2.1. Electrochemical synthesis of the copolymers

The syntheses of the copolymers were achieved using our electropolymerization method, as reported earlier [25, 26]. It corresponds to the addition of the dipyridyl-substituted Lindqvist-type polyoxovanadate [V₆O₁₃{(OCH₂)₃CNHCO(4 − C₅H₄N)}₂]²⁻ (Py-V₆O₁₃-Py) to an electrogenerated dicationic 5,15-ditolylporphyrin (H₂T₂P) or zinc-𝛽-octaethylporphyrin (ZnOEP) obtained by iterative scans between 0 V and +1.60 V versus SCE (Scheme 1) [27].

Using this method, the poly-Py-V₆O₁₃-Py-H₂T₂P and the poly-Py-V₆O₁₃-Py-ZnOEP hybrid copolymers were prepared as described previously [27].

Note that the free base meso-5,15-ditolylporphyrin (H₂T₂P) presents only two meso positions occupied by one substitutable proton at positions C10 and C20 while the zinc-𝛽-octaethylporphyrin (ZnOEP) presents four substitutable meso positions at C5, C10, C15 and C20 (top of Scheme 1). The described electrochemical synthesis of the copolymers uses the previously reported E(EC_NEC_B)_nE process of nucleophilic substitution on porphyrins [17, 18, 19, 26, 27, 28, 29]. As soon as the iterative scans were performed at an anodic potential sufficiently high to allow the formation of the porphyrin dication, the formation of a copolymer coating the working electrode was observed.

The mechanism of the formation of the copolymer can be described such as: first, the porphyrin (abbreviated Porph) radical cation (Porph^∙+, electrochemical step E) and dication (Porph²⁺, electrochemical step E) are electrogenerated. Then, the dication porphyrin Porph²⁺ can be attacked by a two pendant isonicotinate groups of the pyridyl-substituted Lindqvist-type polyoxovanadate [V₆O₁₃{(OCH₂)₃CNHCO(4 − C₅H₄N)}₂]²⁻ (abbreviated Py-V₆O₁₃-Py) at meso-carbon position to yield an isoporphyrin (chemical step C_Nmeso). This later intermediate can be oxidized (electrochemical step E) and the hydrogen atom originally situated on the meso-carbon is released inducing the rearomatization of the porphyrin (chemical step C_B) which leads to the monosubstituted porphyrin Porph-meso-Py⁺ − V₆O₁₃ − Py. At this stage, monosubstituted porphyrin is obtained with one pyridinium covalently connected to the porphyrin and one pendant pyridyl group which is still active for nucleophilic attack. This pendant pyridyl group can further attack oxidized porphyrin continuing the growth of the copolymer. Thus, the polarization of a working electrode at the porphyrin’s second ring-oxidation potential in the presence of Py-V₆O₁₃-Py, leads to the formation of the two hybrid copolymer films with general formula [Py+-V6O132−-Py+-porphyrin]n (bottom of Scheme 1), namely poly-V₆O₁₃-H₂T₂P and poly-V₆O₁₃-ZnOEP.

UV–visible spectra of both copolymers on ITO electrodes presented similar characteristics [27]. They exhibit a large Soret absorption band (Figure 1A), which was red shifted by 15 and 20 nm compared to the ZnOEP and H₂T₂P monomer respectively [27]. It can be explained by intra- or intermolecular excitonic interactions between the porphyrin subunits [17, 30]. The red shifts observed are in agreement with the presence of the disubstituted porphyrin ring by two pyridinium groups [31] in the copolymer.

The films were also examined by scanning atomic force microscopy (AFM) showing tightly packed coils with average diameters of 40–60 nm and a height of 5 nm in the case of poly-V₆O₁₃-H₂T₂P (Figures 1C and C′). The rms surface roughness of the film was 3.5 nm for 1 mm² area. For poly-V₆O₁₃-ZnOEP the diameter and the height of the coils were almost two times larger (Figures 1D and D′) [27].

Thickness of the deposited copolymeric film was measured using AFM by scratching the film with a metallic tip. The thickness increased upon the number of iterative scans as shown in Figure 1B. The values obtained after 20 scans between 0.00 V and +1.60 V were about 23 nm for poly-V₆O₁₃-ZnOEP and 56 nm for poly-V₆O₁₃-H₂T₂P.

2.2. Photocatalytic recovery of silver

The electrochemically deposited poly-V₆O₁₃-ZnOEP or poly-V₆O₁₃-H₂T₂P copolymers were dissolved and removed from ITO with dimethyl formamide (DMF). Subsequently, the copolymer in DMF solution was deposited on a quartz slide, and the DMF solvent was evaporated in air. Then, the quartz slide was plunged in an optical cell containing a deaerated aqueous solution with 8 × 10⁻⁵ mol⋅L⁻¹ Ag₂SO₄ and 0.13 mol⋅L⁻¹ propan-2-ol. The whole sample was illuminated under visible light with a 385 nm cutoff filter to prevent POM photoexcitation allowing only the excitation of the porphyrin. Figure 2 presents the absorption spectra recorded during the experiment performed with the poly-V₆O₁₃-ZnOEP (Figure 2A) or poly-V₆O₁₃-H₂T₂P (Figure 2B) films. The absorbance increases in the whole visible domain during illumination and the solution becomes slightly yellow which reveals the formation of silver nanoparticles [22, 32]. The initial rate of Ag⁺ reduction is 3.4 × 10⁻⁵ mol⋅L⁻¹⋅min⁻¹ for poly-V₆O₁₃-ZnOEP and 3.1 × 10⁻⁵ mol⋅L⁻¹⋅min⁻¹ for poly-V₆O₁₃-H₂T₂P. After almost 95 min (for poly-V₆O₁₃-ZnOEP) and 105 min (for poly-V₆O₁₃-H₂T₂P) of visible light irradiation, the absorption spectrum does not evolve anymore which indicates the end of the reaction.

After removal of the quartz slide covered with the film, the UV–visible spectrum of the solution exhibits a large plasmon band in the whole visible domain with a maximum around 410 nm (Figure 3). This can be attributed to the presence of silver nanoparticles in the solution. No change in the spectrum is observed for one week indicating the good stability of the silver nanoparticles. Moreover, it has been found that the slides of quartz can be reused at least five times with almost no change in the efficiency of the photocatalysis. Silver ions Ag(I) are reduced quantitatively at the interface between water and the copolymeric films without poisoning the surface (as checked by AFM analysis, data not shown).

The TEM micrographs confirmed the formation of silver particles (Figures 4 and 5). The nanoparticles obtained with poly-V₆O₁₃-ZnOEP have an average diameter of 20 nm but are agglomerated (Figure 4).

The presence of Moiré pattern in TEM images of silver particles is due to the presence of metallic silver crystal planes with different spacing and/or orientations. Each silver nanoparticles seems to be crystalline as shown by electron diffraction analysis (Figure 4E).

The 6-fold rotational symmetry displayed by the diffraction spots implied that the faces represented the {111} planes. The first set of spots could be indexed to the formally forbidden 1/3 {422} reflections of face-centered cubic (fcc) silver with a corresponding lattice spacing of 2.48 Å. The second set corresponded to Bragg diffraction from the {220} planes of fcc silver with a lattice spacing of 1.44 Å (1.445 Å in JCPDS file 04-0783). These observations were in agreement with the usual published indexes for silver nanosheets [33, 34]. We explain the observation of 1/3 {422} reflections that are formally forbidden for a perfect fcc structure by the presence of stacking faults [35]. These stacking faults could be caused by bending, which explains the presence of Moiré pattern in TEM images [36]. The EDS spectrum in Figure 4F reveals the presence of metallic Ag in the sample, the C and Cu detected belonging to the TEM grid.

Conversely, the nanostructures obtained with poly-V₆O₁₃-H₂T₂P exhibited larger dispersion in size and shape, because the samples presented not only spherical, elongated particles but also long linear silver nanowires (thickness ca. 20–40 nm and length ca. 200–500 nm) and large silver nanosheets (Figure 5). The EDS spectrum in Figure 5C reveals the presence of metallic Ag in the sample.

Two possible mechanisms can be involved to describe the formation of AgNPs. As shown in Figure 6A, the first mechanism consists in the reduction of the excited porphyrins by propan-2-ol, followed by the cascade electron transfer via the pyridinium to the POM subunit {V₆O₁₃}²⁻ giving the reduced {V₆O₁₃}³⁻. Then, {V₆O₁₃}³⁻ can in turn reduce silver ions. A complexation step between alcohol radical and silver ions can initiate the formation of silver clusters (reactions (4)–(7)) before a thermodynamically possible direct reduction with the reduced POM ({V₆O₁₃}³⁻) or with the alcohol radical (CH₃)₂C∙OH (when the nuclearity of initial silver aggregates is sufficient).

The photoreduction processes should involve the reduction of the excited porphyrin (Porph^∗) by propan-2-ol leading to the formation of the reduced porphyrin (Porph^∙−) and the alcohol radical according to the global reaction (1):

This reaction is thermodynamically favorable. Indeed, the reduction potentials of the excited porphyrins are evaluated according to the relation:

The calculated E(Porph^∗∕Porph^∙−) values are higher than the reported values for (CH₃)₂C∙OH∕ (CH₃)₂CHOH (E = 0.80 V versus NHE [37]).

The second mechanism (Figure 6B) corresponds to a direct intramolecular electron transfer from the excited porphyrins to polyoxovanadate via the reduction of the pyridinium subunit which acts as relay of electron followed by the reduction of silver ions by {V₆O₁₃}³⁻. A fast intramolecular electron transfer agrees with fluorescence quenching of the films observed and is also thermodynamically favorable.

Indeed, the reduction potentials of the excited porphyrins are evaluated according to the relation:

It should be mentioned that in the two proposed mechanisms, the simple consideration of the redox potential of the radical alcohol (CH₃)₂C^∙OH), the reduced porphyrin or the reduced {V₆O₁₃}³⁻ does not allow to explain the subsequent photoreduction of Ag⁺ ions. Indeed, the redox potential of the (Ag⁺∕Ag₁) couple is too low (E° = −1.75 V versus NHE [38]) to allow the direct reduction of Ag⁺ to a single Ag₁ atom either by the reduced porphyrin (E(H₂T₂P^∗∕H₂T₂P^−∙) = 1.60 V versus NHE) or the reduced {V₆O₁₃}³⁻ (E({V₆O₁₃}²⁻∕{V ₆O₁₃}³⁻) = −0.58 V versus NHE).

Referring to radiolytic studies, the reduction of the Ag⁺ might be enabled by a complexation step between alcohol radical and silver ions that initiates the formation of silver clusters (4)–(7) [13, 38, 39]:

Another reasonable explanation is the formation of Ag10 by reaction between Ag⁺ and the alcohol radical (CH₃)₂C^∙OH even if the potential (E((CH₃)₂CO∕(CH₃)₂C^∙OH) = −1.71 V versus NHE [40]) is slightly higher than the one the couple Ag⁺∕Ag₁. This process cannot be excluded.

Finally, as the redox potential of the silver cluster, E(Agⁿ⁺∕Ag_n), increases with the nuclearity n, the direct reduction of large clusters by the reduced porphyrin or the alcohol radical becomes thermodynamically feasible.

Note that the absence of reduced polyoxovanadate at the end of the reaction is in favor of the mechanism (B). It can be explained by an intramolecular back electron transfer between the reduced {Py⁺-V₆O 133−-Py} and the porphyrins oxidized Porph^+∙ which is thermodynamically favorable.

To justify this electron transfer in the case of the mechanism B (Figure 6B), we can calculate the 𝛥_r G of this process using the Rehm–Weller equation: 𝛥G = E(Porph^∙+∕Porph) − E(POM∕POMred) − E_S0–S1 where E(Porph^+∙∕Porph) corresponds to the first oxidation potential of the porphyrin (E(Porph^+∙∕Porph) = +1.25 V versus NHE for H₂T₂P) and E(POM∕POM_red) corresponds to the first reduction potential of the POM (E({V₆O₁₃}²⁻∕{V₆O₁₃}³⁻) = −0.58 V versus NHE) [27]. E_S0–S1 corresponds to the energy of the lowest electronic transition of the porphyrin estimated from its absorbance spectrum in solution. This value leads to negative 𝛥G (− 61.5 kJ⋅mol⁻¹), indicating that the electron transfers between the excited porphyrins and the POMs subunit {V₆O₁₃}²⁻ are thermodynamically possible. These electronic transfers lead to oxidized porphyrins (porph^+∙) and reduced POMs (POM_red, namely {V₆O₁₃}³⁻) via probably a relay including the reduction of the pyridium. Then, oxidized porphyrins porph^+∙ are regenerated by reaction with propan-2-ol which leads to the formation of alcohol radicals (CH₃)₂C^∙OH. This reaction is also really plausible, since the redox potential of the couple (CH₃)₂C^∙OH∕(CH₃)₂CHOH is estimated to 0.80 V versus NHE [37], that is lower than the oxidation potential of the porphyrin.

However, the potential of the couple {V₆O₁₃}²⁻∕ {V ₆O₁₃}³⁻) is too high to permit the reduction of silver ions (E(Ag⁺∕Ag₁) = −1.75 V versus NHE), corresponding to the redox potential of a single Ag(0) atom, different from the one corresponding to metallic silver which is 0.799 V versus NHE. We can also exclude the possibility of a photoinduced electron transfer from the excited porphyrins to the silver ions, because the 𝛥G value of such transfer is positive and is estimated to + 77.2 kJ⋅mol⁻¹.

2.3. Photocatalytic recovery of platinum

To extend the photocatalytic application, we have explored the photocatalysis of the reduction of PtIVCl62− anions using the two covalent porphyrin–Lindqvist-type POM copolymers. The study is conducted under deaerated medium in the presence of 1.6 × 10⁻⁴ mol⋅L⁻¹ H₂Pt^IVCl₆ (to keep the same concentration of metal ions) and 0.13 mol⋅L⁻¹ propan-2-ol. Figure 7 illustrates the change in the UV–Vis absorption spectrum recorded during the visible light irradiation of a quartz slide covered with poly-V₆O₁₃-ZnOEP. The initial absorbance of PtIVCl62− anions decreased during illumination and disappeared completely after 100 min, however, because the Pt nanoparticles do not possess plasmon band or other type of UV–visible absorbance, it is impossible to monitor the creation of the Pt nanoparticles from the UV–visible spectra.

After 24 h of illumination, a yellow sediment on the bottom of the cuvette is obtained, and after stirring, the absorbance in the whole domain increases.

The TEM micrographs confirm the formation of the Pt nanoparticles. The diameter of the basic nanoparticle subunit is 1.0–1.5 nm which forms the closed-packed nanoparticles with a diameter of 10–20 nm (Figure 8). The EDS spectrum in Figure 8E indicates the presence of merely pure Pt in the sample.

Figure 9 presents the absorption spectra of a deaerated aqueous solution containing 1.6 × 10⁻⁴ mol⋅L⁻¹ H₂PtCl₆ and 0.13 mol⋅L⁻¹ propan-2-ol with poly-V₆O₁₃-H₂T₂P covered quartz slide during visible light irradiation. The initial absorbance of PtIVCl62− anions disappeared completely after 110 min which is a little slower than for poly-V₆O₁₃-ZnOEP. After 24 h of illumination, we obtained a uniform and faint yellow solution which is different than the sediment obtained when using poly-V₆O₁₃-H₂T₂P. The TEM micrographs of the obtained Pt nanostructures are presented in Figure 10. The Pt nanoparticles showed again closed-packed spherical structure similar to the ones obtained using poly-V₆O₁₃-ZnOEP film. Nevertheless, such structures are also mixed with large nanosheets. The obtained Pt nanomaterial is in this case inhomogeneous in size and shape. The EDS spectrum in Figure 10E shows the presence of Pt nanoparticles in the sample.

After almost 3 h for poly-V₆O₁₃-ZnOEP and 4 h for poly-V₆O₁₃-H₂T₂P of visible light irradiation, the absorption spectrum does not evolve anymore which indicates the end of the reaction. The duration of the photocatalytic reduction is longer in comparison to the silver recovery which can be explained by the number of electron implied: 4 for the Pt(IV) versus 1 in the case of the Ag(I).

We assume that the mechanism for the photocatalysis Pt(IV) is much more complicated than the one proposed for the reduction of Ag(I) ions probably due to the fact that 4 electrons must be transferred between the catalysts and the metallic ion. We will therefore not provide a mechanism in this present case although there is a good chance that the first step is similar to that of the reduction of Ag(I) ions, that is the electron transfer from the excited porphyrin via the pyridinium to the polyoxovanadates (mechanism B).

Last but not least, the heterogeneous photocatalysis can also be repeated at least five times while keeping the same efficiency. Platinum ions are reduced quantitatively and no deposition at the surface of the slide was observed (as checked by AFM analysis, data not shown).