Photochemical Catalysts for Hydrocarbons and Biomass Derivates Reforming in Intensified Processes

Abstract

Photocatalysts for applications in different sectors, e.g., civil and environmental, are already developed to a mature extent and allow, for example, the purification of gaseous and liquid streams or the self−cleaning surfaces. The application of photocatalysts in the industrial sector is, however, quite limited. The review addresses the specific topic of the photocatalytic reforming of methane and biomass derivates. In this regard, recent advances in materials science are reported and discussed, in particular regarding doped and modified oxides (TiO₂ and ZrO₂) or non−oxidic ceramics. Concerning process integration, a comparison between traditional two−dimensional photoreactors and fluidized bed systems is proposed and design guidelines are drawn, with indications of the possible benefits. Photocatalytic fluidized beds appear more suitable for small− and medium−scale integrated processes of reforming, operating at lower temperatures than traditional ones for distributed hydrogen generation.

1. Introduction

As the principal pathway for hydrogen and syngas production, hydrocarbon reforming is one of the most important processes of the chemical industry. More than 40% of the world’s hydrogen production is obtained through steam reforming of natural gas, and the hydrogen and syngas demands are expected to be ever increasing in the next future [1]. However, the conventional steam reforming reaction, typically performed on methane (R1), is highly endothermic and requires the use of a catalyst, high temperatures, and high amounts of excess water compared to stoichiometry to obtain suitable yields and limit side reactions, such as methane cracking (R2), which result in coke deposition on the catalyst’s surface [2,3].

$\mathrm{C}{\mathrm{H}}_{4\left(\mathrm{g}\right)}+{\mathrm{H}}_2{\mathrm{O}}_{\left(\mathrm{g}\right)}\rightleftharpoons \mathrm{C}{\mathrm{O}}_{\left(\mathrm{g}\right)}+3{\mathrm{H}}_{2\left(\mathrm{g}\right)}$	ΔH° = 204 kJ/mol	(R1)
$\mathrm{C}{\mathrm{H}}_{4\left(\mathrm{g}\right)}\to {\mathrm{C}}_{\left(\mathrm{s}\right)}+2{\mathrm{H}}_{2\left(\mathrm{g}\right)}$	ΔH° = 74 kJ/mol	(R2)

The high reaction temperature required results in a very energy intensive process, and while noble metal−based catalysts would be able to provide very good yields and high reaction selectivity towards syngas production with limited coke formation, they are too expensive to be commercially used, with nickel−based catalysts being the industry standard in modern processes instead [4,5]. However, nickel is a toxic metal [6], leading to safety problems when handling catalysts and during disposal. Coke formation and sulfur poisoning become particularly relevant problems when considering biogas dry reforming, which would otherwise be a much more sustainable process compared to steam reforming of natural gas, being potentially able to achieve negative carbon dioxide emissions [7]. Biogas usually contains high concentrations of carbon dioxide in addition to methane; therefore, the direct application of a dry reforming reaction (R3) would allow methane conversion without expensive separation pre−treatments [8]. Unfortunately, dry reforming is an even more endothermic reaction which is also much more affected by coke deposition compared to steam reforming, as CO₂ is a weaker oxidant, with the Boudouard reaction (R4) and reverse water–gas shift reactions (R5) negatively affecting yield and H₂/CO ratio [9,10].

$\mathrm{C}{\mathrm{H}}_{4\left(\mathrm{g}\right)}+\mathrm{C}{\mathrm{O}}_{2\left(\mathrm{g}\right)}\rightleftharpoons 2\mathrm{C}{\mathrm{O}}_{\left(\mathrm{g}\right)}+2{\mathrm{H}}_{2\left(\mathrm{g}\right)}$	ΔH° = 247 kJ/mol	(R3)
$2\mathrm{C}{\mathrm{O}}_{\left(\mathrm{g}\right)}\rightleftharpoons {\mathrm{C}}_{\left(\mathrm{s}\right)}+\mathrm{C}{\mathrm{O}}_{2\left(\mathrm{g}\right)}$	ΔH° = 172 kJ/mol	(R4)
$\mathrm{C}{{\mathrm{O}}_2}_{\left(\mathrm{g}\right)}+{\mathrm{H}}_{2\left(\mathrm{g}\right)}\rightleftharpoons \mathrm{C}{\mathrm{O}}_{\left(\mathrm{g}\right)}+{\mathrm{H}}_2{\mathrm{O}}_{\left(\mathrm{g}\right)}$	ΔH° = 41 kJ/mol	(R5)

It should be noted that compared to the steam reforming reaction, dry reforming also has the distinction of obtaining a lower H₂ to CO ratio, which is useful for obtaining syngas, which can be directly used in chemical synthesis without post−treatments. The steam and dry reforming reactions can finally also be applied to higher hydrocarbons and are, for example, also utilized for the abatement of tar from biomass gasification [11,12,13]; however, the high temperature requirements of the reaction and the subsequent deactivation of catalysts due to sintering and coke deposition remain great issues in the implementation of the process. Several alternatives in terms of catalyst support, dopants, promoters, and structure have, thus, been investigated to improve reactant conversion and limit poisoning [14,15,16,17,18,19,20,21,22]. In recent years, interest in alternatives to conventional thermochemical processes, such as plasma reforming [23], electrochemical reforming [24], and photocatalytic reforming [25], has also significantly grown, as briefly reported in Figure 1. In conventional thermochemical processes (Figure 1a), the energy required for the reaction is provided in the form of heat by operating at a high reaction temperature. On the other hand, the alternative processes employ a variety of strategies to decrease the temperature of operation: plasma reforming (Figure 1b) utilizes high−voltage electrodes integrated in the reactor design to produce an electrical discharge in the gas phase that causes the formation of plasma with a high electron temperature, which then activates methane and oxidant molecules by forming radical and ionized species which then interact over the catalyst surface [26,27]. Electrochemical processes (Figure 1c) also use electrical current and difference in electrical potential to provide the oxidation of methane and the reduction of the oxidant over the electrodes of a liquid phase or solid phase fuel cell, but they do not involve the formation of plasma and the charge is instead transferred through the diffusion of ions in a liquid or solid media between electrodes. Finally, photochemical processes (Figure 1d) make use of light to provide the energy required for reactant activation. Photocatalytic processes, in particular, can provide very eco−friendly pathways for chemical conversion, especially if directly driven by sunlight [28], and, in recent years, have been studied for a vast number of applications, ranging from wastewater treatment [29] to chemical synthesis [30]. The use of photocatalysis can enable forming reactions in much milder conditions compared to conventional thermochemical processes, lowering CO₂ emissions and limiting problems, such as catalyst sintering [31]. Compared to conventional thermal processes and to electrochemical processes, the photocatalytic approach has the fundamental advantage that, through the activation of chemical bonds by photoexcited electrons, it is possible to drive uphill reactions against thermodynamic restrictions, as the mechanism of reaction is completely changed, allowing researchers to potentially bypass thermodynamic equilibriums [28,32,33]. This advantage is shared with the plasma−reforming processes, where the excited electrons of the non−thermal plasma can also activate chemical bonds; however, photocatalysis has the advantage of not requiring plasma generation, which is an energy intensive process [34,35,36].

One of the most crucial aspects of photocatalysis is, thus, the selection of photocatalysts, which determines the effectiveness of light (LT) utilization through the induction of electrons and charge transfers when exposed to light [37,38]. In the last few years, significant effort has been dedicated to the enhancement of activity of semiconductor photocatalysts for photo−reforming processes by surface and nanostructure engineering, doping, and the utilization of cocatalysts. Previous reviews have been published on the state of the art of the photo−reforming process: Li et al. 2021 [39] and Song and Ye 2022 [40] both reviewed the state of the art of photocatalytic methane conversion, pointing out that the selection of the photocatalyst and the determination of the photocatalytic mechanism are the current bottlenecks of further process development. Recently, Cho et al. [41] provided a detailed review of photocatalysis for methane partial oxidation, steam, and dry reforming, pointing out specifics about reaction mechanisms and requirements for methane activation. Kulandaivalu et al. [42] instead reviewed in greater detail the photocatalytic dry reforming of methane: they noted that for the proper advancement of the process, there is a need to produce novel catalysts able to activate methane as well as carbon dioxide, which are currently lacking in the literature. It should be noted that several of the studied photocatalyst involve the use of expensive noble metal photocatalysts, which would greatly limit the viability of such processes on a large−scale application. Reviews by Liu et al. [25] and Wang et al. [43] discussed the state of the art of semiconductor−based photocatalysts, highlighting the main pathways available for catalyst improvement, such as surface engineering and use of dopants, electron scavengers, and cocatalysts: so far, modified TiO₂, ZrO₂, WO₃, and C₃N₄ have been the most tested photocatalysts. The most recent review by Lei et al. [44] highlights that the study and development of catalysts with a ferroelectric effect may allow researchers to increase catalytic performances. As per the photocatalytic reforming of biomass, recent reviews were provided by Xu et al. [45] and Pan et al. [46]. On the other hand, to the authors’ knowledge, no comprehensive reviews have been published on the use of photocatalysis for tar reforming and abatement. As an alternative to conventional semiconductors, novel covalent organic framework (COF) catalysts have also been proposed [47] but are still in their infancy and suffer from slow preparation, limited versatility in structure modification, and limited efficiency. On the other hand, the development of quantum dots [48,49] and heterojunction−based semiconductor composites [50] could offer great promise in advancing conventional semiconductor materials.

In addition to photocatalyst selection, reaction conditions, such as the selected light source, reaction media, and temperature, as well as reactor design, also play a crucial role in determining process performance. In their reviews, Liu et al. [25] and Wang et al. [43] also briefly discussed the current state of the art of reactor designs for aqueous phase methane conversion. A more extensive discussion on batch and flow reactor designs was presented by Jiang et al. [51], who provided a selection of guidelines for homogenizing experimental practices and comparison of data results to focus on the most promising technologies. Compared to conventional thermal reactors, photoreactors generally face less extreme conditions in terms of reaction temperature and pressure, with their main design challenges instead being the need to properly design the fluid dynamics and the light generation and distribution system [52,53], which are essential to ensure optimal mass transfer and effective light utilization. In the case of aqueous phase reactions, pH and corrosion need to be also accounted for [54,55], as the photocatalytic reaction may require alkaline or acidic environments. The optimization of photocatalytic reactors is, thus, still an open field of investigation.

Finally, the coupling of thermocatalytic and photocatalytic processes, as well as coupled plasma–photocatalytic processes, has also been reported as another interesting perspective to enhance productivity for reforming reactions [56,57].

Despite the increasing effort that photocatalytic reforming has been receiving in the literature, an optimized photocatalytic reforming process is still lacking thus far. Great effort has been dedicated to the reforming of methanol and aqueous organic species, with methane photocatalytic reforming receiving more limited studies, with a great number of available works involving the use of expensive noble metals or toxic species, such as nickel. While solar photocatalytic reforming could allow for substantial reduction in the environmental impact, further improvements in reactor design and catalytic activity are still required [58]. Since photocatalysis is a rapidly growing field, there is a need for frequent reporting of up−to−date information. Thus, this review expands on the previously published literature, with the goal of highlighting new technological developments and strategies to design intensified photocatalytic processes, in particular by adopting fluidized bed (FB) technology.

2. Catalysts for Photoconversion

2.1. General Mechanism of Photocatalysis and Photocatalyst Engineering

An example of a typical photocatalytic mechanism [59,60] for organic compound photooxidation on a semiconductor catalyst is shown in Figure 2.

Generally speaking, the reaction starts with the light−induced excitation of an electron in the valence band of the photocatalyst. The excited electron is promoted to the valence band, leaving a positively charged hole in the conduction band. At this point, the electron can either recombine with the hole and re−emit the captured photon, thus resulting in a zero net−reaction, or it can be transported through the valence band, where it is then captured by an electron acceptor found in the reaction media. At the same time, the valence band hole interacts with the organic molecule, intercepting the electrons of its orbitals (for example the electrons of C−H bonds). The organic molecule is, thus, activated, allowing its reaction with the oxidant and leading to the formation of products and the restoration of the catalyst valence band by the filling of the electron hole due to electron donation, which can be followed by further electron excitation as a new photon is captured, starting the cycle again. Optimization of the band gap energy [61], inhibition of excessive hole–electron recombination [62], distribution and interaction of valence band holes, and conduction bands electrons with organic donors and acceptors [63,64] are, thus, essential aspects of photocatalyst design. In recent years, different approaches have been proposed to optimize activity of semiconductor heterogeneous catalysts. For example, the inclusion of metal and non−metal dopant atoms can induce the formation of localized intermediate energy levels between the conduction and valence bands of the photocatalyst [65,66], which lowers the energy required for electron transitions and widens the radiation spectra that is usable by the material, expanding it to higher wavelengths. Similar effects may also be obtained by defect and surface structure engineering [67,68], and particularly by producing nano−structured photocatalysts [69,70,71] (i.e., nanospheres, nanorods, 2D nanosheets, nanopores, nanotubes, etc.), where the incomplete coordination of atoms in surface and defect sites also causes the formation of energy levels that can act as charge separators, electron traps, or charge recombination sites. Loaded metal species on the catalyst surface demonstrate effective improvement of photocatalyst activity by capturing electrons and reducing charge recombination, as well as improving charge transfer [72,73]. Finally, in the last years, the study of more complex structures, such as quantum dots and heterojunction structures, has also shown promising results. Heterojunctions are formed at the interface by the mixture of different semiconductor materials with an unequal band structure, or at the interface between semiconductors and metals [74]: in these areas, the interaction of the two phases can induce alteration of the energy levels and create an electrical field that enhances charge separation. Heterojunctions between semiconductor surfaces can be distinguished as being type I, type II, type III, S−scheme, or Z−scheme [75], as shown in Figure 3. In type I heterojunctions, also called a “straddling gap heterojunction”, the valence and conduction bands of one of the semiconductors are located in between those of the other. In type 2 heterojunctions, called also “staggered gap” heterojunctions, the VB and CB of one of the catalysts are both higher than the respective bands of the other. Type 3 heterojunctions (“broken gap”), on the other hand, have one of the photocatalyst possess both a VB and CB higher than the other catalyst’s CB. Heterojunctions of the p−n type form at the interface between n− and p−types semiconductors, where the diffusion of electrons from the p−type semiconductor to the n−type one and the counter diffusion of holes from n−type to p−type occurs even before light exposure, leading to the formation of an internal electrical field at the interface that favors electron–hole separation once the material is excited by light. S−scheme heterojunctions (sometimes indicated as direct Z−scheme heterojunctions) are similar to type II, presenting a staggered band gap, but the electron transfer happens between the conduction band of a catalyst to the valence band of the other [76]. Finally, in Z−scheme heterojunctions, the electron transfer is mediated by an intermediate species that acts as an electron carrier [77]. Z−scheme heterojunctions may involve dissolved species in a solution as the electron mediator, or they can involve solid phase mediators. Another type of heterojunction, called a Schottky barrier heterojunction [78], is formed at the interface between a semiconductor and a conductor material, such as a metal, with electrons migrating towards the more conductive material and forming a potential energy barrier that inhibits their back diffusion, thus trapping them on the conductive species.

The type of heterojunction strongly affects the activity of the photocatalyst: type 1 and 3 heterojunctions, for example, cannot inhibit electron–hole recombination, since in the former, both electron and holes tend to migrate in the same direction, while in the latter, diffusion of both electrons and holes is inhibited by the band gap incompatibility of the semiconductors. Thus, type 2, S−scheme, Z−scheme, and p−n type band gaps are the most relevant for photocatalytic applications, with S−scheme and Z−scheme materials attracting much interest due to their superior charge separation ability.

Finally, quantum dots are formed when materials are deposited in near zero−dimensional nanostructures with a radius inferior to the radius of the material’s excitons (excited hole–electron pairs) [79,80]; in these conditions, the band gap of the material becomes size−dependent, offering great tunability of light absorption and activity under visible light, and, due to quantum confinement effects, the number of excitons generated is increased compared to bulk materials.

2.2. TiO₂−Based Catalysts

Among various materials, titanium oxide is a typical choice for photocatalytic processes due to its high activity, good thermal and chemical stability, low cost, and non−toxicity [81]. Unfortunately, titanium oxide on its own is only active under ultraviolet (UV) radiation [82,83], which is only a small fraction of sunlight (3–5%) [60], thus leaving most of the solar spectrum unused and limiting process efficiency. Conventional mercury−based UV lamps pose safety and environmental problems due to the high toxicity of mercury, while they provide a limited light production efficiency with a short lifespan [84]. On the other hand, it should be noted that while the direct use of sunlight is a most desirable outcome, in the last few years, UV technology has also made substantial progress, which can offer the opportunity of efficiently employing UV in photocatalytic processes, thanks to the replacement of conventional mercury lamps with more efficient systems, such as excimer lamps or light−emission diodes (LEDs) [84,85,86,87]. Nevertheless, substantial effort has been dedicated in the last few years to expand TiO₂ photocatalytic activity to the vs., including via surface treatments and functionalization, formation processes, inclusion in composite materials, use of organic and inorganic promoters, and doping of the titania structure with metals and non−metals [88,89,90,91,92,93,94]. In particular, doping titanium oxide and the subsequent formation of lattice defects can introduce intermediate electron levels between the valence and conduction bands and widen the respective bands themselves, thus lowering the band gap and allowing electron excitation and conduction even under the longer wavelengths of visible (VS) and infrared light (IR) [95]. When considering metal promoters and dopants, the addition of noble metals has been shown to significantly enhance TiO₂ photocatalytic activity [96,97]; however, their high cost makes their use unfeasible for large scale operation. Among non−noble metals, both transition metals, such as Ni [98,99], Fe [100,101], Cu [102,103], Co [104], and Mn [105,106], and rare earths, such as Ce [107] and La [108], have all been found to effectively improve photocatalytic activity when used as dopants. While a significant number of the studies on TiO₂ photocatalysis focus on the degradation of organic pollutants, typically in an aqueous environment, photocatalytic reforming and hydrogen production from methane and biomass derivatives have also attracted interest.

Noble metals, and in particular Pt, are able to increase TiO₂ activity for the reforming of methane, biomass derivates, and tar compounds: Yamaguchi et al. [109] observed enhanced performance for steam methane reforming using TiO₂ hollow spheres with spatially separated Pt−Rh noble metal cocatalysts: they proposed that in their hollow TiO₂ sphere structure the excited electrons of the TiO₂ could be transferred to Pt particles loaded in their internal surface, where they then produced hydrogen by water reduction, while the electron holes migrated to the Rh₂O₃ oxide layer deposited on the external surface, where they oxidized methane. This mechanism provided an effective charge separation not only by band gap energy, but also by spatial separation of the electron sink and hole sink materials. On the other hand, Al−Madanat et al. [110] investigated the mechanism of naphthalene aqueous phase reforming for hydrogen formation over Pt−loaded TiO₂, concluding that water is the source of both hydrogen and oxygen. During naphthalene mineralization to CO₂ Pt atoms again act as an electron sink and site for the reduction. Naphthalene is primarily activated through the formation of naphthalene cation radicals. The use of a Pt cocatalyst in conjunction with hydrothermally prepared Nb₂O₅−TiO₂ photocatalysts proved effective for the photoreforming of ethanol aqueous solutions to hydrogen [111], obtaining an apparent quantum yield of 85% under simulated sunlight; the best results were observed for a composite that underwent 3 days of hydrothermal treatment, which outperformed pure P25 TiO₂ catalyst, but in this case no direct investigation of the mechanism was performed, though the authors proposed that the formation of favorable type II and S−scheme heterojunctions is the cause of improved performance for this composite.

The use of nickel has been commonly investigated as an alternative to noble metals. Glycerol photoreforming was investigated by Eisapour et al. [112], who obtained a remarkable 58% glycerol conversion to hydrogen and value−added glyceraldehyde and dihydroxyacetone using a NiO−Ni−TiO₂ sandwich−like heterojunction catalyst, bypassing the limitations of single p−n heterojunction catalysts. Glycerol was converted to glycerate through the formation of glyceraldehyde and to ketomalonate through the formation of hydroxyacetone and hydroxypiruvate. They proposed that the p−n heterojunction formed between NiO and TiO₂ interfaces favored the initial separation of electron–hole pairs, and then the metallic Ni species further captured electrons from TiO₂ by forming a Schottky heterojunction, thus acting as a barrier for charge recombination and functioning as a center for hydrogen production, while electron holes were transferred to the NiO surface and oxidized the organic substrate. On the other hand, Liu et al. [113] obtained a 96.1% selectivity towards syngas formation from glycerol by engineering Ni−O−Ni dimer sites in a TiO₂−derived metal–organic framework (MOF) structure. Single Ni atom deposition reduces the contact angle between TiO₂ and substrate species, favoring their adsorption on the catalyst surface. The electrons of NiO₂ dimers are rapidly redistributed on the TiO₂ surface, where they are used to produce hydrogen, and the formed holes in dimer sites adsorb and activate glycerol, oxidizing it through the formation of formaldehyde and formic acid intermediates. The porosity and structure of the TiO₂ support also contribute to catalytic activity: Xie et al. [114] obtained improved performances for methane dry reforming by depositing Ni on mesoporous TiO₂, while He et al. [115] observed the remarkable effects of the TiO₂ crystal phase for Ni/TiO₂ photocatalysts, with an optimized rutile to anatase ratio leading to better exposed Ni species and remarkable H₂ and CO production rates reaching 87.4 and 220.2 mmol/(g·h), respectively. While the use of nickel is problematic from an environmental perspective, these studies still demonstrate the remarkable effect of catalyst structure engineering on activity and selectivity.

Copper addition may provide an effective alternative to Ni in preparing more active photocatalysts. For example, the sequential deposition of NiCu catalyst over P25 TiO₂ was observed to be effective for the conversion of cellulose to hydrogen and organic acids, with a H₂ yield of 489 μmol/(g·h) for a 1% wt. CuNi−loaded catalyst [116]: sequential deposition of Ni followed by Cu led to a more active catalyst compared to a catalyst produced by simultaneous co−deposition, suggesting improved interaction between metal species, and a low loading was sufficient to enhance activity without incurring deactivation due to excessive TiO₂ site covering observed at higher metal loading, at the same time allowing the reduction in Ni loading. Ni and Cu species were proposed to form Schottky heterojunctions with the TiO₂ surface, acting as electron accumulation and hydrogen generation sites. Aside from the cleavage and oxidation of cellulose by electron holes on the catalyst surface, side reactions of transposition, hydrolysis, and decarboxylation are expected to occur, leading to the formation of subproducts, such as glucose, cellobiose, mannose, maltose, acetic and formic acids, and other. As per pure copper−modified TiO₂−based catalysts, Clarizia et al. [103] provided a review on the aqueous reforming of organic species and observed a great scarcity of hydrogen production efficiency data for such processes. More recently, Muscetta et al. [117] observed a promising performance for aqueous phase reforming of biomass derivates using a ball−milling−prepared Cu₂O−TiO₂ photocatalyst: they observed that a 1% wt. addition of Cu₂O resulted in nearly 60% of the hydrogen yield obtained under sunlight being derived from visible light absorption. The reaction rate could be described by the Langmuir–Hinshelwood model, and it was favored at alkaline pH and a temperature increase from 20 °C to 80 °C. Glycerol and ethylene glycol were the best substrates in terms of hydrogen yield, followed by methanol. Negligible yield was obtained from lactic and formic acid instead, attributed to limits in substrate adsorption on the catalyst surface and the reduced stability of copper species in acidic conditions. They proposed that the reaction mechanism involves the formation of a p−n heterojunction between Cu₂O and TiO₂ surfaces, where Cu₂O can absorb visible sunlight and then transfer electrons to the TiO₂ support. Umair et al. [118] tested a similar catalyst in a 10 L pilot−scale reactor for hydrogen production from biomass derivatives, such as glycerol, ethanol, and glucose, obtaining remarkable hydrogen yields for a 3% wt. Cu₂O−loaded sample ball milled at 150 rpm for 2 h. The addition of Cu₂O effectively shifted the catalyst band gap in the vs. light range, improving light utilization, and the best hydrogen yield was observed for degradation of glycerol. They also supported a mechanism involving the formation of a p−n heterojunction that favors electron–hole separation, where Cu₂O acts as the species active to visible light absorption, with electrons being separated over TiO₂. This represents one of the first efforts to implement photocatalytic hydrogen production on a pilot scale. Khan et al. [119] recently observed high activity for H₂ production from methanol coupled with CO₂ reduction on MOF−structured Cu−TiO₂ composite catalysts, for the first time surpassing the performance of a reference 1% w, Pt−TiO₂ composite after three photocatalytic reaction cycles: the MOF structure highly enhanced catalyst performance compared to simple Cu deposition at the same load, and hydrogen yield improved over subsequent reaction cycles, reaching 13.24 mmol/(g·h) after six cycles. Copper was initially present mainly in the +1 oxidation state in the catalyst, with some metallic Cu nanoclusters also being present, and was reduced to metallic Cu during the reaction. The formation of stable metallic Cu nanoclusters in the MOF structure was credited with the increased catalytic performance of the composite material. The MOF structure prevents aggregation of the nanoclusters and limits their back oxidation, improving catalyst stability. A Schottky heterojunction is formed between these metal nanoclusters and the TiO₂ surface, acting as electron sink which prevents improved charge separation. Carboxylic groups are a fundamental intermediate in the reaction mechanism. The effects of Cu oxidation state on H₂ production from methanol and CO₂ reduction to methane were investigated by Qian et al. [120]: the bandgap of the copper–titania heterojunction catalysts narrowed and the vs. light absorption range significantly broadened with the transition of the Cu valence state from Cu(I) to Cu(0). The more reduced copper species were, thus, found to improve the charge separation of the catalyst and enhance its activity under vs. light. The surface area of the catalyst plays a very significant role in its activity: hydrothermally prepared Cu−TiO₂ samples were found to outperform samples prepared by impregnation [121], even when their oxidation state was unchanged between the two different syntheses; the hydrothermally prepared sample’s increased activity was attributed to improved dispersion of the Cu species and the increased surface area, which favored substrate adsorption and oxidation and decreased the band gap of the photocatalyst due to the better interaction of copper with the TiO₂ substrate. On the other hand, calcination was observed to negatively affect activity, as it reduced the availability of surface copper species. The addition of cerium dioxide together with copper was instead observed by Liu et al. [122], who concluded that a Ti/Ce ratio of 2:1 increases the performance of a Cu/TiO₂−CeO₂ photocatalyst for gas−phase methanol steam reforming, achieving up to 100% methanol conversion and high hydrogen selectivity.

Cobalt is another possible substitute for both Ni and noble metals. A high 95% conversion was observed for reforming glycerol to formic acid and formaldehyde on Co−doped TiO₂ [123]: Co−doping induced lattice strain in TiO₂ and increased the number of oxygen vacancies, which is further increased during light irradiation, providing numerous oxygen adsorption sites; the adsorbed oxygen was reduced to •O²− radicals by electrons concentrated on the Co specie, and these radicals then interacted with the glycerol molecules activated by electron holes, leading to C−C bond cleavage, with the primary oxidation pathway involving glyceraldehyde formation as a reaction intermediate. An S−shaped heterojunction−based 10% wt. −Co_0.3Ni_0.55Se/TiO₂ catalyst achieved a remarkable hydrogen yield of 10,070.9 µmol/(g·h) when degrading 20% vol. triethanolamine (TEOA) aqueous solutions [124]: electrons were concentrated on the CoNiSe clusters and then used to generate hydrogen from H⁺ cations, while holes accumulated in the TiO₂ support and oxidized the organic substrate.

Other metals could also provide useful improvements in photocatalyst activity. Ayodele et al. [125] optimized La−doped TiO₂ through the use of response surface methodology and analysis of variance, obtaining a hydrogen yield of 2.43 μmol/min. In their optimization, they evaluated the effect of four different independent variables (irradiating time, metal loading, methane concentration, and steam concentration) on the rate of hydrogen production, obtaining optimum conditions of 146.15 min, 2.94%, 22.83%, and 1.24% for irradiation time, metal loading, methane concentration, and steam concentration, respectively. Unfortunately, they did not investigate the reaction mechanism and the effects of La addition on the band gap. On the other hand, Zhang et al. [126] tested W−doped TiO₂ nanocrystals for the oxidation of ethylbenzene to acetophenone. The presence of the W atoms induces electron transfer from the TiO₂ to W species, while at the same time inducing lattice distortion and the formation of numerous oxygen vacancies; W species then proceed to reduce oxygen to •O²− radicals, while electron holes activate ethylbenzene. The composite material obtained yields 16 times higher than pure TiO₂ in the same reaction conditions. Finally, Hu et al. [127] demonstrated a 1653 μmol/(g·h) hydrogen yield from MoS₂−modified TiO₂ nanoparticles using agricultural biomass substrates. In this case, the conduction band of MoS₂ acted as sink for the electrons generated by the photoexcitation of TiO₂, which then converted H⁺ cations to hydrogen, while the holes in the titania valence band converted water to hydroxyl radicals that were then scavenged by the biomass substrate. No hole migration is expected to occur between TiO₂ and MoS₂ species, due to the wide difference in energy between the two materials’ valence bands.

Alternatives to metal doping have also been investigated: Falara et al. [128] demonstrated ethanol conversion and H₂ generation on TiO₂ modified by carbon dot deposition, studying quantum dots obtained from hydrothermal treatment of microwave irradiation. They observed a strong dependence of the carbon dots activity on the precursor and methodology, with microwave irradiation−prepared samples outperforming the ones produced by the hydrothermal method. Nitrogen presence in the quantum dots’ precursor is proposed to be important to ensure optimal charge separation efficiency, while excessive quantum dot deposition negatively impacts yield, which was attributed to such phenomena as light scattering recuing the absorption efficiency of light.

Finally, it should be noted that a simple change in catalyst morphology can lead to a very remarkable change in performance even without the presence of dopants or cocatalysts: Ubaidah et al. [129] observed that altering the TiO₂ structure from commercial P25 powder to nanotubes caused the surface area to increase from 55.76 to 120.99 m²/g, while simultaneously reducing the band gap of the material from 3.32 to 3.23 eV and increasing the CH₄ yield from 6560 to 17,020 µmol/(g·h) under UV−VS light. Compared to nanospheres, where electrons can move from the bulk to the surface of the material, in the nanotubes, the electrons are trapped in a new energy band with a lower energy level, and the alternative scenarios for electron and hole dynamic involve the movement of electrons and holes along the axis of the tube, thus making recombination less likely and improving the electron availability for reaction. Jiang et al. [130] also obtained a selective photocatalyst for the photooxidation of methane to formaldehyde by engineering a TiO₂ crystal structure, demonstrating the possibility of controlling process selectivity through photocatalyst structure, and, in particular, they observed that a 90% anatase–10% rutile phase mixture of TiO₂ was able to provide optimized activity towards methane activation through the formation of an heterojunction between the two different crystal phases, with electrons migrating to the anatase and holes migrating to the rutile. The hybrid crystal structure also demonstrated an improved lifetime. Finally, crystal−phase engineering was also demonstrated to be able to drastically enhance the performance of TiO₂ for catalyzing non−oxidative methane coupling, with biphasic TiO₂ (anatase 30% and rutile 70%) exhibiting the highest photocatalytic activity [131].

Morre novel approaches have also been proposed, such as the coupling of photocatalysis with plasma activation. For example, Sun et al. [132] obtained toluene conversion as high as 98% by coupling microwave plasma−catalyzed steam reforming with photocatalysis by anatase TiO₂ at moderate humidities of 38%. Higher humidities were found to decrease conversion.

The results for TiO₂ photocatalysts are reported in

Table 1. TiO₂−based photocatalysts for reforming processes. A = aqueous phase, G = gas phase, AQY = apparent quantum yield, QE = quantum efficiency, and PE = photonic efficiency.

Catalyst	Raw Material	Phase	T (°C)	Product (Max Yield)	Max Light Efficiency (%)	Wavelength	Source
Pt−Rh2O3−TiO2 hollow spheres	Methane	G	300	H2 19.8 mmol/(g·h)	3.51	UV light (Hg−Xe lamp)	[109]
Pt−TiO2	Naphthalene	A	25	−	PE 0.97% H2 production PE 0.33% naphthalene conversion	Simulated sunlight (Xe lamp)	[110]
Pt−Nb2O5−TiO2	Ethanol	A	25 (UV light) 40 (Natural sunlight)	H2 138 mmol/(g·h) UV light H2 40 mmol/(g·h) natural sunlight	AQY 44% (UV light) AQY 85% (UV fraction of sunlight).	365 nm Natural sunlight	[111]
NiO−Ni−TiO2 heterojunction	Glycerol	A	−	H2 24.5 mmol/(g·h) dihydroxyacetone, glyceraldehyde	−	Simulated sunlight (Xe lamp)	[112]
MOF−derived (O−Ni2)/TiO2	Glycerol	A	10	H2 2.5426 mmol/(g·h) CO 0.3617 mmol/(g·h)	−	365 nm	[113].
Ni/mesoporous TiO2	Methane	G	500–600	H2 44.85 mmol/(g·h) CO 68.78 mmol/(g·h)	−	Simulated sunlight (Xe lamp)	[114]
Ni−TiO2	Methane	G	−	H2 87.4 mmol/(g·h) CO 220.2 mmol/(g·h)	−	Simulated sunlight (Xe lamp)	[115]
CuNi/TiO2	Cellulose	A	15	H2 0.489 mmol/(g·h) CO2 0.224 mmol/(g·h) Sugars * Anions **	−	365 nm	[116]
Cu2O−TiO2	Methanol, ethanol, ethylene glycol, formic acid, glycerol, lactic acid	A	20–80	H2 250 μmol/L	8	>400 nm Natural sunlight	[117]
Cu2O−TiO2 (P25)	Glycerol, glucose, ethanol	A	Max 45	H2 7.54 mmol/(g·h) Carboxylic acids	1.7	Natural sunlight	[118]
Cu−based MOF/TiO2	Methanol	A	−	H2 13.24 mmol/(g·h)	−	Simulated sunlight (Xe lamp)	[119]
Cu2O/TiO2, Cu2O/TiO2/Cu TiO2/Cu heterojunctions	Methanol	A	−	H2 0.27953 mmol/(g·h) CO 0.01058 mmol/(g·h)	−	Simulated sunlight (Xe lamp)	[120]
Cu−TiO2 (one−pot hydrothermal method)	Cellulose	A	25	H2 2.752 mmol/(g·h) CO2 1653 mmol/(g·h) Sugars Organic acids		365 nm	[121]
Mesoporous Cu/TiO2−CeO2	Methanol	G	200–300	H2 78.8 mmol/(g·h)	−	UV light (280–400 nm)	[122]
Co−TiO2	Glycerol acetonitrile	A	34.85	Formic acid 57% Formaldehyde 21%	−	365 nm	[123]
CoxNi1−xSe/TiO2	TEOA	A	30	H2 10.079 mmol/(g·h)	−	Simulated sunlight (Xe lamp)	[124]
La−TiO2	Methane	G	−	H2 1.458 mmol/(g·h)	−	UV light (Hg lamp)	[125]
W−TiO2	Ethylbenzene	A	25	Acetophenone 235 mmol/g	0.7	437 nm	[126]
MoS2−TiO2	α−cellulose biomass ***	A	Max 60	H2 1.653 mmol/(g·h) (cellulose) H2 0.05 mmol/(g·h) (corncob)	AQY 5.62% 380 nm	Simulated sunlight (Xe lamp)	[127]
Carbon dot−TiO2	Ethanol	A	−	H2 0.5436 mmol/(g·h)	−	Simulated sunlight (Xe lamp)	[128]
TiO2 nanotubes	Cellulose	A	−	CH4 17.020 mmol/g CO 213,690 mmol/g Glucose	−	Simulated sunlight (Xe lamp)	[129]
Biphasic TiO2	Methane	A	25	Formaldehyde 24.27 mmol/g	QE 3.28% 313 nm QE 2.52% 365 nm	Simulated sunlight (Xe lamp)	[130]
MOF−derived engineered TiO2	Methane	G	−	C2H6 0.0037 mmol/(g·h)	−	Simulated sunlight (Xe lamp)	[131]
Anatase TiO2	Toluene	G	−	98% toluene abatement H2 yield > 60%	−	UV−VS light generated by microwave–metal discharge	[132]

* Arabinose, glucose, xylose, mannose, cellobiose, and maltose; ** acetate, formate, and oxalate; *** rice straw, corn straw, wheat straw, rice husk, soybean straw, and corncob.

.

2.3. ZrO₂−Based Catalysts

ZrO₂−based photocatalysts are a common substitute for TiO₂, sharing very similar properties in terms of their low cost and high thermal and chemical stability [133,134]. Unfortunately, they also share the problem of having a large band gap energy, thus being active only under UV light and needing modification to operate under vs. radiation. Metal doping of ZrO₂ has, thus, been extensively investigated to enhance its photocatalytic activity [133]. Aside from noble metals, transition metals are the most studied for the enhancement of ZrO₂’s performance, with a great number of recent applications using Ni deposition for enhancing the photoreforming activity of organics. For example, Tian et al. [135] observed enhanced performance of Ni/ZrO₂ nanorods for ethanol photothermal dry reforming in a temperature range of 350–550 °C, with samples prepared by precipitation outperforming samples prepared by impregnation thanks to stronger NiO−ZrO₂ interaction. Ni acted as the electron sink, capturing photoexcited electrons which were then transferred to CO₂ for reduction to CO, while electron holes migrated to the ZrO₂−Ni interface, were they activated ethanol for reaction with the O²⁻ species produced by the previous reduction of CO₂. Interestingly, the use of Ni and Ce as co−dopants for ZrO₂ appears greatly favorable for the methane dry reforming reaction: Du et al. [136] observed that the addition of a small amount of CeO₂ to a Ni/ZrO₂ photocatalyst led to a dramatic increase in hydrogen and CO yields, going from a 542 mmol/(g·h) H₂ yield and a 534 mmol/(g·h) CO yield for the Ce−free sample to a H₂ yield of 713 mmol/(g·h) and CO yield of 693 mmol/(g·h) for a catalyst with 5% wt. Ce addition at 700 °C and a 30 suns irradiation intensity for methane reforming. The addition of cerium dioxide provided sites for oxygen vacancy formation and the activation of carbonated intermediates and contributed to reduced coke formation, with light increasing the generation of surface oxygen vacancies and, thus, increasing the number of active adsorption sites while also providing the active electrons. Similarly, Cheng et al. [137] observed high CH₄ and CO₂ conversions of 51.6% and 56.4%, respectively, for methane photothermal dry reforming at 600 °C over Ce−doped Ni−ZrO₂ catalyst, with CO and H₂ yields of mmol/(g·h) and 83.8 mmol/(g·h), respectively. The addtion of Ce led to a drop in band gap energy from 3.08 to 2.04 eV, and, at the same time, a charge redistribution occurred on material surface, leading electrons to migrate from Ce species to O atoms. An internal electric field is formed in the material, which enhances electron redistribution to Ni clusters during light exposure and enhances charge separation. The formation of oxygen vacancies activates CO₂ and enables the avoidance of coke deposition. Tengfei et al. [138] enhanced the performance of Ni/CeO₂−ZrO₂ catalyst by encasing it in a core–shell structure using SiO₂: the resulting catalyst proved stable over 60 h of prolonged reaction, providing CH₄ and CO₂ conversions as high as 63.5% and 55.9%, respectively, with H₂ and CO yields of 137.0 mmol/(g·h) and 182.9 mmol/(g·h), respectively, at 600 °C. Light irradiation considerably lowers the activation energy of the reaction compared to pure thermal process. The addition of nickel–cerium and the core–shell structure were observed to greatly reduce the band gap energy compared to pure ZrO₂. Most recently, Liu et al. [139] conducted a detailed investigation of tailoring of surface properties of a Ni/CeZrO₂ photocatalyst through a MOF structure and achieved conversions of CH₄ and CO₂ as high as 63.9% and 71.1%, respectively, at 650 °C, with the catalyst remaining stable for 250 h of continued reaction. The MOF structure showed resistance to carbon deposition and improved charge separation capacity. The reaction mechanism involved the transfer of electrons to Ni clusters and CO₂ activation to carboxylic intermediates, similar to previous studies.

The deposition of nickel nanoparticles combined with La₂O₃ and ZrO₂ was tested by Meng et al. [140], where 10Ni/La₂Zr₂O₇ demonstrated the highest catalytic activity, with production rates of H₂ and CO of 4572 mmol/(g·h) and 5946 (g·h), respectively. Ni acted as the methane activation site, while oxygen vacancies on the La₂O₃ surface activated carbon dioxide, with light irradiation improving the stability of the vacancies. Light was also observed to inhibit reverse water–gas shift reaction.

As per nickel−free catalysts, Mendoza et al. [141] synthetized nanorods of Bi₂S₃ supported on ZrO₂ and found them to be effective for the production of hydrogen from methanol in an aqueous solution: the H₂ yield was 4440 μmol/(g·h) for 6% wt. Bi₂S₃/ZrO₂ under UV light, six times the yield for bare ZrO₂, and 1476 μmol/(g·h) under vs. light. The Bi₂S₃ nanorods acted as electron sinks for the photoexcited electrons of ZrO₂. On the other hand, Yu et al. [142] obtained promising performance for a mixed Cu/Zn/Zr oxides catalyst for the reforming of aqueous−phase methanol at temperature as low as 130 °C under 16 suns of irradiation. ZrO₂, ZnO, and CuO oxides are proposed to work as sequential type I heterojunctions, with both electrons and holes migrating to CuO from ZrO₂ using ZnO as a carrier. At the same time, ZnO accepts hydrogen cations produced by the activation of methanol, and the photo−excited electrons accelerate the production of formate intermediate species.

Table 2. ZrO₂−based photocatalysts for reforming processes. A = aqueous phase; G = gas phase.

Catalyst	Raw Material	Phase	T (°C)	Product (Max Yield)	Max Light Efficiency (%)	Wavelength	Source
Ni/ZrO2 rods	Ethanol	G	350–550	Max ethanol conversion = 70% H2 and CO selectivity > 40%	−	Simulated sunlight (Xe lamp)	[135]
Ni−CeO2−ZrO2	Methane	G	700	H2 713 mmol/(g·h) CO 693 mmol/(g·h)	−	Simulated sunlight	[136]
Ni−Ce−ZrO2	Methane	G	400–600	H2 83.8 mmol/(g·h) CO 59.1 mmol/(g·h)	−	Simulated sunlight (Xe lamp)	[137]
Ni/CeO2–ZrO2@SiO2 core–shell structure	Methane	G	400–600	H2 137.0 mmol/(g·h) CO 182.9 mmol/(g·h)	−	Simulated sunlight (Xe lamp)	[138]
MOF−derived Ni/CeZrO2	Methane	G	400–650	H2 171.7 mmol/(g·h) CO 182.6 mmol/(g·h)	−	Simulated sunlight (Xe lamp)	[139]
Ni/La2Zr2O7	Methane	G	700	H2 4572 mmol/(g·h) CO 5946 mmol/(g·h)	−	Simulated sunlight (Xe lamp)	[140]
Bi2S3 nanorods/ZrO2	Methanol	A	−	H2 4.440 mmol/(g·h) in UV light H2 1.476 mmol/(g·h) in vs. light		254 nm UV light (high pressure Hg lamp pen−ray) 540 nm vs. light (blue LED lamps)	[141]
Cu/Zn/Zr oxide nanocatalysts	Methanol	G	130	H2 67.37 mmol/(g·h)	45.6% solar–chemical conversion efficiency	Simulated sunlight (Xe lamp)	[142]

summarizes the ZrO₂ catalysts herein discussed.

2.4. C₃N₄−Based Catalysts

Graphitic carbon nitride (C₃N₄)−derived materials are another common choice for photocatalytic reforming applications [143,144]: compared to TiO₂, C₃N₄ possesses a narrower band gap and is, thus, intrinsically more active under vs. light irradiation, while being metal free and chemically stable; however, it also generally suffers from a more complex preparation, which is an obstacle to scaling up the process, and still requires modification to achieve adequate performance under solar irradiation, particularly to avoid excessive charge recombination [145,146]. Again, noble metals are found to be effective cocatalysts for these applications. Speltini et al. [147] compared the performance of Pt and Cu−Ni cocatalysts supported on either TiO₂ or C₃N₄ for aqueous−phase H₂ production from biomass derivatives and observed overall lower performance for C₃N₄ photocatalysts; it should, however, be noted that C₃N₄ still managed to achieve H₂ yields within 40 to 75% of those obtained on TiO₂. In both cases, the metal cation acted as the hydrogen generation site, while electron holes on the semiconductor surface oxidized the organic substrate. Most recently, Bianchini et al. [148] reported successful pre−pilot scale generation of hydrogen from rice industry wastewater using C₃N₄ with H₂PtCl₆ used as a metal cocatalyst under natural sunlight. They reported an analysis of the process through a response surface model that was used to determine the optimal dilution of the waste to degrade, the optimal catalyst amount, and the optimal loading of the Pt cocatalyst, obtaining their best results with undiluted waste, a 0.5 g/L catalyst concentration, and a 35 wt. Pt loading. Considering the cost of noble metals, the optimization of their load is of great importance in their use as catalysts. While they did not investigate in detail the reaction mechanism, tests of the catalyst in pure water and in dark conditions produced a negligible hydrogen yield, confirming the crucial role that biomass plays as an electron donor and the fundamental role of light in promoting the reaction.

Single−atom metal deposition may offer the possibility for preparing selective catalysts with a low metal load and, thus, reduce cost of noble metal catalyst use: Lazaar et al. [149] deposited highly dispersed Pt atoms on exfoliated C₃N₄, obtaining a maximized H₂ production of 1660 mmol/(g_Pt·h) from 10% vol. aqueous triethanolamine solution with a total Pt loading of only 0.03% wt. A dark reactive deposition approach was used to obtain stable single−atom surface species chemically bound to carbon nitride, with Pt atoms being bound to N4 sites with high charge transfer ability. The high dispersion of Pt atoms increased their activity and prevented agglomeration, contributing to catalyst stability and allowing the catalyst to surpass the performance of other catalysts loaded with conventional techniques. The morphology of C₃N₄ also plays an important role in its activity: for example, Pt−loaded ultrathin C₃N₄ nanosheets were found to be more effective than Pt loaded on bulk C₃N₄ for the aqueous reforming of cellulose [150]: the nanosheet structure increased the oxidation potential of the carbon nitride, favoring the formation of hydroxyl radicals that could then react with biomass, while at the same time improving electrical conductivity, reducing band gap, and favoring electron–hole separation. The load of the cocatalyst was essential to optimize the yield, with a 2% wt. load providing optimum performance. On the other hand, the addition of other cocatalysts is also a potentially effective pathway. For example, Gao et al. [151] observed a 3.6−times increase in hydrogen production with a 5% wt. addition of SnS₂, also obtaining an increase in degradation efficiency for methylene blue dye. A type II heterojunction was formed between carbon nitride and SnS₂, with SnS₂ acting as the electron sink. The addition of methyl viologen to the C₃N₄ structure was instead observed to enhance hydrogen production in the aqueous phase with the triethanolamine sacrificial agent [152], with methyl viologen species accumulating electrons formed by the photoexcitation of carbon nitride and accelerating their transfer to Pt cocatalyst, thus inhibiting charge recombination.

Other metal species have been tested as cheaper alternatives to noble metals. Similarly to the previously discussed investigation of Lazaar et al. for single Pt atom deposition, Li et al. [153] maximized the yield of direct methanol synthesis from CH₄ by controlling the coordination number of single iron atom loading on carbon nitride surface, obtaining a remarkable methanol yield of 928.27 μmol/g for the Fe₁/C_3−xN₄ with a Fe−N₃ coordination. In these conditions, they observed that the reaction proceeded through a three−electron mechanism involving methane adsorption and activation as methyl surface group and the production of hydroxyl radicals from oxygen.

Hierarchical carbon nitride structure (H−C₃N₄) also demonstrated enhanced and stable activity for methane dry and bi−reforming, as well as methanol reforming [154], which was further enhanced by cobalt doping: under dry methane reforming with stoichiometric feed, an optimized 2% wt. Co/H−C₃N₄ achieved CO and H₂ evolution rates of 555 and 41.2 μmol/(g·h), respectively, an increase of 18.28− and 1.74−fold than when using H−C₃N₄, itself possessing 1.85− and 1.81−fold higher yields than graphitic C₃N₄. The higher yields can be attributed to higher charge carrier separation induced by doping and structural modification. For photocatalytic bi−reforming of methane, the production of CO and H₂ was reduced, whereas significantly higher CO and H₂ evolved using the bi−reforming of methanol, achieving 10.77− and 1.39−fold more H₂ and CO efficiency compared to dry reforming of methane in the latter case. In both cases, cobalt acted as the electron sink enhancing charge separation, while organizing the carbon nitride into layered hierarchical structure improved its own charge separation capacity by forming defects. CO and H₂ were the main reaction products, produced from the transfer of two electrons.

The formation of heterojunctions of carbon nitride with other semiconductors was investigated by other authors. Khan et al. [155], for example, investigated the use of cobalt aluminum lanthanum−layered double hydroxide (LDH) as a support for a TiO₂−Ti₃C₂ and C₃N₄ mixture in methane bi−reforming with water and carbon dioxide. The strong interaction between the three species allowed the researchers to reduce the charge carrier recombination, and the material reached maximum production of 24.35 μmol of production for CO at a catalyst loading of 0.13 g with a feed ratio of 1.67 at 4.47 h, while the maximum H₂ production of 21.91 μmol was achieved at a catalyst loading of 0.14 g with the feed ratio of 1.41 at 4.93 h. A four−step heterojunction was formed between Ti₃C₂, TiO₂, C₃N₄, and the cobalt−based LDH, with electrons concentrated on the carbide and LDH species. S−scheme heterojunctions formed between TiO₂ and LDH and between TiO₂ and carbon nitride, allowing for efficient charge transfer and separation. The good integration of all the catalyst species also allowed the enhancement of the surface area, thus favoring reactant adsorption. Tahir et al. observed an enhancement in reforming performance by addition of 15% wt. La_xCo_yO₃ perovskite for aqueous phase reforming of methanol [156], also forming an S−scheme heterojunction. Similarly promising results were also observed for S−scheme heterojunction composites prepared by the combination of C₃N₄ with MnMgPO₄ [157]. S−scheme C₃N₄ heterojunctions have also received interest for the photodegradation of plastic waste, with Mohanty et al. [158] observing a hydrogen evolution rate of 12.6 ± 2 mmol/(g·h) for combined C₃N₄/α−MnO₂ with the concurrent formation of value−added organic molecules (benzaldehyde, benzoic acid, toluene, benzene, and carbonic acid) from polystyrene reforming.

A metal−free modification of carbon nitride was instead attempted by Ikreedeegh and Tahir [159], who deposited graphene oxide (GO) nanosheets on a carbon nitride surface by a sonochemical thermal−assisted method and then tested the material for methane dry reforming under vs. light: they observed that a 0.5% wt. addition of graphene oxide increased the yield of CO up to 5 times compared to unmodified material, while a 0.25% wt. addition maximized H₂ production. They also observed that dry reforming carried out with a stoichiometric CO₂/CH₄ ratio (1:1) performed better than both steam and bi−reforming on this catalyst; however, the quantum yields of the catalysts were still below 0.4% even for the optimized graphene oxide loading. The activity of the material in this case was enhanced by the graphene oxide acting as electron sink, similarly to what is commonly obtained through the loading of metal species. On the other hand, Xie et al. [160] demonstrated that carbon ring inclusion in C₃N₄ can potentially provide a pathway for photocatalyst activation under low−energy near−infrared light (NIR) for the degradation of lignocellulosic material coupled with H₂ production. The inclusion of carbon graphitic rings in the catalyst structure introduced intermediate energy levels in the band gap structure, allowing the adsorption of near−infrared radiation, and created regions of low electron density, polarizing the surface and facilitating electron migration and charge separation.

Table 3. C₃N₄−based photocatalysts for reforming processes. A = aqueous phase, G = gas phase, AQY = apparent quantum yield, APE = apparent photon efficiency, GO = graphene oxide, LDH = layered double hydroxide, and NIR = near−infrared.

Catalyst	Raw Material	Phase	T (°C)	Product (Max Yield)	Max Light Efficiency (%)	Wavelength	Source
Pt−oxidized g−C3N4	Glucose	A	−	H2 1.370 mmol/(g·h)	AQY 0.8% (native pH) AQY 1.3% (pH 11).	Simulated sunlight, natural sunlight	[147]
Pt−Poly(heptazine imide)−C3N4	Glucose, rice milk wastewater	A	−	H2 1.000 mmol/(g·h) (glucose) H2 0.150 mmol/(g·h) (rice milk wastewater) H2 0.11P5 mmol/(g·h) (rice milk wastewater, natural sunlight)	−	Simulated sunlight, natural sunlight	[148]
Single−atom Pt−C3N4	TEOA	A	−	H2 1660 mmol/(gPt·h)	−	LED 365 nm	[149]
Pt−2D C3N4 nanosheet	Cellulose	A	−	H2 0.09433 mmol/(g·h) in NaOH 3M		Xe lamp with a UV cut−off filter (400 nm < λ < 780 nm)	[150]
Pt−SnS2/C3N5	TEOA	A	60	H2 0.9225 mmol/(g·h)	−	Xe lamp with a vs. light cut−off filter (λ: 400−780 nm)	[151]
Pt−methyl viologen−C3N4	TEOA	A	−	H2 1.65 mmol/(g·h)	−	Xe lamp with a UV cut−off filter (λ > 420 nm)	[152]
Fe−C3N4	Methane	A	25	Methanol 0.92827 mmol/g	−	Simulated sunlight (Xe lamp)	[153]
Co−C3N4	Methane, methanol	G		Methane dry reforming: CO 0.555 mmol/(g·h) H2 0.0412 mmol/(g·h) Methanol bi−reforming: CO 0.771 mmol/(g·h) H2 0.444 mmol/(g·h)	−	Hg lamp	[154]
g−C3N4/TiO2A/R@Ti3C2/CoAlLa−LDH	Methane	G	27	H2 0.04150 mmol/(g·h) CO 0.03734 mmol/(g·h)	AQY 0.407% CO AQY 0.363% H2	Xe lamp 420 nm	[155]
LaxCoyO3−3D g−C3N4 hollow tube	Methanol	A	25	H2 0.070 mmol/(g·h)	APE 0.58%	Xe lamp λ ≥ 420 nm	[156]
MnMgPO3−C3N4	Methylene blue, oxytetracycline, tetracycline	A	25	H2 3.595 mmol/(g·h)	−	VS light (Xe lamp)	[157]
α−MnO2/g−C3N4	Polystyrene	A	−	H2 12.6 mmol/(g·h) benzaldehyde, benzoic acid, toluene, benzene, and carbonic acid	−	Xe lamp	[158]
GO−g−C3N4	Methane	G	−	H2 0.03024 mmol/(g·h) (0.25% wt. GO) CO 0.3991 mmol/(g·h) (0.5% wt. GO)	AQY 0.9% (0.5% wt. GO)	Simulated sunlight (Xe lamp)	[159]
g−C3N4−Cx carbon ring Modified C3N4	α−cellulose Lignocellulose	A	25	H2 2.1 μmol/h (NIR irradiation) mannose, galacturonic acid, glucuronic acid, galactose, lactic acid, formic acid, and 5−hydroxymethyl furfural	AQE 4.7% at 550 nm	Xe arc lamp coupled with a λ ≥ 630 and λ ≥ 800 nm filter	[160]

summarizes the literature reviewed for C₃N₄−based photocatalysts.

2.5. Other Catalysts

While the previously discussed TiO₂, ZrO₂, and C₃N₄ remain the most investigated materials for photocatalytic reforming processes, a number of other materials have also been tested as possible alternatives. Perovskite−based photocatalysts, for example, have been tested by Chung et al. [161] in an innovative combined plasma–photocatalytic process for methane dry reforming: synergistic effects between plasma− and photocatalysis contributed to increasing the H₂ yield, and among LaFeO₃, NiTiO₃, and AgNbO₃, the former displayed the highest performance in terms of both reactant conversion and energy efficiency. Not only are methane and CO₂ pre−excited by exposure to the high electron temperature plasma, but the photocatalyst can be excited by plasma−produced photons, as well as by the capture of previously emitted free electrons. Plasma exposure can modify the photocatalyst surface structure and reduce particle aggregation, and the presence of a photocatalyst may enhance the power of plasma discharge thanks to photocatalyst polarization. In their further work [162], they concluded that LaFeO₃ calcined at 600 °C demonstrates optimized CH₄ and CO₂ conversions and syngas generation efficiency of 53.6%, 40.0%, and 18.4 mol/kWh, respectively. On the other hand, LaFeO₃ doped with 0.12% wt. Ni was observed to be active also for photoreforming of glucose to hydrogen in aqueous solution, with H₂ yield of 2574 μmol/L after 4 h of UV irradiation [163]. Hydrogen was primarily produced from water reduction over Ni atom electron sinks, while glucose played the role of the hole scavenger. Hydroxyl radicals were identified as the main reactive species involved in the oxidative degradation process. Another perovskite, LaMn_xNi_1−xO₃ (0 ≤ x ≤ 1), proved effective for the coupled thermo−photocatalytic conversion of toluene as a tar model compound [164], with 90% carbon conversion of toluene obtained on LaMn_0.4Ni_0.6O₃ at 400 °C. The formation of oxygen vacancies and activated oxygen species under material irradiation was observed to play an important role in the adsorption and conversion of toluene. Overall, temperature and light displayed a synergistic effect on pollutant degradation.

Cerium dioxide also found interest in its use as a primary photocatalyst rather than only as an additive: Tavasoli et al. [165] investigated the use of Ni−supported catalysts on selectively phosphated CeO₂ nanorods for the hybrid photothermal dry reforming of methane, obtaining results close to the ones obtained for noble metal catalysts. The phosphate groups allowed the researchers to tune the ceria surface alkalinity, and again temperature and light adsorption were observed to synergistically contribute to conversion. Light exposure enhances formation and promotes the reaction of surface carbonyl species, The combined catalyst proved to be active even under visible light−only monochromatic irradiation. Balsamo et al. [166] coupled cerium dioxide and graphene oxide to the use of an Au cocatalyst and obtained a 270 μmol/(g·h) H₂ evolution rate from aqueous−phase 10% vol. glycerol solution under simulated sunlight. The material displayed stable operation, and hydrogen yield increased following the reduction of graphene oxide after five operation cycles. The graphene oxide acted as a mediator for the transfer of electrons from cerium to gold atoms. Iannaco et al. [167] performed photocatalytic hydrogen generation from lactic acid rich solutions, obtaining a hydrogen production of 3989 μmol/L for sample prepared with the supercritical antisolvent technique, higher than those obtained for commercial CeO₂ powder (2519 μmol/L). The addition of an optimal CuO amount of 0.5% wt., determined an increase in hydrogen production of up to 9313 μmol/L after 4 h of UV irradiation time. They also confirmed that water acted as the hydrogen source, while lactic acid acted as sacrificial agent to refill the electron holes of the catalyst. The ceria and CuO interaction formed a type I heterojunctions, with electrons and holes being transferred to CuO. Zhang et al. [168] observed that the surface dispersion of Cu atoms over the CeO₂ surface affected greatly the degradation of aqueous−phase toluene, with more even distribution of active sites Cu⁺−Ov−Ce³⁺ improving the adsorption and activation of reactants. On the other hand, surface lattice oxygen accelerates the rate−determining step of benzoate deep decomposition. The toluene degradation pathway was toluene → benzyl alcohol → benzaldehyde → benzoate →anhydride → CO₂ and H₂O. Aoun et al. [169] observed that a 15% wt. SrNiO₃−CeO₂ composite rapidly oxidated 2−propanol under both UV−LED and vs. light. A type I heterojunction was formed between ceria and the perovskite, with electrons and holes migrating to SrNiO₃. Li et al. [170] synthetized a S−scheme heterojunction by depositing AgInS₂ quantum dots on CeO₂ surface and obtained the selective oxidation of xylose to xylonic acid and CO under vs. light (xylonic acid yield = 60.0%, CO evolution rate = 3689.9 μmol/(g·h)). On the other hand, Wang et al. [171] prepared a noble metal−free CeO₂/CdS/NiS triple heterojunction with remarkable photocatalytic performance for the degradation of plastic waste, obtaining a hydrogen production rate of 32.40 mmol/(g·h) at the fifth hour, 29 times that of bare CdS and 463 times that of bare CeO₂. In this case, Ni S was proposed to act as electron mediator between CdS and ceria conduction bands, with ceria acting as hydrogen production site and CdS electron holes carrying out the oxidation of waste.

WO₃ is another material that has been tested for photocatalysis. Li et al. [172] reported the synthesis of a 2D CoP S−scheme heterojunction over a 0−dimensional WO₃ support for H₂ generation from a triethanolamine aqueous solution. CoP acted as the electron sink, while WO₃ accumulated the electron holes. They obtained 218.63 mmol of hydrogen under vs. light exposure within 5 h of reaction, 287 times the amount of hydrogen formed for pure WO₃ and 1.95 multiples the one produced from pure CoP. Similarly, the addition of CuO to WO₃ increased the hydrogen yield from aqueous methanol solution by 64 times [173]. Surface oxygen vacancies capture the oxygen atom of CH₃OH, while the Cu²⁺ cations activate the C−H bond. During the process, CH₃OH accepts photogenerated electrons from WO₃ and provides electrons to Cu²⁺. The addition of WO₃ and graphene to TiO₂ increased the catalytic activity for gaseous phase benzene degradation and hydrogen production [174]: the addition of WO₃ only forms a type II heterojunction, improving charge separation, but reduces the activity compared to pure TiO₂ due to the low potential of the photoexcited electrons. On the other hand, the addition of graphene not only facilitates electron transfer, but also offers a surface with a potential more suitable for hydrogen generation, thus compensating activity loss. However, despite the higher hydrogen production, the graphene−modified catalysts still display low oxidation capacity towards benzene. N−n heterojunction CuMn₂O₄/WO₃ nanocomposites constructed by a sol–gel procedure were evaluated for H₂ evolution from aqueous glycerol under vs. light by Albukhari et al. [175]. Their optimized 12% wt. CuMn₂O₄/WO₃ nanocomposite exhibited a high maximal H₂ evolution rate of 2856.6 μmol/(g·h). An S−scheme heterojunction transfers electrons from WO₃ to CuMn₂O₄, which are then concentrated on Pt species.

Other catalytic systems were also investigated. Ye et al. [176], for example, investigated the use of 2D V_xW_1−xN_1.5 bimetallic systems as cocatalysts coupled with CdS for the reforming of formic acid to syngas. A formulation of V_0.1W_0.9N_1.5 boosted performance by over 60% compared to the unmodified W₂N₃ system. The V_xW_1−xN_1.5 species acted as electron sinks for electrons of CdS. Zhou et al. [177] used Ni nanoparticles loaded on mesoporous silica for the photo−thermocatalytic steam reforming of cellulose under UV−VS light and obtained high production rates of H₂ and CO (1966.2 and 1257.7 mmol/(g·h)) together with a light−to−fuel efficiency of 5.5%. On the other hand, Zhong et al. [178] performed the same reaction while using Ni nanoparticles loaded on θ−Al₂O₃, and they obtained significantly higher production rates of syngas (H₂ 3776.3 and CO 2028.1 mmol/(g·h)) and no deactivation after four cycles. In both cases, light and heat acted synergically to enhance reaction rate, and the degradation of cellulose followed a pyrolytic mechanism.

The literature on the catalysts discussed is summarized in

Table 4. Other photocatalysts for reforming processes. A = aqueous phase, G = gas phase, L = liquid phase, and AQY = apparent quantum yield.

Catalyst	Raw Material	Phase	T (°C)	Product (Max Yield)	Max Light Efficiency (%)	Wavelength	Source
NiTiO3, LaFeO3 AgNbO3	Methane	G	650–680	LaFeO3 calcined at 600 °C: CH4 conversion 61.9% CO2 conversion 61.9% H2 selectivity 82.5% CO selectivity 69.9%	−	Light from plasma generation system, simulated sunlight (Xe lamp)	[161]
LaFeO3	Methane	G	650–680	CH4 conversion 53.5% CO2 conversion 40.0% H2 selectivity 85.0% Co selectivity 71.8%	−	Light from plasma generation system	[162]
Ni−LaFeO3	Glucose	A	−	H2 2.573 mmol/L	−	UV lamps (emission peak at 365 nm)	[163]
LaMnxNi1−xO3	Toluene	G	400	90% toluene conversion	−	Xe lamps λ = 350–780 nm	[164]
Phosphated Ni−CeO2 nanorod	Methane	G	350	Stable H2 and CO rate ≈ 5 mmol/(g·h) at 48 suns for sample with P/Ce = 0.65	−	UV light, red light, blue light, green light	[165]
Au−CeO2−rGO	Glycerol	A	−	H2 0.270 mmol/(g·h)	−	Simulated sunlight (Xe lamp)	[166]
CuO−CeO2	Lactic acid	A	450	H2 ≈ 2.6 mmol/(g·h)	AQY 0.459%	UV light 365 nm	[167]
Cu−CeO2	Toluene	A	200–231	Toluene conversion 89.9%	−	Simulated sunlight (Xe lamp)	[168]
SrNiO3/CeO2	2−propanol	G	−	100% propanol conversion	−	UV lamp 365 nm Xe lamp 315–400 nm	[169]
AgInS2 quantum dots−CeO2	Xylose	A	30–80	Xylonic acid yield 60.0% CO 3.6899 mmol/(g·h)	−	VS light	[170]
CeO2/CdS/NiS	Polylactic acid hydrolysate	A	6	H2 32.40 mmol/(g·h) pyruvate, acetate, and formate	−	VS light (Xe lamp)	[171]
2D CoP supported 0D WO3	TEOA	A	−	H2 0.21863 mmol in 5 h	AQY 2.02% 520 nm	VS light 450, 475, 500, 520 and 550 nm	[172]
CuO/WO3	Methanol	L	25	H2 0.0996 mmol/(g·h)	−	VS light	[173]
TiO2/WO3/graphene	Benzene Methanol	A	−	H2 0.054 mmol/g in 4 h from methanol	−	Simulated sunlight	[174]
Pt−CuMn2O4/WO3	Glycerol	A	−	H2 3.472 mmol/(g·h)	−	Xe lamp λ > 420 nm	[175]
2D VxW1−xN1.5 + CdS	Formic acid	A	25	H2 0.18404 μmol/h	−	Simulated sunlight (Xe lamp)	[176]
Ni−SiO2	Cellulose, rice straw, wheat straw, corn stalk, kitchen waste		500–700	Cellulose: H2 1966.2 mmol/(g·h) CO 1257.7 mmol/(g·h) Rice straw: H2 1488.1 mmol/(g·h) Wheat straw: H2 728.4 mmol/(g·h) Corn stalk: H2 1036.5 mmol/(g·h) Kitchen waste: H2 760.4 mmol/(g·h)	Light−to−fuel efficiency 5.5% (cellulose)	Xe lamp	[177]
Ni/θ−Al2O3,	Cellulose	A	200–700	H2 3776.3 mmol/(g·h) CO 2028.1 mmol/(g·h)	−	λ > 420 nm, and λ > 560 nm	[178]

.

3. Reactor and Process Design

Like photovoltaic generation, the requisite for the utilization of a photocatalyst in chemical reactors is the availability of a suitable light source and low attenuation of the irradiated power. Packed bed reactors, filled in with monoliths or granules, that are normally used for catalytic reactions even at high superficial velocity [179], are not usable because of the large attenuation of the irradiation in the bulk.

Conversely, a typical photoreactor has a planar or two−dimensional geometry, like in the “carpet”, “pancake”, “toothpick”, and “honeycomb” schemes reported in Figure 4 [180], with a transversal or longitudinal irradiation. The photocatalyst powder is dispersed and fixed over a suitable support that can be easily irradiated by natural or artificial light. The reactant is in contact with the catalyst mainly by a rather laminar flow, resulting in poor process intensification. Apart from reactions occurring in liquids, this configuration is suitable for the removal of pollutants from the environment, e.g., indoor air purification [181]. Photocatalytic CO₂ reduction by steam has recently been addressed at laboratory−scale or modelled, demonstrating the possibility of distributed H₂ production, even with solar irradiation [182,183].

Fluidized bed (FB) technology has been largely adopted for catalytic heterogeneous reactions thanks to its good mixing, excellent heat and mass transfer, and process reliability [184]. Normally, the FB reactor operates with a dense phase that is fluidized by a gas stream at velocities higher than the minimum for fluidization (U_mf). Diluted fluidized beds are operated at a high gas velocity equal or higher than the terminal velocity U_t of the particle, thus allowing large plant potentiality. Therefore, FBs perform well when large diffusion coefficients and high heat transfer are needed for the catalytic process, for instance in the case of steam [185] and dry [186] reforming of hydrocarbons. They also allow easy regeneration of spent catalysts by looping with an external unit operated at different temperatures and atmosphere [187]. Nowadays, the utilization of photocatalytic fluidized beds is limited to only a few research activities dealing with pollutant conversion [188]. The photocatalytic elimination of phenol−derived compounds was studied in an FB reactor with a liquid/solid emulsion provided with a concentric UV light source [189]. The process efficiency increased with decreasing particle size, even for the same surface area concentration, with larger particles causing a blocking effect between the light source and the catalyst surface. Similarly, an excess of particle holdup caused light shielding, and an optimum bed concentration of 20 g/L of mixture was determined. Photocatalytic degradation of aromatics was successfully investigated in a FB photoreactor at lab−scale equipped with a UV source, taking advantage of the increased mass transfer [190]. For a non−purification−related application, Vaiano et al. [191] addressed the photocatalytic reduction of CO₂ in FB via a steam reaction. They reported that an optimized formulation of the photocatalyst based on Ti and Pd, along with a UV−illuminated fluidized bed, can effectively allow the photocatalytic conversion of CO₂, under mild reaction conditions.

The utilization of fine photocatalytic particles that are typically used over supports in other devices, e.g., self−cleaning TiO₂ coating [192], is problematic in fluidized beds due to channeling and poor fluidization according to the Geldart classification of particulates [193]. In this respect, the synergy with other physical phenomena, like magnetism or sound [194,195], may significantly improve the quality of fluidization and the whole performance of the photocatalytic reactor. More frequently, the active phase is deposited over easily fluidizable and mechanically resistant particles, for instance γ−alumina, as reported for tar abetment in a FB gasifier [196]. In the case of photocatalytic reactions, the internal surface area could result inaccessible to irradiation, but inner sites may be suitable for combined steps, e.g., adsorption/release of CO₂, in the whole process [197]. The use of two−dimensional or toroidal geometry systems favors uniform illumination of the catalyst but limits the potential of the system due to the smaller cross−section. Other FB configurations, e.g., conical geometry with an internal jet [198], could be particularly effective in gas–solid mixing and generation of diluted regions for vs. or UV irradiation. Figure 5 displays the different FB configurations.

In a pioneering work [199], light transmission in a dense 2D fluidized bed was measured and correlated in regimes with up to 10 times U_mf showing a rapid attenuation of the flux after a few centimeters. Thus, dense fluidized bed photoreactors can be limited to small−scale applications, as in the case of bidimensional devices, while large reaction chambers can effectively operate with a disperse particle phase [200] and a gas velocity approaching or exceeding U_t. In such transport regime conditions, the solid volume fraction can be limited to about 1%, despite having high solid flows, e.g., 20 kg m⁻² s⁻¹ [201].

Table 5. Comparison between different photoreactor configurations.

	Two−Dimensional	Fluidized Bed
Ease of construction	High	Low
Catalytic device complexity	High	Moderate
Pressure drop	Low	Moderate
Scalability	High	Moderate
Heat removal/supply	Moderate	High
Mass transfer	Moderate	High
Process integration	Low	High
Process intensification	Low	High

reports a short comparison between bidimensional and fluidized bed photocatalytic reactors. The 2D fixed configuration is more suitable for carrying out unitary operations, like purification, in easier, scalable, and less demanding equipment. In contrast, combined/intensified processes, requiring mass and heat transfer, are better conducted in FB configuration. The photocatalyst is more easily prepared in the form of fluidizable granules, for example by spray drying [202], rather than demanding two−dimensional devices, e.g., honeycomb or toothpick devices [180].

Large scale plants for steam or dry reforming may convert up to 300,000 Nm³/h of methane [203] in multiple channels catalytic reactors, for instance the “Steam reformer tube assembly and method of assembling or retrofitting same” [204]. Since the gas velocity in the packed bed is limited to a few cm/s because of the increased pressure drop at higher velocities, the transversal size of large−scale plants likely exceeds 10 m with a height of 2–3 m [203]. In comparison, the possibility to operate at a higher gas velocity, up to 1 m/s, in fluidized bed reactors [193], may largely reduce the volume of the catalytic reactor. In the case of photocatalysis, the need to provide effective irradiation calls for diluted FB systems (Figure 5b,c) or an array of two−dimensional fluidized beds (Figure 6). Thanks to the great flexibility of FB technology, the development of small−scale integrated systems, e.g., biomass gasification and photoreforming of the generated hydrocarbons, could enable distributed H₂ production, with greater environmental sustainability [205].

The development of the integrated process could benefit from the application of artificial intelligence (AI) and machine learning, both in the development of photocatalytic materials [206] and the optimization of the whole process [207], speeding up the technological progress thanks to increasingly precise predictive models.

4. Conclusions

Titanium oxide is the most diffused material used for photocatalytic processes due to its high activity, good thermal and chemical stability, low cost, and non−toxicity. In the last decades, a lot of research has been carried out to enhance the performance of TiO₂ and to adapt it to different irradiation sources by doping it with other elements or producing nanostructured photocatalysts.

ZrO₂−based photocatalysts possess similar properties as TiO₂, in terms of low cost, high thermal and chemical stability, and mechanical resistance. They can be used under UV light and need modification to operate under VS radiation. Graphitic carbon nitride (C₃N₄) is non−oxidic material suitable for photocatalytic reforming applications, characterized by a narrower band gap and, thus, is intrinsically more active under vs. light irradiation. Conversely, it suffers from a more complex preparation method and related costs.

Overall, prior research has already developed photocatalysts with formulations based on a few elements that are relatively cheap, which allow the development of hydrocarbon reforming processes supported by VS− or UV light.

The application of photocatalysis in large−scale plants and the integration of more processes can be conceived with equipment from the process industry that has so far been scarcely applied in this sector, in particular through the use of fluidization technology. Although these devices may give rise to light attenuation, their superior performance in terms of mixing and energy transport can make up for the limitations of traditional photocatalysis technologies.

With reference to hydrocarbon reforming, the combination of photocatalytic devices and mature plant technologies, such as fluidization, can allow new strategies to obtain better performance and greater flexibility and efficiency, especially in relation to the lower temperature required for the process. Such integration could allow the development of more sustainable processes on a small scale, for example for the photoreforming of biomethane and bio−derivatives obtained from agro−industrial sector.

Although the proposed process integration is still at a low level of readiness, the development of research also supported by powerful artificial intelligence and machine learning tools could allow researchers to achieve pre−industrial solutions in less than 10 years and, therefore, pave the way to commercialization.