and climate change concerns, it is essential to reduce CO2 emissions. Separating from the feed gas will lead to a major reduction of CO2 emissions from a CTL/CTO plants which then needs to be concentrated to >95%vol. Biogas purification (CH4/H2S separation) is as well a hot topic due to the explosive growth of biogas production worldwide. CO2 purification for Carbon Capture and Storage (CCS) is still a quite recent field without to data any EU directive on CO2 quality required for geological storage. Amine capture processes can result in high purity CO2. However, CO2 resulting from oxy-combustion capture processes can contain significant amounts of impurities, requiring a further purification step prior to storage.
Nitrogen/hydrocarbon separations of interest for Total in the project emerge from the recovery of unreacted propylene after polymerisation. High polymer purity is increasingly an issue thus requiring optimisation of this nitrogen purge step. With 300 polyolefin plants worldwide, monomer losses cost the industry nearly 300 million € per year, representing a considerable resource-recovery opportunity. As demonstrated in the project, MOF-based separation processes could replace condensation which shows quite poor efficiencies. Total, KRICT and the University of Porto filed in July 2012 the PCT/FR2012/051623 patent “Cyclic process for producing high purity nitrogen and optionally high purity hydrocarbon from a feed containing nitrogen and a hydrocarbon”.
Deep desulfurization is required to comply with requirements for transportation fuels (currently at 10 ppmw S for diesel) or to enable use of liquid fuels in fuel cells (S < 1 ppmw). In addition, hydrodesulfurization would be easier if first nitrogen compounds, such as carbazoles, indoles and their benzo-analogs could be removed, as these severely inhibit hydrotreating catalysts at 50 ppmw. Preliminary results showed that some MOFs, e.g. MIL-47 from CNRS-ILV, can absorb up to 32 wt. % thiophene, or 12 wt.% S. KU Leuven also demonstrated high uptakes of N-heterocyclic aromatics in competition with aromatic hydrocarbons.
Even though all precautions have been taken to perform the evaluation of the performances of the shaped Cu3(BTC)2 in the standard operating conditions typically used at Total Research&Technology Feluy (Belgium) to qualify the sorbent batches, the performances of Cu3(BTC)2 for nitrogen removal were not conclusive. From the results of batch experiments at KU Leuven, Cu3(BTC)2 was expected to have a better selective recovery of S-heterocycles in comparison to N-heterocycles. However, surprisingly, whatever the spiked feed considered (diluted or concentrated case), almost no sulfur removal was observed. N-compounds were only slightly retained on Cu3(BTC)2, and the elution was proceeding very quickly. Repeatability of the tests was demonstrated since the second test performed under concentrated feed gave the same results.
MOF catalysis has the potential to significantly contribute to the greening of chemistry. The reactions selected, viz. epoxide polymerization and alkane autoxidation, are test cases in which MOFs will be key in superseding existing technologies. Progress achieved in MACADEMIA on the Lewis-acid activation of epoxides with MOF catalysts to produce polyether polyols would bring significant advantage with respect to the current technology (double metal cyanide (DMC) catalysts: very high cost and induction time with ethylene oxide). However, the activity of MIL-101(Cr)nano catalyst in consecutive polymerization runs dropped significantly. For fine chemicals production, autoxidation of alkanes such as cyclooctane has a moderate conversion rate, 25 g cyclooctane per g of catalyst. This number needs to be increased several orders of magnitude in order to become a competitive process. However, this is a minimum productivity number, since it would be possible recover, reactivate and reuse the catalyst. Further progress of MOFs is needed in both selected catalytic applications. Development of a specific European knowledge on improved selective oxidation of alkanes is critical to maintain on the long-tem the position of the European chemical industry, including also other applications of selective oxidation such as short-chain alcohols for chemical and clean fuels (e.g. butanol) applications.
Finally, in the rapidly growing field of MOF research, the MACADEMIA project timely addressed major global challenges such as gas- and liquid-phase separations, catalysis for specialty chemicals, and energy-saving processes. Business needs should be addressed in terms of favourable framework conditions for the commercialisation of technologies, adequate numbers of well-trained and mobile researchers responsive to the needs of industry and finally to an excellent public research base (research institutions and infrastructures) with strong interactions with industry.
The main dissemination activities in the 2nd Period have included:
Staff Training: Beneficiaries have employed staff to work on the project who are therefore gaining project related knowledge and experience.Training Workshops on Metal Organic Frameworks were held on 22nd June 2011 in Porto, Portugal and on 14th June 2012 in London, UK. ICON supported the attendance of 9 young researchers to the MACADEMIA workshop on in Porto, Portugal and 2 in London, UK by providing financial assistance for their travel and subsistence costs as described in the Description of Work. ICON also supported the attendance of seven PhD student participants to the 18th Zeolite Forum in Zakopane, Poland (13/09 – 17/-9/2011). ICON supported travel and accommodation costs for a 1-month training exchange visit for one PhD student from JUK to the laboratories of Professor Morris at St. Andrews University. The student was exposed to techniques that currently are not available at JUK. MACADEMIA members participated in 27 meetings, conferences and symposia, presenting material from the project. MACADEMIA members visited each other’s laboratories and promoted student exchanges.
The first issue of the MACADEMIA Project Brief was published. Dissemination activities in the 3nd Period have included Staff Training: Beneficiaries have employed staff to work on the project who are therefore gaining project related knowledge and experience.A training workshop was held on the 11th and 12th June 2013 in Brussels, Belgium which included external speakers from the Free University of Brussels, Belgium, the University of Turin, Italy, University of Granada, Spain, University of Hanover, Germany, University of Gent, Belgium and the Technical University of Delft, The Netherlands.
ICON supported the attendance of three young researchers to the MACADEMIA workshop in London (June 2012, not included in the previous report) by providing financial assistance for their travel and subsistence costs as described in the Description of Work. ICON also refunded travel and accommodation costs to external speakers at the workshop held in Brussels (11th and 12th June 2013). The website (www.macademia-project.eu) has been maintained since going live in March 2010. A dissemination strategy seminar, facilitated by the Commission’s IPR help team was held on June 13th, 2013 in Brussels, Belgium.
97 scientific, peer reviewed articles were published and/or submitted in high impact journals in total over the three years of the project (47 in Period 3).
Participation of MACADEMIA partners at 30 external meetings, conferences and symposia, in the 3rd reporting period presenting material from the project. MACADEMIA partners visiting each other’s laboratories and student exchanges. Publication of the second issue of the MACADEMIA Project Brief.
Exploitation of results:
Three projects partners (FEUP, TOTAL, and KRICT) filed a patent for a process, aiming the recovery of nitrogen or recovery of nitrogen and propylene using MIL-100(Fe) shaped material. The patent is titled “Procédé cyclique de production d'azote de haute pureté et éventuellement d'hydrocarbure de haute pureté à partir d'une charge contenant de l'azote et un hydrocarbure.”, having as inventors F. Luck, C. Leroi, A. Fernandez, A. M. Ribeiro, J. C. Santos, A. E. Rodrigues, J-S. Chang, U- H. Lee, Y-K. Hwang, J-P. Bonne, Patent Application PCT/FR2012/051623, filled on July 9, 2012.
An exploitation seminar was held in Brussels, June 13th, with 13 participants from 9 organisations, animated by Salvador Pastor, ESIC2 Services. At the present moment, only two Exploitable Results had been clearly identified:
● Synthesis optimization and shaping of new MOF materials.
● Workshop / Training Module on MOFs: from synthesis to applications.
It is expected to have specific results for additional MOFs from partner TOTAL within a few months (January 2014); at that time, more relevant Exploitable Results may appear, and a second version of this ESS Report could be delivered with complete new sets of Contribution-Benefit Matrices,
Risk analysis & Priority Maps, and Background/Foreground Information.
List of Websites:
www.macademia-project.eu
A major challenge facing European industry involves the development of more specific, energy saving processes with less environmental impact. The recent development of Metal Organic Frameworks (MOFs) may prove a major milestone in achieving these goals. The MACADEMIA project is an extension to the FP6 project DeSANNS, which highlighted some MOF materials for CO2 capture and storage. It expanded and continued this work on a much larger scale, with additional industrial applications: TOTAL’s focus is on bringing applications of MOFs to key market sectors - gas separation and storage, and liquid phase separations. The TOTAL-led consortium, with Total Petrochemicals and Refining-Marketing as industrial partners, 11 academic partners from across EU, one leading South Korean partner, among world leaders in their particular domain of MOF science, contributed enthusiastically to the project, with the constant and dedicated support of the management partner.
MACADEMIA aimed at producing new MOFs and optimizing those already of promising interest, characterizing MOFs using specialised techniques, testing MOFs using a three-tiered process, using predictive modelling and demonstrating the use of MOFs in key industrial processes. More specifically, the main targets of the project were:
- separation processes in gas/vapour phase (propene/propane, acid gases separation, CO2 and hydrogen purification);
- liquid phase separations (xylene separations, recovery of N- and/or S-compounds from hydrocarbons);
- catalysis for specialty chemicals (Lewis-acid MOFs as catalysts for epoxide polymerization, redox-active MOFs as catalysts for hydrocarbon autoxidation).
While most of the deliverables contributed to expand considerably the knowledge on synthesis, characterization and industrial applications of MOFs, some limitations were encountered in some specific domains, such as:
- Catalysis: in epoxide polymerization, the activity of theMIL-101(Cr)nano MOF in consecutive polymerization runs drops significantly. Cyclooctane autoxidation has a moderate conversion rate, 25 g cyclooctane per g of catalyst: this number needs to be increased several orders of magnitude in order to become a competitive process. However, this is a minimum productivity number, since it would be possible recover, reactivate and reuse the catalyst.
-Purification of fuels by S- and N-adsorption on the Cu3(BTC)2 MOF : All the precautions have been taken to perform the evaluation of the performances of the shaped MOF supplied by KRICT (glove-box, repeatability of the tests,…). However in the standard operating conditions traditionally used at Total Research&Technology Feluy (Belgium) to qualify the sorbent batches, the performances of Cu3(BTC)2 were not conclusive.
Overall, 97 scientific, peer reviewed articles were published in high impact journals in total over the three years of the project. Three partners, FEUP, TOTAL, and KRICT filed jointly a patent for a process aiming at recovery of nitrogen or recovery of nitrogen and propylene using MIL-100(Fe) shaped material. The patent “Procédé cyclique de production d'azote de haute pureté et éventuellement d'hydrocarbure de haute pureté à partir d'une charge contenant de l'azote et un hydrocarbure.”, having as inventors F. Luck, C. Leroi, A. Fernandez, A. M. Ribeiro, J. C. Santos, A. E. Rodrigues, J-S. Chang, U- H. Lee, Y-K. Hwang, J-P. Bonne, Patent Application PCT/FR2012/051623, was filed on July 9, 2012.
Project Context and Objectives:
A major challenge facing European industry involves the development of more specific, energy saving processes with less environmental impact. The recent development of Metal Organic Frameworks (MOFs) may prove a major milestone in achieving these goals. The MACADEMIA project is an extension to the FP6 project DeSANNS, which highlighted some MOF materials for CO2 capture and storage. It allowed expanding MOF synthesis on a much larger scale, as well as multiple applications. TOTAL’s focus is on bringing applications of MOFs to key market sectors -gas separation and storage, and liquid separation. The TOTAL-led consortium, with Total Petrochemicals and Refining-Marketing as industrial partners, 11 academic partners from across EU, one leading South Korean partner, among world leaders in their particular domain of MOF science, contributed to the project, with a dedicated management partner.
The main objectives of the MACADEMIA project are:
- production of new MOFs and optimisation of those already of promising interest;
- characterization of MOFs using specialised techniques,
- use of predictive modelling for gas- and liquid phase applications, with Grand Canonical Monte-Carlo and Moleculr Dynamics tools,
- testing MOFs in a wide range of applications, such as gas and vapour recovery and separation (propene /propane, acid gases separation, CO2 and H2 purification), liquid phase adsorption and separation (xylene separations, recovery of N- and/or S-compounds from hydrocarbons), catalysis for specialty chemicals (Lewis-acid MOFs as catalysts for epoxide polymerization, redox-active MOFs as catalysts for hydrocarbon autoxidation).
Project Results:
WP1 - MOF discovery and synthesis optimisation
The discovery and synthesis optimisation of new MOFs is a tedious, time- and starting material consuming process. Therefore, in addition to the trials and errors solvothermal synthesis, high-throughput methods were employed in this step at CAU and CNRS-ILV and only minute amounts of materials were obtained. Based on these results first scale up reactions to the gram scale were carried out and selected materials were further synthesized in even larger amounts at CNRS-ILV (5-40 g) and then KRICT (0.1-2 kg).
Since the main application focus of MOFs was in the fields of separation and catalysis, stable MOFs were targeted and hence tri- and tetravalent metal ions (Fe3+, Al3+, Cr3+, Ti4+, Zr4+) were mainly used. These lead mostly to the formation of microcrystalline reaction products which hampers the structure determination of the MOFs.
WP 1.1. Synthesis optimization, scale-up and shaping of selected MOFs
Based on the screening of the properties of known MOFs, the synthesis optimization of series of selected functionalised solids (-NH2, -NO2, -Br, -CO2H, -2CO2H…) based on the robust metal terepthalates UiO-66(Zr) and MIL-125(Ti) or the hydrothermally stable iron polycarboxylates MIL-127(Fe) and MIL-100(Fe), was investigated. Ambient pressure and dynamic conditions were selected when possible. As high as possible space time yields, cheap and safe precursors as well as green conditions (water, alcohols) were privileged for each solid. In order to gain understanding prior to the synthesis optimisation work, in situ energy dispersive crystallization studies were carried out in parallel at the Synchrotron for some of the selected solids. Finally, the scale-up from the multigram scale to the kg scale was performed further optimizing the synthesis conditions (STY, temperature, concentrations of precursors, type of metal precursors, etc.) while new shaping methods were developed, leading to very little loss of performances compared with the powdered materials, in order to cope with the weaker mechanical stability of MOFs compared to inorganic porous solids.
WP 1.2. Synthesis of functionalized MOFs
Within the project functionalized derivatives of the framework types MIL-53(Al, Fe),MIL-88(Fe), MIL-125(Ti), CAU-1(Al), CAU-10,1UiO-66(Zr) and Cr-MIL-101 were synthesized in characterized in detail. In addition to the use of functionalized terephthalic and isopthalic acid, post-synthetic modification reactions were employed to introduce new functional groups. An important example is the nitration of Cr-MIL-101 using nitric and sulfuric acid and the subsequent reduction with SnCl2. yielding the Cr-MIL-101-NH2 which can be used in PSM reactions.1j The first keto-functionalized Al-MOF (CAU-8) was obtained using benzophenonedicarboxylic acid (H2BPDC).
WP 1.3 New MOFs based on mixed-linker or mixed-metal systems
Mixed-linker MOFs were investigated in order to tune the sorption properties and to study synergestic effects. First, in order to develop new porous architectures, di and tri-topic polycarboxylate linkers (1,4 BDC and BTB) were used to produce new micro- and meso-porous iron(III) MOFs of the MIL-142 (∅ = 7-12 Å)and MIL-143 (∅ = 12-25 Å) structure types. Noteworthy, functionalized terepthalates could also be used to produce these architectures. An alternative consisted in mixing linkers within the same structure type. For instance, starting from CAU-10 ([Al(OH)(1,3-BDC)] (1,3-BDC = isophthalalte) mixed linker compounds CAU-10-H/SO3H, CAU-10-H/Br and CAU-10-NH2/NO2 were synthesized and characterized. Through the amount of -SO3H groups the hydrophilicity of the pores can be tuned. Cr-MIL-101 derivatives were synthesized in order to make use of its stability in possible post-synthetic modification reactions. Thus, terephthalic acid containing various functional groups (F, Cl, Br, OH, NO2, NH2, CH3, COOH) were tested and synergistic effects were observed when TA-COOH was employed.
WP 1.4. Exploratory syntheses
Exploratory syntheses using Al salts employing the high-throughput method led to a number of new Al-based MOFs: CAU-1, CAU-3, CAU-4, CAU-6, CAU-8, CAU-10, CAU-13. Using non-aqueous reaction conditions unprecedented inorganic building units such as octameric, dodecameric and polymeric (based on the well known Al13-ion) were observed in CAU-1, -3 and -6 respectively. Variation of the linker geometry from linear (1,4-H2BDC) to V-shaped (1,3-H2BDC) opened the way to a new series of compounds (CAU-10). Extension of the linker (benzophenonedicarboxylic acid H2BPDC) resulted in yet a new MOF (CAU-8) which contains small pores (∅ = 4Å) in addition to one-dimensional channels (∅ = 8Å). Employing microwave assisted heating the manganese analogue of Cr-MIL-100 could be synthesized and the extension of the linker from 1,4-H2BDC to the naphthalene derivative 2,6-H2NDC allowed us to synthesize the Cr-MIL-101-NDC derivative which contains pores with a diameter of up to Å.
Using iron(III) salts, one could obtain a new porous iron(III) biphenyldicarboxylate MO, denoted MIL-126(Fe), that bear an interwoven MIL-88D architecture. The interpenetration ruled out any flexibility of the structure leading to a highly microporous rigid MOF (∅ = 4-10 Å) possessing a high amount of Lewis acid sites.
In the case of Zr, one could isolate and characterize a whole new series of micoporous rigid Zr4+ dicarboxylate MOFs, denoted MIL-140’s or ZrO[O2C-R-CO2] (R=benzyl, napthyl, biphenyl or dichloro-azobenzene).2c Noteworthy, the inorganic secondary building unit is comprised of a Zr oxide chain while the connection through the dicarboxylate anions leads to a 1D triangular pore system (∅ = 4-10 Å). Remarkably, despite their higher density and lower surface areas, the stability (hydrothermal, mechanical) of this series is much higher than those of their UiO-66 polymorphs while a significant amount of Lewis sites was also evidenced.
NB: note that series of powdered samples of new porous Al, Ti or Zr MOFs were still under structural investigation at the end of the project and will not be described here. WP2 Detailed structural and surface characterisation
The studies within WP2 allowed the full characterisations of many MOF materials. The results highlighted here are important:
- for materials optimisation by the characterisation of their chemical sites and evolution
- for gas processing applications by the study of gas diffusivities and the study of the adsorption sites
- for liquid phase separations
- for catalysis applications by the study of the reaction steps involved.
JUK
Infrared characterisation has been applied to ascertain surface properties of defined target samples: acidity, redox and exchange properties have been analysed on CuBTC samples by JUK using CO, CO2, pyridine, acetonitrile, thiophene and H2O together with D2O as probe molecules. Specific water interactions were remarked, being different depending on the sample preparation protocol proving that for most cases thermal treatment (and not hydrothermal treatment) was responsible for reported instability of MOF materials.
For most cases delivered MOF samples were investigated in terms of their optimal activation conditions, thermal stability, traces of impurities, hydrophilicity and affinity towards specific compounds.
Hybrid nanomaterials were obtained by post-synthesis combination of the two components: tungsten heteropolyacid (HPW) and Basolite™ F 300 used as a support showing the influence of the location of the active sites on the internal versus external surfaces of the crystals in hybrid catalysts on selectivity in ethanol conversion. Such influence is very important for the reaction with the use of bulky substrates/products, its understanding may be helpful for the overcoming of the diffusion limitation in nanoporous materials, like zeolites or MOFs.
As the highlights of the work, carried out during the MACADEMIA grant, several achievements can be named, most probably applicable to the majority of MOFs:
1. possibility of formation of anhydride in the case of MOFs bearing not saturated –COOH functionalities. Such transformation enabled increase of the sorption capacity and subsequent post-synthesis treatment leading to valuable catalysts (including NHPI: N-Hydroxyphthalimide, a catalyst of oxidation synthesis reactions);
2. reversible/irreversible storage of NO (and NO2) on amine-functionalized MOFs, which reversibility is dependent on the fact whether primary and secondary amines are introduced into the linker molecules.
CNRS-LCS
In situ and operando IR spectroscopy have been used as a technique particularly adapted for the investigation of porous compounds. A large amount of the MOFs studied in the project have been characterised at CNRS-LCS. This allowed notably the optimisation of the activation process for the selected solids before their upscaling. Chemical properties were evidenced on each material, which allowed to finely tune their properties throughout the project.
Besides monitoring from both the qualitative and quantitative point of view the physico-chemical properties of solids, these tools have been effective for the description of the properties of selected MOFs under duty.
In view of the propane/propene separation, the study on the MIL-100(Fe) sample, revealed that from a gas flow, both gases are adsorbed on the solid. An increase in the activation temperature resulted in a greater amount of propene adsorbed onto the solid, while the amount of propane remained constant .
Using quantitative IR spectroscopy of the NO and CO probe molecules, we demonstrated that the thermal treatment at increasing temperatures leads to the formation of Fe2+ sites having a higher affinity for the unsaturated C=C bond of propene than Fe3+. Interesting insights in the formation of the reduced iron sites and their role in the C3 separation were obtained by the selective poisoning of Fe2+ entities by strong NO adsorption. This knowledge of the structure–activity relationship gives precious indications to optimise the sample. In other words, an iron MOF with a larger number of FeII CUSs is required in order to increase the separation effect for an alkane/unsaturated hydrocarbon mixture at low partial pressure.
A possible reactive transformation of the ligand was shown to occur by the formation of diazonium species on functionalized MIL-125(Ti)-NH2. This could allow alternative post-synthesis ligand modification which can favour materials modification. On this solid, the formation of diazonium species was observed, verified and explained through a mechanism via the reaction of nitrosation of the amine by nitrous acid, giving a N-nitroso compound as intermediate, which tautomerizes to a diazohydroxide, successively protonated and dehydrated. The main reaction intermediate was identified as being NO2, issued by the dismutation of NO and producing nitrous acid via a reaction with adsorbed water or OH groups.
A catalytic transformation of methanol was demonstrated on MOFs. Methanol was found to adsorb dissociatively on different MOFs and to react into dimethyl ether and/or CO2. On MIL-100(Fe), these phenomena could be associated with the oxidation states of iron, while on the other non-reducible samples they would be related to the acid/base properties of the materials, favouring the dehydration/dehydrogenation steps of methanol and dimethylether into methyl formate, successively decomposing into CO2.
UWAR
In situ diffraction of guest adsorption UWAR has tested the structural modifications induced by liquid phase sorption of various organic guests by the host MOF MIL-53(Fe). This has allowed the swelling of the structure to be tracked in real time and the kinetics of uptake of various guest molecules to be compared directly. In the case of N/S heterocycles, molecules that must be separated from crude petroleum to avoid poisoning of processing catalysts, the solvent not only influences pore opening but is also a competing guest. Successful uptake by MIL-53(Fe) requires that the adsorbate is primarily a good hydrogen bond acceptor; additionally, based on UV-visible spectroscopy, a charge-transfer interaction between the S atoms of benzothiophene and the aromatic rings in the MOF pore wall is proposed. Since there is not a strong guest-metal interaction this system may be useful as a recyclable host for N/S adsorption since the guest molecules can be released readily after uptake.
Characterisation of Mixed-Metal MOFs Compared to mixed-ligand systems, mixed-metal-organic frameworks where the structure is preserved upon partial metal substitution are relatively rare in the literature, but should provide a subtle way of tuning properties such as magnetism or catalysis. We have prepared and characterised novel mixed-metal iron-vanadium analogues of the 1,4-benzenedicarboxylate (BDC) metal-organic framework MIL- 53. The identity of the materials was confirmed using high resolution powder X-ray diffraction and the oxidation states of iron and vanadium in all samples are verified using XANES spectroscopy at the metal K-edges, and for iron confirmed by 57Fe Mössbauer spectrometry. Infrared and Raman spectroscopies as a function of temperature allowed the conditions for removal of extra-framework species to be identified, to form single-phase MIL-53-type materials, (Fe,V)III(BDC)(OH,F). The iron-rich phase shows structural flexibility that is distinct from either the pure Fe material or the pure V material, with a thermally induced pore opening upon heating that is reversible upon cooling.
CNRS-IRCE
Knowledge of gas dynamics in MOFs is important for the design of gas processes.
By combining two experimental techniques operating at different time scales with molecular simulations, we characterized the peculiar dynamics of benzene confined in MIL-47(V): quasi-elastic neutron scattering (QENS) probes diffusion at short time scales (10-13 – 10-8 s), and simulations yield diffusivities in agreement with QENS but show in addition less frequent motions which can only be observed by 2H NMR. The picture we obtain leads to a molecule which is diffusing along the 1D channels, while randomly rotating by 90° jumps, performing thus a corkscrew motion. Benzene is expected to travel along a minimum energy pathway involving jump sequences between these regions.
Two diffusion regimes were found for xylenes isomers in the same MOF: a low temperature regime where the xylene molecules are close to the MIL-47(V) pore walls with a high activation energy barrier for diffusion, and a high temperature regime (above 300 K) where the xylenes are mainly located in the center of the channels associated with a lower activation energy. It has been further shown that the diffusivity is only slightly affected on the ns timescale when one compares xylenes in single component and in mixtures. However, large changes are found by 2H NMR (time scale > 10-7 s) for reorientation motions in mixtures. For example, para-xylene is much more mobile in a 50/50 mixture with meta-xylene. This indicates that packing effect and guest-guest interactions are crucial on long time scales and will influence separation processes.
WP3 Molecular modelling
The accurate computational description of interaction energies of molecules with MOFs containing coordinatively unsaturated transition metal sites (cus) represents a significant challenge for contemporary quantum chemistry. Commonly used force fields or methods based on density functional theory often fail to provide even qualitatively correct results. As an important achievement within the MACADEMIA project we consider the DFT/CC method developed for the accurate description of adsorption in MOFs. The method is based on the pair- wise representability of the DFT error E → ECCSD(T) EDFT which is calculated as the difference between the interaction energy obtained at the accurate reference level method, coupled clusters CCSD(T) in our case, and particular DFT functional. The DFT/CC method has been applied for the description of adsorbate/adsorbent systems (including H2O, CO, CO2, CH4, C3H6, C3H8 adsorption on CuBTC). The agreement was demonstrated for experimental and calculated adsorption heats at low coverage (experimental data obtained in collaboration with CNRS-MADIREL). Based on the excellent agreement between the experiment and calculations the mechanisms of adsorption have been proposed. In addition, DFT/CC method has served as a benchmark for fitting of new highly accurate force fields in collaboration with UEDIN and CNRS- ICGM.
In complement to this, high-throughput screening approach based on Monte Carlo simulations has been developed to not only predict the optimal existing MOF candidate for xylene separations but also to design a novel MOF with outstanding performance for the selective adsorption of acid gases (CO2, H2S) which was further successfully synthesized in the frame of MACADEMIA.
Indeed, studying xylene separation in a large number of MOFs we discovered a few simple rules that allowed for quick and efficient screening of MOFs and could guide the experimentalist searching for promising structures. Simply determining the pore diameter using geometrical methods, we found that MOFs selective to p-xylenes should ideally have one dimensional pores with pore sizes in the range between 4.2 Å and 5.5 Å. Electrostatic interactions which could be introduced to a framework e.g. by choosing an appropriate linker or post-synthetic modification further enhance this selectivity. Our simulations which showed good agreement with available experimental results also evidenced that sometimes very subtle effects can have a significant impact on the selectivity. For example in MIL-125 in the octahedral cages, separation of the xylene molecules happens on the basis of differences in packing and interaction with the pore walls, whereas the smaller tetrahedral cages are capable of separating molecules by molecular sieving.
Regarding the selective capture of acid gases, the first step consisted of considering a collection of recently synthesized MOFs (more than 100 candidates), which can a priori lead to a breakthrough in terms of separation performance for CO2/N2 and H2S/CH4 mixtures. Regarding this sampled series of materials, our grand canonical Monte Carlo (GCMC) simulations evidenced that the UiO-66(Zr) solid shows the highest CO2/N2 and H2S/CH4 selectivities when its terepthalate ligands are grafted by one – COOH group. This conclusion prompted us to further imagine a hypothetical UiO-66(Zr) material containing a second grafted carboxylic function in order to verify whether its separation performance is even higher. This modeling effort was based on a computational assisted structure determination strategy (force field based and DFT geometry optimizations) to propose a plausible structure candidate for this MOF. Starting with the parent UiO-66(Zr), the crystallographic structure of the material was first built using a ligand replacement strategy proposed in our previous studies. This newly designed solid labeled as UiO-66(Zr)-(COOH)2 exhibited simulated CO2/N2 and H2S/CH4 selectivities reaching ~90 and ~70 respectively, values significantly higher than that for the commonly used zeolite 13X. These excellent predicted separation performances further initiated a synthesis effort that was deployed to prepare a la carte this novel material. UiO-66(Zr)-(COOH)2 was indeed successfully synthesized by means of an environmentally friendly synthetic route with the involvement of an aqueous media. The simulated performances were confirmed by real co-adsorption measurements performed by the FPMS partner. Molecular Dynamics simulations further predicted that despite a relatively bulky environment of the cages due to the presence of the two carboxylic functions that at first sight would suggest a very slow diffusion process or even the absence of any possible mobility, the resulting diffusivity for the CO2 species is finally very similar to the values previously reported in the conventional faujasite 13X for the same range of temperature. This was confirmed by the quasi-elastic neutron scattering measurements performed by CNRS-IRCE. Indeed, such a conclusion emphasizes that the kinetics will not be a drawback for the use of such a novel material in physisorption-based processes.
Throughout this project, we have seen that even MOFs that are relatively rigid (i.e. do not show large structural changes such as breathing) often show inherent flexibility in their linkers. However, taking this flexibility into account in molecular simulation through flexible force fields is computationally expensive not least as such force fields have so far only been developed for a limited number of MOFs. Whereas the frameworks for MOFs can be treated as rigid when simulating equilibrium adsorption, framework flexibility has to be considered carefully in Molecular Dynamics simulations when studying the diffusion of larger molecules in order to get a realistic description of the movement of the molecules. For example, in MIL-47 framework flexibility only plays a significant role at lower temperatures (at higher temperatures the predicted diffusivities in a rigid and a flexible framework are similar). In contrast, in UiO-66 either no movement (xylenes) or limited motions (carbon dioxide, methane) of molecules from pore to pore is observed when treating the framework rigidly, which would mean that experimentally no diffusion should be measured which is clearly not the case when the partner CNRS-IRCE performed quasi-elastic neutron scattering measurements. When framework flexibility is incorporated in the simulations, the size and shape of the window is altered through the rotation and distortion of the window-defining BDC linkers. This distortion is in response to guest-framework interactions and is sufficient to allow adsorbate molecules to move between pores. Thus our work has shown that the flexibility of MOFs must be considered in cases where the dominant diameter (i.e. the smallest pore or window diameter) and kinetic diameter of guest molecules are similar. Additionally, in those cases where specific interaction sites are of importance, the impact of framework flexibility should be considered. This is case even for those MOFs which show no significant flexibility or structural changes in experiments.
Process simulation is a tool that can successfully be applied for a better understanding of all the phenomena involved in a given process without the need of doing unnecessary, time consuming and sometimes expensive experiments, otherwise necessary for achieving the same level of expertise.
The knowledge gained through the process simulation is extremely important for guidance in the experimental runs. The accuracy of the simulator depends on the mathematical model used. The complexity of the model should be large enough to account for all the main mechanisms involved in the process while, at the same time, keeping the computational effort necessary to simulate as low as possible. The process simulation should then account for all the phenomena at the particle scale (equilibrium and kinetics of adsorption), all the phenomena at the bed scale (axial dispersion, pressure drop and heat exchanges among the particle, fluid, columns and its surroundings), and all the aspects of the unit such as number of layers, number of columns, valve equations, compressors, vacuum pumps, storage tanks, cycles and all the flows in and out of the columns. From the process simulation it is then possible to predict the unit’s productivity, the product purity and recovery, as well as the energy consumption.
Therefore, particle scale and bed scale adsorption models were presented in Deliverables D3.22 (gas phase) and D3.25 (liquid phase). These models were subsequently used in the process simulation of PSA and SMB units, their sizing, and optimization of such units (D3.26 D3.27 D3.30 D3.31 D3.32 and D3.33). The separation of propane/propylene mixtures in pilot scale units was studied by simulation. Two technologies were considered: pressure swing adsorption (PSA) and simulated moving bed (SMB). The best adsorbent was considered to be the CuBTC spheres provided by KRICT. In the pressure swing adsorption process two units in series were required to obtain polymer grade propylene (purity > 99.5%) and high purity propane (> 96.5%). In the first unit, a cycle with five steps was used (adsorption, rinse, blowdown, purge, countercurrent pressurization) for the production of a high purity propylene stream. For a feed stream with 75% of propylene in propane, a 99.5% purity propylene stream and a 39% propane stream are obtained in the first unit. The propane enriched stream obtained in the first unit is separated in the second unit for which a five step cycle was designed. A high purity propane stream (96.7%) is produced in the second unit and the propylene enriched stream is recycled to the feed of the first unit. The performance of the global PSA unit was estimated to be:
Propylene purity/recovery: 99.5%/98.9%, Propylene productivity: 0.37 mol/kg/h, Propane purity/recovery: 96.7%/98.6%, Propane productivity: 0.12 mol/kg/day, and Power requirements: 116.8 Wh/molC3H6. In the simulated moving bed process, isobutane was used as desorbent. Using a 1-3-3-1 configuration, the separation region was reported for purity values in the extract outlet over 99.5 % and raffinate outlet over 96.5 % considering operating pressures of 1.5 bar and 2.5 bar. The performance of the SMB process is calculated for an operation point selected within the separation region obtained for 1.5 bar. The performance of the process for operation point selected within the separation region for 1.5 bar was: Propylene purity/recovery: 99.9%/99.8%, Propylene productivity: 0.97 mol/kg/h, Propane purity/recovery: 99.4%/99.8%, Propane productivity: 0.32 mol/kg/day, and Desorbent consumption: 1.973 NL/min. For the nitrogen/propylene separation, an industrial-scale PSA column was designed, several PSA cycles and processes based on these cycles were proposed for recovery of only nitrogen (Case 1) and for recovery of both components, i.e. nitrogen and propylene (Case 2). The effect of operating parameters on the process performance parameters were examined based on simulations. For the nitrogen recovery, the best result was obtained with the 5-step (pressurization, adsorption, cocurrent depressurization, blowdown, and purge), 2-column PSA process. From this cycle a nitrogen product stream of 95.4 % purity was obtained with a recovery of 85.2 %. The adsorbent productivity was 6.0 mol N2/kg adsorbent/h and power consumption was 156 W/kg N2 produced/h. Among the PSA processes proposed for recovery of both components, the one which is extended from the 6-step cycle to a 4-column process exhibited better performance than those (3-column) based on the 5-step cycle. Nitrogen with 96.2 % purity and propylene with 97.6 % purity were obtained from the 6-step 4-column process proposed for recovery of both components. The nitrogen and propylene recoveries were determined as 98.4 % and 91.0 %, respectively. The nitrogen and propylene productivities of the process were estimated as 4.61 and 1.83 mol product/kg adsorbent/h and the power consumption as 383 W/kg nitrogen produced/h. In deliverables D3.27 D3.31 and D3.33 the separation of xylene isomers mixture was studied in different materials (MIL-53(Al) – Basolite A100, UiO_66(Zr) and MIL_125(Ti)_NH2), by simulated moving bed (SMB) technology. Although MIL-125(Ti)-NH2 is a para-xylene selective material, all
studied material still present poor behavior in for this separation. However, some new ideas can be explored using MOF based adsorptive processes. The reverse shape selective of UiO-66(Zr) brought the possibility of finding materials that can produce para-xylene as the light component, and highly ortho-selective materials like MIL-53(Al) open the possibility of new hybrid processes combining SMB and crystallization.
To simulate the performances of different adsorbents for natural gas purification (H2S/CH4), syngas purification (H2S/CO2) and nitrogen-propene separation, four macroscopic thermodynamic models were developed: the Ideal Adsorbed Solution Theory (IAST), the Real Adsorbed Solution Theory (RAST) and the Vacancy Solution Theory with the activity coefficients of Flory-Huggins (VST-FH) and with the coefficients of Wilson (VST-Wilson). These models were able to predict co- adsorption isotherms and selectivities for a series of MOFs (MIL-47(V), MIL-101(Cr), MIL-125(Ti), MIL-125(Ti)-NH2, UiO-66(Zr) and UiO-66(Zr)-NH2) with respect to the gas mixtures mentioned above.
While all macroscopic models converged toward similar single component adsorption isotherms, they led to some deviation for the predicted adsorbed amounts. Significant differences have been pointed out for the predicted selectivities, in the curve shapes and in the calculated adsorbed amounts. Moreover, the IAS model seems to be inefficient to simulate the presence of a minimum in the selectivity profile with the loading. This conclusion clearly emphasizes that it is of great importance to perform real co-adsorption for an accurate estimation of the selectivity of MOFs containing specific adsorption sites. Furthermore, these real co-adsorption measurements allowed us to use improved models such as (i) the Wilson activity coefficient model, which takes into account the binary interactions, or (ii) the RAST model which both require mixture adsorption data to simulate the adsorption behaviour of mixtures.
WP4 Applications of MOF materials in gas adsorption/separation
High-throughput analysis of materials
Hundreds of samples have been prepared in the MACADEMIA project. Many of these have never left the synthesis labs as they were deemed not of sufficient quality. In some cases, we were not sure whether a given sample was of interest or whether it was optimised. To screen these, as well as optimised materials, a unique high throughput high pressure adsorption apparatus was developed to measure single gas isotherms to up to 40 bars. With less than 100mg of sample it is now possible to evaluate whether the porosity is optimal, its cycling behaviour (reversibility/regenerability), effect of adsorption temperature and to predict mixture properties using IAST. One problem which arises from the results obtained in this work package is how to compare these results with those obtained with the reference materials (zeolite 13X and Takeda 5A active carbon). This was eventually carried out with the development of an ‘Adsorbent Performance Indicator’ (API). This indicator was then used in other workpackages to select the most appropriate materials for upscale and analysis at stage 2.
Separation of propene from propane
Propane/propylene separation is commonly carried out by low-temperature distillation at high pressures; however, this is an energy-intensive distillation. Adsorption is a separation technology with a great potential for reducing energy consumption compared to the conventional distillation. Separation by adsorption could be carried out either by a cyclic pressure swing adsorption (PSA) process or by means of simulated moving bed (SMB) technology. The primary requirement for any adsorption process is to find an adsorbent with sufficiently high selectivity, capacity and life. A wide variety of adsorbents have been studied to carry out propane/propylene separation: silica gel, activated carbons, carbon molecular sieves, zeolites, chemical adsorbents with complexing metals (i.e. copper or silver ions), and most recently, MOFs. Although promising, the huge number of MOFs turns difficult the identification of individual candidates for specific applications.
Two generations of shaped MOFs have been evaluated as potential adsorbents to carry out propane/propylene separation by adsorption. Isobutane is proposed as a possible desorbent for Simulated Moving Bed (SMB) operation. In order to evaluate these materials, propane, propylene and isobutane adsorption isotherms were obtained by means of a gravimetric method, and single, binary, pseudobinary and pseudoternary breakthrough curves were obtained in a fixed-bed adsorption unit. Seven different shaped samples were evaluated, the 1st generation comprised Cu-BTC tablets (provided by BASF), MIL-100(Fe) (provided by KRICT) and Mg-Formate (BASF). Within these samples the best performing one was Cu-BTC tables, which present high selectivity towards propylene, and good capacities far higher than the capacities presented by 13X zeolite. Within the 2nd generation we could find the following samples: Cu-BTC tablets (KRICT), Cu-BTC spheres (KRICT), UiO-66(Zr) (KRICT), and MIL-125(Ti)_NH2 (KRICT). The UiO-66(Zr) and MIL-125(Ti)_NH2 presented poor selectivity between propane and propylene. However, the Cu-BTC spheres provided by KRICT we very selective, as selective as the CU-BTC material provided by BASF, but with far higher capacity. This material presented very low capacity loss when compared to the Cu-BTC powder material. Furthermore, it was demonstrated that the production of polymer grade propylene (purity above 99.5 %) is possible by PSA using CuBTC from BASF with a recovery of 28.5 % and productivity of 0.8 mol kg-1 h-1. Additionally propane was also obtained with a high purity (99.5%) and a recovery of 99.5 %. For CU-BTC from KRICT (spheres), the cycle carried out produced propylene with purity of 99.8 % (above polymer grade specifications), with a productivity of 0.6 mol kg-1 h-1. However the propylene recovery was low: 15 %. If part of the raffinate streams produced during the feed and rinse steps are to be recycled for their use in the pressurization and purge steps, the recovery can be higher.
Nitrogen recovery from light hydrocarbons
The separation process for the nitrogen purge gas recovery from light hydrocarbons is an important part of the polypropylene production. The operation conditions for this process were provided as the feed composition of 30% C3H6 and 70% N2, 70 °C and 1 bar by TOTAL. The process aims to recover nitrogen with purity higher than 95 mol%.
During manufacture of polypropylene, nitrogen gas is used in a polymer-degassing step to remove unreacted monomer, solvents, and additives from the polypropylene. The off-gas stream is processed to recover a portion of the hydrocarbons, but the remaining nitrogen and hydrocarbons are disposed of through combustion either in a flare or by diluting a fuel gas stream. The N2 recovery process enables reuse of the substantial raw material, reduces the polypropylene manufacturing cost and results in a more ecofriendly polypropylene manufacturing system. In this task N2 recovery from the N2/C3H6 mixtures using one of the second generation MOFs, MIL-100(Fe) granulates supplied by KRICT, was investigated by means of adsorption dynamic experiments using fixed bed. The single and binary breakthrough curves of nitrogen and propylene were obtained at 343 K and 1.3 bar. This was followed by the validation, with lab-scale PSA experiments, of the mathematical model, previously employed in the optimization of the industrial-scale PSA processes for the recovery of nitrogen and propylene from the propylene production purge. Three PSA cycles, one for recovery of only nitrogen (Case 1) and two for recovery of both nitrogen and propylene (Case 2), were proposed and carried out experimentally on a pilot-scale single-column PSA unit available at LSRE. The temperature, pressure, and concentration histories acquired during the experimental runs of the cycles were compared with the results of simulations employing the mathematical model. By simulation of an industrial scale process it was possible to design a cycle where a nitrogen product stream of 95.4 % purity was obtained with a recovery of 85.2 %. The adsorbent productivity was 6.0 mol N2/kg adsorbent/h and power consumption was 156 W/kg N2 produced/h. Among the PSA processes proposed for recovery of both components, the one which is extended from the 6-step cycle to a 4-column process exhibited better performance than those (3-column) based on the 5-step cycle. Nitrogen with 96.2 % purity and propylene with 97.6 % purity were obtained from the 6-step 4-column process proposed for recovery of both components. The nitrogen and propylene recoveries were determined as 98.4 % and 91.0 %, respectively. The nitrogen and propylene productivities of the process were estimated as 4,61 and 1.83 mol product/kg adsorbent/h and the power consumption as 383 W/kg nitrogen produced/h. Regarding to the most interesting applications (Biogas and Natural Gas purification, CTL and CTO processes), the measurement of the mixture adsorption equilibrium was performed for H2S/CH4,k,. Nevertheless, as far as purification of adsorbable gases is concerned, it appears necessary to obtain experimental adsorption data at low concentrations for the several components of the mixture (acid gases, sulfur compounds and/or carbon dioxide compounds in natural gas, in air or impurities in gases for semi-conductor applications). After performing pure adsorption isotherms measurements, 10 MOFs, regenerable and stable under H2S atmosphere, have been identified. The MIL-100(Al) presents the biggest adsorption capacities but the CH4 adsorption is too high for the H2S/CH4 separation process. MIL-125(Ti)-NH2 presents the biggest H2S slope at zero coverage with a good adsorption capacity. The UiO-66-BTEC has lower adsorption capacities than most of the other studied adsorbents but the lowest ones correspond to CH4 Co-adsorption measurements were performed on the most promising adsorbents. The H2S/CH4 selectivities on shaped MIL-125(Ti)-NH2 are good: 22 and 35, at 5 bar and 50 bar respectively, and the adsorption capacities are correct. It is a very promising material for H2S removal from natural gas or biogas. The best H2S/CH4 selectivities are obtained on shaped UiO-66(Zr)-BTEC: more than 40 at the all range on H2S concentration (0-10%) at 5 and 50 bar, but unfortunately the adsorption capacities are lower. It is also very promising materials for H2S removal from natural gas or biogas. It is important to note that the result of the mixture adsorption equilibrium predictions shows how it is necessary, for selective systems with different kind of adsorption sites as the main interesting systems on MOFs, to realize mixture measurements and not to only use macroscopic models to provide selectivity.
C6 / aliphatics separations
We could design and build a set up allowing for the study of the co-adsorption as well as the separation of benzene from hydrocarbons. This set-up also allows for measuring transient partial pressure of each component of the mixture being adsorbed on a MOF. Furthermore the enthalpy of adsorption can also be directly measured. The most significant result is the co-adsorption of benzene and n-hexane onto MIL-101(Cr) at ambient conditions. We could show that benzene is preferentially adsorbed as compared to n-hexane, even in the very early stages of the adsorption process. It has been shown that once benzene is adsorbed, n-hexane can also adsorb in the same material, likely on different surface sites.
WP5 Application of MOF materials in liquid phase adsorption/separation
Xylene separation
Because of their close boiling points, xylene isomer separation is a challenging process. Industrially, these isomers are separated using a KBaY zeolite in a simulated moving bed (SMB) process. The most important isomer is p-xylene which is used in large quantities for polyester and polyethylene terephthalate production. During the project, MOFs with high separation factors were discovered and their application in an SMB process evaluated.
To complete task 5.2 with success, an iterative process was developed between the different partners involved in this task, CNRS-ILV: synthesis of new structures, KRICT: synthesis scale up and material shaping, KUL: small scale screening, and FEUP: large scale screening and performance testing. As result of this process in the end of the MACADEMIA project a p-xylene selective material was found, characterized and tested in pilot-scale SMB unit. Furthermore, within this task many new structures were tested by KUL and FEUP, such as MIL-53(Al), MIL-47(V), MOF-14, UiO-66(Zr), and MIL-125(Ti)_NH2, just to name the most important ones.
From the screening of the newly developed MOFs, one class of MOFs with identical topologies was found to be highly para-selective adsorbents. MIL-125(Ti) ([Ti8O8(OH)4(BDC)6]), MIL-125(Ti)-NH2 ([Ti8O8(OH)4(BDC-NH2)6]), and CAU-1(Al)-NH2 ([Al8(OH)4(OCH3)8(BDC-NH2)6]) (BDC = 1,4-benzenedicarboxylate) are built up from octahedral cages and tetrahedral cages. These cages both exhibit the para-selectivity. In the octahedral cages, the molecules can be separated on the basis of differences in packing and interaction with the pore walls; in the smaller tetrahedral cages, the xylene isomers are separated by molecular sieving.
Shaped MIL-125(Ti)-NH2 was further tested in breakthrough experiments for selective adsorption and separation of xylene isomers, in liquid phase and using n-heptane as eluent. This material is para-selective, allowing the separation of p-xylene from its isomers in ternary or quaternary mixtures.
In another approach, UiO-66(Zr) was chosen to be the 2nd generation MOF, due to its interesting feature of reverse shape selectivity, which could allow recovering p-xylene as the light product (raffinate), besides the obvious possible separation of o-xylene in the heavy product (extract), in a SMB unit. Powder and tablets were tested for selective adsorption and separation of xylene isomers, in liquid phase and using n-heptane as eluent. In the liquid phase, this material exhibited a reverse shape selectivity. The selectivities of 1.6 and 2.3 were determined from the breakthrough experiments, for o-xylene over m-xylene and over p-xylene at 313 K, respectively, in the presence of n-heptane.
Nitrogen/Sulfur Adsorption
As regulation dictates that the S-level in fuels needs to be decreased below 10 ppm, catalytic hydrodesulphurization becomes increasingly difficult. In principle, adsorptive separation could be a solution, but currently available adsorbents, such as zeolites, have insufficient capacity and stability to realistically meet the industrial requirements. Because of their high adsorption capacities and metal content that can be used to selectively adsorb N or S-containing compounds, MOFs were investigated for this purification.
During the project, several structures were identified with superior affinity and capacity for S-and N-containing compounds. CPO-27 and [Cu3(BTC)2] were found to have the highest adsorption capacities for the selective adsorption of the S-heterocyclics, reaching weight% uptakes of elemental sulfur of up to 6 wt%.
Early in the project, MIL-100(Fe) was found to not only selectively adsorb S-heterocyclics, but also to have a very high affinity for N-heterocyclics. The adsorption of these compounds is crucial because they can poison the catalyst used in the hydrodesulphurization. Using a representative raw fuel feed containing 950ppmw S and 750 ppmw N, it was demonstrated that a column filled with MIL-100(Fe) could effectively adsorb all S and nitrogen containing compounds.
In a follow-up study, the influence of the metal ion in MIL-100(Al3+,Cr3+,Fe3+,V3+) on the adsorptive removal of N/S-heterocyclic molecules from fuels was investigated. The results show a clear influence of the different metals (Al, Fe, Cr, V) on the affinity for the heterocyclic compounds, on the integral adsorption enthalpies, and on the uptake capacities. Among several factors, the availability of coordinatively unsaturated sites and the presence of basic sites next to the coordinative vacancies are important factors contributing to the observed affinity differences for N-heterocyclic compounds.
MIL-100(V) was found to be the best adsorbent for the deep and selective removal of nitrogen contaminants out of fuel feeds.
WP6 Applications of MOF materials in catalysis
Lewis Acid catalyzed reactions over MOFs
The possibility to synthesize MOFs with any desired metal, large metal content, large pore size and surface area make them more advantageous than such conventional solid catalysts as zeolites or metal oxides for liquid phase reactions. Another important feature of MOFs consists in the periodicity and regularity of their active sites. Since that, MOFs may be considered as catalysts with single-sites in contrast to metal oxides and zeolites with the random distribution of the first and second coordination sphere around a metal. In this contribution, we aimed to compare the catalytic behavior of selected MOFs (Cu3(BTC)2, Fe(BTC), MIL-53(Al), MIL-100(Cr/Fe), ZIF-8) and some zeolites (TS-1, Beta, USY) in Prins reaction of β-pinene with paraformaldehyde, and Knoevenagel condensation of aldehydes with active methylene compounds.
In Prins condensation of β-pinene and paraformaldehyde to produce nopol, activity of MOFs increased with increasing of the accessibility of Lewis acid centers in the order ZIF-8 < MIL-53(Al) < Fe(BTC) < MIL-100(Cr) < MIL-100(Fe). In contrast to MOFs, the presence of relatively strong Brønsted acid sites within zeolites Beta and USY resulted in the formation of side products of β-pinene isomerisation, decreasing the selectivity of these catalysts. The yield of nopol formed over MIL-100(Fe) increased with the increasing of the relative polarity of the used solvent in the order: dodecane < p-xylene < cyclohexanone < acetonitrile. The high activity, as well as the preservation of the framework and the active centers of MIL-100(Fe) makes possible the reuse of this catalyst without significant loss of the activity at least in 3 catalytic cycles.
Cu3(BTC)2 and Fe(BTC) appear to be more active and selective catalysts for the Knoevenagel condensation than zeolites Beta and USY due to the absence of consecutive reactions of the primary condensation products and favorable diffusion of reactants and products providing by mild acidity of the MOFs and their larger pore size, respectively. Interestingly, while yields of target products over zeolites and MOFs were comparable in condensation with ethyl acetoacetate and methyl cyanoacetate, unexpectedly high activity of Cu3(BTC)2 was observed for benzaldehyde condensation with malononitrile. In the last case aldehyde totally transformed to corresponding unsaturated nitrile after 1 h at relatively low reaction temperature as for acid-catalyzed Knoevenagel condensation (60 °C). Under the same conditions, reaction on the zeolite does not proceed at all. By using computational methods, it was established that the dual coordination of malononitrile molecule to both active sites of Cu3(BTC)2 takes place. It was calculated, that such adsorption complex is characterized by interaction energy that is twice as large as the one for single adsorption site. Presumably, Knoevenagel reaction catalyzed by Cu3(BTC)2 proceeds through the transfer of proton from malononitrile to carboxylic group of framework (formation of defect Brønsted acid site) that can be mediated by the benzaldehyde molecule. Interaction of benzaldehyde with Brønsted site almost entirely compensates the energy loss due to the formation of deprotonated malononitrile and defect acid center. After the final reaction step, dehydration, the original paddlewheel structure is fully recovered.
Consequently, it was shown that the preferences of MOFs over traditionally used zeolites in acid-catalyzed reactions are caused not only by the textural or acid characteristics, but mainly by the regular organisation of the active centers in the framework. In some cases (as it was established for Knoevenagel reaction) the last feature may result in changing of reaction mechanism over MOFs in comparison to zeolites, which provides the higher catalytic activity of metal-organic framework materials. Thus the deep experimental and theoretical investigation of the structure of active centers and their modification during catalytic process seems to be a key not only for the purposeful development of the effective catalysts but also for the understanding of the fundamental perspectives of MOFs in heterogeneous catalysis.
WP7 Scale-up of MOF synthesis, shaping and pilot applications
FEUP
Pilot-scale tests for determination of long-term integrity, separation performances and catalytic activity/selectivity using shaped MOFs were conducted at TOTAL. However, these tests can be time consuming and sometimes expensive experiments, if run without the knowledge gained through process simulation. Therefore, in this work package FEUP partner played an important role with the development of an accurate of the simulator based on a detailed adsorption fixed-bed mathematical model. For instance, in the nitrogen/propylene separation case, a PSA cycle was proposed for recovery of nitrogen from a purge stream of the polypropylene production, with a product purity of 98 %, recovery of about 83 %, and productivity of 1.42 mol kgads-1 h-1 (see Table 2). From this collaboration three partners (FEUP, TOTAL, and KRICT) filed a patent for a process, aiming the recovery of nitrogen or recovery of nitrogen and propylene using MIL-100(Fe) shaped material. The patent is titled “Procédé cyclique de production d'azote de haute pureté et éventuellement d'hydrocarbure de haute pureté à partir d'une charge contenant de l'azote et un hydrocarbure.”, having as inventors F. Luck, C. Leroi, A. Fernandez, A. M. Ribeiro, J. C. Santos, A. E. Rodrigues, J-S. Chang, U- H. Lee, Y-K. Hwang, J-P. Bonne, Patent Application PCT/FR2012/051623, filled on July 9, 2012.
TOTAL
A gas phase pilot was set up successfully for separations like propane/propylene, nitrogen/propylene, syngas purification and natural gas treatment. For each separation, one or several MOFs have been tested. If the propane/propylene separation wasn’t suitable by adsorption with MOF, the nitrogen/propylene separation with MiL-100(Fe) gave good results which were patented.
It is still too early to get an idea of the MOF performances for syngas purification and natural gas treatment. In the first case, the results were very close to those obtained with activated carbon used as the reference. Even with a process optimization, activated carbon might be as good as MOFs, or even better. For natural gas treatment, no reference was available but results with MOF seemed pretty interesting. However, more work is required to estimate the real performances of MOFs in different applications. It is however likely that there is room for adsorption and separations with MOFs in this application.
Potential Impact:
Starting from a strong knowledge based on structure-performance relationships of MOFs, the primary result of this project will be to speed up time-to-market for the development of metal- organic frameworks materials, their processing and applications. In terms of industrial use of MOF materials, for certain applications such as gas sorption/separation, the potential is certain whereas in other fields (liquids, catalysis), additional R&D is required for major key breakthroughs. Gas separation & purification, separation of liquid mixtures and catalysis are three potential high-volume applications, and essential to sustainable, resource-economical chemical production. Propene/propane separation is relevant for propylene production (annual worldwide production 69 million tons), for production of polypropylene. Propane/propylene mixtures from a conventional FCC are separated by distillation in a C3 splitter in quite a complicated set up which can have up to 200 plateaus. In the project, this separation was performed using Pressure Swing Adsorption and Simulated Moving Bed technologies. In the EU, production of a petrochemical intermediate such as propene faces fierce competition with large petrochemical platforms especially in the Middle East and Asia, which have access to locally cheaper feedstocks. Any technical/economical advantage would be most welcome to allow such energy-intensive activities to remain based in Europe.
Both acid gases H2S and CO2 have been chosen by Total with respect to the Coal-to-Liquids (CTL) and Coal-to-Olefins (CTO) processes, which will have to be developed commercially in the next decades. Various solvents can be used to recover the H2S and CO2, such as Selexol and Rectisol. These processes are high capital expenditure intensive and require low temperatures. A sorption based process could prove an interesting alternative; especially as the raw syngas pressure is around 30 bar: the most promising adsorbent for the purification of CH4 containing CO2 and H2S is the MIL-125(Ti)-NH2. Moreover, MIL-125(Ti)-NH2 presents the best kinetic properties, an important advantage allowing better regeneration and improved performances in a PSA process.
Furthermore, with increasing environmental and climate change concerns, it is essential to reduce CO2 emissions. Separating from the feed gas will lead to a major reduction of CO2 emissions from a CTL/CTO plants which then needs to be concentrated to >95%vol. Biogas purification (CH4/H2S separation) is as well a hot topic due to the explosive growth of biogas production worldwide. CO2 purification for Carbon Capture and Storage (CCS) is still a quite recent field without to data any EU directive on CO2 quality required for geological storage. Amine capture processes can result in high purity CO2. However, CO2 resulting from oxy-combustion capture processes can contain significant amounts of impurities, requiring a further purification step prior to storage.
Nitrogen/hydrocarbon separations of interest for Total in the project emerge from the recovery of unreacted propylene after polymerisation. High polymer purity is increasingly an issue thus requiring optimisation of this nitrogen purge step. With 300 polyolefin plants worldwide, monomer losses cost the industry nearly 300 million € per year, representing a considerable resource-recovery opportunity. As demonstrated in the project, MOF-based separation processes could replace condensation which shows quite poor efficiencies. Total, KRICT and the University of Porto filed in July 2012 the PCT/FR2012/051623 patent “Cyclic process for producing high purity nitrogen and optionally high purity hydrocarbon from a feed containing nitrogen and a hydrocarbon”.
Deep desulfurization is required to comply with requirements for transportation fuels (currently at 10 ppmw S for diesel) or to enable use of liquid fuels in fuel cells (S < 1 ppmw). In addition, hydrodesulfurization would be easier if first nitrogen compounds, such as carbazoles, indoles and their benzo-analogs could be removed, as these severely inhibit hydrotreating catalysts at 50 ppmw. Preliminary results showed that some MOFs, e.g. MIL-47 from CNRS-ILV, can absorb up to 32 wt. % thiophene, or 12 wt.% S. KU Leuven also demonstrated high uptakes of N-heterocyclic aromatics in competition with aromatic hydrocarbons.
Even though all precautions have been taken to perform the evaluation of the performances of the shaped Cu3(BTC)2 in the standard operating conditions typically used at Total Research&Technology Feluy (Belgium) to qualify the sorbent batches, the performances of Cu3(BTC)2 for nitrogen removal were not conclusive. From the results of batch experiments at KU Leuven, Cu3(BTC)2 was expected to have a better selective recovery of S-heterocycles in comparison to N-heterocycles. However, surprisingly, whatever the spiked feed considered (diluted or concentrated case), almost no sulfur removal was observed. N-compounds were only slightly retained on Cu3(BTC)2, and the elution was proceeding very quickly. Repeatability of the tests was demonstrated since the second test performed under concentrated feed gave the same results.
MOF catalysis has the potential to significantly contribute to the greening of chemistry. The reactions selected, viz. epoxide polymerization and alkane autoxidation, are test cases in which MOFs will be key in superseding existing technologies. Progress achieved in MACADEMIA on the Lewis-acid activation of epoxides with MOF catalysts to produce polyether polyols would bring significant advantage with respect to the current technology (double metal cyanide (DMC) catalysts: very high cost and induction time with ethylene oxide). However, the activity of MIL-101(Cr)nano catalyst in consecutive polymerization runs dropped significantly. For fine chemicals production, autoxidation of alkanes such as cyclooctane has a moderate conversion rate, 25 g cyclooctane per g of catalyst. This number needs to be increased several orders of magnitude in order to become a competitive process. However, this is a minimum productivity number, since it would be possible recover, reactivate and reuse the catalyst. Further progress of MOFs is needed in both selected catalytic applications. Development of a specific European knowledge on improved selective oxidation of alkanes is critical to maintain on the long-tem the position of the European chemical industry, including also other applications of selective oxidation such as short-chain alcohols for chemical and clean fuels (e.g. butanol) applications.
Finally, in the rapidly growing field of MOF research, the MACADEMIA project timely addressed major global challenges such as gas- and liquid-phase separations, catalysis for specialty chemicals, and energy-saving processes. Business needs should be addressed in terms of favourable framework conditions for the commercialisation of technologies, adequate numbers of well-trained and mobile researchers responsive to the needs of industry and finally to an excellent public research base (research institutions and infrastructures) with strong interactions with industry.
The main dissemination activities in the 2nd Period have included:
Staff Training: Beneficiaries have employed staff to work on the project who are therefore gaining project related knowledge and experience.Training Workshops on Metal Organic Frameworks were held on 22nd June 2011 in Porto, Portugal and on 14th June 2012 in London, UK. ICON supported the attendance of 9 young researchers to the MACADEMIA workshop on in Porto, Portugal and 2 in London, UK by providing financial assistance for their travel and subsistence costs as described in the Description of Work. ICON also supported the attendance of seven PhD student participants to the 18th Zeolite Forum in Zakopane, Poland (13/09 – 17/-9/2011). ICON supported travel and accommodation costs for a 1-month training exchange visit for one PhD student from JUK to the laboratories of Professor Morris at St. Andrews University. The student was exposed to techniques that currently are not available at JUK. MACADEMIA members participated in 27 meetings, conferences and symposia, presenting material from the project. MACADEMIA members visited each other’s laboratories and promoted student exchanges.
The first issue of the MACADEMIA Project Brief was published. Dissemination activities in the 3nd Period have included Staff Training: Beneficiaries have employed staff to work on the project who are therefore gaining project related knowledge and experience.A training workshop was held on the 11th and 12th June 2013 in Brussels, Belgium which included external speakers from the Free University of Brussels, Belgium, the University of Turin, Italy, University of Granada, Spain, University of Hanover, Germany, University of Gent, Belgium and the Technical University of Delft, The Netherlands.
ICON supported the attendance of three young researchers to the MACADEMIA workshop in London (June 2012, not included in the previous report) by providing financial assistance for their travel and subsistence costs as described in the Description of Work. ICON also refunded travel and accommodation costs to external speakers at the workshop held in Brussels (11th and 12th June 2013). The website (www.macademia-project.eu) has been maintained since going live in March 2010. A dissemination strategy seminar, facilitated by the Commission’s IPR help team was held on June 13th, 2013 in Brussels, Belgium.
97 scientific, peer reviewed articles were published and/or submitted in high impact journals in total over the three years of the project (47 in Period 3).
Participation of MACADEMIA partners at 30 external meetings, conferences and symposia, in the 3rd reporting period presenting material from the project. MACADEMIA partners visiting each other’s laboratories and student exchanges. Publication of the second issue of the MACADEMIA Project Brief.
Exploitation of results:
Three projects partners (FEUP, TOTAL, and KRICT) filed a patent for a process, aiming the recovery of nitrogen or recovery of nitrogen and propylene using MIL-100(Fe) shaped material. The patent is titled “Procédé cyclique de production d'azote de haute pureté et éventuellement d'hydrocarbure de haute pureté à partir d'une charge contenant de l'azote et un hydrocarbure.”, having as inventors F. Luck, C. Leroi, A. Fernandez, A. M. Ribeiro, J. C. Santos, A. E. Rodrigues, J-S. Chang, U- H. Lee, Y-K. Hwang, J-P. Bonne, Patent Application PCT/FR2012/051623, filled on July 9, 2012.
An exploitation seminar was held in Brussels, June 13th, with 13 participants from 9 organisations, animated by Salvador Pastor, ESIC2 Services. At the present moment, only two Exploitable Results had been clearly identified:
● Synthesis optimization and shaping of new MOF materials.
● Workshop / Training Module on MOFs: from synthesis to applications.
It is expected to have specific results for additional MOFs from partner TOTAL within a few months (January 2014); at that time, more relevant Exploitable Results may appear, and a second version of this ESS Report could be delivered with complete new sets of Contribution-Benefit Matrices,
Risk analysis & Priority Maps, and Background/Foreground Information.
List of Websites:
www.macademia-project.eu
