Efficient Dehydration of Fructose to 5-Hydroxy-methylfurfural Catalyzed by Heteropolyacid Salts

Abstract

5-Hydroxymethylfurfural (5-HMF), which is derived from numerous industrial biomass resources, has attracted attention in recent years due to its potential as a building block. In this paper, a range of heteropolyacid salts had been investigated for the dehydration of fructose to 5-HMF. CePW₁₂O₄₀ demonstrated the best catalytic activity. Effects of fructose concentration, reaction temperature and reaction time on 5-HMF yield were investigated and optimised through a central composite design and response surface methodology. The optimal 5-HMF yield was 99.40% under the optimized reaction conditions of 5.48 mg/mL fructose loading, 158 °C temperature and 164 min reaction time. A kinetic analysis of the fructose conversion was also performed, and the activation energy and pre-exponential factor were obtained.

1. Introduction

The massive consumption of unsustainable fossil resources has stimulated the search for renewable resources [1,2]. Biomass has the potential to serve as sustainable resources for the production of fuel and chemicals [3,4]. 5-Hydroxymethylfurfural (5-HMF) is one of the top biomass-derived building blocks from which a wide range of value-added compounds can be synthesised, such as 5-dimethylfuran, 2,5-furandicarboxylic acid, levulinic acid, and so on [5,6,7,8,9]. The dehydration of fructose to 5-HMF has attracted considerable attention in the past few years [10]. Water, organic solvents, organic/water mixtures and a few other biphasic systems have been used as reaction solvents during this process [11,12,13].

Preparation of 5-HMF by chemical catalysis involves mainly dehydration of sugars, which can be efficiently catalyzed by acid catalysts [14]. Mineral acids (H₂SO₄, HCl and HNO₃) and transition metal salts have been widely used in the production of 5-HMF [15,16,17,18,19]. However, the homogeneous methods suffers from complicated separation processes especially considering the regeneration of the system. Therefore, heterogeneous catalysts have been proposed. It’s worthwhile to mention that the heterogeneous acid catalysts have shown superior activity to homogeneous catalysts in terms of 5-HMF selectivity [20]. H-form zeolites and metal phosphates showed 5-HMF selectivities of 60%–90% with fructose conversions in the 30%–60% range [21,22]. Dumesic et al. reported that acidic ion-exchange resin could effectively catalyze the dehydration of 30 wt. % fructose to 5-HMF, resulting into 60% selectivity with 89% fructose conversion [13]. Sulfated zirconia was also demonstrated to be active for the production of 5-HMF with yield as high as 72.8% [23]. Heteropolyacid salts are well-known solid acid catalysts that have been utilised in various important commercial reactions, such as esterifications, isomerisations and etherifications. Shimizu evaluated the activity of heteropolyacid by simple water removal methods and found that Amerlyst-15 and FePW₁₂O₄₀ could catalyze the conversion of high concentration fructose (50 wt. %) to 5-HMF with yields reached approximately 100% and 50%, respectively [24]. In addition, H₃PW₁₂O₄₀ and the corresponding cesium salts were used for dehydration of 50 wt % fructose in a DMSO-polyvinylpyrrolidone system with 52% and 58% 5-HMF yields [25]. Wang et al. demonstrated that Cs_2.5H_0.5PW₁₂O₄₀ exhibited 5-HMF selectivity of 94.7% with a final yield of 74% in a biphsic system of water-methylisobutylketone [26].

In the present study, several types of heteropolyacid salts were synthesised and utilised for the catalytic dehydration of fructose to 5-HMF with optimum yields over 99%. sec-Butanol was demonstrated to be the best reaction solvent. A central composite design (CCD) and response surface methodology (RSM) were employed to design experiments and optimise the conversion. Finally, several kinetic parameters for the reaction were obtained.

2. Results and Discussion

2.1. Effect of Reaction Medium on the Fructose to 5-HMF Conversion Yield

The comparative XRD and IR spectra of the heteropolyacid salts are presented in Figure 1 and Figure 2. In the wavenumber region from 700 to 1100 cm⁻¹, the fingerprint vibrational bands of Keggin anions are observed. The XRD measurement indicated that the crystalline structures of the HPAs still remained in the metal-substituted form. Different heteropolyacid salts and solvents were employed for the dehydration of fructose. The blank references of DMSO, n-butanol and sec-butanol showed that DMSO was the most effective solvent without catalyst. Different catalysts significantly affected the conversion of fructose into 5-HMF (

Table 1. Conversion of fructose to 5-HMF with different solvents and catalysts.

Entry	Solvent	Catalyst	Reaction Time/min	Conv. %	Yield %
1	DMSO	Blank	60	18.11	17.53
2	n-Butanol	Blank	60	1.56	0.58
3	sec-Butanol	Blank	60	2.15	1.04
4	sec-Butanol	K3PW12O40	60	75.85	71.43
5	DMSO	Zn1.5PW12O40	60	27.52	26.22
6	DMSO	Ni1.5PW12O40	60	62.42	60.67
7	DMSO	Cs3PW12O40	60	75.73	73.59
8	DMSO	Cu1.5PW12O40	60	85.68	82.71
9	DMSO	CePW12O40	60	89.61	88.78
10	DMSO	Cu1.5PW12O40	120	93.79	86.22
11	DMSO	CePW12O40	120	96.23	93.86
12	DMAc	CePW12O40	120	91.35	90.13
13	MIBK	CePW12O40	120	88.92	86.74
14	n-Butanol	CePW12O40	120	98.56	97.48
15	sec-Butanol	CePW12O40	120	99.10	98.15

, entries 5–9). Among the catalysts tested, CePW₁₂O₄₀ and Zn_1.5PW₁₂O₄₀ showed the strongest and weakest catalytic activities, affording 88.78% and 26.22% 5-HMF, respectively (Table 1, entries 5 and 9). In addition, the yield of 5-HMF increased as the reaction time was prolonged with only slightly higher conversion of fructose (Table 1, entries 9 and 11).

Different protic (butanols) and aprotic (MIBK, DMSO, etc.) solvents were employed in the dehydration of fructose with CePW₁₂O₄₀ as the catalyst (Table 1, entries 11–15). Protic solvents performed better than the aprotic ones and sec-butanol was determined to be the most efficient solvent (Table 1, entry 15), showing that CePW₁₂O₄₀ exhibited the highest catalytic activity in the fructose conversion among the protic solvents. Moreover, the CePW₁₂O₄₀-sec-butanol combination behaved as a thermoregulated system and there was a phase change from heterogeneous towards homogeneous catalysis during the reaction (Figure 3). After cooling down, the reaction solution turned heterogeneous again and the catalysts could be removed from the reaction mixture by simple filtration.

2.2. Optimization of Process Conditions

The conversion of fructose into 5-HMF catalyzed by CePW₁₂O₄₀ in sec-butanol was optimised by CCD in the RSM. The correlation among fructose loading (X₁), temperature (X₂) and reaction time (X₃) was estimated by ANOVA. The 5-HMF yield was used as the response to evaluate the conversion efficiency and the targets of optimisation.

The results of ANOVA with the 5-HMF yield as the response are shown in

Table 2. ANOVA table for the quadratic model.

Source	Sum of Squares	DF *	Mean Square	F-Value	Probability > F
Model	14047.75	13	1080.60	1275.81	<0.0001	Significant
A	2.62205	1	2.62	3.10	0.1290	-
B	250.6561	1	250.66	295.94	<0.0001	-
C	923.6402	1	923.64	1090.50	<0.0001	-
AB	308.7613	1	308.76	364.54	<0.0001	-
AC	641.8945	1	641.89	757.85	<0.0001	-
BC	790.8265	1	790.83	933.69	<0.0001	-
A2	2237.45	1	2237.45	2641.65	<0.0001	-
B2	5554.388	1	5554.39	6557.81	<0.0001	-
C2	2010.072	1	2010.07	2373.20	<0.0001	-
A2B	33.64851	1	33.65	39.73	0.0007	-
A2C	23.48121	1	23.48	27.72	0.0019	-
AB2	18.70431	1	18.70	22.08	0.0033
A2B2	50.90518	1	50.91	60.10	0.0002	-
Residual	5.081933	6	0.85	-	-	-
Lack of Fit	2.42	1	2.42	4.55	0.0862	Not significant
Pure Error	2.661933	5	0.53	-	-	-
Cor Total	14052.83	19	-	-	-	-

* DF = degree of freedom.

. The Model F-value of 1275.81 implies that the model is significant. The “Model F-Value” could obtain this value with only a 0.01% chance because of noise. Values of “Prob > F” less than 0.05 indicate that the model terms are significant. In this case, B, C, AB, AC, BC, A², B² and C² were the significant model terms. The “Lack of Fit F-value” of 4.55 and “p-value” of 0.0862 (>0.05) imply that the Lack of Fit is not significant. “Adeq Precision” measures the signal-to-noise ratio. A ratio greater than 4 is desirable. The ratio of 107.66 indicates an adequate signal. The results showed that this model was statistically sound and can be utilised to navigate the design space. The final equation in terms of the coded factors can be expressed as:

$$y=98.22-0.68x_1-6.66x_2+12.78x_3-13.65x_1^2-21.51x_2^2-12.94x_3^2-6.21x_1{x}_2-8.96x_1{x}_3-9.94x_2{x}_3+3.19x_1^2{x}_2+2.66x_1^2{x}_3+2.38x_1{x}_2^2+3.99x_1^2{x}_2^2$$

(1)

The similarity between the predicted and actual values is described in Figure 4. The coefficient of determination (R²) value was 0.9001, indicating a close agreement between the experimental and theoretical values.

The 3-D response surface and the 2-D contour plots that show the effects of the variables on the 5-HMF yield are displayed in Figure 5. The interactive effects of independent variables on the responses were obtained by remaining one variable constant and changing the other two variables. The interaction between factors A and B on the 5-HMF yield is shown in Figure 5A. As the fructose loading and temperature increased, the 5-HMF yield rapidly increased, peaked and then decreased when the reaction time was prolonged. This result indicated that the degradation of 5-HMF was significant. Tarry materials were also formed; these materials were likely generated by the polymerisation and cross-polymerisation of 5-HMF and intermediates that readily occur at high temperatures and fructose amount. This finding could account for the decreased yield of 5-HMF. Figs. 5B and 5C also show the presence of two-way interactions between variables, namely, fructose loading and temperature, fructose loading and reaction time as well as temperature and reaction time. Moreover, the sequence of the three factors that affected the 5-HMF yield from strong to weak was C > B > A. This finding is consistent with the results listed in Table 2.

2.3. Optimisation and Confirmation Experiments

The optimal 5-HMF yield was 99.40% under the following reaction conditions: 5.48 mg/mL fructose loading; 158 °C temperature; and 164 min reaction time. Under the same conditions, the actual value (99.60%) obtained from the experiments was within the 95% prediction interval, implying that the prediction models were reasonable and credible.

2.4. Catalyst Recycling

After the reaction, the mixture was distilled and collected to pure sec-butanol. The remaining mixture was extracted with ethyl acetate after the addition of 0.5 mL water according to the procedure reported previously. CePW₁₂O₄₀ and sec-butanol were recycled at least six cycles without significant loss of activity and the yield of 5-HMF was still 93.3% after six runs (Figure 6). The slight decrease in 5-HMF yield could be ascribed to the intensive formation of humins during the reaction; this could be observed from the viscous solutions produced after long reaction times.

2.5. Kinetic Model

Various kinetic analysis of acid-catalysed hydrolysis of carbohydrates have been reported in the literature [27,28]. In the present study, a kinetic analysis of the fructose dehydration catalyzed by CePW₁₂O₄₀ was performed in DMSO so as to make direct comparison with other catalysts used in the same reaction solvent. The kinetic curves shown in Figure 7 correspond to temperatures from 80 to 160 °C. The values of ln(1 − X) (where X is the conversion of fructose) were plotted against the reaction time (t) to obtain the pseudo-first-order rate constant (k). This model has been employed in most of the kinetic studies because it provides reasonable levels of agreement with experimental data [29].

Table 3. Rate constants (k) of fructose conversion at different reaction temperatures with CePW₁₂O₄₀ as the catalyst.

Entry	Temperature (°C)	k (min−1)	Correlation Coefficient
1	80	0.001131 ± 9.1531 × 10−5	0.9963
2	100	0.001761 ± 4.0587 × 10−5	0.9934
3	120	0.002755 ± 3.3841 × 10−4	0.9959
4	140	0.004421 ± 2.4758 × 10−4	0.9872
5	160	0.006069 ± 1.1199 × 10−4	0.9973

shows that the value of k increased when increasing the reaction temperature, indicating that a higher reaction temperature is beneficial for the dehydration process.

An Arrhenius plot was generated by using the values of k. The activation energy and pre-exponential factor for the CePW₁₂O₄₀-catalysed dehydration of fructose to 5-HMF are listed in

Table 4. Kinetic parameters of the degradation of fructose with the CePW₁₂O₄₀ catalyst.

Kinetic Parameters	Value
Reaction order, n	1.0
Activation energy, Ea (kJ/mol)	27.21
Pre-exponential factor, A (min−1)	11.72
Correlation coefficient	0.9869

. The activation energy was 27.21 kJ/mol, which is much lower than the value (55.77 kJ/mo) obtained while using 1-hydroxyethyl-3-methylimidazolium tetrafluoroborate [(C₂OHMIM)BF₄] as the catalyst and DMSO as the solvent [30]. This result indicated that the system involving CePW₁₂O₄₀ efficiently reduced the activation energy and increased the reaction rate, indicating the high catalytic activity of CePW₁₂O₄₀.

3. Experimental Section

3.1. Materials and Catalyst Preparation

5-HMF, cesium carbonate (Cs₂CO₃) and cerous carbonate (Ce₂(CO₃)₃) were purchased from Shanghai Jingchun Chemical Reagent Company (Shanghai, China); sec-butanol, dimethyl sulfoxide (DMSO), dimethylacetamide (DMAc), N-methylpyrrolidone, fructose, zinc nitrate [Zn(NO₃)₂], cupric nitrate [Cu(NO₃)₂] and nickel nitrate [Ni(NO₃)₂] were purchased from Beijing Chemical Reagent Company (Beijing, China). All reagents were utilised as received without further purification.

The heteropolyacid salts were synthesised through the following method: phosphotungstic acid (30.0 g, 9.6 mmol) was dissolved in deionised water (100 mL) followed by the dropwise addition of cesium carbonate (4.7 g, 14.5 mmol). The obtained solution was then stirred for another 2 h at 40 °C under nitrogen atmosphere. Cs₃PW₁₂O₄₀ was obtained after evaporating under vacuum at 60 °C for 6 h. The yield of Cs₃PW₁₂O₄₀ was 90.1%. Other heteropolyacid salts were prepared using similar procedures.

3.2. General Procedure for the Conversion of Fructose to 5-HMF

In a typical reaction, 1.0 g of fructose, 0.1 g of catalyst and 60 mL of solvent were placed in a 100 mL flask fitted with a condenser under a nitrogen atmosphere and heated to the appropriate temperature for the selected amount of time. After the reaction, the mixture was analysed by high-performance liquid chromatography (ProStar 310 with a UV detector, Varian, Palo Alto, CA, USA) equipped with a C18 column using methanol/water (40/60, v/v) as the eluent at a flow rate of 1.0 mL /min. The yield of 5-HMF (%) was evaluated on a carbon basis as:

$$Y=\frac{\text{Moles of }5-\text{ HMF}}{\text{Staeting amount of fructose}}\times 100\%$$

(2)

3.3. Design of Experiments

The effect of three independent variables (temperature, reaction time and fructose loading) on the response (5-HMF yield) was studied by employing a factorial CCD of RSM, which was a collection of mathematical and statistical techniques for designing experiments, analysing the effects of variables, developing models and optimising the process variables to obtain the optimum response [31,32,33,34]. The experimental data were fitted with a low-order polynomial equation to evaluate the effect of each independent variable to the response, which was later analysed to determine the optimum process conditions. In this study, the following polynomial quadratic equation was employed:

$$y={\mathsf{\beta }}_0+{\displaystyle \sum }_{i=1}^3{\mathsf{\beta }}_i{\mathrm{x}}_i+{\displaystyle \sum }_{i=1}^3{\mathsf{\beta }}_{ii}{x}_i^2+{\displaystyle \sum }_{i=1}^3{\displaystyle \sum }_{j=i+1}^3{\mathsf{\beta }}_{ij}{x}_i{x}_j$$

(3)

where y is the response function, x_i and x_j are the independent variables, β₀ is the constant coefficient, β_i is the linear coefficient, β_ii is the quadratic coefficient and β_ij is the interaction coefficient among the variables. For fitting second-order models, CCD is one of the most commonly utilised response surface designs.

Each variable was investigated at three coded levels (−1,0,1), as listed in

Table 5. Range and levels of independent variables.

Variable	Unit	Symbol	Range and Level
			−1	0	1
Fructose loading	mg/mL	x1	5.0	7.5	10
Temperature	°C	x2	140	160	180
Reaction time	min	x3	60	120	180

. The complete experimental design matrix and 5-HMF yields are shown in

Table 6. Experimental design matrix and results.

Run	Experimental Variable			Conv.%	5-HMF Yield (% Mole)
	A (mg/mL)	B (°C)	C (min)
1	5.00	140.00	180.00	85.61	83.45
2	5.00	180.00	60.00	78.23	75.81
3	7.50	160.00	120.00	99.63	99.06
4	7.50	193.64	120.00	30.31	26.17
5	10.00	140.00	60.00	26.87	24.31
6	5.00	140.00	60.00	37.25	35.69
7	10.00	180.00	180.00	93.92	42.10
8	10.00	180.00	60.00	53.45	50.12
9	7.50	160.00	180.00	88.32	79.10
10	5.00	180.00	180.00	95.64	70.15
11	10.00	140.00	180.00	83.59	82.45
12	7.50	140.00	120.00	50.71	47.32

. The data obtained from the CCD were analysed by employing Design-Expert software version 8.0.6 (Stat-Ease Inc., Minneapolis, MN, USA).

4. Conclusions

The conversion of fructose into 5-HMF catalyzed by heteropolyacid salts was investigated. Among the catalysts tested, CePW₁₂O₄₀ showed the highest catalytic activity for the dehydration of fructose to 5-HMF. The optimal 5-HMF yield was 99.40% under the following reaction conditions: 5.48 mg/mL fructose loading; 158 °C temperature and 164 min reaction time. A kinetic analysis was also conducted, and the activation energy and pre-exponential factor for the CePW₁₂O₄₀-catalysed conversion of fructose into 5-HMF were obtained.