Underlying Roles of Polyol Additives in Promoting CO2 Capture in PEI/Silica Adsorbents

Abstract

Solid-supported amines having low molecular weight branched poly(ethylenimine) (PEI) physically impregnated into porous solid supports are promising adsorbents for CO₂ capture. Co-impregnating short-chain poly(ethylene glycol) (PEG) together with PEI alters the performance of the adsorbent, delivering improved amine efficiency (AE, mol CO₂ sorbed/mol N) and faster CO₂ uptake rates. To uncover the physical basis for this improved gas capture performance, we probe the distribution and mobility of the polymers in the pores via small angle neutron scattering (SANS), solid-state NMR, and molecular dynamic (MD) simulation studies. SANS and MD simulations reveal that PEG displaces wall-bound PEI, making amines more accessible for CO₂ sorption. Solid-state NMR and MD simulation suggest intercalation of PEG into PEI domains, separating PEI domains and reducing amine-amine interactions, providing potential PEG-rich and amine-poor interfacial domains that bind CO₂ weakly via physisorption while providing facile pathways for CO₂ diffusion. Contrary to a prior literature hypothesis, no evidence is obtained for PEG facilitating PEI mobility in solid supports. Instead, the data suggest that PEG chains coordinate to PEI, forming larger bodies with reduced mobility compared to PEI alone. We also demonstrate promising CO₂ uptake and desorption kinetics at varied temperatures, facilitated by favorable amine distribution.

Keywords: Additives; CO2capture; Neutron scattering; Solid-supported amines; Structure-property relationships.

Figure 1

Hypothesized causes for improved CO₂ capture performance of PEG+PEI/SBA‐15. (A) PEG preferentially binding to pore walls, reducing PEI‐wall coordinations. (B) PEG intercalation into PEI clusters, surrounding PEI chains. (C) Improved PEI mobility due to the presence of fast‐moving PEG.

Figure 2

Comparison of SANS spectra for PEG+PEI/SBA‐15 with varying PEG/PEI ratios. (A) Illustration of structures within a mesopore of SBA‐15. R_p stands for the radius of a true void space, t_c means corona layer thickness, R_tot is the summation of R_p and t_c. The pore surfaces are rough, and there are micropores as well as occluded pores. (B) Hypothetical distribution of PEI and PEG. With systematic deuterium labelling, we can discern which is mostly represented in the corona layer. (C) SANS spectra with the logarithm of intensity normalized to the diffraction peak [10], where five major peaks are highlighted and assigned ([10], [11], [20], [21], [22]), responsible for arrangements of mesopores. (D) SANS spectra highlighting diffraction peaks.

Figure 3

Analysis on SANS spectra and extracted parameters. (A) Example fits of SANS spectra for evacuated SBA‐15, dPEI in SBA‐15, 0.4 hPEG/dPEI in SBA‐15, and 1.6 hPEG/dPEI in SBA‐15. All polymer/silica samples had similar pore fill fractions (62–66 %). (B) Schematics showing different functions comprising of total scattering law. Component functions include: particle form factor at low q (I~q⁻⁴) following Porod′s law, unit cell form factor (black dotted line spanning whole q range), structure factor (red dotted line), and diffuse scattering function $i_D\left(q\right)$ at high q.

Figure 4

Polymer morphology around pore walls assessed by SANS and MD simulation. (A) Corona layer SLD estimated by SANS curve fitting and hypothesized values based on proportions of PEG and PEI added (with varying PEG/PEI values). Error bars represent standard deviation of SLD values yielded from three best fit models. (B) Schematic showing plausible PEI and PEG distribution in the corona layer, with upper one showing hypothesized ratio of PEG and PEI and lower one illustrating estimated PEI and PEG distribution based on observation via SANS. (C) Normalized density distribution functions of PEI and PEG under varying PEG/PEI. The pore wall is located at z~10 Å. $\rho \left(z\right)$ is the local density distribution a $z$ , while ${\rho }_B$ is the density in the bulk (obtained bulk simulations).

Figure 5

¹H T ₁‐T ₂ relaxation correlation plots with varied PEG/PEI ratio with unchanged polymer fill fractions. Signals are mostly tied to hPEI (with signals from exchangeable sites in dPEG or SiO₂). Horizontal lines are drawn to guide T ₂ ranges, and vertical lines are drawn to guide T ₁ ranges. Signal intensities are marked on the bar next to T₁‐T₂ plots. (B) Expected structures of PEI and PEG based on observation via ¹H T ₁‐T ₂ NMR. PEG may coordinate to PEI, forming larger bodies, which potentially have lower mobility.

Figure 6

Plausible distribution of PEI and PEG assessed by MD simulation. (A) Square root of the mean‐squared radius of gyration of PEI in the bulk under various PEG/PEI mass ratios. PEG/PEI=0 represents pure PEI. (B) Radial distribution functions of amine groups (PEI) and hydroxyl groups (PEG) in the bulk for three types of couplings–amine‐amine, amine‐hydroxyl, and hydroxyl‐hydroxyl, for 0.4 PEG/PEI. (C) Radial distribution for 1.6 PEG/PEI.

Figure 7

CO₂ sorption and desorption under varied temperatures, covering sorption at subambient temperatures. (A) Amine efficiency under varied temperatures. X‐axis refers to the mass ratio of PEG and PEI. Error bars represent standard deviation. Corresponding CO₂ uptakes are plotted in Figure S7. (B–D) Temperature‐programmed desorption of CO₂ from CO₂‐sorbents with a temperature ramp of 0.5 K/min. Dotted vertical lines are inserted as a guide for desorption temperatures of different sorbents.

Figure 8

Chemistry of CO₂ sorption to PEI/SBA‐15 and dPEG+hPEI/SBA‐15 investigated by DRIFTS. (A–B) IR spectra for PEI and PEG+PEI systems after CO₂ sorption (under 400 ppm CO₂/N₂) for 2 h. (C) Proposed sorbed species based on the literature and our IR data.