Carbon capture in polymer-based electrolytes

Abstract

Nanoparticle organic hybrid materials (NOHMs) have been proposed as excellent electrolytes for combined CO₂ capture and electrochemical conversion due to their conductive nature and chemical tunability. However, CO₂ capture behavior and transport properties of these electrolytes after CO₂ capture have not yet been studied. Here, we use a variety of nuclear magnetic resonance (NMR) techniques to explore the carbon speciation and transport properties of branched polyethylenimine (PEI) and PEI-grafted silica nanoparticles (denoted as NOHM-I-PEI) after CO₂ capture. Quantitative ¹³C NMR spectra collected at variable temperatures reveal that absorbed CO₂ exists as carbamates (RHNCOO⁻ or RR′NCOO⁻) and carbonate/bicarbonate (CO₃²⁻/HCO₃⁻). The transport properties of PEI and NOHM-I-PEI studied using ¹H pulsed-field-gradient NMR, combined with molecular dynamics simulations, demonstrate that coulombic interactions between negatively and positively charged chains dominate in PEI, while the self-diffusion in NOHM-I-PEI is dominated by silica nanoparticles. These results provide strategies for selecting adsorbed forms of carbon for electrochemical reduction.

CO₂ capture in NOHM solutions was systematically studied by a variety of NMR techniques and molecular dynamics simulations.

INTRODUCTION

Over the past several decades, the rapid development of industries supporting intensive human activities has caused a rapid increase of CO₂ in the atmosphere, leading to pressing global warming via radiative forcing. To mitigate the effects of climate change, substantial efforts have been devoted to decarbonizing the industrial, chemical, and power sectors. Post-combustion carbon capture technologies, such as absorption in monoethanolamine solutions, have been developed over the last couple of decades and are considered the most industrially mature systems (1, 2). However, the regeneration of CO₂ from aqueous amine solutions is energy intensive due to the large heat capacity of water, requiring up to 30% of the power output when applied as capture techniques in a coal-fired power plant (3). In addition, the transport of CO₂ for sequestration also increases the societal cost of carbon emission mitigation. Today, there are few large-scale CO₂ sequestration projects because, until the launch of recent tax incentive structure amendments, it was not economically feasible to capture and store CO₂ underground. As a result, there has been considerable interest in converting captured CO₂ to a variety of value-added products with a lower carbon intensity to provide a clear economic incentive that encourages the adoption of carbon management technologies more broadly. More recently, reactive capture and conversion have been proposed as an approach whereby CO₂ is captured and then converted in a single-reactor system, to avoid the energy penalty of regenerating the CO₂ capture medium and subsequently pressurizing and transporting the CO₂ before conversion (4). To realize these proposed energy reductions, efficient reactors and ingenious chemistries that couple CO₂ capture and conversion are urgently needed.

Combined carbon capture and reaction are ideally matched to renewable energy technologies in spite of intermittency and storage issues (5–7). The electrochemical conversion of CO₂ coupled with renewable energy is a promising option to mitigate the effects of greenhouse gas emissions while simultaneously producing value-added chemicals and fuels (8, 9). Now, the CO₂ reduction reaction (CO₂RR) is limited by slow kinetics (10) as well as a typically low CO₂ solubility in aqueous electrolytes (e.g., 34 mM) (9, 11). Despite remarkable breakthroughs in catalyst synthesis (12–14) and electrochemical reactor design (12, 15, 16), there is a paucity of understanding about the role of electrolytes in improving these electrochemical conversion pathways (11, 17). Interactions between the electrolyte and electrode at the interface have been shown to noticeably stabilize their electrochemical reaction intermediates, thus highlighting the importance of the electrolyte phase in CO₂ reduction (18, 19).

To overcome the limitation of mass transfer of CO₂ to the catalyst surface, nonaqueous electrolytes, such as acetonitrile and propylene carbonate, have been used (9, 11). Moreover, additives to aqueous electrolytes such as ionic liquids (18, 20), ligands (21), surfactants (14), deep eutectic solvents (22), and/or amines (23–25) have been found to act as cocatalysts for the CO₂RR owing to their remarkable ability to reduce the required overpotential, tune product selectivity, and improve reaction rates. The use of ionic liquids in the electrochemical conversion of CO₂ has been extensively studied, although the exact mechanism is still unclear due to the lack of operando electrochemical measurements (9, 26). The experimental and computational evidence collected to date suggests that physical and chemical interactions between the ionic liquid and the reaction intermediates play a crucial role in facilitating the CO₂RR (20, 21). Thus, it would be ideal to design an electrolyte with high CO₂ solubility and favorable interactions at the electrode-electrolyte interface that could enhance conversion rates while eliminating the energy-intensive solvent regeneration step associated with CO₂ absorption technologies.

Owing to their high CO₂ solubility (27), chemical tunability (28), and thermal-oxidative stability (29), liquid-like nanoparticle organic hybrid materials (NOHMs) are proposed as electrolyte additives for combined CO₂ capture and conversion applications (30). NOHMs are composed of organic polymers ionically or covalently tethered to an inorganic nanoparticle core, usually via a chemical linker (31, 32). These materials have been extensively studied as water-lean CO₂ capture solvents (33, 34) and electrode-stabilizing electrolyte additives for Li-metal batteries (35, 36). Although NOHMs have been shown to appear as very promising candidates for energy storage applications, the main challenge associated with their use in electrochemical systems is their high viscosities in the neat state (28); thus, the addition of NOHMs to aqueous electrolytes has been recommended to overcome this limitation (37).

Because of their high ionic conductivity, CO₂-loaded aqueous amine solutions under certain reaction conditions have recently been demonstrated as media capable of hosting the efficient upgrade of CO₂ to compounds such as carbon monoxide or formate (23, 25, 32). We recently reported on the importance of the binding energy between polymer canopy functional groups and CO₂ toward the electrochemical conversion of CO₂ (32). For example, CO₂ captured by NOHM-I-PEI in the form of a carbamate was not able to be electrochemically converted at room temperature, while NOHMs containing ether groups along the polymer canopy were able to selectively form CO, the desired product for CO₂RR over silver nanoparticle catalysts, at very negative applied potentials (−1.2 V versus reversible hydrogen electrode) (32). Increasing the temperature of CO₂-loaded amine electrolytes was shown to lead to efficient CO₂ reduction, yet the role of the carbamate species in CO₂ conversion has recently been debated (24, 25).

Here, we have prepared NOHM-I-PEI by ionically tethering branched PEI to ~10-nm silica nanoparticles, which works as an excellent CO₂ capture material (Fig. 1A) (29, 32). After mixing the NOHM-I-PEI (Fig. 1B) with water and salt (e.g., KHCO₃) to form NOHM-based electrolytes, we measured the CO₂ capture capacity and transport properties (e.g., ionic conductivity and viscosity) as a function of NOHM-I-PEI concentration to understand how CO₂ loading affects properties that are relevant to electrochemical processes, such as the speciation of carbon species. In addition, we used a variety of nuclear magnetic resonance (NMR) techniques to further probe the reaction mechanisms, chemical equilibria, and transport behavior in these nanoscale-hybrid electrolyte systems. These results provide a detailed mechanism for CO₂ capture and, along with molecular dynamics simulations, show the role that these reactions play in the dynamics of the tethered PEI chains. Overall, the results of this study inform the optimal NOHM-I-PEI preparation and reaction conditions to enhance reactive capture and conversion of CO₂ in NOHM-based electrolytes.

Fig. 1. CO₂ capture in branched polyethylenimine modified nano-silica (NOHM-I-PEI).

(A) Schematic of CO₂ capture in NOHM-I-PEI solution. (B) Chemical structure and one-dimensional (1D) ¹³C nuclear magnetic resonance (NMR) spectrum of NOHM-I-PEI in D₂O [structural units: inset in (B)] measured at room temperature (~22°C). (C) CO₂ capture capacity and (D) initial CO₂ capture rate and viscosity of PEI-based and NOHM-I-PEI–based fluids/electrolytes as a function of polymer concentration.

RESULTS AND DISCUSSION

Effect of polymer tethering and salt on CO₂ capture capacity

The CO₂ capture capacity (P_CO2 = 1 atm) of various PEI-based and NOHM-I-PEI–based fluids/electrolytes were measured by the CO₂ capture setup (fig. S1), and the results are summarized in Fig. 1 (C and D). There is an apparent linear relationship between the equilibrium CO₂ capture capacity and the concentration, measured in wt %, of PEI and NOHM-I-PEI (Fig. 1C), as expected. To provide proper quantification of the CO₂ capture in NOHM-I-PEI, water-ratio normalization [i.e., water content (λ_H2O)] was applied, enabling the concentration of the PEI and NOHM-I-PEI samples to be directly comparable to one another (37). The water content (λ_H2O) can be determined by eqs. S1 and S2 shown in the Supplementary Materials (see table S1 for the conversion between water content and the polymer concentration). It can be found that this normalization results in a converging of these curves into a single curve (fig. S2), suggesting that polymer tethering does not markedly affect the amine efficiency of the polymer, which is especially useful given that ionically tethered NOHM-I-PEI systems have been shown to demonstrate substantially enhanced thermal stability over the untethered polymer (27). On the basis of the CO₂ capture capacity, the amine efficiency (α) of each sample can be calculated by eq. S3 shown in the Supplementary Materials. The amine efficiency of all samples was between 0.4 and 0.5 (fig. S3), which indicates that the resulting CO₂ capture process mainly consists of two reactions (Eqs. 8 and 9) with a molar ratio of 1:2 for CO₂ and amines, consistent with literature reports at similar experimental conditions (T = 25°C, P_CO2 = 1 atm) (38–40).

The bulk CO₂ uptakes, as well as viscosity measurements, further suggest the efficacy of the NOHM-based electrolytes. Figure 1C shows that the addition of 0.1 M KHCO₃, one of the most common electrolytes used for CO₂ electrochemical reduction, negligibly affects their total CO₂ capture capacity in the concentration range of 10 to 30 wt % (Fig. 1D). The viscosity of PEI and NOMH-I-PEI solutions (Fig. 1D and fig. S4) increases rapidly in both solutions when the concentration increases over 20 wt % or λ decreases below 10 (fig. S4). We surmise that the CO₂ uptake rate in PEI and NOMH-I-PEI solutions will become mass-transfer limited as the concentration increases over ~20 wt % (or λ decreases below 10). This demonstrates that the optimal concentration of NOHM-I-PEI for CO₂ capture is ~10 wt % in our system.

Reaction mechanisms and chemical equilibria for CO₂ capture in PEI and NOHM-I-PEI

Quantitative ¹³C NMR (Fig. 2A) quantitively assessed the carbon species after CO₂ capture in PEI and NOHM-I-PEI solutions. Figure 2B summarizes our proposed mechanism for CO₂ capture by the NOHM-I-PEI composites as discerned from ¹H NMR results (middle of Fig. 2B) and ¹³C NMR results (right-hand side of Fig. 2B). The structure of NOHM-I-PEI dissolved in D₂O (10 wt %) was studied by ¹³C NMR with fully assigned peaks spanning the range of 36 to 56 parts per million (ppm), as shown in Fig. 1B. Likewise, the structure of PEI (8 wt % in D₂O) was also studied with the fully assigned ¹³C spectrum shown in fig. S5, which is consistent with the results reported previously (41). The nearly identical spectra for NOHM-I-PEI and pure PEI indicate that the presence of silica has negligible effects on the chemical environment of PEI.

Fig. 2. ¹³C NMR measurements of PEI and NOHM-I-PEI solutions at variable temperatures.

(A) Schematic of ¹³C NMR measurements in polymer solutions at variable temperatures (VT). (B) Chemical reactions on polymer before and after CO₂ capture. (C) Quantification of carbon species in 8 wt % PEI-D₂O and 10 wt % NOHM-I-PEI–D₂O solutions after ¹³CO₂ capture via ¹³C NMR at variable temperatures. (D) 1D ¹³C NMR spectra of 8 wt % PEI and 10 wt % NOHM-I-PEI in D₂O loaded with ¹³CO₂ or different concentrations of KH¹³CO₃ at room temperature (~22°C).

Carbon-13 NMR provides the most direct assessment of the mechanisms shown in Fig. 2B. Resonance peaks in the range of 36.4 to 55.8 ppm are omitted from display because they are easily assigned to polymer carbons with characteristic chemical shifts of PEI backbone carbons (Fig. 1B and fig. S5). Free CO₂ species are characterized by a chemical shift of 124.5 ppm (42). The 0.8 wt % PEI-D₂O-¹³CO₂ sample exhibits a weak peak at ~124.5 ppm consistent with dissolved ¹³CO₂, and there is also an intense and narrow peak at 160.3 ppm (figs. S6 and S7) that is assigned to HCO₃⁻/CO₃²⁻ species (43). Because of the rapid equilibration between the bicarbonate and carbonate ions (Eq. 5), the peaks for these two ions appear as a single peak whose frequency offset depends on the relative amounts between them [equation 9 shown in (44) was supposed to be [HCO₃⁻]/[CO₃²⁻] = (S − 168.09)/(160.33 − S) with S being the chemical shift of HCO₃⁻/CO₃²⁻]. Thus, the molar ratio of CO₃²⁻ to HCO₃⁻ is seen to increase with the concentration of PEI or NOHM-I-PEI solution (figs. S8 and S9), as evidenced by higher frequency shifts at a higher PEI or NOHM-I-PEI concentration (43, 44). A collection of peaks also appears at ~164.7 ppm that we associate with reaction products of CO₂ with amines, i.e., carbamates (¹⁵N NMR data confirm this assignment, vide infra; as detailed in figs. S10 and S11) (45).

Upon fitting the peaks located in the range of 163 to 165 ppm for PEI after ¹³CO₂ capture, two peaks at 164.4 and 163.6 ppm can be assigned to primary and secondary amine-derived carbamate (fig. S12) (46, 47), respectively. Both primary (α) and secondary (β) amine-derived carbamate peaks are split into two peaks, likely due to the unreacted neighboring amino groups (moieties shown on the top in fig. S12) (46, 47). To further prove the assignment of the α- and β-carbamate species, linear PEI (LPEI), mainly consisting of secondary amines with a small number of primary amines, is compared with branched PEI after CO₂ absorption (fig. S13). The carbamate resonance in LPEI after ¹³CO₂ absorption, mainly appearing around 163.6 ppm, is assigned to β-carbamates, while the higher-frequency, weaker peak at 164.4 ppm belongs to the α-carbamates (located at the end of the LPEI chain). Although dissolved in different solvents (LPEI is not soluble in D₂O at room temperature), the carbamate resonances appear to have the same chemical shifts in both deuterium-substituted methanol (MeOD-d4) and water (D₂O). Dissolved CO₂ can also react with MeOD to form MeOCO₂⁻ in nonaqueous solvents (48), as is apparent in fig. S13. In addition, because of the protonation of amino groups in PEI, the 8 wt % PEI solution presents a high pH value (~11.4; figs. S14 and S15) that facilitates the reaction between HCO₃⁻ and OH⁻ to form CO₃²⁻ (Eq. 5).

The present results are informed by considering the equilibrium reactions involved in the absorption of ¹³CO₂ in both PEI and NOMH-I-PEI solution, as follows (49–51).

Protonation reactions before CO₂ absorption

$${\text{RNH}}_2+{\mathrm{H}}_2\mathrm{O}\rightleftharpoons {\text{OH}}^{-}+{{\text{RNH}}_3}^{+}$$ (1)

$$\text{RR}\prime \text{NH}+{\mathrm{H}}_2\mathrm{O}\rightleftharpoons {\text{OH}}^{-}+\text{RR}\prime {{\text{NH}}_2}^{+}$$ (2)

$$\text{RR}\prime {\mathrm{R}}^{″}\mathrm{N}+{\mathrm{H}}_2\mathrm{O}\rightleftharpoons {\text{OH}}^{-}+\text{RR}\prime {\mathrm{R}}^{″}{\text{NH}}^{+}$$ (3)

Carbamate formation reactions after CO₂ absorption

$${\text{CO}}_2+{\text{OH}}^{-}\rightleftharpoons {{\text{HCO}}_3}^{-}$$ (4)

$${{\text{HCO}}_3}^{-}+{\text{OH}}^{-}\rightleftharpoons {{\text{CO}}_3}^{2-}+{\mathrm{H}}_2\mathrm{O}$$ (5)

$${\text{RNH}}_2+{{\text{HCO}}_3}^{-}\rightleftharpoons {\text{RNHCOO}}^{-}+{\mathrm{H}}_2\mathrm{O}$$ (6)

$$\text{RR}\prime \text{NH}+{{\text{HCO}}_3}^{-}\rightleftharpoons \text{RR}\prime {\text{NCOO}}^{-}+{\mathrm{H}}_2\mathrm{O}$$ (7)

$${\text{CO}}_2+2{\text{RNH}}_2\rightleftharpoons {\text{RNHCOO}}^{-}+{{\text{RNH}}_3}^{+}$$ (8)

$${\text{CO}}_2+2\text{RR}\prime \text{NH}\rightleftharpoons \text{RR}\prime {\text{NCOO}}^{-}+{{\text{RRNH}}_2}^{+}$$ (9)

In these chemical equilibria, the R (R′ or R″) groups represent the rest of the organic polymer. As PEI contains primary, secondary, and tertiary amines, it enables several reaction pathways with CO₂ or HCO₃⁻ to form the products listed above [e.g., ─CH₂CH₂─NHCOO⁻, (─CH₂CH₂─)₂NCOO⁻, and (─CH₂CH₂─)₃NH⁺].

The high pH of the polymer solution (fig. S14) facilitates the conversion of CO₂ to HCO₃⁻, CO₃²⁻, and carbamate species in PEI or NOHM-I-PEI solution (47). To quantitate the ratio of carbon species after ¹³CO₂ absorption, the ¹³C NMR spectrum was collected using inverse gated decoupling with adequate recycle delay (108 to 360 s) to ensure total spin-lattice (T₁) relaxation. On the basis of the integrated peak areas from the ¹³C NMR at variable temperatures (Fig. 2C), the molar ratio between carbonate/bicarbonate (CO₃²⁻/HCO₃⁻) and carbamate (RHNCOO⁻ or RR′NCOO⁻) moieties increases with temperature from 22° to 85°C (Fig. 2C), suggesting a dynamic conversion between the carbamate and bicarbonate species in the polymer solution at elevated temperature (reverse reactions of Eqs. 6 and 7). In addition, the CO₃²⁻/HCO₃⁻ resonance shifts to a lower frequency, indicating an increased ratio of HCO₃⁻ to CO₃²⁻ when increasing the temperature (reverse reaction of Eq. 5) or increasing the concentration of added KH¹³CO₃ (Fig. 2D; as detailed in the fig. S9 caption). The net result of increasing temperature is to drive the equilibria for Eqs. 5 to 7 to favor bicarbonate species (reverse reactions), thereby shifting the equilibria in Eq. 4 toward CO₂. The electrochemical reactor implications are clear: The thermal regeneration of NOHM-I-PEI–CO₂ solutions is more favorably operated near 60° or 80°C with accompanying dissociation of carbamate for electrochemical applications (24).

Transport properties of PEI and NOHM-I-PEI solutions and the impact of CO₂ capture

Besides the reaction mechanisms and the chemical equilibria for CO₂ capture in PEI or NOHM-I-PEI solutions, the influence of CO₂ capture on bulk properties is also of great interest as they are vital factors for electrolytes that determine macroscopic operating electrochemical unit design. When studying the conductivity of PEI and NOHM-I-PEI solutions at different water contents, it is apparent that after CO₂ absorption, the conductivity of both PEI and NOHM-I-PEI solutions have been enhanced 20-fold (Fig. 3A, fig. S16A, and tables S2 and S3). The considerably enhanced conductivity is ascribed to the formation of several charged species after CO₂ chemisorption, including the carbamate anion (RNHCOO⁻ and RR′NCOO⁻), protonated cation (RNH₃⁺ and RR′NH₂⁺), bicarbonate ion (HCO₃⁻), and carbonate ion (CO₃²⁻) (Eqs. 4 to 9) (52). The presence of silica appears to have little to no effect on the conductivity in solutions of NOHM-I-PEI (Fig. 3A). At a more microscopic level, the question becomes what role charges on the PEI chains play in conductivity. By using electrophoretic NMR, we observed that, when adding KHCO₃ to PEI and NOHM-I-PEI solutions, the electrophoretic mobility of PEI increased (fig. S17), further demonstrating that the polymer chains acquired negatively charged carbamates after amines react with HCO₃⁻. With the addition of the potassium salt, K⁺ may tailor the electrochemical double layer by pairing with RNHCOO⁻ or RR′NCOO⁻ and facilitate the electron transfer from the cathode through K⁺ to RNHCOO⁻ or RR′NCOO⁻, influencing the conversion of carbamates into higher-value carbon molecules according to the literature (24).

Fig. 3. Transport properties of PEI-based and NOHM-I-PEI–based electrolyte solutions.

(A) Conductivity and (B) viscosity of PEI and NOHM-I-PEI solutions before and after CO₂ capture. (C) Schematic of ¹H pulsed-field-gradient (PFG) NMR measurements in polymer solutions at variable temperatures. (D) Distribution of self-diffusion coefficient of 8 wt % PEI and 10 wt % NOHM-I-PEI solutions before and after CO₂ capture with or without the addition of KHCO₃ at variable temperatures.

The mass transport properties of PEI and NOHM-I-PEI solutions are studied from both macroscale and microscale aspects: viscosity and self-diffusion. The bulk viscosity measurements of both PEI and NOHM-I-PEI solutions are consistent with the hypothesis that PEI chains entangle at a higher concentration (i.e., lower water content) and present increased hydrodynamic radii when tethered to silica particles. For example, the viscosity of both PEI and NOHM-I-PEI solutions increases as the water content decreases (Fig. 3B), consistent with increasing PEI chain entanglement. Either before or after CO₂ absorption, at the same water content, NOHM-I-PEI solution shows a higher viscosity than that of PEI solution (Fig. 3B). In addition, NOHM-I-PEI has an enhanced viscosity gap (the viscosity difference of the same sample before and after CO₂ absorption at the same concentration or water content; Fig. 3B and fig. S16B) compared with PEI solution. The above results demonstrate that the addition of silica particles to PEI leads to a higher viscosity, attributed to the increased hydrodynamic radii of the grafted nanoparticle units compared to untethered polymer solutions and restrictions on polymer mobility imposed by space-filling requirements (see molecular dynamics simulations below). This is consistent with our previous studies on NOHMs viscosity (53, 54). Last, the viscosity of the aqueous PEI solutions remained relatively unchanged before and after CO₂ capture (Fig. 3B); conversely, the viscosity in NOHM-I-PEI solutions after CO₂ capture is much higher than that of the NOHM-I-PEI solution before CO₂ absorption (which was especially noticeable at a concentration higher than 5 wt % (fig. S16B) or water content below 115 (Fig. 3B). Thus, these findings revealed that carbamate formation from CO₂ chemisorption reactions affects transport properties of NOHM-based electrolyte solutions, a notable concern in electrochemical device design.

At the microscale, we measured the self-diffusion coefficients of polymer via pulsed-field-gradient (PFG) NMR based on the ethylene protons in the PEI units (blue integrated area in fig. S18) or NOHM-I-PEI (blue integrated area in fig. S19). The distribution of self-diffusion coefficients in our polymer solutions is analyzed by Laplace inversion (ILT) (55) of the Stejskal-Tanner plots (Fig. 3C) (56). The mean self-diffusion coefficients (peak value of self-diffusion coefficients in Fig. 3D) for both PEI and NOHM-I-PEI solutions increase slightly with temperature and follow an Arrhenius-like behavior over the investigated temperature range (298.0 to 310.0 K) (fig. S20, A to C). The activation energy of self-diffusion for PEI solution is calculated to be 17.6 kJ mol⁻¹, reaching 37.4 kJ mol⁻¹ when KHCO₃ and CO₂ were introduced. Here, we surmise that the formation of negatively charged carbamates and positively charged ammonium on the polymer chains (Eqs. 8 and 9) leads to coulombic interactions between carbamate and ammonium in different polymer chains, thereby hindering the self-diffusion of polymer chains and presenting higher activation energies. In contrast, the activation energies for the four NOHM-I-PEI samples (NOHM-I-PEI solution before and after CO₂ capture with or without the addition of KHCO₃) are almost the same (fig. S20C): Self-diffusion of PEI depends on the silica nanoparticles whereon the PEI is tethered.

An advantage of the ILT method is the determination of the distribution width of derived parameters; in the present case is the self-diffusion coefficients (Fig. 3D). While polydispersity may play some role (fig. S21), we cannot account for the observed distributions of D_self without considering other factors (57). The width of the distribution of the self-diffusion coefficients of the PEI solution decreases monotonically with temperature (fig. S20B); in contrast, the width of the distribution of the self-diffusion coefficients of NOHM-I-PEI stays the same (fig. S20D). This is expected as the transport properties of bound PEI are expected to be dominated by silica particles due to the smaller size of PEI (~2 nm) compared with that of silica (~10 nm). Similar trends of temperature-dependent self-diffusion coefficients and their distribution width are also observed in lower concentrations of PEI solution (fig. S22) or PEI solution treated with HCl (figs. S23 and S24). These findings demonstrate that the self-diffusion properties of NOHM-I-PEI are dominated by silica nanoparticles rather than PEI.

We quantitatively elucidate the correlation between the dynamics and the state of PEI chains by molecular dynamics simulation of the protonation and carbamation on PEI chains. The size of PEI increases monotonically with the chain protonation fraction due to the more pronounced intra-chain repulsion (Fig. 4, A to C). Consequently, the self-diffusion coefficients decrease with chain protonation fraction (Fig. 4D), in qualitative agreement with experimental results. This occurs in both moderately (fig. S25) and highly branched (fig. S26) chains used here, as well as linear chains reported in the literature (58, 59). Therefore, the dependence of chain size and self-diffusion coefficient on protonation fraction is qualitatively not affected by the degree of branching of PEI, an important factor in industrial scale sourcing of PEI. After adding potassium bicarbonate or introducing CO₂, negatively charged carbamates form on PEI or NOHM-I-PEI, with the resulting accumulation of negative carbamates on PEI chains yielding higher charge repulsion and increasing the polymer size (Fig. 4, E to G). However, unlike chain protonation, further carbamation does not monotonically increase the chain size. At a very high chain carbamate fraction (0.75), the PEI chain does not extend but contracts as a single counterion (Na⁺, K⁺) tends to bind multiple carbamate groups (fig. S27). The self-diffusion coefficients accordingly decrease rapidly when increasing the chain carbamate fraction at first but remain stable or increase slightly when the fraction is over 0.2 (Fig. 4H). We surmise that this non-monotonic relationship explains the results in Fig. 3D where the mean diffusion coefficient of PEI is larger than that of PEI with the addition of CO₂ or KHCO₃ (lower fraction of carbamate) yet slightly lower than that of PEI with the addition of both CO₂ and KHCO₃ (higher fraction of carbamate) at 310.0 K. PEI chains in NOHM-I-PEI probably share the same properties due to the same mechanism of forming carbamate. Therefore, both protonation and carbamation have an essential influence on the self-diffusion coefficient of PEI chains, implying a nontrivial interplay between the dynamics of PEI and electrolyte speciation.

Fig. 4. Molecular dynamics of protonation and carbamation on radius of gyration (R_g) and self-diffusion coefficients.

(A) Moderately and (B) highly branched PEI with protonation fractions of 0 and 0.5. (C) The size and (D) self-diffusion coefficient of PEI as a function of chain protonation fraction in moderately and highly branched PEI. (E) Moderately and (F) highly branched PEI with chain carbamate fractions of 0, 0.5, and 0.75. (G) The size and (H) self-diffusion coefficient of PEI as a function of chain carbamate fraction in moderately and highly branched PEI.

In summary, we provide direct evidence that the temperature determines the ratio of CO₂-chemisorbed products in both PEI and NOHM-I-PEI solutions by variable temperature quantitative ¹³C NMR. The molar ratio between carbonate/bicarbonate (CO₃²⁻/HCO₃⁻) and carbamates (RHNCOO⁻ and RR′NCOO⁻) (i.e., chemisorbed carbon species) increased with the temperatures (22° to 85°C) and the CO₃²⁻/HCO₃⁻ equilibrium ¹³C NMR peak shifted to a lower frequency, indicating reversible conversion between carbamates and carbonate/bicarbonate, thereby suggesting the more favorable operation near 60° or 80°C for potential integrated electrochemical applications. Combining ¹H PFG NMR techniques and inverse Laplace analysis, the average and distribution of self-diffusion coefficients of PEI and NOHM-I-PEI at different conditions (temperature, ¹³CO₂ absorption, the addition of KHCO₃, and polymer concentrations) have been obtained and activation energies discerned. The self-diffusion in the PEI systems is determined by the PEI chain states where the columbic interactions formed between the chains of negatively charged carbamates and those of positively charged ammonium, while, in the NOHM-I-PEI systems, the self-diffusion merely depends on the silica nanoparticles whereon the PEI chains are tethered. Protonation of PEI chains leads to a lower mean self-diffusion coefficient that is further proved by molecular dynamics calculations. This work has revealed the mechanism of CO₂ absorption in PEI and NOHM-I-PEI solutions by distinguishing and quantifying the species generated after CO₂ capture, as well as probing the role of PEI and NOHM-I-PEI self-diffusion mechanisms on mass transfer limitation. These insights are expected to assist in the design of efficient electrolytes combining CO₂ chemisorption and electro-conversion.

MATERIALS AND METHODS

NOHM-I-PEI was synthesized by tethering PEI onto the surface of silica nanoparticles, as previously reported (27, 29). The CO₂ capture capacity of various PEI and NOHM-I-PEI solutions with co-adding electrolytes was measured on the homemade setup (fig. S1). The pH, viscosities, and conductivity of PEI-based and NOHM-I-PEI–based solutions were measured by using a pH meter, a piston-based viscometer, and a conductivity meter, respectively. PEI and NOHM-I-PEI samples for NMR measurements were prepared by dissolving them in D₂O. Variable temperature quantitative ¹³C NMR spectra were recorded on Bruker Avance I 500-MHz NMR spectrometer equipped with a 5-mm ¹H/BB Smart solution probe (Broadband Observe), while ¹H PFG and electrophoretic NMR experiments were performed on Bruker Avance I 700-MHz NMR spectrometer equipped with a 5-mm triple resonance ¹H/¹³C/¹⁵N TXI solution probe. The ¹H-¹⁵N heteronuclear multi-bond correlation (HMBC) spectra were collected on a Bruker Avance II 900-MHz NMR spectrometer equipped with a proton-optimized triple resonance NMR inverse cryoprobe (CP TCI) with the temperature set to 298 K. Additional explanatory text including an extended technical description of results can be found in the Supplementary Materials.

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S27

Tables S1 to S3

References