*1.3. Catalysts for CO2 Conversion to Light Olefins*

As can be seen above, because each route has its own unique pathway of species and intermediates, different catalysts must be employed for the hydrogenation of CO2 to olefins depending on the chosen route. In the CO2–FTS path, Fe is one of the most widely used components in the catalysts, as catalysts containing Fe offer less methanation activity under higher reaction temperatures. As described above, it has been reported that Fe3O4 was the active phase responsible for RWGS; the metallic Fe and iron carbides could activate CO and produce hydrocarbons [31,32]. When incorporating alkali promoters, Fe-based catalysts showed greater olefin selectivity. The alkali metals, acting as electron donors to the Fe metal, facilitate the adsorption of CO2 while lowering the affinity for H2. The net result is a higher olefin yield [33–35]. There is also some indication that doping the catalyst with an additional metal may promote even higher olefin yields by forming a highly active interface. The second metal components allow for even greater adjustment of the CO2 and H2 adsorption and activation, shifting the distribution of the product more towards the desired hydrocarbons. By supporting the Fe-based catalysts on supports such as silica (SiO2), alumina (Al2O3), titania (TiO2), zirconia (ZrO2) and carbon materials (i.e., carbon nanotubes (CNTs), carbon nanospheres (CNSs), graphene oxide (GO)), the catalytic performance may be further enhanced by improving the active metal dispersion and slowing down the sintering of the active particles [36,37]. Controlling hydrocarbon chain growth to achieve a desired carbon range (i.e., C2–C4) remains a challenge for CO2 conversion due to the product selectivity limit governed by the Anderson–Schulz–Flory (ASF) distribution with a maximum achievable C2–C4 hydrocarbons selectivity of less than 60%, as shown in Figure 2 [30,38].

**Figure 2.** Product distribution as predicted by the Anderson–Schulz–Flory (ASF) model. Adapted with permission from ref. [30]. Copyright 2021 Elsevier.

Alternatively, in the MeOH path, light olefins can be synthesized with selectivity as high as 80–90% among hydrocarbons, exceeding the ASF product distribution limit for FTS reactions [18,39–41]. Some plausible reasons for the reported ASF distribution deviation are the blockage of surface polymerization by intermediates, (e.g., ketene (CH2CO)), space confinement, or the use of catalysts with two types of active sites (i.e., bifunctional catalysts) [27]. Regardless of the reasons, the observed deviation from the ASF distribution offers opportunities to increase the selectivity to olefins [27]. Several recent studies have reported the results for the combination of MeOH synthesis catalysts (i.e., In2O3, In-Zr, ZnGa2O4, MgGa2O4, ZnAl2O4, MgAl2O4, ZnZrO and In2O3-ZnZrO2) with an MTO catalyst (i.e., SAPO-34, SSZ-13 and ZSM-5), and their ability to produce light olefins with enhanced selectivity for CO2 hydrogenation [13,42–44]. It has been proposed that the secondary functionality of acid–base sites on the catalytic support significantly impacts the light olefin selectivity. For example, by passivating the Brønsted acid sites of In2O3- ZnZrOx/SAPO-34, the secondary hydrogenation reaction is inhibited, thereby improving the olefin selectivity [27,30].

#### *1.4. The Main Focus of This Review*

Even though significant efforts have been made, considerable challenges remain in the development of highly efficient catalysts with selective pathways to light olefins due to the thermodynamically stable nature of the CO2 molecule, the complexity of the reaction networks, and catalyst deactivation [8,45,46]. Several recent reviews have summarized CO2 hydrogenation to value-added products, including light olefins [25,27,30,38]. However, it is necessary to present a review focused on the recent advances in the mitigation of the catalyst deactivation of CO2 hydrogenation to light olefins, as catalyst deactivation has been a big challenge that provides economic hurdles to the adoption of the new technologies.

Because catalysts and mechanisms have been extensively reviewed in numerous review papers [25,27,30,38], the focuses of the current article are to identify possible causes that trigger catalyst deactivation and summarize recent advances on catalyst development with enhanced catalyst stability and light olefin selectivity for CO2 hydrogenation. In this review, we first provide a brief summary of the two dominant reaction pathways (CO2–FTS and MeOH-mediated), mechanistic insights and catalytic materials for CO2 hydrogenation to light olefins. We then list the deactivation mechanism caused by carbon deposition, water formation, phase transformation and metal sintering/agglomeration. Finally, we summarize the recent progress published within five years on catalyst development that improves catalyst deactivation by the following catalyst functionalities: (1) the promoter effect, (2) the support effect, (3) the hybrid functional effect, and (4) the structure effect.

Each one of these aspects is accompanied by a suitable table in which the most significant literature findings are comparatively presented. To the best of our knowledge, no review has ever directly correlated the causes of catalyst deactivation and catalyst mitigation for CO2 hydrogenation to light olefins. Herein, we attempt to provide a useful resource for researchers to correlate the catalyst deactivation and the recent research effort on catalyst development for enhanced olefin yield and catalyst stability.

#### **2. Causes of Catalyst Deactivation**

During CO2 hydrogenation, catalyst deactivation can occur via several mechanisms, resulting in decreased activity and selectivity toward the desired olefins. The determination of the mechanism of deactivation is an important step toward mitigation. The primary causes of catalyst deactivation are the sintering (or agglomeration) of metal particles, phase transformation at the catalyst's surface, and catalyst poisoning by water or carbonaceous deposits (i.e., coke). An understanding of the deactivation causes is necessary to develop a mitigation strategy and sustain high selectivity toward the desired olefins during CO2 hydrogenation. For context, we present brief descriptions of each of these causes with a few representative examples from the literature that demonstrate the necessity of robust

and novel mitigation studies. More thorough reviews of the deactivation causes and their mechanisms can be found elsewhere [47–49].

#### *2.1. Sintering*

Catalyst sintering can occur through either Ostwald ripening or particle migration and coalescence, as shown in Figure 3 [50]. Through sintering, the agglomeration of smaller catalyst crystals into larger ones will bring about the loss of the pore structure, which lowers the internal surface area of the catalyst, leading to the deactivation. In the area of FT by cobalt catalysts, several groups have determined that the particle growth of cobalt is the largest factor causing deactivation [51,52].

**Figure 3.** Diagram of active phase sintering occurring over a support material: the blue ring represents atomic migration to form larger crystallites; the red ring represents the coalescence of crystallites. Adapted with permission from ref. [50]. Copyright 2021 Elsevier.

Sun et al. [53] examined sintering in zinc- and alumina-supported copper catalysts (Cu/ZnO/Al2O3). It was found that the presence of CO in the process employed for CH3OH synthesis strongly contributed to the deactivation of the catalysis over 0 to 50 h. Taken with corroborative evidence from the Cu surface area determination, the deactivation was likely attributed to the sintering of the Cu metal.

As mentioned above, sintering negatively affects the catalytic performance due to many reasons: for example, the overall catalytically active surface area is reduced due to the collapse of the structure and the chemical alteration of the catalytically active phases to non-active phases [50,54,55]. As this form of deactivation involves the coalescence of larger particles from smaller, it is extremely difficult to reverse. Sintering, therefore, is easier to prevent through careful catalyst design [50,56]. For example, Li et al. observed remarkable metal sintering on supported FeCo/ZrO2 catalysts [56]. As shown in Figure 4A(a), for the 13Fe2Co/ZrO2 supported catalyst precursor prepared using the conventional impregnation method, the Co and Fe are distributed into separate oxide particles, which increased the possibility of sintering. As confirmed in Figure 4A(b,c), aggregates composed of Fe and Co oxide nanoparticles were observed on the ZrO2 fibers, with an average diameter of ca. 15 nm before the reaction. The particle size increased to 48 nm after the reaction, which was responsible for the rapid deactivation of activity (Figure 4A(d–f)). By comparison, Fe-Co-Zr polymetallic fibers obtained via a one-step electrospinning technique showed that Fe and Co were dispersed in proximity to ZrO2, as shown in Figure 4C(a), but separately from each other. In order to reduce the possibility of sintering, as demonstrated in Figure 4B(a–f), the Fe and Co oxides nanoparticles successfully dispersed with the ZrO2 particles for the polymetallic oxide fibers, with an average size of roughly 1–2 nm before the reaction, and after the reaction, the particle size barely changed, which contributed to the stable catalytic activity after 500 mins on stream (Figure 4C(a,b)).

**Figure 4.** (**A**) (**a**–**f**) Schematic illustration of the metal distribution and TEM images of the 13Fe2Co/ZrO2-supported catalyst precursor (**a**–**c**) and the spent catalyst (**d**–**f**). (**B**) (**a**–**f**) Schematic illustration of the metal distribution and TEM images of the 13Fe2Co100Zr polymetallic oxide fiber catalyst (**a**–**c**) and the spent catalyst (**d**–**f**). (**C**) CO2 conversion (**a**) and the C2+/C2 <sup>=</sup> –C4 <sup>=</sup> selectivity and C2 <sup>=</sup> –C4 <sup>=</sup> yield (**b**) over different catalysts after 8 h TOS (testing conditions: H2/CO2 molar ratio = 3/1, GHSV = 7200 mL g−<sup>1</sup> h−1, P = 3 MPa, T = 673 K). Adapted with permission from ref. [56]. Copyright 2019 Elsevier.
