**1. Introduction**

The global emission of greenhouse gases, such as CO2, continues to rise by ~3% each year [1]. Higher atmospheric concentrations of greenhouse gases lead to surface warming of the land and oceans [2]. According to computational results based on climate models, the average global value of the temperature rise of +2 ◦C will be surpassed when the concentration of CO2 reaches 550 ppm [1]. If current trends are kept, the CO2 concentration will reach this threshold value by 2050. In addition to warming the Earth, CO2 emissions have increased the ocean's acidity. Thirty million of the 90 million tons of CO2 discharged each day end up in the oceans as carbonic acid, lowering the ocean's pH level [3]. Both the increased temperatures and the higher acidity of the ocean have initiated a set of other impacts including the melting of glaciers, rising sea levels, deeper and longer droughts, more and larger forest fires, migration of tropical diseases, accelerated extinction rates, increased destructive power of tropical storms, and increasingly large downpours of rain and snow [3,4]. Natural processes can potentially remove most of the CO2 that human activities are adding to the atmosphere; however, these processes operate very slowly and will take too long to prevent rapid climate change and its impacts [2].

Three theoretical approaches have been widely suggested for solving the climate problem: (1) direct reduction of CO2 emissions by changing industrial and urban processes and habits, (2) CO2 capture and storage (CCS), and (3) CO2 capture and utilization (CCU). Due to the increasing population

rate and the increasing demand for high-quality life, the direct reduction of CO2 emissions seems infeasible [5]. To keep up with energy needs of a growing population, governments are compelled to continue with their current industrial activities despite the large amounts of resulting greenhouse gas emissions. Carbon capture and sequestration (CCS) is considered to be a very promising solution to removing excess CO2 from the atmosphere. The idea behind CCS lies in storing large quantities of captured CO2 in underground geological formations. The stored CO2 may be used for recovering oil and gas from partly exploited fields, thus giving an economic value to CO2 [6]. The two major issues associated with CCS are cost and storage. The high cost of CCS is due to the large amounts of energy required to separate CO2 from the emission stream. It is estimated that this separation process could account for 70–90% of the total operating cost of CCS [7]. Also, the storage of CO2 is considered challenging due to the large amounts of CO2 that needs to be stored and the risk of leakage [2]. Although CCS technologies may appear to be very promising in removing excess CO2 from the atmosphere, they are expensive, energy-intensive, and require large capital investment for industrial application [8].

To make CCS technologies more economically feasible, carbon capture and utilization (CCU) technologies have emerged as a feasible and promising technique that can complement the storage of huge quantities of CO2 in geological and ocean formations. Industrial applications of CO2 are present in numerous sectors including chemical, oil and gas, energy, pulp and paper, steel, food, and pharmaceuticals [9]. Currently, the commercial CCU technologies use CO2 for enhanced oil recovery (EOR) applications. The use of CO2 as a raw input material in the chemical industry is limited to a few processes such as the production of salicylic acid, urea and polycarbonates [8,10]. Recently, new research is looking into converting the captured CO2 into valuable products including chemicals, polymers and fuel. It is estimated that 5 to 10% of the total CO2 emissions may be utilized for the synthesis of value-added products [11]. The specific application of converting CO2 into fuel is being actively studied by researches and is showing great promise for future industrial applications. A wide variety of fuels, including methanol, ethanol, methane, dimethyl ether, formic acid, petroleum-equivalent fuels, and others may be produced through different CO2 conversion processes. However, these conversion processes produce large quantities of CO2 and will therefore increase the concentration of CO2 in the atmosphere instead of reducing it [11,12]. Thus, to stay within the targeted goal of decreasing CO2 emissions, low-carbon energy sources, such as renewable resources, must be used as the primary energy input in the CO2-to-fuels conversion process. Other drawbacks of CO2-to-fuels conversion processes include intensive energy requirements and low energy conversion efficiency [12]. The main methods of CO2 conversion are thermochemical [13–15], electrochemical [12,13], biological [16–18], and photocatalytic [19–23].

Photocatalysis is considered to be a promising method for the conversion of CO2 into valuable products, such as methane, hydrogen, methanol, formaldehyde, ethanol, and higher hydrocarbons [6,19]. One major advantage of photocatalysis over other conversion methods is that it can take place at room temperature and under atmospheric pressure conditions [19]. In addition to that, it utilizes a renewable and sustainable form of energy, namely solar energy, for the conversion of CO2. Note that, unlike conventional processes, the photoreduction of CO2 does not increase net CO2 emissions or consume additional energy [23]. Despite the intensive research, the photocatalytic reduction of CO2 is still considered inefficient. This is mainly due to the absence of scalable reactor designs able to simultaneously introduce reactants, photons, and visible light-responsive catalysts to produce specific fuels in significant quantities [23]. The main challenge in developing an economically feasible process for the photocatalytic conversion of CO2 is finding a suitable catalyst [24]. The most common type of catalysts used in the photocatalytic conversion of CO2 are the inexpensive and naturally abundant transition-metal oxides, such as titanium dioxide (TiO2). TiO2 is one of the most widely used and commonly investigated semiconductors for photocatalytic applications [25]. This is attributed to its low toxicity, low cost, high efficiency, and high stability. Other several types of metal oxide semiconductors, including zirconium oxide, gallium oxide and thallium oxide, have also been investigated for their use

in the photocatalytic reduction of CO2 [25]. The poor light absorption and the low product selectivity of these photocatalysts pose a challenge for researchers aiming to efficiently reduce CO2 [26]. Currently, the development of new, stable, inexpensive, abundant, nontoxic, selective, and visible light-responsive photocatalysts is being actively investigated [19,27]. In this regard, some state-of-the-art photocatalysts, such are perovskite oxides and III-V semiconductors, have shown some great promise in driving the photocatalytic reduction of CO2 specifically under direct sunlight. Perovskite oxides exhibit interesting compositional flexibility that allows for precise band gap tuning and defect engineering [28]. On the other hand, III-V semiconductors can easily meet the thermodynamic requirements of CO2 reduction to CO due to their higher (less positive) conduction band when compared to that of other metal oxide semiconductors [29]. Nonetheless, the current photocatalytic efficiency of perovskite oxides and III-V semiconductors is too low for practical CO2 reduction applications. Furthermore, their low selectivity and high instability have urged researches to continue with their investigations on TiO2-based photocatalysts for the reduction of CO2 [28].

Besides enhanced photocatalysts, researchers have been also coupling photocatalysis with other CO2 conversion methods aiming at improving the process efficiency. For instance, in a study proposed by Zhang et al. [30], a photo-thermochemical process was used for the reduction of CO2. Combining both processes together allowed for the utilization of both solar and thermal energy. The major limitation of thermochemical conversion techniques is the high temperature (>1273 K) step required to reduce the metal oxide catalyst and create an oxide vacancy. In the photo-thermochemical process, this step is replaced by photocatalysis where UV light is used instead of high temperatures to form vacant sites. Next, the temperature needs to be only slightly raised (573–873 K) to accelerate the dissociation of CO2 to CO. Electrocatalysis is another CO2 conversion method that has been coupled with photocatalysis. As previously mentioned in this work, photocatalysts with outstanding structural and optical properties need to be developed for an efficient CO2 photoreduction process. From a thermodynamic perspective, these photocatalysts must possess favorable band gap energies for CO2 reduction. In photo-electrocatalysis, this limitation may be overcome if an appropriate external bias is applied to the system [31]. Therefore, a broader range of photocatalysts with lower thermodynamic restrictions can be used for the reduction of CO2. Halmann [32] was the first to report the successful photo-electrocatalytic conversion of CO2 into formic acid, formaldehyde, and methanol.

In this review, the current advances on CO2 photoreduction over TiO2-based catalysts are critically discussed. The photocatalytic mechanism of CO2 reduction has been explained particularly in the case when water is used as a reductant. Furthermore, the various modification techniques of TiO2, including metal deposition, metal/non-metal doping, carbon-based material loading, formation of semiconductor heterostructures, and dispersion on high surface area supports, have been summarized. Although a number of review articles [33–35] have highlighted the possible surface modifications of TiO2 photocatalysts, this paper provides emphasis on modification techniques that will specifically enhance the CO2 photoreduction efficiency of TiO2 photocatalysts. Future directions toward efficient photocatalytic systems for the reduction of CO2 have been also presented.
