Introduction

Polyoxometalates (POMs), soluble anionic metal oxide clusters, are comprised of early-transition metal with d0 or d1 electronic configurations (usually MoVI/V, WVI, VV, NbV or TaV), which have attracted intense attention during the last two decades not only for their remarkable structural and electronic properties, but also for their intriguing applications ranging from catalysis to medicine1,2,3,4,5,6,7,8. The metal oxide clusters entirely consisted of late transition metal ions with d10 electronic configuration are rarely available owing to the instability of extremely high negative charge, though a series of addenda atoms substituted by late transition metal ions (such as FeIII, CoII, CuII, MnII and ZnII) were commonly reported9,10,11. Compared with conventional POM clusters, d10 metal-oxygen clusters may have some special properties resulting from the inherent characteristics of late transition metals. In this regard, how to lower the overall negative charge to obtain d10 metal-oxygen clusters is a significant and challenging task in synthetic chemistry and material sicence12,13.

Biodiesel, a renewable source of energy, seems to be an attractive candidate for the solution of global energy demands due to its environmental and sustainable benefits14,15. However, biodiesel is currently not competitive with the conventional diesel fuel because of its high costs of raw material and production processing16,17,18. That is, the presence of free fatty acids (FFAs) and water in low quality feedstocks would result in two undesirable side reactions, saponification and triglyceride hydrolysis, making the production process complicated. One way to overcome this problem is to subject FFAs to esterification before the transesterification of triglycerides. Thus, esterification is an essential step in the production of biodiesel and the search for highly efficient catalysts becomes greatly desirable in esterification reaction. In this context, heterogeneous catalysts, which are insoluble, easily separated from products and reusable, exhibit important advantages with respect to homogeneous catalysts and become particularly important19,20,21,22. POM-based materials with Lewis and Brønsted acidity and reversible redox activity have been utilized as efficient solid acid and electron transfer catalysts for the production of industrially important chemicals since the late 1970s23,24,25,26,27,28. Recently, POM-based materials were also used as acid catalysts for biodiesel production29,30,31,32. In most cases, Lewis acidic POM-based catalysts could be obtained through the following strategies: (i) exchanging protons with Lewis acidic cations, including ZnII, AlIII, SnIV, FeIII and so on32, (ii) incorporation of rare earth metal ions into the vacant sites of POM ions33,34 and (iii) supporting POMs on Lewis acidic porous materials, for instance, Ta2O5 and zeolites31,35,36. However, the above materials often encounter disadvantages, such as being homogeneous forms, difficulty in controlling synthesis, leaching of active sites from supports, hindering substrate accessing to catalytically active sites and inferior activities compared to homogeneous analogues. Meanwhile, crystalline catalysts have caught increasing attention of researchers worldwide due to their well-defined crystalline structures, regular and tunable cavities and high catalytically active site density37,38,39. If organic ligands can directly bond to metal-oxygen clusters to achieve metal-oxygen clusters-based crystalline catalysts that meet the above-mentioned requirements, it may be a ‘one stone two birds’ method. So far, this kind of catalyst has not been reported for biodiesel preparation. Therefore, the search for design synthesis and development of environmentally benign catalysts which are highly active in water, higher thermodynamic stable, highly accessible to acid sites and recyclable is one of the most challenging issues in the production of biodiesel.

These inspired us to employ a semirigid ligand containing eight carboxyl groups, 5,5′-(2,2-bis((3,5-dicarboxyphenoxy)methyl)propane-1,3-diyl)bis(oxy) diisophthalic acid (H8L, Figure S1)40 and H3PO3 for the isolation of crystalline catalysts containing d10 metal-oxygen clusters, which gives the following considerations: (i) It is difficult to isolate classical metal-oxygen clusters from late transition metal ions due to the instability of extremely high negative charge. The substitution of terminal−O and μ2−O of POM clusters with rich carboxyl groups of H8L ligand would greatly reduce negative charge, which is beneficial to the formation of late transition metal-oxygen clusters. (ii) It is difficult to directly bond organic linkers to the traditional POM clusters owing to their oxygen-rich surface, so we intend to take the advantage of late transition metal ions to replace early ones to synthesize metal oxide cluster-organic frameworks. (iii) The introduction of H3PO3 is favorable to induce the formation of metal-oxygen clusters, which perhaps plays the role of heteroatom in the formation of hetero-POM anions. Fortunately, a novel heteropolyoxozincate containing hexadecanuclear zinc nanocluster, [Zn16(HPO3)4]L3 (IFMC-200, IFMC = Institute of Functional Material Chemistry), was successfully synthesized and structurally characterized. IFMC-200 is stable in air and common organic solvents and remarkably insensitive to the presence of water, even in acidic and basic aqueous solution within a certain range of pH values. Furthermore, IFMC-200 was employed as a heterogeneous catalyst for the preparation of biodiesel, which is the first time to employ a crystalline material of entire d10 metal oxide cluster with Lewis acid nature.

Results

Single-crystal X-ray analysis. IFMC-200 crystallizes in the cubic space group F-43c (219, Table S1) and consists of two kinds of ZnII cations (Figure S2), HPO32− anion and deprotonated L8− ion. Each Zn(1) ion is tetrahedrally coordinated by three carboxylic oxygen atoms derived from three bridging L8− fragments and one oxygen atom originated from HPO32− ion. And each Zn(2) atom shows a octahedral geometry, which is completed by six oxygen atoms originated from three bridging L8− ions and three HPO32− ions. The P element in HPO32− ion is in +3 oxidation state and exhibits a tetrahedral configuration with the lone pairs pointing into the tetrahedral cavity, which has been observed in arsenolite (As2O3)41. In addition, twelve Zn(1) and four Zn(2) cations were connected by four HPO32− anions, resulting in a nanocluster subunit [Zn16 (HPO3)4]24+ (noted as {P4Zn16}, Figure 1a, left). To a certain extent, it is similar to classical Keggin-type cluser42, which is composed of four internally edge sharing triads (M3O13) that share corners with each other and are disposed tetrahedrally around a central heteroatom with Td symmetry (Figure 1a, right). As a result of the expansion of space among four HPO32− ions, the size of {P4Zn16} cluster in IFMC-200 (with a diameter of ~13 Å) is clearly larger than that of Keggin-type anion (~10 Å). For conventional fully oxidized POM anions, especially Keggin ions, the highest occupied molecular orbital (HOMO) is mostly from 2p orbitals of the bridging-O atoms, whereas the lowest unoccupied molecular orbital (LUMO) is from 4d orbitals of Mo or W atoms43. However, the frontier molecular orbitals of IFMC-200 show significant differences with those of traditional POM clusters owing to d10 electronic configuration of ZnII cations. In IFMC-200, the carboxyl-O atoms that coordinate to Zn(2) atoms contribute to the HOMO and the LUMO is mainly from p orbitals of carboxyl-C atoms (Figure S3). An interesting structural feature of IFMC-200 is that each {P4Zn16} cluster is stabilized by 24 carboxyl groups from twelve L8− moieties and the connection of eight {P4Zn16} clusters and six L8− fragments generates one nanocage with dimensions of 11 × 11 × 11 Å (Figure 1b,c and S4). Such connectivity leads to a 3D open architecture (Figure 1d). PLATON calculations reveal that a free volume in IFMC-200 is about 7786.2 Å3 per unit cell (27.1% of the cell volume)44.

Figure 1
figure 1

(a) The structures of {P4Zn16} (left) and Keggin-type cluster (right); (b) and (c) the cage structure in IFMC-200; and (d) the 3D framework of IFMC-200.

On the basis of topological analysis, the structure of IFMC-200 can be considered as a 3D (3,4,24)-connected framework, if the center of the L8− ligand is regarded as a 4-connected node, the C6H3(COO)2 moiety as a 3-connected node and the {P4Zn16} cluster as a 24-connected node with a distorted rhombicuboctahedral geometry (Figure 2). The underlying trinodal net, which has been assigned the reticular chemistry structural resource (RCSR) symbol45, ddy, was analyzed with the program Systre46 which shows that it has intrinsic symmetry Pm-3m and is one of a special group of nets with the minimum number of kinds of edge (link) – in this case two for a trindal net. A maximum symmetry embedding is presented as a systre file in the Supplementary Information. It is worthy of note that the 24-connected node is essentially different from that documented in the literatures47. The previously-reported 24-connected node is really a tertiary building unit, constructed from twelve “paddle-wheel” dimetal tetracarboxylate secondary building units and may be further extended by functionalization at 5-position of 1,3-benzenedicarboxylate ligand (Figure S5)48,49,50. It should be pointed out that {P4Zn16} cluster is the first real 24-connected node, which is an important discovery in topological crystal chemistry. From the data in the RCSR, we found that IFMC-200 is the first example of (3,4,24)-connected topology and the first undoubted 24-connected node constructed from a single cluster. Moreover, it was also not hypothesized in Euclidean Patterns in Non-Euclidean Tilings (EPINET)51. Offering further insight into this intricate architecture, the overall structure of IFMC-200 can also be considered as a (4,12)-connected ftw topology framework when L8− fragments are considered as 4-connected nodes and each {P4Zn16} cluster corresponds to a 12-connected vertex (Figure S6)52. So far, only three topologies combine 4,12-connected components in the RCSR database, that is, ith, shp and ftw. Notably, our group has reported a (4,12)-connected ith topology network (Figure S7)53.

Figure 2
figure 2

The (3,4,24)-connected structure of IFMC-200.

(a) the center of H8L ligand as a 4-connected node, the C6H3(COO)2 moiety as a 3-connected node; (b) and (c) the {P4Zn16} cluster as a 24-connected node; and (d) the (3,4,24)-connected ddy topology.

Discussion

Crystals of IFMC-200 were isolated by hydrothermal reaction of ZnCl2, H3PO3, Na2MoO4·2H2O, Mo powder, H8L and TBAOH (Tetrabutylammonium hydroxide). The parallel experiments suggest that IFMC-200 cannot be obtained if there is no addition of Na2MoO4·2H2O and Mo powder. Meanwhile, H3PO3 not only plays the role of phosphorus source but also structure-directing agent in the formation of {P4Zn16} cluster because there is no product if H3PO3 was replaced with H3PO4. It is clear to see that IFMC-200 formed a cubic shape under both optical microscopy and scanning electron microscopy (Figure S8). The corresponding EDS analysis reveals that there is no Mo element in IFMC-200 although sodium molybdate dihydrate and Mo powder were introduced to the reaction system (Figure S9), which is in agreement with the result of single crystal X-ray diffraction. Phase purity of the bulky crystals was confirmed by the similarity between the experimental and simulated X-ray powder diffraction (XRPD) patterns (Figure S10). IFMC-200 was generated in aqueous media; theoretically, it is stable in water even at different temperatures, as confirmed by subsequent XRPD and FT/IR measurements (Figure S10 and S11). It was also found that IFMC-200 is stable in acidic and basic aqueous solution in the pH range of 2–12 (Figure 3 and S12). The vapour adsorption experiments for IFMC-200 suggest that water is hardly adsorbed, but certain amounts of ethanol and cyclohexane are adsorbed (Figure S13). Thus, IFMC-200 possesses hydrophobic cages. Furthermore, the thermogravimetric analysis (TGA) suggests that IFMC-200 has exceptional thermal stability with decomposition starting at 450°C (Figure S14), which is much higher than that of Nafion NR50 (280°C), one of the most stable resins54.

Figure 3
figure 3

The XRPD patterns of IFMC-200 after heated in Teflonlined stainless steel container at 65°C (a, b) and 140°C (c, d) for 12 h at different pH values, respectively.

Exp represents XRPD pattern of IFMC-200 without further treatment.

Generally, biodiesel can be prepared by transesterification of triglycerides or esterification of fatty acids with alcohols55,56. In terms of both atomic economy and green synthesis, heterogeneous solid state acid catalysts that work actively with easily separation for recyclability are highly desirable. The acid value of 0.61 mmol·g−1 for IFMC-200 was detected by titration with aqueous ammonia, which is smaller than those of ZnCl2 (7.96 mmol·g−1) and Keggin structure Zn1.2H0.6PW12O40 (1.14 mmol·g−1)30, but higher than those of other Lewis solid state acid catalysts (e.g., Y-zeolites, 0.54 mmol H+·g−1)57. In this regard, IFMC-200 was evaluated as a heterogeneous catalyst in the esterification of fatty acids with a series of alcohols (Figure 4). The reactions were performed in the presence of an excess amount of alcohols with oil/alcohol molar ratio of 1:10 at 65°C. Preliminary studies were carried out to optimize the reaction conditions and the results are presented in Figure S15, Table S2 and S3. The esterification reaction of palmitic acid with ethanol catalyzed by IFMC-200 reached its highest conversion (44.7%) and turnover frequency (TOF, 26.4 h−1) with catalyst amount of 1% wt total reactant weight in 10 h. In comparison, a blank experiment was also performed under the given conditions. The conversion of IFMC-200 catalyzed esterification of palmitic acid with ethanol was 10 times higher than that in the case of no catalyst (4.3%). Furthermore, we used the conventional Lewis acid ZnCl2 with higher acid value than IFMC-200 as catalyst under the identical conditions and the result suggested that IFMC-200 showed higher TOF than ZnCl2 (8.6 h−1). This may be attributed to the poor water-tolerance of ZnCl2. It is known that ZnCl2 requires strictly anhydrous conditions because it immediately reacts with water forming in esterification of acids with alcohols, resulting in its serious decomposition58. IFMC-200 is much more stable than ZnCl2 in esterification reaction and could not be decomposed by water. With the optimal reaction time and catalyst amount, a series of fatty acids and alcohols were then examined under this condition and the results are collected in Table 1 and Figure S16. As can be seen from Table 1, the biodiesel conversion increased as the chain length of free fatty acids was increased, maintaining other constants. This result is opposite to the previous report59, in which Goodwin et al. illustrated that the porous silica support would diminish the likelihood of mass diffusion limitation, making it favorable for the esterification reaction of low-molecular weight carboxylic acids. This abnormal result perhaps can be attributed to the positive effects of inorganic-organic hybrid framework of IFMC-200 and the presence of hydrophobic cage structure in IFMC-200. It is well-known that the acidic catalysis depends on the acidic strength, as well as the concentration of substrates around the catalytic sites. IFMC-200 located its Lewis acidic sites on the backbone of framework and hydrophobic surroundings inside the pores. This unique structure feature provides long chain fatty acid molecules more chance to access to the catalytic sites due to their large lipophilicity. In other words, the long chain fatty acid molecules could be concentrated around the catalytic sites compared to short-chain ones. In addition, the hydrophobic cages could promote the water generating during the reaction to leach from the catalytic sites, which avoids the deactivation of the acid sites by water. Up to now, there are only few reports on solid porous catalysts with hydrophobic cages, just Inumaru's group firstly demonstrated H3PW12O40 immobilized in hydrophobic nanospaces of organomodified mesoporous silica to show water-tolerant properties60. Therefore, IFMC-200 exhibited water-tolerant properties compared to conventional Lewis acid, ZnCl2, generated by the existence of hydrophobic cages, which makes it favorable for esterification of long-chain carboxylic acids rather than short-chain ones. It might be the only one report on this opposite result. More importantly, IFMC-200 showed very good recyclability. For example, there was almost no activity loss over IFMC-200 after six times (Figure S17). This perhaps associated with the stable structure of IFMC-200 (Figure S18 and S19). The excellent stability and recyclability of IFMC-200 is potentially important for industrial applications.

Table 1 The Con. (%) and TOF (h−1) of IFMC-200 catalyzed esterification of FFAs with alcohols. Conditions: 1% wt catalyst; 65°C; oil/alcohol molar ratio = 1:10; 10 h
Figure 4
figure 4

The equation of IFMC-200 catalyzed esterification of FFAs with alcohols.

In summary, we synthesized an unprecedented 3D (3,4,24)-connected heteropolyoxozincate organic compound, IFMC-200, which contains the first example of 24-connected node constructed from a single metal cluster. We have developed a new strategy with “carboxyl group-stabilized polyoxozincate cluster” mode that helps to lower the overall negative charge and successfully proved that it is an efficient method to isolate late transition metal-oxygen cluster. IFMC-200 exhibits superior thermal stability, good water-stability, tolerance to acid and base and Lewis acidity, which enable its catalytic activity in the esterification of carboxylic acids with alcohols. Surprisingly, IFMC-200 exhibits abnormal catalytic result that is favor for the conversion of long-chain fatty acids under the same conditions compared to the previous reports. Further studies concerning this strategy for the synthesis of more late transition oxide cluster-based heterogeneous crystalline catalysts and their catalytic activities in the production of biodiesel are currently in progress.

Methods

Sample synthesis

6 mL of distilled water was added to a beaker, which contains Na2MoO4·2H2O (0.63 g, 2.6 mmol), Mo powder (0.05 g, 0.52 mmol), ZnCl2 (0.14 g, 1.0 mmol), 50% H3PO3 (0.10 mL) and TBAOH with stirring for 15 min. Next, the pH value of the mixture was adjusted to 4.8 with 2 mol·L−1 HCl and then the mixture was stirred for another 15 min followed by the addition of H8L (0.12 g, 0.15 mmol). Then the mixture was transferred and sealed in a 23 mL Teflonlined stainless steel container and cooled to room temperature at 5°C·h−1 after heated at 180°C for 5 d. Cubic crystals of IFMC-200 were collected and dried in air (yield 65%, based on H8L). Elemental analysis: Anal. Calcd for C111H64O72P4Zn16: C 35.84, H 1.73, Zn 28.13, P 3.33; found: C 35.78, H 1.82, Zn 28.05, P 3.26. IR (KBr, cm−1): 3448 (s), 2939 (m), 1626 (s), 1561 (s), 1469 (s), 1425 (s), 1393 (s), 1266 (m), 1138 (m), 1082 (m), 1046 (m), 889 (w), 815 (w), 775 (m), 730 (m), 594 (w), 458 (w).

Sample characterization

The starting material H8L was synthesized according to modified procedure of the reported literature40. All other chemicals were obtained from commercial sources and were used without further purification. Elemental analyses (C, H and N) were measured on a Perkin-Elmer 2400 CHN elemental analyzer; Zn and P were determined with a Plasma-SPEC(I) ICP atomic emission spectrometer. IR spectrum was performed in the range 4000–400 cm−1 using KBr pellets on an Alpha Centaurt FT/IR spectrophotometer. X-ray powder diffraction (XRPD) measurement was recorded radiation ranging from 3 to 60° at room temperature on a Bruker D8 Advance diffractometer with Cu-Kα (λ = 1.5418 Å). Thermogravimetric analysis (TGA) was performed using a Perkin-Elmer TG-7 analyzer heated from room temperature to 900°C under nitrogen at the heating rate of 5°C·min−1. Field-emission scanning electron microscopy (FE SEM) images were obtained with a XL30 ESEM FEG microscope equipped with energy dispersive X-ray spectroscopy (EDS).

Determination of acid strength

Titration was used to evaluate the acidic characteristics of the solids. 100 mg of solid samples suspended in 25 mL of NaCl aqueous solution (0.05 mol·L−1) was stirred for 30 h. Then, acid value was measured by titration with a solution of aqueous ammonia (0.01 mol·L−1) using phenolphthalein as an indicator.

Catalytic experiments

A 100 mL three-necked glass flask charged with cooled condenser with ~5.0 g of FFAs, a certain amount of anhydrous alcohols and catalyst (IFMC-200) in it were vigorously stirred and refluxed (65°C) for the required reaction time. After reaction, the mixture was rotary evaporated at 50°C to separate the excess of alcohols. Then the mixture was left divided into two layers by centrifugation with 4000 rpm. The upper layer was esters of FFAs and the lower layer was IFMC-200 with small amount of water. The conversion of esters was calculated by measuring the acid value of the products and the yield was detected by gas chromatography (GC). The catalyst was decanted at the bottom of centrifuge, which was easy to be separated, washed by methanol to remove the remaining reactants and water, then dried in air and reused for the next experiment.

Adsorption experiments

The adsorption isotherms of water, ethanol and cyclohexane are measured at 298 K for IFMC-200. Before the measurements, the crystalline sample of IFMC-200 was immersed in CH2Cl2 for three days and then heated at 100°C for 12 h under a vacuum to evacuate the samples.