*2.5. Computational Studies*

The molecular geometries of **SiW10-APTES**, **SiW10-monoB10**, and **SiW10-diB10**, as well as those of P2W17-APTES, P2W17-monoB10 and P2W17-diB10 were fully optimized at a DFT level including implicit solvent effects (see Figure 7 and Supplementary Materials for computational details). We considered the most relevant plausible conformers. Firstly, regarding **SiW10-APTES**, it exhibits two main conformers as defined by the orientation of the two amine organic arms, which we called them open and closed forms. The small difference in their relative energy, less than 1 kcal·mol−<sup>1</sup> in favor of the closed form (represented in Figure 1a), forecasted that further substitution would easily overcome any initial geometric preference in the reactants. Indeed, upon B10 incorporation a much more complex situation arises. For **SiW10-monoB10** we characterized five conformers, two arising from the closed reactant and three species from the open reactant form. In the most stable conformer (Figure 7a), which arise from the closed form, the decaborate moiety interacts favorably with the amine hydrogens (d' and d") of the unreacted arm through strong dihydrogen contacts. In the most stable open form (Figure 7b), although interaction between arms is almost neglected, the H amide atom develops other interactions. Overall, the most stable conformer given in Figure 7a is 11 kcal·mol−<sup>1</sup> below the second one (Figure 7b). All five conformers lie in a narrow 20 kcal·mol−<sup>1</sup> range. For the double substituted **SiW10-diB10**, since the additional repulsion arising from the negatively charged B10 groups, we could only characterize two forms: an open (not shown) and a closed one (two views on Figure 7c,d). The energy difference between both species was computed to be just only 5 kcal·mol<sup>−</sup>1. We highlight, as dashed lines in Figure 7, those hydrogen atoms of the organic arm and the B10 moiety that lie close in three-dimensional space.

Due to the monovacant character of P2W17, the Si-O-Si angle of the APTES moiety grafted to the POM strongly differs from that observed for the divacant SiW10 POM (see Figure 1). The topology of the two arms, and thus the connectivity of the two POM-APTES derivatives, strongly differs. Therefore, as seen in Figure 1, closed form is not possible for P2W17-APTES. Only one geometry could thus be considered. Then, for the monosubstituted P2W17-monoB10, two conformers were characterized, one open and one folded, the folded one being more stable by only 2.2 kcal·mol<sup>−</sup>1. For the di-substituted Dawson derivative, only one open shaped product could be characterized (see Figure 7e,f).

Hydrogen atoms of the decaborate moieties possess a hydride character. Consequently, they can establish hydrogen-hydrogen contacts with protic solvent or with functional groups like amines or amides. In the present structures, many H-H dihydrogen contacts between the amine organic arms from APTES moiety and hydrogen atoms from decaborate clusters were observed. For instance, for **SiW10-monoB10** (Figure 7a), the hydrogen atoms d' and d" from the «free» amino group are found 2.21 and 1.99 Å far from an H atom belonging to the B10 cluster, which are quite short distances. This fact agrees with the couplings observed in the NMR experiments.

The thermodynamics of the formation of the mono- and di-adducts of the Keggin and the Dawson species is computed exergonic in all cases as seen in Figure 8. For **SiW10- APTES**, both the formation of the mono- and bi-derivative were computed exergonic, 22.1 kcal·mol−<sup>1</sup> for the **SiW10-monoB10**, and 13.0 kcal·mol−<sup>1</sup> for the **SiW10-diB10**. The formation of **SiW10-monoB10** is clearly favored and the strong dihydrogen contacts between amino group and the grafted B10 cluster undoubtedly strongly stabilize such a species compared to the di-adduct **SiW10-diB10**. For P2W17-APTES, also the two substitutions are favorable, 6.8 kcal·mol−<sup>1</sup> for the first, and 7.9 kcal·mol−<sup>1</sup> for the second.

**Figure 8.** Energetic profiles of the formation of mono- and di-adduct from the starting precursors in CD3CN. (**a**) **SiW10-APTES**, **SiW10-monoB10** (open and closed isomers), and **SiW10-diB10;** (**b**) **P2W17- APTES**, **P2W17-monoB10**, and **P2W17-diB10.**.

The computed reaction free energies for the **SiW10-APTES** and P2W17-APTES derivatives are fully consistent with the experimental findings. For the keggin derivatives, the strong stabilization of the mono-adduct allows isolating both mono and di-adduct thanks to the formation of strong H-H contacts. In contrast, for the Dawson derivatives, the small difference of energies between mono- and bi-adducts (1.1 eV only) does not permit isolating the mono-adduct. Besides, an excess of [B10H9CO]<sup>−</sup> (3 equivalents/POM instead of 2) is needed to get the pure di-adduct compound to avoid the formation of a mixture between mono and di-grafted adducts. A similar situation was previously obtained with

Anderson-type derivatives since the mono and di-adduct of B10 with [MnMo6(Tris)2] 3− are only separated by 6 eV and it was not possible to get mono-adduct [22]. This result highlights the role of the topology of the POM-APTES compounds and their faculty to stabilize species thanks to intramolecular interactions.

Finally, DFT studies provided the frontier orbitals for each compound (see Figure 9 and Figure S33 in Supplementary Materials). The results obtained for Keggin and Dawson derivatives exhibit the same feature. For POM-APTES the HOMO is located on one (for **SiW10-APTES**) or two (**P2W17-APTES**) amines of the APTES part, while the LUMO are localized on the W atoms of the POM part. By grafting the B10 clusters, the LUMO levels are slightly affected. LUMO remains localized on W atoms and only minor changes in energy are observed.

**Figure 9.** Frontier orbitals energies and energy gaps (eV) for the **SiW10-APTES**, **SiW10-monoB10** and **SiW10-diB10** species. Color code: W green, O red, Si light brown, B dark brown, C grey, N blue, H white; HOMO: red/blue; LUMO: orange/cyan. MO surfaces plotted at a 0.03 isovalue.

Conversely, the HOMO levels are drastically modified by the introduction of B10 clusters. Electrons of the HOMO orbitals are now mainly localized on one grafted B10 cluster. Interestingly, for **SiW10-monoB10** the HOMO is delocalized between one B10 cluster and the amine of the second arm, which strongly interacts with the B10 through H-H contacts. The HOMO energy level increases in all cases within the range 0.78 to 0.92 eV. The HOMO-LUMO gaps, are thus significantly reduced upon the B10 grafting. Indeed, for **SiW10-APTES**, the gap decreases from 1.67 to 0.90 eV for the first substitution, and to 0.94 eV for the second. For **P2W17-APTES**, the gap evolves from 1.56 to 0.77 eV for the first substitution, and to 0.76 eV for the second.

#### *2.6. Electrochemical Properties*

The electronic spectra of compounds **SiW10-monoB10**, **SiW10-diB10** and **P2W17-diB10** recorded in CH3CN containing 0.1 mol.L−<sup>1</sup> TBAClO4 (TBAP, tetrabutylammonium perchlorate)

at room temperature and 2.10−<sup>4</sup> mol.L−<sup>1</sup> concentration are depicted in Figures S35 and S36 (Supplementary Materials).

The electronic spectra of precursors **SiW10-APTES** and **P2W17-APTES** display absorption bands in ultraviolet region corresponding to transition between p-orbitals of the oxo ligands and d-type orbitals centered on tungsten [33,34], while the cluster [B10H9CO]<sup>−</sup> exhibits weak absorption band between 300 and 200 nm notably assigned to π−π\* transitions [35]. Considering that the main contribution of the spectra comes from the LMCT band involving the W atoms and that the LUMO band centered on tungsten atoms are only slightly modified upon grafting of B10 cluster, no drastic changes are expected in the POM-B10 adducts. Indeed, the electronic spectra of **SiW10-monoB10** and **P2W17-diB10** match well with the sum of the spectra of **SiW10-APTES** or **P2W17-APTES** and one or two times that of [B10H9CO]−, respectively. The spectrum of **SiW10-diB10** slightly differs from the sum of the component's spectra probably due to a larger variation of LUMO energy level from **SiW10-APTES** to **SiW10-diB10** and additional constraints due to the vicinity of the two boron clusters.

Since no evolution of the spectra were observed within 24 h in such a medium, the compounds appear chemically stable in these experimental conditions. The cyclic voltammograms (CVs) were thus recorded for all the SiW10 and P2W17 derivatives and are given in Figure 10 and in Figures S37 and S38 (Supplementary Materials), while the anodic and cathodic potentials are gathered in Table S3 (Supplementary Materials).

**Figure 10.** Comparison of cyclic voltammograms (**a**) for the three compounds **SiW10-APTES**, **SiW10 monoB10** and **SiW10-diB10**, and (**b**) for **P2W17-APTES** and **P2W17-diB10** in the reduction part. The electrolyte was CH3CN + 0.1 M TBAClO4. Dashed lines are only guide for eyes.

As depicted in Figure 10a and Figure S37, the CV of **SiW10-APTES** is poorly resolved and it is difficult to identify confidently all the reduction processes corresponding to the successive reduction in WVI centers into WV, well-known to be monoelectronic in nonaqueous solvents [36]. These waves seem nevertheless reversible with processes better resolved in oxidation. Besides, an irreversible process attributed to the oxidation of amine function of the APTES linker is also observed in oxidation around +0.452 V vs. Fc+/Fc

(see Supplementary Materials). As for their parent precursor, CVs of **SiW10-monoB10** and **SiW10-diB10** display poorly resolved reversible electronic transfers, which appear shifted towards more negative potentials and one irreversible oxidation process around +0.452 V vs. Fc+/Fc assigned to the oxidation of the remaining amine function and/or of the B10 cluster (see Figure S39, Supplementary Materials). This behavior agrees with the increase in the charge from 4- in **SiW10-APTES** to 6- in **SiW10-monoB10** and 8- in **SiW10-diB10** and the electron donating character of the boron cluster [37] but does not evidence a strong electronic effect of the boron cluster on the POMs electronic properties.

The CVs of **P2W17-APTES** and **P2W17-diB10** are given in Figure 10b and in Figure S38 (Supplementary Materials) and appears much more resolved than those of SiW10 derivatives. The CV of **P2W17-APTES** displays four reversible electronic transfers with cathodic peak potentials assigned to successive mono- or bi-electronic reductions of WVI centers into W(+V) [38] and two irreversible oxidation processes at Epa = +0.452 and +0.759 V, assigned to the oxidation of the terminal amine groups of the APTES linkers. Conversely to the di-adduct compound **SiW10-diB10**, the CVs of the Dawson derivative **P2W17-diB10** give four reversible reduction processes significantly shifted towards the more positive potential compared to **P2W17-APTES** and one irreversible reduction process at Epc = −2.030 V vs. Fc+/Fc, which was not observed in the precursor. The opposite effect was expected. This effect probably results from a combination of a charge effect, the presence of protons (in DIPEAH+ cations) and of an electronic effect of boron cluster on P2W17 moiety but at this stage it is difficult to have a clear explanation of the contribution of all these effects which can be antagonist.

Although reduction waves in the Dawson derivatives are not very well resolved, it can be observed that the di-substituted species (green line in Figure 10) is reduced at lower potentials than **P2W17-APTES**, in agreement with the fact that the LUMO and LUMO+1 raise in energy upon B10 attachment. Also, the successive reduction waves seem just shifted left, which would conform with the almost constant difference in the LUMO and LUMO+1 energies along the series.

### *2.7. Electrocatalytic Properties for the Reduction in Protons into Hydrogen (HER)*

Many POMs are known to catalyze protons reduction into hydrogen in aqueous or in non-aqueous conditions [7,39,40]. We verified by UV-Vis spectroscopy that [B10H9CO]<sup>−</sup> and its adducts with POMs are stable in CH3CN in the presence of excess acetic acid (20 equivalents). In these conditions, it was interesting to study the reactivity of these compounds in regard to the electro-catalytic reduction in protons into hydrogen. The experiments were performed in CH3CN + 0.1 M TBAP by using acetic acid as a source of protons, and as a weak acid in such a medium (pKa = 22.3) [41].

Figure 11 and Figure S40 (Supplementary Materials) show the evolution of CVs upon stepwise addition of acetic acid up to 20 equivalents of acid/POM for all P2W17 and SiW10 derivatives, respectively. For all the compounds, the addition of acetic acid, gives a new irreversible reduction wave, which grows gradually with the amount of acid, expressed as γ = [acid]/[POM]. As shown in Figure 11 and Figure S40, at a given potential of −2.2 V vs Fc+/Fc, a linear dependence of the catalytic current versus γ is obtained, a behavior featuring the electro-catalytic reduction of protons. However, the effect of the addition of acetic acid in the solution appears stronger for P2W17 derivatives than for SiW10 based compounds.

**Figure 11.** Cyclic voltammograms of (**a**) **P2W17-APTES** and (**b**) **P2W17-diB10** after addition of variable amounts of acetic acid. (**c**) Plots of the cathodic currents measured at −2.2 V vs Fc+/Fc as a function of the ratio [acid]/[POM] for **P2W17-APTES** and **P2W17-diB10**. (**d**) Comparison of HER with and without catalyst **P2W17-diB10**, and with platinum after addition of an excess of acetic acid corresponding to the quantity added for a ratio [acid]/[POM] = 20. In all cases, the electrolyte was CH3CN + 0.1 M TBAClO4. The reference electrode was a saturated calomel electrode (SCE). Reproduced with permission from the doctoral thesis manuscript of Dr Manal Diab, University Paris Saclay/Lebanese University, May 2018.

To evidence the electrocatalytic process, linear voltammetry of **P2W17-diB10** in the presence of 20 equivalents of acetic acid was performed and compared to similar experiments performed without catalyst or on platinum electrode (Figure 11d). We notice that in the presence of the catalyst **P2W17-diB10**, the current density is almost doubled and there is a 250-mV overvoltage decrease compared to the solution without catalyst. Indeed, the proton reduction with respect to platinum starts at −1.400 V vs. Fc+/Fc, while it starts at −1.750 V vs. Fc+/Fc with catalyst and at −2.000 V vs. Fc+/Fc without catalyst. Finally, the formation of hydrogen is unambiguously demonstrated by gas chromatography analysis during electrolysis performed at −2.200 V vs Fc+/Fc during 4.5 h (Figure S40, Supplementary Materials).

To compare the efficiency of all compounds, the catalytic efficiency (CAT) can be estimated using Equation (1):

$$CAT = \frac{100 \ast \left(f\_{\text{(POM} + 20 \text{ } eq. \text{ CH3COOH)} - f\_{\text{(POM alone)}}\right)}{f\_{\text{(POM alone)}}} \tag{1}$$

Table 1 summarizes the *CAT* values measured for our products at −2.2 V vs Fc+/Fc. Interestingly, the two precursors **SiW10-APTES** and **P2W17-APTES** exhibit similar efficiency. Also, the efficiency of **SiW10-monoB10** and **SiW10-diB10** adducts are lower than that of **SiW10-APTES**, while it is the opposite for **P2W17-diB10**, which appears much more efficient than its parent precursor, in agreement with cyclic voltammetry experiments. Indeed, a less negative reduction potential of the POM part should facilitate the electro-catalytic reduction in protons.


**Table 1.** Electrocatalytic efficiency for the reduction of protons into hydrogen at E = −2.2 V vs. Fc+/Fc for 20 equivalents of CH3COOH added in CH3CN.

In terms of mechanism, three key steps have to be considered: the protonation, the reduction in the catalyst and the transfer of electron towards the protons to give dihydrogen. For protonation step, since catalysis is observed in all compounds, it must occur on the most basic sites, either on the oxo groups of the POM moiety, on the free amine groups in **SiW10-APTES** and **P2W17-APTES** or on boron clusters for **SiW10-diB10** and **P2W17-diB10**. DFT calculations evidence that the most nucleophilic sites are found on the oxo ligands of the POM parts which are consequently the preferential sites for protonation (see Figure 12 and Figure S34 in Supplementary Materials).

**Figure 12.** Two views of the molecular electrostatic potential in atomic units (a.u.) projected onto an electron density isosurface (0.03 e·au<sup>−</sup>3) for **P2W17-APTES**, **P2W17-monoB10** and **P2W17-diB10** species.

For the reduction step, as seen in Figure S41c,d in Supplementary Materials, during electrolysis, the P2W17 derivatives turned to blue as expected for the reduction in such species before returning back colorless when the current is stopped indicating that the reduced POM probably transfers electrons to protons to produce hydrogen. We understand well that if this reduction occurs at higher potential, it should favor the process. **P2W17-diB10** is thus logically the most efficient compound.

To sum up, even if the decaborate cluster is probably not directly involved in the HER process, it plays two indirect roles: (1) the covalent grafting on POMs increases the electronic density on the POM which should facilitate the protonation step, and (2) the covalent grafting can modifies the reduction potential of the POM moieties in POM-borate adducts, which favors the reduction step of the POM species when shifted towards more positive potentials as observed in **P2W17-diB10**.
