Transfer hydrogenation of biomass-derived levulinic acid to γ-valerolactone over supported Ni catalysts

Abstract

A sustainable process of catalytic transfer hydrogenation (CTH) of levulinic acid (LA) to γ-valerolactone (GVL) was investigated over Ni on various supports (Al₂O₃, ZnO, MMT and SiO₂) in the presence of isopropanol (IPA) as the H-donor. Among these, the montmorillonite (MMT) supported Ni catalyst showed almost complete LA conversion (>99%) and selectivity (>99%) to GVL within 1 h. XRD and XPS results showed that the concentration of the metallic species significantly enhanced (two to four times) in the recovered sample as compared to the freshly prepared Ni/MMT. This was due to the in situ reduction of Ni²⁺ species present on the catalyst surface, through liberated H₂ under the reaction conditions. The strong acid strength of MMT, evidenced by NH₃-TPD and py-IR, facilitated the esterification of LA as well as cyclization to GVL. The conversion–selectivity pattern was found to decrease in the IPA–water mixture while, it remained unchanged in the IPA–acetone mixture. Our catalyst could be efficiently recycled up to five times with consistent CTH activity and selectivity to GVL. The plausible mechanism of LA to GVL conversion involves the formation of a levulinate ester with IPA that favours its simultaneous hydrogenation and cyclization in a spontaneous manner to give GVL and regenerating IPA for sustainability.

---

Introduction

The production of γ-valerolactone (GVL) by hydrogenation of levulinic acid (LA) is a vital step in the conversion of biomass to platform chemicals and their subsequent downstream valorization to commodity, fine chemicals, fuel and fuel additives.^1,2 Hydrolysis of lignocellulosic biomass to hexose sugars followed by their selective dehydration and rehydration to levulinic acid demonstrates the viability of chemo catalytic transformations from basic biomass feedstock for the production of chemicals for transportation fuels.^3–6 GVL possesses potential applications as a new generation fuel due to its energy density and allied chemical properties that are comparable to the conventional fuels.^7–11 In addition, GVL is a versatile bio-fuel platform chemical as it can be converted to valeric acid (VA) and subsequently into pentyl-valerate, which is a renewable diesel additive. VA esterification with ethylene glycol, 1,2-propanediol or glycols gives di- and tri-valerates,^12–14 while decarboxylation of GVL to butene followed by oligomerization results in the formation of alkenes (>C8) for fuel application.¹⁵ Thus, the catalytic conversion of LA to GVL is considered a primary process for the production of bio-derived fuels and fuel additives. For this transformation, noble metal based catalysts are extensively explored in the presence of molecular hydrogen; although use of precious metal is unfavorable for the overall process economy.^16,17 On the other hand, use of non-noble metals is restricted due to their extensive leaching in aqueous medium posing a serious stability issue.^18–25 Also, the majority of studies on both the types of catalyst systems for LA hydrogenation are on the use of external H₂ under high pressure conditions. This not only makes the process less sustainable as hydrogen is obtained by reforming of fossil derived from hydrocarbons but also needs special norms for safe handling of H₂ at higher pressure.

In order to overcome these lacunas, catalytic transfer hydrogenation (CTH) has often been attempted using various H-donors like formic acid (FA) and isopropyl alcohol (IPA) for ensuring that hydrogenation becomes an inherently safe and environmentally benign process. Such a strategy is considered to be sustainable as FA is a byproduct formed in equimolar proportion during the decomposition of biomass while IPA is produced from cellobiose via degradation of beta-glucosidase and/or by reduction of bio-derived acetone.^26–28 Although hydrogen generation from FA has been reported, there are several issues like the need for precious metals in the form of soluble complexes (e.g. Pd, Rh), addition of an external H-donor to stimulate the activity of the catalyst, addition of copious amounts of base to improve reduction kinetics and moisture free conditions.^3,12,29,30 On the other hand, IPA being a good solvent and possessing α-hydrogen, exhibits beneficial hydrogen generation capacity under ambient conditions as well as the generated byproduct acetone being easily converted to IPA, thereby maintaining optimal hydrogen concentration during CTH.³¹ Yang et al. studied CTH of ethyl levulinate to GVL over the RANEY® Ni catalyst in the presence of isopropanol at room temperature; however, ethanol as a solvent yielded a byproduct during the cyclization step and therefore required a dedicated separation methodology.³² CTH of ethyl levulinate to GVL has also been reported over zirconium oxide and hydroxide based catalytic systems using ethanol and isopropanol as a medium as well as a H₂ source. However, due to a higher reaction temperature of 250–300 °C as well as appreciable catalyst deactivation and a high catalyst to substrate ratio (1 : 1), this approach may not be practical.^33–38 Recently, zeolite and Zr-based catalysts have been reported for CTH of furfural and levulinate esters to GVL under milder conditions.^39–41 Nevertheless, catalyst stability over multiple reuses would not be possible unless the spent catalyst was pretreated by calcination in some cases.³⁸

We report here, nanosized Ni supported on a MMT catalyst developed for CTH of LA to GVL in the presence of IPA as the H-donor (Scheme 1). The rationale behind such a strategy stems from the fact that hydrogenation of LA can be catalyzed by Ni while the lactonisation shall be favored by an acidic support like MMT. Moreover, a non-aqueous solvent like IPA hinders the formation of 4-hydroxy LA; which is an undesirable reaction associated with an aqueous medium. Ni/MMT employed in the present investigation exhibited >99% conversion with the highest GVL selectivity of >99% and could also be efficiently recycled up to five cycles without sacrificing the conversion and selectivity pattern.

Scheme 1 Catalytic transfer hydrogenation of LA to GVL over Ni/MMT catalyst.

Experimental

Materials

LA (99%) and methyl, ethyl, butyl levulinates were purchased from Sigma-Aldrich, Bangalore, India while isopropanol (>99.9%) was purchased from Rankem, India. Iron(II) sulphate (FeSO₄·7H₂O), nickel(II) sulphate (NiSO₄·6H₂O), sodium borohydride (NaBH₄), and MMT {(Na,Ca)0.33(Al,Mg)₂(Si₄O₁₀)(OH)₂·H₂O} were purchased form Thomas Baker, India while nitrogen (>99.99%) was obtained from Inox, India.

Catalyst preparation

Pretreatment of MMT. 6.0 g of MMT was dispersed in 300 mL 1 N NaCl, stirred for 1 h at 600 rpm and the suspension was left to stand overnight at 250 rpm at room temperature. The slurry was centrifuged at 3000 rpm and the solid was dried in an oven at 105 °C for 4 h. The powder so obtained was ground to 200 mesh size and used for preparation of nanocomposites.

50%-Ni/MMT. To a suspension of 1 g of pretreated MMT in 100 mL of Milli-Q water, an aqueous solution of 0.017 moles NiSO₄·6H₂O in 100 mL of Milli-Q water was added with constant stirring under N₂ atmosphere and the whole mixture was stirred for 30 min. To this mixture, 0.017 moles of NaBH₄ in 100 mL of water was added over a period of 2 h and the reaction mixture was stirred further for 1 h under N₂ atmosphere. This procedure afforded 50%-Ni/MMT and a similar procedure was adopted for other compositions like 10%, 20%, 30% and 40% Ni/MMT with appropriate quantities of NiSO₄·6H₂O and MMT. Similarly, 50% Ni on various supports like Al₂O₃, SiO₂ and ZnO were prepared as per the procedure mentioned above.
Hydrogenation experiments and analysis. LA hydrogenation reactions were carried out in a 300 mL capacity autoclave (Parr Instruments Co., USA) at a stirring speed of 1000 rpm. The typical hydrogenation conditions were: temperature, 200 °C; LA concentration, 5–20 mL; IPA 95 mL; 0.5 g of catalyst loading and N₂ atmosphere. Liquid samples were withdrawn periodically and analyzed by GC (Thermo-Trace-700) having an HP-5 column with a FID detector.

Catalyst characterization

The BET surface area of catalysts was measured by means of N₂ adsorption at 77 K preformed on a Chemisoft TPx (Micromeritics-2720) instrument. X-ray diffraction patterns were recorded on an Analytical PXRD Model X-Pert PRO-1712, using Ni filtered Cu Kα radiation (λ = 0.154 nm) as a source (current intensity, 30 mA; voltage, 40 kV). NH₃-TPD experiments were carried out on a Chemisoft TPx (Micromeritics-2720) instrument. In order to evaluate acidity of the catalysts, ammonia TPD measurements were carried out by: (i) pre-treating the samples from room temperature to 300 °C under a helium flow rate of 25 mL min⁻¹, (ii) adsorption of ammonia at 50 °C and (iii) desorption of ammonia with a heating rate of 10 °C min⁻¹ starting from the adsorption temperature to 973 K. FTIR spectra were recorded on a Perkin-Elmer instrument. The pellets for analysis were prepared by mixing 2 mg of the catalyst with 150 mg of KBr. FTIR spectra were recorded between 400 and 4000 cm⁻¹ with the accumulation of 20 scans and 4 cm⁻¹ resolution. Py-IR spectra were recorded on a Shimadzu FTIR 8000 attached with DRIFT assembly. The sample was placed in the DRIFT cell and heated to 400 °C in the flow of inert gas (N₂) for 1 h. It was cooled to 100 °C and pyridine was adsorbed on the sample. Physisorbed pyridine was removed by flushing the cell with N₂ for 30 min at 100 °C. The temperature programmed desorption of pyridine was studied at different temperature. The spectra were recorded after maintaining the temperature for 30 min. The spectrum of the neat catalyst (before pyridine adsorption) at 100 °C was subtracted from all spectra. The particle size and morphology were studied using transmission electron microscopy (HRTEM), model JEOL 1200 EX. XPS spectra were recorded using ESCA-3000 (VG Scientific Ltd., England) equipped with a CLAM4 analyzer at 1 × 10⁻⁸ Torr vacuum using MgKα radiation. The sample analysis of metal leaching experiments was carried out by using an instrument (ICP-OES), the supernatant liquid was evaporated and resulting solid was treated with aqua regia (HNO₃ : HCl = 1 : 3), 60 °C on a sand bath for 2 h and then made up to 25 mL with distilled water.

Results and discussion

Characterization

XRD patterns of Ni NPs, MMT, fresh and used 50%-Ni/MMT catalyst are shown in Fig. 1. XRD of the fresh catalyst (pure Ni NPs) shows broadening of the Ni peak at 2θ = 45° corresponding to the (111) phase of Ni, indicating smaller size nanoparticles of Ni. The presence of a peak at 2θ = 26.2° for MMT (Fig. 1b) was attributed to the quartz phase of silica which disappeared (Fig. 1c) after insertion of Ni NPs. It implies that Ni NPs were uniformly distributed within the MMT matrix thereby forming the homogenous phase of the Ni/MMT composite. The retention of the broad nature of the peak at 2θ = 44.5° in the 50%-Ni/MMT composite (Fig. 1a and c) suggests that smaller size nanoparticles were well dispersed in the lattice structure of MMT. Meanwhile the recovered sample (Fig. 1d) showed more crystalline phases of metallic Ni, indicating agglomeration of Ni particles after exposure to reaction conditions of high temperature and pressure. The average crystallite size evaluated from the Scherer equation was found to be 30–40 nm.^42,43 This was also separately confirmed by treatment of the catalyst in a flow of H₂ at 200 °C which showed crystalline phases of Ni (Fig. S1†).

Fig. 1 XRD patterns for: (a) Ni NP, (b) Na-MMT, (c) fresh 50%-Ni/MMT and (d) used 50%-Ni/MMT.

The HRTEM image (Fig. 2a) of 50%-Ni/MMT revealed a spherical shape forming a chain of beads of metal nanoparticles⁴⁴ with the particle size in the range of 20–30 nm. As can be seen from Fig. 2b, the used catalyst sample after 5 recycles showed a morphology similar to that of the fresh one with a slight increase in particle size in the range of 20–35 nm. HRTEM images at different higher resolutions (Fig. 3A–D) also confirmed that Ni particles formed a chain of bead like morphology in the MMT matrix. Fig. 3C and D also showed that the Ni particle size (21–34 nm) before and after the reaction matched very well with the XRD results as discussed previously.

Fig. 2 HRTEM images of (a) fresh and (b) used 50%-Ni/MMT.

Fig. 3 HRTEM images of fresh 50%-Ni/MMT at different resolutions.

The XPS spectrum of Ni NPs (Fig. 4a) originating from Ni 2p_3/2 at 852.1, 854.2 and 855.4 eV suggested that Ni was present as Ni⁰ and NiO, respectively. Upon their composite formation with MMT, these peaks shifted by about 1 eV (Fig. 4b) to higher B.E. implying that Ni NPs were strongly bonded with the lattice structure of MMT plausibly through intercalation. The intense doublet nature also reflects the formation of oxide and hydroxide phases, while a small peak at 852.4 eV is attributed to metallic nickel (Ni⁰) on the catalyst surface.

Fig. 4 XPS study of nickel (a) fresh 50%-Ni/MMT, (b) Ni NPs and (c) used 50%-Ni/MMT.

Although used Ni-MMT catalysts showed a similar chemical environment as Ni 2p_3/2 having metallic Ni, NiO and Ni hydroxide (Fig. 4c), the extent of the Ni metal was evaluated as 35%, which was two times more than that of Ni NPs and Ni-MMT fresh samples, respectively, from the XPS results (ESI, Table S1†).

Similarly, the O 1s spectra of Ni NPs show two peaks at 529.1 and 529.9 eV corresponding to oxygen present in an environment of NiO (Fig. S2b†) while 50%-Ni/MMT exhibited (Fig. S2a†) well resolved peaks at 530.1, 532.2 and 533.2 eV due to oxide, surface hydroxyl and adsorbed water, respectively. FT-IR of MMT (Fig. S3†) was helpful in estimating the degree of dissolution of the layered structure. The parent MMT exhibited an intense absorption band at ∼1034 cm⁻¹ for Si–O stretching vibrations of the tetrahedral layer, while the bands at 522 and 460 cm⁻¹ were due to Si–O–Al and Si–O–Si bending vibrations, respectively. Another band near 800 cm⁻¹ was characteristic of amorphous silica which was ascribed to the pretreated MMT as well as the absorption bands at 3633 and 1645 cm⁻¹ were due to stretching and bending vibrations of –OH groups of Al–OH, respectively. The 50%-Ni/MMT sample showed a shift in the band from ∼1039 cm⁻¹ to ∼1003 cm⁻¹, indicating changes in the bonding environment in a tetrahedral layer of MMT. As expected, bare Ni did not show any prominent peaks of IR stretching.^45,46

Catalytic activity of Ni composites

Since the theme of the present investigation was to explore the feasibility of the Ni/MMT catalyst towards CTH of LA to GVL initially, different Ni loadings on MMT were evaluated in order to know the optimum Ni loading that could achieve the highest conversion and selectivity towards GVL (Fig. 5). It was observed that although LA conversion marginally increased from 95 to 99%, selectivity to GVL increased substantially from 61 to 99% with an increase in Ni loading from 10–50%. Enhancement in GVL selectivity can be ascribed to the formation of esterification products at lower metal loading due to less availability of the active Ni sites on the catalyst surface, necessary for the hydrogenation of an intermediate isopropyl levulinate to 4-hydroxy isopropyl levulinate followed by its cyclization to GVL. This aspect was evaluated by performing a control experiment with 10% Ni/MMT for a prolonged reaction time of 5 h (Fig. 6) that resulted in 98% LA conversion. However, GVL selectivity was restricted to 81% which was still lower than that (99%) obtained in a 1 h reaction over 50% Ni/MMT catalyst. It implies that a higher loading of Ni is essential for the availability of requisite active sites responsible for further hydrogenation of intermediate IPA levulinate to GVL at a faster rate.

Fig. 5 Effect of Ni loading on CTH of LA. Reaction conditions: LA (5 mL), IPA (95 mL); temperature, 200 °C; N₂ atm; catalyst, 0.5 g; reaction time, 1 h.

Fig. 6 CT profile for CTH of LA over 10% Ni-MMT catalysts for prolonged reaction time. Reaction conditions: LA (5 mL), IPA (95 mL); temperature, 200 °C; N₂ atm; 10% Ni/MMT catalyst, 0.5 g.

In view of the fact that the choice of a suitable solvent as well as H₂ donor is a key issue in the CTH reaction, various C1–C4 alcohols were screened for the CTH of LA to GVL with 50%-Ni/MMT (Table S2†). As disused earlier, primary alcohols have a slow rate of decomposition to give the H-atom, retarding the transfer hydrogenation rate as compared to that of secondary alcohols. Hence when secondary alcohols were employed as solvents and H-donors, 99% selectivity to GVL was achieved without hampering the conversion of LA. These findings clearly demonstrated that the hydrogen donor played an important role in the release of H-atoms required for the reduction step.^32,38

Furthermore, the effect of LA concentration on conversion and selectivity was performed for the transfer hydrogenation of LA to achieve the maximum productivity of GVL. Fig. 7 shows that the conversion of LA decreased from 99 to 79% with an increase in LA concentration from 5–20% with a concomitant decrease in the selectivity to GVL (99–39%). Such a decrease in conversion and selectivity could be ascribed to (i) the number of available active sites of the catalyst essentially remained the same at both a lower and higher LA concentration and (ii) the accumulation of unconverted IPA-LA and 4-hydroxy IPA-LA that ultimately hinders the cyclization step.

Fig. 7 Effect of substrate concentration on CTH of LA. Reaction conditions: LA (5–20 mL); IPA (95 mL); Ni loading, 50%; temperature, 200 °C; N₂ atm; catalyst, 0.5 g; reaction time, 1 h.

To assess the stability of the catalyst and the optimal concentration of the H-donor required for transfer hydrogenation, LA to GVL conversion was monitored in mixtures of IPA with water and acetone possessing varying compositions. It was observed that the conversion of LA decreased from 99 to 79% (Fig. S4†) with an increase in water content with a simultaneous decrease in selectivity to GVL (99 to 39%) due to deactivation of the metal sites caused by Ni corrosion to form Ni²⁺ species in the presence of excess water.⁴⁷ Nevertheless, esterification seems to be more favored under these conditions implying that a strong aqueous medium may be influencing the cyclization step through ring opening of GVL under acidic conditions.

Regeneration of IPA via in situ catalytic hydrogenation of acetone which was formed as a consequence of IPA oxidation was also monitored by varying the acetone : IPA ratio. It is interesting to note that the conversion–selectivity pattern (Fig. S5 and S6†) remained unchanged with an increase in acetone content under the conditions for complete as well as partial LA conversion. This observation clearly confirmed that co-generated acetone during the LA conversion in IPA has no adverse effect on the activity and selectivity of the catalyst and the H-donor can be recycled for CTH of LA to GVL.

Effect of the catalyst support

LA to GVL conversion predominantly proceeds through two steps: (i) hydrogenation and (ii) cyclization.

The former step essentially depends on the catalyst and its activity while the later one is favoured under strong acidic conditions. Having established the efficacy of Ni NPs towards CTH, it was imperative to evaluate the role of various supports towards effective cyclization and eventually the selectivity to GVL. For this purpose, Al₂O₃, SiO₂ and ZnO were employed with 50% Ni loading; since 50% Ni loading on MMT was found to be optimum (Fig. 5).

It was observed that although MMT exhibited negligible selectivity to GVL (entry 1; Table 1), 58% LA conversion was due to the formation of IPA-levulinate favored by a strong acidic nature of MMT. On the other hand, Ni NPs showed 69% conversion with 52% selectivity to GVL (entry 2; Table 1). Upon composite formation with MMT, Ni NPs quantitatively converted LA to GVL with >99% selectivity to GVL (entry 3; Table 1).

Table 1 Screening of supported Ni catalysts for LA to GVL conversion^a

Entry	Catalyst	Conv. (%)	Selectivity (%)
			GVL	IPA-levulinate	4H-IPA-levulinate
a Reaction conditions: LA (5 mL); IPA (95 mL); Ni loading, 50%; temperature, 200 °C; N2 atm; catalyst, 0.5 g; reaction time, 1 h.
1	Na-MMT	58	2	97	1
2	Ni NPs	69	52	40	8
3	50%-Ni-MMT	99	99	0.01	1
4	50%-Ni/Al2O3	99	90	0.01	10
5	50%-Ni/SiO2	95	40	58	2
6	50%-Ni-ZnO	92	35	64	1

---

As can be seen from Table 2, the active metal leaching was at a maximum (125 ppm) for the Ni-MMT catalyst in a water medium. Even in the IPA medium Ni NPs showed some leaching (9 ppm). However, no metal leaching was observed in the case of the Ni-MMT catalyst in IPA in spite of the metal loading decreasing from 50 to 10% w/w. Although LA conversion was found to be ≥92% for other supports (entries 4–6; Table 1), only the Al₂O₃ supported catalyst exhibited >90% selectivity to GVL while it was very low for ZnO and SiO₂ supports suggesting that MMT was a better dispersion media for Ni NPs. To corroborate these findings, various physico–chemical parameters such as surface area, pore volume and pore diameters were also evaluated using BET surface area measurements as shown in Table 3.

Table 2 Metal leaching studies of bare and supported Ni catalysts^a

Entry	Catalyst	Solvent	Conv. (%)	Metal leaching (ppm)
				Ni
a ND – not detectable, reaction conditions: LA (5 mL); IPA (95 mL); Ni loading, 50%; temperature, 200 °C; N2 atm; catalyst, 0.5 g; reaction time, 1 h.
1	Na-MMT	IPA	58	ND
2	Ni NPs	IPA	69	10
3	50% Ni-MMT	IPA	99	<0.01
4	10% Ni-MMT	IPA	92	<0.01
5	50% Ni-MMT	Water	72	125

---

Table 3 Physicochemical characterization of supported Ni catalysts

Sr. No.	Catalyst	Surface area (m^2 g^−1)	Pore volume (cm^3 g^−1)	Average pore diameter (Å)
1	Na-MMT	98	0.104	51
2	Ni NP’s	12	0.012	11
3	50%-Ni/MMT	34	0.074	43
4	50%-Ni/Al2O3	18	0.059	29
5	50%-Ni/SiO2	60	0.099	34
6	50%-Ni/ZnO	20	0.084	78

---

Fig. 8a–d shows H₂ TPR of the Ni catalyst on various supports studied in this work. All the samples exhibited a broad band of H₂ consumption in the range of 400–700 °C. The shape of the H₂ consumption peak was asymmetric with a shoulder or a tail, as a result of a complex overlapping of several elemental reduction processes such as a sequential reduction from NiO to Ni⁰. The TPR peak of 50% Ni-MMT was observed at the lowest temperature of 534 °C while other catalysts showed TPR peaks in the range of 570–600 °C. The total amount of H₂ required for the complete reduction of 50% Ni-MMT was also much less (6.9 mmol g⁻¹) as compared to the other catalysts. This confirmed the easy reducibility of 50% Ni-MMT which contributed to its greater activity for the catalytic transfer hydrogenation reaction.⁴⁸ The metal dispersion of the prepared catalysts was measured using Micromeritics Chemisorb 2720 instrument, by the nitrogen adsorption/desorption (77 K) and chemisorption of 5% H₂ in Ar, respectively. 50% Ni/MMT showed the highest metal dispersion of 23% as compared to that of other catalysts (Table S3†).

Fig. 8 H₂-TPR of (a) 50%-Ni/Al₂O₃, (b) 50%-Ni/ZnO, (c) 50%-Ni/SiO₂ and (d) 50% Ni/MMT.

In order to understand the role of the support in this CTH reaction, acidity measurements (NH₃-TPD) were carried out on MMT, Al₂O₃ and ZnO supported catalysts. Among these, Ni-MMT showed broad peaks of NH₃ desorption in a high temperature region of 550–600 °C indicating the presence of strong acid sites. Other two catalysts (Ni/ZnO and Ni/Al₂O₃) showed NH₃ desorption peaks in the temperature region from 300–400 °C (Fig. 9). The strength of acid sites of all screened catalysts were determined by NH₃-TPD and the results are shown in Table S4†.

Fig. 9 NH₃-TPD of (a) 50%-Ni/MMT, (b) 50%-Ni/Al₂O₃ and (c) 50%-Ni/ZnO.

The distribution of acid sites was observed in low (50–200 °C), medium (200–400 °C) and high (400–700 °C) temperature regions. It was found that the acid strength of the Ni-MMT catalyst is low at both low and mid temperature regions as compared to Ni/ZnO and Ni–Al₂O₃, respectively. However, no NH₃ desorption from both Ni supported ZnO and Al₂O₃ catalysts was observed. In contrast, Ni-MMT showed strong NH₃ desorption in a high temperature region, which is two fold higher in total acid strength (0.458 μmol g⁻¹) as compared to both the catalyst systems.

The acidic character of the catalyst was further established by Py-IR of composites wherein the MMT supported catalyst showed the presence of peaks at 1539 and 1645 cm⁻¹ (Fig. 10), although with low intensity. Meanwhile, Ni/ZnO and Ni/Al₂O₃ catalysts showed peaks in the region of 1454–1458 cm⁻¹; characteristic of Lewis acid sites. This study suggests that the higher acidic strength and BrØnsted acidic nature of MMT supported Ni played an important role in the release of hydrogen from secondary alcohol by selective β-H elimination as well as esterification and cyclization involved in the LA hydrogenation pathway to form GVL.⁴⁹

Fig. 10 Py-IR of (a) 50%-Ni/MMT, (b) 50%-Ni/ZnO and (c) 50%-Ni/Al₂O₃.

Catalyst reusability studies

Recycling studies of our 50%-Ni/MMT catalyst not only established its excellent stability and efficiency but the characterization of the used catalyst also revealed the role of the active species. After the first run was complete, the catalyst was allowed to settle down and the supernatant reaction crude was removed from the reactor. A fresh charge of reactants was added to the catalyst residue retained in the reactor and the subsequent run was continued and this procedure was followed for subsequent cycles. It was observed that the catalyst could be efficiently reused up to five cycles (Fig. 11) without a substantial decrease in the activity and selectivity under identical experimental conditions. A slight decrease in selectivity from 99 to 95% was possibly due to the slower rate of product desorption as well as due to the presence of a small amount of esters on the active sites during successive cycles. The stability of the Ni-MMT catalyst was also demonstrated by deliberately operating at a conversion level <100% for a lower reaction time of 0.5 h. Here again, the catalyst could be efficiently reused up to five cycles (Fig. S7†). In addition, a standard leaching test involving hot filtration was also carried out. As can be seen from Fig. S8,† the hot reaction crude filtrate after catalyst separation at partial levulinate conversion did not show any activity.

Fig. 11 Reusability studies for CTH of LA to GVL. Reaction conditions: LA (5 mL); IPA (95 mL); Ni loading, 50%; temperature, 200 °C; N₂ atm; catalyst, 0.5 g; reaction time, 1 h.

The catalyst recovered after the 5th cycle of the 100% conversion experiment, was subjected to surface characterization using XRD, TEM and XPS analysis. It was interesting to note that the sharp peaks observed for the recovered catalyst in the XRD spectrum (Fig. 1d) were corresponding to metallic Ni species possessing a fcc structure. Such an alteration for Ni sites could be explained on the basis of two reasons: (i) homogeneously dispersed Ni NPs in the MMT matrix tend to agglomerate under high temperature reaction conditions (200 °C) employed for LA to GVL conversion and (ii) the vertical movement of the bigger sized Ni NPs towards the MMT surface. These facts were confirmed by TEM measurement of the recovered catalyst (Fig. 2b) which revealed the agglomerated phase with increased particle size. To substantiate this finding, XRD analysis was carried out for the 50%-Ni/MMT catalyst obtained by calcination at 400 °C followed by hydrogenation of the Ni²⁺ species adsorbed on MMT (Fig. S1†). It was observed that the peaks corresponding to the Ni site for the Ni-catalyst synthesized via direct hydrogenation (Fig. S1a†) and recovered catalyst (Fig. 1d) exactly matched with each other. Thus during CTH, the evolved hydrogen was partially utilized for the reduction of NiO sites present in the lattice structure of the composite that accounted for increased intensities for the metallic Ni peak.

This was also confirmed by XPS analysis of both the fresh and recovered catalyst. The deconvoluted XPS spectrum (Fig. 4c) of the Ni region suggested that the Ni⁰ content of the reused catalyst increased by about 4 times as compared to the fresh catalyst (Table S1†) with a corresponding decrease in the NiO content. Therefore, it could be deduced that the conversion of the NiO species to Ni⁰ during CTH was beneficial for sustained catalytic activity up to 5 cycles for LA to GVL conversion. This was supported by a control experiment in which the externally hydrogenated Ni-catalyst showed the same conversion–selectivity pattern for LA to GVL conversion (Table S5†). Moreover like the previous catalyst, it also retained its catalytic activity when used after 1st cycle. The XRD pattern of this catalyst exhibited similar characteristic features (Fig. S1b†) as that of the fresh one implying that the presence of Ni⁰ on the MMT surface was indeed responsible for better catalytic activity during successive cycles.

Based on these findings, a plausible reaction pathway (Scheme 1) is proposed involving the following steps: (i) esterification of LA in the presence of isopropanol to give isopropyl levulinate (ii) simultaneous elimination of β-H from IPA to giving 4-hydroxy IPA-LA (iii) its cyclization to GVL with IPA as a byproduct.

Conclusions

The bi-functional Ni-MMT catalyst showed an excellent activity with complete conversion and selectivity (>99%) for CTH of LA to GVL in the presence of IPA as the H-donor. Use of IPA for CTH offered two major advantages: (i) a protecting group for LA thus, preventing metal leaching due to free carboxylic acid, (ii) an efficient hydrogen donor for hydrogenation and forming acetone which could be recycled over the metal surface to regenerate IPA. The highest activity of the Ni/MMT catalyst could be explained by its highest acidity and the presence of both Brønsted/Lewis acid sites as evaluated by NH₃-TPD and py-IR studies, respectively. Such strong acid sites of MMT were responsible for LA esterification as well as for the cyclization step while metallic sites of Ni were involved in selective hydrogenation of the LA ester to GVL. The decrease in Ni loading from 50 to 10% led to a substantial decrease in selectivity to GVL from 99 to 61% due to less available active sites on the catalyst surface for the hydrogenation of an intermediate isopropyl levulinate to 4-hydroxy isopropyl levulinate followed by its cyclization to GVL. Another critical parameter controlling the GVL selectivity was the higher LA concentration. Selectivity to GVL lowered by two and half times (39%) with increasing concentration of LA up to four times because of the limiting number of catalytic active sites available and the accumulation of unconverted IPA-LA and 4-hydroxy IPA-LA. The sustainable activity of Ni/MMT can be vouched from the fact that ∼99% selectivity and conversion could be achieved up to 5 cycles. This feature was attributed to the surface properties of the Ni/MMT composite where Ni⁰ content increased 4 times at the expense of a decrease in the NiO content after the repetitive usage of the catalyst.

Acknowledgements

One of the authors, AMH thanks the Council of Scientific and Industrial Research (CSIR) New Delhi, for the award of a senior research fellowship.

References

1. J. Tukacs, D. Kiraly, A. Stradi, G. Novodarszki, Z. Eke, G. Dibo, T. Keglb and L. Mika, Green Chem., 2012, 14, 2057–2065.
2. T. Pham, D. Shi and D. Resasco, Appl. Catal., B, 2014, 145, 10–23.
3. H. Heeres, R. Handana, D. Chunai, C. Rasrendra, B. Girisuta and H. Heeres, Green Chem., 2009, 11, 1247–1255.
4. Y. Leshkov, J. Chheda and J. Dumesic, Science, 2006, 312, 1933–1937.
5. H. Zhao, J. Holladay, H. Brown and Z. Zhang, Science, 2007, 316, 1597–1600.
6. R. Palkovits, Angew. Chem., 2010, 49, 4336–4338.
7. M. Climent, A. Corma and S. Iborra, Green Chem., 2011, 13, 520–540.
8. I. Horvath, H. Mehdi, V. Fabos, L. Boda and L. Mika, Green Chem., 2008, 10, 238–242.
9. A. Dunlop and J. Madden, US Pat., 2786852, 1957.
10. S. Wettstein, J. Bond, D. Alonso, H. Pham, A. Datye and J. Dumesic, Appl. Catal., B, 2012, 117, 321–329.
11. S. Fitzpatrick, R. Bilski and J. Jarnefeld, Resour., Conserv. Recycl., 2000, 28, 227–239.
12. J. Serrano-Ruiz, D. Braden, R. West and J. Dumesic, Appl. Catal., B, 2010, 100, 184–189.
13. C. Chan-Thaw, M. Marelli, R. Psaro, N. Ravasio and F. Zaccheria, RSC Adv, 2013, 3, 1302–1306.
14. T. Pan, J. Deng, Q. Xu, Y. Xu, Q. Guo and Y. Fu, Green Chem., 2013, 15, 2967–2974.
15. D. Braden, C. Henao, J. Heltzel, C. Maravelias and J. Dumesic, Green Chem., 2011, 13, 1755–1765.
16. D. Alonso, S. Wettstein and J. Dumesic, Green Chem., 2013, 15, 584–595.
17. A. Hengne, N. Biradar and C. Rode, Catal. Lett., 2012, 142, 779–787.
18. L. Manzer, Appl. Catal., A, 2004, 272, 249–256.
19. C. Delhomme, L. Schaper, M. Zhang-Preße, G. Raudaschl-Sieber, D. Weuster-Botz and F. J. Kühn, Organomet. Chem., 2013, 724, 297–299.
20. W. Li, J. Xie, H. Lin and Q. Zhou, Green Chem., 2012, 14, 2388–2390.
21. J. Lange, J. Vestering and R. Haan, Chem. Commun., 2007, 3488–3490.
22. R. Bourne, J. Stevens, J. Ke and M. Poliakoff, Chem. Commun., 2007, 4632–4634.
23. M. Al-Shaal, W. Wright and R. Palkovits, Green Chem., 2012, 14, 1260–1263.
24. P. Upare, J. Lee, Y. Hwang, D. Hwang, J. Lee, S. Halligudi, J. Hwang and J. Chang, ChemSusChem, 2011, 4, 1749–1752.
25. A. Hengne and C. Rode, Green Chem., 2012, 14, 1064–1072.
26. H. Kobayashi, H. Matsuhashi, T. Komanoya, K. Hara and A. Fukuoka, Chem. Commun., 2011, 47, 2366–2368.
27. T. Thananatthanachon and T. Rauchfuss, Angew. Chem., 2010, 122, 6766–6768.
28. J. Jae, W. Zheng, R. Lobo and D. Vlachos, ChemSusChem, 2013, 6, 1158–1162.
29. H. Mehdi, V. Fabos, R. Tuba, A. Bodor, L. Mika and I. Horvath, Top. Catal., 2008, 48, 49–54.
30. D. Kopetzki and M. Antonietti, Green Chem., 2010, 12, 656–660.
31. Y. Soma, K. Inokuma, T. Tanaka, C. Ogino, A. Kondo, M. Okamoto and T. J. Hanai, J. Biosci. Bioeng., 2012, 114, 80–85.
32. Z. Yang, Y. Huang, Q. Guo and Y. Fu, Chem. Commun., 2013, 49, 5328–5330.
33. X. Tang, L. Hu, Y. Sun, G. Zhao, W. Hao and L. Lin, RSC Adv., 2013, 3, 10277–10284.
34. X. Tang, H. Chen, L. Hu, W. Hao, Y. Sun, X. Zenga, L. Lin and S. Liu, Appl. Catal., B, 2014, 147, 827–834.
35. J. Wang, S. Jaenicke and G.-K. Chuah, RSC Adv., 2014, 4, 13481–13489.
36. Y. Kuwahara, W. Kaburagi and T. Fujitania, RSC Adv., 2014, 4, 45848–45855.
37. S. Enumula, V. R. Gurram, M. Kondeboina, D. Burri and S. R. Kamaraju, RSC Adv., 2016, 6, 20230–20239.
38. M. Chia and J. Dumesic, Chem. Commun., 2011, 47, 12233–12235.
39. L. Bui, H. Luo, W. R. Gunther and Y. R. Leshkov, Angew. Chem., Int. Ed., 2013, 52, 8022–8025.
40. H. Y. Luo, D. F. Consoli, W. R. Gunther and Y. R. Leshkov, J. Catal., 2014, 320, 198–207.
41. J. Song, L. Wu, B. Zhou, H. Zhou, H. Fan, Y. Yang, Q. Meng and B. Han, Green Chem., 2015, 17, 1626–1632.
42. L. Wang, A. Lu, C. Wang, X. Zheng, D. Zhao and R. J. Liu, J. Colloid Interface Sci., 2006, 295, 436–439.
43. P. Sharma, S. Bhorodwaj and D. J. Dutta, J. Sci. Conf. Proc., 2009, 1, 40–43.
44. B. Kadu, Y. Sathe, A. Ingle, R. Chikate, K. Patil and C. Rode, Appl. Catal., B, 2011, 104, 407–414.
45. V. Kshirsagar, A. Garade, R. Mane, K. Patil, A. Yamaguchi, M. Shirai and C. Rode, Appl. Catal., A, 2009, 370, 16–23.
46. P. Komadel and J. Madejova, Acid activations of clay minerals in Handbook of Clay Science: Developments in Clay Science, ed. F. Bergaya, B. K. G. Theng and G. Lagaly, Elsevier, Amsterdam, 2006, vol. 1, pp. 263–271.
47. S. Ghosh, M. Mandal, S. Kundu, S. Nath and T. Pal, Appl. Catal., A, 2004, 268, 61–66.
48. R. P. Hernandez, G. M. Galicia, A. A. Maravilla and J. Palacios, Phys. Chem. Chem. Phys., 2013, 15, 12702–12708.
49. C. Reddy, Y. Bhat, G. Nagendrappa and B. JaiPrakash, Catal. Today, 2009, 141, 157–160.

---

Footnote

† Electronic supplementary information (ESI) available: Catalyst characterization (XPS, FTIR, XRD), additional activity (effect of water : IPA and acetone : IPA ratio, catalyst reusability, solvent screening) and metal leaching results are provided. See DOI: 10.1039/c6ra08637c

---