An investigation on the influence of support type for Ni catalysed vapour phase hydrogenation of aqueous levulinic acid to γ-valerolactone

Abstract

Ni (20 wt%) supported on SiO₂, γ-Al₂O₃ and ZrO₂ catalysts was examined for hydrogenation of aqueous levulinic acid (LA) to γ-valerolactone (GVL) at 270 °C and ambient pressure. The band intensities of Brønsted (BAS: 1540 cm⁻¹) and Lewis acid sites (LAS: 1450 cm⁻¹) estimated by pyridine adsorbed DRIFT spectra revealed a lower ratio of BAS/LAS over the Ni/SiO₂ catalyst than over the Ni/ZrO₂ and Ni/γ-Al₂O₃ catalysts. The rate of angelica lactone (AL) formation was lower than the rate of AL hydrogenation over the Ni/SiO₂ catalyst. The poisoning and regeneration of the Ni/SiO₂ catalyst using pyridine and 2,6-dimethylpyridine demonstrated that Lewis acid sites influenced the conversion of LA to AL and subsequent hydrogenation of AL to GVL occurred on surface Ni sites. In contrast, Brønsted acid sites were responsible for the ring opening of GVL to valeric acid (VA). Kinetic data emphasized that the hydrogenation activity and product distribution were dependent on the type of acid site, and the Ni sites in close proximity to Brønsted acid sites are prone to hydrogenolysis of GVL to valeric acid and hydrocarbons.

---

1. Introduction

The conversion of levulinic acid (LA), a compound that is obtained from inedible, woody biomass feedstock, namely lignocellulose, to fuels and fine chemicals is a topic of increasing interest.^1,2 In order to achieve the conversion of LA to fuels, the first step required is to convert LA to γ-valerolactone (GVL). This can be done via catalytic dehydration of LA to angelica lactone (AL) followed by subsequent catalytic hydrogenation of AL, giving rise to GVL. The GVL produced can then be converted to a range of valuable compounds (Scheme 1).

Scheme 1 Possible products in the catalytic hydrogenation of levulinic acid.

Numerous studies have been conducted on the catalytic hydrogenation of LA to GVL and also the catalytic conversion of GVL to valuable products.^3–7 The majority of the aforementioned studies have been reported in the liquid phase with very few studies being examined in the vapour phase.^8–14 Zhang et al. reported the selective hydrogenation of LA to GVL in methanol over magnetic Ni_4.59Cu₁Mg_1.58Al_1.96Fe_0.70 catalyst with a GVL yield of 98.1% at 142 °C.¹⁵ Yan et al. reported a 91% GVL yield at 200 °C and 70 bar H₂ pressure over Cu-based catalysts derived from hydrotalcite precursors during the LA hydrogenation.¹⁶

The vapour phase conversion of LA to GVL does however have a number of advantages with respect to the liquid phase process, such as having no requirement for high pressure conditions, purification, space-to-time yield productivity, environmental impact and generally a reduced likelihood of catalyst deactivation, as well as additional benefits in terms of efficiency, safety and waste emission.¹⁷ Dumesic et al. investigated the vapour phase conversion of LA to GVL using a carbon supported bimetallic Ru–Sn catalyst with a turnover rate of 0.051 s⁻¹ using 2-sec-butyl-phenol as the solvent at 180 °C and 35 bar H₂ pressure.¹⁸ Upare et al. studied vapour phase LA hydrogenation at ambient pressure and 265 °C over a commercial Ru/C catalyst and achieved 100% LA conversion with 100% GVL selectivity.¹⁹ Earlier, we reported LA hydrogenation over hydroxyapatite (HAP) supported metal (Ru, Pt, Pd, Ni, Cu) catalysts in the vapour phase at ambient H₂ pressure and 275 °C with 92% conversion and 99.8% GVL selectivity over the Ru/HAP catalyst.²⁰

Although the studies on Ru based catalysts have led to promising results, there is scope for development of cheaper catalysts, such as those based on transition metals. Upare et al. investigated the conversion of aqueous LA to GVL using a Ni promoted copper–silica nano composite catalyst in the presence of formic acid.²¹ Mohan et al. examined the Ni/H-ZSM-5 catalyst for LA to GVL at 250 °C and 1 bar H₂ pressure with 100% LA conversion and 92.2% selectivity to GVL.²² The mechanism of hydrogenation of LA to GVL occurs via either dehydration to angelica lactone followed by its reduction or via reduction of LA to 4-hydroxyvaleric acid and subsequent dehydration. For both of these routes, an active metal combined with an acid site should be present on the catalyst surface. It has been reported that Brønsted acid sites are more prone to ring opening of GVL to valeric acid (VA: another fuel additive) and hydrocarbons.^23,24

However, studies pertaining to the support effect for the vapour phase hydrogenation of LA to GVL have not been investigated in detail except for recent interesting reports from the Weckhuysen group over supported Ru catalysts at high pressures.^25,26 Particularly, the influence of acid–base characteristics of the various conventional and non-conventional supports, such as γ-Al₂O₃, SiO₂, ZrO₂, TiO₂, MgO, ZnO, La₂O₃, hydroxyapatite and carbon, as supports for either noble metals or base metals have not been explored for the vapour phase conversion of LA to GVL. In our continued activities on the selective hydrogenation of LA to GVL over Ru based catalysts, we have examined several support materials for Ru; for brevity their catalytic activities are given in Table S1 (ESI†).^10,20 Hydrogenation of levulinic acid was also carried out at high H₂ pressures over various noble and non-noble metal catalysts.^27–30 It has been reported that the nature and type of (Brønsted and/or Lewis) acid site would influence the product distribution during the hydrogenation of levulinic acid.²⁴

The present investigation is focussed on the characteristics of Ni supported on γ-Al₂O₃, SiO₂, and ZrO₂ catalysts; the role of the metal site and/or the acid sites would influence the product distribution. The catalysts were characterized by DRIFT analysis and the role of Brønsted and Lewis acid sites is rationalized by poisoning of the catalysts using pyridine and 2,6-dimethylpyridine during the course of the reaction. The catalytic activities were evaluated strictly under a kinetic regime. Some of the samples were examined by BET-SA, TEM, and H₂-TPR, and the used catalysts were analysed by CHNS to investigate potential coking.

2. Experimental

2.1 Preparation of catalysts

The SiO₂ (BET-SA: 395.0 m² g⁻¹), γ-Al₂O₃ (BET-SA: 230.5 m² g⁻¹) ZrO₂ (BET-SA: 52.8 m² g⁻¹) Ni(NO₃)₂·6H₂O, levulinic acid (98%), γ-valerolactone (99%), angelica lactone (98%), valeric acid, and methyl tetrahydrofuran (MTHF) were purchased from Sigma-Aldrich and used as received. The α-Al₂O₃ was received from NORPRO-R97655. All the catalyst samples were prepared by incipient wet impregnation method. Briefly, the required amount of solid support (e.g. SiO₂, γ-Al₂O₃, ZrO₂), and in some cases α-Al₂O₃, was added to an aqueous solution of Ni(NO₃)₂·6H₂O (an amount required to obtain a 20 wt% Ni loading) and the suspension was then stirred at 100 °C until the water had evaporated. The recovered solid was then oven dried at 120 °C overnight and calcined in static air at 500 °C for 5 h.

2.2 Characterization of the catalysts

The powder XRD analyses of the catalysts were recorded with an Ultima-IV X-ray diffractometer (M/s. Rigaku Corporation, Japan) using a Ni-filtered Cu Kα (λ = 0.15418 nm) radiation source and a scintillation counter detector. The diffraction patterns were recorded with a scan rate of 5° min⁻¹ in the 2θ range of 10–70° at 40 kV and 20 mA. The average crystallite size (D) of the catalysts was determined by applying the Debye–Scherrer equation with respect to the Ni(111) plane. The surface areas of the reduced samples were measured by N₂ adsorption at −196 °C (Micromeritics ASAP 2010 surface area analyzer). The nature of the acid sites of the catalysts was examined by pyridine adsorbed FT-IR spectroscopy (Carry 660, Agilent Technologies). Spectra were obtained in the range of 1400–1700 cm⁻¹ with a resolution of 2 cm⁻¹ with 64 scans for each spectra collection. The experiments were performed in situ using a purpose-made IR cell connected to a vacuum adsorption setup. In a typical method, the reduced samples were pressed into self-supporting wafers (density ∼ 40 mg cm⁻³) at a pressure of 10⁵ Pa. Subsequently, the wafers were transferred into the IR cell and were pre-treated in N₂ flow by heating at a rate of 10 °C min⁻¹ up to 400 °C for 1 h. After cooling down to 150 °C, the spectrum was collected in the drift mode. The sample was then exposed to pyridine until surface saturation in successive pulse injections at 150 °C and subsequently the sample was purged for 30 min in N₂ flow before recording the spectrum. The drift spectra after pyridine adsorption were subtracted from the spectra of the untreated catalyst to obtain the vibrational bands owing to pyridine acid site interaction. Finally, the spectra were quantified with the Kubelka–Munk (K–M) function and the fitted curves were used to measure the relative ratios of Brønsted (BAS) and Lewis acid sites (LAS) for the corresponding spectral lines at full width at half maximum (FWHM). The H₂-TPR analysis was carried out in a quartz micro-reactor interfaced to a gas chromatograph (GC) equipped with a thermal conductivity detector (TCD) unit. Prior to TPR analysis, the catalyst was degassed at 300 °C in helium gas for 30 min and then cooled to room temperature. The helium gas was switched to 4.97% H₂ in argon with a flow rate of 30 mL min⁻¹ and the temperature was increased to 900 °C at a ramping rate of 5 °C min⁻¹. The hydrogen uptakes of the samples were measured using a calibration curve of Ag₂O TPR under a similar protocol. Calibrated mass flow controllers (Alicat Scientific, USA) were used to regulate the flow rates for all the gases used. The carbon contents in the used samples were measured using a VARIO EL, CHNS analyzer. The elemental analysis of the fresh and used samples was carried out by atomic absorption spectroscopy (AAS) PerkinElmer, Analyst-300. The AAS analysis of the fresh and used samples indicated no leaching of metal during the course of the reaction (Table S2†). The CO pulse chemisorption experiments were carried out using a pulse titration procedure at 40 °C on an AUTOSORB-iQ automated gas sorption analyzer (M/s. Quantachrome Instruments, USA). In a typical method, the sample was reductively pre-treated at 450 °C for 2 h in 4.97% H₂, balance Ar, then the sample was flushed in helium gas for 30 min at 450 °C followed by titration with 5.02% CO, balance He, at 40 °C. The Ni dispersion, Ni metal surface area and Ni particle size were calculated using the following equations:

Metal area = metal cross sectional area × no. of metal atoms on surface (i.e. CO uptake)

2.3 Activity measurements

The vapour phase hydrogenation of aqueous levulinic acid was carried out in a fixed bed quartz reactor (i.d. = 10 mm, length = 420 mm) in down flow mode. The experimental conditions and product analysis details were similar to those reported by us earlier.²⁰ The carbon mass balance in all the experiments was >99% unless otherwise stated. The influence of both Brønsted and Lewis acid sites in the LA conversion was examined by carrying out two independent experiments: (1) by using 2,6-dimethylpyridine (2,6-lutidine) as a selective Brønsted acid site blocker and (2) pyridine (Py) as both a Brønsted and Lewis acid site blocker.

In a typical method, about 12.4 mmol of probe was injected successively in 4 pulses (3.1 mmol each) into the aqueous LA stream. After each pulse, the samples were collected and analyzed by GC-MS. After the dosage, the catalyst was regenerated at 450 °C in flowing air and the catalysts were subsequently reduced at 450 °C with 4.97% H₂, balance Ar, before the aqueous LA was subjected to the catalyst. The conversion, selectivity, rates and turnover frequency (TOF) on product formation were calculated using the equations given below:

3. Results

3.1 X-ray diffraction and BET surface area analysis

The XRD patterns of Ni supported on SiO₂, γ-Al₂O₃, and ZrO₂ samples are shown in Fig. 1. Two main diffraction lines at 2θ = 44.5° corresponding to Ni (111) plane (ICDD #: 04-0850) and 2θ = 52.0° (Ni (110); ICDD #: 04-0850) are ascribed to the metallic Ni phase present in all the samples. The other diffraction lines present in the XRD patterns were owing to their respective supports, as confirmed by the ZrO₂ phase (ICDD # 37-1484). The 20 wt% Ni/SiO₂ catalyst displayed only a broad peak centred around 2θ = 23° showing the micro-crystalline nature of SiO₂. From the previous reports, it is clear that because of the close proximity between the Ni, NiAl₂O₄ and γ-Al₂O₃ phases, it is difficult to discriminate the isolated phases. However, the formation of the NiAl₂O₄ phase was justified by H₂-TPR analysis of the Ni/Al₂O₃ catalyst (Fig. 2). The BET surface area and Ni domain size of 20 wt% Ni supported on SiO₂, γ-Al₂O₃, and ZrO₂ catalysts are reported in Table 1. The 20 wt% Ni/SiO₂ and 20 wt% Ni/γ-Al₂O₃ samples showed significantly higher surface areas than that of the 20 wt% Ni/ZrO₂ catalyst. The XRD patterns (Fig. 1) indicated a higher Ni crystallite size on the 20 wt% Ni/ZrO₂ (37.8 nm), while the 20 wt% Ni/SiO₂ and 20 wt% Ni/γ-Al₂O₃ catalysts both had a domain size of Ni of ∼21 nm. The crystallite size obtained for the respective catalysts indicates that the lower the surface area of the support, the higher the extent of Ni agglomeration, as a result large sized Ni particles are formed during the preparation (Table 1).

Fig. 1 XRD patterns of 20 wt% Ni supported on (a) SiO₂, (b) γ-Al₂O₃ and (c) ZrO₂ samples reduced at 450 °C for 3 h.

Fig. 2 The H₂-TPR profiles of 20 wt% Ni supported on (a) SiO₂, (b) ZrO₂ and (c) γ-Al₂O₃ catalysts.

Table 1 Physicochemical characteristics of the 20 wt% Ni supported on SiO₂, γ-Al₂O₃ and ZrO₂ samples

20 wt% Ni supported on	BET surface area (m^2 g^−1)	Ni crystallite size^a (nm)	H2 uptake^b (μmol gcat^−1)	BAS/LAS ratio^c
a Calculated from XRD spectra using Scherrer equation w.r.t Ni(111) plane.b Obtained from H2-TPR analysis; calibrated with Ag2O TPR.c BAS/LAS ratio obtained from pyridine adsorbed FT-IR spectra.
SiO2	155.5	20.5	1237	0.05
γ-Al2O3	108.8	21.4	1308	0.27
ZrO2	23.5	37.8	1204	0.38

---

3.2 H₂-temperature programmed reduction (TPR)

The H₂-TPR profiles of 20 wt% Ni supported on SiO₂, ZrO₂ and γ-Al₂O₃ catalysts are presented in Fig. 2 and their H₂ uptakes are reported in Table 1. The reduction maxima (T_max) were different for all the catalysts, although the samples have constant Ni loading. This suggests the variation in the Ni interaction with varied supports. The reduction signals appearing in the TPR patterns are associated with NiO reduction to Ni⁰. For the 20 wt% Ni/SiO₂ sample, T_max found at 322 °C is an indication of dispersed NiO particles. These findings are in good agreement with the results reported by Mile et al.^31,32 In the case of the 20 wt% Ni/ZrO₂ catalyst, the reduction maximum occurred at 420 °C, which is assigned to relatively large size NiO particles.^13,33 The reduction profile of the 20 wt% Ni/γ-Al₂O₃ sample indicated very high temperature signals; the one at T_max ∼ 550 °C is attributed to NiO interacting with γ-Al₂O₃ and the high temperature peak (T_max > 800 °C) is possibly owing to an NiAl₂O₄ spinel, which requires high temperature for its reduction.^34,35 However, XRD analysis did not confirm the reflections owing to the NiAl₂O₄ phase because of the superimposition of diffraction lines with that of the γ-Al₂O₃ phase. The TPR results revealed that the reduction behaviour of NiO particles was strongly influenced by the type of support.

3.3 Pyridine adsorption studies

To determine the proportion of BAS to LAS on the catalyst surface, pyridine adsorption was used in conjunction with in situ DRIFTS analysis (Fig. 3) and the ratios of Brønsted (BAS) to Lewis acid sites (LAS) are reported in Table 1. The relative ratios of pyridine adsorbed spectral bands at 1450 cm⁻¹ (Lewis acid site) and ∼1540 cm⁻¹ (Brønsted acid site) emphasized that all the catalysts had a higher number of LAS compared to BAS.³⁶ The proportion of BAS to LAS (0.38) was significantly higher on the 20 wt% Ni/ZrO₂ catalyst compared to the 20 wt% Ni/γ-Al₂O₃ (0.27) and 20 wt% Ni/SiO₂ (0.05) catalysts (Table 1).

Fig. 3 Pyridine adsorbed drift spectra of 20 wt% Ni supported on (a) ZrO₂, (b) γ-Al₂O₃ and (c) SiO₂ catalysts.

3.4 Catalytic hydrogenation of levulinic acid

The vapour phase hydrogenation of aqueous levulinic acid over SiO₂, γ-Al₂O₃ and ZrO₂ supported 20 wt% Ni catalysts is reported in Table 2. It can be seen that the 20 wt% Ni/SiO₂ catalyst exhibited better LA conversion and selectivity of GVL compared to that of the 20 wt% Ni/γ-Al₂O₃ and 20 wt% Ni/ZrO₂ catalysts. The low conversion of LA over the 20 wt% Ni/ZrO₂ catalyst may have been partly owing to larger Ni particles and also owing to lower Ni dispersion of the catalyst.

Table 2 Product distribution in the hydrogenation of LA over 20 wt% Ni supported on SiO₂, ZrO₂ and γ-Al₂O₃ catalysts. Reaction conditions: 10 wt% LA in H₂O, 270 °C, catalyst wt ∼ 0.05 g, H₂ = 20 cm³ min⁻¹, GHSV = 19.5 mL s⁻¹ g_cat⁻¹, activity data measured after 6 h on stream

20 wt% Ni on	% XLA	% selectivity					CO uptake^b (μmol gcat^−1)	SNi^c (m^2 gNi^−1)	TOFGVL^d (s^−1)	Carbon^e (%)
		GVL	AL	VA	MTHF	Others^a
a 1,4-Pentanediol, pentane, butane, CO2.b CO uptakes measured from pulse chemisorption.c Ni metal surface areas calculated from CO uptakes.d Rate/CO uptake.e Obtained from CHNS analysis of catalysts recovered after 15 h on stream.
SiO2	15.7	90.6	1.0	0.4	3.9	2.9	74.8	14.6	1.82	0.52
ZrO2	3.5	60.2	16.9	15.8	2.4	4.7	39.3	7.7	0.52	2.57
γ-Al2O3	10.7	71.9	4.2	6.6	8.2	9.0	56.5	11.1	1.32	2.28

---

The 20 wt% Ni/SiO₂ catalyst demonstrated very good selectivity of GVL (90.6%) while the 20 wt% Ni/γ-Al₂O₃ and 20 wt% Ni/ZrO₂ catalysts showed moderate selectivity of 60% and 72%, respectively. In addition to AL, valeric acid (VA), methyltetrahydrofuran (MTHF), 1,4-pentanediol and hydrocarbons such as butane and pentane are observed. Formation of AL is likely owing to the rate at which AL is formed being faster than the rate at which it is hydrogenated to GVL. The hydrogenation reactivity of these samples is related to the surface Ni metal area (Table 2). The presence of VA and MTHF, however, indicates that either GVL had reacted further and/or AL had undergone ring opening (Scheme 2). Higher selectivity of AL over the 20 wt% Ni/ZrO₂ catalyst is most likely owing to this intermediate not undergoing further reaction to GVL where this conversion would rely on the hydrogenation activity of the surface Ni. One possible reason is that the rate of hydrogenation of the double bond (in AL) on the metal site (Ni) seems to be very high on the 20 wt% Ni/SiO₂ catalyst compared to the Ni on ZrO₂ and γ-Al₂O₃ supports. Consequently, AL undergoes ring opening to produce VA, MTHF and HCs. Another reason may presumably be owing to the conversion of GVL to VA, MTHF and hydrocarbons, since both the 20 wt% Ni/γ-Al₂O₃ and 20 wt% Ni/ZrO₂ catalysts exhibited higher numbers of Brønsted acid sites.

Scheme 2 Reaction pathways during LA hydrogenation in the vapour phase.

The difference in selectivity of GVL, VA and MTHF is explained in terms of the variation in the nature and type of (Brønsted and/or Lewis) acid sites present on the catalyst surface. The sample (20 wt% Ni/SiO₂) with a lower ratio of Brønsted to Lewis acid sites (Table 1) demonstrated lower selectivity for VA and MTHF, while both the 20 wt% Ni/γ-Al₂O₃ and 20 wt% Ni/ZrO₂ catalysts had relatively higher BAS/LAS ratios and exhibited high selectivity for GVL ring opening products.²⁴ The cyclisation of LA to AL is initiated by Lewis acid sites present on the catalyst surface.²³ In the presence of a significantly high ratio of BAS/LAS on the 20 wt% Ni/ZrO₂ and 20 wt% Ni/γ-Al₂O₃ catalysts, the rate of ring opening of either AL and/or GVL to form VA is higher; as a result, a decrease in GVL selectivity was observed over both the catalysts (Table 2).

4. Discussion

The product distribution is strongly influenced by the type of acid sites (Brønsted and/or Lewis) as well as their strength on the catalyst surface. A detailed investigation is carried out using two different probe molecules while co-feeding along with the aqueous LA, such as pyridine (both Brønsted and Lewis acid site blocker) and 2,6-dimethylpyridine (as a selective Brønsted acid site blocker; Scheme 3).

Scheme 3 Surface poisoning of Brønsted and Lewis acid sites using pyridine (both Brønsted and Lewis site blocker) and 2,6-dimethylpyridine (a selective Brønsted acid site blocker) as probes.

An equimolar concentration of both pyridine (12.4 mmol) and 2,6-dimethylpyridine (12.45 mmol) was introduced in the reaction stream in (4 equal) successive pulses in separate experiments. The activity data obtained after the addition of pyridine and/or 2,6-dimethylpyridine are given in Table 3. It can be seen that a significantly lower conversion was obtained over the 20 wt% Ni/SiO₂ catalyst when pyridine was fed along with aqueous LA. The conversion of LA was decreased to 9.7% from 29.9% (activity loss of ∼60%); however, the GVL selectivity was marginally decreased when pyridine was co-fed over the 20 wt% Ni/SiO₂ catalyst (Table 3). From this, it can be inferred that the active Ni sites are not affected by the addition of pyridine. Since the BAS/LAS ratio on the 20 wt% Ni/SiO₂ catalyst is less than 0.1, it is therefore concluded that Lewis acid sites are responsible for the dehydration of LA. According to the reaction mechanism, the acid sites are crucial for the dehydration of LA to AL (1^st step in Scheme 2). The poisoned catalyst was then regenerated by replacing the aqueous LA with air flow (∼20 mL min⁻¹) at 450 °C for 3 h and subsequently reducing the catalyst in H₂ stream at 450 °C for 30 min⁻¹. After regeneration, the sample was examined for LA hydrogenation under similar experimental conditions (Table 3). Although the catalyst suffered a loss in LA conversion after probe addition, the catalytic activity was regained after the successive treatments at 450 °C in flowing air and in diluted H₂. In contrast to this, no significant change in LA conversion and GVL selectivity was observed when 2,6-dimethylpyridine was co-fed with aqueous LA. Pyridine would block both the Brønsted and Lewis acid sites when the catalyst is exposed at 270 °C (Scheme 3). A drastic reduction in LA conversion indicates the importance of Lewis sites for LA conversion to AL. On the other hand, when 2,6-dimethylpyridine was added, a marginal decrease in LA conversion implies the existence of available Lewis acid sites since the 2,6-dimethylpyridine is a selective Brønsted acid site blocker (Table 3). Therefore, it can be concluded that Brønsted acid sites play a vital role in the ring opening of GVL, which is further supplemented by the pyridine adsorbed FT-IR analysis. This could be explained owing to the presence of strong acid sites (Brønsted); LA chemisorption occurs strong enough to cause the ring opening of AL.²⁴ It is thus suggested that the 20 wt% Ni/γ-Al₂O₃ and 20 wt% Ni/ZrO₂ catalysts possess weaker hydrogenation activity than the 20 wt% Ni/SiO₂ catalyst. Finally, it can be concluded that a metal site (Ni) in close proximity to a Lewis acid site is active for the selective conversion of LA to GVL. These results suggest modification of surface properties can alter the product distribution thus the desired product selectivity can be fine-tuned.

Table 3 Influence of probe addition on aqueous LA hydrogenation over 20 wt% Ni/SiO₂ catalyst. Reaction conditions: 10 wt% aqueous LA, 270 °C; catalyst wt: ∼0.075 g, H₂ = 20 cm³ min⁻¹, GHSV = 9.74 mL s⁻¹ g_cat⁻¹

LA hydrogenation reaction condition and analysis	XLA	SGVL	Sothers
a Activities over 20 wt% Ni/SiO2.b Activities after the addition of 12.4 × 10^−3 moles of pyridine.c Activities after regeneration of catalyst.d Activities after the addition of 12.45 × 10^−3 moles of 2,6-dimethylpyridine.e Activities after regeneration of catalyst.
Without probe addition^a	29.9	90.5	10.4
After pyridine addition^b	9.70	81.1	18.9
After regeneration^c	27.8	88.8	11.2
Without probe addition^a	35.3	92.5	7.45
After 2,6-dimethylpyridine addition^d	25.2	91.5	8.50
After regeneration^e	30.7	89.6	10.4

---

The time on stream (TOS: 15 h) analysis emphasized the long-term stability of the 20 wt% Ni/SiO₂ catalyst (Table S3†). The CHNS analysis of the used catalysts revealed that there was a very low amount of carbon deposition over the 20 wt% Ni/SiO₂ catalyst compared to that over the other catalysts (Table 2). The lower activity of the 20 wt% Ni/γ-Al₂O₃ and 20 wt% Ni/ZrO₂ catalysts may also possibly be owing to the deactivation of the catalysts owing to fouling agents, like angelica lactone formed as an intermediate, which leads to coke deposition (∼2%) on the catalyst surface.¹⁹

To understand the nature of bulk NiO, samples using an inert support, i.e. α-Al₂O₃ and 20 wt% Ni/α-Al₂O₃, were examined for LA hydrogenation at a GHSV of 1.95 mL g_cat⁻¹ s⁻¹ and the results are reported in Table 4. The α-Al₂O₃ (known as an inert material possessing neither acid nor basic sites) showed minute activity at 325 °C with 100% selectivity towards AL, while the bulk NiO demonstrated about 9.5% LA conversion with 98.8% selectivity towards GVL. On the other hand, the 20 wt% Ni/α-Al₂O₃ catalyst exhibited 11.2% LA conversion and 93.6% GVL selectivity (Table 4). To further confirm the influence of acidity type and the role of metal in this hydrogenation, CeO₂ (a Lewis acidic material)³⁷ was taken and tested in the LA hydrogenation reaction, which showed about 4.4% conversion of LA with 100% selectivity of angelica lactone. This result thus emphasizes the role of Ni and the support material that is being used to disperse the active metal. We believe that the transformation of LA is occurring at the interface of the catalyst.

Table 4 LA hydrogenation activities over bulk NiO, α-Al₂O₃ and 20 wt% Ni/α-Al₂O₃ catalysts; reaction conditions: 10 wt% aqueous LA, 325 °C; catalyst wt: ∼0.5 g, H₂ = 20 cm³ min⁻¹, GHSV = 1.95 mL s⁻¹ g_cat⁻¹

Catalyst	XLA	SAL	SGVL	rAL/10^−9 (mol gcat^−1 s^−1)	rGVL/10^−9 (mol gcat^−1 s^−1)
α-Al2O3	0.2	100	0.0	0.98	0.0
NiO	9.5	1.7	98.8	0.79	46
20 wt% Ni/α-Al2O3	11.2	6.4	93.6	8.60	133

---

Therefore, we can conclude that the LA conversion is initiated by the Lewis acid sites present on the catalyst surface. A combination of surface Ni metal with adjacent Brønsted acid sites on the catalyst surface leads to ring opening of either GVL and/or AL to form VA and MTHF along with hydrocarbons. As aforementioned, a Lewis site combined with nickel is suitable for the selective hydrogenation of LA to GVL. Finally, it can be concluded that Lewis acid sites in close proximity to Ni sites (i.e. at interface) are responsible for dehydrocyclization of LA to form AL, followed by its hydrogenation, thus producing GVL. In contrast to this, although Ni supported on the α-Al₂O₃ demonstrated lower LA conversion, the reaction proceeds to cyclisation followed by hydrogenation to GVL. The high intrinsic activity of the 20 wt% Ni/SiO₂ catalyst may be attributed to the higher proportion of Lewis acid sites than Brønsted acid sites. We also believe that a uniform distribution of Ni particles on the 20 wt% Ni/SiO₂ catalyst may be another reason for selective conversion of LA to GVL (Fig. S1†).

5. Conclusions

The influence of supports was assessed for Ni in the LA hydrogenation by means of different conventional supports. It was found that SiO₂ support is suitable for Ni for high selectivity of γ-valerolactone. In the case of other supports, such as ZrO₂ and γ-Al₂O₃, the selectivity of VA and MTHF were detected at the cost of GVL formation. The 20 wt% Ni/ZrO₂ and 20 wt% Ni/γ-Al₂O₃ catalysts demonstrated mild hydrogenation activity and consequently high selectivity of angelica lactone is observed. A high ratio of Brønsted acid sites on the 20 wt% Ni/ZrO₂ and 20 wt% Ni/γ-Al₂O₃ catalysts resulted in ring opening of GVL. Finally, it can be concluded that the product distribution can be tuned to get the desired fuel additive (GVL or VA) as the strength of the surface acid–base sites regulates the selectivity in the hydrogenation of levulinic acid.

Acknowledgements

VVK and GN thank RMIT, Australia for the award of fellowships; MS, CA and VK thank CSIR New Delhi for the award of fellowships. All authors acknowledge Frank Antolasic, RMMF staff, RMIT, Australia for the support during characterization.

References

1. M. J. Climent, A. Corma and S. Iborra, Green Chem., 2014, 16, 516.
2. J. C. S. Ruiz, D. J. Braden, R. M. West and J. A. Dumesic, Appl. Catal., B, 2010, 100, 184.
3. J. M. Bermudez, J. A. Menéndez, A. A. Romero, E. Serrano, J. G. Martinez and R. Luque, Green Chem., 2015, 15, 2786.
4. J. Xin, D. Yan, O. Ayodele, Z. Zhang, X. Lu and S. Zhang, Green Chem., 2015, 17, 1065.
5. P. P. Upare, J. M. Lee, Y. K. Hwang, D. W. Hwang, J. H. Lee, S. B. Halligudi, J. S. Hwang and J. S. Chang, ChemSusChem, 2011, 4, 1749.
6. P. P. Upare, Y. K. Hwang, J. M. Lee, D. W. Hwang and J. S. Chang, ChemSusChem, 2015, 8, 2345.
7. R. A. Bourne, J. G. Stevens, J. Ke and M. Poliakoff, Chem. Commun., 2007, 4632.
8. W. Luo, P. C. A. Bruijnincx and B. M. Weckhuysen, J. Catal., 2014, 320, 33.
9. B. P. Kumar, N. Nagaraju, V. P. Kumar and K. V. R. Chary, Catal. Today, 2015, 250, 209.
10. M. Sudhakar, M. L. Kantam, V. S. Jaya, R. Kishore, K. V. Ramanujachary and A. Venugopal, Catal. Commun., 2014, 50, 101.
11. W. R. H. Wright and R. Palkovits, ChemSusChem, 2012, 5, 1657.
12. J. M. Tukacs, R. V. Jones, F. Darvas, G. Dibo, G. Lezsak and L. T. Mika, RSC Adv., 2013, 3, 16283.
13. V. Mohan, V. Venkateshwarlu, C. V. Pramod, B. D. Raju and K. S. R. Rao, Catal. Sci. Technol., 2014, 4, 1253.
14. J. Zhang, S. Wu, B. Li and H. D. Zhang, ChemCatChem, 2012, 4, 1230.
15. J. Zhang, J. Chen, Y. Guo and L. Chen, ACS Sustainable Chem. Eng., 2015, 3, 1708.
16. K. Yan, J. Liao, X. Wu and X. Xie, RSC Adv., 2013, 3, 3853.
17. C. M. Marrodan and P. Barbaro, Green Chem., 2014, 16, 3434.
18. S. G. Wettstein, J. Q. Bond, D. M. Alonso, H. N. Pham, A. K. Datye and J. A. Dumesic, Appl. Catal., B, 2012, 117–118, 321.
19. P. P. Upare, J. M. Lee, D. W. Hwang, S. B. Halligudi, Y. K. Hwang and J. S. Chang, J. Ind. Eng. Chem., 2011, 17, 287.
20. M. Sudhakar, V. V. Kumar, G. Naresh, M. L. Kantam, S. K. Bhargava and A. Venugopal, Appl. Catal., B, 2016, 180, 113.
21. P. P. Upare, M. G. Jeong, Y. K. Hwang, D. H. Kim, Y. D. Kim, D. W. Hwang, U. H. Lee and J. S. Chang, Appl. Catal., A, 2015, 491, 127.
22. V. Mohan, C. Raghavendra, C. V. Pramod, B. D. Raju and K. S. R. Rao, RSC Adv., 2014, 4, 9660.
23. V. V. Kumar, G. Naresh, M. Sudhakar, J. Tardio, S. K. Bhargava and A. Venugopal, Appl. Catal., A, 2015, 505, 217.
24. K. Kon, W. Onodera and K. I. Shimizu, Catal. Sci. Technol., 2014, 4, 3227.
25. W. Luo, U. Deka, A. M. Beale, E. R. H. V. Eck, P. C. A. Bruijnincx and B. M. Weckhuysen, J. Catal., 2013, 301, 175.
26. W. Luo, P. C. A. Bruijnincx and B. M. Weckhuysen, J. Catal., 2014, 320, 33.
27. A. M. Ruppert, J. Grams, M. Jedrzejczyk, J. Matrs-Michalska, N. Keller, K. Ostojska and P. Sautet, ChemSusChem, 2015, 8, 1538.
28. J. Lv, Z. Rong, Y. Wang, J. Xiu, Y. Wang and J. Qu, RSC Adv., 2015, 5, 72037.
29. K. Shimizhu, S. Kanno and K. Kon, Green Chem., 2014, 16, 3899.
30. R. Rodiansono, M. D. Astuti, A. Ghofur and K. C. Sembiring, Bull. Chem. React. Eng. Catal., 2015, 10(2), 192.
31. B. Mile, D. Stirling, M. Zammitt, A. Lovell and M. Webb, J. Catal., 1988, 114, 217.
32. J. Xiong, J. Chen and J. Zhang, Catal. Commun., 2007, 8, 345.
33. H. S. Roh, K. Y. Koo, J. H. Jeong, Y. T. Seo, D. J. Seo, Y.-S. Seo, W. L. Yoon and S. B. Park, Catal. Lett., 2007, 117, 85.
34. A. J. Akande, R. O. Idem and A. K. Dalai, Appl. Catal., 2005, 287, 159.
35. F. Marino, E. Cerrella, S. Duhalde, M. Jobbagy and M. Laborde, Int. J. Hydrogen Energy, 1988, 23, 1095.
36. H. Knözinger and P. Ratnasamy, Catal. Rev., 1978, 17, 31.
37. Y. Wang, F. Wang, Q. Song, Q. Xin, S. Xu and J. Xu, J. Am. Chem. Soc., 2013, 135, 1506.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra24199e

---