All chemicals and reagents were analytical grade. N′N dimethylformamide (DMF) and Zinc Nitrate were purchased from ACE Chemicals, South Africa. Titanium (IV) oxide nanoparticles (21 nm primary particle size), Fluorine doped tin oxide (FTO) (surface resistivity of ∼7Ω/sq) coated glass substrates, ethanol, acetone, Iron(III) chloride FeCl₃·6H₂O, terephthalic acid H₂BDC (benzene-1,4-dicarboxylic acid), 2-aminoterephthalic acid H₂N-BDC (2-aminobenzene-1,4-dicarboxylic acid), ethanol, L-cysteine, R-camphorsulfonic acid, S- camphorsulfonic acid, acetonitrile, graphite powder (<20 µm) and potassium chloride (KCl) were purchased from Sigma, Germany and used with no further purification. Aqueous solutions were prepared using milliQ water.

Original Fe and Zn based doped MOFs (eight different compounds) were synthesized in a Digital oven, Labotec EcoTherm Economy Oven, South Africa. Details concerning the synthesis are reported in the Supporting Information. Infrared measurements were performed using the FT-IR spectrometer, Perkin-Elmer Spectrum Two, United States, equipped with a transmission mode interferometer Dynascan, Perkin-Elmer, United States. IR spectra were recorded in the range 400–4,000 cm⁻¹ with a spectral resolution of 2 cm⁻¹ and 32 scans. Powder X-ray diffraction (PXRD) measurements were carried out using a Bruker D2 phaser XRD machine, United States, with DIFFRAC-SUITE software for controlling the instrument and analyzing the data. Scanning electron microscope (SEM) images were investigated using a Scanning Electron Microscope, SU3800, Hitachi, Japan, equipped with an EDS detector UltimMax, Oxford instrument, United Kingdom. For PXRD measurements MOF were immobilised on a copper surface by dropcast of an acetonitrile dispersion dried in a nitrogen stream. UV-Vis spectra were recorded by using a UV-1900i Shimadzu UV-Vis Spectrophotometer using 1 cm path length polystyrene cuvette for photocatalytic measurements. CH Instruments Electroanalyser model 608E and electrochemical workstation (potentiostat/galvanostat) PGSTAT204, Metrohm AutoLab, Switzerland, were used to perform electrochemical measurements, driven by Nova2.1 software.

The MOF electrochemical behaviour was investigated within the so-called solid-state electrochemical paradigm by using an in-house flat-cell of original design in a typical three-electrodes electrochemical cell set-up. The GCE/MOF served as the working electrode (WE), while a coiled Pt wire and an Ag/AgCl (saturated KCl) electrode served as the counter (CE) electrode and reference electrode (RE), respectively. The special flat-cell configuration of the cell, developed in the past for ECALE depositions (Forni et al., 2000; Innocenti et al., 2001; Giurlani et al., 2018; Salvietti et al., 2018; Giurlani et al., 2020) and adapted for solid-state measurements, features the working electrode placed at the bottom, to help maintaining the graphite-powder/MOF mixture in position. A suitable arrangement was adopted (Supplementary Figure S1), a cylindrical Teflon cell featuring a hole (7 mm diameter) in the bottom was used in a vertical configuration, where the glassy carbon electrode (GC) (1.5 cm diameter) was tightened from below. The GC electrode is held in place by a 1.5 cm diameter screw, which also ensures the electrical contact. A Teflon ring was used to ensure no solution leakage from the cell. The CE and RE were placed at the top of the cylinder in contact with the electrolytic solution. Prior to surface modification, the GC working electrode was polished using 0.05 μm alumina slurries on a polishing cloth; it was cleaned carefully, and then sonicated for 10 min in milliQ water to clean the surface and remove any polishing residue. Finally, the GC working electrode was dried in a nitrogen stream. The effective cell set-up and GC electrode activation procedure was cross-checked by recording CVs in a solution of 1 mM ferricyanide in 0.1 M KCl, in the −0.3 to +0.7 V potential range, we obtained a ΔEp (peak-to-peak separation between the oxidation and reduction potentials) at 25 mVs⁻¹ of 60 mV. MOFs crystals (1.0 mg) were placed over the WE and covered with of graphite powder (10.0 mg). A drop of acetonitrile is added to make the mixture compact and then the GC modified electrode was left “to dry” for 10 min in a nitrogen atmosphere. CV measurements were recorded in the −0.8 to +1.2 V potential range vs. Ag/AgCl KCl _sat. Throughout the paper all the potential values are expressed with reference to the Ag/AgCl/KCl_sat reference electrode (RE), unless otherwise stated.

TiO₂ nanoparticulate films were deposited on fluorine-doped tin oxide, FTO (surface resistivity of ∼7 Ω/sq) coated glass, using the electrophoretic deposition (EPD) technique. A suspension of TiO₂ nanoparticles (NPs) was prepared by dispersing 0.4 g TiO₂ NPs in 40 mL of de-ionized water. Prior to making dispersions, TiO₂ nanoparticle powders were heated at 570 K for 1 h. The mixture was stirred overnight to ensure homogeneity. Prior to nanoparticle deposition, the FTO substrates were boiled in acetone for 15 min, followed by 15 min of boiling in ethanol, and finally rinsed with de-ionized water. After rinsing, the substrates were dried in the air for 15 min. Electrophoretic deposition (EPD) was performed with a CHI electrochemical analyser, model 608E, using the Chronopotentiometry mode. During EPD, the suspension was continuously stirred using a magnetic stirrer. After completion of the last cycle, the electrodes were annealed for 2 h at 400K. Functionalization of FTO/TiO₂ surfaces to yield the final FTO/TiO₂/MOF photoelectrodes was achieved by drop casting 20 µL of 1 mM slurry MOF solutions dissolved in ethanol onto the FTO/TiO₂ surfaces. The electrodes were then left to dry under controlled humidity for a period of 3 weeks.

Calculations were performed in the framework of ab initio quantum mechanical based methods with Gaussian (Frisch et al., 2017) and Quantum Espresso (Giannozzi et al., 2009) programs, using C1 symmetry and unrestricted wave function. Chemcraft (Andrienko, 2023) is used to display molecular structures and molecular orbitals. Molecular orbitals were obtained by full optimization carried out at UB3LYP/6-31G(d) levels of theory. Geometry optimization was carried out by using Barone and Cossi’s polarizable conductor model (CPCM) (Cossi et al., 2003) to account for N,N-dimethylformamide interaction.

Figures 1A, B show CV curves recorded for the chiral “doped” MOF MIL53 S-CSA and Zn-MOF S-CSA, respectively. The black solid line is the control CV recorded using graphite powder only, i.e., the “blank” baseline. It must be noted that a prominent capacitive contribution in the—0.7–1.0 V potential range yields a quite regular “rectangular box” CV pattern as depicted in Figures 1A, B. The red solid curves in Figures 1A, B are the CVs recorded after mixing the graphite powder with the chiral MOF (details of the electrode preparation are in the experimental section). The CV of the MIL53 S-CSA MOF features two redox peaks (labelled as Eox and Ered in Figure 1A) centred at about 0.5 V. This suggests that the redox process underlying the presence of the two current peaks in Figure 1A can be assigned to the Fe (III)/Fe(II) redox couple. The peak-to-peak separation is about 0.3 V, which is rather large, such a potential difference is due to the solid-state nature of the electroactive system. Moreover, iron is in an octahedral coordination (a quite stable state environment).

In addition, the overall MOF framework is based on organic molecules which are well known to behave more closely to a dielectric than to a conductive material, this further hinders the Fe(III)/Fe(II) charge transfer kinetics. Supplementary Figure S2 shows the CV curves obtained for the remaining four Fe based MOFs studied in this paper and Supplementary Figure S3 shows the CV curves for ferrocene a well-studied material used as a control and compared with MIL53 S-CSA. Table 1 summarizes the electrochemical results.

Please note the effect of “doping” due to the presence of different organic “dopants”. The selected four dopants actually affect the electrochemical behaviour by both shifting the redox peak potential and CV pattern. CSA derivatives induce an overall shift of the peak potential to slightly more positive values, featuring a much prominent quasi-reversible behaviour. Current peaks are definitively much more symmetric (compare Figure 1A main; Supplementary Figure S2C Supporting Information). A rather peculiar result is obtained in the case of cysteine, where a much broader shoulder is present. This is probably due to the overlap of two/three separate peaks, which suggests the existence of two/three different species formed between iron and cysteine (cysteine features three functional groups able to bind iron, i.e. the thio, amino and carboxylic acid moieties). The electrochemistry of Zn based MOFs is characterized by the absence of any reversible (or quasi-reversible) oxidation/reduction current signal, as it can be seen by the inspection of the CVs shown in Figure 1B, Supplementary Figure S4 in the Supporting Information. In this respect Zn based MOFs appear promising materials, to be used in amperometric sensors, because of very low faradaic currents and large volume/surface ratio. On the other hand, the absence of any faradaic current is a direct proof of the electrochemical stability of the “doping” chiral compounds, cysteine and camphor sulphonic acid.

Figure 2A shows the IR spectra of the reference MIL-53(Fe) MOF and chiral MIL-53(Fe) based MOFs, synthesized by including chiral linkers. The IR spectrum of MIL-53 (black line) shows the characteristic absorption peaks of MIL53(Fe) shows the characteristic absorption peaks of MIL53 were obviously observed at 1,668, 1,546, 1,391, 750, and the doublets 552/523 cm⁻¹. These values are in agreement with those reported in the literature for iron-based MOFs (Banerjee et al., 2012). In detail, the peak observed at 1,668 cm⁻¹ is assigned to the carboxylic C=O bonds stretching vibration confirming the absence of free ligand in the synthesized MIL53 sample. Indeed, the free ligand MIL53 precursor H₂BDC shows the carboxylic C=O stretching at about 1,680 cm⁻¹, the decrease in energy confirms the interaction of the ligand with the Fe of the network (Li et al., 2018). The two intense peaks charactering the IR spectrum at 1,546 and 1,391 cm⁻¹ instead correspond to the asymmetric and symmetric vibrations of the C-O of the two carboxyl groups of the H₂BDC ligand, respectively. The intense and sharp peak at 750 cm⁻¹ is assigned to the aromatic C-H bending of the benzene ring. These results confirm the presence of the dicarboxylate linker within frameworks (Yang et al., 2016). The bands in the fingerprint region, between 700 and 400 cm⁻¹, are ascribed to the stretching vibration of Fe–O, indicating the formation of a Fe-oxo bond between the Fe(III) and the carboxylic group of BDC ligand (Gong et al., 2002). NH2 MIL53 shows the same spectral pattern and more features in the range 3,300–3,500 cm⁻¹ which belongs to the N-H stretching of the amino group of the amino-terephthalic acid used for its synthesis. The presence of the amino group on the benzenic ring of the terephthalic acid shifts the C-H bending peak from 750 cm⁻¹ of the MIL53 to higher energy at 769 cm⁻¹. MIL53 MOFs synthesized with chiral ligands, L-cyst, R-CSA and S-CSA, all show the same peaks as described above with the confirmation of the Fe-O bond formation by the peak at 552 cm⁻¹. The IR spectra of the Zinc based MOFs (Figure 2B) shows the typical structure due to the carboxylic moiety in the 1,300 to 1,600 cm⁻¹ range, even if Zn MOFs spectra are not as well defined as the MIL53 ones. A peak in the fingerprint region, at 431cm⁻¹ can be attributed to the Zn-O bond, the peaks observed at 811 cm⁻¹, and 1,043 cm⁻¹, can be N-H wag and a C-N stretch. ZnMOF-NH₂ and ZnMOF L Cysteine show a peak between 3,300 cm⁻¹, and 3,400 cm⁻¹, which can be attributed to N-H stretching. Of interest is, when comparing the Zn MOF spectra and the derivatives, the peak observed at 1923 cm⁻¹ has undergone a red shift and can now be observed at 2081cm⁻¹ in Zn MOF S-CSA and Zn MOF L Cysteine. This shift can be attributed to the formation of hydrogen bonds between the chiral linkers and the linker N,N'-bis(pyridin-4-methyl)cyclohexane-1,4-diamine.

Indeed, there is also a neat difference in the morphological aspect between the iron based and zinc based MOFs; and the crystal structure of the Zn based MOFs is definitively much more disordered than that of the iron based ones.

Since its first synthesis MIL53 has shown some analogies with the other MIL-MOFs based on other metallic cations such as Cr³⁺ (Serre et al., 2002) and Al³⁺ (Loiseau et al., 2004) materials MIL53(Cr) and MIL53(Al) but X-ray showed that they are not isostructural. MIL53(Fe) is built up from corner-sharing trans chains of octahedral Fe linked together by benzene-dicarboxylate (BDC) moieties forming an open framework with channels parallel to the crystallographic c-axis. MIL53 shows a peculiar nature presenting a reversible phase transition between the hydrated and the dehydrated form at near room temperature (below 330 K) (Millange et al., 2008).

Figure 3 sets out X-ray diffraction pattern of iron and Zn MOFs. In particular, Figure 3A reports the comparison between the powder XRD of the MIL53 and MIL53 R-CSA synthesized in the present work and the simulated ones form the literature crystal structure of the two hydrated and dehydrated form of MIL53. MIL53 and MIL53 R-CSA synthesized in this work appear to be a mixture of dehydrated form, as for the presence of the diffraction peak at about 12.7 (°), and of the hydrated form, for matching the peak at 19 (°). Indeed, the reference MIL53 and the CSA doped MOFs result rather similar. From the tight comparison of XRD patterns shown in Figure 3, we can infer that iron is in its usual octahedral (complexed by oxygen) site coordination, and that CSA does not destroy or alter in a significant way the MOF framework, but rather occupies interstitial free space within the MOF framework (see Supplementary Figure S5 for the molecular iron and zinc based MOFs structure).

Figure 4 shows SEM images of iron-based MIL53 and zinc-based MOF. In the two cases the morphology appears different. MIL53 shows a regular morphology while Zn-MOF shows a lamellar morphology. In both cases, the presence of structures of less than 5 µm is observed. The chiral variants of MIL53 show similar structures to that of MIL53 while NH2 MIL53 is instead organised in more spherical structures, all images are shown in Supplementary Figures S6–S10. Zn-MOF R-CSA and Zn-MOF S-CSA show the typical lamellar structure of Zn-MOF differently, the lamellar structure is less visible in NH2 Zn-MOF and Zn-MOF L-Cys. All SEM images of Zn-MOF are shown in Supplementary Figures S11–S15. EDS spectra performed on the MOF powder are shown in Sections 6 and 7 of the supporting information. In particular, EDS spectra make it possible to find the chemical composition of MOFs (expressed as an atomic percentage). Note that in the MIL53-type structures, the presence of iron is noted. In the chiral variants MIL53 L-Cys, MIL53 R-CSA, MIL53 S-CSA, EDS analysis confirms the presence of sulphur (element present in cysteine and camphor sulfonic acid doping molecules). Similar considerations can be made for Zn-MOF, in which Zn is observed. The EDS analysis on the chiral variants Zn-MOF L-Cys, Zn-MOF R-CSA and Zn-MOF S-CSA show the signal related to the presence of sulphur. Figures 4C, D show high resolution FE-SEM images of MIL53 and Zn-MOF, respectively. In the case of MIL53 emerges a morphology rather different from the low resolution one: the octahedral-like shape of the MIL53 crystals show a lamellar substructure revealing a layer substructure.

Quantum mechanical based calculations are carried out to compare the electronic structure information gained by the electrochemical analysis as a function of the MOFs molecular structure, Figure 5. Calculations have been performed using, both localized orbitals and plane wave DFT methods. Figure 5A shows a typical band energy versus wave vector plot. Focusing on the boundary between the valence and conduction band (close to the Fermi energy, Ef) a band gap of about 0.5 eV is found, which is in reasonable agreement with the results obtained in the case of solid state electrochemistry investigations dealing with organic charge transfer crystals (Solano et al., 2021; Solano et al., 2022). The relevant density of state (DOS) plot, Figure 5B, gives due reason to the role of iron in the CV results. In that, the projected DOS (pDOS) plot shows a maximum localization on iron. Please compare the orange peaks around the Fermi energy.

Also localized orbitals calculations yield a picture widely in agreement with results obtained by using plane wave basis with periodic boundary constraints. The Mulliken net charge on iron is positive and both the HOMO and LUMO are delocalized between the centra iron and the oxygens of the carboxylic groups coordinating the iron (compare Section 8 of the supporting information). In tight agreement with the experimental electrochemical results which indicate that the quasi-reversible current peaks evident in the CV curves are due to the iron oxidation in the forward curve.

The MOFs synthesized here, supported on titania, were used as working electrode in the water splitting process, in a photo-electrochemical experiment. In particular, MOFs supported on titania served as the working electrode (anode) for the oxygen evolution reaction (OER). MIL53 S-CSA was selected as a candidate for photo-electrochemical measurements, because of CV and UV-Vis results. UV-Vis spectra of MIL53 and Zn-MOF are reported in Supplementary Figures S16, S17 respectively. Chronoamperometry curves (current vs. time) at constant 1.0 V potential under illumination of an AM1.5G solar simulator were recorded. Supplementary Figure S18 shows the Air Mass (AM) experimental spectrum used as excited light source in photoelectrochemical experiments. To investigate the effect of photo-excitation, dark and light conditions were obtained by chopping the light source ON and OFF, in 20 s intervals between light-ON and light-OFF single cycles. Figure 6A sets out the absorption UV-Vis spectra of both MIL53 (black line) as a reference, and the chiralized MIL53 S-CSA (red line), Figure 6B shows current vs. time chronoamperometry curves. 1.0 V was selected as a potential close to the onset of the OER, but still without oxygen evolution. Figure 6B shows that the chiralized MIL53 S-CSA is characterized by a substantial increase in the photocurrent (light on) with respect to the achiral one. On one side, the advantage of the chiralized MIL53 S-CSA is due to the larger absorption with respect to the achiral MIL53, as depicted in Figure 6A. MIL53 S-CSA takes advantage of both being a more efficient light antenna and as a spin-filter (being chiral) as a result the OER reaction is catalyzed. All in all, this outcome is consistent with results present in the literature, where the redox processes concerning the oxygen are shown to be sensitive to the handedness of the electrode surface via chirality-spin interaction (Mtangi et al., 2015; Mtangi et al., 2017; Gazzotti et al., 2020; Sang et al., 2022).

From literature cited above, it has been noted when chiral photoelectrodes are used in photoelectrochemical water splitting, there is spin filtering of the transmitted electrons i.e., there is preferential transmission of electrons spinning one direction. When there is electron spin filtering, the formation of triplet oxygen (³O₂) is promoted, while formation of hydrogen peroxide is minimised. When the formation of hydrogen peroxide is minimized the yield of hydrogen improves, this can also be correlated to the current densities produced by achiral MIL-53 vs. MIL53 doped with R CSA.

During water splitting, Mott-Schottky analysis was carried out to determine the electronic properties of the photoelectrodes (Figure 7). The measurements were performed while sweeping voltages from 0.75 V to −0.5 V using an AC Voltage with an amplitude of 5 mV and frequency of 1000 Hz. Measurements were carried out under illumination and under darkness.

The flat band potentials for MIL 53 R-CSA under illumination and under darkness are similar at + 0.08 V, while for Zn MOF NH₂ the flat band potentials are – 0.19 V, due to the fact that the flat band potentials are similar, we can conclude that the molecules did not affect the electronic properties of the TiO₂ and that the observed current densities, were due to the activity of the molecules deposited.