Nanostructured Nickel/Silica Catalysts for Continuous Flow Conversion of Levulinic Acid to γ-Valerolactone.

Selective transformation of levulinic acid (LA) to γ-valerolactone (GVL) using novel heterogeneous catalysts is one of the promising strategies for viable biomass processing. In this framework, we developed a continuous flow process for the selective hydrogenation of LA to GVL using several nanostructured Ni/SiO2 catalysts. The structural, textural, acidic, and redox properties of Ni/SiO2 catalysts, tuned by selectively varying the Ni amount from 5 to 40 wt %, were critically investigated using numerous materials characterization techniques. Electron microscopy images showed the formation of uniformly dispersed Ni nanoparticles on the SiO2 support, up to 30% Ni loading (average particle size is 9.2 nm), followed by a drastic increase in the particles size (21.3 nm) for 40% Ni-loaded catalyst. The fine dispersion of Ni particles has elicited a synergistic metal-support interaction, especially in 30% Ni/SiO2 catalyst, resulting in enhanced acidic and redox properties. Among the various catalysts tested, the 30% Ni/SiO2 catalyst showed the best performance with a remarkable 98% selectivity of GVL at complete conversion of LA for 2 h reaction time. Interestingly, this catalyst showed a steady selectivity to GVL (>97%), with a 54.5% conversion of LA during 20 h time-on-stream. The best performance of 30% Ni/SiO2 catalyst was attributed to well-balanced catalytic properties, such as ample amounts of strong acidic sites and abundant active metal sites. The obtained results show a great potential of applying earth-abundant nickel/silica catalysts for upgrading biomass platform molecules into value-added chemicals and high-energy-density fuels.

Introduction

Efficient upgrading of biomass-derived platform molecules to chemicals and biofuels has received tremendous attention in view of carbon-neutral society and a more sustainable economy.[1−8] Levulinic acid (LA) is one of the promising biomass derivatives that can be largely obtained via cellulose hydrolysis reaction.[9,10] Owing to two versatile functional groups (ketone and acid), LA can be transformed into a variety of valuable chemicals, such as γ-valerolactone (GVL), 1,4-pentanediol, succinic acid, 3-hydroxypropanoic acid, and 2-methyltetrahydrofuran.[11] In particular, the selective hydrogenation of LA to GVL is a vital step in biomass processing. GVL is an appealing chemical for the synthesis of 1,4-pentanediol, 2-methyltetrahydrofuran, adipic acid, biofuels, and fuel additives.[8,12−15] GVL can also be used as a solvent for biomass conversion reactions, as a precursor for the production of food additives and drugs as well as a gasoline blender.[16,17] In addition, GVL is an important chemical for the synthesis of monomers used in the manufacturing of nylon and other important polymers.[18]

Various homogeneous noble-metal-based catalysts, such as RuCl2(PPh3)3, Ru(acac)3 ligated with PBu3, tris(3-sulfonatophenyl)phosphine (TPPTS), P(Oct)3, RuCl3 in combination with PPh3 or TPPTS, and [Ir(COE)2Cl]2 are reported to show a high catalytic performance in LA to GVL transformation.[19−25] However, homogeneous catalysts have several disadvantages in terms of synthesis, cost, toxicity, stability, and recovery/reusability. Alternatively, heterogeneous solid catalysts could offer great potentials for viable biomass upgrading, attributed to their versatile benefits, such as low production costs, strong hydrothermal stability, and efficient recovery/reusability. Recent works demonstrated that Ru-based solid catalysts show a good activity in LA-to-GVL transformation, compared to other noble-metal-based catalysts (Pd, Pt, Ir, or Au).[26−31] For instance, Tukacs et al.[31] investigated the performance of three different catalysts (Pd/C, Ru/C, and Raney Ni) for LA hydrogenation. Among them, Ru/C catalyst showed high LA conversion (82.9%) and GVL yield (97.5%). In another work, Upare et al.[32] reported that the Ru/C catalyst exhibits a 98.6% yield of GVL, while about 90 and 30% yields of GVL over Pd/C and Pt/C catalysts, respectively. However, high cost of Ru metal could limit the practical applications of Ru-based catalysts, which stimulated the search for promising non-noble-metal catalysts not only for LA-to-GVL transformation, but also for other vital reactions in biomass processing.

The catalytic performance of several earth-abundant transition metals (Ni, Cu, Fe, or Co) deposited on high-surface-area supporting materials has been investigated for the hydrogenation of LA to GVL in liquid phase.[26,33−37] However, leaching of active metal species is a key concern in the liquid-phase (high-pressure) reactions, resulting in low stability of the catalyst and a futile reusability. Performing the catalytic reactions in a continuous flow vapor-phase system could provide promising solutions to overcome the drawbacks associated with the liquid-phase reactions. Numerous research efforts have been recently attempted toward developing Ni-based catalysts for continuous flow LA hydrogenation, attributed to their attractive structural and redox properties, along with the economic and stability advantages. For example, Sun et al.[38] studied the vapor-phase hydrogenation of LA and methyl levulinate over Cu/Al2O3, Ni/SiO2, and Co/SiO2 catalysts (20 wt % metal loading). Among them, the Ni/SiO2 catalyst showed the best performance in LA hydrogenation, with a 85.7% yield of GVL. Hengst et al.[39] found that 5 wt % Ni/Al2O3 catalyst prepared by a wet-impregnation method shows a good activity in continuous flow LA hydrogenation (90% LA conversion and 75% GVL yield). In a recent work, Yoshida et al.[40] have investigated the effect of the preparation method on the structure–activity properties of bimetallic Cu–Ni/SiO2 catalysts, prepared using 4 and 16 wt % loadings of Cu and Ni, respectively, for vapor-phase LA hydrogenation. The above studies demonstrate that the properties of Ni catalysts and their role in tailoring the conversion/selectivity in continuous flow LA hydrogenation may strongly depend on the metal composition. The key reason for this observation is that metal composition can determine the dispersion of active sites as well as metal–support interaction toward selective hydrogenation of LA to GVL product.

On the basis of the above background, we developed nanostructured Ni/SiO2 catalysts without using any base and organic protective agents. Stabilization of smaller metal nanoparticles, which are determined to be active sites for several catalytic reactions, on the catalyst surface remains a foremost challenge against the application of high-temperature (HT) conditions, but it is crucial for obtaining improved reaction rates as well as highly durable catalysts. Owing to high specific surface area and excellent thermal stability, silica as a support can stabilize nanosized metal particles by controlling their aggregation at high-temperature conditions via strong metal–support interactions. Hence, in this work, we have chosen silica as a support for the preparation of nanostructured Ni-based catalysts. The structural, morphology, acidic, and redox properties of Ni/SiO2 catalysts, tuned by varying the Ni loading from 5 to 40 wt % based on the SiO2 support, were systematically analyzed using X-ray diffraction (XRD), transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), temperature-programmed reduction (TPR), temperature-programmed desorption (TPD), H2 pulse chemisorption, and Brunauer–Emmett–Teller (BET) surface area techniques. The performance of nanostructured Ni/SiO2 catalysts was studied for continuous flow LA hydrogenation at mild reaction conditions. The effects of reaction temperature, reaction time, and concentration of LA were optimized to obtain good catalytic results in terms of LA conversion and GVL selectivity. Much attention has been paid to correlate the properties of nanostructured Ni/SiO2 materials with their catalytic activity in continuous flow LA hydrogenation.

Results and Discussion

Characterization Studies

Powder XRD profiles of Ni/SiO2 catalysts are shown in Figure . All catalysts exhibit various XRD peaks at 2θ = 36.6, 42.7, 62.2, 74.8, and 79.2°, which can be assigned to (111), (200), (220), (311), and (222) crystal facets of NiO, respectively.[41,42] In addition, a broad peak is observed at 2θ = 22°, which reveals the presence of amorphous SiO2.[41] Interestingly, the intensity of the XRD peaks increases with the increase of Ni loading, which indicates greater crystallinity and larger crystallite size in higher Ni loading catalysts.[42] To understand this, the average NiO crystallite size of the catalysts was calculated using the Debye–Scherrer equation (Table ). It was found that crystallite size increases gradually up to 30% Ni loading and then shows a rapid increase for 40% Ni/SiO2 catalyst (13.68 nm). This indicates that the crystal growth of NiO can be controlled up to 30% Ni loading, a key criterion for achieving abundant surface-exposed active sites. Powder XRD patterns of reduced Ni/SiO2 catalysts are shown in Figure S1. The diffraction peaks noted at 2θ = 44.49, 51.85, and 76.38° can be indexed to the (111), (200), and (220) crystal facets of metallic Ni with a face-centered cubic structure, respectively.[12,43] Here also, the average crystallite size of metallic Ni for all of the reduced Ni/SiO2 catalysts was estimated using the Debye–Scherrer equation (Table ). A similar trend was noted between the Ni crystallite size of the reduced Ni/SiO2 catalysts and the NiO crystallite sizes obtained for the as-synthesized Ni/SiO2 catalysts, i.e., 40% Ni/SiO2 catalyst exhibits a larger crystallite size under both reduced and nonreduced treatments.

Figure 1

Powder XRD patterns of Ni/SiO2 catalysts.

Table 1

Crystallite Size (D), BET Surface Area (SBET), Concentration of Acidic Sites, and Active Metal Surface Area (AMSA) of Ni/SiO2 Catalysts

catalyst	DNiO (nm)a	DNi (nm)a	SBET (m2/g)b	acidic sites (μmol/g)c	AMSA (m2/g of Ni)d
5% Ni/SiO2	9.74 ± 0.5	7.83 ± 0.5	163 ± 3	154.24 ± 2	6.4 ± 0.5
10% Ni/SiO2	10.47 ± 0.5	8.28 ± 0.5	152 ± 3	168.46 ± 2	9.4 ± 0.5
20% Ni/SiO2	11.31 ± 0.5	9.46 ± 0.5	144 ± 3	184.37 ± 2	12.5 ± 0.5
30% Ni/SiO2	11.39 ± 0.5	9.52 ± 0.5	138 ± 3	206.09 ± 2	16.5 ± 0.5
40% Ni/SiO2	13.68 ± 0.5	11.47 ± 0.5	117 ± 3	178.65 ± 2	14.2 ± 0.5

From XRD analysis.

From N2 adsorption−desorption studies.

Obtained from NH3-TPD studies.

Calculated by H2-pulse chemisorption studies.

XPS analysis has been undertaken to estimate oxidation states of the elements as well as to understand metal–support interactions in Ni/SiO2 catalysts. The Ni 2p XPS images of Ni/SiO2 catalysts are shown in Figure . Two major peaks were noted at 854.9 and 872.6 eV, corresponding to Ni 2p3/2 and Ni 2p1/2.[44−46] The estimated splitting between Ni 2p3/2 and Ni 2p1/2 XPS peaks of Ni2+ was found to 18.4 eV, which is well correlated with the literature reports.[46,47] In addition, two shake-up satellite peaks were observed at 861.3 and 879.2 eV, which confirm the +2 oxidation state of Ni species in the developed Ni/SiO2 catalysts.[44,48] As shown in Figure , Ni 2p3/2 peak is split into two peaks: the first peak at 854.3 eV is attributed to Ni2+species present within the NiO phase and the second peak at 856 eV is assigned to Ni2+ species present in the composite oxide material.[47,48] This observation indicates the existence of different types of Ni2+ species in the Ni/SiO2 catalysts. The O 1s XPS images of Ni/SiO2 catalysts are presented in Figure . The peak observed at 529.6 eV corresponds to lattice oxygen ions bonded to Ni cations, whereas the peak noted at 532.4–533.1 eV indicates O2– species in the SiO2 support (i.e., Si–O–Si environment).[49−51] The O 1s XPS peak of Si–O in Ni/SiO2 catalysts is shifted to lower binding energies compared to that of pure SiO2, which reveals the charge transfer from Si to O in Ni/SiO2 catalysts.[52] The Si 2p XPS images of Ni/SiO2 catalysts are shown in Figure 2S, Supporting Information. A major XPS peak was noted at around 103.1–103.7 eV, which indicates the presence of Si4+ species in the Ni/SiO2 catalysts.[53] The intensity of the Si 2p peak gradually decreases with the increase of Ni loading, which was due to the coverage of SiO2 surface by Ni phase at higher Ni loadings.[51]

Figure 2

Ni 2p XPS images of (a) 5% Ni/SiO2, (b) 10% Ni/SiO2, (c) 20% Ni/SiO2, (d) 30% Ni/SiO2, and (e) 40% Ni/SiO2 catalysts.

Figure 3

O 1s XPS images of prepared Ni/SiO2 catalysts with pure SiO2 sample.

The morphology, particle size, and the size distribution of NiO and SiO2 for all catalysts were estimated using TEM and scanning transmission electron microscopy-energy-dispersive spectrometry (STEM-EDS) studies. Figure shows the TEM images of 5% Ni/SiO2, 30% Ni/SiO2, and 40% Ni/SiO2 catalysts. Irregular shaped SiO2 nanoparticles were formed in all Ni/SiO2 catalysts. As shown in Figure , a number of tiny NiO particles (black dots) were found in 5% Ni/SiO2 and 30% Ni/SiO2 catalysts. In contrast, largely agglomerated particles were noted in 40% Ni/SiO2 catalyst. To understand this, the size of the NiO particles was estimated and the obtained data are summarized in histograms, as shown in Figure . It was noted that 40% Ni/SiO2 catalyst contains larger particles (average value is 20.3 nm) with a broad size distribution, compared to 5% Ni/SiO2 (7.5 nm) and 30% Ni/SiO2 (9.2 nm) catalysts. This observation is well correlated with the crystallite sizes estimated from the XRD studies (Table ). With the aim of confirming these results, STEM-EDS characterization was performed and the obtained images are shown in Figure . It is clearly noted that 30% Ni/SiO2 catalyst exhibits homogeneous dispersion of Ni and Si species. This is a key criterion for achieving promising active sites (particularly, enhanced redox properties) not only for LA hydrogenation, but also for other vital catalytic reactions. Irrespective of Ni loading, smaller nanoparticles were found on the surface of SiO2 support, as evidenced by TEM studies (Figures and 5), especially in the case of 30% Ni/SiO2 catalyst. The absence of macroscopic metal particles indicates beneficial effect of SiO2 support in stabilizing nanosized particles against higher calcination conditions. The BET surface areas of Ni/SiO2 catalysts are presented in Table . It was obvious that the BET surface area of Ni/SiO2 catalysts slightly decreases up to 30% Ni loading and then a drastic decrease for 40% Ni/SiO2 catalyst. This observation is due to the agglomeration of NiO particles at higher Ni loadings, which was evident from the both crystallite sizes (Figure and Table ) and particle sizes (Figures and 5) estimated for all of the catalysts.

Figure 4

TEM images and the corresponding particle size distribution of 5% Ni/SiO2, 30% Ni/SiO2, and 40% Ni/SiO2 catalysts.

Figure 5

STEM-EDS elemental mapping images of 5% Ni/SiO2, 30% Ni/SiO2, and 40% Ni/SiO2 catalysts.

The reduction behavior of Ni species in Ni/SiO2 catalysts, a determinant factor for choosing promising catalysts for LA hydrogenation, is estimated using H2-TPR analysis (Figure ). In addition, this analysis allows to qualitatively evaluate the metal–support interaction in the supported metal catalysts.[54−56] All catalysts show a major peak centered at 850 K, along with a shoulder-like peak at around 1022 K (Figure ). These peaks suggest the existence of different types of NiO species interacting with the SiO2 support.[57] The high-intensity peak noted at 850 K indicates the reduction of NiO interacting with the SiO2 support.[58] This peak intensity gradually increases with the increase of Ni loading. Another peak observed at 1022 K can be attributed to the reduction of nickel silicate species.[42] These nickel silicate species could strongly interact with the support and hence, their reduction is only possible at higher temperatures. The results of active metal surface area (Table ) revealed that 30% Ni/SiO2 catalyst shows a larger amount of H2 consumption. This observation indicates the existence of high concentration of active metal sites in the 30% Ni/SiO2 catalyst.[59]

Figure 6

H2-TPR studies of synthesized Ni/SiO2 catalysts.

The acidic properties of Ni/SiO2 catalysts were investigated using NH3-TPD analysis (Figure ). Various NH3-TPD peaks were found for all of the catalysts, indicating the presence of different types of acidic sites. To define the strength of the acidic sites present in the Ni/SiO2 catalysts, the NH3-TPD profiles can be categorized as low-temperature (LT) and high-temperature (HT) regions, before and after 673 K, respectively. The LT region peak reveals the presence of weak acid sites, whereas the HT region peak corresponds to strong acidic sites. It is evident from Figure that the intensity of LT region peak decreases with the increase of Ni loading. Interestingly, the intensity of HT region peak increased up to 30% Ni loading and then showed a sudden decrease for 40% Ni/SiO2 catalyst. To understand this, the concentration of acidic sites is estimated for all of the catalysts (Table ). Among them, 30% Ni/SiO2 catalyst exhibits high concentration of acidic sites (192 μmol/g), which could play a key role in the LA hydrogenation, as discussed in the following paragraphs.

Figure 7

NH3-TPD profiles of (a) 5% Ni/SiO2, (b) 10% Ni/SiO2, (c) 20% Ni/SiO2, (d) 30% Ni/SiO2, and (e) 40% Ni/SiO2 samples.

Continuous Flow Hydrogenation of Levulinic Acid

Figure shows the catalytic activity results obtained for LA-to-GVL transformation as a function of reaction time over the Ni/SiO2 catalysts. The reaction conditions are 30 mL/min H2 flow, 523 K reaction temperature, 1 g of catalyst, and 1 mL/h LA feed rate. Among the catalysts tested, 30% Ni/SiO2 catalyst shows a 100% LA conversion at 2 h reaction time, whereas only 40, 61.9, 80.3, and 91.5% LA conversions were achieved for 5% Ni/SiO2, 10% Ni/SiO2, 20% Ni/SiO2, and 40% Ni/SiO2 catalysts, respectively. However, the conversion of LA continuously decreases with reaction time for all of the catalysts, which could be due to the blockage of active sites by the reaction substrates and/or coke formation. However, the unique feature of 30% Ni/SiO2 catalyst is that it shows a steady conversion of LA from 16 (59.8%) to even after 20 h reaction time (57%). This stable LA conversion was not achieved over the remaining catalysts. As shown in Figure B, LA is mainly converted into the GVL product, with the considerable amounts of angelica lactones and pentanoic acid products (Scheme ). For 2 h reaction time, 5% Ni/SiO2 catalyst shows 60% selectivity to GVL and the remaining Ni/SiO2 catalysts show more than 96% selectivity of GVL. Interestingly, there was not much difference in GVL selectivity up to 6 h reaction time for 10% Ni/SiO2, 20% Ni/SiO2, 30% Ni/SiO2, and 40% Ni/SiO2 catalysts. After 6 h reaction time, the selectivity of the GVL is found to decrease for 10% Ni/SiO2 and 20% Ni/SiO2 catalysts, whereas 30% Ni/SiO2 and 40% Ni/SiO2 catalysts exhibit a stable selectivity to GVL (95.1 and 95.4%, respectively) up to 12 h reaction time. Afterward, a drastic decrease in GVL selectivity was noted over 40% Ni/SiO2 catalyst (70.1%), whereas there was no much variation in GVL selectivity (93.2%) for 30% Ni/SiO2 catalyst even after 20 h reaction time. Therefore, among all of the catalysts tested, 30% Ni/SiO2 catalyst exhibits the best performance in terms of LA conversion and GVL selectivity. Importantly, this catalyst showed a stable performance for GVL selectivity even after 20 h reaction time, which is crucial for the practical application of earth-abundant Ni/SiO2 catalysts, not only for continuous flow LA hydrogenation, but also for other vital reactions in biomass upgrading.

Figure 8

(A) Conversion of levulinic acid (LA) and (B) selectivity to γ-valerolactone (GVL) obtained over nanostructured Ni/SiO2 catalysts in LA hydrogenation. Reaction conditions: 30 mL/min H2 flow, 523 K reaction temperature, 1 g of catalyst, and 1 mL/h LA feed rate.

Scheme 1

Plausible Reaction Pathway for Continuous Flow Hydrogenation of Levulinic Acid To Produce γ-Valerolactone

To study the variation of LA conversion and GVL selectivity with the reaction temperature, we have studied the hydrogenation of LA at different reaction temperatures from 498, 523 to 548 K over the 30% Ni/SiO2 catalyst (Figure ). It was found that the reaction temperature plays a significant role in the LA hydrogenation. Results reveal that the conversion of LA is very low (59.1%) at 498 K reaction temperature for 2 h reaction time, whereas a 100% LA conversion is obtained at 523 and 548 K reaction temperatures. With the increase of reaction time, the conversion of LA is found to decrease at all reaction temperatures. At 498 K reaction temperature, a drastic decrease in LA conversion is noted (5.2% for 20 h reaction time). In contrast, the LA conversion decreases gradually at 523 and 548 K reaction temperatures, with 57 and 63.9% conversions of LA for 20 h reaction time, respectively. This indicates the necessity of the reaction temperate higher than 498 K for achieving maximum conversion of LA under continuous flow conditions. Interesting results were observed toward GVL selectivity. For 2 h reaction time, a high selectivity of GVL is obtained at all reaction temperatures. Interestingly, the GVL selectivity (93%) is maintained with the reaction time at 523 K reaction temperature. In contrast, a poor selectivity (22%) of GVL (the remaining byproducts are pentanoic acid, 1-pentanol, 1,4-pentanediol, and methyltetrahydrofuran) is obtained at 548 K reaction temperature for 20 h reaction time. At 498 K reaction temperature, the selectivity of GVL notably decreased from 12 to 20 h of reaction time and only a 49.2% selectivity of GVL is obtained after 20 h reaction time. These results therefore demonstrate that a better catalytic result in terms of LA conversion and GVL selectivity were achieved at 523 K reaction temperature.

Figure 9

Effect of reaction temperature on the conversion of levulinic acid (A) and the selectivity of γ-valerolactone (B) over 30% Ni/SiO2 catalyst as a function of reaction time in levulinic acid hydrogenation. Reaction conditions: 30 mL/min H2 flow, 1 g catalyst, and 1 mL/h LA feed rate. LA—levulinic acid and GVL—γ-valerolactone.

We have also studied the effect of LA feed rate (0.5 and 1.0 mL/h) on the continuous flow hydrogenation of LA to GVL as a function of reaction time (20 h) using 30% Ni/SiO2 catalyst (Figure ). The obtained LA conversions at 2 and 20 h reaction times are 100 and 80.4% for 0.5 mL/h of LA feed rate, while 99.6 and 54.5% of LA conversions were obtained for 1 mL/h of LA feed rate, respectively. These results clearly indicate a high and steady conversion of LA at 0.5 mL/h of LA feed rate. On the other hand, the selectivity of the GVL was independent of the feed rate of LA. The obtained GVL selectivities for 0.5 and 1 mL/h feed rates of LA were 92.7 and ∼93% at 20 h reaction time, respectively.

Figure 10

Effect of levulinic acid (LA) feed rate on the LA conversion and the selectivity of γ-valerolactone (GVL) over 30% Ni/SiO2 catalyst as a function of reaction time. Reaction conditions: 30 mL/min H2 flow, 523 K reaction temperature, and 1 g of catalyst.

The hydrogenation of LA to GVL typically follows two reaction pathways: (i) angelica lactones pathway and (ii) 4-hydroxypentanoic acid pathway.[13,26,60,61] However, both reaction pathways require well-balanced acidic and active metal sites. In the angelica lactones pathway, the acidic sites of the catalyst initiate the dehydration of LA to form intermediate products (angelica lactones). Further hydrogenation of these intermediates over the metallic Ni sites yields the GVL product. In contrast, active metal-catalyzed hydrogenation of LA to 4-hydroxypentanoic acid intermediate occurred in another reaction pathway, as shown in Scheme . The subsequent cyclization of this intermediate to produce GVL is catalyzed by acidic sites. The detailed physicochemical characterization studies revealed the presence of both acidic and active metal sites in the investigated Ni/SiO2 catalysts (Table ). The formation of angelica lactones reveals that the LA hydrogenation follows the angelica lactones pathway over Ni/SiO2 catalysts. Simultaneously, pentanoic acid is also formed in LA hydrogenation, which is due to the dehydration–hydrogenation reactions of 4-hydroxypentanoic acid.[13,26,60] Hence, both reaction pathways (Scheme ) are possible for continuous flow LA hydrogenation over Ni/SiO2 catalysts. By comparing the activity data with the characterization studies, a clear structure–activity correlation of Ni/SiO2 catalysts was achieved for continuous flow LA hydrogenation. As shown in Figure , 30% Ni/SiO2 catalyst contains higher amounts of acidic and active metal sites, resulting in higher yields of GVL from LA hydrogenation. This is because both acidic and active metal sites play a favorable role in LA hydrogenation (Scheme ). Although SiO2 could not show a direct effect on the hydrogenation of levulinic acid, due to the lack of catalytic active sites, it is essential for obtaining finely dispersed active phase nanoparticles, which was clearly noted in the present study (Figures and 5). As a result, adequate amounts of redox (Figure ) and acid (Figure ) properties were found in the nanostructured Ni/SiO2 catalysts, which played a considerable effect in the LA hydrogenation to selectively obtain GVL product. The better performance of 30% Ni/SiO2 catalyst, along with the economic and the stability benefits of nickel and silica, could enable it a new appealing catalyst for various important hydroprocessing reactions in biomass upgrading.

Figure 11

Structure–activity correlations of Ni/SiO2 catalysts for γ-valerolactone (GVL) yields obtained from levulinic acid hydrogenation under continuous flow conditions.

To understand the key reasons for the observed catalyst deactivation in LA hydrogenation, XRD and TEM studies of used 30% Ni/SiO2 catalyst were carried out (Figure ) and the obtained results were compared to those of fresh catalyst. As shown in Figure A, both reduced and used 30% Ni/SiO2 catalysts show similar XRD patterns in terms of diffraction angle and peak intensity. It indicates that the crystal structure of nickel is unaffected by the employed reaction conditions in LA hydrogenation. However, largely aggregated Ni particles with a broad size distribution (15.2 nm of mean particle size) were found in the used 30% Ni/SiO2 catalyst. It was already noted that fresh 30% Ni/SiO2 catalyst contains finely dispersed nanoparticles with a 9.2 nm of mean particle size (Figure ). Hence, it appears that the increased particle size of 30% Ni/SiO2 catalyst is one of the key reasons for the catalyst deactivation in LA hydrogenation. The development of supported bimetallic catalysts, such as Ni–Cu/SiO2, would be a promising solution for improving catalyst’s stability as well as for achieving durable results in terms of conversion/selectivity in the hydrogenation of levulinic acid. This is because the addition of another metal to Ni may provide synergistic metal–metal interactions, due to the active interplay of electron transfer at the interface structures. Hence, large amounts of redox properties near the vicinity of the interfaces can be obtained in the supported bimetallic catalysts. In addition, synergistic Cu–Ni nanoalloys dispersed on a high-surface-area support like SiO2 could show high resistance to particle sintering compared to the individual nanoparticles and hence, improved stability of supported nanoalloys against higher-temperature conditions. Therefore, the application of supported nanoalloy catalysts in biomass valorization could overcome the drawbacks associated with the supported monometallic catalysts.

Figure 12

(A) Powder XRD patterns of calcined, reduced, and used 30% Ni/SiO2 catalyst and (B) TEM image of used 30% Ni/SiO2 catalyst.

Conclusions

A continuous flow process was developed for the selective conversion of LA to GVL over a variety of nanostructured Ni/SiO2 catalysts. Among them, the 30% Ni/SiO2 catalyst exhibits the best performance in terms of LA conversion and GVL selectivity. The activity order of Ni/SiO2 catalysts in LA hydrogenation is 5% Ni/SiO2 < 10% Ni/SiO2 < 20% Ni/SiO2 < 40% Ni/SiO2 < 30% Ni/SiO2. Importantly, the 30% Ni/SiO2 catalyst showed a stable selectivity to GVL (>97%), with a 54.5% conversion of LA during 20 h time-on-stream. The systematic characterization of the catalysts reveals a strong dependence of textural, acidic, and redox properties on the Ni loading. A linear correlation between the acidic-active metal sites and the GVL yield was achieved in this study. Characterization of used 30% Ni/SiO2 catalyst reveals the formation of largely aggregated particles with a broad size distribution, which might be the reason for the catalyst deactivation in LA hydrogenation.

Experimental Section

Catalyst Preparation

Nanostructured Ni/SiO2 catalysts with various Ni loadings from 5 to 40 wt % (with respect to SiO2) were synthesized using a modified wetness impregnation method. In a typical procedure, an appropriate amount of colloidal SiO2 (Aldrich, AR grade) was dispersed in double-distilled water under mild stirring conditions, followed by the addition of a required quantity of Ni(NO3)2·6H2O (Aldrich, AR grade) dissolved in aqueous medium. The mixture solution was stirred at 400 rpm for 20 min at 333 K before the temperature was raised to 373 K and stirred until the liquid had evaporated. The obtained samples were oven-dried at 373 K for 12 h and then calcined in static air at 773 K for 5 h with a heating ramp of 5 K/min. For convenience, the catalysts prepared using 5, 10, 20, 30, and 40 wt % of Ni loadings are labeled as 5% Ni/SiO2, 10% Ni/SiO2, 20% Ni/SiO2, 30% Ni/SiO2, and 40% Ni/SiO2, respectively.

Catalyst Characterization

Powder X-ray diffractograms of Ni/SiO2 catalysts were recorded using a Rigaku diffractometer, equipped with a Cu Kα radiation (0.1540 nm) source, operating at 40 kV and 40 mA. Diffractograms were collected over a 2θ range of 10–80°, with a step size of 0.02° and a step time of 2.4 s.

The BET surface area, pore size, and pore volume of Ni/SiO2 catalysts were obtained via N2 adsorption–desorption studies carried out using a Micromeritics ASAP 2020 instrument at liquid N2 temperature (77 K). Prior to analysis, the samples were degassed under vacuum at 573 K for 3 h to remove residual moisture and other volatile species adsorbed on the catalyst surface.

The oxidation states of Ni, O, and Si elements in the Ni/SiO2 catalysts were estimated using X-ray photoelectron spectroscopy (XPS) analysis. For this, a VG Scientific ESCALAB-210 spectrometer was used with Mg Kα radiation (photon energy of 1253.6 eV) as the X-ray source. The adventitious C 1s peak at 284.6 eV was used for the charge correction of the binding energies obtained for Ni 2p, O 1s, and Si 2p spectra.

Transmission electron microscopy (TEM) experiments were performed to estimate particle size and its distribution in Ni/SiO2 catalysts. For this, a Tecnai G2 TEM microscope equipped with a slow scan charge-coupled device camera and operating at an accelerating voltage of 200 kV was used. Sample preparation involves sonication of about 10 mg of catalyst in acetone for 5 min before deposition onto a holey carbon-coated copper grid.

The reducible nature of Ni/SiO2 catalysts was estimated using H2-temperature-programmed reduction (H2-TPR) analysis performed over a Micromeritics model AutoChem 2910 instrument. For this analysis, the required amount of calcined catalyst (about 50 mg) was placed into a quartz tubular reactor. The reactor was then gradually heated up to 573 K (5 °C/min) in He flow (30 mL/min). The temperature of the reactor was maintained at 573 K for another 60 min to completely remove hydroxyl and other volatile species adsorbed on the catalyst surface. Afterward, the reactor was slowly cooled to 313 K, followed by switching of the gas flow to 10% H2 in Ar (30 mL/min). Finally, the temperature of the reactor was increased to 1073 K at a heating ramp of 5 K/min. The purified gas, obtained after removing water molecules from the reactor effluent gas using a molecular sieve trap, was analyzed by gas chromatography (GC), equipped with a thermal conductivity detector.

The active metal surface area of Ni/SiO2 catalysts was estimated using H2 pulse chemisorption experiments performed over a Micromeritics AutoChem II 2920 instrument. Prior to analysis, the Ni/SiO2 catalyst was reduced at 973 K for 1 h under 10% H2/Ar atmosphere (30 mL/min flow). After cooling the reactor to room temperature under Ar flow, H2 pulses were immediately injected until the eluted peak areas of at least three successive pulses become constant. The active surface area of Ni in Ni/SiO2 catalysts was estimated from H2 volume adsorbed on the catalyst surface by assuming the value of H/Ni (surface nickel atom) is one and a surface area of 6.5 × 10 to 20 m2 per Ni atom.where Vchem (mol/g) is the chemisorption volume; σNi (nm2) is the Ni cross section area; and SF is the stoichiometry factor.

The acidic strength of Ni/SiO2 catalysts was determined using NH3-TPD analysis carried out on a Micromeritics AutoChem 2910 instrument, equipped with a thermal conductivity detector. About 30 mg of catalyst was initially preactivated at 573 K for 1 h to remove adsorbed moisture and other volatile species from the catalyst surface and then, the reactor was cooled to 323 K. Afterward, 4.95% NH3 gas balanced with He (20 mL/min flow rate) was passed through the catalyst surface for 60 min. The physisorbed NH3 was immediately removed from the catalyst surface by flushing with He gas. The chemisorbed amount of NH3 was then estimated in He flow (20 mL/min) while gradually increasing the reactor temperature from 323 to 1073 K (10 K/min).

Catalytic Activity Performance

The catalytic activity measurements for continuous flow LA (Sigma-Aldrich, 99.9%) hydrogenation were carried out in a down flow fixed-bed reactor under atmospheric pressure. In a typical experiment, the mixture of 1 g of catalyst and 1 g of quartz particles was placed in a tubular reactor and held between two end plugs of quartz wool. The reaction bed temperature was monitored at the wall of reactor with a type K thermocouple and controlled with a programmable temperature controller. In each experiment, the catalyst was reduced in situ for 6 h at 873 K temperature in 10% H2/Ar (30 mL/min). Afterward, the reactor was cooled to a desired reaction bed temperature and LA was fed into the vaporizer at a flow rate of 0.5 or 1 mL/h using a syringe feed pump under the H2 flow rate of 30 mL/min. Finally, the liquid products were collected at the bottom of the reactor with an ice-cooled trap in regular time intervals. These liquid products were analyzed by a GC (Nucon 5785) equipped with a PEG-20 M capillary column and a flame ionization detector. The products were identified using a GC–mass spectrometer, equipped with QP5050A (Shimadzu Instruments, Japan). The conversion of LA and the selectivity of GVL were estimated according to the following equationswhere nLA and nLA represent initial and final concentrations of the reactant, respectively.