Surface Pore Engineering of Covalent Organic Frameworks for Ammonia Capture through Synergistic Multivariate and Open Metal Site Approaches.

Ammonia (NH3) is a commonly used industrial gas, but its corrosiveness and toxicity are hazardous to human health. Although many adsorbents have been investigated for NH3 sorption, limited ammonia uptake remains an urgent issue yet to be solved. In this article, a series of multivariate covalent organic frameworks (COFs) are explored which are densely functionalized with various active groups, such as -N-H, -C=O, and carboxyl group. Then, a metal ion (Ca2+, Mn2+, and Sr2+) is integrated into the carboxylated structure achieving the first case of an open metal site in COF architecture. X-ray photoelectron spectroscopy reveals conclusive evidence for the multiple binding interactions with ammonia in the modified COF materials. Infrared spectroscopy indicates a general trend of binding capability from weak to strong along with -N-H, -C=O, carboxyl group, and metal ion. Through the synergistic multivariate and open metal site, the COF materials show excellent adsorption capacities (14.3 and 19.8 mmol g-1 at 298 and 283 K, respectively) and isosteric heat (Qst) of 91.2 kJ mol-1 for ammonia molecules. This novel approach enables the development of tailor-made porous materials with tunable pore-engineered surface for ammonia uptake.

Introduction

Industrial development has led to a great demand for ammonia as a chief raw material, with annual production exceeding 200 million tons.[1] However, its strong causticity and toxicity give rise to the identification of threats from industrial emissions and spills, which can cause convulsions, coma, and death to all vertebrates.[2] To eliminate such hazards, engineered materials that can adsorb large amounts of ammonia are desperately needed. Although numerous adsorbents, such as activated carbons, zeolites, and metal–organic frameworks (MOFs), have shown potential for applications in NH3 sorption, the progressive performance of ammonia uptake remains a challenging issue.[3−7] To address this objective, two qualities are considered as vital ingredients for an efficient adsorbent:[8] (1) a highly porous framework consisting of unique microsized pores that can increase the physical adsorption of ammonia; and (2) effective functional groups in the porous structure that can interact with ammonia and favor chemical adsorption.

Covalent organic frameworks (COFs) are an emerging class of crystalline porous materials that are composed of light elements (e.g., H, B, C, N, O) via covalent bonds.[9−15] Using state-of-the-art molecular design, COF-based structures can exhibit a large surface area, unique porosity, high crystallinity, and tunable pore chemistry, with promising applications in the fields of catalysis, sensing, optoelectricity, and gas adsorption and separation.[16−20] Therefore, one can create tailor-made binding sites through suitable selection of building block or modification strategies for target functions. Desirable properties, including extensive pore surface area and tunable surface chemistry, attract great interest for the development of novel COF-based adsorbents for NH3 sorption.

Herein, we present a series of multivariate COF materials to integrate various binding groups into the porous network for surface pore engineering. A three-component system comprising pan class="Chemical">triformylphloroglucinol (TFP), 2,5-diaminobenzoic acid (DAA), and p-phenylenediamine (PA-1) at different molar ratios (X = [DAA]/([DAA] + [PA – 1]) × 100 = 0, 17, 33, 50, and 100) is explored to prepare [HOOC]-COFs (X = 0, 17, 33, 50, and 100), as shown in Figure . The [HOOC]17-COF, with its remarkable NH3 adsorption capacity, was then poured into the chloride salt to anchor Lewis centers (Ca2+, Mn2+, and Sr2+) onto the pore surface. The metallized COF product first obtains the open metal site to enhance the chemical adsorption affinity for NH3 molecules by forming acidic–basic adducts. X-ray photoelectron spectroscopy (XPS) and infrared (IR) spectroscopy provided solid evidence of the multiple interactions with ammonia, including the formation of hydrogen bonds with —N—H and —C=O and acidic–basic adducts with carboxyl group and metal ions. Due to the synergistic effect of multivariate components and open metal site, the COF materials exhibited considerable total adsorption capacities for ammonia realized via multiple binding interactions.

Figure 1

(a) Scheme for surface pore engineering of COFs with various groups and (b) possible pore structure of COFs with various groups (gray, C; blue, N; red, O; yellow, metal).

Results and Discussion

In this context, we use TpPa-1 ([pan class="Chemical">HOOC]0-COF) material as a scaffold[21,22] which features a well-defined hexagonal structure with space group (P6/m), high BET surface area of 713 m2 g–1, and unique pore size distribution (PSD) centered at 1.50 nm calculated by the nonlocal–density functional theory (NL–DFT) method. Two particularly salient features of [HOOC]0-COF suggest that it will be an excellent scaffold for the development of novel ammonia adsorbents. (1) The unique microchannel in the porous skeleton favors physical adsorption of ammonia via van der Waals force.[21] (2) Diverse fragments, including —N—H and —C=O group, can serve as proton-donating or proton-accepting groups for hydrogen-bonding interactions, which can significantly enhance the affinity to ammonia.[23−26] As illustrated in Figure and Figure S1, [HOOC]0-COF shows a meaningful NH3 (2.56 Å) capacity of 6.85 and 9.23 mmol g–1 at 298 and 283 K, respectively. The isosteric heat of NH3 adsorption based on the Clausius–Clapeyron relation is determined to be 43.5 kJ mol–1 from independent fits to the 283 and 298 K isotherms.

Figure 2

NH3 adsorption (solid symbols) and desorption (open symbols) at 1 bar and 298 K for activated samples of [HOOC]0-COF (a), [HOOC]17-COF (b), [HOOC]33-COF (c), [HOOC]50-COF (d), [HOOC]100-COF (e), [CaOOC]17-COF (f), [MnOOC]17-COF (g), and [SrOOC]17-COF (h).

Compared with [HOOC]0-COF, the carboxyl group is introduced into the porous skeleton to produce the multivariate COFs by use of different ratios of linkers. The presence of the carboxyl group in [HOOC]-COFs was confirmed by FT-IR spectroscopy. As shown in Figure S2, the consumption of starting materials can be observed from the appearance of —N—H bending at 1520 cm–1 and —C=O stretching at 1625 cm–1 in the COF skeleton. The main bands at 989 and 1657 cm–1 are assigned to the plane bending vibration of —O—H and stretching vibration of —C=O in the carboxyl group, respectively. CHN elemental analysis reveals the contents of the carboxyl group in [HOOC]-COFs (X = 0, 17, 33, 50, and 100), with the measured result in concert with theoretical calculations (Table S1). The architectural stability is manifested by thermogravimetric (TG) analysis, with no weight loss found for each COF until decomposition at 250 °C (Figure S3). Powder X-ray diffraction (PXRD) patterns show the same ordered structure as that found for TpPa-1, but the crystallinity of [HOOC]-COF (X = 0, 17, 33, 50, and 100) decreases for increasing X, as shown in Figure S4. Scanning electron microscopy (SEM) images show that [HOOC]-COFs (X = 0, 17, 33, 50, and 100) are crystallized as microsized (1–5 μm) aggregated particles (Figure S5). The transmission electron microscopy (TEM) images show that some of the COF samples exhibit a sheetlike structure due to π–π stacking of the COF layers (Figure S6).

The porosity of the COFs was measured by N2 adsorption isotherms for the fully activated samples. The surface areas for the [HOOC]-COFs with carboxyl group content (X) varying from 0 to 17, 33, 50, and 100 was calculated by the BET model to be 713, 652, 458, 279, and 150 m2 g–1, respectively (Figure S7). This descending trend in the surface area is due to the electron withdrawing efficiency of the carboxyl group, which reduces the activity of the coupling reaction during the polymerization process leading to ruptures in the integration of the COF structure.[27,28] The pore sizes varied from 1.50 to 1.48, 1.44, 1.11, and 0.97 nm as the amount of carboxyl group increased from 0 to 17, 33, 50, and 100, respectively (Figure S7).

[HOOC]-COFs with X values of 17, 33, 50, and 100 showed an NH3 capacity of 9.34, 8.21, 6.67, and 4.14 mmol g–1, respectively, at 298 K and 1 bar (Figure ). These decreased NH3 uptakes are attributed to the reduced porous surfaces and lack of long-range order in the [HOOC]-COFs. The [HOOC]17-COF was selected for further investigations despite having a lower surface area (652 m2 g–1) compared to [HOOC]0-COF (713 m2 g–1); we believe this COF should show a strong attractive interaction between the carboxyl group (−COOH) and NH3 molecule,[8,29] this hypothesis is confirmed by the enhanced isosteric heat of NH3 adsorption (55.3 kJ mol–1) measured at 298 and 283 K (Figure S8). Therefore, this result clearly demonstrates the vital role of the additional functional group (−COOH) in the COF pore surface.

The [HOOC]17-COF with the best pan class="Chemical">NH3 uptake is selected as a subject for further surface engineering to improve the performance. Previous studies reported that Lewis centers, such as Ca2+, Mn2+, and Sr2+ ions on the surface of a porous framework, may create a strong affinity for NH3 molecules.[3−8] Therefore, respective metal ions are incorporated into the [HOOC]17-COF architecture to achieve the open metal site for a larger storage capacity by chemical adsorption. After immersing into each chloride salt solution, a series of [MOOC]17-COF materials were obtained: [CaOOC]17-COF, [MnOOC]17-COF, and [SrOOC]17-COF. Infrared spectroscopy (Figure S2) for these materials presents weak bands in the scope of 502–585 and 444–472 cm–1, which are ascribed to the stretching vibrations of M—O and M—N, respectively.[30] The existence of —O—H signal at 989 cm–1 and a reduced intensity for the —C=O band in the carboxyl group (1657 cm–1) of [MOOC]17-COFs suggest the coordination of —C=O and —N—H groups to the metal ion (Figure ). Based on energy minimization optimization calculated by the Materials Studio (MS) simulation, two O (from —C=O of —COOH and COF skeleton) and one N (from —N—H) atoms coordinate to one metal ion in the [MOOC]17-COF framework to form a stabilized double six-membered ring structure (Figure ), which is consistent with the conclusion of a previous investigation.[31]

Elemental analysis reveals the almost equal content of metal ion and —COOH group in [MOOC]17-COFs (Ca2+, Mn2+, and Sr2+), which is consistent with the result of the theoretical model (Table S1). The weight losses of 2.7%, 7.3%, and 4.5% before 100 °C are ascribed to the loss of guest molecules (Figure S3). After heating samples up to 800 °C, ca. 3.2%, 4.8%, and 5.9% of residues remain corresponding to the [MOOC]17-COFs (Ca2+, Mn2+, and Sr2+), which is in agreement with the results of the elemental analysis. PXRD patterns reveal that the [MOOC]17-COFs (Ca2+, Mn2+, and Sr2+) retain the same crystallinity as the [HOOC]17-COF even after metal ion incorporation (Figure S4). As shown in Figure S5, SEM images for the [MOOC]17-COF (Ca2+, Mn2+, and Sr2+) also show aggregated particles. The metal ions are uniformly dispersed in the COF structure, as indicated by no metallic nanoparticles visible in the TEM images (Figure S6). The BET surface areas are determined to be 629, 614, and 587 m2 g–1 for [CaOOC]17-COF, [MnOOC]17-COF, and [SrOOC]17-COF, respectively (Figure S9). This explicit decrease in area results from the molecular weight being scaled up along with incorporation of the Ca2+, Mn2+, and Sr2+ ions.

Upon introduction of the Ca2+, pan class="Chemical">Mn2+, or Sr2+ ions onto the pore surface, the resulting [MOOC]17-COFs with open metal sites present higher sorption amounts (at 298 K) of 12.25, 11.38, and 14.30 mmol g–1, respectively. The strong affinity of open metal site is also demonstrated by the sharp increase of uptake capacity at low pressure range (Figure S10). Likewise, after the desorption processes, the three COF materials exhibit a hysteresis terminating with uptakes of 1.53, 0.85, and 1.99 mmol g–1 for [CaOOC]0-COF, [MnOOC]17-COF, and [SrOOC]17-COF, respectively. After calculation, we speculate that each Ca2+, Mn2+, and Sr2+ ion can uphold 2.01, 1.69, and 2.65 NH3 molecules, respectively. The Sr2+ ion possesses the highest uptake capability because of its specific electronic structure, which favors formation of a multiple (a maximum of 6) coordinated complex with NH3 molecules.[32−35] After interaction with NH3 molecules, the tricoordinated Sr2+ ion in the double six-membered ring structure can combine with three more ammonia molecules to reach its maximum coordination structure, which is basically identical to the experimental result of 2.65 NH3 molecules.

[SrOOC]17-COF possesses an ultrahigh capacity of 19.8 mmol g–1 at 283 K and 1 bar (Figure S11). After introduction of the pan class="Chemical">Sr2+ ion, the [SrOOC]17-COF shows a significant improvement in the isosteric heat for NH3 gas (91.2 kJ mol–1). [SrOOC]17-COF reveals a decent capacity (10.92 mmol g–1) after three consecutive adsorption/desorption cycles (Figure S12). Such a decrease in NH3 uptake is due to strong acid–base interactions that unite for a certain amount of ammonia in the framework, limiting desorption under vacuum conditions. The TGA result indicates that [SrOOC]17-COF will desorb ammonia molecules at ∼150 °C (Figure S13) while retaining its crystalline structure for cycle use. The bound ammonia calculated from ca. 3.4% weight loss among 120–170 °C in the TGA curve also agrees with the sorption calculation that each Sr2+ ion can uphold 2.73 NH3 molecules. After heated at 200 °C under vacuum for 12 h, [SrOOC]17-COF would release the bound ammonia molecules and reabsorb ca. 14.12 mmol g–1 NH3 molecules (Figure S14). The capacity decreased by only 7% after five cycles indicating that [SrOOC]17-COF exhibits good reusability (Figure S15). The large adsorption capacity of [SrOOC]17-COF (∼14.30 mmol g–1 at 298 K) is much higher than those of most other adsorbents such as MCM-41 (7.9 mmol g–1), 13X zeolite (9.0 mmol g–1), Amberlyst 15 (11.0 mmol g–1), Co2Cl2(BTDD)(H2O)2 (12.0 mmol g–1), Ni2Cl2(BTDD)(H2O)2 (12.0 mmol g–1), and PPN-6-SO3H (12.1 mmol g–1) and is close to the highest level of ammonia adsorbents, including Mn2Cl2(BTDD)-(H2O)2 (15.5 mmol g–1), BPP-5 (17.7 mmol g–1), and Cu2Cl2BBTA (19.79 mmol g–1).[3−8,10]

Usually, the absorptivity of a porous material is related to the surface area and pore volume; however, none of these correlations hold in our comparative investigations. For comparison, the classical porous aromatic framework (PAF-1) with ultrahigh surface area (4240 m2 g–1 based on a BET model) and uniform pore channel (1.41 nm) was demonstrated to contrast the impacts of porosity and aromatic rings.[36] With similar pore size but larger surface area, PAF-1 exhibits ∼0.2 mmol g–1 capacity at 298 K.[8] It is evident that the adsorption capability of COF materials in this work is highly dependent on the synergistic effect of multivariate components and open metal site. XPS study affords a surface analytical technique that can be used to accurately determine the chemical state of each element. [HOOC]0-COF, [HOOC]17-COF, and [SrOOC]17-COF samples for XPS were prepared by first heating to 473 K for 24 h under vacuum, followed by exposure of the activated COFs to a dry NH3 environment for 72 h at 283 K (Figure ).

Figure 3

XPS spectra for [HOOC]0-COF (a, b), [HOOC]17-COF (c, d), and [SrOOC]17-COF (e–g). Black curves represent activated samples, and red curves represent exhausted samples following interaction with NH3.

As illustrated in Figure , ammonia with a pan class="Chemical">nitrogen 1s binding energy of 397.9 eV could not be isolated possibly because of the weak signal of gas molecules.[37] The binding energies for N 1s and O 1s in the [HOOC]0-COF center are at 400.0 and 532.3 eV, respectively (Figure a,b). After NH3 (∼397.9 eV) gas adsorption, these binding energies change to ca. 399.6 eV for N 1s and 532.0 eV for O 1s, which manifests the barely exposed H nuclei in NH3 via the hydrogen bond that pulls the electron cloud of the donor atom (N and O) of the framework resulting in a decrease of the binding energies.[38,39] In [HOOC]17-COF (Figure c), N 1s photoemission in the activated sample occurs at 400.0 eV and is decreased to 399.6 eV after exposure to an NH3 environment due to the formation of hydrogen bonds. A shoulder peak for N 1s is located at 401.0 eV due to protonation of NH3, which can be used to identify the formation of the ammonium (NH4+) salt.[40] The reducing binding energies at 513.5 and 533.2 eV can be ascribed to the O atoms in —C=O and —COOH, respectively (Figure d). The high-resolution O 1s spectrum observed at 532.3 eV is resolved into two characteristic peaks for —C=O in H bond and —COO– in the ammonium salt. For [SrOOC]17-COF (Figure e), the N 1s peak component at 400.0 eV is shifted by −0.4 eV compared to the reacted N atom, which is clearly evident of hydrogen bond formation, with protonation of NH3 also shown at 400.8 eV because of the formation of the ammonium salt. Coordination of the O atom to the Sr2+ ion is demonstrated by the lower shifted peak (530.9 eV) for —C=O and the weakened intensity (533.2 eV) for the —COOH (Figure f). The N 1s and O 1s changes for [SrOOC]17-COF in XPS spectra agree with the formation of the stabilized double six-membered ring structure. As for the exhausted [SrOOC]17-COF, the strong signal observed at 532.3 eV fits perfectly with the two characteristic peaks corresponding to —C=O in H bond and —COO– in the ammonium salt. It is evident from the data shown in Figure g that a coordinated structure for Sr2+ with ligand atoms is formed, as the binding energy of Sr 3d reducing from 134.5 eV (SrCl2)[41] to 134.2 eV ([SrOOC]17-COF). This energy shift is due to capable elements clinging to the Sr2+ surface, and withdrawal of the electron cloud from the Sr atom to the coordinated O and N atoms themselves, leading to an enhanced shielding effect that decreases the binding energy.[42−44] After the coordination of NH3 to Sr2+ ion, the electron cloud of the Sr atom is pulled further toward the N atoms, resulting in a decrease in the binding energy for the Sr atom (133.7 eV) in the NH3-adsorbed [SrOOC]17-COF.

The effect of each group is assessed by IR spectroscopy at different temperatures to qualitatively indicate the affinity (Figure ). After adsorption of NH3 molecules, the respective COF samples were heated from 283 K, with the temperature increased in 10 degree steps and constant holding of the temperature for 10 min before further heating. The effect of temperature for the exhausted sample was monitored by an infrared detector until the temperature reached 423 K.

Figure 4

NH3 adsorption against different temperatures at 1 bar for activated samples of [HOOC]0-COF (a), [HOOC]17-COF (b), and [SrOOC]17-COF (c).

From the IR spectroscopy, several features are observed as the exhausted sample desorbs NH3 from the framework at different temperatures. Here, the —C—N bond (1255 cm–1) is selected as the internal standard; variation in peak intensity is confirmed based on the ratio of the intensity of the characteristic bond to the —C—N. As observed in Figure a, the value (relative intensity for the —N—H bond compared to the —C—N bond) increases from 1.0 in the original [HOOC]0-COF to ca. 1.2 for the adsorbed-ammonia COF material (283 K), which indicates interaction between the —N—H group and ammonia molecule.[45] Moreover, the ratio changes back to a value of 1.0 after the sample is heated up to 303 K. The data show a wide band (1630 cm–1) for —C=O in the [HOOC]0-COF structure; following adsorption of ammonia at 283 K, the location of the maximum for the —C=O band toward lower wavenumber (1614 cm–1) can be attributed to the formation of hydrogen bonds. With increasing temperature to 323 K, desorption of ammonia leads to reversion of the band position back to 1630 cm–1. Similarly, we utilize the FT-IR spectra to probe the interactions between [HOOC]17-COF and the ammonia adsorbate before and after adsorption (Figure b). In addition to the two characteristics for [HOOC]0-COF, a new feature at 1430 cm–1 is observed for [HOOC]17-COF exposed to ammonia, which can be assigned to the vibration of —N—H in NH4+.[46,47] This observation suggests that a fraction of the ammonia molecules are converted into the ionic form via the acid–base reaction with the carboxylic group. This band disappeared at 363 K, indicating the removal of ammonia. As to the [SrOOC]17-COF (Figure c), another band at 1496 cm–1 is ascribed to the existence of a scissoring mode for the amide species, —NH2, which is formed from the coordination of the Sr2+ ion and ammonia.[48,49] The relative intensity of the —NH2 bond decreases gradually from 393 K and ultimately disappears at 423 K. One can speculate that the binding capability of —N—H, —C=O, carboxyl group, and metal ion trends from weak to strong.

Conclusions

In summary, surface pore engineering of COF materials is explored via the incorporation of various functional units on pore walls. Due to the synergistic multivariate and open metal site, the porous architecture profoundly enhances the binding affinity for ammonia molecules. Notably, the role and tendency of the —N—H, —C=O, —COOH, and metal ions in interactions with ammonia can guide the functionalization of other porous materials for separation and adsorption of ammonia. Additionally, we expect our approach based on the “surface pore engineering” will motivate research on utilizing COF materials in other fields, with further progress eventually leading to industrial applications.

Experimental Section

Synthesis of [HOOC]-COFs

2,5-Diaminobenzoic acid (DAA) and p-phenylenediamine (PA-1) were added into a solution of triformylphloroglucinol (TFP) (63 mg, 0.3 mmol), mesitylene (1.5 mL), dioxane (1.5 mL), and 3 M aqueous acetic acid (0.5 mL). Molar ratios for DAA and PA-1 of 0:6, 1:5, 2:4, 3:3, and 6:0 in total amount of 0.9 mmol were used to produce [HOOC]0-COF, [HOOC]17-COF, [HOOC]33-COF, [HOOC]50-COF, and [HOOC]100-COF, respectively. The mixture was then frozen by using a liquid N2 bath (77 K) and degassed via three freeze–pump–thaw cycles. After heating at 120 °C for 3 days, the precipitate was collected by filtration and washed with copious amounts of anhydrous dimethylacetamide (DMAC) and acetone. Finally, the resulting red powder was dried under vacuum at 180 °C for 24 h yielding a series of [HOOC]-COF samples.

Preparation of [MOOC]17-COFs

A 0.100 g portion of activated [HOOC]17-COF with the −COOH group (0.65 mmol g–1) was poured into a chloride salt (CaCl2, MnCl2, and SrCl2) aqueous solution (15.6 mmol L–1, 5 mL), respectively. After the solution was stirred for 24 h, the insoluble product was filtered out. Then, the precipitate was thoroughly washed with water three times to afford [MOOC]17-COFs (Ca2+, Mn2+, and Sr2+).

NH3 Release Experiment against Temperature

[HOOC]0-COF, [HOOC]17-COF, and [SrOOC]17-COF were heated under vacuum to approximately 473 K for 12 h and then cooled to room temperature. Next, the activated COFs were exposed to a dry NH3 environment for 24 h at 283 K. The respective COF samples were then heated to 283 K, with the temperature increased 10 degrees at a time and held constant for 10 min at each step before further heating. The effect of temperature for the exhausted sample was monitored by an infrared detector until the temperature reached 423 K.

General experiments for the complete physical characterization of all as-synthesized COF materials are available in the Supporting Information.

No unexpected or unusually high safety hazards were encountered.