The aim of the group is to employ atomistic simulations to understand the mechanisms that govern chemical processes in heterogeneous catalysis. The analysis of reaction networks, activity and selectivity issues and the final tests on the stability of the potential materials are fundamental to establish a solid background to determine when it can be considered as a catalyst candidate for a given chemical transformation.
Our collaboration with several experimental groups is of fundamental importance to retrieve, compare and define models that can later be applied in the new definition of experiments and new materials to be explored. To this end the use of massive computational resources, as those provide by the RES-BSC are required.
During this year our activities have been centred in the analysis of selectivity in different processes but also on the study of the links between homogeneous and heterogeneouscatalyst: i.e. their similarities and differences for the same catalytic property.
The search for new more active, selective, stable, cheap and environmentally friendly heterogeneous catalysts has traditionally been performed by means of trial-and-error methods. While a large amount of heuristic knowledge has been achieved the lack of systematic studies hinders a fast development in an area that is crucial in 90% of the chemical processes performed in industry. Trial-and-error methods are extensive, time consuming and expensive. Although much knowledge has been achieved by the use of experimental techniques, in particular Surface Science, the lack of systematization constitutes a severe problem. In the last 15 years computational methods have reached the sufficient level of accuracy required for the current chemical problems and enough computational resources are available. This has boosted a new area of science that can be termed as “Theoretical Heterogeneous Catalysis”. Nowadays, it is possible to obtain accurate models based on atomistic first principles simulations that determine the descriptors controlling the relevant chemical properties. Moreover, atomistic simulations provide a systematic way to evaluate the descriptors in sufficiently large populations of potential catalysts thus being at the same time a tool for analysis and prediction. The structured knowledge coming from atomistic simulations can shift the present research mainly supported on trial-and-error procedures to more rationalized methodologies.
In this second year of activity of the Theoretical Heterogeneous Catalysis group we have concentrated on two main issues: selectivity and gold catalysis. Selectivity is a must in the chemical industry. Selective processes reduce the amount of lost material but also diminish the environmental impact of the chemical industries and their energetic costs. The search for selective processes is one of the requirements for the chemical industry in the 21st. century. In parallel, gold catalysis has appeared as a new and exciting field in both homogeneous and heterogeneous catalysis showing high activities and surprisingly high selectivities. Gold catalysts can perform low-temperature oxidations, selective hydrogenations and C-C coupling, the only trick being how to prepare them correctly for the desired application.
With the help of highly parallelized codes and the massive computational resources at the RES-Barcelona Supercomputing Center, we have been able to analyse in detail several factors that affect mainly the selectivity of chemical processes that are relevant to the industry.
For instance, in 2007 we have studied the selectivity in propylene epoxidation from propylene and oxygen. Propylene oxide is an attractive compound since all three C atoms do show different chemical characteristics and therefore behaves as a very versatile synthetic intermediate. However, its production from propylene and oxygen is far from being easy, in particular when compared to that of ethylene oxide. We have found out that when oxygen atoms on the metal surfaces are the active oxidation agents the key property in the design of new more selective catalyst is to devise materials that can provide oxygen atoms of lower basicity that those encountered in the surface of silver (the usual catalyst in ethylene epoxidation). This rule is simple, can be easily computed for a large group of compounds and shows why all previous experimental attempts to find a catalyst for this reaction have failed.
Figure 1. Energy diagram for the hydrogenation of propene and propyne on both gold nanoparticles and palladium surfaces. The number of H atoms added to the C-C bond is indicated. Selectivity in Au-based systems lays on the non-adsorbed nature of propene on gold, i.e. thermodynamic control.
In a second study we have investigated the selective hydrogenation of hydrocarbons. In the cracking units in refineries ethylene and propylene fractions are not completely pure. In general they contain a small amount of C2 and C3 molecules with a high degree of insaturation (i.e. triple bonds). Ethylene and propylene are usually polymerized. The presence of impurities, alkynes, affects the performance of the polymerization units leading to low-quality polymers. Under normal operation conditions alkynes have to be removed from the stream and this is done by hydrogenation to the double bond counterparts. We have performed a detailed analysis on the properties of new catalysts devised in the group of Prof. J. Pérez-Ramírez that contain gold as the main component. For these compounds good activities and selectivities towards ethylene/propylene are found and thus under certain conditions they might be considered to replace the present Pd-based catalysts. From a theoretical point of view the high performance in terms of selectivity of the Au-based systems is due to the preferential adsorption of alkynes to the surface of the gold nanoparticles in the catalyst, see Figure 1. Such adsorption is not possible for alkenes and thus the hydrogenation reaction takes place on the alkyne molecules. Furthermore, when the alkene molecule is formed it desorbes from the surface. Again, theoretical methods based on atomistic first principles simulations were able to determine the descriptor for selectivity. In the present case thermodynamic selectivity, i.e. differential adsorption, is at the core of the selective hydrogenation of alkynes in rich alkene streams.
Figure 2. Energy diagram for the formation of NO and N2 on gold (111) surfaces. The relevant structures are shown in the insets and the relative instability of NO on the Au(111) surface is key for the kinetic control found for the selectivity towards N2.
Selectivity has also been studied for the oxidation of ammonia on oxygen-covered gold surfaces. In contrast to the systems described previously: propylene epoxidation and selective alkyne hydrogenation, this constitutes a more basic science study. Gold is known not to split oxygen molecules and therefore it cannot be employed as oxidation catalysts unless when nanoparticles are present. However, when atomic oxygen is dosed to the surface the individual properties as oxygen donor can be studied in a very easy way employing the techniques of ultra-high vacuum that lead to detailed, high accuracy results. Experiments indicate that in contrast with many other metals, the oxidation of ammonia leads to the excess production of nitrogen instead of NO. N2 formation might be relevant in ammonia abatement from waste streams containing ammonia. We have analysed in detail the reaction network for the decomposition of ammonia on Au(111) but also the assisted decomposition by oxygen or hydroxyl groups present on the surface. Upon complete decomposition the fragments on the surface can rearrange and lead to N2 and NO products. Two aspects are relevant in the present study. First, assisted decomposition is a must due to the inert nature of the Au(111) surface, this is likely to be common to other metals since the water produced can lowered the energy requirements for ammonia dissociation leading exothermic dehydrogenation steps. Second, the barriers for the recombination of the atoms on the surface are much lower than on other metals, again this can be correlated to the lack of activity of gold surfaces. Finally, the barriers for N2 formation and NO formation favour the generation of the first, unlike what happens for Pt, Pd and Rh, see Figure 2. In this work our results have clarified the nature of selectivity in a branched reaction and how we can modify the external conditions to observe a better selectivity towards N2 or NO, see Figure 3.
Figure 3. Selectivity towards N2 production as a function of the relative oxygen to nitrogen coverages on the Au(111) surface. The solid line represents the results obtained with our model and represented in Figure 2 while the squares correspond to the experimental measurements by the group of Prof. B. Mullins in Texas.
A fundamental science aspect that has been tackled in this period is the link between the chemistry of homogeneous and heterogeneous catalysts. The comparison between both approaches has been pointed out as one of the challenges in the development of the field of catalysis. Although some groups consider that there is no connexion between the fields of organometallic, heterogeneous and enzymatic catalysis several indications seem to point out that there is possible to establish a full theory on catalysis taking into account all the aspects that might be relevant under different conditions. The development of a comprehensive unique theory for catalysis constitutes an especially attractive field for theoretical methods since the same techniques can be applied for all kinds of systems. This contrasts with experimental results where the methodologies and technical equipments employed in homogeneous and heterogeneous catalysis differ to a very large extent. In 2007, under the framework of the Consolider Ingenio 2010 project we have constituted a large research group that benefits from the expertise of researchers with different backgrounds combining theory and experiment from both homogeneous and heterogeneous points of view. The aim has been to understand some of the properties that are common to both homogeneous and heterogeneous catalysts. The groups of Profs. A. M. Echavarren, F. Maseras and J. Pérez-Ramírez and our own group have joined efforts in the understanding of alkynophilicity in gold catalysis. Alkynophilicity can be described as the high affinity of Au to selectively act on triple bonds in complex molecules or mixtures.
Alkynophilicity in homogeneous catalysis fosters the formation of C-C bonds and very complex structures coming from highly selective processes starting from molecules containing both–yne and –ene functional groups. In heterogeneous catalysis selective hydrogenation of triple bonds in the presence of double bonds can be achieved with a high degree of selectivity. These two cases represent complex mixtures where double and triple C-C bonds coexist and with gold catalysts only alkynes are systematically modified. Theoretical methods have allowed us to determine that the reasons for selective alkyne activation on both systems are not the same. While for heterogeneous catalyst, nanoparticle-based, the selective activation of the triple bonds takes place for homogeneous systems this control does not apply. In homogeneous catalysis both double and triple bonds can bond the metals and the control is of kinetic origin. Moreover, the high interaction between the –yne functions and the catalysts induces the increase of Lewis acidity of the C-C bond and promotes the formation of new C-C bonds with –ene groups on the same molecule. This process is more difficult on gold nanoparticles where the stabilization of the LUMO of the alkyne is much lower as shown in Figure 4.
Figure 4. Orbital energy diagram for the interaction of ethylene and acetylene with different gold catalysts. The key for C-C formation is the differential stabilization of the LUMO of the alkyne, much larger for homogeneous catalyst than for heterogeneous ones.