*3.1. Synthesis and Structure*

Compound **1** was prepared by the hydrothermal reaction of K7HNb6O19·13H2O, VOPO4·2H2O, NaVO3, Ni(CH3COO)2·4H2O, and en at 160 ◦C. The single-crystal structural analysis (Table S1) reveals that **1** crystallizes in the monoclinic *C2/c* space group. Compound **1** is a 3D organic-inorganic framework constructed from [PNb12O40(VO)5] <sup>5</sup><sup>−</sup> ({PNb12V5}) clusters and ten [Ni(en)2] 2+ linkers. The C2h symmetric {PNb12V5} cluster contains a typical Keggin-type [PNb12O40] <sup>15</sup><sup>−</sup> ({PNb12}) capped by five {VO} units (Figure 1a). In the {PNb12} cluster, the centered heteroatom P connects with four edge-sharing {Nb3O13} subunits by sharing μ4-O atoms; the P–Oc (Oc = central oxygen) bond lengths are in the range of 1.539–1.550 Å and the O–P–O angles are in the range of 108.7–110.6◦. Each Nb center is sixcoordinated with octahedral geometry and the bond lengths are 1.730–1.768 Å for Nb–Ot (Ot = terminal oxygen), 1.912–2.035 Å for Nb–Ob (Ob = bridge oxygen), and 2.539–2.583 Å for Nb–Oc. There are six square windows on {PNb12}, which are capped by four 100% occupied {VO} and two 50% occupied {VO} (front ellipses style) (Figure 1a). The pentacapped Keggin-type {PNb12O40(VO)5} in compound **1** presents a new V-containing PONb skeleton. All of the V centers are coordinated with four μ2-O atoms from {PNb12} and one terminal oxygen atom. The bond lengths of V–Ob are 1.929–2.340 Å, and those of V–Ot are 1.605–1.748 Å. According to the bond-valence sum (BVS) calculation, the five-capped V atoms are all in +4 oxidation state (Table S6). The oxidation state of the V atoms was

further confirmed by XPS measurement. In the XPS spectrum of **1** (Figure S5), the peaks at 523.2 eV and 516.0 eV are attributable to V4+ 2p1/2 and V4+ 2p3/2, respectively.

**Figure 1.** Crystal structure of **1**. (**a**) Ball-and-stick representation of {PNb12O40(VO)5} in **1**. (**b**) Connections between {PNb12O40(VO)5} and nickel ions. (**c**) The 3D framework of **1** viewed along the *b* direction. Color codes: Nb, green; V, pink; Ni, dark green; P, yellow; O, red; N, blue; C, gray.

There are four crystallographically independent Ni centers in **1** (Figure 1b); each one is coordinated with four N atoms from two en ligand and two terminal O atoms from two adjacent {PNb12O40(VO)5}. The Ni–N distances are in the range of 2.084–2.118 Å and the Ni–Ot distances are from 2.080 Å to 2.154 Å. Each {PNb12O40(VO)5} cluster was connected by ten adjacent [Ni(en)2] 2+ linkers to form a 3D framework (Figure 1c). To our knowledge, it represents the first extended V-containing PONb connected by Ni-complex units.

When a similar hydrothermal reaction was performed at 140 ◦C without adding NaVO3, compound **2** was obtained. Compound **2** crystallizes in the orthorhombic *Pna21* space group. Compound **2** contains a C2v symmetric discrete [PNb12O40(VO)2] <sup>10</sup><sup>−</sup> ({PNb12V2}) cluster and five [Ni(en)3] 2+ units as counter cations (Scheme 1). The polyanion {PNb12V2} features a {PNb12} cluster capping two {VO} units (Figure 2). As with **1**, the centered heteroatom P is coordinated with four μ4-O atoms, and all of the Nb centers are six-coordinated with octahedral geometry. The P–Oc bond lengths are in the range of 1.549–1.555 Å, and the O–P–O angles are in the range of 108.4–111.3◦. The bond lengths of Nb–Ot, Nb–Ob, and Nb–Oc in {PNb12V2} are in the ranges 1.743–1.776 Å, 1.888–2.125 Å, and 2.498–2.604 Å, respectively. Two {VO} units are located on two opposite square windows on the surface of {PNb12}, and each V center exhibits square pyramidal geometry. The bond lengths of V–Ot and V–Ob are in the ranges 1.620–1.623 Å and 1.853–1.979 Å, respectively. Five free [Ni(en)3] 2+ are distributed around {PNb12V2} and the Ni–N bond lengths are in the range of 2.061–2.164 Å. The XPS spectrum (Figure S6) indicates that there are both V+4 and V+5 oxidation states in **2**. In the V 2p region of **2**, the peaks at 523.4 eV and 515.8 eV are assigned to +4 oxidation state, and the peaks at 524.8 eV and 517.3 eV are attributable to +5 oxidation state, which are consistent with the BVS values of V (3.98 for V1 and 4.64 for V2) (Table S7).

We systematically explored the factors that influence the synthesis of **1** and **2**. It is found that the used en can effectively protect Ni2+ from hydrolysis under alkaline conditions. When 1,2-diaminopropane or 1,3-diaminopropane was used instead of en, the amount of precipitation was obtained. In addition, control experiments show that temperature and vanadium source play important roles in the synthesis of **1** and **2**. Following a procedure similar to that of **1**, compound **2** was obtained by removing NaVO3 at 140 ◦C. In addition, compound **2** cannot be obtained by lowering the hydrothermal temperature or varying the ratio of VOPO4 to NaVO3 in the synthesis of **1**. Therefore, we speculate that the evaluated hydrothermal temperature and the use of NaVO3 might increase the number of vanadium caps in the PONb cluster, and meanwhile the terminal O atoms of the Keggin-type {PNb12} would be activated by introducing additional vanadyl caps. As a result, two-capped {PNb12O40(VO)2} was isolated as a discrete cluster with the Ni-complex as counter cations, and five-capped {PNb12O40(VO)5} gave rise to a 3D framework by using the Ni-complex as linker.

The IR spectra of **1** and **2** (Figure S2) were recorded from 4000 to 400 cm<sup>−</sup>1. The terminal M = Ot (M = Nb, V) vibrations are at 942 and 870 cm−<sup>1</sup> for **1**, and at 947 and 877 cm−<sup>1</sup> for **2**. The peaks at 805, 700, 627, 498, and 473 cm−<sup>1</sup> of **1**, and at 815, 709, 638, 505, and 469 cm−<sup>1</sup> of **2** are attributed to the bridging M–O–M vibrations. The peaks at 1032 cm−<sup>1</sup> for **1**, and at 1027 cm−<sup>1</sup> for **2**, are attributed to P–Oc vibration. The peaks at 1642–1048 cm−<sup>1</sup> for **1,** and at 1593–1035 cm−<sup>1</sup> for **2** can be assigned to the en ligand. In addition, for the two compounds, the broad band above 3000 cm−<sup>1</sup> is attributed to the O–H vibrations of water molecules and/or the hydroxyl groups.

The phase purity of **1** and **2** was confirmed by PXRD (Figure S3), where the collected diffraction peaks match well with the simulated ones. Compounds **1** and **2** are nearly insoluble in water and common organic solvents, such as CH2Cl2, THF, CH3COCH3, CH3CN, and DMF (Figures S8 and S9). Therefore, we tested their pH stability in aqueous solution modified by PXRD (Figure 3) and IR spectra (Figure S7). As shown in Figure 3, **1** and **2** remained stable in the pH range of 4–14 after soaking for 24 h and began to decompose when the solution pH was 3. In addition, the crystals of **1** and **2** can keep their structure integrity after heating in organic solvent in the temperature range of 40–80 ◦C (Figures S10 and S11).

**Figure 3.** PXRD patterns of **1** (**a**) and **2** (**b**) after being soaked in aqueous solutions with different pH values for 24 h.

#### *3.2. Electrocatalytic Selective Oxidation of Benzyl Alcohol*

The selective oxidation of alcohols to aldehydes is one of the important organic transformations [43,44]. Compared with the traditional oxidation processes, the electrochemical oxidation provides an efficient and sustainable alternative [45,46]. Driven by electricity, alcohols can be oxidized on the anodic electrode under ambient conditions with hydrogen released from the cathodic electrode. Although some electrocatalysts have been developed in the anodic oxidation of alcohols, the selective oxidation of alcohols to the corresponding aldehydes remains a challenge, and in the reported system, the reaction activity significantly relies on the addition of alkaline additives [36,47,48]. Therefore, it is necessary to develop efficient and cost-effective electrode materials to realize the selective oxidation of alcohols under alkaline additive free conditions.

Considering that V-containing PONbs **1** and **2** both have both Brønsted basicity and redox activity, we investigate the electrocatalytic activities of **1** and **2** using the selective oxidation of benzyl alcohol (BA) to benzaldehyde as a model reaction. The electrocatalytic activity of **1** and **2** was first evaluated by the cyclic voltammetry (CV) method, which was performed in an acetonitrile solution containing LiClO4 and BA with a carbon cloth modified by **1** or **2** as the working electrode. As shown in Figure 4, for **1** and **2**, the addition of BA leads to the significant increase in the anodic peak currents, indicating that the two compounds have a fast electrocatalytic response to the oxidation of BA. Notably, the anodic peak current of **1** is obviously higher than that of **2**, revealing that **1** has better electrocatalytic performance than **2**. The electrocatalytic activities of **1** and **2** were further verified by bulk electrolysis experiments performed in an undivided cell using **1** or **2** modified carbon cloth as the working electrode. As shown in Figure 5a and Table 1, both **1** and **2** are active for the selective oxidation of BA. Under ambient conditions, 92% of BA was converted by **1** in 6 h at the potential of 1.6 V vs. Ag/Ag+ and the selectivity for benzaldehyde reached 95%, giving the Faradaic efficiency (FE) of 93% (Table 1, entry 2). In addition, a trace amount of N-benzylacetamide as the only by-product was detected (Figures S12 and S14). Under the otherwise identical conditions, the catalytic activity of **2** (conversion: 79%, selectivity: 90%, FE: 84%, Table 1, entry 3) is lower than that of **1**. During the electrolysis process, hydrogen was released on the counter electrode (Figure S13).

**Figure 4.** CV curves of **1** and **2** with BA (0.5 mmol) and without BA, obtained with a carbon cloth working electrode modified with **1** or **2** at a scan rate of 40 mV s<sup>−</sup>1.

**Figure 5.** (**a**) Time profile for the electrocatalytic oxidation of BA by **1** and **2**; (**b**) The scan-rate dependence of current density at potential = 0.07 V vs. Ag/Ag+ for **1** and **2**.


**Table 1.** Electrocatalytic selective oxidation of benzyl alcohol by different catalysts a.

<sup>a</sup> Standard reaction conditions: CH3CN (10 mL), LiClO4 (1.0 mmol), BA (0.5 mmol), reaction time: 6 h, constant potential: 1.6 V vs. Ag/Ag+. <sup>b</sup> The product conversion and selectivity were determined by GC analysis using biphenyl as internal standard.

To investigate the influence of the Ni-complex unit, PONb, and the capped V of **1** and **2** on the electrocatalytic selective oxidation of BA, the following control experiments were carried out (Table 1, entries 4–8). As shown in Table 1, entry 4, the electrocatalytic activity of the Ni-complex unit is negligible, because the catalytic performance of Ni(en)3Cl2 (conversion: 48%, selectivity: 69%, FE: 70%) is almost comparable to that of the blank

test (conversion: 42%, selectivity: 73%, FE: 66%, Table 1, entry 1). When carbon cloth modified by K7H[Nb6O19]·13H2O was used as a working electrode, the conversion of BA (72%) was improved relative to the blank test, but the selectivity (79%) for benzaldehyde was still common (Table 1, entry 5). Moreover, a similar result was obtained by [N(CH3)4]10H5[PNb12O40] (conversion: 70%, selectivity: 78%, FE: 76%, Table 1, entry 6). The above results indicate that PONbs contribute to the conversion of BA because basic PONbs might facilitate the dehydrogenation oxidation of BA. Therefore, the temperatureprogrammed desorption of the carbon dioxide (CO2-TPD) measurement for **1** and **2** was performed, where the desorption peaks at 152 ◦C for **1** and 148 ◦C for **2** corresponding to the weak base site were observed, respectively (Figure S15). In addition, polyoxovanadate, K6[V10O28], can convert 83% of the substrate (Table 1, entry 7), but its selectivity (76%) is lower than that of the V-containing PONb **1** or **2**. As shown in Table 1, entry 8, the catalytic activity of the bicapped Keggin-type [N(CH3)4]9[PNb12O40(VO)2] (conversion: 74%, selectivity: 89%) is similar to that of **2** (conversion: 79%, selectivity: 90%). The control experiments above show that both the PONb cluster and the V caps contribute to the enhancement of BA oxidation. Then, we speculate that the different catalytic activity of **1** and **2** is mainly caused by their different number of V caps. This is further confirmed by the electrochemical surface area (ECSA) measurement: the ECSA of **1** (4.0 mF·cm<sup>−</sup>2) with five V caps is higher than that of **2** (3.3 mF·cm<sup>−</sup>2) with two V caps (Figures 5b and S16).

To explore the optimal reaction conditions, we systematically investigated the influences of electrolyte, solvent, applied potential, and catalyst dosage on the electrocatalytic selective oxidation of BA by **1**. As shown in Figure 6a, compared with other types of supporting electrolytes, **1** exhibits excellent catalytic performance by using LiClO4. Meanwhile, it is found that acetonitrile with excellent conductivity exhibits a better performance than that of acetone, tetrahydrofuran, and N,N-dimethylformamide (Figure 6b). When the applied potential was increased from 1.4 to 1.6 V vs. Ag/Ag+, the conversion of BA was increased from 16% to 92%, but when it reached 1.7 V vs. Ag/Ag+, the selectivity decreased to 79%, although 98% of the BA was converted (Figure 6c). Therefore, 1.6 V vs. Ag/Ag+ is the optimal potential. As shown in Figure 6d, the best catalytic performance was achieved by using 1.0 mg **1**. After that, the catalytic activity was not further improved by increasing the catalyst dosage.

Moreover, the recyclability and stability of **1** were evaluated. As shown in Figure 7a, the catalytic activity of **1** is basically maintained after four cycles. There is no obvious change observed in the IR spectra of 1 before and after the reaction (Figure S17), revealing that compound **1** is basically stable after the recycle test. We compared the XPS spectra of **1** before and after the recycle. As shown in Figure 7b, in the V 2p region, the peaks at 523.2 eV and 516.0 eV are basically unchanged, indicating that the oxidation state of V remains +4 in **1**. Meanwhile, to verify the heterogeneity of **1**, the reaction solution was tested by ICP-OES (detection limit ca. 1 ppm) and no Nb, V, or Ni was detected, indicating that there is no catalyst leaching during the electrocatalytic process. In addition, no characteristic absorption of V-containing PONb is detected by the UV-vis spectrum (Figure S18).

In order to explore the possible mechanism of the electrocatalytic selective oxidation of BA by 1, we performed free radical trapping experiments. Oxygen radical scavenger, diphenylamine, and hydroxyl radical scavenger, *tert*-butanol, were added to the reaction system, respectively. As shown in Table S8, the oxidation of BA was significantly inhibited after the addition of diphenylamine. Therefore, we speculate that a free radical process was involved in the electrocatalytic oxidation of BA. Based on the experimental results, a plausible reaction mechanism was proposed (Figure S19). First, the electrocatalyst 1 in the reduced state (1-Red) is oxidized to its oxidized state (1-Ox) at the anode under constant potential. Then, the hydroxyl group of BA might be activated by the surface bridging O of the Keggin-type PONb cluster due to its Brønsted basicity [49,50]. After that, the BA is oxidized by 1-Ox through a −1e−/−1H<sup>+</sup> process, generating the oxygen radical species (PhCH2O•). PhCH2O• is further oxidized to benzaldehyde through another −1e−/−1H+

process, and meanwhile, 1-Ox is reduced to 1-Red, releasing protons to complete a catalytic cycle. The released protons are reduced at the cathode to produce H2.

**Figure 6.** The influence of supporting electrolytes (**a**), solvent (**b**), applied potential (**c**), and catalyst dosage (**d**) on the electrocatalytic oxidation of BA by **1**.

**Figure 7.** (**a**) Recycle test for the electrocatalytic selective oxidation of BA by **1**; (**b**) V 2p XPS spectra of **1** before and after the oxidation.
