Catalytic Activity of MW Prepared Ru/TiO2
Initially, a series of Ru/TiO2 catalysts was prepared from Ru(acac)3 at 150, 175 and 200 °C to assess the effect of preparation temperature and time on catalyst activity for LA hydrogenation. The catalysts were denoted in the manuscript using the following nomenclature: reductant-preparation temperature (°C)-preparation time, e.g., Acac-200-10 was prepared using Ru(acac)3, at 200 °C for 10 min.
The LA hydrogenation activity of the catalysts is presented in Figure 2
. For all the catalysts, the sole product observed was γ-valerolactone. It was found that for a given processing time, increasing preparation temperature resulted in an increase in catalyst activity. In the case of the catalysts prepared with 5 min processing time, LA conversion increased from 41% in the case of Acac-150-5 to 51% and 61% for Acac-175-5 and Acac-200-5, respectively. A similar trend was also observed when the preparation time was increased to 10 min, with these catalysts showing higher activity than the 5-min prepared analogues.
The longest processing time, 15 min, was found to yield catalysts with relatively poor activity; increasing preparation time from 10 to 15 min resulted in a decrease in LA conversion from 69.4% to only 27% for Acac-200 series. The activities of the catalysts are comparable to previously published 1 wt%Ru/TiO2
prepared by an aqueous colloidal method [20
]. Given the similar catalyst activities, the principle benefit of a microwave assisted solvothermal method is the rapid preparation of highly active catalysts using a minimal amount of solvent, and therefore, highly concentrated metal precursors. For comparison, preparation of the previously reported sol immobilization 1 wt%Ru/TiO2
catalysts required over 30 times the solvent volume of the analogous microwave prepared catalyst on a per gram basis, principally to avoid colloidal agglomeration.
The weight loading of the catalysts was measured using MP-AES. Due to the insolubility of Ru species in aqua regia, the most commonly used laboratory acid digestion matrix for AES, the weight loadings were instead determined by comparison between the initial concentration of the dissolved Ru precursors pre-preparation and comparing this to Ru concentration measured in the supernatant solvent post catalyst preparation. The results in Table 1
show that the 1 wt% denotation of metal loading was very close to the measured weight loading and that almost all of the metal in the precursor solution was deposited on to the support.
To better understand the relationship between catalyst preparation parameters and activity, the most and least active Ru(acac)3
-derived catalysts were investigated by TEM, displayed in Figure 3
. The catalysts Acac-200-10 and Acac-200-15 were found to contain highly dispersed Ru nanoparticles, with the former exhibiting a mean particle diameter of 2.35 nm versus 2.55 nm for the latter. The increase in mean particle size could be a result of the sintering of formed nanoparticles or alternatively, surface-induced growth from increased precursor nucleation on the formed nanoparticles. Previous work by Regalbuto et al. investigated the effect of particle size on the activity of alumina and carbon supported Ru nanoparticle catalysts prepared by selective electrostatic adsorption (SEA) for Levulinic acid hydrogenation [17
]. The authors showed that the activity of the catalysts correlated with Ru nanoparticle size, yielding a volcano trend with a maximum around 1.5 nm average particle diameter. This contrasts to the structure-activity relationship observed in this work, wherein a modest increase in average particle size of only 0.2 nm between the acac-200-10 and acac-200-15 catalysts was accompanied by a 61% decrease in LA conversion, suggesting that the broad range of catalyst activities presented cannot be ascribed to particle size effects alone.
The two catalysts, Acac-200-10 and Acac-200-15, were further studied by x-ray photoelectron spectroscopy (XPS) to determine whether their respective catalytic activities were a consequence of Ru surface oxidation state variations. The Ru3d/C1s and O1s narrow scan spectra are presented in Figure 4
. In the case of the catalyst Acac-200-10, fitting of the Ru3d/C1s spectrum suggests the presence of both a metallic and oxide Ru species with Ru3d5/2
binding energies of 279.5 and 280.4 eV respectively [30
]. These binding energies are similar to those reported by Okal and co-workers for analogous Ru/Al2
catalysts, with the presence of oxidic Ru species being attributed to surface oxidation between preparation and analysis [31
]. The Ru 3d/C1s region also shows that both catalysts contained significant amounts of carbonaceous material, likely a consequence of the preparation method. The XPS derived atomic surface composition, presented in Table 1
. shows that both the most and least active Ru(acac)3
-derived catalysts exhibit similar surface carbon concentrations, 48.22 and 42.06 at% respectively, suggesting that the formation of carbonaceous deposits is not detrimental to catalyst activity. Research by Skrabalak and co-workers found that in the case of polyol prepared Ag nanoparticles, both the precursor salts and resultant nanoparticles were highly active for not only the oxidation of the ethylene glycol solvent, but also the oxidation by-products of nanoparticle formation, such as glycolaldehyde [32
]. Similarly, Feldmann et al. reported that colloidal Zn3
nanoparticles prepared by the polyol method exhibited unusual fluorescence activity, which was determined to be due to the dehydration and carbonization of the solvent yielding 3–5-nm carbon dots in addition to the desired metal phosphate nanoparticles [33
]. It can therefore be suggested that the high carbon content of the catalysts prepared in this work is a result of either complete carbonization of the solvent and resultant oxidation products during catalyst preparation, or alternatively, due to the formation of large oligomeric condensation products as a result of the condensation reaction also between the solvent and associated oxidation products.
The presence of complex carbonaceous species on the surface of the catalyst is consistent with the C1s XPS contributions at 284.7, 286.2 and 288.6 eV, corresponding to aliphatic C, C-O and C = O environments, respectively [34
]. Whilst the aliphatic C and C-O components could be indicative of latent ethylene glycol from the preparation procedure, the presence of C = O contributions and high overall carbon surface concentration is indicative of the formation of carbonaceous deposits resulting from the oxidation of the solvent during catalyst preparation. Similarly, the relative contributions of the three carbon environments and their overall surface concentrations are inconsistent with the previously reported spectrum of solid Ru(acac)3
, supporting the hypothesis that the high surface carbon content is due to the formation of insoluble carbonaceous deposits, rather than an incomplete decomposition of the Ru precursor [30
]. Fitting of the O1s XP spectrum similarly indicates the presence of three oxygen environments at 529.7, 531.8 and 533.32 eV, consistent with Ti-O, C-O and C = O species, providing further evidence for the formation of oxygenated carbonaceous deposits.
Given that the XP spectrum of the Acac-200-10 catalyst shows high bulk carbon concentrations, SEM-EDX was undertaken to determine the dispersion of the carbon layer on the catalytic material, as shown in Figure 5
. Consistent with the TEM presented in Figure 3
, low resolution SEM-EDX mapping of a representative catalyst particle shows that the Ru was well dispersed across the support surface, with a complete absence of larger nanoparticles agglomerates. The EDX C Lα map agreed with the XPS spectra, showing an even distribution of carbon across the surface of the catalyst particle.
A series of further catalysts were prepared using RuCl3
with a range of preparation temperatures and processing times, given the large variation in activity of the Ru(acac)3
catalysts. The LA hydrogenation activity of the catalysts are presented in Figure 6
. Unlike the catalysts prepared using Ru(acac)3
, the most active RuCl3
catalysts were formed at low temperatures and short processing times. Indeed, the catalyst Cl-150-5, prepared at the mildest temperature for the shortest time was the most active catalyst of the series with an LA conversion of 67%. Increasing the microwave irradiation time at 150 °C to 10 or 15 min yielded less active catalysts with LA conversions of 28% and 37% respectively. Likewise, the catalysts prepared at either 175 or 200 °C showed much reduced activity compared to those prepared at 150 °C, with catalyst activity also decreasing with increasing microwave processing time.
The catalysts Cl-150-5 and Cl-150-10 were further studied by TEM, shown in Figure 7
. Similar to the Ru(acac)3
derived catalysts, increasing the processing time resulted in an increase in average particle size from 1.87 to 2.26 nm, suggesting that lower temperatures and short processing times may be employed favorably to yield small well-dispersed supported nanoparticles. In the case of the Cl-150-10 catalyst, the particle size distribution was found to be skewed towards smaller particle diameters (<2 nm), whilst larger (>5 nm) particles, which were not found in other samples, were readily observed for this catalyst. This suggests that the decreased activity of the catalysts prepared with long processing times can be attributed to some extent to sintering of the active Ru nanoparticles.
XPS Ru 3d/C1s and O1s narrow scans of the RuCl3
derived catalysts Cl-150-5 and Cl-150-10 are presented in Figure 8
and surface elemental compositions in Table 2
. Consistent with the Ru(acac)3
derived analogues, the catalysts Cl-150-5 and Cl-150-10 also exhibit very high surface carbon concentrations, suggesting that the formation of insoluble carbon species is independent of the choice of Ru precursor. In addition, peak fitting of the Ru 3d region is consistent with the presence of both Ru(0) and Ru(IV), likely due to the formation of a RuO2
shell in the case of both catalysts.
Given that both the RuCl3
- and Ru(acac)3
-derived catalysts were found to contain significant surface carbon concentrations by both XPS and SEM-EDX analysis, a series of catalysts was further investigated by thermogravimetric analysis (TGA) under oxidising conditions in an attempt to quantify the amount of carbon present. The results are shown in Figure 9
. The mass loss from the catalyst was found to vary between 2.4% and 4.5% for the Acac-200 and Cl-150 catalyst series, and no clear correlation is observed linking catalyst preparation conditions/carbon concentration and catalytic activity. This finding is consistent with the XPS derived atomic surface concentrations presented in Table 2
, which shows the presence of high carbon surface concentrations in every catalyst analysed. The surface area of the catalysts was measured to determine whether the contribution of the carbon over layer had an effect on the surface area of the catalyst. The results (Table 3
) clearly show that no increase in total surface area was observed, and the total surface area of the catalyst remained close to the measured area of the P25 support. Coupled with the SEM results, this suggests that the carbon over layer formed on the surface of the support material was relatively thin. BET surface area measurements (Table 3
) indicate no increase in measures surface area of these catalysts, again indicating that the amount of carbon present is minor compared to the bulk.