Yulong
Deng
abc,
Binyu
Zhang
abc,
Huiru
Wu
abc,
Zhuo
He
abc,
Xiaorui
Du
abc,
Jiayi
Ou
abc,
Tianyu
Ren
bc,
Haiyong
Wang
abc,
Yuhe
Liao
abc,
Qiying
Liu
*d,
Chenguang
Wang
*abc and
Yanbin
Cui
*abc
aSchool of Energy Science and Engineering, University of Science and Technology of China, Hefei 230026, PR China
bCAS Key Laboratory of Renewable Energy, Guangzhou Institute of Energy Conversion, Chinese Academy of Sciences, Guangzhou 510640, PR China. E-mail: wangcg@ms.giec.ac.cn; cuiyb@ms.giec.ac.cn
cGuangdong Provincial Key Laboratory of Renewable Energy, Guangzhou 510640, P.R. China
dJiangsu Co-Innovation Centre of Efficient Processing and Utilization of Forest Resources, International Innovation Centre for Forest Chemicals and Materials, Nanjing Forestry University, Nanjing 210037, PR China. E-mail: liuqy@njfu.edu.cn
First published on 2nd December 2024
Ru/Nb2O5 is effective for furfural aqueous reductive conversion. Systematic characterization, kinetic studies and in situ DRIFT tests demonstrated that the Ruδ+ species abundance is highly correlated with the adsorption behavior of the substrate, key intermediate and product (furfural, 2-cyclopentenone, and cyclopentanone), which governs the FFR conversion rate and cyclopentanoid product selectivity.
The reaction proceeds through a series of CO hydrogenation, ring-opening, isomerization, dehydration and C
C hydrogenation steps. Therefore, to achieve high selectivity for cyclopentanoid products, it is essential to develop a catalyst that preferentially hydrogenates the C
O bond over the C
C bond and excels in acid-catalysed ring rearrangement. Ru as an oxophilic metal with excellent H2 dissociation and deoxygenation activity is one of the most widely applied metal catalysts for aqueous reductive conversion of biomass-derived carbonyl compounds.7–9 Support matrices such as metal oxides can act as promoters by anchoring metal species, introducing defects, interacting with supported metal and stabilizing particular oxidation states of the active metal.10 Nb2O5 is widely applied as a stable support in catalytic processes involving the synergy of Brønsted and Lewis acid sites, making Ru/Nb2O5 a versatile catalyst with outstanding performance for hydrodeoxygenation (HDO) of a variety of substrates, such as phenols,11 alkenes,12 ethers,13 and particularly unsaturated aldehydes.14
The electronic structure is a key factor determining the intrinsic activity of the catalyst because it affects the stability of the transition state during the reaction and determines the interaction strength between the catalyst and reactants. Numerous studies reported that electron-deficient Mδ+ species on the catalyst surface is responsible for the high selectivity toward FAOL conversion of furfural.15,16 The Nb2O5 support could boost the HDO performance of Ru by forming a Ru–RuOx–Nb2O5 interface through tunable electron transfer between the support and metal,17,18 which is critical in manipulating substrate adsorption behaviour. Moreover, the desorption behaviour of CPO can also be tuned on electron-deficient metal species derived from heteroatom doping, which further governs the distribution of over-hydrogenation products.15 However, the impact of the electronic structure of various Ru species on cyclopentanoid compounds from FFR aqueous reductive conversion has been elusive, and the correlative analysis between the abundance of reactive Ruδ+ species and catalytic performance is scarcely investigated.
In this contribution, a series of Ru/Nb2O5 catalysts with varied reduction temperatures were employed as model catalysts to investigate the metal–support interaction in FFR reductive conversion. Systematic characterization was applied to investigate the physicochemical properties of the as-prepared catalysts, and the abundance of partially oxidized Ruδ+ species was analysed by XPS and H2-TPR on activated catalysts. The kinetic study profiled the transition from the substrate to key intermediates, and to final products, providing a comprehensive perspective to investigate the relationship between the Ruδ+ species and catalytic performance. The role of Ruδ+ species in FFR conversion and product selectivity was further elucidated by interpreting the adsorptive behaviour of FFR, furfural alcohol (FAOL), 2-cyclopentenone (CPEO), and CPO using in situ DRIFTS.
Entry | Catalyst | Conversion of FFR (%) | Yield (C-mol%) | Carbon balance of the product (C-mol%) | |||
---|---|---|---|---|---|---|---|
THFAOL | FAOL | CPO | CPOL | ||||
Catalyst preparation conditions: calcined at 500 °C and reduced at 450 °C. Reaction conditions: 0.20 g furfural, 0.04 g catalyst, 20 mL of H2O, p (initial, H2) = 4.0 MPa, reaction temperature is 160 °C, heating rate = 5 °C min−1, stirring speed = 800 rpm, reaction time = 6 h. | |||||||
1 | 5% Ru/MoO3 | >99.9 | 10.0 | — | 0.2 | 60.5 | 70.7 |
2 | 5% Ru/Al2O3 | >99.9 | 31.6 | — | 0.8 | 42.5 | 74.9 |
3 | 5% Ru/ZrO2 | >99.9 | — | 27.6 | 0.2 | 43.4 | 71.2 |
4 | 5% Ru/TiO2 | 25.9 | — | 2.0 | 23.7 | — | 25.7 |
5 | 5% Ru/Nb2O5 | >99.9 | — | 2.8 | 21.1 | 66.3 | 90.2 |
6 | Pristine Nb2O5 | 1.9 | — | — | — | — | — |
7 | Nb2O5-H(450) | 3.1 | — | — | — | — | — |
Due to its satisfactory catalytic performance, Ru/Nb2O5 was selected as the model catalyst for further investigation via material manipulation techniques. The reaction temperature and pressure were optimized for Ru/Nb2O5 catalysed FFR conversion (Fig. S1†). At low reaction temperatures, considerable FAOL accumulation was observed due to the low efficiency in initiating the acid-driven ring rearrangement.21 However, carbon balance dropped sharply at elevated reaction temperatures, associated with extensive self-polymerization of FFR and formation of humins under such conditions.22,23 Low initial hydrogen pressure was shown to be ineffective in hydrogenating the key intermediates, leaving considerable amounts of FAOL and CPEO unconverted, while over hydrogenation of the furan ring occurred and generated the THFAOL by-product when the hydrogen pressure was beyond 5 MPa. Overall, 4 MPa and 160 °C were selected due to the satisfactory conversion, carbon balance and cyclopentanoid product selectivity.
No characteristic XRD patterns of Ru or RuO2 phases are detected for all catalysts, indicating that Ru particles are highly dispersed on the support with their sizes below the limit of phase detection (Fig. 1a). The sharp peaks for all samples indicate the high degree of crystallinity of the Nb2O5 support. Catalysts reduced at temperatures below 650 °C share multiple sharp diffraction peaks of different planes of Nb2O5 (PDF#72-1484) and Nb16.8O42 (PDF#71-0336). In Ru/Nb2O5-H(850), the XRD pattern exhibits new diffraction peaks at 26.0°, 35.2°, and 52.1°, corresponding to the presence of a newly formed NbO2 phase (PDF#71-0020). The BET test indicates that the crystallized support showed a very limited porous structure (Table S2†). In Fig. 1b, the Raman spectra show a prominent vibrational peak centred at 130 cm−1, attributed to the stretching vibration between the Nb–Nb bond.24,25 The Raman vibrational peak at 267 cm−1 and 627 cm−1 is assigned to the bending vibration and bridging vibration mode of the Nb–O–Nb bond, respectively.26,27 The vibrational peak at ∼997 cm−1 is attributed to the symmetric stretching vibration mode of the NbO bond between edge-sharing NbO6 octahedra, which contribute to Lewis acidic sites (LASs) on the surface.28,29Fig. 1c displays the Py-FTIR spectra of Ru/Nb2O5 catalysts after desorption at 150 °C, showing characteristic bands for Lewis acidic sites (1450 cm−1, 1610 cm−1) and minor bands (1490 cm−1) for both Brønsted and Lewis acidic sites.30 For all tested Ru/Nb2O5 catalysts, LASs present the dominant acid species and their density remains essentially consistent. The LA abundant sites in Nb2O5 promote cyclopentanoid selectivity by suppressing hydrogenation to THFAOL, as reported in previous research.31 H2-TPR shows a dominant H2 uptake peak (β peak) at 193 °C, and the reduction peak slightly shifts to a higher temperature as the metal loading increases, which is associated with the broader size distribution. The γ reduction peak centred at 814 °C is attributed to the removal of surface oxygen (Fig. 1d).32,33
The surface electronic structure/chemical states of Ru and Nb were determined by X-ray photoelectron spectroscopy (XPS). In Fig. 1e, the deconvoluted peaks at ∼280.0 eV and ∼280.8 eV are assigned to the Ru0 3d5/2 state and Ruδ+ 3d5/2 state, respectively.34 With the increase of reduction temperature, the proportion of Ru0 in total ruthenium species increases along with a shift to a lower binding energy, which is associated with intensified electron transfer from the support to metallic Ru sites.35,36 Detailed Nb 3d spectra of these Ru/Nb2O5 catalysts present a broad Nb 3d5/2 peak between 205 and 209 eV (Fig. 1f), which can be deconvoluted to two peaks at ∼207.1 and ∼207.4 eV, respectively, corresponding to Nb4+ and Nb5+ states.37 The ratio of the Nb4+ peak increases gradually with reduction temperature. As a result of the intensified electron transfer effect, the Ruδ+/(Ru0 + Ruδ+) ratio decreased from 0.41 to 0.08, with the reduction temperature increasing from 250 °C to 850 °C (Table S3†).
The apparent reaction rate constants for FFR conversion of various Ru/Nb2O5 catalysts were calculated using the pseudo-first-order model (Table S4†), assuming that hydrogen is in significant excess.36,37 The calculated reaction rate constant decreases as the reduction temperature of Ru/Nb2O5 catalysts increases (Fig. 3a). Considering the similar properties of the Ru/Nb2O5 catalysts, such as porosity, surface acidity, geometric features and dispersion, we infer that the main factor attributed to the different catalytic performances is the electronic structures of reactive Ru species. XPS results showed that electron transfer from the Nb2O5 carrier to the Ru metal active site was promoted, leading to a decreased ratio of Ruδ+ among Ru species. It has been extensively reported that electron-deficient metal sites are critical for furfural adsorption, and partially oxidized Ruδ+ sites were also reported as reactive sites dominating furfural conversion.15,38
To quantitatively interpret the role between active sites and catalytic activity, the activated catalysts were subjected to H2-TPR analysis, and hydrogen uptake was used to determine the abundance of reactive Ruδ+ species in Ru/Nb2O5 catalysts (Fig. 3b, Table S5†). No β reduction peak was found in the activated Ru/Nb2O5 catalysts, indicating that the dominant species in these reduced catalysts is metallic Ru. By contrast, a distinct α reduction peak can be observed at around 100 °C for all reduced Ru/Nb2O5 catalysts, which is attributed to the reduction of the reactive Ruδ+ species on the surface that can be easily oxidized upon air exposure.39 The hydrogen consumption of the α reduction peak decreased gradually with the increase of reduction temperature, indicating that the abundance of Ruδ+ species on the catalyst surface was negatively affected by reduction temperature.
This is consistent with the observed trend presented by XPS analysis. Notably, a good correlation between reactive Ruδ+ species on the surface of the Ru/Nb2O5 series catalysts and the apparent reaction rate constant k for FFR conversion was found, with the corresponding fitting curve's correlation coefficient R2 = 0.975 (Fig. 3c). The H2-TPR hydrogen consumption increases as the reduction temperature of the catalyst decreases, and the conversion rate of FFR by the catalyst increases accordingly.
![]() | ||
Fig. 4 Time-resolved in situ DRIFT spectra for FFR, FAOL, CPEO, and CPO adsorption over various 1% Ru/Nb2O5 catalysts reduced at varied temperatures. Note: FAOL was subjected to programmed heating to 200 °C, for detailed conditions see Fig. S5–S8.† |
Notably, the retention time of the CO signal upon heating treatment decreases from 13 min (Ru/Nb2O5-H(250)) to 3.5 min (Ru/Nb2O5-H(850)), indicating weaker FFR adsorption with increasing reduction temperature due to decreasing active Ruδ+ sites. Also, the intensity of the characteristic C
O signal after consistent heating treatment displayed a decreasing trend as the catalyst reduction temperature increases (Table S6†). In fact, a good linear fitting relationship was found between the H2-TPR uptake and the adsorption intensity of the C
O group of FFR on the catalyst surface (Fig. 3d). These results indicate that the presence of active Ruδ+ sites is highly related to the substrate adsorption orientation and adsorption strength. Kim et al. reported that the oxygen atoms of the C
O groups of these carbonyl containing compounds preferentially adsorbed at the coordinatively unsaturated Ruδ+ sites in an η1(O)-aldehyde configuration,40 which is consistent with the result presented in our study. In comparison, the adsorption behaviour of FAOL onto different Ru/Nb2O5 catalysts showed no obvious difference. A strong signal assigned to –OH group adsorption centred at ∼1000 cm−1 was detected, corresponding to the adsorption configuration of η1(O)-alcohol hydroxyl of FAOL (Fig. S6†).16 It has been reported that the main sites for the adsorption of η1(O)-alcohol hydroxyl groups in FAOL are Lewis acid sites.24,45 This result is consistent with the Py-FTIR result that showed similar abundance and distribution of LA sites among the catalysts.
Similar adsorption tests were conducted for other carbonyl containing compounds CPEO and CPO (Fig. S7 and S8†). Both molecules show a band at around 1750 cm−1 attributed to ν(CO),46 and the peak shows a red-shift (30 cm−1) with significant improvement of relative intensity, indicating chemically adsorbed carbonyl groups onto Ruδ+ sites. As the catalyst reduction temperature increases, the C
O signal intensity follows the same decreasing trend as that of FFR, correlating with the H2-TPR uptake well (Fig. 3d). The decreasing CPO adsorption strength reveals that high CPO selectivity can be obtained by avoiding over-hydrogenation using a catalyst produced at elevated reduction temperature. The adsorptive behaviour of these different types of carbonyl containing intermediates and the target product affords a panoramic vision regarding FFR conversion and final product selectivity in aqueous reductive conversion processes.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4cy01272k |
This journal is © The Royal Society of Chemistry 2025 |