Selective Hydrodeoxygenation of 5-Hydroxymethylfurfural to 2,5-Dimethylfuran over Ni Supported on Zirconium Phosphate Catalysts.

Crystal α-zirconium phosphate (α-ZrP) was prepared by a hydrothermal method and exfoliated into a layered structure by n-hexylamine (C6H13NH2). Ni-based catalyst (Ni/ZrP) was promoted by loading nickel on the layered α-ZrP via ion exchange. The catalyst was performed to catalyze hydrodeoxygenation of 5-hydroxymethylfurfural (HMF) to 2,5-dimethylfuran (DMF), and a 68.1% yield of DMF and 100% conversion of HMF were achieved at 240 °C, 5 MPa H2, and 20 h. The DMF yield can still retain 52.8% after five cycles. The characteristics of the catalyst were investigated via N2 adsorption-desorption, X-ray diffraction, field emission scanning electron microscopy, high-resolution transmission electron microscopy, pyridine-adsorbed Fourier transform infrared (FTIR) spectra, FTIR spectra, inductively coupled plasma mass spectrometry, and thermogravimetric analysis-mass spectrometry, as well as Raman spectroscopy. A pathway from HMF to DMF was found with MF as the intermediate product, and DMF production was preferable via the -CH2OH group hydrogenolysis of HMF over Lewis acidic sites of Ni/ZrP, which is caused by the zirconium vacant orbits.

Introduction

As the world population grows and the economy expands, only fossil fuels have been unable to meet the class="Species">human survival and development about the huge demand for chemicals and energy. Abundant biomass resources are a promising alternative for the sustainable supply of valuable chemicals (such as alcohols, aldehydes carboxylic acids, and gasoline alkanes) to the chemical industry for the production of drugs, polymeric materials, and fuels.[1−3] As an inexpensive and easily available feedstock, lignocellulosic biomass is the most abundant class among the biomass resources. Cellulosic biomass can be converted into liquid biofuels by thermochemical and hydrolysis routes, and lignocellulose can be transformed into a platform material—5-hydroxymethylfurfural (HMF)—by means of chemical catalysis.[4] As a versatile platform chemical, HMF can form a number of important C6 compounds, such as 2,5-dimethylfuran (DMF), 2,5-furandicarboxylic acid, 5-hydroxymethylfuroic acid, 2,5-dihydroxymethylfuran (DHMF), alkoxymethylfurfurals, and so forth.[5] Among these compounds, DMF (belonging to the second-generation biofuels[6]) was proposed as a promising transportation fuel with higher boiling point, higher energy density, and lower solubility in aqueous solutions compared to ethanol.[7] Román-Leshkov et al. have converted carbohydrates to DMF for use as a liquid transportation fuel earlier.[8] Recently, Zu et al. have utilized Ru/Co3O4 to produce DMF from HMF under relatively mild conditions.[9]

class="Chemical">Zirconium phosphate (ZrP), a kind of layered solid acid,[10] has been developed to be a class of multifunctional materials in recent years.[11,12] α-ZrP (Zr(HPO4)2·H2O) is one of the most important ZrPs that has been prepared by various methods, among which crystalline ZrPs can be prepared by a hydrothermal synthesis method.[13,14] Synthetic α-ZrP can be easily intercalated by amines and then exfoliated to prepare polymer nanocomposites.[14,15] More importantly, the ZrP-type materials have been widely used in biomass catalytic conversion. Some studies have been developed to apply α-ZrP with its acid property to produce biomass-derived platform chemicals, such as levulinic acid,[16] HMF,[17] polyols,[18] and so forth. However, using ZrP as the support to load metal species for hydrodeoxygenation (HDO) of HMF to DMF was rarely reported in recent work. Moreover, many studies about the HDO process are involved in the application of high-price noble metals. The development of non-noble substitutes (base metal) has attracted great attention in both academic and industrial aspects.[19,20] Because of the different functional groups (such as C=O, CH2–OH, and furan ring) in HMF, HDO of HMF to DMF needs C=O hydrogenation to generate the corresponding −OH groups and subsequent C–OH group deoxygenation,[21] and the hydrogenation of C=O bonds over the non-noble metal catalytic sites can promote a selective production of DMF.[22,23]

In this work, we prepclass="Chemical">ared a Lewis acid pan class="Chemical">ZrP-supported Ni catalyst to achieve the aim. Briefly, crystal α-ZrP was prepared through a hydrothermal method and exfoliated into a layered structure by n-hexylamine (C6H13NH2). Ni/ZrP was manufactured by inserting Ni2+ into the layers of amine-intercalated ZrP via ion exchange. After reduction, the Ni/ZrP catalyst was used for catalyzing HDO of HMF to DMF in tetrahydrofuran (THF). The catalyst was investigated by a series of characteristic methods, and the transformation pathway of HMF to DMF was also elucidated.

Results and Discussion

Catalyst Characterization

The data of Bclass="Chemical">runauer–Emmett–Teller (BET) and Barrett–Joyner–Halenda (BJH) pore size distribution of pristine α-ZrP and exfoliated α-ZrP are demonstrated in Figure and Table , respectively. In Figure a, pristine α-ZrP has a BET surface area of 22.5 m2/g and a pore volume of 0.2 cm3/g. In terms of the BJH method, pristine α-ZrP has a pore size of 26.4 nm as displayed in the inset of Figure a. Exfoliation of α-ZrP causes the coexistence of mesopores and macropores, as well as the scattered pore size distribution, which leads to the appearance of artificial pores at around 4 nm (see the inset of Figure b–d).[24] In addition, exfoliated α-ZrP and freshly reduced Ni/ZrP have been changed with the BET surface area of 26.1 m2/g, pore volume of 0.1 cm3/g, and pore diameter of 14.3 nm compared with those of α-ZrP (Table , entries 1 and 2), which is explained by the intercalation of C6H13NH2 into Zr(HPO4)2·H2O layers shown in Figure b,c.

Figure 1

BET isotherms and corresponding BJH plots of (a) pristine α-ZrP, (b) exfoliated α-ZrP, (c) freshly reduced Ni/ZrP, and (d) spent Ni/ZrP.

Table 1

Textural Properties of α-ZrP and Ni/ZrP Based on BET Measurements

entry	sample	surface area (m2/g)	pore volume (cm3/g)	pore size (nm)
1	pristine α-ZrP	22.5	0.17	26.4
2	exfoliated α-ZrP	26.1	0.11	14.3
3	freshly reduced Ni/ZrP	29.3	0.12	16.2
4	spent Ni/ZrP	20.4	0.10	9.1

BET isotherms and corresponding BJH plots of (a) pristine αclass="Chemical">-ZrP, (b) exfoliated αpan class="Chemical">-ZrP, (c) freshly reduced Ni/ZrP, and (d) spent Ni/ZrP.

Scanning electron microscopy (SEM) of pristine αclass="Chemical">-ZrP and transmission electron microscopy (TEM) of class="Chemical">pristine α-ZrP, exfoliated α-ZrP, NiO/ZrP, and freshly reduced Ni/ZrP are shown in Figure a–g, as well as the Ni particle size distribution of freshly reduced Ni/ZrP and spent Ni/ZrP is depicted in Figure h,i. We can see that the average particle size and thickness of the α-ZrP hexagonal nanoplatelets are about 288 and 20 nm, respectively, in Figure a, which demonstrates the overall appearance of pristine α-ZrP. α-ZrP has a crystal structure with a measured interplanar spacing d = 0.35 nm shown in the high-resolution TEM (HRTEM) image of Figure b, belonging to the (001) facet[25] of the hexagonal crystal. The crystal structure is composed of alternating HPO4 tetrahedra and ZrO4 tetrahedra (as the structure unit) to form the layered structure (for the structure unit of layered α-ZrP, see Scheme ).[26,27]

Figure 2

(a) SEM and (b) HRTEM images of pristine α-ZrP; (c) TEM and (d) HRTEM images of exfoliated α-ZrP; (e) TEM images of NiO/ZrP, (f) freshly reduced Ni/ZrP, and (g) spent Ni/ZrP; and particle size distribution of (h) freshly reduced Ni/ZrP and (i) spent Ni/ZrP.

Scheme 1

Structure Unit of Layered α-ZrP

Gray ball: Zr atom, purple ball: P atom, red ball: O atom, and white ball: H atom.

(a) SEM and (b) HRTEM images of pristine αclass="Chemical">-ZrP; (c) TEM and (d) HRTEM images of exfoliated αpan class="Chemical">-ZrP; (e) TEM images of NiO/ZrP, (f) freshly reduced Ni/ZrP, and (g) spent Ni/ZrP; and particle size distribution of (h) freshly reduced Ni/ZrP and (i) spent Ni/ZrP.

Gray ball: class="Chemical">Zr atom, class="Chemical">purple ball: P atom, red ball: O atom, and white ball: H atom.

Compclass="Chemical">aring Figure b with 2c, we can see that the crystal α-ZrP nanoplatelets tend to be transparent and rolled up on the edge of the nanoplatelets, which shows that α-ZrPs are exfoliated by C6H13NH2 successfully. In Figure d, we can observe that very tiny particles of NiO/ZrP are evenly dispersed over the surface of α-ZrP, whereas the Ni particles of freshly reduced Ni/ZrP were aggregated into larger particles with the Ni(111) facet, possessing metal stripes with d = 0.2 nm[28] (see Figure e). Figure f shows the TEM image of Ni/ZrP, from which we can observe that the reduced Ni particles have an average size of about 30.3 nm on the α-ZrP layers.

The elemental mapping images (P, class="Chemical">Zr, and Ni) of the freshly reduced Ni/pan class="Chemical">ZrP catalyst are shown in Figure . We can observe that the P and Zr elements are homogeneously distributed on the ZrP carrier in Figure a,b, respectively, whereas the Ni element is distributed with cluster shapes, which is consistent with the Ni species distribution shown in Figure c.

Figure 3

Elemental mapping images of (a) P, (b) Zr, and (c) Ni in the freshly reduced Ni/ZrP catalyst.

Elemental mapping images of (a) P, (b) class="Chemical">Zr, and (c) Ni in the freshly reduced Ni/pan class="Chemical">ZrP catalyst.

The conventional X-ray diffraction (XRD) patterns of pristine αclass="Chemical">-ZrP, exfoliated α-ZrP, Ni2+/ZrP, NiO/ZrP, freshly reduced Ni/ZrP, and spent Ni/ZrP are given in Figure I(a–f), respectively. In the case of α-ZrP, the peaks appearing at 2θ = 10.5°, 2θ = 20°, and 2θ = 25° are the characteristic peak of α-ZrP, corresponding to the (002), (110), and (112) planes, respectively.[25] After being exfoliated by C6H13NH2, the reflected peaks of (110) and (112) planes have been transferred to 2θ = 7.7° and 11.5° by comparing Figure I(a) and 4I(b). In addition, the (002) planes transferred to 2θ = 3.8° can be observed via the analysis of the small angle (2θ = 0.6°–5°) of pristine α-ZrP and exfoliated α-ZrP (Figure II). According to Bragg’s law, the d of exfoliated α-ZrP at 2θ = 3.8° has been increased to 2.32 nm compared with that of pristine α-ZrP at 2θ = 10.5° with 0.74 nm. In Figure c, the peaks at 2θ = 9.3° and 28.2° show the formation of nickel phosphate hydrate (PDF-#31-0909), which can prove the cation exchange of (C6H13NH3)22+ with Ni2+. Ni(111), Ni(200), and Ni(220) appear at 2θ = 44.6°, 51.4°, and 75.8°, respectively, by comparing Figure I(d) and 4I(e),[29] which illustrates the reduction of Ni oxidation species on the ZrP carrier. Additionally, the Ni particles agglomerate into larger ones with an average particle size of 52 nm (Figure h) after the reaction, facilitating that the XRD intensity of the spent catalyst is enhanced (see Figure If).

Figure 4

(I) Conventional XRD patterns of (a) pristine α-ZrP, (b) exfoliated α-ZrP, (c) Ni2+/ZrP, (d) NiO/ZrP, (e) freshly reduced Ni/ZrP, and (f) spent Ni/ZrP. (II) Small angle of XRD patterns of (a) pristine α-ZrP and (b) exfoliated α-ZrP.

(I) Conventional XRD patterns of (a) pristine αclass="Chemical">-ZrP, (b) exfoliated αpan class="Chemical">-ZrP, (c) Ni2+/ZrP, (d) NiO/ZrP, (e) freshly reduced Ni/ZrP, and (f) spent Ni/ZrP. (II) Small angle of XRD patterns of (a) pristine α-ZrP and (b) exfoliated α-ZrP.

As shown in Figure , the weight of Ni/class="Chemical">ZrP loses 6.7% during the whole heating class="Chemical">process, which is caused by the decomposition of pan class="Chemical">Zr(HPO4)2·H2O. Three strong ion current signals of H2O+ (m/z = 18) at about 310, 450, and 670 °C also prove the fact that H2O is released from the dehydration of P–OH groups in ZrP.

Figure 5

TGA–mass spectrometry (MS) curve of freshly reduced Ni/ZrP under N2 flow with the heating rate of 20 °C/min.

TGA–mass spectrometry (MS) curve of freshly reduced Ni/class="Chemical">ZrP under pan class="Chemical">N2 flow with the heating rate of 20 °C/min.

To understand the change in the acidic sites of the class="Chemical">ZrP-type catalyst, the pyridine-adsorbed Fourier transform infrared (Py-FTIR) spectra and acid density of pristine α-ZrP, exfoliated α-ZrP, freshly reduced Ni/ZrP, and spent Ni/ZrP at different desorption temperatures (40, 150, and 240 °C) are shown in Figure and Table , respectively. In Figure a and Table , entry 1, there are two bands at around 1450 and 1540 cm–1, which represent the Lewis (L) and Brönsted (B) acidic sites, respectively.[30,31] The L and B acidic sites are originated from the vacant orbits of Zr4+-contained framework of four-coordination and P–OH groups, respectively.[32,33] In Figure b and Table , entry 2, the L acid density increases and the B acid density is seriously lost, which are caused by the intercalation of C6H13NH2 into the Zr(HPO4)2 layer. In Figure c,d and Table , entries 3 and 4, the acidity of the Ni/ZrP catalyst is mainly in the form of L acid, along with the L acid density of spent Ni/ZrP seriously lost. The L acid densities of all the samples tend to decrease via exfoliation, ion exchange, calcination, and reduction, and they decrease with the increase of evacuation temperature simultaneously (Table ).

Figure 6

Py-FTIR spectra of (a) pristine α-ZrP, (b) exfoliated α-ZrP, (c) freshly reduced Ni/ZrP, and (d) spent Ni/ZrP after pyridine desorption at (A) 40, (B) 150, and (C) 240 °C.

Table 2

L Acid Density of α-ZrP and Ni/ZrP after Evacuation at Different Desorption Temperatures

entry	sample	t (°C)	L acid density (μmol/g)
1	α-ZrP	40	114.8
		150	63.1
		240	31.9
2	exfoliated α-ZrP	40	478.2
		150	388.8
		240	257.0
3	freshly reduced Ni/ZrP	40	192.8
		150	88.3
		240	59.0
4	spent Ni/ZrP	40	69.2
		150	8.3
		240	0.0

Py-FTIR spectra of (a) pristine αclass="Chemical">-ZrP, (b) exfoliated αpan class="Chemical">-ZrP, (c) freshly reduced Ni/ZrP, and (d) spent Ni/ZrP after pyridine desorption at (A) 40, (B) 150, and (C) 240 °C.

The FTIR spectra of αclass="Chemical">-ZrP, exfoliated α-ZrP, Ni2+/ZrP, NiO/ZrP, and freshly reduced Ni/ZrP are shown in Figure . The peaks at 3594, 3510, and 1620 cm–1 are from H2O symmetric and asymmetric stretching and bending vibrations.[34] In Figure a, the peak at 532 cm–1 ascribes to Zr–O group deformation vibration,[35] and the strong band in the range of 965–1120 cm–1 is characteristic of PO4 group stretching vibrations.[36] In Figure b, the peak at 2960 is assigned to −CH3 stretching vibrations, the peaks at 2920 and 2850 are from −CH2 stretching vibrations, and the band in the range of 1590–1650 cm–1 and the peak at 1202 cm–1 ascribe to −NH and −N–C stretching vibrations from C6H13NH2, which demonstrates the exfoliation of pristine α-ZrP by C6H13NH2. In Figure c, all the peaks associated with C6H13NH3+ disappear, and an absorption band at 1383 cm–1 related to the NO3– residue[37] was observed during the cation exchange of C6H13NH3+ with Ni2+, indicating that almost all C6H13NH3+ is replaced by Ni2+. The NO3– peak becomes broad gradually after catalyst calcination and reduction shown in Figure d,e (for specific process, see Scheme ). There is no P–OH stretching at around 2300 cm–1[38] associated with B acid, showing the fact that the acidity of Ni/ZrP is mainly in the form of L acid.

Figure 7

FTIR spectra of (a) pristine α-ZrP, (b) exfoliated α-ZrP, (c) Ni2+/ZrP, (d) NiO/ZrP, and (e) freshly reduced Ni/ZrP.

Scheme 2

Schematic Diagram of the Catalyst Synthesis Process; (A) Pristine α-ZrP, (B) α-ZrP Exfoliated by C6H13NH2, (C) Ion Exchange of Exfoliated α-ZrP with Ni(NO3)2, and (D) Formation of Ni/ZrP via Calcination and Reduction

FTIR spectra of (a) pristine αclass="Chemical">-ZrP, (b) exfoliated αpan class="Chemical">-ZrP, (c) Ni2+/ZrP, (d) NiO/ZrP, and (e) freshly reduced Ni/ZrP.

Catalytic Performance

The catalyst was tested in the HDO of class="Chemical">HMF to class="Chemical">produce pan class="Chemical">DMF. During the reaction process, DHMF and MF were the main intermediates along with different reaction pathways. Meanwhile, the products mainly contained the O-containing intermediates [e.g., 5-methyl-2-furfurylalcohol (MFA) and 2,5-dihydroxymethyltetrahydrofuran (DHMTHF)] and the over-hydrogenated products such as DMTHF and 2,5-hexanedione (HD) from the furan ring opening of DMF. Figure shows the effect of different reaction parameters in the HDO reaction of HMF, such as (a) temperature, (b) time, (c) pressure, and (d) HMF dosage.

Figure 8

Variation of process parameters of (a) temperature, (b) reaction time, (c) hydrogen pressure, (d) HMF dosage, and (e) HDO of HMF in 5 h. Reaction conditions: (a) catalyst = 0.1 g, HMF = 0.25 g, 5 MPa H2, rotating speed = 500 rpm, and 40 mL THF; (b) 5 MPa H2, 240 °C, catalyst = 0.1 g, HMF = 0.25 g, rotating speed = 500 rpm, and 40 mL; (c) 240 °C, 12 h, catalyst = 0.1 g, HMF = 0.25 g, rotating speed = 500 rpm, and 40 mL THF; (d) 5 MPa H2, 12 h, catalyst = 0.1 g, 240 °C, rotating speed = 500 rpm, and 40 mL THF; and (e) 5 MPa H2, 240 °C, catalyst = 0.1 g, HMF = 0.25 g, rotating speed = 500 rpm, and 40 mL THF.

Vclass="Chemical">ariation of class="Chemical">process class="Chemical">ppan class="Chemical">arameters of (a) temperature, (b) reaction time, (c) hydrogen pressure, (d) HMF dosage, and (e) HDO of HMF in 5 h. Reaction conditions: (a) catalyst = 0.1 g, HMF = 0.25 g, 5 MPa H2, rotating speed = 500 rpm, and 40 mL THF; (b) 5 MPa H2, 240 °C, catalyst = 0.1 g, HMF = 0.25 g, rotating speed = 500 rpm, and 40 mL; (c) 240 °C, 12 h, catalyst = 0.1 g, HMF = 0.25 g, rotating speed = 500 rpm, and 40 mL THF; (d) 5 MPa H2, 12 h, catalyst = 0.1 g, 240 °C, rotating speed = 500 rpm, and 40 mL THF; and (e) 5 MPa H2, 240 °C, catalyst = 0.1 g, HMF = 0.25 g, rotating speed = 500 rpm, and 40 mL THF.

Figure a shows the effect of temperature on the catalytic reaction, and we observed that the class="Chemical">HMF conversion changes from 84.8 to 97.2% as the temperature increases from 180 to 260 °C. DMF reaches a higher yield of 63.9% at 240 °C, accompanied with the intermediate: MF (6.3%), over-hydrogenated products: DHMTHF (6.2%) and tetrahydro-5-methyl-2-furanmethanol (MTHFA) (1.3%), and the furan ring-opening product: HD (4.0%). Among them, the HD yield continues to increase caused by furan opening at higher temperature. However, the DMF yield tends to decrease at 260 °C, which is caused by the production of small molecular alkanes from the cracking of chains on the furan ring[39] and coke formation at the higher temperature. Therefore, too high temperature is not favorable for DMF formation.

Figure b shows the effect of reaction time on class="Chemical">HMF conversion under 5 MPa H2 and 240 °C: HMF is almost converted after 8 h, and its yield reaches a maximum yield of 68.1% at 20 h, followed by MF (8.4%), DHMTHF (5.2%), MTHFA (0.2%), and HD (2.9%) formation. Thus, a reasonable control of temperature and pressure with an appropriately prolonged reaction time can suppress the production of byproducts and intermediate products.

Figure c represents the effect of pressure on class="Chemical">HMF conversion, from which we observe that pan class="Chemical">HMF conversion reaches more than 97% when H2 pressure is more than 2 MPa, and the DMF yield and DHMTHF increase consecutively as the reaction pressure increases. Thus, increasing H2 pressure is conducive for DMF and byproduct (such as DHMTHF) production.

Figure d shows the effect of class="Chemical">HMF dosage on catalytic conversion: pan class="Chemical">HMF is almost converted under all HMF dosages, with 250 mg as the optimal dosage, and DMF gains the highest yield under the corresponding reaction conditions. However, the DMF yield decreases and the MF yield increases with an increase of HMF dosage, which is caused by the insufficient Ni/ZrP catalytic activity in this HDO process.

Conclusively, class="Chemical">DMF reaches a maximum yield of 68.1%, with pan class="Chemical">HMF converted completely under the reaction condition of 5 MPa H2, 240 °C, and 20 h.

To further clclass="Chemical">arify the MF distribution, we conducted an HDO experiment of pan class="Chemical">HMF within the initial 5 h under optimal conditions. In Figure e, the MF yield trend undergoes an increase first before 2.5 h and then decreases after 2.5 h, whereas the DMF yield continues to increase for 5 h, demonstrating that part of MF has been converted to DMF. The other byproducts, such as DHMTHF and HD, retain a low yield as the time goes on, showing the existence of MF as the important intermediate.

Possible Reaction Pathway

The two distinct pathways were associated with the production of class="Chemical">DMF from HMF, and MF is produced as an intermediate in the pathways (see Scheme ).[20,40] Interestingly, MF is detected during all the batch catalytic experiments in our study. Moreover, it possesses a higher proportion and plays a key role in all of the over-hydrogenated products, which also proves HDO reaction pathway from HMF into DMF via MF. Additionally, the detection of a small amount of DHMTHF (over-hydrogenated product) explains another HDO pathway from HMF into DMF via DHMF (as well as MFA, all with low productivity, and the yield cannot be calculated by the minimum detection limit of GC). Meanwhile, the byproducts mainly also contained over-hydrogenated products such as tetrahydro-5-methyl-2-furanmethanol (MTHFM) and DHMTHF and HD generated by DMF furan ring opening and hydration. L acidity of Ni/ZrP favors the mentioned conversion pathway compared with other researchers’ works. For example, Nakagawa and Tomishige have reported that HMF cannot produce MF catalyzed by Ni–Pd/SiO2 without L acid,[41] which may be explained by the fact that the L acid sites on Ni/ZrP would effectively accelerate the cleavage of the −CH2OH bond prior to C=O bond in HMF.[42] In addition, Saha and Abu-Omar have studied the HDO of HMF over Ni/C without L acids, and Ni/C can be accounted for only 10% HMF conversion with the formation of a trace amount of DMF.[43] Therefore, the insight into the conversion from HMF to DMF confirms the reaction pathway proposed by some teams.[20,44]

Scheme 3

General Reaction Pathway for DMF Production from HMF

To corroborate this hypothesis, we cclass="Chemical">arried out the controlled HDO reactions over Ni/pan class="Chemical">ZrP by taking intermediate products (MF and DHMF) as the substrate (see Table ). When MF was used as the substrate, DMF acquires a yield of 42.6%. On the contrary, when DHMF was employed as the substrate, the DMF yield is just 6.9%.

Table 3

Controlled Experiments for DMF Production from Intermediates HDOa

			product yields (%)
entry	substrate	conversion (%)	DMF	DHMTHF	MTHFA	HD	MFA
1	MF	98.8	42.6	3.8	0.5	1.5
2	DHMF	96.3	6.9	3.7	0.2	8.4	0.7

Experimental conditions: 1.98 × 10–3 mol substrate, 0.1 g catalyst, 240 °C, 5 MPa H2, 12 h, rotating speed: 500 rpm, and 40 mL THF.

Experimental conditions: 1.98 × 10–3 mol substrate, 0.1 g catalyst, 240 °C, 5 MPa class="Chemical">H2, 12 h, rotating speed: 500 rpm, and 40 mL THF.

According to the obtained experimental results and reported studies, a brief reaction pathway for the HDO of class="Chemical">HMF to DMF is proposed in Scheme . First, Ni/ZrP is facilely converted into Ni and ZrP with metal and acid functionalities for activating H2 and −CH2OH groups, respectively. The oxygen in the −CH2OH of HMF is activated by the Lewis acidic sites of Zr4+ (originated from the vacant orbit of Zr), which is easily hydrodeoxygenated to MF as the primary intermediate because of attacking of H atoms activated by the neighboring Ni metal. Second, the oxygen in the carbonyl group of MF is activated by the Zr4+ species, followed by hydrogenolysis via MFA. MFA is a precursor to be converted into DMF, which means that HMF was transformed into MFA directly or indirectly during the hydrogenolysis process. Besides the main pathway, the side products such as MTHFA and HD were observed in these cases, which is responsible for the unwanted furan ring hydrogenation of MFA and furan ring opening and hydration of DMF.[45] In terms of these facts, the reaction pathway mentioned via the MF intermediate was further confirmed.

Scheme 4

Proposed Reaction Mechanism for the HDO of HMF to DMF via MF

Catalyst Stability

To investigate the stability of the prepclass="Chemical">ared catalyst, we selected the reaction condition of 240 °C, 12 h, and 5 MPa H2 pressure as our testing object. After each reaction, the catalyst was filtered, washed three times with ethanol (through 0.22 μm Millipore filter), and dried at 60 °C overnight. The collected catalyst was recalcined at 400 °C under air atmosphere for 3 h and reduced at 400 °C for 2 h with a heating rate of 4 °C/min before the next recycling experiments. Figure shows the yield of DMF and Ni content of Ni/ZrP versus Ni/ZrP catalyst usage cycles. It was found that the yield of DMF and Ni content of Ni/ZrP show a positive correlation during the process. We summarize that DMF can retain 52.8% of its original yield with Ni elemental content decreasing from 18.2 to 12% after five consecutive cycles. The continuous leaching Ni species can explain the decreasing activity of Ni/ZrP during the batches of testing experiments. As we know, the Ni species performs a key role in the selective conversion of C=O, C=C, and C–O bonds of HMF, and leaching Ni also breaks the balanced surface metal–acid sites of the catalyst.[46]

Figure 9

Yield of DMF and Ni content (wt %) of Ni/ZrP vs Ni/ZrP recyclability.

Yield of class="Chemical">DMF and Ni content (wt %) of Ni/pan class="Chemical">ZrP vs Ni/ZrP recyclability.

The Raman spectra of the freshly reduced Ni/class="Chemical">ZrP (a) and spent Ni/ZrP (b) are demonstrated in Figure . Comparing Figure a with 10b, we can see that the significant bands at 1375 cm–1 from C–C skeleton modes associated with chemical structures consist of condensed benzene rings and the G peak at around 1590 cm–1 from disordered graphite.[47,48] Therefore, the coke deposition can explain the activity decline of the catalyst to some extent. Besides, the position between 22° and 35° of the peak in the low angle region corresponds to the (002) peak of graphite, caused by the stacking of the graphitic basal planes of char crystallites (see Figure d).[49] As the reaction proceeds, the pore volume and pore size of the catalyst become smaller, thereby resulting in the decreased BET surface area (see Figure d and Table , entry 4).

Figure 10

Comparison of Raman spectroscopy of the (a) freshly reduced Ni/ZrP catalyst and (b) spent Ni/ZrP catalyst.

Compclass="Chemical">arison of Raman spectroscopy of the (a) freshly reduced Ni/pan class="Chemical">ZrP catalyst and (b) spent Ni/ZrP catalyst.

In order to observe the coke deposition more intuitively, the TEM images of the spent Ni/class="Chemical">ZrP before and after recalcination at 400 °C under air atmosphere are shown in Figure . Figure a,b depicts the layers of dark spots, which were resulted from coke deposition, while the spots disappear after the catalyst is recalcined, shown in Figure c,d. Thereupon, Ni/ZrP recalcination we mentioned is an effective way to dispel the coke. In a word, the catalytic activity loss of Ni/ZrP can be summarized as follows: (i) Ni particle size increases; (ii) Ni species is lost, which breaks the balance of metal–acid; and (iii) formation of coke after the catalytic reaction. However, based on Figure , the main deactivation of the Ni/ZrP catalyst is deduced by Ni leaching. The efforts for suppressing Ni leaching to enhance the catalytic stability are in progress.

Figure 11

Spent Ni/ZrP catalyst (a,b) before and (c,d) after re-calcination and reduction treatments.

Spent Ni/class="Chemical">ZrP catalyst (a,b) before and (c,d) after re-calcination and reduction treatments.

Conclusions

Pristine αclass="Chemical">-ZrP was synthesized via a hydrothermal technology, followed by exfoliation with C6H13NH2 to expand the surface area and porosity. Ni/ZrP was prepared by ion exchange of exfoliated α-ZrP and Ni(NO3)2·6H2O, followed by calcination and reduction at elevated temperatures. Ni/ZrP is characterized by a series of techniques, and its catalytic performance was tested in the HDO of HMF to DMF. We proposed a conversion path from HMF to DMF with MF as the main intermediate and found that the synergistic effect of metallic Ni and L acidic sites created from the zirconium vacant orbits on α-ZrP is favorable for the hydrogenolysis of C–OH groups of HMF. Ni/ZrP has a better performance on HDO of HMF to DMF with the target product yield of 68.1% in the first run and presents the stability by retaining its original activity of 52.8% after five runs.

Experimental Section

Materials

All chemical reagents were used directly without further purification. class="Chemical">Dichlorooxozirconium (ZrOCl2·8H2O, AR, ≥98%), C6H13NH2 (AR, ≥99%), THF solvent (AR, ≥99%) and nickel nitrate (Ni(NO3)2·6H2O, AR, ≥98%) were purchased from Shanghai Macklin Biochemical Co., Ltd. Phosphoric acid (H3PO4) solution (85 wt %) was purchased from Tianjin Damao Chemical Reagent Factory. HMF (C6H6O3, GC, ≥95%), DMF (C6H8O, AR, ≥99%), and HD (C6H10O2, AR, ≥97%) were purchased from Aladdin Industrial Corporation. MTHFA (C6H8O2, GC, ≥97%) was purchased from Jiangsu Aikon Biopharmaceutical R&D Co., Ltd. MF (C6H6O2, AR, ≥99%) was bought from J&K Scientific Ltd. DHMTHF (C6H12O3, AR, ≥95%) was purchased from Maya Reagent Co., Ltd. MFA (C6H8O2, AR, ≥97%) was bought from Energy Chemical Co., Ltd.  Ethanol absolute (AR, ≥99.7%) was purchased from Tianjin Yongda Chemical Reagent Co., Ltd.

Catalysts Preparation

Preparation of Pristine α-ZrP

According to the literature,[13] 4.0 g of class="Chemical">ZrOCl2·8pan class="Chemical">H2O was mixed with 40.0 mL (3.0 M) of H3PO4 solution and sealed into a Teflons-lined pressure vessel and heated at 200 °C for 24 h. After the reaction, the solid was filtered with deionized (DI) water through a 0.22 μm Millipore filter (Jinteng, China) until the filtrate was neutral. The product was dried at 60 °C overnight to obtain 2.4 g of pristine α-ZrP.

Exfoliation of Pristine α-ZrP

Referring to the literature,[50] pristine αclass="Chemical">-ZrP was interstratified by pan class="Chemical">C6H13NH2. Briefly, 0.839 g of C6H13NH2 was dissolved in 20 mL of water, and then 1.0 g of solid α-ZrP was added into the mixture. The mixture was stirred for 24 h at room temperature. The solid phase was centrifuged at 10 000 rpm for 10 min (RCF 10610, 10610TG16-WS, Hunan Xiangyi Centrifuge Co. Ltd., China) in DI water twice and then centrifuged in acetone to remove extra water and C6H13NH2. The sample was dried at 60 °C to obtain the exfoliated α-ZrP.

Preparation of NiO/ZrP and Ni/ZrP

A solution was prepclass="Chemical">ared by dissolving 4.5 g of pan class="Chemical">Ni(NO3)2·6H2O in 50 mL of DI water. An amine intercalation sample (0.5 g) was added to this solution, and the mixture was refluxed for 24 h at 80 °C. In this step, Ni2+ was loaded on the layers of α-ZrP via ion exchange. The solid Ni2+/ZrP was recovered from the solution by centrifugation, washed extensively with DI water, and dried at 60 °C overnight. Last, the solid was calcined under air atmosphere at 500 °C for 4 h in a muffle furnace (FO310C, Yamato Scientific Co., Ltd., Tokyo, Japan), denoted as NiO/ZrP.

Prior to the experiment, class="Chemical">NiO/pan class="Chemical">ZrP was reduced in flowing H2 (30 mL/min, purity ≥ 99.9%) at a heating rate of 4 °C/min to 400 °C for 2 h to obtain freshly reduced Ni/ZrP.

XRD (X’Pert-ProMPD, PANalytical Company, Almelo, Netherland) operated at 40 kV and 40 mA with Cu Kα radiation (λ = 0.154 nm) was performed to compclass="Chemical">are the crystallinity of class="Chemical">pristine α-ZrP, exfoliated α-ZrP, NiO/ZrP, freshly reduced Ni/ZrP, and spent Ni/ZrP at room temperature.

The specific surface class="Chemical">area of samples was measured by the BET equation, and the class="Chemical">pore size distribution was investigated by the BJH method depending on N2 adsorption–desorption at −196 °C (IQ-2, Quantachrome Instruments, Boynton Beach, FL).

The stclass="Chemical">ructural and morphological information of class="Chemical">pristine α-ZrP was obtained by using a cold field emission scanning electron microscope (S-4800, Hitachi, Tokyo, Japan). Pristine α-ZrP, exfoliated α-ZrP, and freshly reduced Ni/ZrP were observed by a high-resolution transmission electron microscope (JEM-2100, JEOL Ltd., Tokyo, Japan) equipped with an energy-dispersive X-ray spectrometer (Thermo Scientific UltraDry, Waltham, MA) to compare their microstructure. All samples were dispersed in ethanol absolute and ultrasonicated 10 min for TEM analysis.

Coke deposition on the catalyst was analyzed by a Raman microscope equipped with the 523 nm excitation wavelength laser diode and back-scattering configuration (Lab RAM HR800-LS55, Horiba Jobin-Yvon, Pclass="Chemical">aris, France), and the spectral resolution was about 0.7 cm–1.

The Ni elemental content (wt %) of freshly reduced Ni/class="Chemical">ZrP and used Ni/ZrP was measured by an inductively coupled plasma mass spectrometer (Agilent 7900, Agilent Technologies, Santa Clara, CA). All the samples were digested with 1 mL (30 wt %) of H2O2 and 2 mL (65 wt %) of HNO3 at 100 °C for 12 h and diluted to the desired concentration for test.

The thermal stability and composition of the prepclass="Chemical">ared catalyst are investigated by thermogravimetric analysis (TGA; Linseis STA PT1600, Selb, Germany) under N2 flow (purity ≥ 99.9%) with a heating rate of 20 °C/min up to 1100 °C. H2O released from the catalyst decomposition is monitored via the mass spectrometer (Pfeiffer Omnistar, Asslar, Germany). In order to investigate whether the B and L acidity was present in the catalysts, the Py-FTIR (Nicolet 6700, Nicolet Instrument Company, PA) spectrum at an average of 32 scans and a resolution of 4 cm–1 was obtained to analyze the feature. Sample (10 mg) was activated at 200 °C for 0.5 h under vacuum (10–4 mmHg). The background spectrum was recorded after cooling the sample to 50 °C. Afterward, the sample was exposed to pyridine (Aldrich, GC, ≥99%) vapors for 15 min. The FTIR spectrum for each sample was obtained after pyridine desorption by evacuation for another 0.5 h at 40, 150, and 240 °C, respectively. All the spectra were recorded at room temperature after pyridine desorption at each temperature. The spectra were finally obtained by subtracting the background spectrum previously recorded.

Samples involved in our reseclass="Chemical">arch were chpan class="Chemical">aracterized by an FTIR spectrometer (Nicolet Nexus 10, Thermo Fisher Scientific, Waltham, MA) via the KBr pellet method (samples were analyzed at an average of 32 scans from 400 to 4000 cm–1 at 2 cm–1 resolution).

Catalytic Performance

Freshly reduced Ni/class="Chemical">ZrP (0.1 g) was mixed with 0.25 g (1.98 × 10–3 mol) of HMF in 40 mL of THF solvent, and the reactions with specific reaction time were performed in a 100 mL autoclave (MS-100-C276, Anhui Kemi Machinery Technology Co., Ltd, Hefei, China) under the stirring rate of 500 rpm. Before each run, the autoclave was purged three times with H2 to exclude the air residual in the reactor. The reaction was conducted at a certain temperature, H2 pressure (inflated at room temperature), and time. After the reaction, the reactor was quickly quenched in an ice–water bath. Subsequently, filtration steps (through a 0.22 μm Millipore filter) were conducted, and the reaction solution was collected to analyze the products. The products were quantitatively analyzed by a gas chromatograph (GC-2014C, Shimadzu, Kyoto, Japan) equipped with a flame-ionized detector and a HP-Innowax capillary column (30 m × 0.25 mm, 0.25 μm film thickness, Hewlett Packard, Palo Alto, CA).

The products were qualitatively analyzed by a gas chromatograph–mass spectrometer (Trace 1300-ISQ, Thermo Fisher Scientific Inc., Waltham, MA) with a HP-Innowax column, and their yields were calculated based on the internal standclass="Chemical">ard by using cyclohexanone as the internal standard. The conversion (C mol, %) of HMF and the yield (C mol, %) of product were calculated as follows: