The Relationship Between Catalyst and Solvent in Hydrogenation via Condensed Phase Heterogeneous Catalysis

Abstract

To understand a system is to understand its components and their sum. Cascading interactions between catalyst, solvent, and reagent create a complex web of influences when heterogeneous catalysis meets the condensed phase. Due to the importance of heterogeneous catalysis in chemical manufacturing, and the present and growing potential of condensed phase chemistries, the understanding of these interactions is of paramount importance. To develop condensed phase heterogeneous catalysis, the field needs to develop understanding of the role of solvent in heterogeneous catalytic hydrogenation. While no small feat, fields such as biofuel and petroleum refining have established certain applicable generalities that can bridge the knowledge gap in emerging technologies such as integrated carbon capture and conversion to materials (IC³M). In this review, we thoughtfully probe the current paradigm of condensed phase catalysis by challenging the idea that catalyst and solvent are independent reaction design choices. Challenges such as lack of experimental stability studies and poor resolution on our conceptualization of the condensed phase environment are discussed. Parameters such as viscosity and the dielectric constant, and their role on reaction activity and stability are explored. Knowledge gained from established biomass and petroleum processes is discussed and used to anticipate behavior in novel processes.

Heterogeneous catalysis has historically been developed in the gas phase, however, we have found it pertinent to highlight solvent effects that must be accounted for in liquid environments. Condensed phase catalysis, like that in biomass, petroleum and the conversion of CO₂, stands to benefit from the typically overlooked discussion regarding the relationship between catalyst, solvent and reagent.

1. Introduction

Catalysis is a backbone in chemical manufacturing, with approximately 85%–90% of products of the chemical industry made via catalytic processes [1]. Homogenous catalysis and enzymology (biocatalysis) have critical roles to play in industrial production. Typically, homogenous catalysis has superior selectivity to a comparable heterogeneous counterpart. However, homogenous catalysts are typically only stable at mild conditions and often impractical to recycle postuse [2, 3]. Enzymes boast typically incredible selectivity but have limited stability at elevated temperatures and require advanced techniques for stabilization [4]. A significant advantage of heterogeneous catalysts compared to homogeneous and enzyme‐based catalysts is the ease of recycling of the catalyst material following the reaction. As with other catalysts, the stability of heterogeneous catalysts remains a front‐line concern both for the catalyst and the total system. An ideal heterogeneous catalyst is active, selective, and stable. In 2022 Martín et al. [5] assessed a decade's worth of catalysis literature (∼513,000 reports) and found that 72% of the articles directly addressed matters of catalytic activity. 20% addressed product selectivity, an important parameter in assessing catalysts, especially in the current era of atom efficiency. Interestingly and surprisingly, only 8% of reports dealt with stability. Catalyst stability is typically discussed in the context of gas phase reactions, but the addition of a solvent provides a new axis of degradation. Despite the broad dependence of humanity on heterogeneous catalysis, Martín et al. [5] points out the unfortunate reality that “catalysts are not immortal”. Catalyst stability is important, but just one component of a multiphasic system. For a multiphasic system to possess stability, the catalyst, solvent, and reagent must coexist harmoniously. This is especially clear in systems such as integrated carbon capture and conversion, where a CO₂ capture agent must experience minimal degradation through the cyclic capture and release process.

Interest in the utilization of CO₂ as a carbon feedstock has driven research into condensed phase heterogeneous catalytic chemistry. As is often the case, nature already uses atmospheric CO₂ to create glucose and its polymer cellulose. The average tree in tropical climates stores 50 pounds of CO₂ per year as plant matter [6] with carbon making up is ∼50% of the dry weight of wood (species dependent) [7]. Leaving the tree intact represents a form of sequestration, but wood can also be used as fuel. This provides a parallel to the anthropogenic suggestion of integrated carbon capture and utilization and potential inspiration for humanity to utilize captured CO₂ for direct conversion to beneficial products.

Condensed phase heterogeneous catalytic chemistries have gained increasing attention in the field of carbon capture and conversion. CO₂ exists as a waste stream in many industrial processes, including power plants [8], cement production [9], and steel manufacturing [10]. The petroleum refining sector produces 478,000 metric tons of CO₂ a year [11], with ∼65% attributed to stationary combustion processes [12]. Ideally, this carbon can be recovered and converted into a product stream, offsetting the cost of the capture process in the short term and potentially generating additional profit in the long term. Significant ongoing research on cutting edge capture agents, such as N‐(2‐ethoxyethyl)‐3‐morpholinopropan‐1‐amine (2‐EEMPA), has yielded water lean, single component solvents with elevated durability, low capture cost and high capture efficiency [13]. It has been demonstrated that via an integrated carbon capture and conversion process, heterogeneous catalysis can successfully convert the captured form of CO₂ into valorized products such as methane [14], methanol [15], and cyclic carbonates [16]. Barpaga et al. [17] recently used 2‐EEMPA to demonstrate integrated semi batch simultaneous CO₂ capture and conversion to methanol. Impressively, the capture agent was not loaded with pure CO₂. The CO₂ was captured at >90 mol % efficiency from simulated coal‐derived flue gas, achieving  >60 C‐mol% conversion and >80 C‐mol % selectivity to methanol. Interesting solid‐state chemistries for integrated carbon capture and conversion to materials (IC³M) have been demonstrated, such as Cu(I)‐MOF NU‐2100, which selectively captures CO₂ and H₂ and converts them to formic acid [18]. Relatedly, Wang et al. [19] used an Ni–MgO–Al₂O₃ dual functional material for IC³M to produce synthetic natural gas (SNG), reporting a flue gas treatment efficiency of 280 L h⁻¹ kg⁻¹ and a 95% conversion to methane at 100% selectivity. The growing feasibility of this technology gives promise to the actualization of increased atom efficiencies for reclaiming otherwise wasted materials in process streams for CO₂ generating processes.

Despite an increasingly promising outlook, there are significant challenges to implementing IC³M. Freyman et al. [20] identified two key challenges: (1) reducing the energy input required to produce products form CO₂ and (2) increasing reactor/process productivity. Given the goals of increased stability, productivity, and efficiency, how should one go about improving the IC³M process? Typically, condensed phase reactions are designed by screening working gas phase catalysts in a variety of solvents, before selecting the best performing solvent. Given the plethora of differences between the gas and condensed phases (Table 1), there are many unexplored distinctions that could be used to better IC³M outcomes.

TABLE 1.

Comparison of gas and condensed phase catalysis.

Challenge area	Gas phase catalysis	Condensed phase catalysis
Solvent effects	No solvent present	Effects of pH, bulk solvent structure, reagent solubility, chemical reagent modification, hydrogen shuttling and physical parameters such as viscosity.
Mass transfer	Typically unhindered	Solubility of reagents and viscosity of solvent limit the movement of reagents to the catalyst surface. Significantly greater mass transfer resistances.
Reaction complexity	Reagents diffuse to catalyst, adsorb to surface, react, desorb from surface, and migrate to bulk.	Same as gas phase, with added interactions with a liquid phase molecularly dense solvent. The solvent can interact with the reagent, catalyst, and itself, complicating reaction mechanisms.
Heat transfer	Typically, lower heat capacities and thermal conductivity of bulk phase compared to liquid phase. Transport of heat‐dissipating mass faster than in the condensed phase.	Higher heat capacities and thermal conductivities of the condensed phase aid in heat dissipation and abatement of hotspots; however stagnant liquid pools can form in catalyst pores.
Catalyst stability	Sintering, poisoning, coking, volatilization, and phase transformation, among other phenomena, threaten catalyst stability in the gas phase.	All degradative phenomena from the gas phase, with the additional concern of leaching. Solvents can poison and mechanically obscure active sites on the catalyst.
Modeling and simulation	Most modern modeling techniques can adequately simulate gas phase reactions.	Simulations are typically limited to a few solvent molecules and have difficulty capturing the true “crowded” nature of the condensed phase.

We propose that rather than circumnavigating solvent effects to mirror gas phase catalytic results in a liquid media, that researchers utilize these effects to achieve outcomes impossible in the gas phase. Admittingly, applying this approach in the carbon capture and utilization field will require a deep and rigorous study of solvent effects, a large whitespace in the current literature. Parameters such as pH, relative permittivity, hydrogen solubility, solvent coordination, and many others have limited documented exploration in condensed phase catalytic CO₂ hydrogenation. Therefore, we put forward that through a systematic review of previous successful hydrogenation systems, such as those found in the biomass and petroleum fields, the condensed phase valorization of CO₂ can move beyond boundaries of the gas phase. Through a principle guided conceptualization of the liquid environment, a compilation of documented hydrogenation solvent effects, and systematic tabulation and evaluation of relevant solvent parameters, we can gain more control over the activity, durability, and efficiency of catalysis in the condensed phase.

2. Principles of Heterogeneous Catalysis in the Condensed Phase

2.1. Mass Transport in the Condensed Phase

Heterogeneous catalysis has demonstrated broad and historical success in the efficient conversion of gas phase reagents. Gas phase heterogeneous catalytic processes, such as Fischer–Tropsch and Haber–Bosch, have had a truly enormous impact on society. These reactions fundamentally rely on gas phase reagents engaging over a solid interface. Large swaths of literature covering gas phase heterogeneous catalysis and related phenomena exist. However, complexities arise upon the addition of a solvent. To start, the molecular density of an ideal gas at 298 K and 1 atm is 2.46 $\times$ 10²⁵ molecules/atoms per m³. In comparison, liquid phase water at the same conditions (ρ_{298K, 1 atm} = 0.997 g/ml [21]) possesses a molecular density of 3.33 $\times$ 10²⁸ molecules per m³, a ∼124,000% increase in the number of molecules per given area. This trivial calculation illuminates a simple challenge in conceptualizing the condensed phase, as being able to adequately imagine/observe the reaction system is paramount to creating accurate models and feasible hypotheses. This large difference in molecular densities is a major reason for differences between gas and liquid phase reactions.

A typical catalytic surface reaction can be broken down into steps: (1) reagents transport to surface, (2) reagents adsorb on surface, (3) reagents react on surface/products form, (4) products desorb from surface, and (5) products transport away from surface. In both the gas and condensed phases, the ability of a reagent to reach the catalyst surface is governed by convection and diffusion. Momentum, heat, and pressure gradients drive convective flow. However, as this flow approaches a solid surface, such as a catalyst particle, it is often assumed that the velocity vector of the fluid possesses a magnitude of zero relative to the surface of the particle [22]. It is important to note that are many exceptions to this “no‐slip” boundary condition, with factors such as wettability and roughness playing critical roles in determining the amount of slip [23, 24]. In either case, there exists a regime in which diffusion takes over as the dominant mass transfer mechanism either near the catalyst surface or within the pore complex. Common catalytic supports such as silica [25], alumina [26], and carbon [27] contain pore systems with large amounts of internal surface area, often with an internal surface area much greater than the external surface area. Within these pores bulk flow is negligible and Fickian diffusion typically reigns as the dominant mass transport mechanism (dependent on pore/diffusing molecule size). A consequence of the increasing molecular density between the gas and condensed phase is changes in the Fickian diffusion behavior, directly impacting steps 1 and 5. For a typical gas phase heterogeneous catalytic reaction

$$\begin{array}{c}{\mathrm{\upsilon }}_{\mathrm{A}}\mathrm{A}\left(\mathrm{g}\right)+{\mathrm{\upsilon }}_{\mathrm{B}}\mathrm{B}\left(\mathrm{g}\right)\hspace{0.17em}\underset{\mathrm{Cat}.}{\leftrightarrow }\hspace{0.17em}{\mathrm{\upsilon }}_{\mathrm{C}}\mathrm{C}\left(\mathrm{g}\right)+{\mathrm{\upsilon }}_{\mathrm{D}}\mathrm{D}\left(\mathrm{g}\right)\end{array}$$ (1)

where gaseous reagents A and B reversibly react to form gaseous products C and D, A and B must initially diffuse through one another to reach the catalytic surface (Figure 1a). As products form, and begin to leave the catalyst surface, A and B must continue to diffuse to the catalyst surface through C and D (Figure 1b). Upon the addition of solvent, A and B must now diffuse through C, D, and solvent molecules, with the density of solvent molecules often rendering the concentration of C and D negligible (Figure 1c).

FIGURE 1.

(a) Gas phase reagents diffusing through a gaseous medium to the catalytic surface, (b) gas phase products leaving the catalytic surface to the bulk, and (c) solvent hindered transport of product and regent molecules diffusing through crowded solvent molecules to reach the bulk solvent and catalytic surface, respectively.

In triphasic reaction systems where a gaseous reagent reacts on a solid catalyst surface in a solvent, the first transport step can be divided into the diffusion of the gas reagent through the gas phase, absorption into the liquid phase, and then transport through the liquid phase to the catalyst surface. Fick's law of diffusion relates the gradient mass vectors ($\mathrm{\nabla }{w}_i$) to the mass fluxes (${\mathit{j}}_i$) of species [28, 29] commonly shown in matrix notation,

$$\begin{array}{c}\overline{\mathit{j}}=-\rho \mathit{D}\nabla \mathit{w}\end{array}$$ (2)

where for a mixture of N species with density $\rho$, where species $i$ exists in mass fraction $w_i$. $D_{ij}$ is the diffusivity of species $i$ through species $j$. $\overline{\mathit{j}}$ is the vector of molecular flux and $\mathrm{\nabla }\mathit{w}$ is the mass gradient vector. $\mathit{D}$ represents the matrix of diffusivities, and is represented by

$$\begin{array}{c}\mathit{D}={\mathit{B}}^{-\mathbf{1}}\mathbf{\Gamma }\end{array}$$ (3)

where matrix $\mathit{B}$ is a function of mass fractions and inverse drag coefficients or Maxwell‐Stefan diffusivities [28, 29]. These Maxwell‐Stefan diffusivities work well for ideal, dilute gas cases. To adjust these to the Fickian counterpart, nonidealities are accounted for in the $\mathbf{\Gamma }$ matrix [30, 31], whose elements are given by

$$\begin{array}{c}{\text{Γ}}_{\mathit{ij}}={\delta }_{\mathit{ij}}+w_i\frac{\partial {\mathit{ln\gamma }}_i}{\partial w_i}\end{array}$$ (4)

where ${\delta }_{ij}$ is the Kronecker delta [32] ($i=j\to {\delta }_{ij}=1\wedge i\ne j\hspace{0.17em}\to \hspace{0.17em}{\delta }_{ij}=0$) and ${\gamma }_i$ is the activity coefficient for the $i$‐th component of the mixture. For gas mixtures that deviate from ideal behavior, ${\gamma }_i$ is replaced with the fugacity coefficient (${\varphi }_i$) [28]. ${\gamma }_i$ is a parameter that appears on many occasions in nonideal condensed phase theory and will be discussed in further detail in Section 2.4. In short, the ${\gamma }_i$ of a species change as a function of interactions between species $i$ and the environment it moves through and is typically higher for bulkier molecules with interacting substituents. Intuitively, as the molecular density of the working volume increases, the frequency of these interactions increases, leading to deviations from ideality. In an ideal mixture with negligible ${\gamma }_i$, diffusivities are independent of the mass fraction of the species (eloquently derived by Upreti and Mehrotra [29]). Kundra et al. [33] demonstrated the nonideality of CO₂'s diffusivity through polypropylene with results indicating that the diffusivity coefficient of CO₂ is dependent on the concentration of CO₂ in the mixture, at a range of temperatures (170°C–190°C).

The mean free path of particle is the average distance traveled by that particle prior to a collision with another particle [34]. Knudsen diffusion occurs when the size of the pore that the particle is in is smaller than the mean free path of the particle [35]. While this can be observed in the gas phase with microporous pore systems [36, 37], different considerations must be made in the condensed phase given the significantly higher molecular densities. Knudsen diffusion is rarely achieved in the condensed phase due to the incredibly short mean free paths observed. Knudsen diffusion in condensed phase heterogeneous catalysis could arise if the pore diameter of the catalyst is larger than the gas phase reagent, smaller than the liquid solvent molecules, and smaller than the mean free path of the gas molecule. These conditions are commonly met in gas–liquid separation membranes [38]. A key component of study in triphasic catalysis is the phase interface and mass transport through it. After all, a molecule of reagent that begins in the gas phase must diffuse through the gas face to the interface, travel through the interface, then diffuse through the liquid phase to reach the catalyst surface. A present challenge in the condensed phase is that diffusivities in liquids are ∼4–5 orders of magnitude lower than in gas [39]. Upreti et al. [29] succinctly shows that the expression for mass flux of a gas diffusing into a liquid combined with the mass balance over the system. If the concentration of gas in the liquid is minimal, one can assume that the density of the media can be treated as constant and the diffusivity of the gas is not a function concentration yielding a usable form of Upreti's derivation. From this a general solution is found, and the boundary condition of setting equal the diffusive flux of gas into the liquid and the gas mass flux across the interface is applied to yield

$$\begin{array}{c}-D\frac{\partial c}{\partial z}=k(c_{\mathit{eq}}-c)\end{array}$$ (5)

where $c$ is the gas concentration in the liquid, $z$ is the distance into the liquid layer away from the diffusing gas, $D$ is the diffusivity of the gas into the liquid, $k$ is the mass transfer coefficient of the interface, and $c_{eq}$ is the equilibrium concentration of the gas in the liquid. In a review [29], Upreti compiles examples where Equation (9) has been used to determine diffusivities in scenarios involving interfacial mass transfer at gas–liquid interfaces [40, 41]. $c_{eq}$ can be found using Henry's Law [42], which relates the concentration of a species a ($c_a$) in a solvent, at an equilibrium partial pressure ($p_a$):

$$\begin{array}{c}{c}_a=k_H{p}_a\end{array}$$ (6)

where $k_H$ is the Henry's Law constant. Here, the word “constant” can be misleading, as the Henry's Law constant is dependent on the identity of the gas and liquid species as well as the temperature of the process. Equation (6) must be modified for liquids that perform reactive capture, such as those found in carbon capture [43]_. This nonconstant behavior is readily observed for CO₂ and H₂ in water. For H₂, $k_H$ (in water) is found to increase from the range of ∼300–350 K before rapidly approaching 0 from ∼350–600 K [44]. Comparatively, $k_H$ of CO₂ (in water) increases over the range of ∼300–400 K before a steady decay from ∼400–600 K [45]. This creates complexity in systems such as condensed phase CO₂ hydrogenation to methanol, where the concentration of reagents in the bulk fluid phase near the catalyst is a function of temperature. The selectivity to products of the CO₂ to methanol reaction also depends on the concentration of reagents in the bulk [46]. By extension, this provides another mechanism by which temperature affects selectivity in the condensed phase, whereas temperature dependent selectivity is typically governed by the equilibrium thermodynamics of competing reactions in the gas phase (see CO₂ to methanol vs. the reverse water gas shift reaction (RWGS) [46]). These temperature considerations mandate managing heat effects in the condensed phase to effectively design active catalytic reactions.

There are methods of abating mass transfer limitations seemingly intrinsic to the condensed phase. In structures such as zeolites, the hydrophobicity of the pores can be tuned to control the structure of the solvent facilitating mass transfer [47]. Further, the hydrogen donation ability of a solvent plays a key role in the mechanism of hydrogen shuttling. While studying a Ru/Co₃O₄ catalyst for CO₂ to CH₄ in a batch reactor, Song et al. [48] performed a solvent screening using water, butyl alcohol, dimethylformamide, n‐nonane, decalin, cyclohexane, isooctane, and 1,4‐butyrolactone. In general, it was found that the hydrogen solubility in a solvent had a positive relationship with the yield of the hydrogenation product, CH₄. Despite this, decalin with its hydrogen solubility of 67.5 mmol L⁻¹ had a CO₂ conversion of 45.6%, while n‐nonane, with solubility 93.5 mmol L⁻¹, demonstrated a conversion of 31.9%. Using NMR evidence, it was determined that decalin could more readily donate its tertiary carbon proton versus n‐nonane's secondary carbon and that the decalin species could adsorb on the catalyst surface, giving proximity to the proton transaction. Similarly, Liu et al. [49] studied the hydrogen‐shuttling mechanism of liquefaction solvents using Fe and Ni catalysts. They found that hydrogen shuttling exists as a set of reactive pathways, in which hydrogen is maneuvered via the aid of a solvent, rather than a clearly defined mechanism. The condensed phase provides another axis of complexity where the above‐mentioned effects can significantly affect catalyst activity and selectivity. It is necessary to understand the solvent parameters, mainly as a means of categorization. It should be noted that many of these parameters are interconnected and isolating a single parameter to relate it to activity may be impossible in some cases. Therefore, intelligent solvent choice relies on familiarity with solvents descriptors, their relationships, and the role of key parameters on catalyst activity and stability. As is a theme throughout this review, IC³M processes have potential to have time dependent solvent viscosity. Many proposed capture agents become increasingly viscous upon loading of CO₂ and as the CO₂ is converted, the capture agent returns to its preloading viscosity. The effects of this time and conversion dependent viscosity on mass transfer in the IC³M process remain underexplored.

2.2. Heat Transport in the Condensed Phase

A brief review of heat transport principles can help develop an intuition for assessing solvent choices for condensed phase heterogeneous catalysis. Classically, the three modes of heat transfer are conduction, radiation, and convection [50]. Analogous to Fick's law of diffusion, the heat convection of a system is described with Fourier's law [51],

$$\begin{array}{c}{\overline{\mathit{j}}}_{\mathbf{H}}=-\mathit{\kappa }\mathrm{\nabla }\mathit{T}\end{array}$$ (7)

where ${\overline{\mathit{j}}}_{\mathit{H}}$ is the heat flux vector, $\mathrm{\nabla }\mathit{T}$ is the temperature gradient, and $\mathit{\kappa }$ is a 3 x 3 thermal conductivity matrix. For isotropic materials, $\mathit{\kappa }$ is simplified to a scalar value. In the context of fixed bed catalytic reactors, particles or pellets possess a temperature and concentration gradient. With increasing particle depth, the concentration of reagent decreases and the concentration of product increases. Temperature gradients are drawn based on the exothermicity or endothermicity of the reaction. Simply put, the nature of the reaction governs these gradients within the particle [52]. At steady state for a spherical particle, a two‐dimensional Fourier's law in conjunction with an energy balance over the particle, yields an expression to determine the temperature difference between the surface and a point within the particle [52]

$$\begin{array}{c}\left(T-T_{\mathrm{s}}\right)=\frac{D_{\mathit{eff}}}{K_{\mathit{eff}}}\left(\left[\mathrm{X}\right]-{\left[\mathrm{X}\right]}_{\mathrm{s}}\right)\left(-\text{Δ}\mathrm{H}\right)\end{array}$$ (8)

here $T$ is the temperature at the given interior point, $T_s$ is the surface temperature, $D_{eff}$ is the effective diffusion, $K_{eff}$ is the effective heat transfer coefficient, $[\mathrm{X}]$ is the concentration of reagent at the given interior point, ${[\mathrm{X}]}_{\mathrm{s}}$ is the concentration of the reagent at the surface, and $\text{Δ}\mathrm{H}$ is the heat of reaction. A maximum temperature difference can be found by setting $[\mathrm{X}]=0$. These temperature gradients can lead to localized increased rates of reaction via the Arrhenius equation and differing product selectivity's via Le Chatelier's principle.

In gas phase catalysis, phenomena such as “hot spots” are often reported. Hot spots typically occur during exothermic reactions, where regions of a catalyst become significantly hotter than the desired surface temperature, on the order of 100 K greater [53]. To maintain homogenous temperature conditions, heat must be effectively transported away from the “hot”’ location. In the gas phase, this is often done using SiC as a heat homogenizer [54]. α‐SiC has a high thermal conductivity of 146–270 W m⁻¹ K⁻¹ (varies by type) compared to other “inert” materials like SiO₂ (0.015⁻¹ W m⁻¹ K⁻¹) [55]. Due to this high thermal conductivity, SiC can rapidly distribute heat in a catalytic bed, aiding in the elimination of hot spots [54]. In the condensed phase, solvents can serve a similar function to SiC in removing heat, with a balance of parameters dictating the rate and effectiveness of heat removal. For continuous flow systems, the flow rate, thermal conductivity, and specific heat play significant roles in this determination. The larger the thermal conductivity, the less contact time is needed to remove heat from the surface of the catalyst. The higher the specific heat of a solvent, the more heat it can contain without a correlating elevation of temperature—analogous to the number of passengers a train can hold. Finally, the flow rate dictates how quickly the heat can leave the system boundary and be effectively removed. Table 2 shows some select thermal conductivities and specific heats for common gases and solvents used in relevant systems.

TABLE 2.

Thermal conductivities ( κ ) and isobaric specific heat capacities (C_p) of select species at 0.1 MPa. MEA = monoethanol amine, DEA = diethanol amine, TEA = triethanol amine.

Species	T, K	$\kappa$, W m−1K−1	Ref.	T, K	Cp, J g−1 K−1	Ref.
N2(g)	300	0.0259	[56]	300	1.04	[57]
Air(g)	300	0.0262	[58]	300	1.01	[57]
H2(g)	300	0.1866	[59]	300	14.3	[57]
He(g)	300	0.1557	[60]	300	5.19	[57]
CO2(g)	300	0.0168	[61]	300	0.85	[57]
Water (l)	300	0.6096	[62]	298	4.19	[63]
n‐hexane(l)	303	0.1189	[64]	298	2.27	[65]
Toluene(l)	298	0.1305	[66]	298	1.71	[67]
MEA(l)	298	0.2399	[68]	303	2.68	[69]
DEA(l)	296	0.2171	[68]	303	2.45	[70]
TEA(l)	297	0.1884	[68]	303	2.42	[70]

Typically, liquids possess higher thermal conductivity than gases due to their higher molecular densities. However, notable exceptions include H₂ and He, which both have thermal conductivities on the same order of magnitude of liquids. Thermal conductivities are a function of temperature and pressure. Typically, solids experience an increase in thermal conductivity with temperature, whereas liquids and gases experience a decrease. Heat capacities are also typically higher for liquids than gases. A condensed phase system will typically have higher thermal conductivities and heat capacities around the catalyst surface and can more readily remove heat than a pure gas system. In the case of hydrogenation reactions, where the atmosphere is nearly pure H₂, high thermal conductivity and heat capacity are present. However, due to the small amount of H₂ mass present, heat transfer does not readily occur from H₂ flow relative to condensed phase heat removal. Both heat capacities and thermal conductivities are functions of temperature, with the typical trend being a decrease in both with increasing temperature. It is also important to note that often higher heat capacities require increased heating requirements. When designing a reaction system, solvent choice plays a role in reaction design as well as upstream/downstream processes. In reactive capture systems, such as aqueous MEA in carbon capture, changing heat capacities can occur as a function of CO₂ loading. Weiland et al. [71] demonstrated that a 30%_wt MEA in H₂O shows a decrease in heat capacity from 3.734 to 3.359 J g⁻¹ K⁻¹ from 0%_mol to 50%_mol CO₂ loading. Despite the decrease in thermal conductivity upon loading, a significant challenge with using aqueous amines for carbon capture is the large energy loss in regenerating the solvent. Due to the significant heat capacity difference between water and MEA (4.19 and 2.68 J g⁻¹ K⁻¹, respectively), a water rich system requires more energy to regenerate on top of MEA's relatively large enthalpy upon binding with CO₂ relative to other amines [72, 73]. As previously mentioned, many solvent parameters change as CO₂ undergoes reactive separation in an IC³M system. As the reaction coordinate increases, the viscosity of the capture agent may decrease by facilitating mass transfer, potentially increasing reaction rates if the system is mass transfer limited. Parallel asymmetries can also occur with regards to the thermal properties of the solvent. A large gap in the current IC³M literature is how the changing thermal conductive and heat capacities affect the catalytic reaction. If the capture agent's thermal conductivity decreases down the length of the reactor (in the direction of conversion), then there may be problems with establishing a homogeneous temperature profile throughout the catalyst bed or hot spots that threaten the integrity of the capture agent. How the physical and thermodynamic properties of capture agents change as CO₂ is separated remains unexplored and ripe for questioning.

2.3. Kinetic and Thermodynamic Considerations in Condensed Phase Reactions

Kinetics is as central to catalysis as material design. It is rare that the series of elementary steps constituting an overall reaction is known, which convolutes the prediction of key kinetic parameters. For instance, take a gas phase surface reaction with an ideal gas

$$\begin{array}{c}A\stackrel{K_{\mathit{ads}}}{\leftrightarrow }{A}_{\mathit{ads}}\stackrel{K_{\vartheta }}{\leftrightarrow }{A}^{\vartheta }\stackrel{k_B}{\to }B\end{array}$$ (9)

where $A$ is the reagent in the bulk phase, $A_{ads}$ is the reagent adsorbed on the catalyst surface (associative adsorption), $A^{\vartheta }$ is the activated complex, and B is the product. Further, the assumptions that $A_{ads}$ is in equilibrium with the bulk phase and that $A_{ads}$ is the first step in the surface reaction and in is equilibrium with $A^{\vartheta }$ are taken. Assuming each step is elementary, where [i] denotes the concentration of species and i, and $r_x$ and $K_x$ are the rate of formation and equilibrium constant of the relevant steps, respectively, transition state theory (TST) would show

$$\begin{array}{c}{r}_{\mathit{TST}}={\mathit{\upsilon K}}_{\vartheta }\left[A_{\mathit{ads}}\right]=\upsilon K_{\vartheta }{K}_{\mathit{ads}}\left[A\right]\end{array}$$ (10)

Here $\upsilon$ is the frequency of vibration, or the ratio of the product of the Boltzmann constant and temperature to Plank constant. The above derivation is similar to one performed by Li et al. [74]. For a solvated surface reaction, with the form of Equation (9), the interactions between the reacting species and solvent are accounted for in activity coefficients (${\gamma }_i$), related to the activity of the species ($a_i$) via the equality

$$\begin{array}{c}{a}_i=\left[i\right]{\gamma }_i\end{array}$$ (11)

where ${\gamma }_i$ is directly related to the excess Gibbs energy

$$\begin{array}{c}{\gamma }_{\mathrm{i}}=\exp \left(\frac{G_i^E}{\mathit{RT}}\right)\end{array}$$ (12)

Here $G_i^E$ is the excess Gibbs energy of species i and R is the ideal gas law constant [75]. The excess Gibbs free energy is the difference between the actual Gibbs energy and the Gibbs energy if the solution behaved ideally at the same temperature and pressure. Therefore, the activity coefficient is directly related to the aggregate of nonideal interactions. As a heuristic rule, when solvent–reagent interactions differ in magnitude and nature from solvent–solvent interactions, $G_i^E$ and ${\gamma }_i$ increase. Accounting for the interactions between solvent and reagent species, Li shows that for reaction 11

$$\begin{array}{c}{r}_{\mathit{TST}}=\mathrm{\upsilon }{K}_{\vartheta }{K}_{\mathit{ads}}\left(\frac{{\mathrm{\gamma }}_{\mathrm{A}}\left[\mathrm{A}\right]}{{\gamma }_{\mathrm{\vartheta }}}\right)\end{array}$$ (13)

An important immediate deviation from the ideal case is that the rate of the reaction is not strictly proportional to concentration of the reagent in the bulk [74]. With this, one may be eager to determine the activity coefficients of the desired reagent and solvent. However, these coefficients can be difficult to find experimentally. Instead, methods such as UNIFAC (Universal Quasi‐Chemical Functional‐group Activity Coefficients [76]) can be used to calculate an approximation of these activity coefficients. The UNIFAC method decomposes the activity coefficient into a combinatorial and a residual for the ith species in a multicomponent mixture

$$\begin{array}{c}\ln {\gamma }_i=\ln {{\gamma }_i}^C+\ln {{\gamma }_i}^R\end{array}$$ (14)

with each of these components equal to

$$\begin{array}{c}\ln {{\gamma }_{\mathrm{i}}}^C=\ln \frac{{\varphi }_i}{x_i}+\frac{z}{2}{q}_i\ln \frac{{\theta }_i}{{\varphi }_i}+l_i-\frac{{\varphi }_i}{x_i}\hspace{0.17em}\sum _j\hspace{0.17em}{x}_j{l}_j\end{array}$$ (15)

$$\begin{array}{c}\ln {{\gamma }_i}^R=\hspace{0.17em}\sum _k\hspace{0.17em}{\nu }_k^i[\ln {\text{Γ}}_k-\ln {\text{Γ}}_k^i]\end{array}$$ (16)

The combinatorial accounts for differences in molecular size and shape, while the residual factors in effects from differences in intermolecular forces of attraction. The components of the combinatorial are as follows [77]

$$\begin{array}{c}{l}_i=\frac{\mathrm{z}}{2}(r_i-q_i)-(r_i-1)\end{array}$$ (17)

$$\begin{array}{c}{\theta }_i=\frac{q_i{x}_i}{\hspace{0.17em}\mathit{\sum }_j\hspace{0.17em}{q}_j{x}_j}\end{array}$$ (18)

$$\begin{array}{c}{\varphi }_i=\frac{r_i{x}_i}{\hspace{0.17em}\mathit{\sum }_j\hspace{0.17em}{r}_j{x}_{\mathrm{j}}}\end{array}$$ (19)

where $l_i$ is a quantity related to the shape and size of species i and ${\theta }_i$ and ${\varphi }_i$ are the area and segment fraction, respectively, of species i. $z$ is the coordination number, taken as 10, and $q_i$ and $r_i$ are the molecular surface area and volume. After analysis, one can see that the combinatorial component of the activity coefficient for a component will increase with increasing surface areas and volumes of the species. This is intuitive, as the larger a molecule, the more likely that it will collide and interact with neighboring molecules, decreasing ideality. The residual, on the other hand, is assessed on a per functional group basis. For k group on i species, the group residual activity coefficient, ${\text{Γ}}_k,$ is expressed as

$$\begin{array}{c}\ln {\text{Γ}}_k=Q_k[1-\ln \hspace{0.17em}\sum _m\hspace{0.17em}{\theta }_m{\text{Ψ}}_{\mathrm{mk}}-\hspace{0.17em}\sum _{\mathrm{m}}\hspace{0.17em}\left(\frac{{\theta }_{\mathrm{m}}{\text{Ψ}}_{\mathrm{mk}}}{\hspace{0.17em}\sum _{\mathrm{n}}\hspace{0.17em}{\theta }_{\mathrm{n}}{\text{Ψ}}_{\mathrm{nk}}}\right)]\end{array}$$ (20)

where for group m of type k, $Q_k$ is the total group contribution surface area, and ${\text{Ψ}}_{mk}$ is the group interaction parameter, which increases with temperature. Returning to Equation (16) ${\nu }_k^i$ is the number of groups of type k on molecule i and ${\text{Γ}}_k^i$ is found in the same manner as ${\text{Γ}}_k$, except that x _i is set to 1. ${\text{Γ}}_k^i$ is meant to normalize the contributions of interactions to the activity coefficient by using a solution of a pure substance as a reference state. The result of analyzing Equation (22) is rather intuitive, in that molecules with many highly interacting substituent groups have deviations from ideal behavior. These interactions increase the value of the Gibbs excess energy, with increased activity coefficients observed. When thinking of catalysis, and applying the above discussion to Equation (19), one can see that in scenarios where the reagent in the bulk and the activated complex have equivalent Gibbs excess energy or rather are solvated to the same extent by the solvent, the rate of reaction is proportional to the concentration of the reagent in the bulk phase. However, in situations where the activated complex has significantly different interactions with the solvent, one can expect significant deviations from ideal kinetic behavior. Due to this effect of solvation on the reaction rate, solvent choice can lower the apparent Gibbs energy barrier [74]. Through stabilizing the transition state and destabilizing the absorbed bulk reagent, reactions can be driven forward via lowering the energy barrier.

There exists, however, a large white space in the literature directly correlating solvent effects to perceived reaction rates. The understanding of how key solvent parameters affect reaction rates, catalyst stability, and product selectivity is paramount to intelligently and purposely evolving the field of condensed phase heterogeneous catalysis. From the preceding sections, the role of dielectric constant (ε), Kamlett and Taft (KT) α, KT β, KT π*, Henrys Law constant (k_H), density (ρ), viscosity (µ), and molecular diameters are some of many possible critical parameters to monitor to understand solvent effects in the condensed phase. Solvent polarity is a key determinant in whether a solvent will stabilize a given species and therefore plays a direct role in determining the excess Gibbs energy and activity coefficients. Defined as the ratio of the absolute permittivity of a substance to the absolute permittivity of free space [78], the dielectric constant is classically used as a measure of solvent polarity and is a physical property that measures a solvent's ability to separate electrolytes into ions [79]. Similarly, Abraham, Taft, and Kamlett [80], developed the KT π* parameter as an index of the ability of a solvent to stabilize a charge or dipole. Other KT parameters include KT α and Kt β, which measure hydrogen bond donation and acceptance abilities, respectively. Recently, Aryafard et al. [81] tuned KT parameters by adding chloroform/dichloromethane to a hydrocarbon solvent for the methylation of the hydroxy group on N‐(2‐hydroxyethyl) pthalimide over an Ag₂O catalyst. They found that adjusting the KT α and β parameters to be close to zero and having an intermediate value of KT π* results in optimum yields.

The Henry's law constant is important for mass transfer but can also be an important kinetic parameter. In cases where the chemical potential of the reagent in the gas phase is equal to the condensed phase, Henry's law can be used to relate the condensed phase concentration to the gas phase partial pressure. In the case of equilibrium of gas and bulk condensed phase, the free energy change of a solvated reaction can be derived from gas phase concentrations, independent of solvent, as k_H includes the entropy and enthalpy of solvation [74]. It is important to note that for chemically absorbing systems, such as reactive CO₂ capture in amines, Henry's law requires modification to account for the kinetics of capture [43]. µ can pose challenges to scale up and have mass transfer implications [82, 83] that can affect the overall apparent activation energies. Finally, molecular diameter directly impacts the ability of a gas reagent to diffuse through a given solvent [84]. Any mass transfer hindrance can present itself as a lowering of the apparent activation energy, due to the lower temperature dependency of mass transfer phenomena than chemical reaction rates. It remains paramount that kinetic studies account for this and perform measurements in regimes where mass transfer is not the dominant limiting factor. In general, conversion as a function of kinetic rates is well studied. In the context of IC³M, a significant knowledge gap exists in understanding how the solvent effects kinetic parameters. Typically, the most understood solvent effect is the chemical binding of CO₂ to a chemisorbing capture agent, however, at typical CO₂ loadings the minority of the solution is the bound form. Bulk solvent effects remain unstudied in IC³M and rarely influence the design of capture agents. This paradigm will need to shift for the development of effective IC³M process.

3. On the Role of Solvent in Catalyst Stability

3.1. Overview of Decomposition Mechanisms in Heterogeneous Catalysis

Catalyst deactivation can be divided into seven phenomena: poisoning, coking (fouling), sintering, phase transformation, masking, volatilization, and leaching (Figure 2). While there are other deactivations pathways, the seven mentioned will remain the scope of this review. To rationally design a stable catalyst for condensed phase usage, it is paramount to understand how they can deactivate.

FIGURE 2.

Overview of catalytic deactivation mechanisms.

Catalyst poisons/inhibitors (typically a distinction is made based on permanency) involve a species bonding to the surface of the catalyst and either electronically or geometrically inhibiting active sites. Mitigating the damage from a poisoning environment is challenging. Practically, this is achieved by upstream treatment of the feed to remove poisoning agents prior to their entry to the catalytic bed [85]. For example, ZnO is used to remove H₂S, as sulfur is known to poison group VIII B and V A (Fe, Ru, Os, Co, Rh, Ir, Ni, Pd, Pt and Cu, Ag, Au) metals [86]. Besides sulfur, other common poisons are elements from groups V A and VI A (N, P, As, Sb and O, S, Se, Te). In gas phase heterogeneous catalysis, the most notorious poison to noble metals is CO [87, 88]. In condensed phase processes, a critical part of solvent selection is ensuring that the solvent itself does not act as a poison. While using a Pt‐promoted sulfated zirconia alumina catalyst (PSZA) for n‐hexane isomerization, Zhou et al. [89] found that water bound to surface Zr^δ+ sites forming Zr‐OH and leading to rapid deactivation of the catalyst. While this study was conducted in the gas phase, it is reasonable to expect similar chemistries in the condensed phase.

Importantly, in the condensed phase solvents themselves can act as poisons, such as the irreversible binding of water to active sites [89, 90]. In the field of biomass refining, sulfur containing species in the feed are well‐established catalyst poisons [91, 92]. Organic sulfur compounds such as dimethyl sulfoxide (DMSO) can act as a poison in the condensed phase. Turkin et al. [93] showed that the presence of DMSO during the hydrogenation of 5‐hydroxymethylfurfural shifts product selectivity from furan ring hydrogenation to ring unsaturated products due to the formation of sulfide and thiol groups on the Pt, Rh, Ir, and Ru catalysts. Aside from outright avoidance of poisons entering the reaction system, a key catalyst parameter in the prevention of poisoning is wettability [94]. Ni catalysts have an established ability to hydrogenate in the presence of an organic solvent but suffer stability issues in water. To address this, Lin et al. [95] coated Ni/TiO₂ with a layer of carbon for the water‐mediated hydrogenation of nitrobenzene. This tuning of surface wettability led to both activity and stability increase. Similarly, Jia et al. [96] tuned the wettability of a Ru/TiO₂ using C for use in a biphasic, one‐pot, condensed phase system. The Ru/C‐TiO₂ catalyst showed good activity and stability converting fatty acids to alkanes at 180°C.

Substrates do not actually need to chemically bind to the surface to diminish activity. Masking occurs after the deposition of species on a catalytic surface, causing physical blockage of active sites and pores. Coking is a common type of masking where carbonaceous species form on the catalyst surface, usually driven by side reactions. These species build up as a residue and physically cover the active sites of the catalyst, leading to deactivation [86]. Often a distinction is made between carbon and coke buildup. Typically, carbon is the result of CO disproportionation, where CO decomposes into C and CO₂, and coke is the result of the decomposition or buildup of hydrocarbons [97]. In the gas phase, coking can dramatically affect porous materials via clogging, seen in microporous zeolites during the methanol to olefins (MTO) reaction [98]. To address the problem of the rapid deactivation of the MTO active SAPO‐34 zeolite, Castellanos‐Beltran et al. [99] used an extrusion synthesis technique to create a mesoporous version of the typically microporous SAPO‐34. Their approach resulted in a catalyst with improved coking resistance and high activity. The abundance of hydrocarbons and hydrocarbon‐like materials present in the conversion of biomass makes these systems particularly prone to coking. The formation of deactivating carbonaceous species for the hydrodeoxygenation of bio‐oils has been a keystone challenge in the field [100, 101]. A stand‐out solution to this issue is the incorporation of Pt into catalytic materials, which slows the rate of coke formation by promoting hydrocracking of coke precursors [102, 103]. It is established that this phenomenon occurs within the condensed phase via similar mechanisms [104].

Metal nanoparticles tend to grow larger at elevated temperatures in a process known as sintering, as particles with a smaller surface area to volume ratio become increasingly stable [105]. A common gas phase CO₂ to methanol catalyst, Cu/ZnO/Al₂O₃ (CZA), is subject to poisoning and sintering at typical reaction conditions [106]. This leads to deactivation as the total exposed surface area of the metal is reduced, decreasing the collision rate of active metal and reagent. Three main mechanisms of sintering have been categorized in the literature: crystallite migration, atomic migration, and vapor transport [107]. In the first two, crystallites or single atoms of the metal particle transport across the supported surface. In vapor transport metal atoms are vaporized and redeposited on a growing crystallite. A common pitfall when dealing with sintering is the belief that sintering is solely a function of temperature and metal species. The rate and extent in which sintering occurs depends on many factors: temperature, atmosphere, metal species, support identity, pore size, and the presence of promoters. Water has been found to exacerbate sintering and lead to increased metal species oxidation in the case of Co in the Fischer–Tropsch reaction [108]. In the condensed phase, phenomena such as hydrothermal sintering remain a challenge [109]. To address this, Walter et al. [110] found that by adding high dipole moment additives (such as ethylene glycol, PEG‐400, and propylene carbonate) to the CZA catalyst surface during solvent assisted methanol synthesis inhibited the growth of ZnO crystallites. A variety of methods have been applied to limit effect of sintering. Sintering can be inhibited via strong metal support interactions (SMSIs) [111, 112]. SMSIs were first reported in 1978 when large changes in the chemisorption properties of TiO₂ supported group VIII metals were observed [113], where the support is strongly attracted to and encapsulates the metal nanoparticle [114, 115]. Recently, SMSIs have been used to stabilize metals on support surfaces to inhibit sintering phenomena [116]. The strong electronic interaction between the support and active phase changes the nature of the active site [117], meaning that SMSIs cannot be retrofitted with a new support and be expected to exhibit the same activity. Additionally, zeolite frameworks have been used to fix active metal species and prevent the accumulation of particles [118, 119].

Phase transformation is self‐defining and refers to the phenomenon where the crystalline nature of the catalyst changes when exposed to reaction conditions. Zhou et al. [89] determined that not only was Zr^δ+ poisoned by water, but also that water initiated a phase transformation converting S₂O₇ ^{2 −} to SO₄ ^{2 −} and further hindering the activity of PSZA. Water has also been found to facilitate the formation of ZnCO₃ on CZA [120]. While unwanted phase transformation can be a problem for both the gas and condensed phase, solvent induced phase change can also help create desired materials [121, 122]. By varying the solvent environment, different solid phases are stabilized. Therefore, catalysts used in a condensed reaction environment can undergo different phase and structure changes to those observed in the same catalyst in the gas phase. Volatilization is a deactivation mechanism by which metal atoms leave a catalyst surface by gas phase transport. This can occur during direct vaporization of the metal species, typically negligible at standard catalytic conditions. Another, far more common vaporization mechanism is a metal atom binding to a ligand, forming a more easily vaporized species. For example, the more volatile ruthenium carbonyl can form on Ru/Al₂O₃ [123]_. Leaching can be thought of as the liquid mediated “volatilization” and is discussed in further detail in future sections. While there are increased challenges with catalyst characterization in the presence of a solvent, many cutting‐edge techniques allow for deeper exploration of deactivation mechanisms [124]. The low typical temperatures of condensed phase reactions mean that sintering and volatilization due to thermal effects can heuristically be neglected. However, one must ensure that the introduction of solvent does not aggravate any of the above listed deactivation mechanisms. Sintering and volatilization can occur in the condensed phase, however, the mechanism by which solvents aggravate or dispel these decomposition pathways is not well understood in IC³M. Current methodology fails to assess the effects of bulk solvent effects on catalytic activity in the composite system. A chief decomposition concern is leaching as hetero‐atom‐inclusive organics can famously bind to the metals commonly used for hydrogenation, potentially forming nonstable species.

3.2. Condensed Phase Catalyst Decomposition

In gas phase heterogeneous catalysis, the interactions between the catalyst and gas phase reagents govern the products, rate of the reaction, and the rate of degradation of the catalyst. The addition of solvent into the system makes a new dimension of interactions available. Leaching requires the existence of a liquid phase, making it entirely unique to the condensed phase. Its gas‐phase parallel, volatilization, was discussed briefly in preceding sections. Sádaba et al. [125] created a thorough review regarding leaching as a deactivation mechanism in biomass conversion reactions. In this review, two modes of deactivation via leaching are presented: (1) direct solubilization and (2) chemical transformations. In direct solubilization, the materials present on the catalyst are readily dissolved by the reaction media. Many metal oxides used as catalytic supports are readily soluble in water [126], making their structure subject to deterioration in the presence of polar solvents. On the other hand, solvent‐influenced chemical transformations, such as the creation of hydroxyl groups, can increase the solidity of a metal–ligand complex. The leaching of metal atoms, in short, is directly based on the reaction medium's pH, oxidation potential, and chelating properties [127].

A large gap in the literature exists about dedicated studies on leaching in the condensed phase, despite many reports of its occurrence. There is need for a single cohesive work that compiles these observed effects. While leaching exists in catalysis, it is primarily a separation concept. Significant knowledge exists in separation and metallurgy that can be applied to condensed phase heterogeneous catalysis. For example, Dai et al. [128] studied the leaching behavior of V, Cr, Ni, Cu, Zn, and As while developing methods for recovering metals from spent selective catalytic reduction (SCR) catalysts. The leaching rate of each metal depended on the pH of the leaching fluid, where Cr, Ni, Cu, and Zn leached faster in acidic media, whereas V and As leached in both acidic and basic (pH < 3 or pH > 11). In similar study, Mishra et al. [129] found that 90% of Mo–Ni–V could be recovered from a spent petroleum catalyst through leaching with an acidic liquor followed by an ammonium carbonate wash. The study also shows that Mo and V were extracted at low pH. Mo has a higher extraction rate at lower pHs and V has higher extraction efficiencies at >2 pH. Al, Ni, and Fe were not extracted until the pH reached 3.5, requiring an acidic medium. Mishra proposes the following equilibrium to explain the observed results

$${{\mathrm{MoO}}_2^{2+}}_{\left(\mathrm{aq}\right)}+2{\mathrm{HL}}_{\left(\mathrm{org}\right)}\leftrightarrow {\mathrm{MoO}}_2{{\mathrm{L}}_2}_{\left(\mathrm{org}\right)}+2{\mathrm{H}}^{+}$$ (21)

In this proposed chemical transformation leching mechanism, the organic ligand from the acidic leaching aching can bind to MoO _x to form a soluble complex that leaves the solid matrix. Both Mishra and Dai show that pH is a very important variable in the rate of leaching. This is intuitive as deprotonation of the ligand is a required step for ligand coordination to the metal oxide. The importance of pH is supported by Marafi et al. [130] who studied the leaching of Mo, V, and Ni from hydroprocessing catalysts with inorganic and organic acids. They found that the dissociation constant of an acid is an important factor than influence leaching rate. Interestingly, the inorganic acid used (H₂SO₄) leached metals faster than the organic acid (citric acid), despite H₂SO₄ having a greater dissociation in water than citric acid. This demonstrates that the rate and totality of deprotonation are not the only factors that influence the rate of leaching, but that the ease of complex formation and the solubility of the metal complex in the media must be considered [130]. They found thatthe oxidic forms of metals leach more easily than sulfided metals, explaining the results of this work [131].

In a recent review, Kolbadinejad and Ghaemi [132] compiled reports of platinum extraction from spent catalysts. As platinum commonly used in hydrogenation, it would be valuable to know the conditions under which it leaves the metal matrix. Regarding hydrothermal leaching of Pt, Kolbadinejad and Ghaemi compiled 13 cases in which Pt leaches at acidic conditions from temperatures of 25°C–300°C. In many cases, Cl⁻ ions acted as an oxidizing agent, bonding to Pt and forming water soluble platinum chloride [133]. It is important to note that in the context of condensed phase reaction chemistry water's pH decreases with elevated temperature and systems containing CO₂ and low temperature water can possess a pH ∼3–4 [134]. It is critical that in the condensed phase, all factors are taken into consideration when attempting to design novel systems. Performing water mediated CO₂ hydrogenation at elevated temperatures using a Pt based catalyst would create an excellent environment for leaching. Organic solvents with heteroatoms that can complex with the metal species are more likely to draw these species out into the bulk phase. Additionally, leaching occurs along with other decomposition issues noted in the gas phase. Cases of leaching in basic media are present, such as aqueous ammonia dissolving silica and reforming Ti in zeolites [135], but basic conditions in heterogeneously catalyzed condensed phase hydrogenation are currently a literature white space. Many efforts have been made to adapt gas phase CO₂ hydrogenation catalysts to work in established CO₂ capture agents. We argue that the inherent issue with this approach is that the capture agents were not designed to be stable under catalytic hydrogenation conditions and that the gas‐phase catalysts were not designed to be stable in the typically basic conditions provided by a capture agent mediated reaction. Therefore, finding stable systems using this screening process makes researchers reliant on chance. We propose that systematic studies of solvent effects and the stability of capture agents in reductive catalytic conditions will yield information that can streamline the development of IC³M. While much information regarding catalyst leaching can be obtained from the field of metallurgical catalyst recovery, a literature gap exists in leaching in the basic catalytic condensed environments relevant to the IC³M process.

4. Utilizations and Lessons of Condensed Phase Hydrogenation

4.1. Overview

Condensed phase hydrogenation has been established as a valid method of production for a variety of materials. While the processes that utilize condensed phase catalytic hydrogenation chemistry have significant commonalities, there are key differences that need to be addressed on a case‐by‐case basis. Far too often, prior art solves system optimization by holding a working gas phase catalyst fixed and screening solvents to identify those that work with the catalyst. The catalyst and solvent permutations that work are carried on to optimization studies, while systems that fail are benched and not further explored or understood. There is much information to be gained in the “negative space” of the picture; a solvent may poison a catalyst leading to lower activity toward a desired product but increase the selectivity to another reaction. While this may serve as a starting point, it fails to account for any new interactions between the catalyst and solvent present in the system. These effects may be synergistic in nature leading to increased yields, they may modify the selectivity of the reaction, or they may be detrimental. The introduction of a solvent adds a third axis to a previously two‐dimensional coordinate system, providing potential for both improvement and deterioration. Examples of these phenomena are outlined throughout this section.

The condensed phase is complex. Pioneering new condensed phase chemistries or even adapting gas phase chemistries to a liquid medium is a daunting task. While liquid phase CO₂ hydrogenation in the context of integrated capture and conversion is a frontier in this field, prior art shows pitfalls and success stories of condensed phase hydrogenation. Many previous applications of condensed phase hydrogenation (Table 3) can provide insight into the development of new hydrogenation systems. Specifically, key solvent parameters, such as acidity/basicity, hydrogen solubility and proticity/aproticity, should be studied in the context of their interactions with known catalysts. However, drawing generalized conclusions is an almost insurmountable task. The increased complexity of the system means that the catalyst and solvent should be treated as “symbiotes in the same ecosystem” rather than two individual components. There is no single parameter that can be used to predict condensed phase activity, with mass transfer, heat transfer, kinetics, and thermodynamics all affected by the relationship of catalyst and solvent.

TABLE 3.

Summary of condensed phase catalytic hydrogenation reactions.

Reaction type	Reaction description	Example	Catalystsx	Application
Alkenes and Alkynes	Conversion of C=C and C≡C bonds to saturated hydrocarbons	Acetylene to ethylene [136]	Supported noble metals [136]	Petrochemical refining [137], edible oil processing [138]
Aromatic compounds	Polysaturated aromatic rings to cylcoalkenes and cycloalkanes	Benzene to cyclohexane [139]	Raney nickel [140], Pt [141], Pd [142], Rh [143], Ru [139]	Nylon precursor production [144, 145], fuel additives [146]
Carbonyl and oxygenated compounds	Aldehydes and ketones to alcohols	Acetone to isopropanol [147]	Ru [96], Pd, Rh [148], or Ni [149]	Fine chemicals, pharmaceutical intermediates [150]
Reductive amination	Aldehydes and ketones react with an amine and H2 to form a new amine	Benzaldehyde and ammonia react to form benzylamine [151]	Fe, Ru, Co, Rh, Ir, Ni, Pd, Pt, Cu, Ag, Au [151]	Drug synthesis, agrochemicals [151]
CO2 hydrogenation	CO2 to methanol, formic acid, and C2+ hydrocarbons	CO2 to methanol [152]	Cu/ZnO [15], supported Pd [152], and Pt [153]	Green chemistry, carbon capture and utilization [15]

4.2. Hydrogenation of Alkenes, Alkynes, and Aromatics

Perhaps the most classic example of hydrogenation in heterogeneous catalysis is the use of metal active phases for the reduction of alkenes and alkynes [154]. The presence of unsaturated bonds lessens the oxidative stability of produced biofuels, leading to the use of hydrogenation to reduce the number of double bonds in unsaturated fatty acids [155]. The interconversion of triple and double bonds can also be used as a hydrogen storage method [156]. Hou et al. [157] demonstrated enhanced selectivity to ethylene when acetylene hydrogenation was performed in N‐methyl‐2‐pyrrolidone (NMP) compared to the gas phase. By contrast, when the reaction was performed in decane, selectivities were similar to that of the gas phase. This observation was attributed to the fact that in NMP, acetylene is more soluble than ethylene. This is true in the tested temperature range of 40°C–100°C, with the molar solubility of acetylene ∼9–17 times higher than that of ethylene depending on temperature. In decane, the solubility ratio favored ethylene and lacked a large temperature dependence. Recently Kang et al. [158] found similar results using a PdAg/SiO₂ catalyst to investigate solvent effects on liquid phase selective hydrogenation of acetylene in methyl amide and alkane solvents. Kang chose dipolar aprotic solvents (NMP, N, N‐dimethylformamide, 1,3‐dimethyl‐2‐imidazolidinone, and 1,3‐dimethyl‐3,4,5,6‐tetrahydro‐2(1H)‐pyrimidinone) as this class of solvents possesses high selective solubility to acetylene compared to ethylene [157, 159]. They also tested a variety of alkane solvents (dodecane, tetradecane, and hexadecane) for comparison. In all cases, the methyl amides outperformed the alkanes. All the solvents gave similar product distributions, suggesting that the solvent plays a minimal role in the reaction mechanism.

Similarly, Huang et al. [136] performed hydrogenation of acetylene using Pd nanoparticles in N, N‐dimethylformamide. Good activity was obtained, with 90% selectivity to ethylene and ∼80% acetylene conversion. Interestingly, Huang tested a series of reactions using sodium borohydride (NaBH₄) as a hydrogen source and achieved similar conversions and selectivities to the pure hydrogen feedstock. However, this was only viable for a single reaction, as catalyst activity rapidly diminished by the 5th cycle with conversion dropping from ∼80% to below 40%. The authors show energy‐dispersive X‐ray (EDX) analysis demonstrating the coexistence of sodium in the agglomerated Pd nanoparticles. However, the more likely culprit for the poisoning of palladium is the BH _x species [160, 161]. The low energy X‐rays of boron make it difficult to detect on unspecialized EDX equipment, which may explain the potential oversight.

The reversible hydrogenation of aromatic compounds has been proposed as a potential approach for two‐way liquid phase hydrogen carriers [162]. Further, the hydrogenation of aromatic compounds occurs frequently in hydro refining crude oil and in petrochemical production. Much work has been done regarding the liquid phase hydrogenation of benzene [142, 163, 164]. Struijk and Scholten [139] used an unsupported Ru catalyst for the partial reduction of benzene to cyclohexene. Compared to the gas phase, they found that diluting the benzene in n‐hexane increased the selectivity to cyclohexene, achieving a maximum yield of just 2%. Performing the reaction methanol increases this maximum yield further to 8%. The time to total conversion significantly differed in the two solvents, with n‐hexane requiring ∼450 min compared to ∼1800 min in methanol. Similar phenomena were observed in the gas phase by Struijk and Scholten [139, 165, 166], serving as an early case of selective solvent poisoning for enhanced selectivity. As cyclohexene is a hydrogenation intermediate between benzene and cyclohexane, competitively absorbing a solvent on the surface diminishes the rate at which cyclohexene is reabsorbed and further hydrogenated. The decreased rate leads to enhanced selectivities in the liquid phase as opposed to the gas phase. Observation of the cyclohexene intermediate in gas phase reactions provides evidence that benzene is first hydrogenated to cyclohexene, which can be desorbed from the surface of the catalyst. While not discussed, it is possible that the methanol used as a solvent was first converted to CO. The CO could in turn poison the Ru catalyst more acutely than the steric coverage of the n‐hexane, explaining the activity difference. A possible method of increasing selectivity to cyclohexene is using a weaker benzene hydrogenation metal such as Pd, Ni, or Co—catalyst activity for benzene hydrogenation has been reported as Rh > Ru >> Pt > Pd >> Ni > Co [167].

In similar work, Takagi et al. [168] hydrogenated toluene, phenol, benzyl alcohol, and benzoic acid in alcohols using Ru/Al₂O₃ and Pt/Al₂O₃ catalysts. Seemingly contradicting the work of Struijk and Scholten, the Ru based catalyst had similar conversion results hydrogenating benzyl alcohol in hexane and methanol (∼75–78 mol%). However, the dominant product in methanol was toluene, while the dominant product in hexane was cyclohexanecarbaldehyde. In both solvents, toluene and cyclohexanecarbadlehyde were the primary products and only slightly dominated the mixture. A key feature of this study was its use of a solvent parameter to guide experimentation. Solvents were chosen based on a “δ” value, where

$$\begin{array}{c}\mathrm{\delta }=\mathrm{DN}-\mathrm{AN}\end{array}$$ (22)

The donor number (DN) and acceptor number (AN) were established originally by Gutmann as empirical parameters to measure solvents’ electron donor and acceptor properties [169]. DN is a measure of the Lewis basicity of a solvent and is measured by the enthalpy of the tested solvent reacting with the Lewis acid SbCl₅. AN reflects the Lewis Acidity and is based on the ³¹P NMR shift of triethylphosphine oxide in a solvent. Therefore, solvents with a positive δ are associated with having a higher electron donating ability relative to their accepting ability or are more Lewis basic. Conversely, a negative δ indicates higher Lewis acidity relative to a solvent's basicity. A key finding from the work of Takagi et al. was that solvents with negative δ values (methanol, ethanol, acetic acid) did not affect the hydrogenation of benzyl alcohol over Ru. However, solvents with positive δ did play a role, showing decreased activity for hydrogenation. Systems with increased Lewis basicity were found to be detrimental to hydrogenation. Additionally, Takagi et al. demonstrated that in three tested solvents (ethanol, acetone, and tetrahydrofuran) tuning the relative permittivity of the system had drastic effects on the hydrogenation conversion of benzyl alcohol. The relative permittivity for ethanol, acetone, and tetrahydrofuran is 24.3, 20.7, and 7.4, respectively, and the hydrogenation conversion of benzyl alcohol followed the same order as the permittivity trend (ethanol > acetone > tetrahydrofuran). Reactions were conducted in each solvent, adding acetic acid (relative permittivity = 6.2) and formic acid (relative permittivity = 58.0). The hydrogenation conversion increased with the addition of acetic acid and decreased for reactions with formic acid, relative to conversion in solvent without additives. Similar findings have also been reported by Khodadadi‐Moghaddam et al. [170], who found a positive, logarithmic relationship between relative permittivity and the reaction rate constant for cyclohexene hydrogenation over a Pt/Al₂O₃ catalyst. It has been established that solvents play a key role in the manipulation of reaction outcomes and catalyst stability. Most studies involving solvent effects in catalysis focus on polarity, typically using relative permittivity as the guiding metric. However, polarity rarely exists as a single dimension scaler. Manipulating polarity requires molecular change or a shift in electron density, the latter of which is rarely permeant and may be impractical in a reaction environment. Adjusting the polarity of organic solvents comes in the form of moiety addition. Many studies that find correlations between catalytic activity and “polarity” fail to address changes made in other parameters such as KT α, KT β, and KT π*. Even simpler are the changes in pH, viscosity, and thermal stability that occur with the addition of substituent groups. We argue that these parameters work in a web of influence and that isolating a single variable such as polarity is rarely done with extreme prejudice, instead catching a range of related solvent qualities. We believe that this is the reason a literature gap exists and explains conflicts in the field exist regarding a given solvent's ability to enhance a reaction.

4.3. Hydrogenation of Carbonyl and Oxygenated Compounds

Following fast pyrolysis or liquefaction, biomass feeds can contain oxygenates such as carboxylic acids, aldehydes, ketones, and phenolic compounds which have low energy densities compared to conventional high‐energy fuels [171, 172, 173]. A growing number of studies focused on solvent effects relevant to the deoxygenation and reduction of these compounds have given rise to better understanding of the interactions between solvents and catalysts. Using ab initio molecular dynamics and density functional theory (DFT) calculations, Cao et al. [174] demonstrated a synergistic effect between oxygen vacancies found on TiO₂ and liquid phase water molecules in the catalytic reduction of aldehydes. This was done by creating two models of a Pt/TiO₂ catalyst, one with oxygen vacancies (Pt/TiO_2‐x) and one without oxygen vacancies (Pt/TiO₂). Cao found that the Pt clusters function as an electron reservoir for both the oxygen vacancy and the solvent water. Relative to the gas phase, the Pt cluster on TiO₂ experienced an average charge decrease of 0.26 $|e|$ and 0.30 $|e|$ for Pt/TiO₂ and Pt/TiO_2‐x, respectively, in the presence of liquid phase water. Additionally, the liquid phase water was found to have the ability to spontaneously protonate the O in the aldehyde group via acid base exchange, generating OH* at the metal–support interface.

The observations made by Cao are reminiscent of catalytic transfer hydrogenation (CTH). In CTH, organic molecules donate their hydrogen to reagents and reduce them [175]. This donation is done either via direct hydrogen transfer or formation of a metal hydride [176, 177]. A popular hydrogen donator used is formic acid [178] and the placement of the donated hydrogen can be tuned via thoughtful catalysis. For example, Fu et al. [179] conducted a study using Al₂O₃ supported Ni and Cu catalysts to facilitate the CTH of furfural using formic acid as a hydrogen donor. Fu found that Cu/Al₂O₃ facilitated hydrogenation and subsequent hydrogenolysis, leading to the formation of methyl furfuran from furfuran. Ni/Al₂O₃ produced a mixture of fururyl alcohol, methyl furan, and furfuran. The Cu/Al₂O₃ possessed superior selectivity to the hydrogenation of the carbonyl due to its inability to catalyze the decarbonylation reaction but suffered from poor TOF (643 h⁻¹) compared to its Ni counterpart (1099 h⁻¹). Of the catalysts screened by Fu, the 10%Ni‐10%Cu/Al₂O₃ demonstrated a beneficial synergistic effect, producing a TOF of 1408 h⁻¹ and selectivity comparable to the Cu only catalyst. Beyond catalytic design, the presence of a polar protic species in a system can clearly lead to hydrogen transfer, a fact that can be utilized in the condensed phase.

There is evidence to suggest that the presence of solvents can alter a reaction pathway via direct participation in kinetically relevant steps. The hydrogenation of fufural to tetrahyrofurfuryl alcohol can precede by either first hydrogenating the carbonyl substituent or the conjugated π‐bond system in the ring of the furan body. Zhao et al. [180] found that the pathway is solvent dependent; when performed in cyclohexane, the selectivity to tetrahydrofurfural, the intermediate formed if the furan ring π‐bonds are hydrogenated first, was almost twice that compared to when the reaction was performed in water. The reaction in water had greater than four times higher selectivity to the carbonyl‐first hydrogenated product, furfuryl alcohol, and boasted more than double the total conversion of furfural. Zhao cites the mobility of hydrogen through the first solvation shell and the bulk hydrogen bonded structure as the reason for the activity difference between the two solvents. Through DFT calculations, Zhao shows that the reaction energetics are altered in the presence of water. They propose that water can coordinate on the catalytic surface with furfuran, donate a proton through hydrogen bonding, and receive a proton from the catalytic surface. Due to the limited presence of gas phase hydrogen in the liquid phase, creative ways of delivering hydrogen to catalytic surface should be implemented to overcome mass transfer barriers. The bulk polar protic solvent provides a scaffolding for proton shuttling, facilitating the hydrogenation of surface bound species. This was observed by Chin et al [181], who reported kinetic and isotopic evidence for proton–electron transfer events in polar protic solvents for the hydrogenation of aldehydes and ketones. However, as with any solvent choice, stability of the solvent must be considered. The heteroatom that gives a molecule its polar protic status may also be ripe for hydrogenolysis, creating solvent stability concerns. Often, catalysts selectively target a C–C bond affixed to an oxygenated substituent, such as in the hydrodeoxygenation of bio‐oils.

Condensed phase catalysis has been utilized to achieve the creation of jet fuel products from increasingly renewable sources such as biomass. Among these approaches, the deoxygenation of fatty acids to produce aliphatic jet fuel components has been established as a potentially viable means of fuel additive production. The hydrogenation of esters and fatty acids (HEFA) is a mature technology, converting bio‐oil based triglycerides to aliphatic hydrocarbons. The utilization of solvent effects in the HEFA process has been reported. Xie et al. [138] studied the decarboxylation of steric acid to form heptadecane using a Pd/C catalyst. A critical finding in this study was that the hydrotreatment of steric acid performed in a biphasic system of cyclohexane and water outperformed the process done in water or cyclohexane alone. After 1 h of reaction, the yield of heptadecane from the biphasic system was triple that of the single‐phase systems. In 6 h, the biphasic tandem catalytic process (biTCP) could achieve 91.7% carbon yield of heptadecane from steric acid. Further, it was shown that the ratio of cyclohexane to water in the system played a critical role in the rate of the reaction, with optimal results existing within the range of 12:12–2:22 in the volumetric ratio of cyclohexane:water. Xie et al. [138] hypothesized that the organic and aqueous phases selectively stabilize the aliphatic tail and polar carboxylic head of the fatty acid. However, the bulk solvent structure of this system at the reaction temperature of 260°C remains an open question.

Building on the work of Xie, Jia et al. [96] used the biTCP and a Ru based catalyst for the hydrodeoxygenation of oleic acid at just 150°C. The activity of Ru/C‐TiO₂ was greater than Ru/C and Ru/TiO₂, with an optimal C:TiO₂ ratio of 3:3. Using oleic acid as a model compound, Ru/C‐TiO₂ achieved liquid product yields greater than four times that of the noncomposite Ru/C and Ru/TiO₂ at 150°C. At 200°C, the composite material was able to hydrocrack C17 and C18 products into the much more desirable C8–C16 at rates much higher than the C and TiO₂ counterparts, potentially its most exciting advantage. Again, the “solvent cage” [182] was proposed to explain the biphasic systems success, with the added acidity [174] and oxygen vacancies [183, 184] of the TiO₂ support potentially aiding the composite catalysts’ increased success. Interestingly, the catalyst wettability is altered with the addition of TiO₂. Of the catalysts tested by Jia, a range of contact angles of 110° to 32° was found, with Ru/C and Ru/TiO₂ possessing the highest and lowest contact angles, respectively. Unsurprisingly, the Ru/C‐TiO₂ with equal mass loadings of C and TiO₂ possessed an intermediate contact angle of 80°. As pointed out by Jia, metal oxide nanoparticles have been reported to stabilize oil‐in‐water emulsions, and their hydrophilic nature attracts them to the aqueous phase. However, if the particle is too hydrophilic it can become isolated in the aqueous phase, away from the increased reagent concentrations in the organic phase. The work of Xie and Jia demonstrates a harmony of design between the catalyst, reagent, and solvent(s) rarely discussed in the literature. Of note, the Ru/C‐TiO₂ was reported to have enhanced stability, demonstrated within 5 cycles, despite reports of Ru's instability in other condensed phase HEFA processes [185, 186]. The potential role of TiO₂ in the increased catalyst stability went undiscussed. Broadly, the successful hydrogenation of heteroatom‐inclusive organic species in the condensed phase demonstrates the capability of hydrogenation catalysts to cleave at heteroatom sites. This creates a challenge in the IC³M process. Typically, the reduction of vapor pressure in designer capture agents is achieved through the addition of polar groups whose intermolecular interactions drive down volatility. Significant work needs to be done to either understand how to chemically protection these heteroatom‐inclusive moieties or find a hydrogenation‐safe method of “weighing down” capture agents.

4.4. Reductive Amination

Amines are in many pharmaceuticals, bioactive compounds, CO₂ capture agents, and important bulk and fine chemicals [151, 187, 188, 189, 190]. Reductive amination is a popular amine synthesis technique, with advantages such as mild reaction conditions, inexpensive reagents, and a variety of applicable substrates compared to other amine synthesis techniques [189, 191]. The work of Song et al. [192] highlights the large effects that choice of solvent has on the reductive amination of cyclohexanone using a commercial Ru/C catalyst. Song performed hydrogenation in a range of solvents (methanol, ethanol, isopropanol, water, tetrahydrofuran, dioxane, toluene, and cyclohexane) and noted the productivity of key products such as cyclohexylamine, cyclohexanol, cyclohexylamine, and the Schiff base N‐Cyclohexylcyclohexylideneamine. Broadly, a large solvent dependence was found, with only methanol and toluene containing the cyclohexylamine main product. While performing reductive amination using ketones and ammonia, Song reported that selectivity to imines increased in the presence of aprotic polar solvents, indicating that the hydrogenation of the imine was inhibited. Rationalizing this, Song cites Wan et al. [193] who similarly reports low hydrogenation activity from a Ru/C catalyst in aprotic polar solvents. Song and Wan both hypothesize that interactions between oxygenated solvents and the Ru active site inhibit the hydrogenation rates. Further, Song found that protic solvents were more active in promoting the reaction between ammonia and imines with ketone when compared to aprotic polar solvents. While not explicitly mentioned by Song, it is possible that proton transfer from the polar protic solvent to the imine and iminium intermediates occurs, facilitating the hydrogenation steps of the mechanism. Hydrogen shuttling would also be feasible in polar protic solvent systems, increasing the abundance of H near the desired sites. Given these findings, one would expect this reaction to perform well in water—this is not the case. Song shows that water prevents imine/Schiff base formation while promoting the undesired C = O hydrogenation. While counter to the goal of the reaction, it demonstrates elevated hydrogenation rates in water. An additional consideration is that dehydration is required for both the formation of the imine from cyclohexanone and the formation of the Schiff base intermediate from cyclohexylamine. The abundance of a product‐side species, such as water, forces the equilibrium to the reagent‐side. Supporting the importance of proton shuttling transfer hydrogenation has been reported for use in reductive amination by a homogenous catalyst [194]. Other relevant work with homogenous catalysis exists, including a computational study done by Jameel et al. [195] who found that methanol had high productivity and shallowed the potential energy surface. Riemer et al. [196] reports a selective Rh‐based homogeneous catalyst in a switchable solvent system that uses methanol and dodecane to achieve amine selectivies as high as 96.5%.

Ru‐based catalysts performed poorly at the reductive amination of furfural to furfuralamine in water, as reported by Chatterjee et al. [191]. Despite this, they found that a range of catalysts could perform the aqueous chemistry using Pd, Rh, and Pt as the active phase. In fact, Pd/C, Rh/C, and Pt/C all achieved 100% conversion (conversion for Ru/C = 25.2%). However, the dominant product across the catalysts varied significantly, with the reduced ring adduct Di(2‐tetrahydrofurylmethyl)amine for Pd/C, the adduct Bis[(furan‐2yl)methyl]amine for Rh/C, and furfural alcohol for Pt/C. For the non‐Ru‐based catalysts screened, selectivity to the desired product, furfurylamine, followed the trend Rh/Al₂O₃>Rh/MCM‐41>>Rh/C≈Pt/C>Pd/Al₂O₃≈Pt/MCM‐41≈Pd/C. Comparing the studies from Chatterjee and Song, it becomes evident that the choice of solvent and catalyst needs to be made together. The catalyst and solvent should be designed as a coherent system rather than holding one fixed and varying the other. The success of the catalyst or solvent are frequently separated in the literature, rather than recognizing catalyst‐solvent synergy as a most critical variable.

5. Application to Condensed Phase CO₂ Hydrogenation

5.1. CO₂ Hydrogenation using Gas Phase Methods

Gas phase heterogeneous catalysis has been used to convert CO₂ to valorized products. It is reasonable to expect that there will be similarities between gas and condensed phase chemistries. Therefore, it is justifiable to study the gas phase equivalent of desired condensed phase reaction systems. A broad portfolio of gas phase CO₂ hydrogenation products currently exist, including methane [197], carbon monoxide [198], formic acid [199, 200], dimethyl ether [201, 202], methanol [203, 204], C²⁺ hydrocarbons [205, 206], higher alcohols [207], and aromatics [208]. Reaction mechanisms and catalysts for the CO₂ Fischer–Tropsh process (FT), the Sabatier reaction, CO₂ to methanol (CTM), and the methanol mediated pathway (MMP) to hydrocarbons remain the most discussed methods of CO₂ utilization in the gas phase.

Developed by Paul Sabatier, the Sabatier reaction

$$\begin{array}{c}{\mathrm{CO}}_2+4{\mathrm{H}}_2\hspace{0.17em}\underset{\mathrm{Cat}.}{\to }\hspace{0.17em}{\mathrm{CH}}_4+2{\mathrm{H}}_2\mathrm{O}\end{array}$$ (23)

originally used finely divided nickel to hydrogenate CO₂ to CH₄ [209]. There has been renewed interest in this reaction, as it could be used to harness CO₂ for the creation of SNG and has been studied as a potential source of fuel by space agencies such as NASA [210]. Due to the high cost of electrolysis [211], the application of CO₂ methanation is hindered by the supply of hydrogen. Recent advancements of this reaction increasingly focus on lowering the temperature and pressures required for reaction [212], with Ru [213, 214], Co [215, 216], and Ni [217] based catalysts.

Similar to the Sabatier reaction, the partial reduction of CO₂ to CO or the RWGS,

$$\begin{array}{c}{\mathrm{CO}}_2+{\mathrm{H}}_2\hspace{0.17em}\underset{\mathrm{cat}.}{\to }\hspace{0.17em}\mathrm{CO}+{\mathrm{H}}_2\mathrm{O}\end{array}$$ (24)

often finds itself in thermodynamic competition with the partial reduction of CTM,

$$\begin{array}{c}{\mathrm{CO}}_2+3{\mathrm{H}}_2\hspace{0.17em}\underset{\mathrm{cat}.}{\to }\hspace{0.17em}{\mathrm{CH}}_3\mathrm{OH}+{\mathrm{H}}_2\mathrm{O}\end{array}$$ (25)

as CO is a key reaction intermediate of some CTM reaction pathways. CTM is an exothermic reaction (ΔH_Rxn = −49 kJ mol⁻¹ [218]) and RWGS is endothermic (ΔH_Rxn = 41 kJ mol⁻¹ [219]), leading to temperature dependent selectivities. This was modeled by Stangeland et al. [46], who used Aspen Plus to simulate the competition of these reactions in a gas phase reactor over a range of temperatures and pressures (Figure 3).

FIGURE 3.

Effect of temperature and pressure on (a) CO₂ conversion and (b) methanol selectivity at phase and chemical equilibrium. Dashed lines represent the chemical equilibrium predicted by gas phase thermodynamics. Figure and caption from Stangeland et al. [46].

Additionally, the scenario in which a condensed phase exists within the gas phase reactor was modeled and is represented by the solid lines in panel a of Figure 3. The shape of the conversion curve is explained by the endothermicity and exothermicity of the RWGS and CTM reactions, respectively. At low temperatures, CTM is favored and selectivity to methanol is dominant. As temperatures increase, selectivity to methanol decreases and the endothermic RWGS is favored. In the CTM regime there exists significant difference between the pressure lines, with the trend of higher pressures being increasingly favorable for methanol production in accordance with Le Chatelier's principle [120]. However, in the RWGS regime, the pressure lines converge, indicating the pressure independence of the RWGS product. The RWGS shift reaction is often the first step in a series of reactions constituting FT, making understanding this reaction critical for producing larger hydrocarbons and alcohols from CO₂. Methanol, on the other hand, is a valuable C₁ building block used in the synthesis of olefins, aromatics, gasoline, and other commodity chemicals [220].

Methanol creation in the gas phase is well studied, with Cu/ZnO, In₂O₃, ZnO–ZrO₂, and Pd‐based catalysts being among the frontrunners for utilization. The Cu/ZnO catalyst is used industrially for the conversion of CO₂ inclusive syngas to methanol [221, 222], and modifiers of this catalyst have revealed a variety of beneficial effects [223, 224]. A popular modification is the addition of alumina, forming a CZA catalyst, wherein the alumina improves the activity and stability of Cu/ZnO by increasing the active phase dispersion and modifying the electronic state of Cu [225]. Using a hydrotalcite‐like precursor to synthesize CZA, Zhang et al. [226] reports a methanol space time yield (STY_MeOH) of ∼21,800 mmol_MeOH kg_cat ⁻¹ h⁻¹ (H₂:CO₂ = 3:1, 50 bar, 250°C, weight hourly space velocity (WHSV) = 18,000 mL g⁻¹ h⁻¹). While studying the effect of ZnO geometry on the Cu/ZnO catalyst, Lei et al. [227] report a STY_MeOH for a Cu/ZnO‐Rod, Cu/ZnO‐filament, and Cu/ZnO‐carbonate coprecipitation (a prominent synthesis method of this material) of ∼6,500, ∼17,100, and ∼7,500 mmol_MeOH kg_cat ⁻¹ h⁻¹ (H₂:CO₂ = 3:1, 30 bar, 240°C, WHSV = 12,000 ml g_cat ⁻¹h⁻¹). For In₂O₃, the active site for CO₂ activation is broadly accepted as the oxygen vacancy, which initiates a formate pathway conversion of CO₂ to methanol [208, 228]. While studying the effect of oxygen vacancy density on methanol production, Wang et al. [229] reported an In₂O₃ catalyst that achieves a STY_MeOH of 8,400 mmol_MeOH kg_cat ⁻¹ h⁻¹ (H₂:CO₂ = 3:1, 30 bar, 330°C, WHSV = 12,000 mL g_cat ⁻¹h⁻¹) and a methanol selectivity of 72.3%, with only CO as a side product. Recently, zeolites have been used for converting methanol into hydrocarbons [230, 231]. Directly related to this study, Gao et al. [208] used a mixture of In₂O₃ and H‐ZSM5 to achieve a C₅₊ selectivity of 78.6%, with only a 1% selectivity to methane and a CO₂ conversion of 13.1% (H₂:CO₂ = 3:1, 30 bar, 340°C, WHSV = 9,000 mL g_cat ⁻¹h⁻¹). However, this process required a cofeed of CO where CO/(CO + CO₂) = 80%.

As shown in recent literature, MMP serves as a strategy for creating higher carbon materials but requires the ability to readily synthesize methanol. Congruent with the work of Gao [208], MMP is typically accomplished with a bifunctional catalyst or a tandem process where a series of chemical reactions occur following the creation of methanol [232]

$$\begin{array}{c}{\mathrm{CO}}_2+3{\mathrm{H}}_2\hspace{0.17em}\underset{\mathrm{cat}.}{\to }\hspace{0.17em}{\mathrm{CH}}_3\mathrm{OH}+{\mathrm{H}}_2\mathrm{O}\end{array}$$ (26)

$$\begin{array}{c}2{\mathrm{CH}}_3\mathrm{OH}\hspace{0.17em}\underset{\mathrm{cat}.}{\to }\hspace{0.17em}{\mathrm{CH}}_3{\mathrm{OCH}}_3+{\mathrm{H}}_2\mathrm{O}\end{array}$$ (27)

$$\begin{array}{c}{\mathrm{nCH}}_3\mathrm{OH}+{\mathrm{H}}_2\hspace{0.17em}\underset{\mathrm{cat}.}{\to }\hspace{0.17em}{\mathrm{CH}}_3{\left({\mathrm{CH}}_2\right)}_{\mathrm{n}-2}{\mathrm{CH}}_3+{\mathrm{nH}}_2\mathrm{O}\end{array}$$ (28)

$$\begin{array}{c}{\mathrm{nCH}}_3\mathrm{OH}\hspace{0.17em}\underset{\mathrm{cat}.}{\to }\hspace{0.17em}{\mathrm{CH}}_2=\mathrm{CH}{\left({\mathrm{CH}}_2\right)}_{\mathrm{n}-3}{\mathrm{CH}}_3+{\mathrm{nH}}_2\mathrm{O}\end{array}$$ (29)

with methanol to dimethyl ether (MTDE) possessing a standard reaction enthalpy of ΔH _Rxn = −24 [233]. Both methanol to paraffins [234] and MTO [235, 236], are exothermic, with reaction enthalpies that are a function of the final products. For this system of reactions, all reactions are exothermic and, with the exception of MTDE, produce less moles of gas than they consume. Therefore, all processes benefit from lower reaction temperatures and, except MTDE, higher reaction pressures.

In the case of CO₂‐FT, a typical system consists of the following reactions [232, 237]

$$\begin{array}{c}{\mathrm{nCO}}_2+(3\mathrm{n}+1){\mathrm{H}}_2\hspace{0.17em}\underset{\mathrm{cat}.}{\to }\hspace{0.17em}{\mathrm{C}}_{\mathrm{n}}{\mathrm{H}}_{2\mathrm{n}+2}+2{\mathrm{nH}}_2\mathrm{O}\end{array}$$ (30)

$$\begin{array}{c}{\mathrm{CO}}_2+{\mathrm{H}}_2 \underset{\mathrm{cat}.}{\to } \mathrm{CO}+{\mathrm{H}}_2\mathrm{O}\end{array}$$ (24)

$$\begin{array}{c}\mathrm{nCO}+(2\mathrm{n}+1){\mathrm{H}}_2\hspace{0.17em}\underset{\mathrm{cat}.}{\to }\hspace{0.17em}{\mathrm{C}}_{\mathrm{n}}{\mathrm{H}}_{2\mathrm{n}+2}+{\mathrm{nH}}_2\mathrm{O}\end{array}$$ (31)

$$\begin{array}{c}\mathrm{nCO}+2\mathrm{n}{\mathrm{H}}_2\hspace{0.17em}\underset{\mathrm{cat}.}{\to }\hspace{0.17em}{\mathrm{C}}_{\mathrm{n}}{\mathrm{H}}_{2\mathrm{n}}+{\mathrm{nH}}_2\mathrm{O}\end{array}$$ (32)

where reaction 24, the RWGS reaction, is endothermic, and the subsequent hydrogenation of CO to paraffins is exothermic [238]. While the overall reaction is exothermic [239], the literature agrees that the direct hydrogenation of CO₂ to paraffins does not occur, but the reaction proceeds by first generating CO from CO₂ via the RWGS followed by CO hydrogenation via FTS. Therefore, while higher reaction temperatures favor the generation of CO, they hinder the formation of alkanes. Regarding pressure, the RWGS reaction is independent of Le Chatelier related effects due to pressure, while the FTS is favored at high pressures. An additional challenge of CO₂‐FTS is the binding of the product distribution by the Anderson–Schulz–Flory (ASF) distribution, with some deviations [240]. The Flory–Schulz distribution is a probability distribution that describes the ratio of different chain length polymers in an ideal step‐growth process [241]. In the context of FT it is known as the ASF distribution and suggests that the product stream hydrocarbon mass fraction ($w_i$) of a carbon length i is a function of the chain growth probability α, or

$$\begin{array}{c}{\mathrm{w}}_{\mathrm{i}}={\mathrm{\alpha }}^{\mathrm{i}-1}{(1-\mathrm{\alpha })}^2\mathrm{i}\end{array}$$ (33)

When this distribution is plotted for 0 ≤ α ≥ 1 and 1 ≤ i ≥ 120, methane is a product regardless of the value of α and gasoline range products only dominate in the approximate range of 0.70 ≤ α ≥ 0.85 [242]. There are studies focused on the breaking of the ASF distribution, but it is important to remember its limitations on product selectivity in the FTS process. The gas phase serves as a no‐solvent “control,” allowing insight into reaction mechanics sans‐solvent. A rigorous study of related gas phase materials should be performed prior to using materials in the condenced phase, enabling an enhanced understanding of the role of solvent in complex condensed phase reactions.

5.2. On Reactive Capture, Reagent Modification, and Reaction Mechanism

Within the scope of IC³M arises the demand for an economic, effective, and efficient material capable of CO₂ capture (and release) under reasonable conditions. While the perfect capture agent remains elusive, it is possible to anticipate the qualities of this platonic form solvent based on current innovations. Doing this requires a brief evaluation/consolidation of current liquid phase capture agents. For the scope of this review, we focus on capture agents that are feasible as reaction mediums, mainly, aqueous amine systems (AASs), ionic liquids (ILs), and CO₂ binding organic liquids (CO₂‐BOLs). AASs are the predominant system and have been used industrially since 1930, typically employing monoethanol amine (MEA) [243]. Maina et al. [244], describes a typical carbon capture and utilization process; CO₂ containing gas, usually flue gas, is fed into an absorption column where an amine solvent is used to remove CO₂ from the stream. This CO₂ rich amine is moved to a stripper, where elevated temperatures remove CO₂ that is then purified, compressed, and fed to a packed bed reactor for conversion to valorized products. The drive for integrating capture and utilization is based on removing the separation, purification, and compression steps by directly reacting with the captured form of CO₂ with a catalyst, potentially lowering the energy and capital demands of the overall process [245].

Along with AASs, ILs have been demonstrated for use in CO₂ capture. Room temperature ILs are salts that are liquid at room temperature and are promising CO₂ sorption materials due to their typical nonvolatility, high thermal stability, and tunable properties [246, 247]. A class of chemically‐reactive ILs, known as aprotic heterocyclic anion (AHA) ILs have drawn significant attention as they possess all the benefits of conventional ILs with the ability to reversibly chemically fix CO₂ without suffering from the large viscosity increases [248] or solidification. In a typical AHA IL, the negatively charged anion, usually nitrogen, bonds with incoming CO₂ forming a carbamate intermediate. CO₂‐BOLs are typically amines, with cutting‐edge systems such as 2‐EEMPA having demonstrated low capture costs and viscosities with high capacities and rates [13, 14] (Figure 4).

FIGURE 4.

The complexity of the IC³M system: CO₂ and H₂ are present in the gas phase. Dissolved CO₂ and bound RNH₂CO₂ ⁻ engage the catalyst surface, while moving through a dense condensed media. The products, methanol and water, leave the catalytic surface.

One can anticipate that using capture agent may induce a chemical modification of the inbound reagent CO₂. Heldebrant et al. [245] pointed out that via reactive capture, CO₂ is “activated” prior to reaching the catalytic site. The carbon on linear CO₂ possesses a well shielded sp antibonding orbital. Placing electrons in this antibonding orbital is energy intensive, leading to large activation energies. Upon forming a carbamate or carbonate intermediate, though, this carbon becomes trigonal planar and adopts a sp² hybridization. This brings the p* antibonding orbital out of plane, making this now perpendicular antibonding orbital ripe for nucleophilic attack [245]. This type of modification needs to be accounted for in the design of the catalytic system, as the nature of the reagent has changed. In this case, solvent‐reagent interactions converge, and a solvent–reagent complex is now the reaction target.

Any claim toward a true integrated capture and conversion process must be able to do both capture and conversion adequately. Heldebrant et al. [245] provides a set of criteria that a solvent/capture agent must possess in order to be correctly identified as a successful candidate for CO₂ capture and conversion. (1) The material must be capable of efficient absorption of CO₂ from flue‐gas like streams. This means that they need to be able to separate sufficient CO₂ quickly enough to avoid large upscaling of process units. The maximum amount of CO₂ that any solvent can absorb is equivalent to the equilibrium partial pressure of CO₂ (P*) over the solvent at a given temperature [245]. With the desire to capture CO₂ well enough for successful separation combined with the goal of solvent regeneration, Mathias et al. [73] finds an ideal range of −65 to −85 kJ mol⁻¹ for enthalpy of solution that is strong enough to capture sufficient CO₂ while still allowing for successful solvent regeneration. (2) The material cannot be volatile, viscous, or toxic. Volatile amines are subject to substantial losses as millions of pounds of gas per hour flow through an industrial absorber [245]. Additionally, the engineering goal of keeping fugitive emissions as safe as possible should be maintained, demanding that these emissions are designed to be minimally harmful. As previously noted, increasingly viscous fluids are more expensive to transport and increasing viscosity slows the rate of absorption, leading to the need for larger process units. (3) The material must have a techno‐economic Analysis (TEA) that is competitive against a relevant reference case, such as 5 M MEA or Shell's Cansolv [245]. Additionally, the solvent needs to be tested at relevant conditions for both capture and regeneration. This means that capture must be performed with ∼1.4 kPa CO₂ at 40°C and released at 120°C and ∼2.8 bar [245] under CO₂ atmosphere. If regeneration is performed using nitrogen as a sweep, a dilution has been performed. A notable exception is the piperazine systems of Rochelle, where CO₂ is thermally compressed at 6 bar, allowing for circumnavigation of the energy intensive first stage of compression [249, 250]. Knowing these criteria allows for anticipation of solvent properties in the condensed phase, enabling enhanced design regulations for conversion systems. In short, with current technology and understanding, a catalyst designed for IC³M should be compatible with a single‐component water lean organic secondary amine with a −65 to 85 kJ mol⁻¹ enthalpy of solution. As the reaction environment consists of mostly unbound secondary amines, catalysts must be stable in basic conditions and lack the capability to cleave through the C–N bonds on the pre‐existing amine. The specific intermediate formed via amine capture is a function of the degree of substitution of the amine. 1º and 2º amines will form a carbamate intermediate upon reacting with CO₂, which can be reversibly converted to bicarbonate in the presence of water. In the presence of water, the 3º amine will accept a proton from water. The negatively charged hydroxyl group then performs a nucleophilic attack on the partial positive carbon of carbon dioxide, forming a bicarbonate intermediate. In alcohol, the 3º amine will again accept the acidic proton and the alkoxide performs a nucleophilic attack on CO₂'s partial positive C, forming an alkyl carbonate [15]. Due to the anticipation of a water‐lean secondary amine system as the capture agent of choice, catalysts for this system should be designed to react with a bound carbamate in the condensed phase rather than a gas phase CO₂ molecule. Finally, the demand for low volatility can be achieved through either the addition of polar functional groups or bulky functional groups, but the desire for low viscosity and the avoidance of solidification renders the latter unlikely. This may be beneficial as polar moieties can stabilize polar reaction intermediates, facilitating the production of polar molecules such as methanol in CTM [74].

The targeting of condensed phase CTM has been a concerted effort in recent years. However, significant understanding gaps exist about the impact of capture agents on the reaction mechanism of CTM. The gas phase reaction mechanisms for CO₂ to methanol have been discussed in the literature with 5 key pathways identified: CO associative, CO dissociative, formate via dioxymethylene (HCOO* 1), formate via hydrocarboxyl (HCOO* 2) and the controversial, but relevant, trans‐COOH* pathway (Figure 5) [204, 251, 252, 253]. Typically, the choice of catalytic material is the dominant factor in the determination of the reaction mechanism. Active sites formed by the catalyst will impact the energetics of key reaction intermediates.

FIGURE 5.

Identified CO₂ to methanol reaction pathways that are partially evidenced.

A key component of rational solvent choice in the design of condensed phase reaction systems is the solvent's ability to stabilize reaction intermediates, necessitating the study of key reaction intermediates and mechanisms [74]. For the HCOO* and CO pathways, spectroscopic evidence has been obtained confirming the existence of the CO*, COOH*, and H₃CO intermediates [251, 254]. However, it is crucial to note that all other intermediates are evidenced by DFT calculations [229, 255, 256, 257, 258, 259, 260, 261]. While the reaction mechanism is a function of the catalyst used, the trans‐COOH* pathway is of particular interest in the condensed phase, as it was found to be water dependent. DFT calculations determined that in the presence of water, the formation of the trans‐COOH* species was more energetically favorable than that of formate over a Cu(111) surface [261]. This suggests that the physical parameters of solvents can play a direct role in altering the reaction mechanism without direct reaction with the reagent. These nonchemical interactions between catalyst, reagent, and solvent lack significant study in the current literature using capture agents as reaction media. Some work exists, most notably research from Kothandaraman et al. [14, 153], where thorough studies utilizing in situ ¹³C magic angle spinning NMR provided key insights such as identifying that the deactivating N‐methylation of amines likely occurs through formamide intermediates. Again, significant understanding of the role of amines in the stabilization of reaction intermediates is lacking yet needed to advance IC³M.

5.3. CO₂ Hydrogenation in Capture Agents and Capture Agent Adjacent Media

Broadly, significant work has been performed in the realm of condensed phase CO₂ hydrogenation. Especially, methanol has been a primary target of homogenous and heterogeneous catalysts alike in the condensed phase [152]. C²⁺ products other than dimethyl ether remain a rarity in literature. Of note, He et al. [262] reports the synthesis of higher alcohols from CO₂ using a Pt/Co₃O₄ catalyst in a variety of solvents. Water was found to be beneficial for reaction rates and higher alcohol production. Interestingly, the highest selectivity to methanol (88.1%) was achieved when a mixture of water and 1,3‐dimethyl‐2‐imidozlidnone was used. ¹³CH₃OH and D₂O labeling experiments provided evidence of water's direct participation in the reaction with ethanol forming via an MMP reaction mechanism. The effect of solvent has been somewhat explored in CO₂ hydrogenation, most often in the context of CO₂ inclusive syngas. For example, Zeng et al. [263] used a Cu/ZnO catalyst to investigate the effect of alcohol on the synthesis of methanol from CO₂ inclusive syngas. It was found that methanol yield generally decreased with increasing bulkiness of the alcohol. This aligns with previously discussed transport principles—as solvent molecules become bulkier, mass transfer is hindered and limits the apparent the kinetic rates. Beyond extensive contributions to the field of IC³M, Kothandaraman produced other related studies such as the coproduction of methanol and glycol from epoxides and bound CO₂, mediated by polyethyleneimine [264], demonstrating the feasibility of the many uses of CO₂ in a catalytic condensed phase. Similarly, Kroll et al. [16] used EEMPA captured CO₂ to perform cycloaddition to propylene oxide, forming propylene carbonate at 120°C. Other examples of carbon capture and conversion include the use of a Cu/CaO dual functional materials in a calcium looping‐RWGS reaction to create CO with 100% selectivity at 650°C [265]. In related work, Cu(I)‐MOF NU‐2100, which selectively captures CO₂ and H₂, was shown to create formic acid from these bound species [18].

Earlier homogeneously catalyzed work laid the groundwork for the understanding of amine‐assisted CO₂ conversion technologies we have today. In 2015, five cycle stability was demonstrated by a Ru‐MACHO‐BH homogeneous catalyst in converting CO₂ to methanol using pentaethylenehexamine [266]. Of critical importance, it was the first study demonstrating that CO₂ captured directly from air can be converted to methanol, with remarkable 79% yields. Shortly after, in 2016, Kothandaraman et al. [267] utilized an AAS to capture CO₂ and perform in situ reduction of the bound species to ammonium formate, with a relatively stable Ru‐PNP and Fe‐PNP pincer complexes. In 2017, Kar et al. [268] used a Mn(I)‐PNP pincer catalyst to demonstrate amine‐assisted homogeneous CO₂ hydrogenation to methanol in a range of solvents, obtaining methanol yields as high as 90%. While works regarding the homogenous condensed phase conversion of CO₂ to methanol have been compiled [269], previous sections have enumerated the advantages of heterogeneous processes. Kothandaraman et al. [15] demonstrated the low temperature conversion of captured form CO₂ to methanol by the means of CZA in a triethylamine‐ethanol solution in 2018. This study obtained remarkable yields and selectivity of 100% and 95% for methanol (with respect to the amine) at reaction temperatures of only 170°C, also providing the first in situ spectroscopic evidence for condensed phase hydronation of alkylcarbonate to methanol via ammonium formate and alkylformate intermediates. In 2020, Kothandaraman et al. [270] used CZA to hydrogenate captured form CO₂ in a variety of carbon capture solvents. They observed that a VOC‐free, nontoxic, bioderived, and biodegradable capture solvent of chitosan and PEG₂₀₀ could achieve methanol concentrations of 139.5 mmol L⁻¹. While an interesting result, the same study demonstrated the effectiveness of the 1:10 triethylamine:ethanol reaction medium, which possessed a final methanol concentration of 689.6 mmol L⁻¹ at 170° after 12 h. Shortly after EEMPA was developed and reported in 2020 [13, 271], it was used to mediate a heterogeneous catalyzed conversion of CO₂ to methane over a 5% Ru/Al₂O₃ catalyst [14] in 2021. In a continuous flow reactor at 170°C and 60 bar, a 43.2% CO₂ conversion with 90.1% methane selectivity was obtained with a 0.03 gCO₂ g_cat ⁻¹h⁻¹ WHSV of 10%_wt CO₂ loaded EEMPA. This study included a TEA with the important finding that this process could reduce the total capital investment and minimum SNG selling price by 32% and 12%, respectively, compared to the conventional Sabatier process.

Combining lessons learned by the 2018 and 2021 studies, in 2022 Kothandaraman et al. [153] used EEMPA to target the synthesis of methanol rather than methane. To do this, the previously identified problem of N‐methylation needed to be remedied, accomplished by moving away from CZA and to Pt based chemistry. Pt based catalysts can achieve 100% C—N bond cleavage selectivities, necessary to prevent N‐methylation, compared to CZA which only has a C—N bond cleavage selectivity of, at best, ∼25%. When 5% Pt was supported by TiO₂, it was found that the SMSIs, stability in organic solvents, and acid–base properties of the Pt/TiO₂ aided its success in obtaining a single pass conversion of 29% with a 70% selectivity to methanol at 170°C. At 190°C, a single pass conversion of 86% was observed with a decrease in methanol selectivity as predicted by Le Chatlier's principle. Recently, in 2024, Barpaga et al. [17] demonstrated an integrated capture and catalytic conversion of CO₂ from simulated flue gas to methanol in EEMPA. In this study, a carbon capture unit with EEMPA was allowed to run, capturing CO₂ from a flue gas simulant. The unit was attached to a continuous flow reactor where Pt/TiO₂ was used to produce methanol. In this combined set‐up, a single pass conversion of >60 C‐mol% and methanol selectivities greater than 80 C‐mol% were obtained at 190°C. Water was found to be a critical deactivator of the catalyst, as catalyst activity was only sustained for 3 h when water was included in the feed. Practically, the biggest challenges requiring adjustments are the stability of both the catalyst and solvent. The next step of actualization will require that catalyst design allow for long term use in flue‐gas like conditions. Changing the catalyst will require follow‐up studies to understand the impact of catalyst composition on amine stability and production activity.

6. Conclusions, Uncertainties, and Outlook

When designing condensed phase heterogeneous catalytic reaction systems, one must be able to identify the relevant solvent parameters and approximate their effects on reaction activity, selectivity, and stability. Based on current literature, and in our opinion, studies on the impact of dielectric constant/relative permittivity (ε), pH, KT α, KT β, KT π*, Henrys Law constant (k_H), density (ρ), viscosity (µ), and molecular diameters are needed on secondary amine capture agent mediated reaction systems. Further, little is mentioned in the literature regarding the catalyst–capture agent stability relationship. There are studies regarding the impact on the catalyst, but as catalysts are designed for increasing abilities to cleave the C—N bond, a successful catalyst must not cleave the C—N bonds on the capture moiety. Regarding stability, there are cases reported of catalysts being leached in basic solvents, however a thorough study linking the two in condensed phase heterogeneous catalysis is lacking. This would be directly applicable to IC³M, as the promise of basic capture agents demands catalysts that will be stable in high pH environments. In general, any IC³M process will have the unique challenge of conversion dependent solvent properties. Specifically, how the heat and mass transfer, along with the viscosity and pH will change in this system lacks significant study. Due to the rather important role of these parameters, they cannot be neglected in a system that aims to do catalysis within a carbon capture solvent.

Many uncertainties, or opportunities for exploration, exist within the realm of condensed phase heterogeneous catalysis. One such avenue is catalyst assisted impurity driven solvent degradation. While many studies in the gas phase chronicle the impact of impurities such as O, S, and Hg compounds, there is little mention of similar contamination damaging solvents during catalysis. Amines in particular face stability issues in oxidative environments [272]. With O₂ present (among NO _x , SO _x , and CO) in flue gas streams [273] and the increased solubility of O₂ in organic solvents [274], the use of a single component CO₂‐BOL may lead to enhanced rates of oxidative degradation. Further, exposing this O₂ enriched CO₂‐BOL to reductive catalytic conditions could accelerate the rate of oxidative amine degradation. The added O₂ could also react directly with H₂, forming water and consuming valuable reagent_. Additional studies are needed to quantify the extent to which oxygen impacts the stability of CO₂‐BOLs in reductive catalytic environments. Along with the uptake of oxygen, many promising catalysts used in these systems rely heavily on noble metals. Any CO tracked into the reactor via solvent absorption presents a large poisoning threat. Some recent gas phase literature demonstrates strategies for mitigating CO poisoning on noble metals, such as modification with α‐MoC [275], but these techniques have not been adapted or studied for condensed phase utilization. In general, catalyst stability in the condensed phase remains underdiscussed in the current literature. Many report catalytic activities of catalysts in “capture agents” but fail to investigate the stability of either the catalyst or the solvent. This represents a significant oversight for those developing IC³M systems. Aside from catalyst stability, the consequences of using increasingly exotic amines for carbon capture are not fully understood in the context of thermal‐catalytic degradation. Everything gets complicated in the catalytic environment, as new mechanistic pathways open and create a plethora of end products for amines to degrade into. To explore this, rigorous reaction pathway studies must be carried out to identify critical intermediates. When this pathway is understood, steps can be taken in solvent and catalyst design to better harmonize the interplay of these crucial elements.

As stated in preceding sections, mass transfer remains a challenge for condensed phase catalytic reaction chemistries. To realize economically feasible processes, exceptional reaction rates are needed from catalyst–solvent systems. Slower mass transfer, increased energy demand, complex reaction mechanisms, and convoluted energetic effects all present critical challenges to in‐depth understanding of condensed phase heterogeneous catalysis. The condensed phase provides an additional variable to tune that may lead to enhanced or diminished results. While many case‐by‐case studies identify specific phenomena for an individual solvent and catalyst, few compilations exist summarizing general trends. We would like to highlight the work of Lin et al. [276], where the authors artfully summarize the structure, properties, and catalytic functions of water. This type of work done for additional solvent classes would be beneficial to guide researchers in their design choices for condensed phase heterogeneous catalytic systems.

In addition to experimental work compiling and summarizing results, there are many opportunities for theoreticians. Again, water possesses many studies on its bulk structure [277, 278], with some summaries regarding its arrangements near surfaces [279, 280], and a number of insightful molecular dynamic and DFT investigations [281, 282]. However, prior art exploring the structure of other solvents, specifically many organic solvents, lack fully detailed overviews of their bulk and near‐surface structures. With some designer amines becoming the front runners for utilization in IC³M, studying the structure of these solvents pre and post capture is crucial for enhanced optimization and understanding of these cutting‐edge processes. For example, molecular modeling work with EEMPA has shown that it is capable of tetrameric self‐assembly, behaving as protosurfactant and forming pseudomicelle clusters after CO₂ uptake [283]. This is believed to be the explanation for the low viscosity and high conductivity of EEMPA. It is still unknown, however, how these pseudomicelles interact with the active sites on a catalyst and if this is an inherently beneficial structure for conventional heterogeneous catalysis. The interplay between catalyst and solvent remains paramount for the development of IC³M and an in‐depth understanding of solvent structure is critical for assessing how it will interact with solid catalytic surfaces. On this note, the mechanism through which bound CO₂ reacts with captured form CO₂ remains convoluted, with only select solvents studied. For some capture agents, it is possible that the reaction temperature releases CO₂, making free CO₂ the reaction substrate. This would be disadvantageous, as the rate of reduction for alkyl carbonates has been shown to be faster than that of free CO₂ via a hypothesized inner‐sphere mechanism [284, 285, 286]. For some amines, it is possible that the reaction temperature will drive the release of CO₂ from the amine prior to reaction while for others the bound form may remain the substrate. Kothandaraman et al. [15], using triethylamine and ethanol to mediate a captured CO₂ hydrogenation to methanol, includes an in situ NMR study which shows the presence of ethyl carbonate on the catalyst surface, identifying it as a key precursor to the formate intermediate in classic gas phase CO₂ hydrogenation pathway over CZA. These types of studies provide critical insight into the nature of condensed phase heterogeneous catalysis and are typically forgone in the current literature.

The condensed phase is not without its challenges; mass transfer limitations alone can pose a threat to viability if not properly circumnavigated. However, in specific cases, the added tunability that solvent provides may bring previously neglected chemistries to the forefront of industrial practice. In the case of condensed phase CO₂ hydrogenation, the use of solvent can activate CO₂, decrease activation energy barriers in the gas phase, and reduce the energy required to valorize this stubborn molecule. The potential to excise processes such as separation, compression, and purification from conventional gas phase CO₂ capture and conversion processes demand rigorous study of these admittingly complex environments. In all studies claiming to progress IC³M, we recommend that alongside catalyst productivity and selectivity, catalyst and solvent stability must remain as equivalent engineering goals. The interactions between catalyst, solvent, and reagent must be harmonious if the desire is to actualize these processes.

Funding

This work was supported by U.S. Department of Energy (Grant FWP 81462, FWP 80562).

Conflicts of Interest

The authors declare no conflicts of interest.

Biographies

Jim B. Floyd obtained his bachelor of science in chemical engineering from Washington State University in 2023. Currently, he is a member of the Distinguished Graduate Research Program via Washington State University, finishing his Ph.D. candidacy at Pacific Northwest National Laboratory. Floyd possesses a passion for research, and his research interests include heterogeneous catalysis, solvent effects, green chemistry, and CO₂ and biomass utilization. Jim aspires to be a leader in the field of condensed phase heterogenous catalysis and is grateful for the guidance of his coauthors and mentors.

Robert A. Dagle is a chief research engineer and Project Manager at Pacific Northwest National Laboratory with more than two decades of experience advancing catalytic processing and thermochemical conversion technologies. He leads approximately $3 million annually in DOE‐funded research, directing multidisciplinary teams that develop catalytic pathways converting methane, CO₂, syngas, methanol, and ethanol into lower cost fuels and chemicals. His expertise spans heterogeneous catalyst development, process intensification, and advanced reactor design, including engineered catalyst integration with micro and meso channel systems and additively manufactured architectures. Robert has led numerous industrial‐laboratory partnerships–including collaborations with an oil major, LanzaTech, Bridgestone, Southern California Gas Company, Gevo, STARS Technology Corporation, and NuScale Power–advancing technologies ranging from alcohol‐to‐jet fuel pathways and reactive distillation to catalytic steam methane reforming and nuclear small modular reactor‐integrated chemical synthesis. With more than 70 peer‐reviewed publications, 26 patents, and honors including Battelle Distinguished Inventor and the PNNL Inventor of the Year Award, his work continues to shape next‐generation pathways for producing low‐cost fuels and chemicals from domestically sourced alternative feedstocks.

Jotheeswari Kothandaraman is a senior scientist at the Pacific Northwest National Laboratory (PNNL), one of the U.S. Department of Energy's national laboratories. Dr. Kothandaraman earned a Ph.D. in chemistry from the University of Southern California, specializing in catalytically driven reversible hydrogen storage materials. After joining PNNL in 2017 as a postdoctoral researcher in catalytic CO₂ conversion, Dr. Kothandaraman became a staff scientist in 2019. With more than a decade of experience spanning catalysis and sustainable chemical process development, Dr. Kothandaraman's work includes CO₂ capture and conversion, hydrogen carrier and polymer synthesis, catalyst design, and the development of low‐carbon chemical pathways. Dr. Kothandaraman's research contributions have been recognized with the Phi Kappa Phi Award for Creative and Scholarly Achievements, election as a fellow of the Royal Society of Chemistry (2023), and the PNNL Laboratory Director's Early Career Exceptional Achievement Award (Ronald L. Brodzinski Award).

David J. Heldebrant is a laboratory fellow and Director of the Physical Sciences Division at the Pacific Northwest National Laboratory. He also has a joint appointment with Washington State University as a research associate professor of Chemical Engineering and Bioengineering. Dr. Heldebrant's research interests focus on making energy‐related processes more efficient and environmentally benign, most notably for CO₂ capture and utilization, industrial gas separations, liquid/liquid separations, and catalysis. He has authored over 100 peer‐reviewed publications, 2 book chapters, and has 23 issued US patents. He was recently recognized as a distinguished inventor of Battelle and a fellow of the American chemical Society. Dr. Heldebrant has a B.S. in fiber and polymer science from the University of California at Davis and a Ph.D. in organic chemistry, also from the University of California at Davis. He was a postdoctoral research fellow at the Pacific Northwest National Laboratory in Richland, WA, before becoming a research staff member.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.