Trapping of Ag+ into a Perfect Six-Coordinated Environment: Structural Analysis, Quantum Chemical Calculations and Electrochemistry

Self-assembly of (Bu4N)4[β-Mo8O26], AgNO3, and 2-bis[(2,6-diisopropylphenyl)-imino]acenaphthene (dpp-bian) in DMF solution resulted in the (Bu4N)2[β-{Ag(dpp-bian)}2Mo8O26] (1) complex. The complex was characterized by single crystal X-ray diffraction (SCXRD), X-ray powder diffraction (XRPD), diffuse reflectance (DR), infrared spectroscopy (IR), and elemental analysis. Comprehensive SCXRD studies of the crystal structure show the presence of Ag+ in an uncommon coordination environment without a clear preference for Ag-N over Ag-O bonding. Quantum chemical calculations were performed to qualify the nature of the Ag-N/Ag-O interactions and to assign the electronic transitions observed in the UV–Vis absorption spectra. The electrochemical behavior of the complex combines POM and redox ligand signatures. Complex 1 demonstrates catalytic activity in the electrochemical reduction of CO2.

In our recent studies, we paid particular attention to the Ag + coordination environment in novel [β-{AgL} 2 [46][47][48]. These works focus on the Ag + coordination behavior and changing of the coordination number (CN) initiated by the auxiliary ligand L. In all cases, we used monodentate ligands for L, and found typical coordinations with CN 3 and 4. Moreover, Ag + can form long-range contact with an additional DMF molecule, formally increasing its CN [46]. This situation may be more complicated without any auxiliary ligand when different monomeric and polymeric species have been detected [47]. Concerning bidentate ligands, either rigid (e.g., BIANs) or flexible (e.g., bpy and its derivatives) diimines with redox nature can be used, which opens numerous ways to construct new coordination networks, with relevance in the field of coordination polymers preparation. In this work we report a new complex, viz. (Bu 4 N) 2 [β-{Ag(dppbian)} 2 Mo 8 O 26 ], indicating a unique Ag coordination environment. The electrochemical activity of the complex toward CO 2 reduction was studied.
In our recent studies, we paid particular attention to the Ag + coordination environment in novel [β-{AgL}2Mo8O26] 2− complexes, where the [β-Mo8O26] 4− polymolybdate acts as a ditopic doubly tetradentate ligand towards Ag + [46][47][48]. These works focus on the Ag + coordination behavior and changing of the coordination number (CN) initiated by the auxiliary ligand L. In all cases, we used monodentate ligands for L, and found typical coordinations with CN 3 and 4. Moreover, Ag + can form long-range contact with an additional DMF molecule, formally increasing its CN [46]. This situation may be more complicated without any auxiliary ligand when different monomeric and polymeric species have been detected [47]. Concerning bidentate ligands, either rigid (e.g., BIANs) or flexible (e.g., bpy and its derivatives) diimines with redox nature can be used, which opens numerous ways to construct new coordination networks, with relevance in the field of coordination polymers preparation. In this work we report a new complex, viz. (Bu4N)2[β-{Ag(dppbian)}2Mo8O26], indicating a unique Ag coordination environment. The electrochemical activity of the complex toward CO2 reduction was studied.

Synthesis and Structural Features
Mixing (Bu4N)4[β-Mo8O26] and AgSO3CF3 in DMF results in the coordination of Ag + with [β-Mo8O26] 4− , leading to [{Ag(DMF)x}2Mo8O26] 2− (x = 1, 2) and other species [47]. The addition of dpp-bian to such a solution provided an orange precipitate. Brown blockshaped crystals of 1 suitable for X-ray diffraction analysis were obtained by diffusion of diethyl ether into a solution of 1 in DMF. The resulting crystalline complex 1 crystallized in monoclinic crystal system (C2/c space group) with the following unit cell parameters: a = 38.5675 (16) Table S1). The main building blocks of the structure are hybrid [β-{Ag(dppbian)}2Mo8O26] 2− anions ( Figure 1). It should be noted that using other bians (e.g., tmp-bian, or p-Cl-bian) with less bulky substituents always provided the corresponding [Ag(bian)2]SO3CF3 complexes. The most proposed cause of this is in bulkiness of the substituents.
The main structural feature in this complex is the unusual coordination environment of Ag + (Figure 2, Table S2) consisting of four oxo ligands from polyoxomolybdate and two N-atoms from dpp-bian. The Ag-O bond distances ranged from 2.495 to 2.569 Å, while Ag-N distances were 2.386 and 2.461 Å. The asymmetric coordination of the dpp-bian ligand was likely caused by packing effects. The Ag + , therefore, resides inside a coordination environment that has six coordination bonds, as demonstrated by the quantum chemical calculations (vide postea). Unlike many other structures that involve multiple Ag···L contacts, the bond lengths vary in a relatively narrow range (0.18 Å) and cannot be divided into primary and secondary interactions. The coordination polyhedron of Ag + is irregular, neither octahedral nor trigonal prismatic, but is similar to a monocapped octahedron, It should be noted that using other bians (e.g., tmp-bian, or p-Cl-bian) with less bulky substituents always provided the corresponding [Ag(bian) 2 ]SO 3 CF 3 complexes. The most proposed cause of this is in bulkiness of the substituents.
The main structural feature in this complex is the unusual coordination environment of Ag + (Figure 2, Table S2) consisting of four oxo ligands from polyoxomolybdate and two N-atoms from dpp-bian. The Ag-O bond distances ranged from 2.495 to 2.569 Å, while Ag-N distances were 2.386 and 2.461 Å. The asymmetric coordination of the dppbian ligand was likely caused by packing effects. The Ag + , therefore, resides inside a coordination environment that has six coordination bonds, as demonstrated by the quantum chemical calculations (vide postea). Unlike many other structures that involve multiple Ag···L contacts, the bond lengths vary in a relatively narrow range (0.18 Å) and cannot be divided into primary and secondary interactions. The coordination polyhedron of Ag + is irregular, neither octahedral nor trigonal prismatic, but is similar to a monocapped octahedron, lacking a vertex in the capped face ( Figure 3). The analysis with SHAPE (http://www.shapesoftware.com/ (accessed on 3 September 2022)) shows that this is a transition state between a pentagonal pyramid and a trigonal prism (Table S3). lacking a vertex in the capped face ( Figure 3). (http://www.shapesoftware.com/ (accessed on 3 Septem transition state between a pentagonal pyramid and a trigo   lacking a vertex in the capped face ( Figure 3). The analysis with SHAPE (http://www.shapesoftware.com/ (accessed on 3 September 2022)) shows that this is a transition state between a pentagonal pyramid and a trigonal prism (Table S3).  Similar considerations apply, mutatis mutandis, for the Ag-N bond distances. We can describe this situation as Ag + trapped inside a {N2O4} cage. It is interesting to compare the observed bonding arrangement around Ag + with other reported data. We used the AgO4N2 coordination environment as the criterion for the structural search in CCDC. The relevant entries are summarized in Table 1. All hits were divided into five groups based on the ligand nature. The first one included complexes with sterically fixed pincer-type ligands based on pyridine as a central fragment and two O-donor arms. In such complexes, the Ag-N bonds were significantly shorter than Ag-O bonds. Similar considerations apply, mutatis mutandis, for the Ag-N bond distances. We can describe this situation as Ag + trapped inside a {N 2 O 4 } cage. It is interesting to compare the observed bonding arrangement around Ag + with other reported data. We used the AgO 4 N 2 coordination environment as the criterion for the structural search in CCDC. The relevant entries are summarized in Table 1. All hits were divided into five groups based on the ligand nature. The first one included complexes with sterically fixed pincer-type ligands based on pyridine as a central fragment and two O-donor arms. In such complexes, the Ag-N bonds were significantly shorter than Ag-O bonds. However, in the case of HATSAZ, two tridentate ligands completed the coordination sphere of Ag + . Therefore, the discrimination between Ag-O and Ag-N bond distances cannot be done properly.
The second group combined complexes with crown-ethers and cryptands. In this environment, Ag + was fixed inside the cages wherein Ag-O contacts were largely determined by ligand geometry. The Ag-N bond distances were shorter than Ag-O, except for TIBQIE, wherein the coordination sphere of Ag + may include additional Ag-π donation (c.a. at 2.83 Å) and Ag-F contacts (2.71 Å). In this case, the bonding nature should be further studied through quantum chemical calculations, to shed light on the Ag + complexes with very high coordination numbers. Remarkably, for such a group of complexes, coordination numbers were lower than the number of donor atoms.
The third group was constituted by POM-based complexes. In this group, an uncommon Ag + coordination by the POM was observed in the case of ZULXOV, wherein there were two types of Ag (namely, Ag1 and Ag2), with a significant Ag-X bond distance distribution. Curiously, the values of d(Ag1-O) ranging from 1.960 to 2.097 Å were too short for Ag and might be attributed to the presence of a 3d metal.
Within the fourth group, no anomalies in the Ag + coordination were revealed. The comparison of the bonding distribution around Ag followed the same trends as in the previous groups.
In the fifth group, more structures featured anomalous geometries. In most cases, the atom assigned as Ag looks more like a 3d metal, rather than Ag. In the case of ELEGIM structure, the "Ag" atom had CN 9 and might be a lanthanide.
From this structural analysis, it can be seen that the significant difference between Ag-N (shorter) and Ag-O (longer) bond distances is a rule. There are only two cases wherein this rule is infringed, viz. in SENJIE and TIBQIE. This might be due to a wrong refinement of N and O atoms or impurities in the initial ligands. High coordination numbers can be achieved only with the use of pincer ligands with fixed geometry, and there is no possibility to reach CN 6 (4 O and 2 N donor atoms) for Ag + using mono or bidentate ligands. The example reported in this work, therefore, represents a unique case in the literature, which reveals a hitherto hidden aspect of Ag + coordination behavior.

Quantum Chemical Calculations
The natural bond orbital (NBO) analysis was conducted on the electron density of the optimized cluster to clarify the nature of the Ag-involving electron density delocalization interactions (ESI , Tables S4 and S5). Ag may play the role both of electron density donor and acceptor. As a donor, the NBO of Ag involved are core orbitals (CR 100% p in nature; with occupancy > 1.998) or lone pairs (LP 100% d in nature; with occupancy > 1.990). From Ag, the electron density shifted to empty virtual Rydberg orbitals (RY*) of C atoms or bond orbital BD*(C-C). When Ag plays the role of acceptor, the virtual NBO of Ag involved are 4 LP* and 6 RY*. Their nature and occupancy is variegated; however, 2 LP* are almost 100% p orbitals; 1 LP* is s 33% p 67% ; 1 LP* is s 65% p 35% ; RY* orbitals involve d orbitals (in particular, one RY* is 100% d). In these cases, the donors are CR belonging to O, N, C, or Mo atoms, BD(C-C), BD(C-N), or BD(C-H).
The non-covalent interaction (NCI) analysis confirmed that Ag has CN = 6, showing four strongly attractive interactions (Figure 4, blue regions) with as many O atoms (red arrows) + 2 with as many N atoms (blue arrows), involving each Ag atom. In those regions, bond critical points (BCPs) were identified by means of the topological analysis based on the atoms-in-molecules (AIM) theory (ESI, Table S6).
The UV-Vis absorption spectrum (Figure 5a) indicates two main bands, which do not represent two single transitions, but cover multiple transitions. The four transitions with higher intensities are found at 478 and 303 nm, each of which shows double degeneration. They are characterized by a complex composition in terms of molecular orbitals, and we hence decided to analyze their nature by means of natural transition orbitals (NTO) analysis (Figure 5b). Both absorption bands corresponded to CT transitions, and in both cases, Ag d orbitals donated electron density. The band at 478 nm was mostly Ag → L (the [Mo 8 O 26 ] 4− ligand was almost excluded from this electronic transition), and is dipolar in nature. Contrary to this, the other band at 303 nm involved the octamolybdate and was quadrupolar in nature.

Electrochemistry
The structure of 1 combines a redox-active polymolybdate, Ag + , and non-innocent dpp-bian ligand. Since the [{Ag(L)x}2Mo8O26] 2− complexes are unstable in solution [46], their electrochemical behavior can be correctly studied only in solid state.
The redox behavior of 1 in the paste electrode in acetonitrile was studied with cyclic

Electrochemistry
The structure of 1 combines a redox-active polymolybdate, Ag + , and non-innocent dpp-bian ligand. Since the [{Ag(L) x } 2 Mo 8 O 26 ] 2− complexes are unstable in solution [46], their electrochemical behavior can be correctly studied only in solid state.
The redox behavior of 1 in the paste electrode in acetonitrile was studied with cyclic voltammetry (Figure 6, black curve). For the sake of comparison, a cyclic voltammogram for the (Bu 4 N) 4 [Mo 8 O 26 ] under the same conditions was recorded ( Figure 6, red curve). The comparison of paste electrodes behavior and blank solution is presented in Supplementary  Materials (Figure S2). The corresponding data are provided in Tables 2 and 3. In both cases, a large number of quasi-reversible and irreversible electrochemical transitions were observed, viz. from six (for the precursor) to eight (for the complex). The redox properties of the precursor were due to its ability to undergo multistage reduction processes, which is a characteristic feature of POMs. The presence of an additional redox-active dpp-bian fragment in the complex, known for its ability to multielectron reduction, increased the number of electrochemical transitions.       They assigned these processes to Mo 6+ reduction. However, in our opinion, ligand-centered redox reactions or redox reactions involving a heterometal that might occur in this region cannot be excluded. Indeed, for the initial (Bu 4 N) 4 [Mo 8 O 26 ], we did not observe any reduction peaks in the positive region. Only oxidation peaks (peaks 4, 5, and 6 in the red curve in Figure 6) were found in the anodic curve and they can be attributed to the oxidation of electrochemically generated reduced species during cycling. On the other hand, the absence of cathodic peaks in the positive region for (Bu 4 N) 4  On the other hand, solid-state electrochemistry of bulk molybdates has been attracting increasing interest for possible technological applications, such as MgMoO 4 , which has been used for supercapacitor and battery applications [51], or CoMoO 4 , which has shown a specific capacitance of 79 F·g −1 , an energy density of 21 W h·kg −1 , and excellent cycling stability (retaining over 75% of its initial capacitance after 2000 cycles), which makes it a very promising candidate for large-scale energy-storage applications.
Recently  Figure 7 shows a CV series recorded in the range of −1.4/1.4 V relative to a saturated Ag/AgCl reference electrode at a 10 mV/s scan rate for a paste electrode containing 1. The red arrow in Figure 7 shows the cathodic potential sweep direction; the beginning of the first cycle corresponds to the 0.50 V point. In the region from approx. −0.9 to −1.2 V, there is a cathodic plateau that is most likely related to CO 2 reduction. The black arrow on this plateau displays the sequential increase in current from the first to subsequent scans. At potentials below −1.2 and above 1.0 V, there was a sharp increase in current due to HER and OER, respectively. The voltammograms also revealed regions at potentials of~0.10 (cathodic current peaks) and~0.50 V (anodic current peaks), which are responsible for the presence of complex 1 in the paste. The dependence of the current in the peak on the square root of the potential scan ( Figures S3 and S4 in Supplementary Materials) indicates the diffusion nature of the current for the redox reaction of complex 1 and its stability.
plays the sequential increase in current from the first to subsequent scans. At potentia below −1.2 and above 1.0 V, there was a sharp increase in current due to HER and OE respectively. The voltammograms also revealed regions at potentials of ~0.10 (cathod current peaks) and ~0.50 V (anodic current peaks), which are responsible for the presen of complex 1 in the paste. The dependence of the current in the peak on the square root the potential scan ( Figures S3 and S4 in Supplementary Materials) indicates the diffusio nature of the current for the redox reaction of complex 1 and its stability.  Comparison of CVs (Figure 8) recorded in the absence (red curve, blank) and in the presence of complex 1 (grey curve) in a 1.0 M KHCO 3 solution saturated with CO 2 or in a solution unsaturated with CO 2 (blue curve) showed that, in the region of cathodic currents, the area (on the gray curve) appears with potentials in the region of−1.16/−0.87 V (relative to Ag/AgCl), which, apparently, is responsible for the electrocatalysis of CO 2 reduction to less deeply oxidized products. The conversion to the RHE scale provided the potential ranging from −1.35 to −1.06 V. According to the literature data, the direct reduction involved the following first step: Molecules 2022, 27, x FOR PEER REVIEW 11 of 19 total difference between the red (blank + CO2) and gray currents was 34 mC or 176 mC/cm 2 . The cathodic peak at approximately −0.5 V observed on the red curve ( Figure 8) refers to the Nafion matrix containing acidic SO3H groups capable of ion exchange [53]. This peak is slightly shifted towards positive potentials for the paste containing 1 (up to −0.25 V). On the other hand, in the potential region of 0.1-0.2 and 0.5-0.6 V (gray and blue curves), cathodic and anodic peaks were observed, marking the presence of 1 in the pastes. When CO 2 was saturated in the presence of 1, an increase in the cathodic current was observed (Figure 8). In this case, the difference in an area between the blue (1) and gray (1 + CO 2 ) cathodic currents reached 5.0 mC or 29 mC/cm 2 in the range of −1.16/−0.87 V. The total difference between the red (blank + CO 2 ) and gray currents was 34 mC or 176 mC/cm 2 .
The cathodic peak at approximately −0.5 V observed on the red curve ( Figure 8) refers to the Nafion matrix containing acidic SO 3 H groups capable of ion exchange [53]. This peak is slightly shifted towards positive potentials for the paste containing 1 (up to −0.25 V). On the other hand, in the potential region of 0.1-0.2 and 0.5-0.6 V (gray and blue curves), cathodic and anodic peaks were observed, marking the presence of 1 in the pastes.
Typically, currents observed under anhydrous conditions and in the solvent-water mixtures differ due to the presence of protons accessing alternative reduction mechanisms. It is well known that, without external proton sources, electrocatalytic CO 2 reduction can proceed: (i) as reductive disproportionation forming carbonate and CO, or (ii) as dimerization of CO 2 •− anion-radical to oxalate, while proton addition opens pathways to numerous reduction products, such as CO, formic acid, methanol, methane, and higher carbon products, e.g., ethylene, ethanol, and acetic acid [54][55][56][57][58][59]. The possible reaction mechanism and catalytic activity of Ag + coordinated by different BIANs are currently the object of a dedicated investigation in our group, both from an experimental and computational point of view. However, in the specific case investigated in the present work, taking into account the activity of the complex in the solid state, the through-space electron transfer from the reduced complex to CO 2 is the only possible explanation. In fact, the direct coordination of CO 2 cannot be hypothesized due to lack of coordination space near to Ag + caused by the bulky side groups installed onto BIAN molecules.
To check this hypothesis, we calculated the first reduced state of [β-{Ag(dpp-bian)} 2 Mo 8 O 26 ]. The obtained spin density profile is shown in Figure 9. According to the calculated data, for the first reduced state (induced during electroreduction) the spin density was mostly located over two BIAN ligands. Moreover, its small portion affects both Ag atoms. Thus, we can conclude that electron transfer to CO2 can be indeed realized from the combination of Ag and BIAN electronic states.
The formation of formic acid as a possible product of CO2 reduction could be judged from a comparison of CV data for pure HCOOH in 0.1 M Na2SO4 and the CV data for a paste electrode containing 1 in the presence of CO2 under the same conditions ( Figure 10). According to the calculated data, for the first reduced state (induced during electroreduction) the spin density was mostly located over two BIAN ligands. Moreover, its small portion affects both Ag atoms. Thus, we can conclude that electron transfer to CO 2 can be indeed realized from the combination of Ag and BIAN electronic states.
The formation of formic acid as a possible product of CO 2 reduction could be judged from a comparison of CV data for pure HCOOH in 0.1 M Na 2 SO 4 and the CV data for a paste electrode containing 1 in the presence of CO 2 under the same conditions ( Figure 10). According to the calculated data, for the first reduced state (induced during electr reduction) the spin density was mostly located over two BIAN ligands. Moreover, i small portion affects both Ag atoms. Thus, we can conclude that electron transfer to CO can be indeed realized from the combination of Ag and BIAN electronic states.
The formation of formic acid as a possible product of CO2 reduction could be judge from a comparison of CV data for pure HCOOH in 0.1 M Na2SO4 and the CV data for paste electrode containing 1 in the presence of CO2 under the same conditions (Figure 10

General Information
(NBu 4 ) 4 [β-Mo 8 O 26 ] and dpp-bian [60] were synthesized according to the standard procedures. All other reagents were of commercial purity (Sigma Aldrich, Buchs, Switzerland) and were used without any purification. Organic solvents (DMF and diethyl ether ((C 2 H 5 ) 2 O)) were dried by standard methods before use. Elemental analysis was carried out on a Eurovector EA 3000 CHN analyzer. IR spectrum was recorded in the 4000-300 cm −1 range with a Perkin-Elmer System 2000 FTIR spectrometer (KBr pellets). DR was carried out on a UV-Vis-NIR spectrophotometer UV-3101 PC in an integrating sphere in a range of 240-800 nm (gap width 5 nm) at room temperature.

SCXRD
Single-crystal XRD data for 1 were collected with a Bruker D8 Venture diffractometer (Bruker, Billerica, MA, USA) with a CMOS PHOTON III detector and IµS 3.0 source (Mo Kα radiation, λ = 0.71073 Å, Montel mirror optics, nitrogen flow thermostat). ϕand ω-scan techniques were employed to collect data for crystal structure refinement; full ϕ-scans were used for investigation of temperature dependence of unit cell parameters in 80-280 K range (increasing of temperature by 5-10 K steps was used). The unit cell parameters were refined using APEX3 Determine Unit Cell routines with the refinement of beam position instrumental parameters and may have similar systematic errors (~0.01 Å,~0.1 • ) [61]. Integration and absorption corrections were applied with the use of  [62] and refined by full-matrix least-squares treatment against |F| 2 in anisotropic approximation with SHELX 2014/7 [63] in the ShelXle program [64]. The main geometrical parameters are present in Table S2. H-atoms were refined in calculated positions. One of TBA + was disordered over two closed positions with 0.4/0.6 occupancies. C-atoms of such parts were refined isotropically.
The crystallographic data were deposited In the Cambridge Crystallographic Data Centre under the deposition codes, CCDC 2189249.

Quantum Chemical Calculations
The molecular geometry of the monomer and model dimer was fully optimized using density functional theory (DFT), as well as using the long-range corrected hybrid functional CAM-B3LYP [65] coupled with the Pople 6-311+G** triple-ζ basis set for H, C, N, and O [66][67][68][69][70][71][72][73][74], the LANL2DZ double-ζ basis set on the Mo and Ag valence shell, and the LANL2DZ effective core potential on the core [75][76][77][78]. The vibrational frequencies and thermochemical values were subsequently computed at the same levels of theory, within the harmonic approximation, at T = 298.15 K and p = 1 atm; no imaginary frequencies were found.
The UV-Vis absorption spectra for the equilibrium geometries were calculated at a time-dependent density functional theory (TD-DFT) level, accounting for S 0 → S n (n = 1 to 120) transitions. The nature of the vertical excited electronic state was analyzed via natural transition orbital (NTO) analysis [93].
The integration grid was set to 250 radial shells and 974 angular points. The convergence criteria for the self-consistent field were set to 10 −12 for the RMS change in the density matrix and 10 −10 for the maximum change in the density matrix. The convergence criteria for optimizations were set to 2 × 10 −6 a.u. for the maximum force, 1 × 10 −6 a.u. for the RMS force, 6 × 10 −6 a.u. for the maximum displacement, and 4 × 10 −6 a.u. for the RMS displacement.
The location of the CPs and subsequent calculation of SF values were performed using a modified version of the PROAIMV program [94]. The NCI analyses were conducted and plotted using homemade code. All the other calculations were performed using the GAUSSIAN G16.A03 package [95].

Electrochemistry
The CV measurements were carried out in a three-electrode cell with a paste electrode, an auxiliary electrode, and a reference electrode using a potentiostat-galvanostat P-45X (Elins, Chernogolovka, Russia) at 25 • C. The description of the cell and the measurement procedure have been previously reported [96][97][98]. The composition of the paste included carbon powder:EAS:Nafion = 100:10:4 mg (Nafion, 10% aqueous dispersion calculated on dry matter) for the complex 1 and the precursor. Based on the weights of the dry mixture, the concentration of 1 in the paste was~0.030 and the precursor~0.070 mol/kg. Initially, the measurements were carried out in the acetonitrile solutions containing 0.10 M Bu 4 NPF 6 (supporting electrolyte). In total, in these experiments, three types of pastes were tested, viz. the pastes contained the complex 1, the precursor, and the blank (without EAS). The blank paste contained only the carbon powder and the Nafion dispersion in the ratio 100:4 mg. CO 2 reduction experiments were performed in an aqueous 1.0 M KHCO 3 solution. Based on the weight of the dry mixture, the concentration of 1 in the paste was 0.030 mol/kg. The blank paste contained only the carbon powder and the Nafion dispersion (as in experiments with acetonitrile solutions). First, the electrochemical experiments were carried out in solutions saturated with CO 2 . Then, the measurements were continued in the same solution, but without its saturation with CO 2 . Finally, to compare the obtained results, measurements were also carried out on the blank pastes.
Some additional experiments were done with paste electrodes in an aqueous solution of 0.10 M Na 2 SO 4 . The composition of the paste was the same as in the previous series. In this case, we tried to find the difference in the CV behavior between the blank paste in solution with the addition of formic acid (0.038 M) and the paste after saturation with CO 2 .

Conclusions
Three components, and five building blocks' self-assembly, produced a [{Ag(dppbian)} 2 Mo 8 O 26 ] 2− complex which had a non-typical coordination environment for Ag + cation, wherein Ag had six comparable distances to both O and N donor atoms. There was no remarkable difference between all Ag-N and Ag-O bond lengths; however, the Ag-N bond was revealed to be covalent in nature, whereas Ag-O had a predominant ionic character. Quantum-chemical calculations confirmed the CN of Ag + cation and indicated the charge-transfer character of the two bands in the UV-Vis absorption spectrum. Solid (Bu 4 N) 2 [β-{Ag(dpp-bian)} 2 Mo 8 O 26 ] showed rich electrochemistry which involved both organic components and octamolybdate ligands with an electrochemical window of ca. 2.5 V between two quasi-reversible processes in both extremes of the cycling route. The complex also demonstrated a catalytic activity toward electrochemical CO 2 reduction. CO 2 activation most likely occurs at the Ag-bian redox site.