New methods for the preparation of nanoscale nickel phosphide catalysts for heteroatom removal reactions

Abstract

The removal of heteroatom impurities (S, N, O) from fossil fuel and biomass feedstocks is a critical step in the production of clean-burning transportation fuels. Nickel phosphide (Ni₂P) has emerged as a leading candidate material on which to base a new class of catalysts to compete with the long established Ni–Mo and Co–Mo sulfides for hydrotreating of crude oil feedstocks. A recent assessment by researchers at ExxonMobil suggested that a Ni₂P catalyst having a 3 nm particle size could compete with a state-of-the-art Co–Mo/Al₂O₃ catalyst for the hydrodesulfurization (HDS) of a straight-run gas oil feed. This article reviews recent advances in the low-temperature preparation of highly-active, nanoscale Ni₂P catalysts.

---

Mark E. Bussell

Mark Bussell joined the faculty at Western Washington University in 1990 where he is a Professor of Chemistry and Faculty Associate of the Advanced Materials Science and Engineering Center (AMSEC). Bussell received a B.A. in chemistry from Reed College and a Ph.D. in physical chemistry from the University of California, Berkeley. He was a NSF-NATO postdoctoral fellow at the Université Pierre & Marie Curie in 1988–89, and a postdoctoral research associate at the University of Washington in 1990. Bussell has held sabbatical or leave positions at the ETH-Zürich, Tufts University, the University of British Columbia and Yale-NUS College.

---

Introduction

Hydrotreating catalysis

The development of highly-active catalysts for the hydrodesulfurization (HDS) and hydrodenitrogenation (HDN) processes is critical for the production of ultra-low sulfur transportation fuels, and these catalysts may also have utility for the hydrodeoxgenation (HDO) process used in the conversion of bio-oils to renewable biofuels.^1,2 The combination of tightened regulations on sulfur content in transportation fuels and the need to process lower quality fossil fuel feedstocks necessitates the development of highly active hydrotreating catalysts for the removal of sulfur from refractory organosulfur compounds such as 4,6-dimethyldibenzothiophene (4,6-DMDBT), whose structure is shown below. The removal of sulfur from organosulfur compounds via HDS is inhibited by organonitrogen compounds in refinery feeds, necessitating that hydrotreating catalysts have high activity for HDS and HDN. The chemical structure of carbazole, often used as a model compound in HDN studies, is shown in Fig. 1b.

Fig. 1 Chemical structures of (a) 4,6-dimethyldibenzothiophene and (b) carbazole.

Molybdenum sulfide and nickel phosphide hydrotreating catalysts

The current workhorse catalysts for commercial hydrotreating processes are based on molybdenum sulfide (MoS₂), a highly anisotropic material in which molybdenum atoms are sandwiched between layers of sulfur atoms. The sulfur layers separating MoS₂ sheets are only weakly interacting, which is responsible for the lubricating properties of bulk MoS₂. For heterogeneous catalysis applications, such as the heteroatom removal reactions that comprise hydrotreating, MoS₂ is dispersed in the form of nanoscale slabs on a metal oxide support such as gamma-alumina (γ-Al₂O₃), as depicted in Fig. 2a.

Fig. 2 Schematic representations of (a) Ni- or Co–Mo/Al₂O₃ and (b) Ni₂P/SiO₂ catalysts.

The active catalytic sites of MoS₂ crystallites are located on the edge planes where Mo atoms are exposed; a consequence of the layered structure of MoS₂ is a low exposure of active sites. The last twenty years have seen tremendous advances in the development of metal phosphide catalysts, to the extent that nickel phosphide (Ni₂P) is now viewed as a candidate material for a new class of industrial hydrotreating catalysts.^3,4 In contrast to MoS₂, Ni₂P is an isotropic solid that exposes metal atoms on all surface planes (see Fig. 2b), resulting in a higher density of active sites. A recent assessment of unsupported Ni₂P catalysts suggested that nanoscale Ni₂P materials having particles with average diameters of 3 nm would have HDS activities that are competitive with that of a state-of-the-art, cobalt-promoted Mo sulfide catalyst (Co–Mo/Al₂O₃).⁵

Recent advances in the preparation of nanoscale nickel phosphide catalysts

Background

The most widely employed method for preparing Ni₂P catalysts, either in bulk or oxide-supported form, is temperature-programmed reduction (TPR) in flowing hydrogen of phosphate-based precursors prepared from nickel and phosphate salts (e.g. Ni(NO₃)₂ + NH₄H₂PO₄).³ However, high temperatures (773–973 K) are needed to reduce phosphate-based precursors to Ni₂P owing to the strong P–O bonds in PO₄³⁻.⁶ A consequence of carrying out TPR to these high temperatures is that Ni₂P catalysts prepared from phosphate-based precursors have relatively large Ni₂P particles (>10 nm) and wide particle size distributions. High reduction temperatures can also lead to reactions between surface P and aluminum in Al-containing supports (e.g. γ-Al₂O₃, amorphous SiO₂–Al₂O₃ (ASA)) leading to formation of AlPO₄, which results in poor catalytic performance.^7,8 Finally, high synthesis temperatures also preclude the in situ preparation of metal phosphide catalysts as industrial hydrotreating reactors cannot withstand temperatures in excess of ∼673 K. This mini-review highlights some key advances in the low-temperature preparation of small Ni₂P particles (∼3 nm) on metal oxide supports that are in the size range identified as being necessary for HDS and HDN activities competitive with state-of-the-art Co–Mo and Ni–Mo sulfide catalysts.

Low temperature methods for preparing Ni₂P catalysts

To obviate the need of high temperatures (773–973 K) in the reduction of oxidic precursors to Ni₂P catalysts, alternate synthesis methods have been developed that enable reduction at substantially lower temperatures (473–673 K).³ One promising strategy has been to use P sources in the precursors that have lower oxidation states than phosphate, for which the P is in the +5 oxidation state. For example, phosphite (H₂PO₃¹⁻, P oxidation state: +3) and hypophosphite (H₂PO₂¹⁻, P oxidation state: +1), when used in Ni₂P catalyst precursors, permit lower temperature reduction due to weaker P–O bonds.^8–16 Phosphine (PH₃, P oxidation state: −3) and its organophosphorus analogs (e.g. triphenyl phosphine, P(C₆H₅)₃) can also be used for the phosphidation of Ni²⁺ in catalyst precursors, but the toxicity and high cost of these P sources makes their use in large-scale catalyst synthesis impractical.^17–22 Alternatively, low temperature reduction of Ni₂P catalyst precursors has been accomplished by employing a hydrogen plasma or microwave heating.^23–25 While these recently reported preparation methods have achieved successful synthesis of bulk and supported Ni₂P at temperatures considerably lower than those required by conventional TPR of phosphate-based precursors, the resulting Ni₂P particle sizes have often been large (>10 nm). As a result, continued effort has focused on the development of synthesis methods that consistently yield small Ni₂P particle sizes (<5 nm) and which are amenable for industrial hydrotreating applications.

New developments in the synthesis of nanoscale Ni₂P catalysts

Specific challenges that must be overcome for Ni₂P-based catalysts to successfully compete with state-of-the-art Co–Mo and Ni–Mo sulfide catalysts for heteroatom removal reactions include small particles (<5 nm), versatility of oxide supports (e.g. SiO₂, Al₂O₃, ASA), and a reduction protocol amenable for use in industrial reactors. Recently, we reported a new synthesis method involving the preparation of nickel hypophosphite (Ni(H₂PO₂)₂) precursors and their reduction at 473–673 K to yield well dispersed, small Ni₂P particles (3–4 nm) on oxide supports (Fig. 3).⁸

Fig. 3 The synthesis of Ni(H₂PO₂)₂ precursors and their conversion via TPR to oxide-supported Ni₂P catalysts.

Different from earlier reported syntheses that use nickel and hypophosphite salts (e.g. NiCl₂ and NaH₂PO₂),¹² this synthesis takes advantage of the acid–base chemistry between hypophosphorus acid (H₃PO₂) and nickel hydroxide (Ni(OH)₂) to give nickel hypophosphite and water. Of particular note, the synthesis outlined in Fig. 3 reduces the number of synthesis steps necessary to prepare Ni₂P catalysts as the precursors are not calcined and the final step (i.e. reduction of the precursor via TPR) can be carried out directly in the hydrotreating reactor at relatively low temperatures (≤673 K).⁸

The low temperature synthesis of Ni₂P is enabled by the decomposition of the Ni(H₂PO₂)₂ precursor on SiO₂ and Al-containing supports (Al₂O₃, ASA) to give well dispersed, small Ni₂P particles. The reactivity of Ni(H₂PO₂)₂ on the oxide supports was probed with temperature-programmed reduction – mass spectrometry (TPR-MS) and, as shown in Fig. 4a, the reduction of a Ni(H₂PO₂)₂/SiO₂ precursor exhibited TPR-MS traces for PH₃ (m/z = 34) and H₂O (m/z = 18) in the effluent gases that indicate reaction of Ni(H₂PO₂)₂ starting at ∼475 K.

Fig. 4 (a) TPR-MS traces for H₂O (m/z = 18) and PH₃ (m/z = 34) for a Ni(H₂PO₂)₂/SiO₂ precursor. (b) TEM image of a 15 wt% Ni₂P/SiO₂ catalyst prepared from the Ni(H₂PO₂)₂/SiO₂ precursor. The figures are taken with permission from ref. 8 (Copyright 2016 Elsevier).

The temperature at which Ni(H₂PO₂)₂ reacted shifted higher for Al₂O₃ and ASA (507 and 571 K, respectively), indicating stronger interactions between Ni(H₂PO₂)₂ and these supports. Importantly, the reduction of Ni(H₂PO₂)₂ yielded small, uniformly dispersed Ni₂P particles on the SiO₂, Al₂O₃ and ASA supports as confirmed by X-ray diffraction (XRD) and transmission electron microscopy (TEM); the TEM image in Fig. 4b reveals Ni₂P particles of average diameter 3.2 ± 0.4 nm on the SiO₂ support. TEM images of the Ni₂P/Al₂O₃ and Ni₂P/ASA catalysts prepared similarly from Ni(H₂PO₂)₂ showed average Ni₂P particle sizes of 3.9 ± 0.6 nm and 3.3 ± 0.5 nm, respectively, underscoring the utility of this new synthesis method for preparing small Ni₂P particles on a range of oxide supports. Preparation of Ni₂P on the same supports from phosphate-based precursors yielded significantly larger particles as indicated by average Ni₂P crystallite sizes of 12–19 nm. Despite having substantially smaller particles, the Ni₂P/SiO₂ and Ni₂P/Al₂O₃ catalysts prepared from hypophosphite-based precursors had similar or lower CO chemisorption capacities than did their counterparts prepared from phosphate-based precursors. However, as discussed later the catalysts derived from the hypophosphite precursors had similar or higher HDN and HDS activities, as well as significantly higher turnover frequencies (TOFs).

Lan et al. recently reported the results of a head-to-head comparison of the properties of Ni₂P/SiO₂ catalysts prepared from precursors composed of nickel nitrate co-impregnated with either H₃PO₂, H₃PO₃ or (NH₄)₂HPO₄ (as the P source) onto the silica support.¹⁵ TPR-MS measurements of silica impregnated with H₃PO₂ revealed hypophosphite to decompose at 453 K as indicated by PH₃ evolution, while the decomposition of phosphite was observed at a considerably higher temperature as indicated by PH₃ evolution at 673–823 K. Not surprisingly, the production of PH₃ from the hydrogenation of phosphate (HPO₄²⁻) occurred at an even higher temperature of 873 K. Since PH₃ is a strong reducing (and phosphiding) reagent, the lower decomposition temperature of hypophosphite relative to phosphite and phosphate enables Ni₂P formation at lower temperatures. TEM images of the Ni₂P/SiO₂ catalysts prepared using Ni nitrate and H₃PO₂, and reduced at 623, 723 and 823 K, are shown in Fig. 5 along with the corresponding particle size distributions. Following reduction at 623 K, small Ni₂P particles (1.8 ± 0.3 nm) were apparent on the silica support; the particle size increased to 3.4 ± 0.7 nm for the same hypophosphite-based precursor reduced at 823 K. Only broad and very weak XRD peaks were apparent for the Ni₂P/SiO₂ catalysts reduced at 623–823 K, which is not surprising given the small particle sizes, but the peaks are at the expected positions for Ni₂P with no evidence for additional phases being present.

Fig. 5 TEM images and particle size distributions for Ni₂P/SiO₂ catalysts (6.3 wt% Ni) prepared from a Ni(NO₃)₂ + H₃PO₂ precursor reduced at (a) 623 K, (b) 723 K and (c) 823 K. The figures are taken with permission from ref. 15 (Copyright 2017 Elsevier).

Employing a different synthetic strategy, Liu et al. also used hypophosphite as the P source to synthesize Ni₂P/Al₂O₃ catalysts, but relied on vapor transport of PH₃ generated by the decomposition of the hypophosphite to carry out phosphidation of a Ni-impregnated Al₂O₃ precursor.¹³ The synthesis set-up was comprised of two porcelain containers positioned in a tubular reactor with the upstream container loaded with the hypophosphite compound and the downstream container holding the Ni-impregnated Al₂O₃ precursor. Nitrogen gas was flowed through the reactor during the synthesis. The authors explored a number of synthetic parameters, including the Ni salt impregnated onto the γ-Al₂O₃ support, the amount and type of hypophosphite compound loaded upstream and the phosphidation temperature. The optimized synthesis used Ni acetate-impregnated Al₂O₃, H₃PO₂ in a molar ratio of P/Ni = 2–3, and a reaction temperature of 573 K. The freshly prepared Ni₂P/Al₂O₃ catalysts were passivated using a 10 mol% H₂S/H₂ flow prior to removal from the synthesis reactor. As in the study by Liu et al.,¹³ metal phosphide catalysts are most often prepared in a synthesis reactor and then transferred to a different reactor for catalytic studies. To protect the freshly prepared catalysts from deep oxidation, a dilute oxygen flow (e.g. 1 mol% O₂/He) is typically used for passivation of metal phosphide catalysts, resulting in a thin oxide layer at the surface of the phosphide particles. It has been recently shown, however, that passivation with H₂S may be more effective (i.e. higher HDS activity after subsequent reduction).²⁶

Hypophosphite as a reagent for in situ generation of PH₃

The studies discussed above highlight recent advances in the synthesis of oxide-supported Ni₂P catalysts enabled by the use of hypophosphorus acid or hypophosphite salts (e.g. NaH₂PO₂, NH₄H₂PO₂) as the P source. It is instructive, therefore, to discuss the current understanding of the reactivity of hypophosphite and its relevancy to the synthesis of metal phosphide catalysts. While studied by many researchers over the years, the thermal decomposition of NaH₂PO₂ has only recently been clarified by Liu et al.²⁷ Upon heating, NaH₂PO₂ undergoes disproportionation to give reduced (PH₃) and oxidized (P₃O₁₀⁵⁻) P species according to the following overall reaction (eqn (1)).

10NaH₂PO₂ → 4PH₃ + 2Na₅P₃O₁₀ + 4H₂ (1)

The TPR-MS traces in Fig. 6a give insight into the decomposition of NaH₂PO₂ in flowing Ar.⁸ The evolution of PH₃ at 535 K indicates the onset of the disproportionation of NaH₂PO₂, while a significantly smaller PH₃ peak at 579 K and a high temperature shoulder (∼590 K) are assigned to the disproportionation of the intermediate compounds, Na₂H₂P₂O₅ and Na₂HPO₃.²⁷ The results illustrate the usefulness of NaH₂PO₂ and other hypophosphites as a reagent for the in situ production of PH₃, which can then be used directly for the phosphidation of oxide-supported metal precursors. Most importantly, the in situ generation of PH₃via the disproportionation of hypophosphite obviates the need to transport, store and directly handle highly toxic PH₃ for use as a reagent.

Fig. 6 TPR-MS traces for H₂O (m/z = 18) and PH₃ (m/z = 34) for (a) NaH₂PO₂ and (b) Ni(H₂PO₂)₂. The figures are taken with permission from ref. 8 (Copyright 2016 Elsevier).

In using hypophosphite as a reagent for the phosphidation of catalyst precursors with PH₃, one must be mindful of the additional products that result from the disproportionation reaction and their effect on catalyst properties. As indicated in eqn (1), the disproportionation of NaH₂PO₂ leaves sodium tripolyphosphate (Na₂H₂P₂O₅) as a solid residue, but a range of phosphate products are possible depending on the oxide support used and the hypophosphite salt impregnated. For example, reduction of a support impregnated with NiCl₂ and NaH₂PO₂ is expected to yield NaCl and phosphate species as byproducts on the support surface in addition to the desired Ni₂P. Water washing has been employed in some cases to remove solid residues left on Ni₂P/SiO₂ catalysts prepared using Ni salts and NaH₂PO₂.^12,28 By separating hypophosphite and Ni-impregnated Al₂O₃ into separate containers, Liu et al. avoided the issue of solid residue deposition on the catalyst surface during reduction, but additional complexity is introduced into the catalyst synthesis procedure.¹³

By employing Ni(H₂PO₂)₂ as the precursor impregnated onto the support, d'Aquino et al. minimized byproduct residue in the reduced catalyst.⁸ As probed with TPR-MS, unsupported Ni(H₂PO₂)₂ reacts in flowing H₂ to give Ni₂P at 468 K as indicated by the evolution of PH₃ and H₂O in Fig. 6b. XRD analysis of the solid obtained following TPR to 500 K revealed the presence of Ni₂P, but no evidence for crystalline phosphate phases (e.g. Ni₃(PO₄)₂, Ni₂P₂O₇ or Ni(PO₃)₂). Instead, the oxidized product from the disproportionation reaction was identified by IR spectroscopy to be H₃PO₄, indicating that the overall reaction for the decomposition of Ni(H₂PO₂)₂ can be described with eqn (2).

12Ni(H₂PO₂)₂ + 35H₂ → 6Ni₂P + 14PH₃ + 32H₂O + 4H₃PO₄ (2)

The evolution of water above 800 K in the TPR-MS trace is assigned to the hydrogenation of the H₃PO₄ residue. For the oxide-supported Ni₂P catalysts, the H₃PO₄ byproduct from the reaction of Ni(H₂PO₂)₂ likely remains on the catalyst surface to form Brønsted acid sites (P–OH) since the Ni₂P catalysts are prepared at temperatures (≤673 K) below which the phosphate species undergo reduction. High surface densities of Brønsted acid sites on unsupported Co_xNi₂P were shown to correlate with high activity for the HDS of 4,6-dimethyldibenzothiophene.²⁹

Recent advances in hydrotreating over nickel phosphide catalysts prepared using hypophosphite

The development of Ni₂P-based materials as a new class of hydrotreating catalysts has been shown to require Ni₂P particles having diameters of ∼3 nm if they are to be competitive with state-of-the-art Co–Mo and Ni–Mo sulfide catalysts.⁵ One synthetic approach for achieving small Ni₂P particle sizes takes advantage of the low oxidation state of P and weak P–O bonds in hypophosphite to prepare Ni₂P catalysts at relatively low temperatures, as described earlier. d'Aquino et al. prepared Ni₂P on SiO₂, Al₂O₃ and ASA supports having particle sizes of 3–4 nm, while Lan et al. prepared Ni₂P/SiO₂ catalysts having Ni₂P particle sizes smaller than 3 nm.^8,15 Not surprisingly, the catalysts comprised of small Ni₂P particles on oxide supports exhibited high HDS and HDN activities. Fig. 7a compares carbzaole HDN conversions for Ni₂P on SiO₂, Al₂O₃ and ASA supports prepared using either H₃PO₂ or NH₄H₂PO₄ as the P source. The Ni₂P catalysts prepared using H₃PO₂ were reduced via TPR either in situ (“Hypo-in”) or ex situ (“Hypo-ex”) at 673 K, with the ex situ prepared catalysts then re-reduced at 673 K in the reactor to remove the passivation layer. The phosphate-based catalyst was reduced via TPR at 923 K and passivated, and then re-reduced at 673 K in the reactor.

Fig. 7 Carbazole HDN (a) conversions (WHSV = 49 h⁻¹) and (b) turnover frequencies for Ni₂P on SiO₂, Al₂O₃ and ASA. Fig. 7a and b are based on data from ref. 8 (Copyright 2016 Elsevier).

For the Ni₂P/Al₂O₃ and Ni₂P/ASA catalysts, the hypophosphite-based precursors reduced in situ had significantly higher carbazole HDN (and benzothiophene HDS) conversions than either the hypophosphite-based precursors reduced ex situ or the catalysts prepared from phosphate-based precursors. For Ni₂P/SiO₂, the hypophosphite-based catalyst reduced in situ and the phosphate-based catalyst both had high HDN (and HDS) conversions and, along with the in situ prepared Ni₂P/Al₂O₃ catalyst, had significantly higher activities than did a commercially available Ni–Mo/Al₂O₃ catalyst (Shell 424; 4 wt% NiO, 19.5 wt% MoO₃, 8 wt% P₂O₅). Additional evidence for the superior hydrotreating properties of the Ni₂P catalysts prepared in situ comes from the carbazole HDN TOFs that are plotted in Fig. 7b. The Ni₂P/SiO₂ and Ni₂P/Al₂O₃ catalysts prepared in situ from the hypophosphite-based precursors had higher TOFs than the catalysts prepared ex situ from either the hypophosphite- or phosphate-based precursors. The Ni₂P/ASA catalysts were outliers in that the phosphate-based catalyst had the highest carbazole HDN TOF, but this catalyst had an exceptionally low CO chemisorption capacity (19 μmol g⁻¹) compared to the others (52–113 μmol g⁻¹). On a TOF basis, the Ni₂P/SiO₂ and Ni₂P/Al₂O₃ catalysts prepared in situ surpassed the Ni–Mo/Al₂O₃ catalyst for HDN and HDS efficacy.

The HDS results of Lan et al. shown in Fig. 8 highlight the advantage of low temperature reduction of hypophosphite- and phosphite-based Ni₂P/SiO₂ catalysts in comparison to phosphate-based Ni₂P/SiO₂.¹⁵ While the phosphate-based Ni₂P/SiO₂ catalyst exhibited low HDS activity when the precursor was reduced at 723 K or lower, the hypophosphite- and phosphite-based Ni₂P/SiO₂ catalysts had high HDS activities when the precursors were reduced at a temperature as low as 623 K. Of the Ni₂P/SiO₂ catalysts investigated, the catalyst prepared from a hypophosphite-based precursor reduced at 723 K was the most active. Liu et al. compared the dibenzothiophene HDS properties of two Ni₂P/Al₂O₃ catalysts prepared by TPR at 603 K: 1) Ni-impregnated Al₂O₃ and H₃PO₂ in separate containers (with phosphidation by vapor transport of PH₃), and 2) Al₂O₃ co-impregnated with Ni(CH₃COO)₂ and H₃PO₂.¹³ The Ni₂P/Al₂O₃ catalysts prepared by vapor transport of PH₃ were 20–40% more active for dibenzothiophene HDS than the catalyst prepared from the co-impregnated precursor and, interestingly, water washing of the latter Ni₂P/Al₂O₃ catalyst did not improve its HDS properties.

Fig. 8 Thiophene HDS activities at 643 K and 20 h on-stream for Ni₂P/SiO₂ catalysts as a function of the precursor reduction temperature. Fig. 8 is based on data from ref. 15 (Copyright 2017 Elsevier).

Summary and future directions

Recent advances in the synthesis of Ni₂P catalysts prepared from hypophosphite-based precursors and their promising HDN and HDS properties underscore the potential for a new class of hydrotreating catalysts based on nickel phosphide. In addition to enabling the preparation of small Ni₂P particles (<5 nm), hypophosphite-based precursors facilitate preparation of highly active Ni₂P catalysts on Al-containing supports (Al₂O₃ and ASA) as well as in situ reduction in hydrotreating reactors. To develop the potential of Ni₂P catalysts further, it will be critical to better understand the role of Ni₂P particle size on catalyst activity and selectivity for HDN and HDS reactions. The use of solution-prepared Ni₂P nanoparticles of different sizes and narrow dispersity should provide an excellent system for probing these relationships.³⁰ New insights into the nature of the active sites of Ni₂P catalysts and the role of Brønsted acidity in determining catalytic properties will also aid in the design and optimization of the preparation of these catalysts from hypophosphite and other precursors.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This research was supported by the National Science Foundation under grant number CHE-1361702. The author would also like to acknowledge the financial support of Yale-NUS College (Singapore).

References

1. V. C. Srivastava, RSC Adv., 2012, 2, 759–783.
2. T. V. Choudhary and C. B. Phillips, Appl. Catal., A, 2011, 397, 1–12.
3. R. Prins and M. E. Bussell, Catal. Lett., 2012, 142, 1413–1436.
4. S. T. Oyama, T. Gott, H. Zhao and Y.-K. Lee, Catal. Today, 2009, 143, 94–107.
5. S. Soled, S. Miseo, J. Baumgartner, J. Guzman, T. Bolin and R. Meyer, Catal. Today, 2015, 246, 3–8.
6. J. A. Rodriguez, J.-Y. Kim, J. C. Hanson, S. J. Sawhill and M. E. Bussell, J. Phys. Chem. B, 2003, 107, 6276–6285.
7. S. J. Sawhill, K. A. Layman, D. R. Van Wyk, M. H. Engelhard, C. Wang and M. E. Bussell, J. Catal., 2005, 231, 300–313.
8. A. I. d'Aquino, S. J. Danforth, T. R. Clinkingbeard, B. Ilic, L. Pullan, M. A. Reynolds, B. D. Murray and M. E. Bussell, J. Catal., 2016, 335, 2014–2214.
9. J. A. Cecilia, A. Infantes-Molina, E. Rodríguez-Castellón and A. Jiménez-López, J. Catal., 2009, 263, 4–15.
10. J. A. Cecilia, A. Infantes-Molina, E. Rodríguez-Castellón and A. Jiménez-López, J. Phys. Chem. C, 2009, 113, 17032–17044.
11. L. Song, S. Zhang and Q. Wei, Catal. Commun., 2011, 12, 1157–1160.
12. Q. Guan, W. Li, M. Zhang and K. Tao, J. Catal., 2009, 263, 1–3.
13. D. Liu, A. Wang, C. Liu and R. Prins, Catal. Today, 2017, 292, 133–142.
14. G. Shi and J. Shen, J. Mater. Chem., 2009, 19, 2295–2297.
15. X. Lan, E. J. M. Hensen and T. Weber, Catal. Today, 2017, 292, 121–132.
16. J. Li, Y. Chai, B. Liu, Y. Wu, X. Li, Z. Tang, Y. Liu and C. Liu, Appl. Catal., A, 2014, 469, 434–441.
17. S. Yang, C. Liang and R. Prins, J. Catal., 2006, 237, 118–130.
18. S. Yang and R. Prins, Chem. Commun., 2005, 4178–4180.
19. Z. Wang, L. Zhou, M. Zhang, W. Li and K. Tao, Chem. – Asian J., 2009, 4, 1794–1797.
20. H. Song, M. Dai, H. Song, X. Wan, X. Xu, C. Zhang and H. Wang, Catal. Commun., 2014, 43, 151–154.
21. J. Wang, H. Chen, Y. Fu and J. Shen, Appl. Catal., B, 2014, 160–161, 344–355.
22. J. Chen, M. Han, S. Zhao, Z. Pan and Z. Zhang, Catal. Sci. Technol., 2016, 6, 3938–3949.
23. A. Wang, M. Qin, J. Guan, L. Wang, H. Guo, X. Li, Y. Wang, R. Prins and Y. Hu, Angew. Chem., Int. Ed., 2008, 47, 6052–6054.
24. Y. Xue, Q. Guan and W. Li, RSC Adv., 2015, 5, 53623–53628.
25. L. Ding, M. Zheng, A. Wang and T. Zhang, Catal. Lett., 2010, 135, 305–311.
26. X. Li, Z. Sun, A. Wang, Y. Wang, X. Duan and Y. Chen, Catal. Commun., 2014, 43, 21–24.
27. D. Liu, X. Li, W. Ling, T. Zhang, A. Wang, C. Liu and R. Prins, Dalton Trans., 2017, 46, 6366–6378.
28. Q. Guan and W. Li, J. Catal., 2010, 271, 413–415.
29. I. I. Abu and K. J. Smith, J. Catal., 2006, 241, 356–366.
30. G. H. L. Savithra, R. H. Bowker, B. A. Carrillo, M. E. Bussell and S. L. Brock, Chem. Mater., 2013, 25, 825–833.

---