The XRD diffraction patterns of the Co- and Pt-based catalysts are shown in
Figure 1. The XRD pattern of ZSM-5 is also presented in the same figure for comparison. The diffraction patterns corresponding to the ZSM-5 structure were still detected on the Co/ZSM-5 catalysts with 1, 5, and 10 wt.% of Co loading, but the peak intensity was gradually decreased. A significant loss of zeolite crystallinity was observed on the 10%Co/ZSM-5 with the appearance of new angles at 2θ = 21.0°, 31.8°, 36.0°, 38.6°, and 47.8°, which are assigned to Co
3O
4 and the other peaks at 2θ = 29.3°, 42.4°, and 56.7° which are assigned to cobalt silicate (Co
2SiO
4) species [
27,
28]. The Pt/ZSM-5 catalysts exhibited similar diffraction patterns of ZSM-5 with a slightly lower intensity compared to those of the Co/ZSM-5 catalysts. There were no differences in the crystalline patterns of ZSM-5 after Pt was added. The crystallinities of all the samples were calculated, and the values are presented in
Table 1. For the calculations, the crystallinity of the ZSM-5 was taken as 100%. As expected, both Co- and Pt-based catalysts showed lower crystallinities than the undoped ZSM-5.
The catalyst textural properties were investigated by N
2 adsorption technique. The BET surface area, pore volume, and pore diameter of all the catalysts are shown in
Table 1. For the Co/ZSM-5 catalysts, the surface area and pore volume decreased with increasing Co loading. The decrease in BET surface area may occur due to blocking of certain zeolite pores by cobalt. On the contrary, the surface area and pore volume of the Pt/ZSM-5 catalysts did not change significantly with increasing Pt loading.
The acidic properties of the catalysts were investigated by NH
3-TPD and the results are displayed in
Figure 2. The curves were deconvoluted into individual peaks by Gaussian deconvolution method, which are shown as dashed lines. All the catalysts showed a similar appearance consisting of two desorption peak regions at (I) 100–350 °C and (II) 350–550 °C. Based on other ZSM-5-based catalysts reported in the literature [
29,
30,
31], these peaks could be inferred to be weak acid sites or physically adsorbed ammonia and strong acidic sites. A semi-quantitative comparison of the acid distribution was expressed as μmol NH
3/g of catalyst with respect to weak and strong acidic sites and summarized in
Table 2. It was found that increasing Co loading led to an increase in weak acid sites, but strong acid sites decreased. The total acidity of Co/ZSM-5 catalysts decreased compared to ZSM-5 support. Lu et al. reported somewhat similar results for Co species insertion on Sn-beta zeolite that exhibited a slight decrease of the density of strong acid sites while the weak acid sites markedly increased [
32]. In the same manner as Co-doped ZSM-5 catalysts, the increase of Pt loading also led to an increase in weak acid sites while strong acid sites decreased and the total acidity of Pt/ZSM-5 catalysts slightly decreased compared to the pristine ZSM-5. This phenomenon indicated the addition of Co and Pt on ZSM-5 catalyst was important to improve the proper acid properties which may have contributed to the conversion of FA to GVL.
The acid sites distribution of Lewis (L) and Brønsted (B) of the synthesized Co- and Pt-doped ZSM-5 catalysts were also identified by in situ FTIR spectra of pyridine adsorption, and the results are shown in
Figure 3. Absorption bands at 1445 cm
−1 were associated with pyridine adsorbed on Lewis acid sites. Absorption bands at 1546 cm
−1 were associated with pyridine adsorbed on Brønsted acid sites, while the 1490 cm
−1 was ascribed to pyridine absorbed on both Lewis and Brønsted acid sites or the hydrogen-bonded pyridine [
33,
34,
35,
36,
37].
Table 2 summarizes the ratio (B/L) of Brønsted to Lewis acids of all the catalysts obtained from the FTIR spectra of pyridine adsorption. It was found that the B/L ratio decreased as Co loading increased, indicating that higher Lewis acid sites correlate with increasing Co loading. When Pt metal was incorporated into the ZSM-5, the B/L ratio of the Pt/ZSM-5 catalysts also decreased, indicating that Lewis acid sites also markedly increased compared to bare ZSM-5. From the NH
3-TPD and pyridine-IR results, the density of weak and Lewis acid sites increased proportionately with an increase in Co and Pt loading on the ZSM-5 catalysts. This result suggested that the GVL yield was correlated to the different characteristics of the acidic properties. The proper B/L ratio would promote the GVL yield. In other words, Brønsted and Lewis acid sites in close proximity have a synergistic effect in catalyzing this cascade reaction [
38].
Catalytic Reaction Study
The catalytic activity and selectivity of ZSM-5 were investigated for furfural conversion at 100, 120, and 140 ºC, 0.1 MPa with N
2 for 5 h. The reaction results are shown in
Table 3. The conversion of furfural at 100 and 120 °C was 51.1% and 66.7% with 19.0% and 35.4% GVL yield, respectively. The increased reaction temperature to 140 °C resulted in the conversion to 75.1% with a yield of 53.3% GVL. The yield of the main product (GVL) increased with increasing reaction temperature. The results are in good agreement with those observed by Winoto et al. that furfural conversion and the yield of GVL were improved with the increase in reaction temperature from 120 to 180 °C [
38]. However, the elevation of reaction temperature decreased in the main product GVL. The carbon balance of reaction temperature at 140 °C was relatively low, which may be caused by the instability of product at an elevated temperature, possibly by forming humins or polymers that cannot be detected by using GC [
39,
40]. According to a previous study by Rao et al. [
18], a decrease in GVL yield was observed at a reaction temperature of 200 °C because of the formation of humin by-products. In our study, the optimum reaction temperature at 120 °C was an appropriate choice for further investigation on the effect of catalyst modification by Co- and Pt-doped ZSM-5 catalysts for the conversion of furfural to GVL.
The catalytic activity and selectivity of Co (1, 5 and 10 wt.%) loaded on ZSM-5 were investigated for furfural conversion at 120 °C, 0.1 MPa with N
2 for 5 h. The reaction results are shown in
Table 4. Typically, ZSM-5 contains high amount of Brønsted acid with a B/L ratio of 1.21. Co-doped ZSM-5 catalysts showed an increase in weak and Lewis acid sites. The combination of Lewis acid and Brønsted acid is required for direct production of GVL from furfural [
1]. The Co/ZSM-5 catalysts with an appropriate qualification can help conversion of the intermediate products (AL, LA and IPL) to GVL. Both furfural conversion, as well as GVL selectivity were markedly increased on the 1%Co/ZSM-5 catalyst, comparing to non-modified ZSM-5. The 1%Co/ZSM-5 catalyst presents the highest conversion and GVL selectivity at 68.0% and 86.0%, respectively. With further increase of Co loading to 10 wt.%, both conversion and selectivity of GVL dropped, due probably to the decrease in BET surface area at high Co loading. The high BET surface area of ZSM-5 may accommodate the diffusion of furfural and GVL. From the catalyst characterization catalysts, weak acidic sites was found to favor the formation of GVL, which is in agreement with the results of Lu et al. [
32]. For the Pt/ZSM-5 catalysts, both furfural conversion and GVL selectivity were markedly increased on the Pt/ZSM-5 catalyst, comparing to the non-modified ZSM-5 and the Co/ZSM-5. The high Brønsted acid sites of ZSM-5 catalyst may be necessary for the conversion of FA/FE to intermediate products such as AL or LA. The increase of weak and Lewis acid sites of the Pt/ZSM-5 catalysts was similar to the Co/ZSM-5 catalysts. Metal-catalyzed or Lewis acid sites promote furfural MPV reduction reaction to FA and LA hydrogenation toward GVL [
23,
32,
41]. In the one-pot conversion of furfural to GVL, a suitable B/L ratio on 1%Pt/ZSM-5 catalyst (B/L ratio 0.66) demonstrated the highest conversion, GVL selectivity, and yield of 88.8%, 96.2%, and 85.4%, respectively. When Pt loading was increased to 2 wt.%, the BET surface area and B/L ratio decreased, resulting in lower catalytic activity than 1 wt.% of Pt loading.
The reaction effects with time for the furfural conversion, as well as GVL selectivity were investigated. The furfural conversion, product selectivity, and yield of GVL as a function of reaction time over 1%Pt/ZSM-5 are shown in
Table 5. At a reaction time of 2 h, furfural conversion (46.5%) was just started towards the 19.2% yield of GVL with LA and IPL selectivity of 3.2% and 23.1%, respectively. After 5 h of reaction, the conversion of furfural was promoted to 88.8% with 85.4% yield of GVL while the LA and IPL formations disappeared. The production rates and TOF of 1%Pt/ZSM-5 catalyst were 0.09 mmol/h and 1.05 h
−1 at 5 h of reaction, respectively. The furfural conversion increased with increasing time and reached to the highest conversion at 10 h with 87.8% of GVL yield.
The reusability test was performed on the 1%Pt/ZSM-5 due to its superior performance, and the results are summarized in
Table 6. To explore the reusability performance, the used 1%Pt/ZSM-5 catalyst was separated by high-speed centrifugation after the reaction, repeatedly washed with 2-propanol, and dried at 100 °C. Prior to the next reaction test, the recovered catalyst was reduced under hydrogen (25 cm
3/min) at 500 °C with a ramp rate of 10 °C/min for 2 h. The activated catalyst was reused under the same reaction conditions for the next three consecutive runs. The GVL yield remained almost unchanged at 83–85% during the four cycles of run, indicating the high stability and reusability of the 1%Pt/ZSM-5 catalyst. The slight deactivation was probably due to the adsorption of intermediate compounds, which may block the pore or active sites.
Many catalysts have been used for GVL production via MPV using various primary and secondary alcohols including transition metal oxides, zeolites, and other noble and non-noble-metal catalysts [
42,
43]. It has been shown that Lewis acidic or metal catalysts such as Pt are effective in GVL production from LA/ levulinate ester. However, the presence of Brønsted acid sites is also necessary when starting from furfural. In order to convert furfural to GVL in a one-pot process, it is therefore highly desirable to develop an efficient multifunctional heterogeneous catalyst. Based on the previous reports [
1,
12,
38,
44,
45,
46,
47,
48], a possible reaction network scheme for the one-pot conversion of furfural to GVL over the multifunctional catalysts (Lewis/Brønsted acid and metal sites) is depicted in
Figure 4. It can be seen that the multiple-step reaction was required to complete the conversion of furfural to GVL by using acid and metal catalyzed reaction. Lewis acid or metal-catalyzed reaction promoted furfural Meerwein–Ponndorf–Verley (MPV) reduction into furfuryl alcohol (FA) via selective hydrogenation of C = O bond of an aldehyde group using 2-propanol as a hydrogen donor. Subsequently, etherification was achieved to form furfuryl ether (FE). Hydrolysis and alcoholysis were catalyzed by Brønsted acid of FE with 2-propanol to LA/IPL or AL. Catalyzed by Lewis acid or metal-catalyzed MPV reaction, the Isopropyl levulinate (IPL) was reduced to Isopropyl 4-hydroxypentanoates, followed by lactonization to form γ -valerolactone (GVL).