High-Capacity, Cooperative CO₂ Capture in a Diamine-Appended Metal–Organic Framework through a Combined Chemisorptive and Physisorptive Mechanism

Abstract

Diamine-appended Mg₂(dobpdc) (dobpdc^4– = 4,4′-dioxidobiphenyl-3,3′-dicarboxylate) metal–organic frameworks are promising candidates for carbon capture that exhibit exceptional selectivities and high capacities for CO₂. To date, CO₂ uptake in these materials has been shown to occur predominantly via a chemisorption mechanism involving CO₂ insertion at the amine-appended metal sites, a mechanism that limits the capacity of the material to ∼1 equiv of CO₂ per diamine. Herein, we report a new framework, pip2–Mg₂(dobpdc) (pip2 = 1-(2-aminoethyl)piperidine), that exhibits two-step CO₂ uptake and achieves an unusually high CO₂ capacity approaching 1.5 CO₂ per diamine at saturation. Analysis of variable-pressure CO₂ uptake in the material using solid-state nuclear magnetic resonance (NMR) spectroscopy and in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) reveals that pip2–Mg₂(dobpdc) captures CO₂ via an unprecedented mechanism involving the initial insertion of CO₂ to form ammonium carbamate chains at half of the sites in the material, followed by tandem cooperative chemisorption and physisorption. Powder X-ray diffraction analysis, supported by van der Waals-corrected density functional theory, reveals that physisorbed CO₂ occupies a pocket formed by adjacent ammonium carbamate chains and the linker. Based on breakthrough and extended cycling experiments, pip2–Mg₂(dobpdc) exhibits exceptional performance for CO₂ capture under conditions relevant to the separation of CO₂ from landfill gas. More broadly, these results highlight new opportunities for the fundamental design of diamine–Mg₂(dobpdc) materials with even higher capacities than those predicted based on CO₂ chemisorption alone.

Introduction

Rising atmospheric CO₂ levels are a leading cause of deleterious climate change, and carbon capture from point sources in the power-generation and industrial sectors is being intensively investigated as one of several key mitigation strategies.^1,2 Aqueous amine solutions are the most mature capture technology and are used in a number of large-scale operations around the globe.³⁻⁸ However, these solutions exhibit extremely high regeneration energies and low oxidative and thermal stabilities and generate large amounts of waste, among other challenges, which has limited their more widespread implementation for carbon capture.^6,9−13 To address these challenges, significant research effort has focused on the development of amine-functionalized solid sorbents such as silicas, inorganic oxides, and metal–organic frameworks as alternatives.^5,14−24

Polyamine-appended frameworks of the type amine–Mg₂(dobpdc) (dobpdc^4– = 4,4′-dioxidobiphenyl-3,3′-dicarboxylate),²⁵⁻³⁵ and recently amine–Mg₂(olz) (olz^4– = (E)-5,5′-(diazene-1,2-diyl)bis(2-oxidobenzoate)) materials, have shown particular promise for CO₂ capture,³⁶ and some of these are being developed for testing at the pilot scale.³⁷ The vast majority of these materials selectively capture CO₂ via a cooperative mechanism involving CO₂ insertion into the metal–amine bonds to form ammonium carbamate chains.²⁶ A hallmark of this chemisorption mechanism is step-shaped CO₂ adsorption, where CO₂ uptake occurs within a narrow temperature or pressure window, and as a result, relatively small temperature or pressure swings can be used for material regeneration. Depending on the choice of parent framework and appended amine, these materials can exhibit very high CO₂ capacities exceeding 3 mmol/g at a range of pressures relevant to the capture of CO₂ from diverse flue streams. Fundamentally, the discovery of related materials exhibiting even higher CO₂ adsorption capacities represents an important advance for the field. However, to date, the uptake of CO₂ in diamine–Mg₂(dobpdc) and diamine–Mg₂(olz) materials has consistently been limited to one molecule of CO₂ per diamine (Figure 1a), with small amounts of additional CO₂ uptake due to physisorption.

Figure 1.

(a) Depiction of cooperative CO₂ insertion into diamine–Mg₂(dobpdc) to form chains of ammonium carbamate. (b) (Left) DFT-simulated structure of pip2–Mg₂(dobpdc) and (right) detailed structure of first coordination sphere of a Mg^II site in this structure. Green, blue, red, gray, and white spheres represent Mg, N, O, C, and H atoms, respectively. (c) Structure of linker H₄dobpdc (left) and pip2 (right).

Here, we present a new primary,tertiary (1°,3°)-diamine–appended metal–organic framework, pip2–Mg₂(dobpdc) (pip2 = 1-(2-aminoethyl)piperidine; Figure 1b,c), that can adsorb nearly 1.5 CO₂ molecules per appended amine, as a result of a mixed cooperative chemisorption/physisorption mechanism that is associated with a step-shaped adsorption profile.²⁵⁻³² Together with gas sorption analysis, solid-state nuclear magnetic resonance (NMR) spectroscopy and in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) data support a mechanism involving initial uptake of CO₂ as ammonium carbamate, followed by simultaneous chemisorption and physisorption of CO₂. Powder X-ray diffraction data supported by van der Waals (vdW)-corrected density functional theory (DFT) calculations indicate that CO₂ is physisorbed in a highly ordered fashion, occupying a pocket created by ammonium carbamate chains and the framework linker. Under conditions relevant to landfill CO₂ capture, pip2–Mg₂(dobpdc) adsorbs nearly 5 mmol CO₂ per gram, and extended CO₂ cycling and breakthrough studies under simulated landfill gas³⁸ reveal that this mixed adsorption mechanism is highly robust. These results highlight an opportunity to design a new family of amine-appended MOFs exhibiting enhanced CO₂ capacities via tandem and cooperative chemisorption and physisorption for a range of capture applications.

Results and Discussion

Single-Component Gas Adsorption Experiments

The pip2–Mg₂(dobpdc) material was prepared using a procedure described previously for other diamine–appended Mg₂(dobpdc) materials.²⁵⁻³² Briefly, methanol-solvated Mg₂(dobpdc) (Figures S1–S3) was soaked in a toluene solution of pip2 for several hours, and then the resulting solid was isolated and activated at 130 °C under flowing N₂ for 1 h (see Experimental Section for details and Figures S4–S6). Based on solution-phase ¹H NMR spectroscopy analysis of a digested sample of pip2–Mg₂(dobpdc), diamine loading in the material is quantitative (∼100%). From N₂ adsorption data collected for activated pip2–Mg₂(dobpdc) at 77 K (Figure S5), we calculated a Langmuir surface area of 570 m²/g (Brunauer–Emmett–Teller surface area = 490 m²/g), consistent with surface areas reported for other diamine–Mg₂(dobpdc) variants appended with bulky diamines.³⁴

As an initial assessment of the CO₂ adsorption properties of pip2–Mg₂(dobpdc), we collected adsorption and desorption isobars between 120 and 30 °C under 1 bar of CO₂. The material exhibits double-step CO₂ adsorption behavior with steps located at ∼55 and 40 °C (defined as the midpoint of the step region; Figure 2a, blue trace). Cooperative two-step CO₂ uptake has previously been reported for several diamine–Mg₂(dobpdc) materials featuring bulky 1°,2°-diamines,^33,34 and it was ascribed to steric interactions between resulting ammonium carbamate chains.³⁴ However, for the latter materials, each step is associated with uptake of ∼0.5 equiv of CO₂ per diamine, and the total CO₂ uptake in these materials has not been reported to exceed ∼1.2 CO₂ per diamine, with the additional uptake beyond 1 CO₂ per diamine corresponding to physisorption in the poststep uptake region.^33,34,38 For pip2–Mg₂(dobpdc), the first step in the adsorption isobar is also associated with uptake of close to 0.5 equiv of CO₂ per diamine, although in contrast to previously studied materials, nearly double this quantity is taken up in the second step, resulting in an overall capacity of 1.4 CO₂ per diamine, or 5.1 mmol/g, at 30 °C. Isobaric CO₂ desorption from pip2–Mg₂(dobpdc) also occurs in a two-step fashion, with moderate desorption hysteresis leading to desorption steps at ∼55 and 70 °C.

Figure 2.

(a) Adsorption (blue) and desorption (red) isobars pip2–Mg₂(dobpdc) under pure CO₂ (∼1 bar), as measured by thermogravimetric analysis. (b) Pure CO₂ adsorption isotherms for pip2–Mg₂(dobpdc) obtained at 25, 40, and 50 °C.

To further investigate the CO₂ adsorption properties of pip2–Mg₂(dobpdc), we collected the CO₂ adsorption isotherms at 25, 40, and 50 °C (Figure 2b). The material exhibits double-step profiles at all temperatures, consistent with the isobaric experiments. At 25 and 40 °C, the first adsorption step occurs at 50 and 150 mbar, respectively, and is associated with uptake of 1.3 and 1.5 mmol of CO₂/g; the uptake in the second step (at 200 and 500 mbar, respectively) is nearly double that of the first step (3.3 and 3.1 mmol/g, respectively), giving rise to total capacities of 4.9 and 4.6 mmol/g, respectively. This same trend is observed at 50 °C, although the uptake after each step (∼1.0 and 2.0 mmol/g, respectively) is less than at lower temperatures, consistent with the temperature-dependent uptake characterized in the adsorption isobar data. Notably, the high CO₂ uptake achieved at 300 mbar and 25 °C (4.4 mmol/g) suggests that pip2–Mg₂(dobpdc) may be a promising candidate for CO₂ capture from landfill gas, which is composed of 40–60% CO₂, 40–60% CH₄, and 2–5% N₂, at 25 °C (see further discussion below).³⁹⁻⁴¹ Indeed, single-component CH₄ and N₂ adsorption isotherms collected for pip2–Mg₂(dobpdc) at 25 °C suggest that the material is highly selective for CO₂ over these other gases (Figure S7).

The CO₂ adsorption isotherms at all temperatures were fit using linear interpolation, and the resulting fit data (pressures and loadings) were used with the Clausius–Clapeyron equation to calculate the differential enthalpy (Δh_ads) and entropy (Δs_ads) of CO₂ adsorption as a function of loading (see Experimental Section and Figures S8–S10). From these data, we determined Δh_ads values of −59(2) and −53(1) kJ/mol for the first and second steps, respectively, corresponding to the loading values associated with the midpoint of each step (0.5 and 2.5 mmol/g). Using the reversible heat capacity (1.29 J/g·°C) of pip2–Mg₂(dobpdc) measured by differential scanning calorimetry (Figure S11) and the operating temperature range (adsorption at 25 °C and desorption at 80 °C), we further estimated an approximate regeneration energy of 1.58 MJ/kg CO₂ for a temperature swing adsorption process, which is approximately one-third of the regeneration energy for a monoethanolamine solution used for landfill CO₂ capture (∼4.5 MJ/kg CO₂).^42,43

Spectroscopic Investigation of CO₂ Uptake

To identify the adsorbed species formed upon CO₂ uptake in pip2–Mg₂(dobpdc), we collected in situ DRIFTS data for a sample of the material dosed with CO₂ at room temperature and pressures of 100 and 300 mbar, corresponding to immediately after the first and second adsorption steps in the 25 °C isotherm. In separate experiments, the activated framework was first dosed with 100 mbar of natural-abundance CO₂ (∼99% ¹²CO₂) or ¹³CO₂ and allowed to equilibrate before the spectra were collected (5 h in each case). Two additional experiments were carried out by dosing the activated framework with 300 mbar of CO₂ or ¹³CO₂, and spectra were collected at regular intervals until equilibration occurred after 13 h. Isotopic difference spectra were generated from both sets of data by subtracting the equilibrated ¹³CO₂ spectrum obtained at 100 or 300 mbar from the corresponding equilibrated CO₂ spectrum. As shown in Figure 3a, the two difference spectra both feature characteristic bands associated with the C=O and C–N stretches of carbamate at 1644 and 1325 cm^–1, respectively, and the intensity of both bands increased at higher dosing pressure. These data are consistent with a mechanism involving ammonium carbamate formation upon the CO₂ uptake.

Figure 3.

(a) Isotopic difference spectra generated by subtracting equilibrated DRIFTS data collected for pip2–Mg₂(dobpdc) dosed with ¹²CO₂ and ¹³CO₂ at 100 mbar (gray trace) and 300 mbar (green trace). Blue bands highlight peaks associated with carbamate. (b) Difference spectra generated by subtracting DRIFTS data for bare pip2–Mg₂(dobpdc) from spectra obtained for pip2–Mg₂(dobpdc) dosed with 100 mbar of CO₂ after 5 h and time-resolved spectra for pip2–Mg₂(dobpdc) dosed with 300 mbar of CO₂. Dosing with 300 mbar of CO₂ results in the appearance of peaks at 3696 and 3587 cm^–1 assigned to physisorbed CO₂, concomitant with an increase in the intensity of a peak at 3394 cm^–1 associated with carbamate formation (light to dark green).

We also generated difference spectra by subtracting the spectrum of pristine pip2–Mg₂(dobpdc) from time-resolved spectra collected for the framework after dosing with 300 mbar of CO₂. These spectra are plotted in Figure 3b (pale to dark green) along with a difference spectrum obtained from subtracting the spectrum for the pristine framework from the equilibrated spectrum collected after dosing with 100 mbar of CO₂ (gray data). In the 100 mbar difference spectrum, negative peaks were apparent between 3400 and 3200 cm^–1, along with a large positive feature at 3394 cm^–1, consistent with the conversion of N–H vibrations associated with primary amines to those associated with secondary amines as carbamate is formed upon CO₂ uptake. At 100 mbar, no peak was apparent for physisorbed CO₂ (expected at ∼3696 cm^–1; see Figure 3b, gray trace), consistent with only chemisorption occurring up to adsorption of ∼0.5 equiv of CO₂ per diamine. However, in the time-resolved difference spectra generated after dosing with 300 mbar of CO₂, new peaks were apparent at 3696 and 3587 cm^–1, corresponding to the combination bands of physisorbed CO₂.⁴⁴ Notably, these peaks were visible within 15 min and grew concomitant with an increase in the intensity of the carbamate peak at 3394 cm^–1, indicating that well before equilibration at this pressure, physisorption is occurring simultaneously with ammonium carbamate formation. Both peaks associated with physisorbed CO₂ grew in intensity over the course of several hours (Figure 3b and inset) along with the carbamate peak. These data strongly suggest that CO₂ physisorption is occurring simultaneously with chemisorption in pip2–Mg₂(dobpdc) as the material adsorbs CO₂ beyond 0.5 equiv per diamine.

To gain additional insight into the adsorbed species formed upon CO₂ uptake in pip2–Mg₂(dobpdc), we collected solid-state magic angle spinning (MAS) ¹⁵N and ¹³C NMR spectra after dosing the framework with ¹³CO₂. The ¹H → ¹⁵N cross-polarization spectrum of pip2–Mg₂(dobpdc) dosed with 1 bar of ¹³CO₂ features resonances at ∼85 and ∼51 ppm (Figure 4a). Both peaks are consistent with resonances characterized previously for carbamate and ammonium generated upon CO₂ adsorption in various diamine–Mg₂(dobpdc) materials.³⁸ In addition, a two-dimensional ¹H → ¹³C HETCOR spectrum collected for the same sample features a strong correlation at 4.9 ppm (¹H) and 163.2 ppm (¹³C), consistent with the formation of ammonium carbamate species (Figure 4b,c).³⁸

Figure 4.

(a) Solid-state MAS ¹⁵N NMR spectrum (11.7 T) for pip2–Mg₂(dobpdc) dosed with 1 bar of ¹³CO₂, acquired by cross-polarization. (b) Solid-state ¹H–¹³C HETCOR NMR spectrum for pip2–Mg₂(dobpdc) dosed with 1 bar of ¹³CO₂. (c) Illustration of the proposed mechanism for CO₂ chemisorption in pip2–Mg₂(dobpdc) based on solid-state NMR data. (d) Solid-state MAS ¹³C NMR spectra (11.7 T) for a sample of pip2–Mg₂(dobpdc) dosed at different ¹³CO₂ pressures. See Experimental Section for details.

Solid-state MAS ¹³C NMR spectra were additionally collected for activated samples of pip2–Mg₂(dobpdc) dosed and equilibrated at room temperature with ¹³CO₂ pressures ranging from 0 to 300 mbar (see Experimental Section for details). After dosing with 100 mbar of ¹³CO₂, a clear resonance was apparent at 162.7 ppm, which we assign to carbamate based on ¹³C NMR spectra reported for other diamine–Mg₂(dobpdc) variants dosed with CO₂.³⁸ Upon increasing the ¹³CO₂ pressure to 200 mbar, a new resonance became apparent at 125.1 ppm, which was assigned to physisorbed CO₂.³⁸ Both resonances persisted upon dosing with 300 mbar of ¹³CO₂, consistent with the DRIFTS data. Notably, the presence of chemisorbed and physisorbed CO₂ at 200 mbar—a pressure within the second step of the 25 °C CO₂ isotherm—is further evidence that physisorption indeed occurs simultaneously with chemisorption in pip2–Mg₂(dobpdc) after adsorption of an initial ∼0.5 equiv of CO₂. Finally, integration of a ¹³C NMR spectrum collected after dosing pip2–Mg₂(dobpdc) with 1 bar of ¹³CO₂ (Figure S12) yielded a chemisorbed:physisorbed CO₂ ratio of 1:0.43. Assuming saturation of the material with CO₂ under the experimental conditions, this ratio is consistent with the total uptake of ∼1.4 equiv of CO₂ per diamine in pip2–Mg₂(dobpdc) (Figure 2b), significantly more than the 10–15% physisorbed CO₂ that has been reported for other diamine-appended frameworks.³⁸ Altogether, these results indicate that the unusually high CO₂ capacity of pip2–Mg₂(dobpdc) arises due to a unique cooperative mechanism involving the simultaneous physisorption of CO₂ with CO₂ chemisorption at pressures corresponding to the second isothermal adsorption step. In contrast, for other diamine-appended frameworks, chemisorption alone is associated with the step regions of the isotherm, whereas physisorption is predominantly a poststep phenomenon.^33,34

Structural and Computational Investigation of CO₂ Binding

To elucidate the structure resulting upon CO₂ uptake in pip2–Mg₂(dobpdc) and identify the location of the physisorbed CO₂, we collected in situ powder X-ray diffraction data for a sample of the framework before and after dosing with 1 bar of CO₂ at 298 K (see Experimental Section for details). The diffraction pattern of pip2–Mg₂(dobpdc) features more peaks than that of Mg₂(dobpdc) (Figure S1), for example, three low intensity peaks at low 2θ values, while there are more differences at higher angles. Interestingly, the diffraction pattern collected for the CO₂-dosed sample does not feature these additional peaks (Figure S6), and it is more consistent with typical diffraction patterns of diamine-appended Mg₂(dobpdc) with and without CO₂.^33,34 The disappearance of peaks after CO₂ dosing indicates that CO₂ adsorption gives rise to a more ordered and higher-symmetry space group for the resulting structure. Additionally, the diffraction peaks for the CO₂-dosed sample appear at 2θ values lower than those for pip2–Mg₂(dobpdc), indicating an expansion of the crystal lattice to accommodate CO₂.

Single-crystal and powder X-ray diffraction analysis of other diamine-appended Mg₂(dobpdc) and Mg₂(olz) frameworks has revealed that these structures adopt various trigonal space groups,^34,36 although the diffraction pattern collected for pip2–Mg₂(dobpdc) could not be indexed to any trigonal space group due to the presence of the additional peaks noted above. Attempts to index the pattern of activated pip2–Mg₂(dobpdc) with lower symmetry space groups (either monoclinic or triclinic) were unsuccessful, although some extra peaks that could not be indexed with trigonal space groups were indexed by lower symmetry space groups (Figure S13 and Tables S1 and S2). The difficulties encountered with indexing the diffraction pattern to a single space group could be indicative of the presence of two distinct phases of pip2–Mg₂(dobpdc) due to different orientations of the bulky diamine in the framework pores.

In contrast, the diffraction pattern collected for pip2–Mg₂(dobpdc) dosed with CO₂ could be successfully indexed to the space group P3₂21 (Table S2), consistent with the space group assignment for other diamine-appended Mg₂(dobpdc) and Mg₂(olz) frameworks dosed with CO₂.^33,34 To locate physisorbed CO₂ in the structure, a calculated structure for pip2–Mg₂(dobpdc) featuring one CO₂ per diamine in the form of ammonium carbamate chains was generated using DFT with vdW corrections (see Experimental Section for details).³⁸ We then introduced one additional CO₂ molecule for every two metal sites in the unit cell of the structure to simulate the uptake of half an equivalent of CO₂ per diamine and performed geometry optimization to allow the CO₂ molecule to rearrange in the pore. Repeating this calculation with different starting coordinates for CO₂ yielded three candidate structures that were used for Rietveld refinement of the diffraction pattern for the CO₂-dosed framework. In the first of these structures, the physisorbed CO₂ is located near one of two adjacent pip2 moieties in the ab plane (structure A, Figure S14). In the second structure, CO₂ is located in the center of the pore (structure B, Figure S16), and in the third structure, CO₂ is located in a pocket formed by two carbamates and the linker (structure C, Figure S18).

When the occupancy of the extra CO₂ molecule was freely refined in the candidate structures, structures A and B gave physically unreasonable negative occupancies, while structure C yielded a positive occupancy for extra CO₂ after the initial refinement. Further refinement of the atomic coordinates of structure C yielded a final structure for the CO₂-dosed pip2–Mg₂(dobpdc) (Figure 5, Figures S20 and S21, and Tables S2 and S3). Moreover, our DFT calculations with vdW corrections show that structure C is more stable than structures A and B by 10.1 and 8.0 kJ/mol, respectively. As shown in Figure 5, the physisorbed CO₂ molecules run along the framework channels and are located in the region between adjacent ammonium carbamate chains. The distance between the carbon of the CO₂ molecule and the oxygen atom of each carbamate is 3.4(2) Å (see Table S3). Based on the van der Waals radii of carbon and oxygen (1.70 and 1.52 Å, respectively),⁴⁵ this distance suggests that the physisorbed CO₂ molecules are engaged in stabilizing interactions with the ammonium carbamate chains. The physisorbed CO₂ was ultimately refined to an occupancy of 59(2)%; because it is located at a special position (Wyckoff position 3b) of the P3₂21 space group, this occupancy translates to an uptake of approximately 0.3 equiv per carbamate, consistent with the CO₂ uptake determined from isobaric and isothermal analyses.

Figure 5.

Views of the structure of (CO₂)_1.5–pip2–Mg₂(dobpdc) determined from Rietveld refinement (left) along the c axis and (right) in the ab plane. Green, red, blue, gray, and white spheres represent Mg, O, N, C, and H, respectively.

To further support our assignment of the position of physisorbed CO₂, we used vdW-corrected DFT to simulate geometry-optimized structures for pip2–Mg₂(dobpdc), pip2–Mg₂(dobpdc) loaded with 0.5 equiv of CO₂ per diamine, and pip2–Mg₂(dobpdc) loaded with 1.5 equiv of CO₂, the latter based on the final structure discussed above. Using these optimized structures, we calculated the CO₂ binding energies and NMR chemical shifts associated with species formed after loading of 0.5 and 1.5 equiv of CO₂ per diamine (see Tables S4 and S5). For pip2–Mg₂(dobpdc) loaded with 1.5 CO₂ per diamine, the computed binding energy is within 5 kJ/mol of the corresponding experimental Δh_ads value (−48.6 versus −53(1) kJ/mol, respectively), and the simulated carbamate ¹³C NMR shift of 165.9 ppm is comparable to the experimental peak position at 162.5 ppm. Taken together, the spectroscopic and computational results clearly support a mechanism of CO₂ uptake in pip2–Mg₂(dobpdc) involving initial chemisorption of CO₂ at half of the diamine sites, followed by a dual chemisorptive/physisorptive process wherein binding of CO₂ at the remaining diamine sites is accompanied by uptake of an additional half equivalent of physisorbed CO₂.

Adsorption Performance under Simulated Landfill Gas

As noted above, pip2–Mg₂(dobpdc) exhibits a high CO₂ capacity at pressures relevant to the concentration of CO₂ in landfill gas, which typically contains 40–60% methane, 40–60% CO₂, and a balance of N₂ and small amounts of organic compounds.³⁹⁻⁴¹ Both methane and CO₂ are potent greenhouse gases, and methane has a global warming potential nearly 30 times that of CO₂.⁴⁶ Capturing and refining methane from landfill gases have the potential to significantly reduce greenhouse gas emissions and also provide a valuable energy source. The generation of purified methane from landfill gas involves the removal of moisture and impurities, followed by removal of CO₂.⁴⁶ To evaluate the potential of pip2–Mg₂(dobpdc) for this separation, we carried out breakthrough experiments using a custom-built breakthrough apparatus and binder-free pellets of the material (see Experimental Section for details). For safety reasons, a blend of CO₂ in N₂ was employed for the experiments instead of the use of CO₂ and CH₄. A stream of dry 60% CO₂ in N₂ at 25 °C and ambient pressure was passed through the breakthrough column, and the outlet gas composition and flow rate were monitored over time. As shown in Figure 6a, a sharp breakthrough of CO₂ occurred after 15 min, and the CO₂ uptake capacity of the material under these conditions was determined to be 5.1 mmol/g. This high capacity remained unchanged over the course of four consecutive desorption (80 °C under flowing He) and breakthrough cycles (Figure 6a), and ¹H NMR spectroscopy analysis of a digested sample of the framework following this cycling experiment revealed that diamine loading remained at approximately 100%.

Figure 6.

(a) Breakthrough data for pip2–Mg₂(dobpdc) exposed to flowing dry 60% CO₂ in N₂ (∼1 bar, 5 sscm) at 40 °C. The material adsorbed 5.1 ± 0.2 mmol of CO₂ per gram under these conditions. Note that the nonzero flow detected at the earliest time points is due to residual gas in the tubing downstream from the breakthrough column, resulting from purging the manifold with the gas mixture before the start of each run (see Experimental section for details). This initial flow does not affect the breakthrough capacity calculation. (b) Thermogravimetric temperature–swing cycling data collected for pip2–Mg₂(dobpdc) at atmospheric pressure; adsorption: 30 °C, 60% CO₂ in N₂, 15 min; desorption: 80 °C, CO₂, 1 min.

The framework also exhibits excellent stability to long-term TGA adsorption/desorption cycling (Figure 6b; adsorption: 15 min, dry 60% CO₂ in N₂ at 1 atm (Figure S24); desorption: 1 min, dry CO₂ at 1 atm and 80 °C). Over the course of 500 cycles, pip2–Mg2(dobpdc) exhibited a stable capacity of ∼16.9 g/100 g (or 3.83 mmol/g; note that this capacity is slightly lower than that determined from the breakthrough experiments as a result of the slightly higher adsorption temperature of 30 °C). Powder X-ray diffraction analysis of the material after the cycling experiment revealed that it remained highly crystalline (Figure S25), and ¹H NMR spectroscopy analysis of a digested framework sample revealed that diamine loading remained at ∼100%. These results highlight the excellent stability of pip2–Mg₂(dobpdc) to long-term cycling under conditions relevant to CO₂ capture from landfill gas.

Conclusions

We have discovered a new diamine-appended metal–organic framework, pip2–Mg₂(dobpdc), that captures CO₂ via an unprecedented mixed chemisorption/physisorption mechanism that endows it with a capacity of nearly 1.5 equiv of CO₂ per diamine, significantly more than previously reported materials in this class (typically <1.2 equiv per diamine). This behavior is associated with two-step CO₂ uptake in isothermal and isobaric adsorption data, involving sequential capture of ∼0.5 equiv of CO₂ per diamine in the first step, followed by ∼1 equiv of CO₂ per diamine in the second step. In situ DRIFTS and solid-state magic angle spinning NMR spectroscopy data support a mechanism in which uptake in the first step is associated with the formation of ammonium carbamate chains at half of the diamine sites in the material, followed by uptake of additional CO₂ via chemisorption at the remaining diamine sites and physisorption of CO₂ between resulting adjacent ammonium carbamate chains. Importantly, pip2–Mg₂(dobpdc) retains its high capacity and exhibits exceptional stability over the course of breakthrough and extended cycling experiments using a simulated landfill gas. Ultimately, these results suggest that, with judicious choice of diamine, it may be possible to design a new class of amine-appended Mg₂(dobpdc) (and Mg₂(olz)³⁶ materials with significantly enhanced capacities as a result of a mixed chemisorption and physisorption mechanism, as demonstrated here. Additional studies are ongoing to understand the role of the diamine structure and pore environment in facilitating this adsorption mechanism.

Experimental Section

General Procedures

All of the experiments are carried out in the air, unless noted otherwise. All reagents and solvents were purchased from Sigma-Aldrich at reagent-grade purity or higher and used without further purification. The ligand H₄dobpdc was purchased from Hangzhou Trylead Chemical Technology Co. Ultrahigh purity (>99.998%) gases used for all adsorption measurements were purchased from Praxair, as well as the custom gas blends of 60% CO₂ in N₂. The solution-phase ¹H nuclear magnetic resonance (NMR) spectra were collected on a Bruker AMX 400 MHz NMR spectrometer for digested framework samples, which were referenced to residual dimethyl sulfoxide (δ = 2.50 ppm).

Synthesis of Mg₂(dobpdc)

The metal–organic framework Mg₂(dobpdc) was synthesized following the previously reported procedure.²⁵⁻³² Generally, Mg(NO₃)₂·6H₂O (11.6 g, 45.2 mmol) and H₄(dobpdc) (9.75 g, 35.6 mmol) were dissolved in a 55:45 (v:v) mixture of methanol and N,N-dimethylformamide (DMF) (total volume of 200 mL) with sonication. The resulting solution was filtered to remove undissolved particles and transferred to a 350 mL high-pressure round-bottom flask equipped with a stir bar. The flask was sealed with a Teflon cap and then heated at 120 °C in an oil bath for 20 h with stirring (300 r.p.m.). After this time, an insoluble white solid had formed. The mixture was allowed to cool to room temperature, and the solid product was filtered using a Buchner funnel. The solid was then soaked in 300 mL of fresh DMF at 80 °C for 6 h and then isolated again via filtration. This process was repeated two more times, and the resulting solid was then soaked in 300 mL of fresh methanol at 60 °C for 6 h and then isolated via filtration; this process was repeated two more times. The resulting solid Mg₂(dobpdc) was stored in fresh methanol at room temperature. The powder X-ray diffraction pattern obtained for this material (Figure S1) and the calculated Langmuir surface area (77 K, N₂) of 4000 m²/g (Figure S3) are consistent with that previously reported for Mg₂(dobpdc).²⁵⁻³²

Synthesis of pip2–Mg₂(dobpdc)

Methanol-solvated Mg₂(dobpdc) (150 mg) was dispersed in a 5 mL solution of 20% pip2 in toluene. The powder was soaked for 18 h and then filtered and washed with fresh toluene in a Buchner funnel (3 × 20 mL) at room temperature to remove as much residual diamine as possible prior to activation. The diamine-appended MOF was then activated at 130 °C for 1 h under flowing N₂ in a Schlenk flask equipped with a rubber septum and venting needle (the activation temperature was determined by the thermogravimetric decomposition analysis; see Figure S5). Diamine loading was determined by solution-phase ¹H NMR spectroscopy analysis of a digested sample of the material. Briefly, ∼2 mg of material was suspended in 0.5 mL of dimethyl sulfoxide-d₆ and 100 μL of DCl solution (35 g/100 g in D₂O, ≥ 99 atom % D) was added to dissolve the sample.³² Surface areas, powder X-ray diffraction patterns, and decomposition profiles are presented in Figures S4–S6.

Thermogravimetric Analysis

Thermogravimetric analysis (TGA) data were collected by using a TA Instruments Discovery thermogravimetric analyzer. Thermogravimetric decomposition experiments were carried out under 100% N₂ with a temperature ramping rate of 2 °C/min from 30 to 600 °C with a gas flow rate of 25 mL/min. Masses were not corrected for buoyancy effects. Isobaric data under pure CO₂ were collected at ambient pressure using a gas flow rate of 25 mL/min. Prior to isobar collection, pip2–Mg₂(dobpdc) was first activated by heating at 120 °C for 30 min under flowing N₂. The inlet gas was then switched to 100% CO₂, and the sample was held isothermally at 120 °C under flowing 100% CO₂ for 30 min to completely purge the system of N₂. Adsorption isobar data were obtained while slowly cooling the sample to 30 °C with a ramping rate of 1 °C/min, and desorption data were collected upon then heating the sample to 130 °C at the same ramping rate. The reported two step temperatures from the adsorption isobar were determined using the inflection points of the adsorption steps, and the desorption temperature was determined at the point of closure of the hysteresis loop.

For analysis of adsorption kinetics (see Figure S24), the sample was first activated at 120 °C for 30 min under a 100% dry N₂ stream to remove remaining unreacted amine in the framework pores and then cooled to 30 °C prior to the kinetics experiments. The inlet gas was then switched to 60% CO₂ in N₂ at 1 bar, and the sample was held isothermally for 60 min to study the adsorption kinetics under conditions intended to simulate exposure to landfill gas (Figure S21).

For the cycling experiments, the conditions were based on the adsorption/desorption kinetics results. The materials were first activated at 120 °C for 30 min under a 100% dry N₂ stream to remove the extra diamine in the framework pores and then ramping to 30 °C prior the cycling experiments. The inlet gas was then switched to dry 60% CO₂ in N₂ (∼1 bar) at 30 °C, and the sample was held isothermally for 15 min to adsorb CO₂; the furnace was then ramped to 80 °C and held isothermally for 1 min as 100% dry CO₂ (∼1 bar) was flowed over the sample to desorb CO₂. A total of 500 adsorption/desorption cycles were performed.

Breakthrough Measurements

Breakthrough experiments were conducted using pip2–Mg₂(dobpdc) to gauge the separation performance of the material under a multicomponent dynamic stream intended to simulate landfill gas. Experiments were carried out using a custom-built breakthrough apparatus consisting of Swagelok fittings and 1/8″ copper tubing connecting gas flow to the sample holder or bypassing the sample holder. Cylinders of CO₂ and N₂ were connected to the breakthrough manifold by using Alicat mass flow controllers. Gas flow was controlled to achieve 60% CO₂ and 40% N₂ gas streams with a flow rate of 5 sccm. In particular, prior to flowing the mixture through the pip2–Mg₂(dobpdc) sample, the gas mixture was equilibrated by flowing through the breakthrough manifold with the sample column closed off. The outlet flow rate was verified using an Agilent ADM Flow Meter, and the outlet gas stream composition was verified using an SRI Instruments 8610 C GC equipped with a 6′ Haysep D column maintained at 110 °C. This purging of the manifold leads to the initial nonzero flow rate in the first few minutes of the breakthrough data in Figure 6.

The pip2–Mg₂(dobpdc) sample was pelletized using a pellet die with 1 in. diameter and separated with a 20–40 mesh sieve. A sample of pellets (460 mg) was loaded into the sample holder and activated at 80 °C and under flowing He at 5 sccm for 20 min. Then, the sample was cooled to 25 °C for the breakthrough experiments. The outflow composition and flow rate throughout the breakthrough experiment were analyzed using the same flowmeter and GC specified above. Once CO₂ had broken through the packed pip2–Mg₂(dobpdc) bed, the stream was switched to He gas at a flow rate of 5 sccm, and the sample holder was heated to 80 °C to fully desorb CO₂ from the column prior to subsequent breakthrough measurements.

Gas Adsorption Isotherms

Carbon dioxide adsorption isotherms were collected using a Micromeritics 3Flex gas adsorption analyzer, and 77 K N₂ adsorption isotherms were collected on a Micromeritics ASAP 2420 instrument. All gases were 99.998% pure or higher. The temperature was controlled by an oil bath during CO₂ adsorption isotherm collection and controlled by liquid nitrogen when collecting N₂ adsorption isotherms. Approximately 50 mg of activated pip2–Mg₂(dobpdc) powder was transferred to a glass adsorption tube with a Micromeritics Transeal. Samples were regenerated at 100 °C under dynamic vacuum (<10 μbar) for overnight between isotherms. The isotherm data points were considered equilibrated if the pressure change is <0.01% after 11 consecutive equilibration time intervals (15 s).

Calculation of Differential Enthalpies and Entropies for Adsorption

The differential enthalpy (Δh_ads) and entropy (Δs_ads) of CO₂ adsorption for pip2–Mg₂(dobpdc) were calculated using the Clausius–Clapeyron equation:

From the isotherm fits, the exact pressure (p_q) corresponding to CO₂ loading (q) was determined at different temperatures (T) by plotting ln(p_q) versus 1/T at constant values of q. The y intercepts of these linear trendlines are equal to – Δs_ads/R at each loading, and the slopes are equal to – Δh_ads/R. Further details are shown in Figure S8–S10.

Differential Scanning Calorimetry

Differential scanning calorimetry (DSC) experiments were carried out using a TA Instruments Q1000/RCS90 DSC instrument. A sample of activated pip2–Mg₂(dobpdc) was characterized under an ambient pressure of N₂ using a gas flow rate of 50 mL/min. The temperature was first ramped to 120 °C and held for 30 min to reactivate the material. Then, the material was cooled to 0 °C and heated to 120 °C (ramp rate of 2 °C/min). DSC data were collected between 25 and 80 °C (Figure S11).

Solid-State NMR Spectroscopy

All solid-state NMR spectra were collected at 11.7 T using a 4.0 mm Bruker MAS probe with a MAS rate of 10 kHz, except for the ¹³C{¹H} 2D HETCOR spectrum, which was acquired at 16.5 T using a 3.2 mm Bruker MAS probe with a magic-angle spinning rate of 15 kHz. For data collection, activated pip2–Mg₂(dobpdc) was packed into a 4.0 mm zirconia NMR rotor inside an argon-filled glovebox before being transferred with a rotor cap (to prevent air exposure) to a home-built gas manifold. For the ¹³CO₂ dosing experiment, the rotor was transferred to a home-built gas manifold and degassed for 15 min. Subsequently, variable pressures (100, 200, 300, and 1000 mbar) of ¹³CO₂ gas (Sigma-Aldrich, 99 atom % ¹³C, <3 atom % ¹⁸O) were dosed into the sample at room temperature and allowed to equilibrate over the course of at least 12 h before capping the rotor for data collection. In particular, after the material was initially dosed at each pressure, the pressure was monitored over time and additional ¹³CO₂ was dosed into the system until the final pressure was the same as the desired dosing pressure (see our previous study for details on the gas manifold, which allows samples to be closed inside rotors at controlled CO₂ pressures).³⁸

Solid-state ¹³C{¹H} and ¹⁵N{¹H} cross-polarization (CP) NMR spectra were acquired by using an optimal contact time in the range of 1–5 ms with proton Spinal64 decoupling at a B1 field of 57.1 kHz during acquisition. Solid-state MAS ¹³C NMR spectra (11.7 T) for a sample of pip2–Mg₂(dobpdc) dosed at different ¹³CO₂ pressures were acquired by a direct pulse with recycle delays (∼100–120 s) and with two-pulse phase modulation ¹H decoupling at 32 kHz. The quantitative ¹³C NMR spectrum acquired after dosing pip2–Mg₂(dobpdc) with 1 bar of ¹³CO₂ was acquired by applying a long recycle delay (1000 s) with continuous-wave ¹H decoupling (MAS rate of 15 kHz). The ¹³C{¹H} 2D HETCOR experiments also employed magnetization transfer by cross-polarization with a short contact time of 100 μs, used to selectively show short-range correlations. The ¹H, ¹⁵N, and ¹³C chemical shifts were referenced to 1.85 ppm (adamantane), 33.4 ppm (glycine), and 38.48 ppm (adamantane, tertiary carbon—left-hand resonance), respectively.

In Situ Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS)

In situ DRIFTS data were collected using a Bruker Vertex 70 spectrometer equipped with a glowbar source, KBr beamsplitter, and liquid nitrogen cooled mercury–cadmium–telluride detector. A custom-built diffuse reflectance system with an IR-accessible gas dosing cell was used for all of the measurements. The cell is equipped with a heater controlled by a thermocouple in direct contact with the sample, and the sample atmosphere was controlled by a Micromeritics ASAP 2020Plus gas adsorption analyzer. In a typical experiment, a sample of the activated framework was dispersed in diamond powder (10 wt %) and evacuated at 120 °C before dosing. Known pressures of CO₂ (99.998%) or ¹³CO₂ (Sigma-Aldrich, 99 atom % ¹³C, <3 atom % ¹⁸O) were dosed into the sample using a Micromeritics ASAP 2020Plus gas sorption analyzer. Spectra at 4 cm^–1 resolution were generated from 128 scans collected over the course of approximately 35 s and collected at 1 min intervals until no further changes were observed. All spectra were processed in pseudoabsorbance units. Difference DRIFTS spectra were generated by subtracting the spectrum of the activated framework from the framework under various dosing conditions, and isotopic difference DRIFTS spectra were generated by subtracting the spectrum of the framework under ¹³CO₂ from the spectrum of the framework under the same pressure of natural abundance CO₂.

Powder X-ray Diffraction Data

Laboratory powder X-ray diffraction data were collected on a Bruker AXS D8 Advance diffractometer with Cu Kα radiation (λ = 1.5418 Å) with sample powders placed on an open-air sample holder. Synchrotron powder X-ray diffraction data were collected at Beamline17-BM-B at the Advanced Photon Source at Argonne National Laboratory using an average wavelength of λ = 0.45399 Å. Activated samples were packed in borosilicate glass capillaries (1.0 mm in diameter) under an N₂ atmosphere before being attached to a custom-designed gas-dosing cell equipped with a gas valve.⁴⁷ These cells were then mounted on the goniometer head and connected to a gas-dosing manifold for in situ diffraction measurements. The sample temperature was controlled using an Oxford CryoSystems Cryostream800. A diffraction pattern was first collected for the activated sample at room temperature, and then the sample was briefly heated to 120 °C under dynamic vacuum before 1 bar of CO₂ was dosed to the framework. The sample was then cooled to 298 K under a CO₂ atmosphere. Diffraction patterns were recorded using a PerkinElmera-Si FlatPanel detector and monitored to confirm that the materials had reached equilibrium under gas-dosing conditions. Diffraction patterns were analyzed using TOPAS–Academic v6.1.⁴⁸

Unit cell parameters for CO₂-dosed pip2–Mg₂(dobpdc) were obtained by structureless Pawley refinement using TOPAS–Academic v6.1.⁴⁸ The backgrounds of the pattern were modeled with Chebyshev polynomial functions. Peak shapes were described with the fundamental parameter approach. Using the parameters obtained by the Pawley refinement, Rietveld refinement was performed with the structural models of CO₂-dosed pip2–Mg₂(dobpdc) obtained from DFT calculations, as discussed (see Figures S13, S15, and S17; see DFT Calculations below for details of the geometry optimization).

The occupancy of CO₂ was first freely refined by using these models, and it was found (as discussed above) that only structure C gave a positive occupancy. For Rietveld refinement of structure C, the positions and atomic displacement parameters of the Mg atom and the atoms of the linker were refined with restraints. The positions and atomic displacement parameters of the C, O, and N atoms of the carbamate group attached to Mg²⁺ were also refined with restraints. For the other atoms of the pip2 moiety, only atomic displacement parameters were refined, while the positions were not refined. The positions were fixed based on the DFT structure, the alkyl chain and the six-membered ring of pip2 are expected to be flexible at the analyzed temperature (298 K), and it may not be reasonable to determine the conformation of pip2 from these data. The position, atomic displacement parameters, and occupancy of physisorbed CO₂ were refined with restraints on the C–O bond and O–C–O angle, while the central carbon was freely refined at a Wyckoff position 3b between carbamates. The hydrogen atoms were added to the dobpdc^4– linker, ammonium carbamate, and pip2 after refinement with a C–H distance of 1.09 Å and a N–H distance of 1.02 Å using the structure edit tool of Mercury. Although the conformation of the ethylene moiety and the six-membered ring of pip2 was not refined owing to their flexibility, the refinement of the Mg₂(dobpdc) framework and the physisorbed CO₂ gave a low R_wp value (6.58%). This analysis supports the idea that the refined structure is a reasonable model for the approximate position of the physisorbed CO₂. Note, however, that the bond lengths, bond angles, and atomic displacement parameters of the pip2 moiety are not captured in this model.

DFT Calculations

First-principles density functional theory (DFT) calculations were performed using GBRV pseudopotentials⁴⁹ and the revised Perdew–Burke–Ernzerhof (RPBE)⁵⁰ exchange-correlation functional with the Quantum ESPRESSO plane wave DFT code.⁵¹ To include the effect of the van der Waals (vdW) dispersive interactions on energetics, we performed structural relaxations with Grimme’s D3 correction for all calculations, as implemented in Quantum ESPRESSO.⁵² Additionally, for all calculations, we used (i) a 1 × 1 × 3 k-point grid sampling, (ii) a 60 Ry plane-wave cutoff energy, and (iii) a 600 Ry charge-density cutoff energy. We explicitly treated 10 valence electrons for Mg (2s²2p⁶3s²), 6 for O (2s²2p⁴), 5 for N (2s²2p³), 4 for C (2s²2p²), and 1 for H (1s¹). Using the above input parameters and initial structures obtained from the Rietveld refinement, we fully relaxed both lattice parameters and internal coordinates. The ions were relaxed until the force is less than 1 × 10^–4 Ry/Bohr.

Isotropic NMR chemical shielding values (δ_iso) were computed using the linear response method of Yates et al.⁵³ implemented in the Vienna ab initio Simulation Package (VASP)⁵⁴⁻⁵⁷ according to the equation δ_iso = (−δ_iso-DFT – δ_ref-DFT), where δ_ref-DFT is a reference value from the ¹³C and ¹⁵H chemical shifts of adamantane and glycine, respectively, from prior work.³⁸

To compute CO₂ binding energies, we optimized three different structures: (1) a structure of pip2–Mg₂(dobpdc) prior to CO₂ adsorption (E_MOF), (2) a structure of the framework interacting with CO₂ in the gas phase (E_CO2) within a 15 Å × 15 Å × 15 Å cubic supercell, and (3) a structure of pip2–Mg₂(dobpdc) with adsorbed CO₂ (E_CO2–MOF). The binding energy (E_B) was then obtained by using the following equation:

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.3c13381.

• Additional full experimental characterization data and computational details (PDF)

• Supporting simulation and refined structures (ZIP)

Author Present Address

^○ Department of Chemistry, University of Cambridge, Cambridge CB2 1EW, United Kingdom (A.C.F.)

Author Present Address

^◆ Computational Science Research Center, Korea Institute of Science and Technology (KIST), Seoul 02792, Republic of Korea (J.-H.L.).

Author Present Address

^¶ Department of Chemistry and Chemical Biology, Cornell University, Ithaca, New York 14853, United States (P.J.M.).

Author Contributions

⁺ Z.Z. and H.T. contributed equally and agree that the order of their names may be exchanged as useful to highlight their contributions in individual professional pursuits.

The authors declare no competing financial interest.