CharacterizationTo affirm successful formation of G1-Hal-MET/ZnCl2 and G2-Hal-MET/ZnCl2, their FTIR spectra were recorded and compared with that of Hal. As depicted in Fig. 1A, the main characteristic absorbance bands of pristine Hal are observed at 537 cm−1 (Al–O–Si vibration), 743 cm−1 (stretching vibration of Si–O), 794 cm−1 (symmetric stretching of Si–O), 1030 cm−1 (Si–O stretching), 1104 cm−1 (perpendicular Si–O–Si stretching), 1654 cm−1 (weak stretching and bending vibrations of water), 3697 cm−1 and 3623 cm−1 (internal –OH). All of these characteristic bands are observable in the FTIR spectra of both G1-Hal-MET/ZnCl2 and G2-Hal-MET/ZnCl2, underscoring the fact that the multi-step process used for the introduction of the dendritic moiety on Hal did not demolish its structure. Precise comparison of the FTIR spectra of the catalysts and Hal also implied the appearance of some new bands in the catalysts. In more detail, the absorbance band in 1697 and 2914 cm−1 are related to the –C = N and –CH2 functionalities, which are indicative of conjugation of dendritic functionality. Notably, the stretching band related to Zn–Cl (511 cm−1) overlapped with the that of Hal36.Fig. 1(a) FTIR spectra of Hal, G2-Hal-MET/ZnCl2 and G1-Hal-MET/ZnCl2, (B) XRD patterns Hal and G2-Hal-MET/ZnCl2 and G1-Hal-MET/ZnCl2.XRD was also exploited to explore the crystalline structure of the catalysts and examine the stability of Hal in the course of grafting of dendritic moiety. As displayed in Fig. 1B, Hal characteristic peaks appeared at 2θ = 19.8°, 24.3°, 26.5°, 38.4°, 55.3°, 62.6°, 73.8° and 77.3° (JCPDS No. 29–1487)37,38,39. Upon introduction of dendritic moiety of generation 1, apart from Hal peaks, broadening of the peaks in the range of 2θ = 15.4–24.8° can be observed, which is ascribed to the amorphous dendritic moiety. Similarly, G2-Hal-MET/ZnCl2 XRD pattern exhibited Hal peaks, implying the structural stability of Hal and some broadening, which makes the pattern distinguishable from pristine Hal.The morphology of two as-synthesized catalysts, i.e. G2-Hal-MET/ZnCl2 and G1-Hal-MET/ZnCl2 has been investigated via SEM. The results underlined that in both samples, the Hal tubes were detectable, confirming that Hal instinct morphology has been maintained in the course of conjugation of dendrimer. Comparison of the SEM images of the two catalysts, Fig. 2A and B, underscored that G2-Hal-MET/ZnCl2 exhibited more compact morphology and more aggregates are observable in this sample compared to G1-Hal-MET/ZnCl2. Indeed, as in G2-Hal-MET/ZnCl2 the dendritic moiety is of generation two, it can form more amorphous aggregates.Fig. 2SEM images of (A) G1-Hal-MET/ZnCl2 and (B) G2-Hal-MET/ZnCl2 and (C) EDS analysis of G2-Hal-MET/ZnCl2 and G1-Hal-MET/ZnCl2.EDS analysis of G1-Hal-MET/ZnCl2 and G2-Hal-MET/ZnCl2 was also conducted to approve incorporation of the dendritic moiety. As displayed in Fig. 2C, both catalysts showed the presence of carbon, oxygen, nitrogen, chlorine, aluminum, silicon, zinc and sulfur. Among the aforementioned elements, aluminum, silicon and oxygen are representative of Hal, while the presence of carbon, sulfur, oxygen and nitrogen is a proof for incorporation of the dendritic moiety. Observation of chlorine and zinc atoms is also indicative of formation of Lewis acid. Although EDS analysis is a semi-quantitative one, comparison of the weight and atomic percents of the present atoms in G1-Hal-MET/ZnCl2 and G2-Hal-MET/ZnCl2 clearly underlined that these values for carbon, nitrogen, chlorine, zinc and sulfur, which can be indicative of functional moiety on Hal are higher for G2-Hal-MET/ZnCl2, confirming formation of generation II of dendrimer in this catalyst. Dispersion of the elements present in G2-Hal-MET/ZnCl2 and G1-Hal-MET/ZnCl2 was studied by elemental mapping analysis, Fig. 3. According to the results, the elements related to dendritic moiety have been dispersed uniformly. Similarly, the results approved homogeneous dispersion of chlorine and zinc elements in both catalysts.Fig. 3Elemental mapping analysis of (A) G1-Hal-MET/ZnCl2 and (B) G2-Hal-MET/ZnCl2.TG analysis was also exploited to confirm grafting of the dendritic moiety on Hal. As the comparison of thermograms of G1-Hal-MET/ZnCl2 and G2-Hal-MET/ZnCl2 with that of Hal shows, Fig. 4, both as-prepared catalysts showed lower thermal stability than pristine Hal, which is a proof for the presence of less-thermally stable organic moiety in those samples. In fact, in both G1-Hal-MET/ZnCl2 and G2-Hal-MET/ZnCl2, apart from the weight losses of Hal, which are detected at 150 °C and 500 °C additional weight loss stage at 270–400 °C was observed that is due to the decomposition of dendritic moiety. More accurately, in G1-Hal-MET/ZnCl2 an additional weight loss (14 wt%) was observed in the range of 250–400 °C, which is due to the decomposition of organic moiety (melamine, epoxy and vitamin B1). In the case of G2-Hal-MET/ZnCl2, more weight loss steps were detected, Fig. 4. In this thermogram, the weight loss at 250 °C is attributed to the decomposition of melamine and epoxy group, while the one observed in 350 and 470 °C can be ascribed to the decomposition of second generation of dendritic moiety. It is also worth noting that comparison of thermograms of G1-Hal-MET/ZnCl2 and G2-Hal-MET/ZnCl2 implied that this weight loss in G2-Hal-MET/ZnCl2 is more pronounced and corresponded 33 wt%, while in G1-Hal-MET/ZnCl2 this value is only 14 wt%. This outcome approves successful formation of the second generation of dendrimer in G2-Hal-MET/ZnCl2.Fig. 4Right: Thermograms of Hal, G2-Hal-MET/ZnCl2 and G1-Hal-MET/ZnCl2 and DTG curve of G1-Hal-MET/ZnCl2.The XPS plot (Fig. 5a) clearly shows the presence of peaks associated with carbon, chlorine, nitrogen, oxygen, silicon, aluminum, sulfur, and zinc. The C1s spectrum depicted in (Fig. 5b) reveals binding energies of 284.4 eV (corresponding to C–N and C–C), 285.5 eV (corresponding to C–O and C=N), and 286.1 eV (corresponding to C-N). A high-resolution O 1 s profile, Fig. 5c was deconvoluted into the peaks at 531.8, 532.5, and 531.1 eV, which are related to O 1s, SiO2, and Al2O3. Deconvolution of Cl 2p, Fig. 5d, resulted in peaks at 198.7, 198.8, 198.9, 200.5 eV which are related to Cl 2p, Cl 2p3 and Zn–Cl, Cl 2p1. According to the deconvoluted high-resolution S 2p (Fig. 5e), the peaks at 165.1, 166.6, and 164.8 eV that are related to S, SO, have been detected. The deconvoluted high-resolution Si 2p spectrum, Fig. 5f, showed the presence of peaks at 99.3 eV, 102.6 eV, and 103.3 eV, which indicate Si, SiO, and SiO2. High-resolution Al 2p spectrum, Fig. 5g, can be deconvoluted into the peaks at 72.9 eV, 72.7 eV, 74.3 eV, and 74.4 eV, which are representative of Al (2p1, 2p3), Al2O3, Al2O3-n H2O, and Al2SiO5 respectively. High-resolution N 1s, Fig. 5h, can be deconvoluted into the peaks at 398.4, 397.1, 397.6, 400.2, and 403.2 eV. Deconvoluted high-resolution Zn 2p (Fig. 5I), into the peaks at 1021.8, 1021.7, 1044.8 eV, which are related to Zn 2p1.Fig. 5(a) XPS analysis of G2-Hal-MET/ZnCl2 (b) spectrum of C 1 s, (c) spectrum of O1s, (d) spectrum of Cl 2p, (e) spectrum of S 2p, (f) spectrum of Si 2p, (g) spectrum of Al 2p, (h) spectrum of N 1 s, and (I) spectrum of Zn, 2p.Considering the importance of textural properties of the catalyst, specific surface area (SBET) and total pore volume (Vp) of G2-Hal-MET/ZnCl2 were also measured. Moreover, to shed light into the effect of functionalization of these properties, SBET and Vp of the catalyst were compared with those of pristine Hal. According to the results, the specific surface area of Hal was 52 m2 g−1. This value for G2-Hal-MET/ZnCl2 was ~ 10 m2 g−1, implying coverage of outer surface of Hal with dendritic moiety. Similarly, Vp value of Hal (0.12 cm3 g−1) decreased to 0.09 cm3 g−1 upon introduction of dendritic moiety. Total acidity of G2-Hal-MET/ZnCl2 was also measured using NH3-TPD as 7.34 mmol/g cat.Comparison of the catalytic activity of G2-Hal-MET/ZnCl2 and G1-Hal-MET/ZnCl2
In this study, two catalysts, G1-Hal-MET/ZnCl2 and G2-Hal-MET/ZnCl2 have been prepared, in which the generation of the dendritic moiety was different. In fact, it was believed that in G2-Hal-MET/ZnCl2 that the number of vitamin B1 hydrochloride functionality is higher, the chance of conjugation of ZnCl2, which acts as a Lewis acid is higher and consequently, this catalyst exhibits superior catalytic activity compared to G1-Hal-MET/ZnCl2. To verify this assumption, the catalytic activity of both catalysts for dehydration of fructose to HMF at 100 °C in DMSO as solvent using 0.03 g of catalyst was examined and compared. The outcomes indicated that under the aforesaid reaction conditions, G1-Hal-MET/ZnCl2 led to the formation of HMF in 80% yield, while, G2-Hal-MET/ZnCl2 resulted in 89% HMF, confirming the superiority of G2-Hal-MET/ZnCl2. This issue is due to the presence of higher acidic sites on G2-Hal-MET/ZnCl2.Optimization of the reaction conditionsAs RSM is a precise method that not only can consider the effects of reaction parameter, but also the interactions among them, it was selected to optimize the reaction variables, including catalyst loading, reaction temperature and time have been optimized to achieve the highest HMF yield. To this aim, some initial experiments have been conducted to obtain the suitable range of each variable, Table 1. Then, using a quadratic model the statistic studies was performed. The outcomes of analysis of variance (ANOVA) are summarized in Table 2. RSM calculations also furnished Eq. (1), which includes various parameters with positive and negative coefficients, which indicate their synergistic or antagonistic effects on the HMF yield. More precisely, A, AC and BC parameters with positive coefficients have synergistic effects and ultimately increase the HMF yield, while B, C, AB, A2, B2 and C2 parameters with negative coefficients have antagonistic effects and decrease the yield of HMF.Table 1 The initial tests performed for finding the range of reaction variables for RSM.Table 2 ANOVA results of response surface method using a quadratic model.Furthermore, the coefficients implies that the order of the magnitude of the influence of the parameters are as follow:C2 > BC > AC > AB > B2 > C > B > A > A2.$$ {\text{HMF}}\;{\text{Yield}}\;{\text{(\% )}} = + {95}.{76} + 0.{6}0{44}\;{\text{A}} – 0.{8}00{6}\;{\text{B}} – {2}.{78}\;{\text{C}} – {4}.{63}\;{\text{AB}} + {9}.{2}0\;{\text{AC}} + {1}0.{13}\;{\text{BC}} – 0.{1}0{3}0{\text{A}}^{2} – {6}.{\text{47B}}^{2} – {19}.{\text{38C}}^{2} $$
(1)
A: Temperature, B: Time, C: Catalyst amountBy looking at the coefficients in Eq. (1), it can be seen that certain terms like C2, BC, AC, and AB have a greater impact compared to terms like A2 and other powers. For instance, the coefficient for C2 being − 19.38 is the most negative among all coefficients, indicating a significant influence of this variable in the model. This means that changes in the value of C2 lead to substantial variations in the final result, in this case, the HMF yield (%). Similarly, observing larger values for coefficients of variables like BC, AC, and AB highlights the high importance of these variables in affecting the model’s outcome. On the other hand, the coefficient for A2, being − 0.1030 and having the smallest negative value, suggests less pronounced impact of A2 compared to other variables.According to the results, the Model F-value was 22.49, which implies that the model is significant. There is only a 0.01% chance that a F-value of this large could occur due to noise. Also, P-values less than 0.0500 indicates that the model terms are significant. In this case, AC, BC, B2, and C2 are significant model terms. Values greater than 0.1000 indicate the model terms are not significant. The Lack of Fit F-value of 1.90 implies that the Lack of Fit is not significant relative to the pure error.The R2 value indicates the degree of correlation between the model predictions and the actual results. Here, R2 has a value of 0.910, indicating a good fit between the model predictions and the actual results. Additionally, the Adjusted R2 and Predicted R2 values are 0.90 and 0.728, respectively. Additionally, the 3D graphs (Figs. 6) illustrate the effects of various reaction parameters (temperature, time, and catalyst loading) on the HMF yield. In more detail, in (Fig. 6A), 3D surface plot of the interaction between temperature and time shows that the maximum HMF yield is achieved at 105 °C and 95 min.Fig. 6(A) 3D surface plot of the interaction between temperature and time amount for HMF yield, (B) 3D surface plot of the interaction between temperature and catalyst amount for HMF yield and (C) A 3D surface plot illustrating the interaction between reaction time and catalyst.The results obtained from the 3D surface plot in (Fig. 6B), which illustrates the interaction between temperature and catalyst loading on HMF yield, suggest that increasing the catalyst loading up to 0.035 g resulted in the highest HMF yield. Moreover, the optimal reaction temperature was found to be 105 °C.The 3D surface plot in (Fig. 6C) shows the effect of varying reaction times and catalyst amounts on the HMF yield. As the reaction time increases, the HMF yield increases up to a certain point and then reaches a plateau. Similarly, increasing the catalyst amount initially leads to an increase in HMF yield, but beyond a certain point, further increase in catalyst amount does not result in a significant increase in HMF yield. According to the RSM results, the optimal reaction conditions for maximum HMF yield were achieved using 0.035 g catalyst, at a temperature of 105 °C, and a reaction time of 95 min.Recyclability of the catalystAs recyclability of a heterogeneous catalyst can impact its applicability and potential large-scale use, it is imperative to study this feature of the catalyst. The results of the examination of the recyclability of catalyst for fructose dehydration under the optimal reaction conditions are presented in Fig. 7A. Gratifyingly, the results implied that the catalyst maintained its activity for the second run of the reaction and for the third and fourth runs of the reaction only 1% loss of the catalytic activity was detected. Moreover, recycling of the catalyst for fifth run led to only 3% decrement of HMF yield, which underlined high recyclability of the catalyst. Noteworthy, recycling of the catalyst for sixth and seventh runs led to more pronounced loss of the activity and HMF was achieved in 75% yield upon seventh run of the reaction. The structural stability of the recycled catalyst was studied by recording its FTIR spectrum and comparing it with that of fresh one. As displayed in Fig. 7B, the two spectra are very similar and the spectrum of the recycled catalyst exhibited all of the characteristic bands of the catalyst.Fig. 7(A) Recyclability of the catalyst for the dehydration of fructose to HMF under optimum reaction condition, (B) Comparison of FTIR spectra of fresh and recycled catalysts and (C) The result of hot filtration test of G2-Hal-MET/ZnCl2.Hot filtration testIn heterogeneous catalysis, two distinct routs can be conceived. In the first rout, the catalytic active species remained heterogeneous during the reaction, while in the second rout, leaching of catalytic species in the reaction media and their re-deposition on the catalyst is probable. To distinguish the nature of catalysis, hot filtration test is mainly applied. In this test, the reaction is halted after a short period of time and then the catalyst is removed and the reaction continued in the absence of the catalyst. It is expected that in the case of true heterogeneous catalysis, no improvement of reaction yield occur upon catalyst removal, while in other case, the presence of the leached species can promote the reaction even after its separation. Gratifyingly, the results of hot filtration of G2-Hal-MET/ZnCl2, Fig. 7C, for dehydration of fructose to HMF under the optimal conditions underlined the true heterogeneous nature of the catalysis.Kinetic studyBased on the literature, conversion of fructose to HMF occurs via a first-order process40,41 and the reaction rate constant, k, value and is dependent on temperature. The rate of reaction is calculated using Eq. (2).42 In this equation, “r” is the reaction speed, which is the equal to the change in the reactant or product relative to time.$$ – r\left[ {fructose} \right] = + \frac{{d\left[ {HMF} \right]}}{dt} = – \frac{{d\left[ {fructose} \right]}}{dt} = + K\left[ {fructose} \right] $$
(2)
Using X as a conversion rate, Eq. (3) can be achieved.$$ \left[ {fructose} \right] = \left[ {fructose} \right]_{0} \times \left( {{1} – {\text{X}}} \right) $$
(3)
Furthermore, combination of the so-called equations can give Eq. (4).To calculate k value, the target reaction was conducted at four different temperatures, 90, 94, 98 and 100 °C, and then the plot of − Ln(1-X) vs time was obtained, in which k can be estimated from the slope of the line, Figure S1.$$ – \ln \left( {1 – X} \right) = kt + C $$
(4)
The results in Figure S2 underlined that there is a linear relationship between k value and temperature and upon increment of reaction temperature, k value increased. This value for temperature of 90, 94, 98, and 100 °C was obtained as 89.1.33, 99.1.1, and 23.2, respectively.Having k value in hand, activation energy, Ea, was also estimated using Arrhenius Equation, Eq. (5). In this equation R stands for universal gas constant that is 8.314 J/(mol K). As displayed in Figure, S3 plotting of lnk vs 1000/T(K) leads to the line that its slope is equal to − Ea/R. According to the calculations, Ea for the dehydration of fructose to HMF was 22.85 kJ/mol, indicating that using G2-Hal-MET/ZnCl2 as a catalyst, the activation energy barrier was relatively low and conversion of fructose could proceed at moderate temperatures, which is attractive from industrial viewpoint.$$ lnk = – \frac{Ea}{{RT}} + lnA $$
(5)
Thermodynamic studyTo calculate thermodynamic parameters, i.e. activation enthalpy \(\Delta {\text{H}}^{ \ne }\), activation entropy \(\Delta {\text{S}}^{ \ne }\), and activation Gibbs energy \(\Delta {\text{G}}^{ \ne }\), Eqs. (6 and 7) were employed.In Eq. (6), kb and h represent the Boltzmann constant (1.38 × \(10^{ – 23} {\text{j}}.{\text{K}}^{ – 1}\)) and Planck’s constant (6.626 × \(10^{ – 36} {\text{ j}}.{\text{s}}\)) respectively43.$$ {\text{k}} = \frac{{{\text{k}}_{{\text{b}}} {\text{T}}}}{{\text{h}}}{\text{e}}^{{\frac{{ – \Delta {\text{G}}^{ \ne } }}{{{\text{RT}}}}}} $$
(6)
$$ \Delta {\text{G}}^{ \ne } = \Delta {\text{H}}^{ \ne } – {\text{T}}\Delta {\text{S}}^{ \ne } $$
(7)
Combining Eqs. (6 and 7, Eq. 8) can be derived. Using this new equation and plotting Ln k/T vs 1000/T (Figure S3), The \({ }\Delta {\text{H}}^{ \ne }\) (kJ/mol) and \(\Delta {\text{S}}^{ \ne }\) (J/mol) were estimated from the slope \(\left( { – \frac{{\Delta {\text{H}}^{ \ne } }}{{\text{R}}}} \right)\) and the intercept \(\left( {{\text{Ln}}\frac{{{\text{k}}_{{\text{b}}} }}{{\text{h}}} + \frac{{\Delta {\text{S}}^{ \ne } }}{{\text{R}}}} \right)\) of the line respectively43. Having \(\Delta {\text{H}}^{ \ne }\) and \(\Delta {\text{S}}^{ \ne }\), \(\Delta {\text{G}}^{ \ne }\), was calculated using Eq. (7) as 90.92 kJ/mol.$$ {\text{Ln}}\frac{{\text{k}}}{{\text{T}}} = \left( { – \frac{{\Delta {\text{H}}^{ \ne } }}{{\text{R}}}} \right)\frac{1}{{\text{T}}} + \left( {{\text{ln}}\frac{{{\text{k}}_{{\text{b}}} }}{{\text{h}}} + \frac{{\Delta {\text{S}}^{ \ne } }}{{\text{R}}}} \right) $$
(8)
Comparative studyDehydration of fructose to HMF is such an important chemical process that has been focused by many research groups and multifarious catalysts have been promoted for this catalysts. To elucidate whether the performance of G2-Hal-MET/ZnCl2 is comparable with previously developed catalysts, its activity under the obtained optimal conditions was compared with some of the randomly selected catalysts, listed in Table 3. As displayed, a broad range of catalysts from metal–organic frameworks (MOF) to heteropolyacids has been applied for this conversion. Notably, compared to G2-Hal-MET/ZnCl2, some catalysts, such as SBA-SO3H and Si-3-IL-HSO4 led to moderate HMF yields. The comparison of the reaction yields also indicated that the catalytic activity of G2-Hal-MET/ZnCl2 is superior to some other catalysts, such as Mesoporous TiO2, Fe3O4@SiO2-SO3H and PS-Tet-SO3H. G2-Hal-MET/ZnCl2 also exhibited comparable catalytic activity compared to some other dendritic catalysts, such as SO3H-dendrimer-SiO2@Fe3O4 and Cell-G3-SO3H. It is worth noting that this comparison cannot be deemed as an accurate comparison due to the difference of the reaction conditions and it is not claimed that G2-Hal-MET/ZnCl2 is the best catalyst for the conversion of fructose to HMF, however, this comparison just can give a sense of efficiency of this catalyst.Table 3 Comparison of the activity of some catalysts for conversion of fructose to HMF.Reaction mechanismThe plausible mechanism of dehydration of fructose to 5-HMF is depicted in Fig. 8. Initially, adsorption of fructose on the surface of G2-Hal-MET/ZnCl2 through non-covalent interactions brings it close to the main catalytic sites. As illustrated, the reaction proceeds by activation of fructose through coordination of G2-Hal-MET/ZnCl2 to remove water and form an enolic intermediate, which then tautomerizes to form more stable keto-form. Afterwards, loss of the second molecule of water occurs to give 5-HMF.Fig. 8Plausible reaction mechanism.
Halloysite functionalized with dendritic moiety containing vitamin B1 hydrochloride as a bio-based catalyst for the synthesis of 5-hydroxymthylfurfural
