Photocatalytic Reactions over TiO₂-Based Interfacial Charge Transfer Complexes

1. General Background

The scientific and engineering interest in the photocatalytic reactions, initiated by the pioneering work by Fujishima and Honda in the early seventies of the last century on water splitting using a TiO₂ electrode [1], has grown tremendously, addressing a variety of environmental problems such as water and air purification and inactivation of bacteria and viruses, as well as hydrogen production. Considering the importance of photocatalysis, over the years, many review articles have appeared, so we refer readers to reviews by Hoffman et al. [2] and Kudo et al. [3] for additional background information and relevant literature that covers an initial period of the development of this field.

Among the variety of semiconductors, TiO₂ is the most studied photocatalyst since it is chemically stable and affordable. The anatase has a bandgap energy of 3.2 eV, and the pH-dependent position of valence band maximum (VB_max) and conduction band minimum (CB_min) is presented in Figure 1 [4]. Upon TiO₂ excitation with photons with an energy exceeding the bandgap energy, electrons from the VB are promoted to the CB, leaving holes behind.

$$\mathrm{T}\mathrm{i}{\mathrm{O}}_2\stackrel{\mathrm{h}\mathsf{\nu }}{\to }{\mathrm{e}}_{\mathrm{C}\mathrm{B}}^{‒}+{\mathrm{h}}_{\mathrm{V}\mathrm{B}}^{+}$$

(1)

Photogenerated charge carriers can recombine non-radiatively and dissipate the absorbed energy as heat, recombine radiatively, or get trapped at surface states and react with electron donors and electron acceptors adsorbed on its surface. In most photocatalytic experiments, oxygen is present and serves as the primary electron acceptor.

$${\mathrm{e}}_{\mathrm{C}\mathrm{B}}^{‒}+{\mathrm{O}}_2\to {\mathrm{O}}_2^{•‒}$$

(2)

Further, in water, two protonated superoxide radical anions disproportionate into oxygen and hydrogen peroxide in a bimolecular reaction.

$$\mathrm{H}{\mathrm{O}}_2^{•}+\mathrm{H}{\mathrm{O}}_2^{•}\to {\mathrm{O}}_2+{\mathrm{H}}_2{\mathrm{O}}_2$$

(3)

On the other hand, hydroxyl radicals (HO^∙) are formed in the reaction of holes with adsorbed water or hydroxyl ions.

$${\mathrm{h}}_{\mathrm{V}\mathrm{B}}^{+}+{\mathrm{H}}_2\mathrm{O}\to \mathrm{H}{\mathrm{O}}^{•}+{\mathrm{H}}^{+}$$

(4a)

$${\mathrm{h}}_{\mathrm{V}\mathrm{B}}^{+}+\mathrm{O}{\mathrm{H}}^{‒}\to \mathrm{H}{\mathrm{O}}^{•}$$

(4b)

So, the excitation of aerated aqueous dispersions of semiconductors leads to generating reactive oxygen species (ROS).

The adsorption of reactants, either electron donors or electron acceptors, to metal oxides is governed by their surface chemistry, i.e., the charge of surface hydroxyl groups. Based on the literature survey by Professor Kosmulski [5], it is well-known that the average and median pH values of zero point charge (pH_ZPC) for anatase are 5.9 and 6, respectively, while for the rutile, they are 5.4 and 5.5, respectively.

Considering the position of VB_max and CB_min in TiO₂, the photogenerated holes are powerful oxidants, while photogenerated electrons are good reductants. Besides the limitless use of solar energy, the advantage of photocatalysis is the possibility to carry on photo-driven heterogeneous reactions under mild experimental conditions (room temperature and atmospheric pressure), including non-spontaneous reactions with selective product synthesis [6,7]. However, two obstacles must be overcome to achieve the high yield of photocatalytic reactions over TiO₂. First, based on the laser flash photolysis experiments, it is well known that the photogenerated charge carriers recombination is a fast process occurring on the nanosecond time scale [8,9]. Second, as a wide bandgap semiconductor, TiO₂ absorbs only the UV photons, i.e., 5% of solar radiance. So, efficient separation of photogenerated electrons and holes and improved optical properties are prerequisites to achieve efficient photocatalytic reactions over TiO₂.

Three different approaches are distinguished to address the previously mentioned issues. First, the deposition of co-catalysts (noble metal particles) on the photocatalysts’ surface provides enhanced electron transfer reactions and improved optical properties by plasmon resonance absorption [10,11]. Second, doping with light and heavy elements promotes less energetic excitations of electrons from mid-gap dopant levels to the conduction band of TiO₂ [12,13,14]. And finally, third, surface modification of TiO₂ with organic and organometallic molecules absorbing in the visible spectral range [15], that letter led to the discovery of dye-sensitized solar cells, the so-called Grätzel solar cells [16].

Despite tremendous efforts to use other semiconductor materials for photocatalytic purposes under visible light excitation, improving their optical properties by doping [17], forming heterojunctions to enhance their oxidation or reduction power [18], or synthesizing Z-scheme photocatalysts to improve charge separation [19], TiO₂ remains as a photocatalyst of choice.

The most recent approach to bring the absorption of TiO₂ and other wide bandgap metal oxides in the visible spectral range is interfacial charge transfer (ICT) formation [20]. Colorless aromatic molecules are recognized as suitable ligands to facilitate the formation of ICT complexes. During this time, from fundamental studies, the research in this area evolved to potential applications of TiO₂, including the photocatalytic production of hydrogen and degradation of organic dyes [21,22,23,24], the light-to-current conversion [25,26,27], and photo-induced antimicrobial activity [28]. However, this review will be limited to photocatalytic reactions over TiO₂-based ICT complexes.

2. Interfacial Charge Transfer Complexes

2.1. Synthesis

The formation of interfacial charge transfer (ICT) complexes is facilitated by a polycondensation reaction between hydroxyl groups originating from the surface of metal oxide (Me‒OH) and the colorless aromatic molecules, inducing the formation of a covalent linkage between surface metal, oxygen, and carbon. A schematic presentation for TiO₂-based ICT complex formation is presented in Scheme 1. The successful formation of the ICT complexes is accompanied by the red absorption shift, providing a simple way to bring the optical properties of materials to the more practical spectral range for photo-driven catalytic reactions. As an instructive example, the absorption spectra of pristine 45-Å TiO₂ colloid and TiO₂-based ICT complexes, displaying tunable red absorption shifts induced by different surface-active ligands, are shown in Figure 2 [29].

The appearance of absorption in the visible spectral range in TiO₂-based hybrids is due to the electronic coupling of localized orbitals of surface-attached ligands with the delocalized electron levels from the conduction band of TiO₂ [30]. Consequently, absorption of light by the ICT complex yields direct, single-step promotion of electrons from the ligand directly into the CB of TiO₂ without energy loss [31], resulting in a red absorption shift compared to unmodified TiO₂. It should be emphasized that there is a fundamental difference in the photogeneration of charge carriers in ICT complexes and metal oxides sensitized with dye molecules; a graphical presentation is shown in Scheme 2. In the case of dye-sensitized metal oxides, photogeneration of charge carriers involves two steps: first, excitation of the dye molecules and subsequent electron transfer from the excited state into the semiconductor conduction band. In addition, photogenerated electrons and holes in ICT complexes are separated instantaneously into two phases, so holes are localized on the donating organic modifier, and the electrons are delocalized in the CB of TiO₂ [32].

In the early stages of development of this field, when revealing the basic properties of the ICT complexes, most of the studies were done on small-sized colloidal TiO₂ nanoparticles due to their unique surface structure. It is well-known, based on the X-ray absorption near-edge spectroscopy (XANES), that the bond length and coordination of the surface Ti atoms change from octahedral (six-coordinate) to square-pyramidal (penta-coordinate) when the size of TiO₂ particles is sufficiently small, i.e., when particles have large curvature [20,33]. Later, for practical purposes, instead of using small colloidal nanoparticles, TiO₂-based ICT complexes were prepared using submicronic TiO₂ particles prepared by ultrasonic spray pyrolysis [34,35], TiO₂ nanopowders prepared by sol-gel method [22], and commercial TiO₂ photocatalysts such as Degussa P25 and Aerosil P90 [21,36,37,38,39,40,41].

Having in mind that on one side, the efficiency of the photocatalytic reactions is higher for small-sized particles, while, on the other side, powders are easier to handle, the visible-light-responsive TiO₂ nanoparticles were in-situ prepared on polymer support [24]. Macroporous copolymer, based on glycidyl methacrylate and ethylene glycol dimethacrylate (poly(GMA-co-EGDMA)), with a large specific surface area (36 m²/g) and average pore size (130 nm) was used as a starting material since its functionalization is easy due to the presence of the reactive epoxy group. A two-step synthetic procedure is presented schematically in Figure 3A. The first step includes dopamine modification of poly(GMA-co-EGDMA) copolymer, indicated by the complete disappearance of vibrations that belong to the epoxy ring. In the second step, the hydrolysis of titanium(IV) isopropoxide (TTIP) in the presence of functionalized poly(GMA-co-EGDMA) copolymer dispersed in organic solvent results in the simultaneous formation of the TiO₂ nanoparticles and the appearance of the ICT complex via two adjacent phenolic groups from dopamine coordinated to poly(GMA-co-EGDMA) copolymer over the amino groups. The microscopy data showed that the polymer support is decorated with a large number of well-separated small-in-size TiO₂ nanoparticles (<10 nm) with a reasonably narrow size distribution (Figure 3B). In addition, the spectroscopy measurements (Figure 3C) revealed that the absorption spectrum of the resulting composite displays a significant redshift.

Until recently, the ICT complex formation has been almost exclusively studied using various morphological forms of TiO₂ [20,21,22,23,24,25,26,27,28,29,31,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70]. In the last several years, the successful formation of the ICT complexes for other wide-band-gap oxides, including titanates [71,72,73,74,75,76], CeO₂ [77,78], ZrO₂ [79,80], ZnO [81,82,83,84,85], Al₂O₃ [86,87,88], and hydroxyapatite [89,90], has been reported. However, the number of inorganic components of the ICT complexes is quite limited, while the number of organic components is practically limitless. A literature overview of the ICT complexes between various wide-band-gap oxides and different types of ligands is summarized in Table 1. For clarity reasons, we classified ligands into four groups: ligands forming two neighboring Me‒O‒C linkages (catechol (CAT), salicylic acid (SA), and their derivatives), ligands forming single Me‒O‒C linkage (phenol and its derivatives), ligands forming Me‒S‒C linkage/linkages (thiols), and miscellaneous, refereeing to all other ligand types. It should be emphasized that the formation of the ICT complex between silver nanoparticles and aromatic amino acids was recently predicted by the DFT calculations, providing new insight into peculiar optical properties and reactivity of surface-modified silver nanoparticles [91], indicating that the ICT complex formation is not exclusive for metal-oxides.

2.2. Composition and Surface Structure

Infrared spectroscopy is a method of choice to reveal the coordination of ligand molecules to the surface of wide bandgap oxides. The differences in infrared (IR) spectra between free and bound ligands (appearance or disappearance, as well as the shift of vibrational bands) provide the possibility to understand the surface structure of organic-inorganic hybrids. So far, besides infrared spectroscopy, there is only one attempt, combining solid-state NMR with density functional theory (DFT), to understand the geometry of attached catechol (CAT) to the TiO₂ surface [92].

Knowing that the ICT complex formation occurs by polycondensation reaction between hydroxyl groups from inorganic and organic parts of hybrid, the coordination of phenol (Ph) and its derivatives to the TiO₂ surface is almost intuitive. An instructive example is a study by Fujisawa et al. [37]. The Kubelka-Munk function spectra of pristine TiO₂ and TiO₂‒Ph samples and the IR spectra of free Ph and attached Ph to the TiO₂ surface are shown in Figure 4. The absorption onset in the TiO₂‒Ph sample is red-shifted (around 550 nm), compared to unmodified TiO₂, due to the ICT complex formation. The IR spectrum of Ph shows pronounced peaks at 1472 and 1593 cm⁻¹ that belong to in-plane starching C=C vibration mixed with C‒H and O‒H bending vibrations and two relatively broad peaks at 1223 and 1369 cm⁻¹ that belong to C‒H and O‒H bending vibrations. While the peaks at 1472 and 1593 cm⁻¹ are practically intact upon adsorption of Ph on TiO₂, the lower-energy peaks at 1223 and 1369 cm⁻¹ are drastically changed, indicating a significant structural change around the hydroxy group of Ph and suggesting the formation of a Ti‒O‒C linkage.

However, in the case of ligands with two neighboring hydroxyl groups (CAT and SA, and their derivatives), the IR spectroscopy can’t provide an answer about the mode of their binding to Ti surface atom (Ti_surf), is it bidentate binuclear (bridging) coordination or bidentate mononuclear (chelating) coordination; for clarity reasons chelating and bridging coordination of CAT to Ti_surf is presented in Scheme 3. In other words, the question is, what is the molar ratio between Ti_surf and these types of ligands? Since the ICT complexes exhibit optical properties distinct from their constituents, Job’s method of continuous variation [93] was applied to determine their composition, assuming that a single-type complex is present in the solution. The molar concentration of surface Ti atoms ([Ti_surf]) can be calculated from the following equation [94], knowing the molar concentration of TiO₂ ([TiO₂]) and the diameter of particles in angstroms (D):

$$\left[{\mathrm{T}\mathrm{i}}_{\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{f}}\right]={\displaystyle \frac{12.5\left[{\mathrm{T}\mathrm{i}\mathrm{O}}_2\right]}{\mathrm{D}}}$$

(5)

Of course, the prerequisite for using equation 1 to calculate the molar concentration of Ti_surf is the narrow size distribution of nanometer in size TiO₂ particles. The stoichiometric ratio (n) between Ti_surf and the ligand (L) can be determined by plotting the absorbance of the ICT complex versus the mole fraction (x) of the metal or ligand. An instructive example, the Job’s plot for the surface-modified 45-Å TiO₂ colloids with CAT and SA [29], is presented in Figure 5. Mole fraction (x=[Ti_surf]/([Ti_surf]+[L])) at the absorbance maximum corresponds to the stoichiometric ratio between T_surf and L. The stoichiometric ratio between Ti_surf and various catecholate- and salicylate-type ligands was found to be 2:1, indicating bridging coordination of these ligand types to the surface Ti atoms [22,29,34,35,42,43,44,45,46,47].

Besides composition, the simple spectrophotometric method, based on Benesi-Hildebrand analysis [95], is suitable for the determination of stability constants in transparent heterogeneous colloidal systems, i.e., when the inorganic-organic hybrid particles are sufficiently small, and the ICT complex exhibits optical properties distinct from its constituents. The stability constant of the ICT complex (K_b) can be expressed as follows:

$${\mathrm{K}}_{\mathrm{b}}={\displaystyle \frac{\left[{\mathrm{I}\mathrm{C}\mathrm{T}}_{\mathrm{c}\mathrm{o}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e}\mathrm{x}}\right]}{\left[{\mathrm{T}\mathrm{i}}_{\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{f}}\right]\left[\mathrm{L}\right]}}$$

(6)

Since the absorption in the visible spectral range solely originates from the ICT complex, the concentration of the ICT complex can be expressed by the Lambert-Beer equation ([ICT_complex]=A/εl), so equation 2 can be rearranged in the following way:

$${\displaystyle \frac{1}{\mathrm{A}}}={\displaystyle \frac{1}{{\mathrm{K}}_{\mathrm{b}}{\mathrm{A}}_{\mathrm{m}\mathrm{a}\mathrm{x}}\left[\mathrm{L}\right]}}+{\displaystyle \frac{1}{{\mathrm{A}}_{\mathrm{m}\mathrm{a}\mathrm{x}}}}$$

(7)

where A and A_max are absorbances for a given and saturation concentration of ligand ([L]), respectively.

The stability constants of the TiO₂-based ICT complexes, obtained using this simple spectrophotometric method, are similar in the order 10³ M^-1 [22,29,34,35,42,43,44,45,46,47]. Only when thiosalicylic acid was used as a ligand a slightly smaller stability constant was found [57]. However, the stability constants for surface-modified TiO₂ with SA and CAT, obtained by different methodologies, the adsorption approach after filtration [96,97,98,99,100] or IR measurements [101,102,103,104], are one order of magnitude larger (~10⁴ M^-1). This discrepancy may be due to different applied methodologies and the TiO₂’s different particle sizes. In addition, for spectroscopic determination of the stability constants of the TiO₂-based ICT complexes, only adsorption curves were used, and since the Benesi-Hildebrand analysis requires equilibrium conditions when the adsorption and desorption rates are equal, they might be underestimated [105].

Density functional theory (DFT) calculations, with a time lag of more than a decade, supported the experimental efforts to understand the composition and coordination of ligands to TiO₂ surface. Even initial studies on small model systems provided theoretical results with the same trend as experimental ones [43,44]. For example, the calculated IR spectrum of CAT bound to the TiO₂ surface, assuming bridging coordination and using molecular complex Ti₂O(OH)₄‒L, is quite similar to the measured one, while the calculated IR using chelating model (Ti(OH)₂‒L) does not have resemblance [43] (Figure 6A). Later on, the use of model clusters with increasing complexity, such as Ti₈O₁₄(OH)₃−L [38] and Ti₂₁O₄₀H(OH)₄−L [68], provided a more realistic description of the geometry of the ICT complexes, including the bond length. An instructive example is a study by Fujisawa et al. [38], showing that the calculated IR spectra on model complexes mimicking TiO₂‒aminophenol system (Figure 6C) follow in the same manner the influence of an electron-donating amino group at different positions (o-, m-, and p-) on the Ti‒O‒C vibration observed experimentally in the 1200-1300 cm^-1 spectral range. Optimized structures of model systems are presented in Figure 6B. It is fair to say that during the years, the DFT calculation evolved at the predictability level, providing an opportunity to minimize a trial-and-error approach.

2.3. Fine-Tuning of Optical Properties – The Role of Free Functional Group

The visible-light response of the TiO₂-based ICT complexes originates from the excitation from the donor levels created in the TiO₂ midband. The presence of different ligands attached to the TiO₂ surface leads to different spectral responses. Higashimoto et al. [21], probing the photocatalytic ability of surface-modified TiO₂ with CAT and its derivatives in water splitting reaction, were the first to establish the influence of phenyl-ring-substituted groups on the electronic structures of the ICT complexes, i.e., the possibility to tune optical properties of inorganic-organic hybrids without trial-and-error approach. Simply speaking, the ligands’ free electron-donating groups (EDG) induce the decrease of the energy gap of the TiO₂-based ICT complex compared to the ligand without substituent group. On the opposite, the presence of electron-withdrawing groups (EWG) leads to the enlargement of the energy gap. For clarity, the energy-level diagram displaying the red/blue absorption shift, i.e., the decrease/increase of energy gap in the presence of ligands with EDG/EWG, is presented in Scheme 4.

A recent study by Fujisawa et al. [53] provides an illustrative example of the substituent effect on the optical properties of the TiO₂-based ICT complexes. In this study, the optical properties of the ICT complexes with phenol derivatives, having EDG groups (4-tert-butylphenol and 4-methoxyphenol) and EWG group (pentafluorophenol), were compared with the optical properties of the anatase TiO₂‒phenol complex; the Kubelka-Munk spectra are shown in Figure 7. The absorption of the surface-modified TiO₂ with phenol derivatives having EDGs (‒C(CH₃)₃ and ‒OCH₃) is red-shifted compared to the TiO₂‒phenol complex. Also, a phenol derivative with a stronger electron-donating –OCH₃ group induced a larger absorption shift toward the infrared spectral region. On the other hand, pentafluorophenol, a phenol derivative with five electron-withdrawing fluorine atoms, induced the blue absorption shift compared to TiO₂‒phenol complex. In addition, the experimental findings are well-supported with DFT and time-dependent DFT (TD-DFT) calculations on large model systems, mimicking anatase TiO₂ (Ti₃₄O₆₆(OH)₄) and TiO₂-phenol complexes (TiO₂‒O‒Ph‒R, where R is H, C(CH₃)₃, OCH₃, or F₅).

Although the spectroscopy measurements combined with the DFT calculations provide information concerning relevant electronic levels in the ICT complexes, ionization potential (IP) measurements by photoelectron spectroscopy (PS) offer more precise information. The pinned position of IP in CAT-functionalized TiO₂, SrTiO₃, and BaTiO₃ powders, corresponding to the highest occupied molecular orbital (HOMO) of the chemisorbed CAT to metal oxide surfaces, indicates an alternative way, although limited, to tune optical properties of hybrids by changing their inorganic component while using the same ligand [75].

Step forward in IP determination provided the vacuum-ultraviolet electron imaging angle-resolved photoelectron spectroscopy (VUV ARPES). This unique technique allows the measurements of IP in separated colloidal nanoparticles instead of measuring isolated powders. The absorption spectra of 50-Å TiO₂ colloid and surface-modified TiO₂ colloids with SA and 5-aminosalicylic acid (5-ASA) are shown in Figure 8, together with corresponding photoelectron spectra [49]. The amino group is EDG, and the TiO₂‒5-ASA complex is red-shifted compared to the TiO₂‒SA complex. The ionization energies for TiO₂, TiO₂‒SA, and TiO₂‒5-ASA, determined by VUV ARPES, are 7.2, 6.5, and 5.9 eV, respectively. Differences in ionization energies between functionalized TiO₂ with SA and 5-ASA and pristine TiO₂ (0.7. and 1.3 eV, respectively) correspond to the red absorption shift observed in absorption spectra.

Besides the direct tuning of the optical properties of the ICT complexes, the proper introduction of substituent functional groups provides the possibility to prepare higher hierarchical structures. For example, the amino group introduction with a strong reducing capacity [106,107] offers a simple way for in-situ reduction of silver ions to metallic silver particles linked to metal oxide over ligand [76,80]. So, the presence of Ag nanoparticles, conjugated to the ICT complex, has a twofold effect, simultaneously improving the optical properties of the hybrid due to surface plasmon absorption and separation of photo-generated charge carriers. As an example, the TEM image of the titanate nanotube surface modified by 5-ASA and decorated with Ag nanoparticles, prepared using the free functional amino group to reduce Ag⁺ ions, is presented in Figure 9, together with Kubelka-Munk spectra [76]. This simple synthetic approach leads to the formation of well-separated, nearly spherical nanometer in-size Ag particles linked to the titanate nanotube surfaces, as indicated by microscopy and spectroscopy data. Of course, Ag nanoparticles display antimicrobial activity, and the above-mentioned synthetic procedure makes it possible to support them with a biocompatible material prepared from waste, like hydroxyapatite [89], or by magnetite, simplifying from a technological point of view separation process [108]. In addition, the introduction of desired functionality is pivotal in other potential applications non-related to photo-driven processes, such as enhanced sorption capacity and selectivity of sorbents [90] and visualization/recognition of drug molecules [109,110].

3. Photocatalytic Degradation of Organic Pollutants

Although the photocatalytic degradation of organic pollutants, in particular phenol and its derivatives, has been in this field one of the main topics for decades, the formation of ICT complexes with this class of organic compounds and the possibility to use the visible light to drive photocatalytic reaction was recognized with considerable delay. The reason for that is the most probable use of low concentration of phenol-based compounds (micromolar), typical for photocatalytic experiments, and the low extinction coefficient of the ICT complexes with this type of ligands. The 4-chlorophenol (4-CP) is a frequent representative of phenols to optimize the efficiency of their photocatalytic degradation since degradation of 4-CP leads to complete mineralization and generation of chlorides and CO₂. Because of that, we will use, as an example, the photocatalytic degradation of 4-CP to emphasize the difference in the reaction mechanism under visible light excitation.

The detailed influence of the TiO₂ loading and excitation wavelength in the UV spectral range from 300 to 400 nm on the mineralization rate and the quantum yield of 4-CP degradation is in-depth analyzed in a study by Stafford et al. [111]. The postulated mechanism of photocatalytic degradation of 4-CP under UV light excitation of TiO₂ is complex, occurring in three parallel reaction pathways, including reactive oxygen species, hydroxyl radicals (HO^•), and superoxide radical anion (O₂^•‒), formed from photogenerated electrons and holes, respectively [112,113]. Photomineralization kinetics of 4-CP follow a Langmuir-Hinshelwood kinetics, typical of many similar systems.

However, it is possible to degrade 4-CP using exclusively visible light (hν > 420 nm) by exciting TiO₂‒4-CP‒O₂ system, generating as final products chlorides and CO₂, as pointed out by Agrios et al. [114,115] and Kim et al. [116]. Of course, direct ligand-to-metal electron transfer without involving the excited state of 4-CP is responsible for the visible light reactivity. Then, the injected electron into the CB of TiO₂ reduces oxygen, electron acceptor, forming superoxide radical anion (O₂^•‒). In deaerated suspensions saturated with nitrogen, the photocatalytic degradation of 4-CP does not occur (Figure 10) since photogenerated electrons recombine with the surface complex, making a null cycle. However, when an alternative electron acceptor, ferric ions, is present in the anoxic suspension, the photocatalytic degradation of 4-CP is as fast as in the presence of oxygen. It is important to point out that photooxidation of 4-CP is not affected by the presence of radical scavengers, such as tert-butyl alcohol or enzyme superoxide dismutase.

Over the years, the above-mentioned approach for photocatalytic degradation of organic pollutants, preferably phenol derivatives, was widely exploited [117,118,119,120,121,122,123,124,125]. Since these are examples of self-sensitized degradation of pollutants under visible irradiation, and TiO₂-based ICT complexes are unstable when exposed to light and not characterized in detail, we did not include the corresponding references in the literature overview presented in Table 1.

The prerequisite to consider any material as a photocatalyst is its stability. Higashimoto et al. [21] were the first to report that the visible-light-responsive TiO₂-based ICT complexes with catechol and its derivatives retain their optical properties when exposed to light and display the photocatalytic activities for H₂ evolution, which we will discuss in the following chapter of this review. Simultaneously, Milićević et al. [22] noticed that the same class of hybrid materials is capable of inducing photocatalytic degradation of an organic dye crystal violet under exclusive illumination with photons whose energy is smaller than 2.75 eV, i.e., smaller than the band gap energy of TiO₂. The organic dyes, such as methylene blue (MB), crystal violet (CV), and methyl orange (MO), are frequently used to test the photocatalytic performance of semiconductors because the mechanism of their degradation kinetics is well-established, it is easy to follow their degradation kinetics experimentally, and they do not undergo to direct photolysis, i.e., their degradation in the absence of photocatalyst is negligible [126,127].

An illustrative example of enhanced photooxidative capability is the degradation of CV over ICT complex between TiO₂ and dopamine linked to poly(GMA-co-EGDMA) copolymer with extended absorption in the visible spectral range [24], whose synthesis is described in the previous chapter (see Figure 3). Before photocatalytic experiments, sorption equilibrium was established in the dark, and from absorption spectra, the sorption capacity of the hybrid towards CV was estimated. The exclusive visible light excitation ensured the use of the low-energy band-pass 450 nm cutoff filter. The absorption spectra of CV as a function of time, before (in dark) and under visible light illumination, are presented in Figure 11. The results indicate that the ICT complex between TiO₂ and dopamine linked to poly(GMA-co-EGDMA) copolymer can induce photocatalytic degradation of CV by exclusive visible light excitation. Of course, the degradation kinetics of CV is faster when the light source mimics the solar spectrum (compare kinetic curves given in the inset to Figure 11).

So far, photocatalytic degradation of different organic dyes, taking advantage of enhanced optical properties due to ICT complex formation, has been reported for TiO₂ functionalized with tiron [23,41] and rhodizonic acid [65]. Also, other surface-modified metal oxides, such as titanates [71,72], ZrO₂ [80], and Al₂O₃ [86,88], display photooxidative ability. Among the mentioned studies, the most striking example is photocatalytic degradation of organic dyes (MB and CV) over surface-modified Al₂O₃ with catechol [86] and 5-aminosalicylic acid [88], knowing that Al₂O₃ is an insulator with the band gap of about 8.7 eV [128]. We can say that the ICT complex formation transforms the insulator into a hybrid semiconductor-like material capable of harvesting a large portion of the solar spectrum.

Silver and silver compounds are powerful biocides, and the increased resistance of microbial species towards antibiotics renewed the interest in using Ag nanoparticles as disinfectant agents, either free-standing [129] or deposited onto different supports, inorganic [89] and organic [130,131], or within matrices for water treatment [132] and food packaging applications [133,134]. As an alternative to avoid the undesired impact of silver on the environment, the photocatalytic inactivation of microbial species by TiO₂ thin films [135,136] or deposited TiO₂ nanoparticles onto fibers [137], including textiles [138,139] has arisen. The TiO₂, doped with light [140] and heavy elements [141], with extended absorption in the visible spectral range, was applied to avoid the use of harmful UV light sources and replace them with less expensive, harmless visible light sources. Concerning the use of TiO₂-based ICT complexes, to the best of our knowledge, the biocidal effect against E. coli and S. aureus upon exclusive visible light excitation was reported only for TiO₂ nanofibers surface-modified with rhodizonic acid [28]. Of course, the reaction mechanism includes reactive oxygen species (hydroxyl radical and superoxide radical anion), the same ones participating in the photocatalytic degradation of organic dyes.

4. Photocatalytic Hydrogen Generation

Hydrogen is the ultimate clean energy source that can replace fossil fuels, considerably solving energy and environmental issues. Semiconductors must have the proper energy alignment of the conduction and valence band and the band gap energy to perform as photocatalysts in a water-splitting reaction. Consequently, the potentials of CB_min and VB_max have to be negative compared to the redox potential of H⁺/H₂ (0 V versus NHE) and more positive than the potential of O₂/H₂O (1.23 V), respectively. Therefore, the theoretical minimum band gap for water splitting is 1.23 eV. Like electrolysis, photogenerated electrons reduce water molecules to form H₂ and oxidize water by the holes to form O₂.

Another significant issue is the stability of photocatalysts. For example, CdS has suitable band positions and visible light response but is inactive for water splitting into H₂ and O₂. Instead of oxidizing water, photogenerated holes oxidize S^2‒ in CdS, followed by the release of Cd²⁺ ions [142].

$$\mathrm{C}\mathrm{d}\mathrm{S}+2{\mathrm{h}}_{\mathrm{V}\mathrm{B}}^{+}\to \mathrm{C}{\mathrm{d}}^{2+}+\mathrm{S}$$

(8)

This reaction is called photocorrosion and is characteristic of an entire class of metal sulfides.

In strict terminology, water splitting means splitting water molecules into H₂ and O₂ in a stoichiometric ratio without the sacrificial agents. However, the sacrificial agents, i.e., electron donors, are frequently used, particularly alcohols, since H₂ yield might be low. For example, 2-propanol (isopropyl alcohol) reacts with holes or their successors, hydroxyl radicals, forming alkoxy ((CH₃)₂CHO^∙) or α-hydroxyalkyl ((CH₃)₂C^∙(OH)) radicals (Equation 9).

$$\mathrm{H}{\mathrm{O}}^{•}+(\mathrm{C}{\mathrm{H}}_3{)}_2\mathrm{C}\mathrm{H}\mathrm{O}\mathrm{H}\to (\mathrm{C}{\mathrm{H}}_3{)}_2{\mathrm{C}}^{•}\left(\mathrm{O}\mathrm{H}\right)+{\mathrm{H}}_2\mathrm{O}$$

(9)

The redox potential of α-hydroxyalkyl radical ((CH₃)₂C^∙(OH)) is sufficiently negative (-1.23 V versus NHE [143]) to transfer electrons into the semiconductor conduction band. So, the absorption of one photon leads to the generation of two electrons, and alcohols are frequently termed current doubling agents. But, in this case, compared to the water-splitting reaction, photocatalytic generation of hydrogen is a half-reaction.

Despite the low overlap of its absorption with the solar spectrum, chemically inert and cost-effective TiO₂ remains the most attractive photocatalytic material, and the use of TiO₂-based ICT complexes is the most recent attempt to enhance its photocatalytic performance in water-splitting reactions. However, the number of studies in this research area is still small.

Pioneering work by Ikeda et al. [144] is an instructive example of improved photocatalytic hydrogen evolution due to visible-light absorption of surface-modified TiO₂ with 1,1’-binaphtalene-2,2’-diol. Before modifying TiO₂ with binaphthol, a small amount of co-catalyst, Pt particles, was deposited on TiO₂ to make active sites for hydrogen generation. The photocatalytic experiments were carried out in the presence of sacrificial electron donor triethanolamine (TEOA) under exclusive visible light excitations, filtering out high-energy photons by cut-off filters. The kinetics of hydrogen evolution over surface-modified TiO₂ with binaphthol as a function of excitation wavelength is shown in Figure 12. We can draw the following conclusions from these results: First, hydrogen production does not occur if excitation wavelengths are longer than 580 nm since this wavelength corresponds to the absorption onset of the TiO₂-based ICT complex with binaphthol. Second, hydrogen evolution is becoming more efficient by broadening the visible excitation range. Finally, the hydrogen generation over unmodified TiO₂ is non-existent under visible light excitation (>430 nm).

Another instructive example is a study by Higashimoto et al. [21] concerning the influence of functional groups in catechol derivatives on the photocatalytic hydrogen production of functionalized TiO₂ under visible light excitation. It should be mentioned that the TiO₂ was platinized, and triethanolamine (TEOA) was used as the hole scavenger, similar to the work of Ikeda et al. [144]. Figure 13A shows time-dependent hydrogen production over surface-modified TiO₂ with various catechol derivatives. The photocatalytic activity of TiO₂-based ICT complexes with catechol derivatives having electron-donating groups (4-tert-butylcatechol and 3-methoxycatechol) is lower, almost non-existent, than that of the surface-modified TiO₂ with catechol. On the other hand, hybrids with catechol derivatives having electron-withdrawing groups (2,3-dihydroxybenzoic acid, 3,4-dihydroxybenzonitrile, and tiron) display better photocatalytic performance. The observed photoactivity trend is unexpected since electron-donating groups are decreasing the energy gap of the TiO₂-based ICT complex, while the presence of electron-withdrawing groups leads to the enlargement of the energy gap compared to the ligand without substituent group (Scheme 4). Simple speaking, the better photoactivity displayed photocatalysts absorbing less in the visible spectral region. The explanation for the observed effect is the necessity of using a sacrificial electron donor to facilitate the efficient separation of photogenerated charge carriers. Figure 13B correlates the photocatalytic activities of TiO₂-based ICT complexes with catechol derivatives with oxidative potentials of corresponding catechol derivatives (the oxidative potential of TEOA is marked by an arrow). The catechol derivatives with electron-withdrawing groups have more anodic oxidative potential than the sacrificial hole scavenger (TEOA), and consequently, the larger the difference, the better the photoactivity. On the other hand, the catechol derivatives with electron-donating groups have more cathodic oxidative potential than TEOA, and, of course, TEOA does not perform as a hole scavenger. So, recombination between photogenerated charge carriers prevails, lowering the efficiency of the water-splitting reaction.

Oxidative potentials of small alcohols (methanol, ethanol, and isopropyl alcohol) are significantly more negative (0.016, 0.084, and 0.105 V versus NHE [145], respectively) than the oxidative potential of TEOA (1.14 V versus Ag/AgCl electrode [21]). So, on one side, their ability to scavenge photogenerated holes and hydroxyl radicals is better than TEOA. On the other side, the generated alcohol radicals can transfer electrons to the conduction band of TiO₂, leading to two electrons in the conduction band per one absorbed photon. The time-dependent rate of photocatalytic hydrogen generation, using coordinated over dopamine TiO₂ nanoparticles to polymer support as a photocatalyst, is shown in Figure 14. The schematized synthetic procedure of the photocatalyst, its morphology, and the optical properties are presented in Figure 3, accompanied by comments in section B1 [24]). The photocatalytic hydrogen generation rate over TiO₂-based ICT complex supported by polymer increases and, after approximately one hour, reaches the steady value, which is about two times higher compared to the rate obtained under the same experimental conditions using the most studied commercial TiO₂ photocatalyst Degussa P25. It is worth mentioning that the photocatalytic experiments were performed without a co-catalyst (Pt nanoparticles), unlike the above-described studies by Ikeda et al. [144] and Higashimoto et al. [21].

5. Identification of Reactive Species

Reactive species induced by photoexcitation, primary (electron-hole pair) and secondary (radicals), can be followed either by “fast” time-resolved spectroscopic techniques (pulse radiolysis, pump-probe flash photolysis, and time-resolved EPR techniques) or “static” continuous-wave EPR techniques based on the nature of reactive species, i.e., the presence of paramagnetic centers in photogenerated reactive species. The spectroscopic fingerprint of the hydrated electron [146] and hydroxyl radical [147], discovered by Professor Hart in the early sixties by pulse radiolysis technique, provided a burst in studying the radical reactions. Later on, Professor Grätzel [148] showed in simple experiments that stationary illumination of the deaerated TiO₂ colloid led to the appearance of a blue color with absorption maximum peaking at 700 nm, which is identical to the absorption spectrum of hydrated electron observed in pulse radiolysis experiment. However, time-resolved spectroscopic techniques have never been used to provide in-depth insight into the photocatalytic reaction mechanism of any oxide-based ICT complexes. On the other hand, just a few recent studies employed the EPR technique to identify radical species that participate in photocatalytic reactions, following initial works by Howe et al. [149] and Micic et al. [150] on pristine TiO₂.

Photogenerated charge carriers and radical intermediates are paramagnetic species, so EPR spectroscopy is the technique of choice for their detection and identification. It provides valuable insight concerning the origin of the photocatalytic activity in specific structures. Photogenerated electrons and holes are highly reactive, so a photocatalytic system of interest has to be frozen to suppress their disappearance. The low-temperature EPR spectra of pristine TiO₂ and surface-modified TiO₂ with 4-chlorophenol (4-CP) in dark and under illumination with light sources having different spectral profiles (UV and Vis) are shown in Figure 15 [40]. EPR signals reflecting the effective photoinduced electron transfer were observed upon in situ UV exposure of pristine TiO₂ powder (Degussa P25) [40,151]. An axially symmetric EPR signal with the g-tensor components g⊥=1.988 and unresolved g||=1.945 can be attributed to the Ti(III) bulk centers in anatase lattice representing the trapped photogenerated electrons., while the signals with g > 2.00 are compatible with the formation of the trapped holes [152,153]. Since the commercial TiO₂ powder (Degussa P25) contains about 30% rutile (E_g=3.0 eV) [154]), the Vis light exposure also evokes the generation of analogous paramagnetic signals, however, of lower intensity [40] (Figure 15A).

On the other hand, the EPR spectra of the TiO₂/4-CP preserved the axially symmetric signal of Ti(III) centers, displaying an additional single-line EPR signal with the g-value of 2.005 even in the dark (Figure 15B) due to the formation of persistent free radicals and reduced metal ion via interaction of chemisorbed phenol with metal oxide surfaces [40,155,156]. The intensity of the single-line signal with the g-value of 2.005, characteristic of the oxygen-centered organic radicals [157] increasing under UV excitation, and becomes even more pronounced under Vis light illumination, reflecting the effective photoinduced electron transfer from the organic moiety of the TiO₂/4-CP complex to the conduction band of TiO₂, producing phenoxy radical and Ti(III)_surf characterized with the unresolved broad EPR signals [40]. These experimental data correspond well with the previously published results by Kim et al. [116] and described in Section B, concerning visible-light-induced photocatalytic degradation of 4-CP and phenolic compounds in aqueous suspensions of TiO₂, mediated by a surface complex, where the formation of O-centered phenol radical in the course of the photodegradation process was proposed [158]. To briefly conclude, the low-temperature EPR spectra obtained for unmodified and surface-modified TiO₂ powder with 4-CP [40] strongly supported the visible-light-induced photocatalytic oxidation mechanism of phenolic compounds proposed in the literature [114,115,116].

An alternative, or better to say complementary techniques, to the low-temperature EPR technique, suitable for studying intermediate radicals, are the spin-trapping and spin-scavenging EPR techniques. The modus operandi of spin-trapping relies on the formation of persistent radical species, e.g., nitroxide radicals, via the reaction of diamagnetic spin trap with non-persistent photogenerated radicals. On the other hand, the spin-scavenging approach enables monitoring the disappearance of persistent radical species added to the experimental system that efficiently scavenge the photogenerated reactive radicals. So far, the spin-trapping EPR technique has been used to get a deeper insight into the mechanism of photocatalytic processes over TiO₂-based ICT complexes with squaric acid [67] and polyphenol taxifolin [159], as well as surface-modified titanate nanotubes by 5-aminosalicylic acid decorated with silver nanoparticles [76].

The semi-stable 2,2′-azinobis(3-ethylbenzothiazoline-6-sulfonic acid) radical cation (ABTS^•+) with a characteristic highly-resolved EPR signal is known to undergo reduction to ABTS. Consequently, the decrease of the overmodulated EPR signal upon the excitation in the presence of a photocatalyst reflects the reactions involving the photogenerated electrons [160]. Figure 16 shows the time-dependent relative ABTS^•+ concentration evaluated from the EPR spectra, monitored in the aerated aqueous suspensions of the pristine TiO₂ powder and TiO₂-based ICT complex with squaric acid (SqA) upon UV (λ_max = 365 nm) and Vis excitation (λ > 420 nm) [67]. Control experiments, in the absence of photocatalysts, showed that the intensity of the ABTS^•+ signal does not change upon exposure to UV or Vis light (see Figure 16). However, different behavior displayed pristine TiO₂ and TiO₂-based ICT complex with SqA upon excitation with light sources emitting photons in the UV and Vis spectral range. Upon UV excitation in the presence of both photocatalysts (TiO₂ and TiO₂/SqA), the EPR signal of ABTS^•+ decreases immediately after the beginning of exposure, quickly reaching the zero relative concentration of radical cation (Figure 16A). On the other hand, the ABTS^•+ EPR signal remains intact upon exposure of pristine TiO₂ to Vis light, while on the other hand, in an analogous experiment with the TiO₂/SqA, a continuous decrease of the ABTS^•+ concentration can be noticed (Figure 16B). Of course, these results are a consequence of the optical properties of the TiO₂/SqA, absorbing in the Vis spectral range, which provides the possibility for photoinduced electron transfer from the organic moiety of TiO₂/SqA to the conduction band of TiO₂ upon exclusive Vis light excitation. Of course, the application of spin trap, reactive towards hydroxyl radicals, such as 5,5-dimethy-1-pyrroline N-oxide (DMPO) [160,161], or stable radicals capable of scavenging all reactive paramagnetic species and measuring the photo-induced radical-producing capacity of the system of interest, such 1,1-diphenyl-2-picrylhydrazyl (DPPH) radical and Tempo derivatives, such as 4-hydroxy-2,2,6,6-tetramethylpiperidin-1-oxyl (Tempol) [162] can provide complementary data to previously described.

6. Perspectives for Future Studies

In light of the preceding overview, the development of oxide-based ICT complexes is still at the early stage, including the most studied inorganic-organic hybrid using TiO₂ as an inorganic component, and studies of their potential applications based on photo-induced catalytic reactions are seldom. However, the advantages of using oxide-based ICT complexes, in particular TiO₂-based ICT complexes, are easily recognizable and may be summarized as follows:

The simple synthetic procedure with cost-effective components.

Covalent linkage between inorganic and organic components of ICT complexes.

Enhanced optical properties with absorption in a more practical visible spectral range.

The simple way of fine-tuning optical properties.

Excitation without energy loss.

Efficient separation of photo-induced charge carriers.

The possibility of preparing composite materials with a higher hierarchical structure.

Considering the current status in this field and the emphasized advantages of oxide-based ICT complexes, we believe these novel materials are competitive with other well-developed hybrid materials intended to bring absorption in the desired spectral range. So, in our opinion, oxide-based ICT complexes are worth further studies, and we humbly suggest the following directions:

Diversification of prepared ICT complexes, using other oxides besides TiO₂ with bioactive organic components as ligands.

Extensive use of the DFT calculations at the predictability level is needed to avoid a trial-and-error approach as much as possible.

Photo-induced catalytic reactions (water splitting reaction and degradation of organic molecules) should be carried out on a long timescale to estimate the stability of oxide-based ICT complexes.

The antimicrobial activity of oxide-based ICT complexes should be tested to avoid toxic agents and harmful UV light sources; the literature data are almost non-existent.

Besides photo-induced catalytic reactions, other potential applications of oxide-based ICT complexes should be explored, such as the recognition of organic molecules, particularly drugs, and their sorption abilities towards heavy metal ions, taking advantage of functionalization that can provide selectivity and increase the sorption capacity.