Challenges and Prospects in the Catalytic Conversion of Carbon Dioxide to Formaldehyde

2.1 Hydrogenation Using H₂

Hydrogenation of CO₂ to value-added products is the most atom-efficient route for CO₂ transformation. So far, great progress has been made to generate various C₁ and C₂₊ chemicals, including CO,^{23, 24} methane (CH₄),^{25, 26} CH₃OH,^27-29 formic acid (HCOOH),³⁰ and hydrocarbons.^31-33 Both heterogeneous and homogeneous catalysts were employed for selective conversion of CO₂ to HCHO or derivatives thereof. Hydrogenation of CO₂ to HCHO [CO₂ (g)+H₂ (g)→HCHO (g)+H₂O (g); Δ =59.8 kJ mol⁻¹, Δ =39.8 kJ mol⁻¹] is a strongly endergonic/endothermic reaction,³⁴ which is thus thermodynamically unfavored under ambient conditions. In addition, catalytic transformation of CO₂ with halt at the HCHO level remains challenging, as the resulting aldehyde group is susceptible to further reduction.

In 2001, Lee and co-workers reported Pt/Cu/SiO₂ as heterogeneous catalyst for the selective hydrogenation of CO₂ to HCHO at 150 °C and 6 bar.³⁵ The Pt/Cu-based part, with an optimum atomic Pt/Cu ratio of 0.03 : 1, was found to be crucial for HCHO formation, accompanied by CH₃OH as main by-product. Mechanistically, H₂ was suggested to be adsorbed on Pt before migrating onto Cu to promote reduction of CO₂. Furthermore, the relative rate of HCHO formation increased with the H₂/CO₂ ratio, showing the strong effect exerted by the surface hydrogen concentration on HCHO selectivity. At an optimum H₂/CO₂ of 20 : 1, the yield of HCHO reached 10.4 mmol g_cat⁻¹ over 2 h reaction where g_cat⁻¹ denotes the weight unit of the catalyst. Tanksale and co-workers first demonstrated production of HCHO via hydrogenation of CO₂ using Pt/Ni/Al₂O₃ as catalyst in CH₃OH as liquid medium at 25 °C.³⁶ Increasing the H₂/CO₂ ratio from 2 to 16 led to 50 % increase in the equilibrium HCHO yield, i.e., from 0.06 to 0.09 mmol g_cat⁻¹ over 22 h reaction. Despite the limited availability of heterogeneous catalysts for efficient CO₂-to-HCHO conversion, a few have been reported in the hydrogenation of CO₂ to CH₃OH.³⁷ In addition, insight into the underlying mechanism can help to guide the preparation of novel heterogeneous catalysts for hydrogenation of CO₂ to HCHO, as surface-bound HCHO derivatives have been demonstrated to be the key intermediate along the CO₂-to-CH₃OH formation pathway.^{38, 39}

A disadvantage of direct hydrogenation is associated with the need of applying high pressure of H₂ to enhance its solubility. In various hydrogenations, liquid-phase alcohols have been used as hydrogen donors, due to higher safety and more convenient management and usage. Recently, Liu and co-workers found that transfer hydrogenation of CO₂ into HCHO from aqueous glycerol could be efficiently induced by highly dispersed Ru on layered double hydroxide (LDH) under base-free conditions at low temperature.⁴⁰ The yield of HCHO could be as high as 3.6 mmol g_cat⁻¹ with a turnover number (TON) of 72 over 12 h at 50 °C and 10 bar CO₂. The metal–support interaction over Ru/LDH proved to afford high turnover of hydrogen transfer between C₃ species (glycerol and lactic acid) and CO₂ to generate HCHO. In addition to the direct synthesis of HCHO, conversion of CO₂ to dimethoxymethane (DMM), the dimethyl acetal of HCHO, can be realized by heterogeneous catalysts. Recently, Wang, Tanksale, and co-workers synthesized 3 wt % Ru over novel hierarchical zeolite beta to provide an optimized DMM yield of 0.66 mmol g_cat⁻¹ after 20 min of reaction.⁴¹ Moreover, the activity could be retained even after five recycle steps, demonstrating the reusability of this catalyst.

Homogeneous catalysts in direct hydrogenation of CO₂ to HCHO have been less successful than heterogeneous ones. Stephan, Fontaine, and co-workers demonstrated that the frustrated Lewis pairs (FLP) 1-BR₂-2-NMe₂C₆H₄ (R=2,4,6-Me₃C₆H₂ or 2,4,5-Me₃C₆H₂) are active for CO₂ hydrogenation in the presence of CO₂ and H₂, generating the corresponding formate, acetal, and methoxide species.⁴² However, the H₂ activation and CO₂ reduction steps are stoichiometric and not catalytic in nature.

Several works have studied the mechanistic pathways for CO₂ hydrogenation to CH₃OH from a theoretical point of view which may provide insights into the reduction of CO₂ to the HCHO level, noting once more that HCHO is a key intermediate along the CH₃OH production pathway. Klankermayer, Leitner, and co-workers conducted detailed mechanistic studies on the hydrogenation of CO₂ to CH₃OH using a Ru–triphos (triphos=1,1,1-tris(diphenylphosphinomethyl)ethane) complex.⁴³ The results supported by DFT calculations reveal that a sequential series of hydride transfer and protonolysis steps account for the transformation of CO₂ via HCOO⁻/HCOOH to hydroxymethanolate/HCHO and finally methanolate/CH₃OH within the coordination sphere of the complex. Based on DFT calculations, Yang and co-workers predicted a hydride transfer mechanism for the conversion of CO₂ and H₂ to CH₃OH catalyzed by a half-sandwich Co complex, [Cp*Co(bpy-CH₃)OH₂]²⁺.⁴⁴ Three cascade cycles were proposed: hydrogenation of CO₂ to HCOOH, hydrogenation of HCOOH to methanediol which decomposes to HCHO and water, and hydrogenation of HCHO to CH₃OH. A similar mechanism was proposed by Lei and co-workers for CO₂ hydrogenation catalyzed by a Mn pincer complex, Mn(Ph₂PCH₂SiMe₂)₂N(CO)₂, or a Ru pincer complex, RuH₂(Me₂PCH₂SiMe₂)₂NH(CO).^{45, 46} Here, H₂ activation is the rate-controlling step in each catalytic cycle under solvent-free conditions. In comparison, the reaction is more favorable in the appropriate solvent (water, toluene, THF) than in the gas phase. These works provide guidelines for how to design new homogeneous catalysts exhibiting higher efficiency for hydrogenation reactions.

Most of the currently available homogeneous catalysts for CO₂ hydrogenation have been developed with focus on obtaining DMM as main product instead of HCHO. This is done by performing the hydrogenation reaction in presence of CH₃OH, to make the overall process [CO₂ (g)+H₂ (g)+2 MeOH (l)→MeOCH₂OMe (l)+H₂O (l); Δ =12.5 kJ mol⁻¹, Δ =−73.4 kJ mol⁻¹) exothermic, with an only slightly positive value of the Gibbs energy. The equilibrium towards product becomes even more favored at pressures of 80 bar (Δ =−0.3 kJ mol⁻¹).³⁴ In 2016, Klankermayer and co-workers reported the synthesis of DMM (Figure 2a) using Ru complex catalysts with triphos ligands (Figure 2b) in combination with acidic co-catalyst Al(OTf)₃.⁴⁷ DMM was obtained with TON up to 214 (over 18 h) at moderate temperature (80 °C) and pressure (60 bar H₂/20 bar CO₂). In addition, methylformate was isolated in small amounts. Importantly, DMM can be used as a HCHO synthon in syntheses and if desired, DMM may be hydrolyzed to HCHO and CH₃OH. The acidic co-catalyst was found to be pivotal in enabling fast acetal formation in the hydrogenation of methylformate to DMM. Formation of the relatively stable acetal prevented over-hydrogenation to CH₃OH or CH₄. Similarly, other acetals could be synthesized when using other alcohols, such as ethanol and butanol.

Later, Klankermayer and co-workers developed a tailored non-precious transition-metal catalytic system based on Co salts in combination with selected triphos ligands and acidic co-catalyst (HNTf₂) (Figure 2c).⁴⁸ For example, a TON of 157 (over 22 h) for the formation of DMM, comparable to that of the precious-metal system, was obtained at slightly higher temperature (100 °C), but at the same pressure (60 bar H₂/20 bar CO₂), and with a modified triphos^Tol ligand (Tol=toluene). Inspired by this, Trapp and co-workers developed a highly efficient catalytic system using a Ru catalyst with a N-triphos^Ph (Ph=phenyl) ligand (Figure 2d), raising the TON for DMM to 786 (18 h) with a remarkable selectivity of >90 % at moderate temperature (90 °C) and partial pressure (90 bar H₂/20 bar CO₂).⁴⁹ By substituting the apical carbon atom in the backbone of the triphos ligand platform with silicon or phosphorus (Figure 2e–g), they achieved a TON as high as 685 (18 h) for DMM at optimized conditions, being in the range of the benchmark N-triphos^Ph system.⁵⁰

Although much progress has already been achieved in CO₂ hydrogenation, the transformation of CO₂ to the HCHO level still faces great challenges. Numerous mechanistic studies have inferred the intermediacy of HCHO during CH₃OH generation in both catalytic homogeneous and heterogeneous processes. Nevertheless, the thermodynamically favored over-reduction of HCHO and its derivatives to the CH₃OH level makes it challenging to obtain HCHO via a direct reduction route. To halt the reduction process at this point would require careful manipulation of the reaction conditions, i.e., the chemical environment around the active species, temperature, solvent, reducing agent, etc. Another important aspect is that H₂, as one of the main reactants, is predominantly produced using fossil fuels today.⁵¹ This results in the so-called CO₂-paradoxon in which the CO₂ produced for generating H₂ exceeds the amount of CO₂ consumed in the hydrogenation. Thus, it becomes crucial to sustainably generate carbon-free H₂ on a large scale.

2.2 Hydroboration/Hydrosilylation

Looking at the overall energetics of hydroboration or hydrosilylation of CO₂, the first important point to note is that hydroborane and hydrosilane reductants feature relatively weak and polarized B−H and Si−H bonds compared with the H−H bond. Second, the CO₂ reduction results in formation of strong Si−O and B−O bonds.⁵² As a result, hydroboration and hydrosilylation of CO₂ require significantly milder operating conditions (temperature <100 °C and CO₂ pressure <3 bar) than use of H₂ while enabling the synthesis of the complete series of C₁ products (CO, HCHO, CH₃OH, CH₄). The exact distribution of these products depends on the nature of the catalyst introduced for regenerating the hydroborane/hydrosilane reductants.

Taking the commonly used metal-based catalysts as an example, CO₂ hydroboration usually involves three sequential catalytic cycles (Figure 3). In Cycle I, CO₂ is reduced to the HCOOH level, in the form of R₂B−O(O)CH. This involves initial CO₂ insertion into the metal hydride (M−H), followed by transmetalation of M−OOCH with hydroborane (HBR₂) to regenerate M−H and release R₂B−O(O)CH. In Cycle II, further reduction leads to the HCHO level. Specifically, M−H reacts with R₂B−O(O)CH, resulting in M−OCH₂OBR₂, which is decomposed to M−OBR₂ and HCHO. Meanwhile, M−OCH₂OBR₂ can transmetalate with HBR₂ to generate bis(boryl)acetal (R₂BOCH₂OBR₂) and M−H feeding once again into Cycle I. Cycle III is the CH₃OH level with formation of CH₃OBR₂, starting with insertion of HCHO into M−H to generate M−OCH₃ that reacts with HBR₂.

Hydrosilylation using metal-based catalysts follows an equivalent pathway to that of hydroboration, although a fourth cycle now extends the product portfolio to CH₄. Usually, the main products of CO₂ hydroboration or hydrosilylation are formoxy (Cycle I; single reduction) or methoxy (Cycle III; triple reduction). Selective reduction of CO₂ to the HCHO level (Cycle II; double reduction) is more challenging. Only by carefully manipulating reaction conditions (temperature, CO₂ pressure, concentration, etc.) is it possible to halt the cascade of reduction processes at the HCHO level.

Sabo-Etienne and Bontemps pioneered the catalytic reduction of CO₂ using pinacolborane (HBpin) as reductant and RuH₂(η-H₂)₂(PCy₃)₂ (Cy=cyclohexyl) as catalyst precursor to obtain HCHO in the form of bis(boryl)acetal, PinBOCH₂OBPin, with a low yield of 5 % and low TON of 0.5 over 30 min of reaction at room temperature (Figure 4a1).⁵³ CH₃OH was further used to capture HCHO from the reaction by forming CH₃OCH₂OH.⁵⁴ Later, they disclosed formation of free HCHO (yield=22 %, TON=2.2), along with bis(boryl)acetal (yield=6 %, TON=0.6) over 30 min of reaction, using the analogous complex [RuH₂(H₂)₂(PCyp₃)₂] (Cyp=cyclopentyl) as catalyst precursor (Figure 4a2).⁵⁵ The better performance of the cyclopentyl-based catalyst was attributed to differences in cycloalkyl ring conformations and solubilities. Selective trapping of HCHO was realized by in situ condensation with 2,6-bis(diisopropyl)aniline to afford the corresponding imine. Both the yield of the corresponding imine and TON (74 % and 10.6, respectively) over 1 h of reaction were greatly improved by the presence of the trapping agent. In addition, subsequent hydrolysis of the imine with CH₃OH regenerated aniline and yielded a formalin solution.

In addition to the Ru-based catalysts, they realized Fe-catalyzed reduction of CO₂ into bis(boryl)acetal using Fe(H)₂(dmpe)₂ (dmpe=Me₂PCH₂CH₂PMe₂) (Figure 4a3).⁵⁶ Three different hydroboranes were considered, i.e., HBpin, catecholborane (HBCat), and 9-borabicyclo[3.3.1]-nonane (9-BBN). With HBCat, only methoxyborane was obtained while with HBpin and 9-BBN, the corresponding bis(boryl)acetal was favored over methoxyborane. Optimization of the synthesis using 9-BBN as reduction agent and tetrahydrofuran as solvent provided an excellent yield of 85 % with a TON of 85 for bis(boryl)acetal over 47 min. Inspired by this work, Turculet and co-workers reported a new bis(indolylphosphino)silyl (PSiP) ligand (Figure 4a4), the Ni hydride complex of which exhibited an unprecedented yield of 97 % for hydroboration of CO₂ to the targeted bis(boryl)acetal, with a TON of 487 after 4 h of reaction under 1 atm CO₂ at room temperature.⁵⁷ Hazari and co-workers performed a systematic study on how to control the reduction level of CO₂ hydroboration using the Ni−PSiP complex (Figure 4a4) and (^tBuPCP)−NiH [^tBuPCP=2,6-C₆H₃(CH₂PtBu₂)₂] (Figure 4a5).⁵⁸ They demonstrated that reduction of CO₂ past the HCOOH level could be accomplished by decreasing the CO₂ concentration, increasing the hydroborane concentration, or adding a Lewis acid cocatalyst. With this, borylformate insertion into M−H (Cycle II) became favored over the corresponding CO₂ insertion (Cycle I). In addition, low catalyst loading resulted in precipitation of bis(boryl)acetal, providing a driving force for its formation. Xu and co-workers found that a molecular Zn^II dihydride complex served as active catalyst for CO₂ hydroboration (Figure 4a6), selectively generating bis(boryl)acetal with a yield of 75 % and a TON of 15 after 4 h reaction using 9-BBN as hydroboration reagent.⁵⁹

Leitner and co-workers reported the Mn pincer complex, Mn(Ph₂PCH₂SiMe₂)₂NH(CO)₂Br, that enables reduction of CO₂ in the presence of hydroboranes.⁶⁰ Lei, Cao, and co-workers investigated the reaction mechanism of CO₂ hydroboration catalyzed by the Mn pincer complex.⁶¹ The calculated results showed that the carbonyl association mechanism is more favorable, with a low energetic span of 113 kJ mol⁻¹, than the carbonyl dissociation mechanism. The intrinsic difference between these two pathways is whether the second CO ligand coordinates with the metal center or not. The carbonyl dissociation mechanism mainly includes three catalytic cycles: hydroboration of CO₂ to yield the intermediate product HCOOBpin, hydroboration of HCOOBpin to yield HCHO and O(Bpin)₂, and hydroboration of HCHO to yield methyl boronate (CH₃OBpin). Regarding the same Mn pincer complex, Li, Schaefer, and co-workers reported a novel ionic mechanism for CO₂ hydroboration in the presence of sodium tert-butoxide (NaO^tBu).⁶² The ionic mechanism also contains three tandem stages. Differently, the H−Mn−N−Bpin pincer species generated by the reaction of the Mn pincer complex and HBpin in the traditional mechanism is not viable under basic conditions due to abstraction of the Bpin group by NaO^tBu, while the generated H−Mn−N−Na pincer species leads to a more favorable ionic mechanism. In comparison, the energetic span of the ionic mechanism is calculated to be 94 kJ mol⁻¹ lower than that of the traditional mechanism. Altogether, these experimental and computational works suggest that the Mn pincer complex is a promising catalyst for CO₂ hydroboration to the HCHO level.

With regard to hydrosilylation, Piers, Maron, Eisenstein, and co-workers reported selective reduction of CO₂ to the HCHO level using the [Cp₂Sc][HB(C₆F₅)₃] ion pair (Figure 4b1).⁶³ Berke and co-workers demonstrated a Re−H/B(C₆F₅)₃ Lewis pair catalyst (Figure 4b2), which was able to reduce CO₂ to (Et₃SiO)₂CH₂ within 4 h with a yield of 35 % and TON of 35.⁶⁴ Later, Piers and co-workers designed a novel organoscandium-based hydrosilylation catalyst (Figure 4b3) which, when combined with B(C₆F₅)₃, was capable of effectively hydrosilylating CO₂ almost exclusively to the bis(silyl)acetal, (Et₃SiO)₂CH₂, with an isolated yield of 94 % and TON of ∼1000 for 96 h in the presence of Et₃SiH.⁶⁵ Oestreich and co-workers reported Ru-catalyzed hydrosilylation of CO₂ yielding silylated HCHO with a high yield of 99 % and TON of 25 over 4 h using Et₃SiH as hydrosilane at relatively low temperature (80 °C) (Figure 4b4).⁶⁶ After additional 7 days at 150 °C, 77 % of the formed silylated HCHO were converted to silylated CH₃OH, indicating that the oxidation level of the product (silylated HCHO or CH₃OH) could be readily controlled by temperature.

López-Serrano, Rodríguez, and co-workers found that a bis(phosphino)boryl Ni−H complex (Figure 4b5), in combination with B(C₆F₅)₃, selectively hydrosilylated CO₂ to the bis(silyl)acetal, (Et₃SiO)₂CH₂, with a yield of 60 % and TON of 1200 (21.5 h reaction time) using Et₃SiH as hydrosilylation agent.⁶⁷ Mechanistic studies demonstrated that the metal complex is instrumental in the reduction steps during catalysis via FLP-like bond activation steps involving silylium transfer from [R₃Si−H⋅⋅⋅B(C₆F₅)₃]. This is followed by hydride transfer from [HB(C₆F₅)₃]⁻ at contact ion pairs such as [(PBP)Ni][HB(C₆F₅)₃], which is the species responsible for CO₂ activation. In addition, the metal complex effectively sequesters B(C₆F₅)₃, thus limiting the amount of [R₃Si−H⋅⋅⋅B(C₆F₅)₃] available in the reaction medium, and preventing the latter species from catalyzing the over-reduction of bis(silyl)acetal (Figure 5).⁶⁸ Parkins and co-workers reported that terminal Zn and Mg hydride compounds (Figure 4b6) were capable of hydrosilylation of CO₂ by hydrosilanes at room temperature, when reacting with B(C₆F₅)₃ to afford the ion pairs.⁶⁹ In the presence of Ph₃SiH, the Zn complex could reduce CO₂ to (Ph₃SiO)₂CH₂ with a TON of 12 over 12 h, while the Mg complex selectively afforded (Ph₃SiO)₂CH₂ with a TON of 83 over 70 h. In general, hydrosilylating CO₂ to the bis(silyl)acetal level was favored using a sterically hindered hydrosilane, e.g., Ph₃SiH, while PhSiH₃ would form CH₄. A N,P-heterocyclic germylene/B(C₆F₅)₃ Lewis adduct (Figure 4b7), reported by Baceiredo, Kato, and co-workers, was able to hydrosilylate CO₂ exclusively to (Et₃SiO)₂CH₂ with a TON of 19.9 (25 h reaction time) at 60 °C.⁷⁰ In 2019, Parkin and co-workers demonstrated that (Ph₃SiO)₂CH₂ could be generated using a Mg complex (Figure 4b8) from CO₂ hydrosilylation with an isolated yield of 95 % and TON of 88.6. In addition, HCHO was released from (Ph₃SiO)₂CH₂ upon treatment with CsF at room temperature.⁷¹

Werlé, Leitner, and co-workers demonstrated the possibility of directing the hydrosilylation of CO₂ to either of the HCOOH, HCHO, or CH₃OH levels by adjusting temperature, solvent, and pressure when using a Co triazine pincer complex as catalyst (Figure 4b9).⁷² In general, higher temperature, lower CO₂ pressure, and higher catalyst concentration would favor HCHO formation. Optimally, a yield of 50 % and TON of 56 for HCHO derivatives could be obtained under solvent-free conditions (2.5 mmol silane, 0.2 mol % Co, 0.8 mol % potassium tert-butoxide, 80 °C, 1 bar, 4 h reaction time). Computational results demonstrated that the kinetic energy barrier for the catalytic formation of silyl formate (cycle I, 101 kJ mol⁻¹) is smaller than that of the formation of bis(silyl)acetal (cycle II, 121 kJ mol⁻¹) and that of the formation of methoxysilane (cycle III, 125 kJ mol⁻¹).⁷³ It rationalizes the need of using higher temperature for reducing CO₂ to the HCHO and CH₃OH derivatives. In addition, the similar energy barrier of Cycles II and III indicates that it is difficult to avoid the over-reduction to CH₃OH derivatives. Despite this, the calculated high kinetic hindrance associated with the rearrangement of bis(silyl)acetal to HCHO and 1,3-diphenylsiloxane is identified as crucial to selectively halt the reaction at the HCHO level.

Aldridge and co-workers found that a 3-coordination cationic gallane complex (Figure 4b10), partnered with a hydroborate anion, catalyzed CO₂ reduction selectively to the HCHO level with a TON of 295.8 over 58 h in the presence of Et₃SiH at 60 °C.⁷⁴ Recently, Maron, Xu, and co-workers observed selective generation of bis(silyl)acetal without over-reduction in the catalytic system of homoleptic La aryloxide and B(C₆F₅)₃ (Figure 4b11).⁷⁵ An exceedingly high yield of 99 % was reached, corresponding to a TON of 670 (13 h reaction time), with the bis(silyl)acetal (Ph₃SiO)₂CH₂ formed upon using the sterically hindered Ph₃SiH. Thus, the reaction outcome depended highly on the nature of the silane reductant.

While use of metal complexes as catalysts has dominated the hydroboration/hydrosilylation field, organocatalysts have begun to emerge as more cost-effective and ecologically friendly alternatives for CO₂ reduction. Zhang, Ying, and co-workers reported that N-heterocyclic carbenes promoted the formation of bis(silyl)acetal from hydrosilylation of CO₂ using diphenylsilane as reducing agent.⁷⁶ Unlike the mechanistic pathways shown in Figure 3, the CO₂ reduction was suggested to be initiated by nucleophilic attack of the carbene on CO₂, forming an imidazolium carboxylate. The carboxy moiety would then attack the electropositive silane center and promote hydride transfer to afford the key intermediate, formoxysilane.

Later, Maron, Fontaine, and co-workers found that the ambiphilic organocatalyst 1-Bcat-2-PPh₂C₆H₄ (cat=catechol) can efficiently reduce CO₂ to CH₃OBR₂ in presence of hydroboranes, with the HCHO derivative acting as key intermediate.⁷⁷ Mechanistic investigation revealed that hydroboration occurred through simultaneous Lewis base activation of the hydroborane and Lewis acid activation of CO₂.⁷⁸ In addition, activation of the HBcat moiety by the phosphorus center, while the substrate was fixed and activated by the Lewis acidic boron center, reduced the entropic cost of the catalyzed steps (reductions of HCOOBcat and HCHO) (Figure 6).

Cantat and co-workers revealed that two nitrogen bases, i.e., 1,5,7-triazabicyclo[4.4.0]dec-5-ene (TBD) and 7-methyl-1,5,7-triazabicyclo[4.4.0]dec-5-ene (Me-TBD), were active organocatalysts for hydroboration of CO₂, although they followed two distinct mechanisms.^{79, 80} Notably, the N−H bond of TBD is reactive towards dehydrocoupling with hydroborane and affords a frustrated Lewis pair that can activate CO₂. This mechanism is similar to that of 1-Bcat-2-PPh₂C₆H₄.⁷⁷ In contrast, Me-TBD lacks the N−H moiety and promotes CO₂ reduction through hydroborane activation.

Compared with the hydrogenation of CO₂ using H₂, hydroboration/hydrosilylation can be conducted under much milder conditions with respect to both temperature (<100 °C) and pressure (<3 bar). This facilitates sequential synthesis of value-added compounds from HCHO, let alone mechanistic investigations. An interesting opportunity to explore would be using the HCHO intermediates as methylene transfer agents to generate organic compounds such as amines, as the bis(boryl)acetals thus produced would have value in synthetic chemistry in their own right. On the more problematic side is the high cost of hydroborane and hydrosilane reagents, along with production of stoichiometric waste products (after treatment of the carboxylated borane product with acid to generate HCOOH, HCHO, or CH₃OH). This sums up to large economic and ecological deficiencies of these processes.

3.1 Photochemical Reduction

Photochemical reduction of CO₂ to useful chemicals presents an appealing way of converting sustainable solar energy to chemical energy while helping to mitigate CO₂ accumulation. Figure 7a illustrates how illumination of a photocatalyst with incident light excites electrons from the valence band (VB) to the conduction band (CB), leaving an equal number of holes in the VB. While the photogenerated electrons can drive the reduction of absorbed CO₂, the holes oxidize water to molecular O₂. Both homogeneous molecular and heterogeneous semiconductor catalysts have been widely investigated for photocatalytic CO₂ reduction. Product-wise, the focus has been primarily on CO, ethylene, and ethanol.

In 1979, HCHO was first detected as a photocatalytic product of CO₂ using photosensitive semiconductor powders (TiO₂ and SiC) suspended in water as catalysts.⁸¹ The suspensions were illuminated for 7 h under a 500 W Xe lamp with a cut-off filter (<500 nm). Yields were determined to be 0.11 mmol g_cat⁻¹ for HCHO and 0.023 mmol g_cat⁻¹ for CH₃OH using the TiO₂ catalyst, and 0.1 mmol g_cat⁻¹ for HCHO and 0.54 mmol g_cat⁻¹ for CH₃OH with SiC. Since then, TiO₂ was regarded as one of the best photocatalysts for CO₂ photoreduction because of its high activity and long lifetime. However, high electron–hole recombination rate and limited absorption in the visible region are severe drawbacks, although these can be alleviated by modifying TiO₂ using noble metals through interband electron transition (Figure 7a, right). Thus, a series of Au/TiO₂-based catalysts with low Au loading (0.1–0.5 wt %) were tested for CO₂ photoreduction in the liquid phase over 24 h of continuous irradiation (77 W Hg lamp with a range of emission wavelengths from 254 to 364 nm) at 7 bar pressure at 80 °C.⁸² Major products detected in the liquid phase were CH₃OH, HCOOH, and HCHO, formed in a distribution depending on the exact catalyst formulation used. With 0.1 wt % Au/TiO₂, maximum productivity of HCHO of 0.19 mmol g⁻¹ h⁻¹ was obtained. In another work, Au-NP-decorated TiO₂ nanotube arrays (Au−PMTiNTs) were used as photocatalyst to selectively convert CO₂ to CO (0.32 mmol g⁻¹ h⁻¹) and HCHO (0.42 mmol g⁻¹ h⁻¹) under a 50 W white LED (P_max at 520 nm) for 2 h, followed by irradiating the reactor with simulated sunlight for 2 h.⁸³ In other studies with TiO₂ as catalyst, typically either a low selectivity of HCHO¹⁵ or further conversion of HCHO to CH₃OH^{86, 87} was reported.

In 2013, the design of a bifunctionalized TiO₂ film containing dye-sensitized and catalysis zones was reported for the visible light photocatalytic reduction of CO₂ under a 300 W Xe lamp (λ_max=500 nm).⁸⁶ In the first two hours reaction time, the main products were HCOOH and HCHO, but thereafter the yield of CH₃OH increased sharply because of the preceding buildup of HCHO. This strongly suggests that the stepwise pathway of CH₃OH production proceeding via reduction of HCOOH and HCHO is more favorable than a one-step conversion of CO₂ to CH₃OH. Ananthakrishnan and co-workers introduced benzimidazole-containing Ru metal complexes with two different functional groups (−COOH and −NO₂) for TiO₂ sensitization.⁸⁸ The hybrid catalyst with the -COOH group showed excellent photocatalytic activity towards CO₂ reduction, with a HCHO production rate of 1.1 mmol g⁻¹ h⁻¹ under irradiation of a 250 W tungsten lamp (>420 nm) for 5 h.

In addition to TiO₂, CeO₂ is also a good candidate as photocatalyst for CO₂ reduction, which can activate CO₂ by adsorbing and deforming its linear molecular structure.^{89, 90} Kang and co-workers reported a NiO/CeO₂/rGO (reduced graphene oxide) hybrid composite photocatalyst to produce HCHO selectively (0.42 mmol g⁻¹ h⁻¹) under illumination of a 300 W Xe lamp with a simulated sunlight wavelength for 5 h, around four times faster than if pristine CeO₂ was used (Figure 7b).⁹¹ The high CO₂ reduction ability was ascribed to formation of a p–n junction at the NiO−CeO₂ interface, as this would suppress charge carrier recombination. Furthermore, introduction of rGO allowed accumulation of electrons at the hybrid composite photocatalyst surface with subsequent transfer to the activated CO₂. A carbon-doped NaTaO₃ perovskite annealed at 650 °C presented a HCHO production (0.02 mmol g^-1 h^-1) under irradiation by a pen ray lamp (UV, 254 nm) for 2 h.⁸⁴

In general, homogeneous molecular catalysts pertaining to Ru, Os, Ir, and Co complexes are known to provide good quantum yields for CO₂ reductions.^{92, 93} In the specific case of HCHO as product, Chatterjee and co-workers reported a suitable system consisting of K[Ru(H-EDTA)Cl]⋅2 H₂O as homogeneous catalyst and particulate Pt−CdS−RuO as photon absorber at 505 nm.⁹⁴ Here, the photocatalytic reduction at a constant (dissolved) CO₂ concentration (3.38×10⁻² M) in water, irradiated at 505 nm using a 250 W Xe lamp, produced 0.10 M of HCHO along with 0.22 M of HCOOH during 6 h photolysis at rates of 3.05×10⁻² and 2.0×10⁻² M h⁻¹, respectively.

Currently, state-of-the-art photocatalysts consist of semiconductors with appropriate band gap or homogeneous catalysts such as the costly Ru complexes. The first challenge of CO₂ photoreduction is the low solubility of CO₂ in aqueous solution. Although low solubility can be mitigated by increasing temperature or pressure as reported in some works, this would soon show its costs on the energy account. Second, the selectivity towards HCHO is modest. Third, the quantum yield of the photocatalytic process is comparably low. This low utilization rate of solar energy calls for development of more active and efficient photocatalysts before large-scale application can come true.

3.2 Electrochemical Reduction

Electrochemistry has emerged as a promising platform for CO₂ reduction with electrons from the electrode serving as reductants (Figure 7c). Such reductions can usually be conducted under ambient conditions without necessity of adding chemical reductants. The electrochemical CO₂ reduction reaction (eCO₂RR) has been capable of generating CO, HCOO⁻, hydrocarbons, and oxygenates in high yields and selectivity using various kinds of metal-based catalysts, single-atom catalysts, and molecular catalysts.^95-99

In contrast, reports on HCHO have been relatively scarce because of the difficulty in preventing its further reduction once formed. Nevertheless, HCHO is often identified as minor product (<10 %) in eCO₂RR. In 2014, Dyer and co-workers reported a pterin (natural cofactor for a wide range of enzymes) electrocatalyst, 6,7-dimethyl-4-hydroxy-2-mercaptopteridine (PTE), which catalyzed the reduction of CO₂ on a reticulated vitreous carbon electrode to HCOOH, HCHO, and CH₃OH.¹⁰⁰ It was suggested that PTE acted as a multifunctional catalyst for converting CO₂ to HCOOH, followed by further reduction to HCHO or other products. Later, Proust and co-workers used (TOA)₆[α-SiW₁₁O₃₉Co(_)] (TOA=tetraoctyl ammonium; _=vacant position in the coordination sphere of Co) as homogeneous electrocatalyst and achieved a maximum faradaic efficiency (FE) for HCHO formation of 40 % at −1.5 V vs. saturated calomel electrode (SCE) in CH₂Cl₂.¹⁰¹ Unfortunately, CH₂Cl₂ as solvent greatly limits practical applications due to its high volatility and potential degradation (dehalogenation) during electrolysis.

Nakata, Einaga, and co-workers discovered boron-doped diamond (BDD) as an efficient electrocatalyst in CO₂-to-HCHO conversion under ambient conditions with a FE of 74 % in CH₃OH electrolyte and 62 % in 0.1 M NaCl aqueous electrolyte at −1.5 V vs. Ag/Ag⁺ (Figure 7d).¹⁰² Notably, HCOOH came out with FE <15 % and hydrogen evolution was, likewise, of small concern. Compared with glassy carbon electrodes consisting of sp²-bonded carbons, BDD as electrode material showed much better selectivity for HCHO, inferring that the sp³-bonded carbons on BDD might be the key factor. Yet, current densities were exceedingly low (in the μA cm⁻² range), calling for further performance improvements by, e.g., increasing the number of active sites and tuning the electronic structure of BDD.

Recently, Wang, Liang, and co-workers reported a Co β-tetraaminophthalocyanine molecular catalyst supported on carbon nanotubes which converted CO₂ and NO₃⁻ to methylamine in aqueous media under ambient conditions.¹⁰³ Methylamine was produced at a maximum FE of 13 % with a partial current density of 3.4 mA cm⁻² at −0.92 V vs. reversible hydrogen electrode (RHE). At first, electrochemical reductions of CO₂ and NO₃⁻ proceed independently with formation of HCHO and NH₂OH, but soon adsorbed HCHO is to undergo nucleophilic attack by NH₂OH. This yields formaldoxime, which is reduced to N-methylhydroxylamine and further to methylamine. Later, the authors demonstrated that addition of different nitrogen sources, i.e., amine, hydrazine, hydroxylamines, and nitro compounds, could be feasible for electrochemical reductive N-methylation.¹⁰⁴ This in situ trapping approach can broaden the adoption area of suitable electrocatalysts capable of producing HCHO as intermediate, in spite of the challenges persisting in halting the process at the HCHO level under reductive electrochemical conditions.

Although electrochemical reductions present one of the most sustainable approaches for producing HCHO from CO₂, the number of electrocatalysts capable of realizing this transformation is quite limited. Moreover, currently reported catalysts are met with the serious drawbacks of having both low activity and selectivity. More investigations are therefore needed to improve the overall performance by increasing the number of active sites and tuning the electronic structure.

3.3 Photoelectrochemical Reduction

Photoelectrochemical reduction aims to combine the best of the photochemistry and electrochemistry worlds. While the photoelectrode generates electron–hole pairs upon light irradiation, the applied bias potential drives the electrons to the cathode for CO₂ reduction (Figure 7e). As a result, the recombination probability of photogenerated electron–hole pairs diminishes while the electron–hole transfer to their respective acceptors accelerates, which maximizes the overall yield. In addition, the photovoltage gained at the semiconductor/solution interface can be utilized to generate an underpotential, reducing the required bias potential to activate CO₂.

In 1978, Halmann performed the first study of this kind on CO₂ reduction using p-type Ga phosphide as photocathode.¹⁰⁵ After 18 h irradiation under a high-pressure mercury arc with no filters and application of a cathodic bias of −1.0 V vs. SCE, HCOOH was produced as main product (1.2×10⁻² M), accompanied by a minor amount of HCHO (3.2×10⁻⁴ M) and CH₃OH (1.1×10⁻⁴ M). Similarly, low HCHO selectivity was the result of many subsequent investigations.^{16-18, 106} Then, in 2018, Kang and co-workers reported an efficient system combining a (040)-facet engineered BiVO₄ plate photoanode (040-BVO) and a Cu cathode in NaCl electrolyte (Figure 7f).¹⁰⁷ The maximum FE achieved for HCHO was 85.1 % at a bias potential of −0.9 V vs. RHE under irradiation by a 300 W Xe arc lamp with simulated sunlight. Recently, they designed a Ca–Fe doped TiO₂ photoanode and a Cu cathode modified using, as minimum, rGO.¹⁰⁸ With this system, HCHO was the major product over Cu/rGO with a FE of 25 %, while acetaldehyde became the major product over a Cu/rGO/polyvinylpyrrolidone/Nafion cathode with a FE of 39 % at −0.68 V vs. RHE bias potential under solar simulated radiation for 8 h.

Photoelectrochemical reduction is one of the most interesting technologies proposed for CO₂ conversion. Promising results can be expected with more efforts directed towards the development of photoelectrodes and optimization of reactor configurations.

5.1 Issues That Need to Be Resolved

HCHO is a vital platform chemical used for producing value-added products such as resins, polymers, and adhesives by the chemical industry. Compared with traditional strategies to produce HCHO, based on partial oxidation/dehydrogenation of CH₃OH, catalytic reduction of CO₂ presents a sustainable, yet highly challenging pathway. In this sense, selective generation of HCHO utilizing CO₂ under mild reaction conditions remains an important goal in chemistry.

The last few decades have witnessed intense research in the CO₂-to-HCHO conversion using hydrogenation, hydroboration/hydrosilylation as well as photochemical, electrochemical, photoelectrochemical, and enzymatic reductions. However, despite these advancements, significant issues remain to be addressed carefully before industrial applications can be even considered. In general, hydrogenation of CO₂ includes three cascade catalytic cycles: CO₂ hydrogenation to the HCOOH level, HCOOH hydrogenation to the HCHO level, and HCHO hydrogenation to CH₃OH. The key issue is how to halt the reduction of CO₂ at the HCHO level, i.e., achieve quantitative conversion of HCOOH while suppressing further reduction of the produced HCHO. Unfortunately, the low energy barrier of the latter reduction process makes this task exceedingly difficult, although alteration of the energy span for the HCHO and CH₃OH levels can be realized by modulating the reaction conditions, i.e., temperature, solvent, CO₂ pressure, concentrations, etc.

In the cases of electrochemical and photochemical reduction, the situation often turns out to be more convoluted due to the possibility of having proton-coupled electron-transfer processes. Along with this come CO₂ mass transport limitations in aqueous solution because of the relatively low solubility of CO₂, resulting in low conversion rates and low selectivity. Encouragingly, utilization of gas diffusion electrodes and flow-cell reactors has, by and large, mitigated the gap between laboratory discovery and industrial needs in this respect.

A point, often considered trivial but important, is the procedure used for HCHO quantification. HCHO is generated in small quantities with low selectivity, and under aqueous conditions, hydrated or oligomeric forms of HCHO exist alongside free HCHO, making its quantification challenging. Various techniques have been applied, including high-performance liquid chromatography (HPLC),¹¹¹ gas chromatography,¹⁰⁸ fluorimetry,¹¹² nuclear magnetic resonance,¹¹³ and UV/Visible spectroscopy.⁷ Notably, pretreatments of HCHO are always necessary in order to form more easily detectable derivatives using reagents such as 2,4-dinitrophenylhydrazine,¹⁰⁶ 3-methyl-2-benzothiazolinone hydrazine,⁴⁰ acetoacetanilide,⁷ 2,4-pentanedione,¹⁰¹ and sodium bisulfite.^{104, 114} HPLC based on hydrazine derivatization is the most frequently used method, as the detection limit is exceedingly low (3 nM). However, the hydrazine agents may also react with NO, NO₂ in air, and other aldehydes, decreasing the accuracy and selectivity for HCHO quantification.¹¹⁵

The Hantzsch reaction, based on derivatization of formaldehyde with β-diketone (acetoacetanilide, 2,4-pentanedione), comes with good sensitivity and selectivity for HCHO, but with a much lower reaction rate.¹¹⁶ The chromotropic acid method requires heating of the sample under strongly acidic conditions, which is undesirable in many applications.¹¹⁷ Recently, sodium bisulfite has proved to be an efficient agent for HCHO quantification and H¹³CHO identification using NMR.¹¹⁴ The drawback of this method is the reactivity of sodium bisulfite itself towards CO₂. Consequently, the design of even more specific agents would be of great interest to realize a fast, low-cost, selective, and highly sensitive analysis of HCHO.

5.2 Strategies for CO₂-to-HCHO Conversion

Given the rarity of chemical platforms that can efficiently and selectively reduce CO₂ to the HCHO level, development of new catalysts would be of considerable importance from both an energy cost-efficiency and environmental perspective. Considering that numerous catalytic systems are known already to efficiently reduce CO₂ to HCOOH or CH₃OH, they would be worth looking at in a first approach, once appropriately modified and once reaction conditions have been carefully adjusted, i.e., temperature, CO₂ pressure, reductant type, concentration, etc.

Among the reported catalysts, organocatalysts present a promising research field for achieving high selectivity towards HCHO. In general, these catalysts together with hydroboranes or hydrosilanes provide diverse mechanistic pathways for reduction of CO₂, which offers more chances of halting the cascade of reduction processes at the HCHO level. Although currently available organocatalysts are yet to arrive at a stage where this can be done efficiently, a more elaborate catalyst design, integrating both the Lewis acid and base into a single molecule, would enable direct manipulation of the binding strength of specific intermediates, let alone reduce the entropic cost of the catalyzed step.

In situ trapping of HCHO provides an interesting approach for diverting unwanted reactions and halting CO₂RR at the HCHO level. In hydroborations and hydrosilylations, trapping agents have already enabled the synthesis of complex molecules through formation of new C−C, C−N, C−O, and C−P bonds. Much of this work could easily be extended to electro- or photochemical reductions with the notion that judicious choice of the trapping agent with respect to the catalyst system would be vital to avoid further reduction. As mentioned, the biocatalysis field could also benefit greatly from introducing such approaches.

In general, mechanistic investigations are indispensable during the optimization of catalyst structures. In fact, without a fundamental understanding of the various mechanistic pathways and their energetics as well as identification of key intermediates and/or transition states, optimization of reaction conditions would have to rely on a trial-and-error approach to a large extent. In the quest towards mechanistic understanding, use of both theoretical studies and in situ/operando tools becomes indispensable. To date, research on theoretical models for HCHO formation is too rare to provide precise guidance for experimental operation and should be further developed.

Application of in situ/operando characterization techniques in catalytic CO₂-to-HCHO conversion is highly desirable, as any short-lived intermediates in the stepwise reaction kinetics will be left undetectable by ex situ characterizations. Recently, various techniques of this kind based on X-ray diffraction/scattering, X-ray photoelectron spectroscopy, X-ray absorption spectroscopy, Raman spectroscopy, infrared spectroscopy, and transmission electron microscopy have been used in CO₂RR to resolve many unsettled issues.^{118, 119} In future work, much more attention should be given to the HCHO formation step using these important techniques to provide clearer guidance for catalyst design based on a thorough mechanistic understanding.