Single-Pot Reductive Rearrangement of Furfural to Cyclopentanone over Silica-Supported Pd Catalysts.

Direct one-pot hydrogenation of furfural (FFR) to cyclopentanone (CPO) was investigated over different silica-supported Pd catalysts. Among these, 4% Pd on fumed silica (4%Pd/f-SiO2) showed remarkable results, achieving almost 98% furfural (FFR) conversion with ∼89% selectivity and 87% yield to cyclopentanone at 165 °C and 500 psig H2 pressure. More interestingly, the fumed-silica-supported catalyst tuned the selectivity toward the rearrangement product, i.e., cyclopentanone, whereas all of the other supports were found to give ring hydrogenation as well as side chain hydrogenation products due to their parent Brönsted acidity and specific support properties. X-ray diffraction data revealed the presence of different phases of the face-centered cubic lattice of metallic Pd along with lowest crystallite size of 15.6 nm in the case of the silica-supported Pd catalyst. However, Pd particle size was found to be in the range of 5-13 nm with even dispersion over the silica support, confirmed by high-resolution transmission electron microscopy analysis. While studying the effect of reaction parameters, it was observed that lower temperature gave low furfural conversion of 58% with only 51% CPO selectivity. Similarly, higher H2 pressure lowered CPO selectivity with subsequent increase in 2-methyl furan and ring hydrogenation product 2-methyl furan and 2-methyl tetrahydrofuran. Thus, as per the requirement, the product selectivity can be tuned by varying the type of support and/or the reaction parameters suitably. With the help of several control experiments and the characterization data, a plausible reaction pathway was proposed for the selective formation of cyclopentanone.

Introduction

Ever increasing depletion of fossil resources augmented the research of renewable alternatives, and eventually much of the focus is on biomass valorization as a potential feedstock to meet the growing demand of fuel and chemicals.[1−3] Implementing new concepts of catalysis for the selective and efficient defunctionalization of biomass-derived molecules combined with scale-up strategies becomes the prime agenda for viable solutions in near future.[4,5] Furfural (FFR) as a platform molecule is commercially produced via acid-catalyzed dehydration of the lignocellulosic part of biomass primarily from arabinose and xylose.[5] Because of its oxygen-rich content and the presence of a pendent carbonyl functionality, furfural hydrogenation leads to a spectrum of products such as furfuryl alcohol (FAL),[6] tetrahydrofuran (THF),[7] tetrahydrofurfuryl alcohol (THFAL),[8] 2-methyl furan (2-MF),[9a] 2-methyl tetrahydrofuran (2-MeTHF),[10] pentane diols (PeDO),[11,12,9b] cyclopentanone (CPO),[13] etc. (Scheme ). Because of the multiproduct formation in furfural hydrogenation, one of the key challenges is to achieve the highest selectivity to the desired product. The multifunctionality of furfural offers specific adsorption of C=O and/or C=C bonds on the catalyst surface depending on the type of catalyst/support under optimized reaction conditions, which results in different selectivities for products.[2,4,5] An efficient transformation of furfural at an affordable cost and under ambient conditions is the driving force to develop new catalyst systems and novel conversion routes.[2,5]

Scheme 1

General Scheme for Furfural Hydrogenation

Cyclopentanone is one of the fine chemicals used in pharmaceuticals, insecticides, and rubber chemicals.[14,15] The product mixture of cyclopentanone and cyclopentanol (CPO and CPL) is used in the fragrance and perfume industry as these are the major ingredients of jasmine family.[15] CPO is also used as a solvent in various electronic industries as an antiresin reagent because of its good resin solubilizing capacity. On the other hand, unique physical properties of CPO such as high hydrophobic characteristic, low latent heat, and reluctance to peroxide formations make it an important starting material for Grignard reactions and the formation of various derivatives as well as in coupling reactions, etc.[14] It has a strong potential to be used in polyamide preparations along with the preparation of C15–17 jet fuels and as a stabilizer for polyolefins.[16]

The conventional synthesis of CPO involves catalytic vapor-phase cyclization of 1,6-hexanediols or esters of adipic acid. It is also prepared by liquid-phase catalytic oxidation of cyclopentene using nitric acid. In both the routes, substrates are mainly of petroleum origin.[17−19] As furfural is easily available from several types of agro wastes, cyclopentanone synthesis has been attempted starting from furfural. Hronec et al. have extensively explored the conversion of furfural and different furfural derivatives into CPO using various noble as well as non-noble metal catalyst systems.[20−22] Among several catalysts studied, the highest yield of 82% of CPO and CPL together was achieved over the 5% Pt/C catalyst but at a very high pressure of H2 (8 MPa).[21] Later, the same research group has developed a bimetallic Pd–Cu catalyst system, achieving complete FFR conversion and high CPO yield of 92% within a short period of time (1 h) but using very high metal loading.[22] Some other supported catalysts reported include Ni–Cu/SBA giving 62% yield of CPO at 4 MPa H2 pressure, whereas a mixture of Ru/C and Amberlyst-15 in biphasic systems gave only 16% CPO selectivity.[23,24] However, Ru supported on carbon nanotube (CNT) and MIL-101 catalysts showed almost 90% yield of CPO.[25,26] Copper alone and in combination with other metals were also reported for direct hydrogenation of furfural to cyclopentanone. For example, the Cu–Co catalyst system was found to give ∼67% selectivity to each CPO and CPL by varying Cu loadings and catalyst preparation methods at 160 °C and 4 MPa H2 pressure.[16] Various hydrotalcite-based Cu–Ni–Al, Cu–Zn–Al, and Cu–Mg–Al also showed selective formation of either CPO or CPL in the range of 40–75%.[27−29] Recently, Jiang et al. reported a 20% Cu/CNT catalyst giving >65% CPO yield at 140 °C and at 4 MPa H2 pressure, however requiring a longer reaction time of 12 h.[30]

Thus, developing a suitable catalyst for selective ring opening and closer via rearrangement followed by hydrogenation of furfural at milder conditions still remains a challenge. In this article, we report for the first time Pd on a fumed SiO2 (Pd/f-silica) as a catalyst system for one-pot aqueous-phase hydrogenation of furfural, giving excellent performance in terms of 98% conversion of furfural and almost ∼89% selectivity to cyclopentanone at considerably milder reaction conditions (165 °C and 500 psi H2 pressure). The catalyst was thoroughly characterized by X-ray diffraction (XRD), transmission electron microscopy (TEM), NH3-temperature-programmed desorption (TPD), Brunauer–Emmett–Teller (BET) surface area, etc. Along with this, reaction conditions such as pressure, temperature, and catalyst loading were also optimized to achieve maximum selectivity to CPO. A plausible reaction pathway for furfural hydrogenation to cyclopentanone is also proposed involving structure–activity correlation with the help of different control experiments.

Results and Discussion

In continuation of our previous studies on the role of supports in furfural hydrogenation for product selectivity tuning,[8,11,35,36] here, we could successfully achieve the highest selectivity to CPO using fumed silica as a support for the Pd catalyst.

Catalyst Characterization

Low-angle XRD of 4% Pd/SBA-15 and 4% Pd/hexagonal mesoporous silica (HMS) catalysts in Figure a shows a peak at 2θ = 0.96°, whereas that of 4% Pd/MCM-41 and 4% Pd/MCM-22 shows a peak at 2θ = 2.27°, confirming the presence of the parent ordered hexagonal mesoporous structure.

Figure 1

(a) Small-angle XRD of different Pd-supported silica catalysts. (b) XRD graphs of different Pd-supported silica catalysts.

XRD patterns of different silica-supported Pd catalysts reduced by NaBH4 are shown in Figure b. The appearance of the characteristic broad peak at 2θ = 22.81° corresponding to the (002) plane of silica confirmed its mesoporous nature, which was retained even after impregnation of Pd. The characteristic peaks of Pd were observed at 2θ = 40.05, 46.56, and 68.22° corresponding to the (111), (200), and (220) planes, clearly confirming the face-centered cubic lattice of metallic Pd.[37] Interestingly, Pd(111) peaks observed in MCM-22, MCM-41, and HMS with an intensity lower than that for f-SiO2 and SBA-15 supports suggest the high degree of dispersion of Pd over the support. This ultimately resulted in a more amorphous nature of the catalysts. A similar observation was reported in the literature for Ni-based catalysts.[38] The average crystallite size of Pd nanoparticles (NPs) estimated from the full width at half-maximum of the (111) diffraction peak by applying the Scherrer equation was about 15.6 nm in the case of the Pd/f-SiO2 catalyst, which was much smaller than that of all of the other catalysts prepared with various other supports (Table ), which might be one of the reasons for its better performance. Furthermore, the crystallite sizes of all of the catalysts were found to be in the order of MCM-22 > SBA-15 > HMS > MCM-41 > f-SiO2, which is linearly proportional to the rise in CPO selectivity, discussed later.

Table 1

Texture Properties of Different Silica-Supported Catalysts

sr. no.	catalyst	surface area, m2/g	crystallite size (nm)	acidity, mmol/g	TON	TOF/h–1
1	4Pd/f-SiO2	139	15.6	1.72	105.34	210.6
2	4Pd/SBA-15	147	18.2	3.13	94.65	189.2
3	4Pd/HMS	287	17.9	2.44	74.86	153.6
4	4Pd/MCM-22	197	18.3	3.091	86.25	172.4
5	4Pd/MCM-41	254	16.8	3.27	80.69	161.2

BET surface areas of different silica-supported Pd catalysts are given in Table . Both f-SiO2 and SBA-15 showed a comparatively low surface area of 139 and 147 m2/g, respectively. In the case of the 4% Pd/HMS catalyst, a comparatively high surface area (287 m2/g) was observed, whereas Pd/MCM-22 and Pd/MCM-41 showed moderate surface areas of 197 and 254 m2/g, respectively. The remarkable observation was that although the surface areas of all of the other structured silica-supported Pd catalysts were much higher than that of the fumed-silica-supported Pd catalyst, their activity trend was reverse. Thus, the catalytic activity was governed not only by the surface area but also by other physicochemical properties such as reduced metal species, distribution and strength of acid sites, crystallite size, and morphology.

To evaluate particle sizes and actual distribution of metal particles over the support, the TEM image of the best catalyst, i.e., 4% Pd/f-SiO2, was observed (Figure A). Uniform distribution of Pd NPs was observed, and the size of particles was found to be in the range of 5–13 nm, whereas the morphology of Pd nanoparticles was found to be spherical. The chemical composition and elemental mapping of the 4% Pd/f-SiO2 catalyst analyzed by TEM and high-angle annular dark field (HAADF)-energy dispersive X-ray (EDX) mapping are shown in Figure B. Figure B (i) was considered as a reference image based on which elemental mapping was done. Figure B (ii,iii) showed that Si and O were homogeneously distributed all over the surface with very few dim areas, which might be due to the presence of impregnated Pd particles. Figure B (iv) showed that there was uniform distribution of Pd, with some irregular dim areas all over the sample, which indicates the presence of Si and O. Hence, TEM results (Figure ) revealed that Pd NPs were well distributed over the support without any aggregation, which strongly contributes to its highest activity.

Figure 2

(A) High-resolution (HR)-TEM images of Pd/f-SiO2 with the selected area electron diffraction (SAED) pattern: (i) HR-TEM at low magnification (50 nm), (ii) HR-TEM at high magnification (20 nm), (iii) SAED pattern, and (iv) particle size distribution. (B) HAADF-EDX elemental mapping for 4% Pd/f-SiO2: (i) based drift image, (ii) EDX-map-Si-K, (iii) EDX-map-O-K, and (iv) EDX-map-Pd-K.

X-ray photoelectron spectroscopy (XPS) analysis of the 4% Pd/f-SiO2 catalyst was done to study its chemical composition and electronic states (Figure ). Deconvolution of the 3d spectrum of palladium showed two spin-orbital states, i.e., 3d3/2 and 3d5/2, due to its spin–orbit coupling and was further subdivided into four peaks mainly at binding energies (BEs) of 334.64, 336.64, 339.73, and 342.27 eV. Of them, two peaks observed at 334.64 and 339.73 eV corresponded to the metallic state of Pd with spin-orbital splitting of 5.09 eV, which is in good agreement with the literature report.[39] However, peaks at BEs of 336.64 and 342.27 eV arising due to photoemission of electrons from Pd2+ cations clearly revealed the presence of PdO. It can be observed from XPS data that all of the peaks were shifted to slightly lower binding energies, suggesting the strong interaction between the silica support and palladium. The observed lower shift also suggests the change in charge transfer or relaxation energy from palladium to the silica support.[40,41] Interestingly, no positive shift for Pd5/2 was observed, which clearly suggests that there was no any formation of the Pd–Si alloy over the Pd–Si interface.[40,42] After deconvoluting oxygen O 1s spectra (Figure S1), three peaks were observed with binding energies of 529.53, 530.74, and 531.21 eV. Of these, peaks at BEs of 529.53 and 530.74 eV corresponded to the presence of lattice oxygen and the peak at BE of 531.21 eV suggests the presence of different oxygen defects or is due to the presence of O2–, O– species, which ultimately confirmed the presence of PdO species.[43] Similarly, Si 1s spectra of Pd/SiO2 were also deconvoluted as shown in Figure S2.

Figure 3

XPS spectra of Pd/f-SiO2: palladium (Pd 3d).

The deconvolution of the Pd 3d spectrum (3d3/2 and 3d5/2) of HMS-, SBA-, and MCM-22-supported catalysts confirmed the presence of Pd0 and PdO species, as shown in Figure S3. In the case of 4% Pd/HMS, peaks at BEs of 334.39 and 339.86 eV corresponded to Pd0 and peaks at BEs of 336.73 and 342.13 eV corresponded to PdO. At the same time, for the 4% SBA-15 sample, peaks at BEs of 335.28 and 340.83 eV were attributed to metallic Pd, whereas the single peak at BE of 343.18 eV was attributed to Pd oxide. The MCM-22-supported Pd sample showed peaks at BEs of 335.06 and 339.93 eV corresponding to the metallic Pd, whereas its oxides were completely absent. Interestingly, it showed one distinct satellite peak at BE of 333.36 eV, which might be due to uneven shaking of Pd electrons. To distinguish between the Pd2+ and Pd0 states being formed during activation protocols, the relative intensity ratios of the Pd species from XPS spectra were calculated and are presented in Table S1. It was observed that the Pd2+/Pd0 ratio decreased in the order of Pd/f-SiO2 < Pd/SBA-15 < Pd/MCM-22 < Pd/MCM-22, which directly reflected in their activity performance, particularly, on CPO selectivity, as discussed later, because Lewis acidic sites of PdO play an important role in the rearrangement to yield CPO.

Acidity of the catalyst plays a very crucial role in selective formation of cyclopentanone from furfural. Most of the oxygen-containing groups in silica possess acidity, which could be determined by NH3-TPD analysis, and the results are shown in Figure a,b. All of the catalysts showed large desorption peaks appearing below 150 °C because of the presence of weak acidic sites, indicating weak physisorption of the NH3 molecule. Similarly, all of the catalysts showed a combined broad NH3 desorption peak in the range of 200–700 °C, suggesting the mixture of moderate as well as strong acidic sites. Total acidities of all Pd-supported catalysts were calculated in terms of mmol/g NH3 desorbed and are given in Table .

Figure 4

(a) NH3-TPD of 4% Pd/f-SiO2 catalysts. (b) NH3-TPD of different Pd-supported silica catalysts. (c) Py-IR spectra of Pd-supported silica catalysts.

To confirm the presence of Lewis and Brönsted acidity of all silica supports, pyridine-IR experiments were performed, and the results are given in Figure c. Absorption bands at around 1539–1541 cm–1 in all of the catalysts with different intensities represent Brönsted acidity. In Py-IR spectra of 4% Pd/SiO2 (separately given in Figure c), absorption bands at 1575 and 1627 cm–1 represent the presence of Lewis acidity, whereas small peaks at 1539 cm–1 confirm the presence of Brönsted acidity.[44] The presence of both Lewis and Brönsted acidity plays a key role in rearrangement and dehydration to form cyclopentanone, as discussed later in the reaction mechanism.

Activity Testing

Selective conversion of furfural to cyclopentanone mainly requires a metal function for hydrogenation, Lewis acidity, water as medium, and external H2, and it goes via Piancatelli-type rearrangement.[20−22,45] Therefore, initially, Pd catalysts on different acidic supports were prepared and screened for aqueous-phase hydrogenation of furfural, and the results are given in Table . It was observed that the zirconia-supported catalyst showed almost complete conversion with 40% selectivity toward CPO and 42% selectivity to FAL (Table , entry 1). However, the fumed-silica-supported catalyst achieved remarkable selectivity toward CPO (89%) with 98% FFR conversion, suppressing the formation of other byproducts (Table , entry 2). Furthermore, TiO2- and Al2O3-supported catalysts gave 85 and 76% FFR conversion along with CPO selectivity of 25 and 34%, respectively, with the formation of FAL as a major product (Table , entries 3 and 4). On the other hand, the MMT-KP-30-supported catalyst showed 56% selectivity to THFAL with 90% FFR conversion (Table , entry 5). For all of the acidic supports, selectivity to 2-MF and 2-MeTHF was negligible. Because fumed silica gave the highest performance in terms of the furfural rearrangement product (CPO), further work on the effect of solvent was studied over the 4% Pd/f-SiO2 catalyst. Very surprisingly, in the 2-propanol solvent, the Pd/f-SiO2 catalyst showed almost complete furfural conversion, but maximum selectivity obtained was toward FAL (65%) and THFAL (25%) with no selectivity to CPO and CPL (Table , entry 6). When methanol was used as a solvent, multiproduct selectivities were observed with 90% conversion and only 18% selectivity to CPO (Table , entry 7). Hence, further study was carried out in water as a solvent.

Table 2

Catalyst Screening for Furfural Hydrogenation over Different Acidic Supportsa

			selectivity, %
sr. no.	catalyst	conv., %	FAL	THF	THFAL	2-MF	2-MeTHF	CPO	CPL	other
1	4% Pd/ZrO2	99	42		9	1	2	40	3	2
2	4% Pd/f-SiO2	98	1		8	1	1	89
3	4% Pd/TiO2	85	39	5	28	1	1	25
4	4% Pd/Al2O3	76	56	1	5	2	1	34	1
5	4% Pd/MMT-KP-30	90	29		56	2	1	12
6	4% Pd/f-SiO2*	99	65	2	25	2	1			5
7	4% Pd/f-SiO2#	90	11	10	15	14	21	18	5	6

Reaction conditions: substrate, 5 g; solvent (H2O), 95 mL; loading, 0.5 g; temperature, 165 °C; H2 pressure, 500 psig; agitation speed, 1000; reaction time, 5 h. [*] = 2-propanol; [#] = methanol {others: PeDO’s, furans, etc.}.

Effect of Type of Silica

As the fumed silica support showed CPO selectivity as high as 89%, various other types of silica supports such as MCM-41, MCM-22, SBA-15, and HMS were also screened for the 4% Pd catalyst. However, all of these catalysts showed CPO selectivity lower than that obtained for fumed silica (Table ). For example, with the HMS support, 95% furfural conversion with 48% CPO selectivity was achieved (Table , entry 2), whereas for SBA-15 and MCM-22 supports, comparable selectivities of 30 and 26% to CPO, respectively, were obtained (Table , entries 3 and 4). Interestingly, Pd/MCM-41 showed 52% CPO selectivity with marginally lower FFR conversion of 90% (Table , entry 5). In general, all of these structured silica-supported catalysts showed significant formation of THFAL in the range of 20–30% and almost equal formation of THF and CPL, the products which were completely suppressed in the case of fumed silica support. It is also interesting to note the significant formation of FAL for MCM-22, as against the fumed silica support (∼1%), and FAL selectivity obtained for different structured silica supports was in the order of MCM-22 > SBA-15 > HMS > MCM-41. This can be explained on the basis of specific adsorption of C=C bonds over different Pd catalysts. At the same time, all of the catalysts showed very poor selectivities to 2-MF as well as to 2-MeTHF. The lower selectivity to CPO was due to the formation of resinous products under strongly acidic conditions. Turnover number (TON) and turnover frequency (TOF) values of the respective catalysts were calculated, and it was observed that 4% Pd/f-SiO2 and 4% Pd/SBA-15 catalysts showed highest TON and TOF values as compared with all other silica-supported catalysts (Tables and S2). At the same time, the effect of crystallite size wrt CPO selectivity can be discussed irrespective of its FFR conversion (Figure ). As mentioned above, CPO selectivity decreased as the Pd crystallite size increased. For the lowest crystallite size of 15.6 nm, ∼89% CPO selectivity was observed. With the increase in crystallite size from 15.6 to 18.3 nm, CPO selectivity significantly decreased from 89 to 26%, whereas a subsequent increase in FAL selectivity from 1 to 42% was observed. Thus, selective rearrangement of furfural was preferred over the f-SiO2-supported catalyst only with a smaller crystallite size (15.6 nm) and was limited for other structured silica supports with somewhat larger crystallite sizes. At the same time, manipulation of product selectivities to the rearrangement product (CPO) and the side chain hydrogenation product (FAL) could be achieved as a function of respective silica supports with varying crystallite sizes. Thus, the fumed-silica (f-SiO2)-supported catalyst was the best catalyst for one-pot furfural hydrogenation to CPO.

Table 3

Catalyst Screening for Furfural Hydrogenationa

			selectivity, %
sr. no.	catalyst	conv., %	FAL	THF	THFAL	2-MF	2-MeTHF	CPO	CPL	other
1	4% Pd/f-SiO2	98	1		8	1	1	89
2	4% Pd/HMS	95	12	4	12	2	3	48	5	4
3	4% Pd/SBA-15	97	18	2	30	5	7	30	4	4
4	4% Pd/MCM-22	91	42	2	21	5	2	26	2
5	4% Pd/MCM-41	90	8	1	24	3	1	52	7	4

Reaction conditions: substrate, 5 g; solvent (H2O), 95 mL; loading, 0.5 g; temperature, 165 °C; H2 pressure, 500 psig; agitation speed, 1000; reaction time, 5 h {others: PeDO’s, furans, etc.}.

Figure 5

Crystallite size vs CPO product selectivity in FFR hydrogenation. Reaction conditions: substrate, 5 g; solvent (H2O), 95 mL; catalyst, 4% Pd/f-SiO2; loading, 0.5 g; temperature, 165 °C; H2 pressure, 500 psig; agitation speed, 1000; reaction time, 5 h.

Because 4% Pd/f-SiO2 showed considerably high activity in terms of furfural conversion and CPO selectivity, the Pd loading effect was also studied in the range of 1–5%, and the results are given in Figure . For a lower Pd loading of 1%, FFR conversion achieved was 85% with the maximum selectivity to FAL giving only 35% CPO selectivity and 5% THFAL selectivity. As Pd loading was increased to 2%, conversion reached up to 95% with a drastic increase in CPO selectivity from 35 to 64% at the cost of FAL selectivity from 55 to 26%. The increase in Pd loading to 5% affected the CPO selectivity adversely, reducing it from 89 to 65% with an increase in THFAL selectivity from 8 to 20% and also forming the over-hydrogenation product of CPO, i.e., CPL with 4% selectivity. In addition, a higher metal loading resulted in ring hydrogenation to give ∼5% THF. Therefore, 4% was the optimum Pd loading on the fumed silica support for selective synthesis of CPO (89%).

Figure 6

Effect of Pd metal loading over the f-SiO2 catalyst. Reaction conditions: substrate, 5 g; solvent (H2O), 95 mL; catalyst, Pd/f-SiO2; loading, 0.5 g; temperature, 165 °C; H2 pressure, 500 psig; agitation speed, 1000; reaction time, 5 h {others: 2-MF, 2-MeTHF, THF, PeDO’s, furans, CPL, etc.}.

Furfural conversion and CPO selectivity were also studied as a function of reaction time, and these profiles are shown in Figure . Initially, at the first hour of reaction time, almost half of the substrate was consumed with rapid formation of furfuryl alcohol (69%) along with 29% CPO. As reaction progresses to fifth hour, conversion reached to 98% along with ∼90% selectivity to CPO with the formation of THFAL in a very marginal amount (8%), formed via hydrogenation of the C=C bond of FAL, which was formed at the initial time of reaction. Furthermore, when the reaction progresses to extended hours (10 h), no substantial change in both FFR conversion and CPO selectivity was observed (Table S3, entry 2). For all of the reactions, almost ∼83% mass balance was observed as the balance was for resinous products of furfural.

Figure 7

Conversion and selectivity vs time profiles over the f-SiO2 catalyst. Reaction conditions: substrate, 5 g; solvent (H2O), 95 mL; catalyst, 4% Pd/f-SiO2; loading, 0.5 g; temperature, 165 °C; H2 pressure, 500 psig; agitation speed, 1000; reaction time, 5 h {others: 2-MF, 2-MeTHF, THF, PeDO’s, furans, CPL, etc.}.

Some control experiments were also carried out in which various products of FFR hydrogenation were used as starting materials (Table S3). For the first-step FFR hydrogenation product, FAL as a starting material, CPO selectivity obtained was 78%, whereas (Table S3, entry 3) THFAL did not show any conversion because of its extremely stable nature (Table S3, entry 4). When CPO was also used as the starting material (Table S3, entry 5), selective hydrogenation to CPL was observed with complete CPO conversion. Interestingly, there was hardly any formation of the polymerized product along with negligible selectivity to pentanediol over silica catalysts. In all of the experiments, the formation of intermediate 3-hydroxy-4-cyclopentenone was not identified over GC and gas chromatography–mass spectrometry (GC–MS). The main reason behind this could be its extreme instability at relatively high reaction conditions and which may be readily undergoing reductive rearrangement to form cyclopentanone. The formation of CPO was also evidenced by GC–MS analysis (Figure S4). On the basis of the structure–activity correlation and different control experiments (Table S3), we have elucidated the plausible reaction pathway for selective formation of CPO (Scheme ), which involves (i) initial hydrogenation of furfural to form furfuryl alcohol over metallic Pd sites (Table S3, entry 6), (ii) subsequent attack of the water molecule over the fifth position of the furan ring to form the oxycation, (iii) under acidic conditions, rearrangement of the formed oxycation to 4-hydroxy-cyclopentenone via Piancatelli rearrangement with ring opening and closing and with the loss of one mole of water, (iv) further dehydration of 4-hydroxy-cyclopentenone, catalyzed by Brönsted acidity, giving rise to a very unstable cyclopentadienone intermediate or cyclopentanone intermediate, and (v) further hydrogenation of cyclopentadienone under reductive conditions giving cyclopentanone.

Scheme 2

Plausible Reaction Pathway over the Pd/f-SiO2 Catalyst

Because the 4% Pd/f-SiO2 catalyst showed maximum selectivity to cyclopentanone, further reaction parameter optimization was studied for the same.

Reaction Parameter Optimization

Figure shows the effect of temperature on furfural conversion and CPO selectivity. Obviously, furfural conversion increased from 58 to 99% with an increase in the temperature from 135 to 200 °C. At a lower temperature of 135 °C, 51% selectivity to CPO was observed, which increased to 89% as the temperature increased to 165 °C. However, a further rise in temperature to 200 °C affected the CPO selectivity adversely, which decreased to 76%. At a lower temperature of 135 °C, 30% selectivity to FAL was observed, which became nil with a rise in temperature to 200 °C. At the same time, selectivities to THFAL, 2-MF, and 2-MeTHF remained somewhat constant at all temperatures. Interestingly, 4 and 8% selectivities were observed for the product of over-hydrogenation of CPO, i.e., CPL, at high temperatures of 180 and 200 °C, respectively. A proportional decrease in FAL selectivity resulted in an increase in CPO selectivity to 89% at 165 °C, whereas 76 and 8% for CPO and CPL, respectively, at a higher temperature of 200 °C (Table S3, entry 7).

Figure 8

Effect of temperature on the conversion and selectivity of furfural hydrogenation. Reaction conditions: substrate, 5 g; solvent (H2O), 95 mL; catalyst, 4% Pd/f-SiO2; loading, 0.5 g; H2 pressure, 500 psig; agitation speed, 1000; reaction time, 5 h {others: 2-MF, 2-MeTHF, THF, PeDO’s, furans, CPL, etc.}.

Figure shows the effect of H2 pressure on FFR conversion and selectivity to CPO. Furfural conversion increased from 78 to 99% as the H2 pressure increased from 100 to 700 psig. At a lower pressure of 100 psig, 39% selectivity was observed for FAL along with 57% selectivity for CPO. With a further increase in pressure to 350 psig, a drastic rise in CPO selectivity from 57 to 80% was observed, which further reached 89% at 500 psig. Similarly, at 350 psig, FAL selectivity decreased from 39 to 15%, which again became almost nil with a further increase in H2 pressure to 500 psig. For the highest pressure of 700 psig, CPO selectivity dropped down slightly to 75% and a subsequent increase in over-hydrogenation products, i.e., CPL (5%) and THF (3%), were observed. Significant formation of other byproducts, mainly polymerized products, was observed at a higher H2 pressure (Table S3, entry 8).

Figure 9

Effect of H2 pressure on the conversion and selectivity of furfural hydrogenation. Reaction conditions: substrate, 5 g; solvent (H2O), 95 mL; catalyst, 4% Pd/f-SiO2; loading, 0.5 g; temperature, 165 °C; agitation speed, 1000; reaction time, 5 h {others: 2-MF, 2-MeTHF, THF, PeDO’s, furans, CPL, etc.}.

The effect of catalyst loading was also studied in the range of 0.25–0.7 g, and results are given in Figure . Furfural conversion was found to increase from 70 to 99% with an increase in catalyst loading from 0.25 to 0.7 g. At a lower catalyst loading of 0.25 g, the maximum selectivity was observed for FAL (60%) along with CPO selectivity of 25% with a considerable amount of byproduct formation (10%). On further increasing the catalyst loading to 0.35 g, CPO selectivity increased to 61% with a decrease in FAL selectivity from 60 to 25%. A further increase in catalyst loading (0.7 g) affected the CPO selectivity adversely, which dropped down to 70% along with an increase in THFAL selectivity up to 15%. Therefore, 0.5 g was considered as the optimum catalyst loading for furfural hydrogenation to cyclopentanone. The effect of substrate loading was also studied in the range of 5–10 g, where CPO selectivity decreased to 61% as well as furfural conversion decreased to 51%, as shown in Figure S5.

Figure 10

Effect of catalyst loading on the conversion and selectivity of furfural hydrogenation. Reaction conditions: substrate, 5 g; solvent (H2O), 95 mL; catalyst, 4% Pd/f-SiO2; temperature, 165 °C; H2 pressure, 500 psig; agitation speed, 1000; reaction time, 5 h {others: 2-MF, 2-MeTHF, THF, PeDO’s, furans, CPL, etc.}.

The stability and recyclability of the 4% Pd/f-SiO2 catalyst was confirmed by carrying out the recycle runs under optimized reaction conditions, and the results are shown in Figure . After completion of the first run with the fresh catalyst, the later was allowed to settle down, the reaction crude was separated by decantation, and a new charge was fed for continuing the reaction. Till the second recycle, the catalyst was quite stable and gave almost similar results as the fresh one. For the third recycle, furfural conversion dropped down to 91% along with a decrease in CPO selectivity to 82% with a subsequent increase in FAL selectivity to 11%. The decrease in furfural conversion and CPO selectivity could be due to oligomer formation during the reaction, which might result into deactivation of the catalyst surface or decrease in metallic active sites of Pd. Interestingly, the spent catalyst regained its activity after activating it in the presence of H2. Subsequently, the stability of the Pd/f-SiO2 catalyst was also confirmed by performing the hot filtration test, and the results are shown in Figure S6. In a typical experiment, the reaction was stopped after completion of the first hour. The catalyst was separated out, and the reaction was continued till fifth hour. There was hardly any change in furfural conversion as well as CPO selectivity with no any leaching of active metal during the reaction, confirmed by inductively coupled plasma optical emission spectrometry (ICP-OES) analysis. Later, the actual presence of Pd over the catalysts after subsequent recycle runs was confirmed by EDX analysis (Figure S7).

Figure 11

Catalyst reuse runs. Reaction conditions: substrate, 5 g; solvent (H2O), 95 mL; catalyst, 4% Pd/f-SiO2; loading, 0.5 g; temperature, 165 °C; H2 pressure, 500 psig; agitation speed, 1000; reaction time, 5 h {others: 2-MF, 2-MeTHF, THF, PeDO’s, furans, CPL, etc.}.

To study the effect of successive catalyst recycles on the catalyst composition, XPS analysis of the 4% Pd/f-SiO2 catalyst was done (Figure ). In the used catalyst, two peaks appeared at BEs of 334.87 and 340.69 eV corresponding to metallic Pd, whereas one small peak at 336.41 eV represented the presence of palladium oxide. No significant change in peak positions was observed, but the increase in metallic content and decrease in its oxide during successive catalyst recycling was observed (Table S1) because of in situ reduction under the reaction conditions. After deconvolution of O 1s spectra of the Pd/f-SiO2 catalyst (Figure S8), peaks at 529.94 and 530.90 eV corresponded to the lattice oxygen and surface hydroxyl groups. Interestingly, the peak at BE of 532.46 eV represented the adsorbed C=O bond of adsorbed furfural and cyclopentanone during successive catalyst recycles.[9]

Figure 12

XPS spectra of used Pd/f-SiO2: palladium (Pd 3d).

Conclusions

In the present study, palladium catalysts supported on different silica frameworks were prepared by the standard wet impregnation method keeping 4% metal loading constant and were screened for furfural hydrogenation to achieve the highest selectivity to cyclopentanone. Among all of the Pd-supported catalysts, 4% Pd/f-SiO2 exhibited almost complete furfural conversion with 89% selectivity and with ∼87% yield to CPO. Specific Lewis acidity of the silica (SiO2) support was found to play a crucial role in selective formation of CPO. However, all of the other silica supports tend to give a mixture of products, which may be due to the presence of strong Brönsted acid sites. Even distribution of Pd particles with spherical morphology and with particle size in the range of 5–13 nm was confirmed by HR-TEM. The role of Lewis acidity and water-mediated reaction pathway for furfural to CPO was elucidated by controlled experiments and by XPS and Py-IR characterization of the catalyst. More interestingly, the product selectivity was tuned by varying the different reaction parameters such as temperature, pressure, catalyst loading, etc. The catalyst was also found to be very much stable and can be recycled three times with a marginal change in selectivity. Leaching of catalyst was also cross-checked by the hot filtration test and the ICP-OES technique. Results of some control experiments and structural characterization allowed us to elucidate a reaction pathway for the selective formation of CPO. Hydrogenation of furfural proceeded to give first-step hydrogenation product, furfuryl alcohol, over metallic Pd, followed by the attack of water molecule on the furan ring to form the FAL oxycation. The later rearranges to give 4-hydroxy-cyclopentenone via Piancatelli rearrangement with the loss of one mole of water and its dehydration, followed by hydrogenation, which gives cyclopentanone.

Experimental Section

Materials

Furfural, furfuryl alcohol, tetrahydrofurfuryl alcohols, 2-methylfuran, 2-methyltetrahydrofuran, pentanediols, tetraethyl orthosilicate, cetyltrimethylammoniumbromide, DDS, palladium nitrate, and fumed silica were purchased from Sigma-Aldrich, Bangalore, India. Sodium hydroxide, isopropyl alcohol, and sodium borohydride were purchased from Thomas Baker, India. Hydrogen gas (>99.9% purity) was obtained from Vadilal Chemicals Pvt. Ltd, Mumbai, India.

Preparation of Supports

All catalyst supports, i.e., HMS, MCM-41, SBA-15, and MCM-22, were prepared by the previously reported procedures.[31−34] Fumed silica was used as it is without any chemical treatment.

Catalyst Preparation

All of the catalysts were prepared according to the previously reported procedure.[35] In a typical procedure, for 1 g of catalyst, 85 mg of anhydrous Pd(NO3)2 was first dissolved in 50 mL of deionized (DI) water. To this, a slurry of 0.96 g of fumed silica in water was added slowly with constant stirring, and the whole mixture was stirred for 1 h. NaOH solution (2 M) was then added (∼5 mL) to the above solution till the pH became almost neutral, and the stirring was continued further for 30 min. To the above solution, 0.5 g of NaBH4 was added in small portions, and stirring was continued for next 30 min. The resultant solid catalyst was filtered, washed with DI water, and dried in an oven at 100 °C. A similar procedure was adopted for the preparation of catalysts with different Pd loadings in the range of 1–5% as well as for Pd on different silica supports.

X-ray diffraction measurements were carried out on Shimadzu Labx-6100 series instrument using Cu Kα crystallinity. The presence of Pd loading was confirmed by EDX analysis on a LEO-LEICA STEREOSCAN 440 instrument. HR-TEM and HAADF-STEM images of the catalyst were obtained using a transmission electron microscope, model JEOL 1200 EX. For this purpose, a small amount of the solid sample was sonicated in 2-propanol for 10 min. A drop of the prepared suspension was deposited on a Cu grid coated with a carbon layer, and the grid was then dried at room temperature before analysis. BET surface areas of all of the catalysts were measured using a Micromeritics Chemisorb 2720 instrument. Temperature-programmed desorption of ammonia (NH3-TPD) was also performed on a Micromeritics Chemisorb 2720 instrument. In a typical experiment, 0.05 g of catalyst was taken in a U-shaped, flow-through, quartz sample tube. Prior to measurements, the catalyst was pretreated in He (25 cm3/min) at 200 °C for 2 h. The mixture of NH3 in He (30%) was passed (25 cm3/min) at 50 °C for 1 h. TPD measurements were carried out in the range of 50–700 °C with a heating rate of 10 °C/min. Ammonia concentration in the effluent was monitored with a gold-plated-filament thermal conductivity detector. For determining the Pd content in the filtrate of the leaching experiment, the supernatant liquid was evaporated and the resulting product concentrate was treated with aqua-regia (HNO3/HCl = 1:3) at 60 °C on a sand bath for 2 h and then diluted up to 25 mL with distilled water. This sample was then analyzed by an ICP-OES. Py-IR spectra of all of the prepared catalysts were recorded on a Perkin Elmer 2000 FTIR spectrometer in the range of 4000–400 cm–1 using the KBr pellet technique. XPS analysis was carried out on Thermo Fisher Scientific Instruments. The spectra were excited by a low-power Al Kα X-ray source, and the analyzer was operated in the constant analyzer energy mode. For the individual peak energy regions, a pass energy of 20 eV set across the hemispheres was used. The sample powders were analyzed as pellets without exposing to air. The pressure in the analysis chamber was on the order of 10–8 torr during data collection. The constant charging of the samples was eliminated by referencing all of the energies to the Si 2p set at 103.5 eV. Analyses of the peaks were performed with the software provided by VG.[9b]

Typical Procedure for Catalytic Hydrogenation of Furfural

Catalytic hydrogenation of furfural was performed in a 300 mL stainless steel autoclave (Parr reactor) equipped with an overhead stirrer, a pressure gauge, and an automatic temperature control facility. In a typical experiment, 5 g of furfural, 95 mL of water, and 0.5 g of catalyst were charged into the reactor. It was then sealed and purged with N2 two times to exclude air and then by H2 by a deep tube. After attaining the desired reaction temperature, the reactor was pressurized with hydrogen to the desired value and the reaction was started by stirring the contents at 1000 rpm. The liquid samples were withdrawn from the reactor with every 1 h time interval and subjected for GC analysis, for monitoring the progress of the reaction. H2 consumption as a function of time was also recorded, and the reaction was carried out at a constant pressure by filling it with fresh H2 as per the consumption.

Product Analysis

Samples collected from time to time were analyzed by gas chromatography (Shimadzu GC-2025) using a capillary column of free fatty acid phase (30 m (length) × 0.53 mm (i.d.) × 1 μm (film thickness)) connected to a flame ionization detector. The identification of products was carried out by a comparison of retention times of particular samples. The catalyst was recycled and reused three times to check its activity and effect on the rate of reaction. The performance of different Pd-supported silica catalysts was studied in terms of (%) conversion of furfural and (%) product selectivity as defined below