**2. Photocatalytic Conversion of CO2: Thermodynamics and Kinetics**

In the photocatalytic approach, semiconductors irradiated by UV or visible light are used for the reduction of CO2. As can be seen in Figure 1, the light energy (*hv*) absorbed by the photocatalyst is used to produce electron–hole pairs, which are generated when that energy is equal or greater than the band gap energy (Eg) of the photocatalyst. Only then will the electrons (e−) be promoted from the valence band (VB) to the conduction band (CB) of the semiconducting material, simultaneously creating a hole (h+) at the VB [20]. The e<sup>−</sup> and h<sup>+</sup> are then transferred to active redox species present across the photocatalytic interface in order to participate in the conversion process [26].

**Figure 1.** Generation of an electron–hole pair in a photocatalyst. Reproduced from work in [20]. Copyright 2011 Royal Society of Chemistry.

The photocatalytic reduction of CO2 into value-added products involves radical-chain reactions that form cation, anion, and electrically neutral or charged radicals. These radicals are produced as a result of the reaction with photogenerated electrons and holes between the metal oxide photocatalyst and the reactants [26,36]. As CO2 will be reduced, the presence of a co-reagent, or an electron donor, that will be simultaneously oxidized, is necessary [37]. A sacrificial donor (D), usually water, is oxidized by the photogenerated holes in the valence band, while the electrons in the conduction band reduce CO2 [38]. Hydroxyl groups present on the surface of the photocatalyst might also be oxidized by the holes [39]. The end products of the photocatalytic reduction of CO2 depend on both the redox potential (thermodynamics), and surface electron density and photoadsorption/photodesorption thermodynamics and kinetics [40].

The overall efficiency of the photocatalytic reduction of CO2 is governed by the effectiveness of three processes: the photogeneration of the electron/hole pair, the interfacial transfer of electrons between the photocatalyst surface and the adsorbed CO2, and the conversion of redox species to valuable products [20]. The features of the photocatalyst, the energy of light, and the concentration of reactants are all factors that greatly influence the reaction rate of the photocatalytic process. Other important parameters include pH (when the reaction takes place in water), which affects the charge of the photocatalytic surface, and temperature, which impacts the collision frequency between the reactants and the photocatalyst [21]. Moreover, the surface of the photocatalyst must be thoroughly cleaned prior to any photocatalytic test. Otherwise, carbon-containing species, resulting from the synthesis procedure, will be present on the photocatalyst's surface and will contribute to product formation. Dilla et al. [41] developed an efficient photocatalytic cleaning step that may be performed right before the start of the CO2 photoreduction test. Herein, humid helium is flushed through the reactor under light irradiation. The cleaning progress is then monitored and, as soon as the concentration of products from the cleaning reaction is low, CO2 is fed into the reactor. This method not only ensures a clean photocatalytic surface but also rids the reactor from any trace hydrocarbon contaminants that might be present inside.

If the participating species are adsorbed on the semiconductor surface, then the photocatalytic process will certainly be more efficient due to the decrease in activation energy and increase in substrate concentration near reactive sites. When CO2 adsorbs onto the photocatalyst surface, its structure transforms from linear to the more reactive bent form (Figure 2). More specifically, the carbon atom of CO2 binds to a surface oxygen site, while the oxygen atom binds to the surface metal center of the metal oxide [42]. As a result of the structural transformation, the lowest unoccupied molecular orbital (LUMO) level of the metal oxide decreases lowering its activation energy [39]. However, density functional theory (DFT) calculations performed by Sorescu et al. [43] demonstrated that CO2 preferentially binds to the metal oxide surface in a linear geometry. This is most probably due to the low binding energy (−46 kJ/mol) associated with the linear configuration, the considerable energy

required to bend CO2 (38.6 kcal/mol) and the significant deformation the metal oxide surface undergoes when binding to bent CO2 [42,43].

**Figure 2.** Five possible CO2 adsorption configuration models on metal oxide semiconductors. M: Metal C: Carbon O: Oxygen. Reproduced from work in [44]. Copyright 2014 Wiley-VCH.

To enhance the adsorption of CO2, basic sites and/or micropores must be introduced on the surface of the photocatalyst. The introduction of sites that can store electrons and acid sites that can stabilize the species derived from CO2 must also be considered [37]. Therefore, researchers have to focus on developing such complex photocatalysts with a combination of acidic and basic sites. Additionally, the adsorption of CO2 could also be improved by synthesizing metal-oxide photocatalysts with oxygen vacancy sites. These vacancies may then be filled by the oxygen atoms of CO2 promoting the adsorption of the latter [45]. The generation of an unexpected attraction between the oxygen vacancy and CO2 subsequently lowers the reactive barrier [44].

A large and negative reduction potential (*E* = −1.90 V) is required for the single electron reaction that coverts CO2 to CO2 •− [46]:

$$\text{CO}\_2 + \text{e}^- \rightarrow \text{CO}\_2\text{}^{\bullet-},\tag{1}$$

This reaction is thermodynamically unfavorable due to the high stability of CO2 and the large amount of energy needed to change the geometrical configuration of CO2 from linear to bent. CO2 •− is a highly unstable radical anion that almost immediately converts to CO3 <sup>2</sup><sup>−</sup>, CO, and C2O4 <sup>2</sup><sup>−</sup> [46]. If the transfer of the single electron occurs simultaneously with a proton transfer, the CO2 reduction potential can be significantly lowered (i.e., more positive) due to the stabilization of CO2 •− by the proton [37]. The reduction potential can be further reduced to a more positive value when multiple electrons and holes are simultaneously transferred to CO2. Therefore, the multiple electron reduction of CO2 is considered more thermodynamically favorable. Here, CO2 typically converts to CO, CH2O2, CH2O, CH3OH, and CH4. Although more thermodynamically favorable, these multiple electron reactions are kinetically more challenging and lead to difficulties in process efficiency and selectivity.

As depicted in Figure 3, any photocatalytic reaction mechanism generally involves the following steps [47]; (1) reactant adsorption onto active sites, (2) light absorption from the catalyst with subsequent photogeneration of e<sup>−</sup> and h+, (3) interaction between charges and adsorbed reactants, (4) redox reactions, and (5) product desorption. Nonetheless, the particular reaction mechanism of CO2 photocatalysis is fairly complex. The reduction process, which is significantly influenced by the type of photocatalyst used, can give rise to a range of various possible products. The mechanism is also highly dependent on the energy of the photoexcited charges. If the excited charge carriers do not have the sufficient energy to react with the intermediate species, the reaction will not proceed and the final expected product from the photocatalytic reaction pathway will not be formed. Other factors that may inhibit the formation of products include product accumulation on the photocatalytic surface and photocatalyst deactivation. Moreover, the formed products might re-oxidize back to CO2 in oxygen-rich environments. As a result of the aforementioned challenges, the yield of CO2 photoreduction is usually very low and, as such, the formed products are generally difficult to detect [39].

**Figure 3.** Basic illustration describing the reaction mechanism of CO2 photoreduction. Reproduced from work in [47]. Copyright 2016 Elsevier B.V.

The kinetic modeling of CO2 photoreduction has only been investigated by few. In these studies, two insights have been suggested on the rate limiting step. The first proposition claims that the activation of CO2 through charge transfer is rate limiting [48]. In their mechanistic study, Uner et al. [49] suggested that the rate of the overall process of CO2 photoreduction was limited by the production of electrons and protons. The second proposition claims that the reactant adsorption/product desorption is rate limiting [48]. A kinetic model proposed by Saladin et al. [50] demonstrated that product desorption was rate limiting at low temperatures while reactant adsorption was rate limiting at high temperatures. Further studies need to be conducted to help advance the understanding of reaction kinetics in CO2 photoreduction.

Due to its availability and low cost, water is a desirable reducing agent to provide hydrogen atoms in the photocatalytic conversion of CO2. It reacts with holes (h<sup>+</sup>) to form O2 and H+. The H<sup>+</sup> ions later interact with the excited electrons (e<sup>−</sup>) producing H2 [51]:

$$2\text{H}\_2\text{O} + 4\text{h}^+ \rightarrow \text{O}\_2 + 4\text{H}^+ \qquad \qquad E = 0.81\text{ V}\_\prime\tag{2}$$

$$2\text{H}^+ + 2\text{e}^- \rightarrow \text{H}\_2 \qquad \qquad E = -0.42\text{ V} \tag{3}$$

Together, the CO2 and the H<sup>+</sup> react and lead to the formation of stable molecules as shown in the equations below [51].

$$\text{CO}\_2 + 2\text{H}^+ + 2\text{e}^- \rightarrow \text{CO} + \text{H}\_2\text{O} \qquad \qquad E = -0.53 \text{ V} \tag{4}$$

$$\text{CO}\_2 + 2\text{H}^+ + 2\text{e}^- \rightarrow \text{HCOOH} \qquad \qquad E = -0.61 \text{ V} \tag{5}$$

$$\text{CO}\_2 + 4\text{H}^+ + 4\text{e}^- \rightarrow \text{HCHO} + \text{H}\_2\text{O} \qquad \qquad E = -0.48 \text{ V} \tag{6}$$

$$\text{CO}\_2 + 6\text{H}^+ + 6\text{e}^- \rightarrow \text{CH}\_3\text{OH} + \text{H}\_2\text{O} \qquad \qquad E = -0.38 \text{ V} \tag{7}$$

$$\text{CO}\_2 + 8\text{H}^+ + 8\text{e}^- \rightarrow \text{CH}\_4 + 2\text{H}\_2\text{O} \qquad \qquad E = -0.24\text{ V} \tag{8}$$

Studies have shown that the presence of water in the reaction system favors the formation of methane as the major CO2 reduction product [37]. Two fundamental mechanisms have been proposed for this specific reaction. One mechanism suggests that methane is produced in series [52]:

$$\text{CH}\_2 \rightarrow \text{HCOOH} \rightarrow \text{HCHO} \rightarrow \text{CH}\_3\text{OH} \rightarrow \text{CH}\_4\tag{9}$$

The second mechanism assumes that methane forms in parallel with methanol [53]:

$$\text{CO}\_2 \rightarrow \text{CO} \rightarrow \text{CH}\_3\text{OH/CH}\_4\tag{10}$$

Evidently, the reaction pathway of CO2 photoconversion also greatly depends on the phase (liquid/vapor) of water, which has to be considered as an important reaction parameter [39]. The major product of CO2 photoreduction in aqueous media is usually formic acid (Figure 4), whereas the main product in gaseous phase is typically CO (Figure 5). It is important to note that although methanol and methane are the desired end products, the reaction usually does not proceed to that point. Instead, it typically stops after the formation of formic acid or CO.

**Figure 4.** Aqueous phase photocatalysis of CO2 and water: proposed reaction pathway. Reproduced from work in [39]. Copyright 2016 Elsevier Ltd.

**Figure 5.** Gaseous phase photocatalysis of CO2 and water: proposed reaction pathway. Reproduced from work in [39]. Copyright 2016 Elsevier Ltd.

Despite the aforementioned, the use of water has some limitations, including low CO2 solubility (relevant when H2O is used as the solvent), competing reduction reaction leading to hydrogen, and weak reducibility, eventually lowering the CO2 conversion efficiency [38]. The photoreduction of water to hydrogen requires lower energy (*E* = 0 V) than the photoreduction of CO2 (*E* = −1.90 V) and is, therefore, considered a strongly competing reduction process [39]. Furthermore, the photoreduction of CO2 involves a multi-step charge transfer mechanistic approach in contrast to the single electron transfer step sufficient to initiate the photoreduction of water to hydrogen [37]. Due to these thermodynamic and kinetic limitations, and depending on the reductant used to obtain hydrogen, the generation of hydrogen from water photoreduction might occur with efficiency much higher than that of CO2 photoreduction. Thus, when water is used as a reductant in the photocatalytic conversion of CO2, hydrogen can appear as the predominant product. In addition, the higher dipole moment of water (D = 1.85), when compared to that of CO2 (D = 0), favors its adsorption on the surface of the photocatalyst leading to another competing process [54]. The adsorption equilibrium constants of CO2 and water vapor were determined in a study performed by Tan et al. [47]. Results showed that water vapor had a considerably higher adsorption equilibrium constant (8.07 bar<sup>−</sup>1) than that of CO2 (0.019 bar<sup>−</sup>1). This implies that CO2 adsorbs weakly on the surface of the photocatalyst. Ideally, and as discussed earlier in this section, CO2 adsorbs to a photocatalyst surface in a bent configuration, whereby the carbon atom of CO2 binds to a surface oxygen site while the oxygen atom binds to the surface metal center. This, however, is only applicable in the absence of water. Molecular dynamics studies were employed by Klyukin et al. [55] to help investigate the surface adsorption of CO2 in the presence of water. Simulations showed that CO2 adsorbed at oxygen sites while water saturated the metal sites. To manage the competitive adsorption between CO2 and water, an optimum concentration of both reactants must be carefully considered. If CO2 was present at extremely high concentrations, then it would effectively compete with water for the reactive sites. On the other hand, at CO2 concentrations lower than that of water, only a limited number of CO2 molecules would adsorb on the photocatalyst surface [47]. The formation of water-soluble products that are difficult to recover is another challenge faced in presence of water [37].

To help overcome the drawbacks of water as a reducing agent, many researches are currently investigating the use of alternative reductants for the photocatalytic conversion of CO2. In this respect, the dielectric constant (κ) of the reducing agent significantly impacts the conversions and yields of the photocatalytic process. Solvents with varying dielectric constants (κ), including water (78.5), acetonitrile (37.5), 2-propanol (18.3), and dichloromethane (9.1), were tested as reducing agents in the photocatalytic reduction of CO2 [56]. As κ increased, the conversion to formate increased while that to carbon monoxide decreased. This may be explained by the stabilization of the CO2 •− radical by solvents with higher κ. Consequently, the more stable anion radical will interact more weakly with the photocatalytic surface.

As methanol is a strong reducing agent and has high CO2 solubility, Qin et al. [38] investigated its use as a reductant for the photoreduction of CO2 on a CuO-TiO2 composite catalyst. Results show that CO2 was reduced to produce methyl formate with a yield of 1602 μmol/g-cat/h. In addition, other studies on the photocatalytic reduction of CO2 in a solution of NaOH showed that the use of NaOH as a reductant enhances the photocatalytic efficiency [57,58]. In this respect, CO2 absorbs chemically in NaOH solutions due to its acidic properties leading to a concentration much higher than in DI water, and moreover the OH− present in solution may serve as strong hole scavenger, enhancing charge separation. The optimal concentration of NaOH was reported to be 0.2 M [57,58].

Studies have also shown that the conversion of CO2 using hydrogen as a reducing agent is higher than that when water is used [59]. This is because the photoreduction of CO2 with hydrogen is more thermodynamically favorable than that with water [39]. Lo et al. [60] studied the photocatalytic conversion of CO2 using bare TiO2 (Evonik P-25) under UV irradiation. Three different reductants were tested: water, hydrogen, and water + hydrogen. The results, which are presented in Figure 6, show that the highest product yields of CO2 photoreduction were obtained when hydrogen and saturated water vapor were used together as reductants, producing methane, carbon monoxide and ethane with a yield of 4.11, 0.14, and 0.10 μmol/g-cat/h, respectively. The authors propose that water accelerated the photoreduction of CO2 with hydrogen by donating electrons and subsequently inhibiting charge recombination. They also suggest that water supplied more hydrogen atoms for the photoreduction of CO2.

**Figure 6.** Effect of reductant on the yield of CO2 photoreduction products, i.e., CH4 (**a**) and CO (**b**). Reproduced from work in [60]. Copyright 2007 Elsevier B.V.

The photocatalytic reduction of compressed CO2 (either liquid or supercritical) might be considered as an alternative solution to the use of organic solvents, but it has been rarely investigated [61]. Kaneco et al. [62] performed a study on the photoreduction of liquefied CO2 using TiO2 as the catalyst and water as the reducing agent. Results show the highly selective formation of formic acid with no other reduction products detected. Kaneco et al. [63] also investigated the photocatalytic reduction of CO2 into formic acid in supercritical CO2. The photoreduction reaction was conducted at 9.0 MPa and 35 ◦C with TiO2 as the catalyst. Under these conditions, the production rate of formic acid reached a maximum yield of 1.76 μmol/g-cat/h. Kometani et al. [64] examined the photocatalytic reduction of CO2 in a supercritical mixture of water and carbon dioxide (400 ◦C and 30 MPa) using Pt–TiO2 as the catalyst. Carbon monoxide, methane, formic acid, and formaldehyde were all detected as reaction products with yields much higher than those obtained from the same reaction but at room temperature. Despite the enhanced product yields, the major drawback in using compressed CO2 is the large amount of energy required to change its state from gas to liquid or to supercritical. One possible suggestion to help mitigate this limitation could be to target industrial processes that already utilize CO2 in its compressed form. Thus, the overall energy efficiency of the photoreduction process could be greatly enhanced.
