Polyethyleneimine-Modified UiO-66-NH₂(Zr) Metal–Organic Frameworks: Preparation and Enhanced CO₂ Selective Adsorption

Abstract

UiO-66-NH₂, a zirconium-based functional metal–organic framework (MOF), was postsynthetically modified via Schiff base reaction between aldehyde groups in glutaraldehyde and amino groups in UiO-66-NH₂ and CO₂-preabsorbed polyethyleneimine (PEI). The resulting PEI-modified MOFs, abbreviated as PEI^C@UiO, were characterized with ¹H NMR, Fourier transform infrared, powder X-ray diffraction, Brunauer–Emmett–Teller, scanning electron microscopy, and thermogravimetric analysis and evaluated as CO₂ adsorbents. In comparison with pristine UiO-66-NH₂, the PEI^C@UiO adsorbents have reduced specific surface area (7–150 m²/g) but maintained the same crystal structure. Particularly, the PEI^C96@UiO adsorbent exhibited significantly improved CO₂/N₂ adsorption selectivity (48 vs 25) and higher CO₂ adsorption capacity (3.2 vs 2.7 mmol/g). The adsorbent also displayed moderate desorption energy (68 kJ/mol CO₂), superior moisture endurance, and recyclability, which are very desirable for practical applications.

1. Introduction

To control the concentration of carbon dioxide (CO₂) in the air, carbon capture, utilization, and storage have received extensive attention in recent years.¹⁻³ Currently, the chemical absorption using aqueous organic amines like monoethanolamine as absorbent is the most important method for CO₂ capture from industrial waste gases.⁴⁻⁶ However, some inherent drawbacks such as degradation or volatilization of organic amines, equipment corrosion, and high energy consumption limit its practical application. Therefore, various kinds of solid porous adsorbents have been proposed for CO₂ capture.

Metal–organic frameworks (MOFs) are three-dimensional porous crystalline materials made up of metal ions or clusters coordinated to organic linkers.⁷⁻¹⁰ As a new type of promising adsorbent materials,^11,12 MOFs have shown enormous potential for CO₂ capture due to large specific surface area, high crystallinity, tunable chemical structure, and pore properties.⁸ However, it remains a challenge to design MOFs with quite high selectivity as well as CO₂ adsorption capacity because most MOFs do not possess high CO₂ adsorption selectivity. Therefore, much research has been done to improve the CO₂ adsorption selectivity of MOFs, including incorporation of open metal sites,¹³ construction of size/shape with specific pores,¹⁴ ligand functionalization to introduce strongly polar functional groups,¹⁵ and amine grafting.¹⁶

As a zirconium(IV)-based MOF, UiO-66 has attracted great interest because of its superior thermal and chemical stability.¹⁷ By functionalizing the ligand of UiO-66 with polar groups such as −NH₂, −Br, −NO₂, −(CF₃)₂, −SO₃H, and −CO₂H,¹⁸⁻²⁰ the CO₂ adsorption selectivity was improved while retaining the stability, especially when amino group was used as the functional group.²⁰ Cmarik et al. reported that UiO-66-NH₂ showed enhanced CO₂ adsorption capacity (2.98 vs 1.75 mmol/g) and adsorption selectivity over N₂ (13.2 vs 12.5) at 298 K and 1 bar²⁰ when compared with UiO-66. The selectivity obtained by applying ideal adsorption solution theory (IAST) for a 15:85 CO₂/N₂ gas mixture was 52.²⁰ Higher CO₂ adsorption capacity (3,²¹ 3.5²² mmol/g) and CO₂/N₂ selectivity (31²¹) of UiO-66-NH₂ were reported by Molavi et al.²¹ and Huang et al.²² Improvement of CO₂ capture performance of UiO-66 by polyethyleneimine (PEI) impregnation²³ or ethanolamine grafting²⁴ was also reported.

For UiO-66-NH₂, postsynthetic modification can further improve the CO₂ capture performance.^21,25,26 Molavi et al. modified UiO-66-NH₂ with glycidyl methacrylate (GMA) by epoxy-amino reaction, resulting in improvement in CO₂ adsorption capacity from 3.0 to 4.3 mmol/g and CO₂/N₂ selectivity from 31 to 46.²¹ The resulting GMA-UiO-66 was further functionalized with ethylenediamine through aza-Michael addition reaction to improve CO₂ adsorption capacity up to 5.0 mmol/g.²⁵ Piperazine-grafted UiO-66-NH₂ was reported to have IAST selectivity up to 19 for a 15:85 CO₂/CH₄ gas mixture at 298 K at the expense of lower CO₂ adsorption capacity.²⁶ Interaction between amines with the open metal sites of MOFs has also been reported to improve CO₂ adsorption capacity of MOFs at low pressure and room temperature, such as N,N′-dimethylethylenediamine (mmen)/[Mg₂(dobpdc)],²⁷ ethylenediamine/ZIF-8,²⁸ tetraethylenepentamine (TEPA)/MIL-101-NH₂,²⁹ polyethyleneimine (PEI)/MIL-101,^30,31 PEI/MIL-101-NH₂,³²³² and PEI/HKUST³³ composites. In all these amine-modified MOFs, the MOFs act as supports with large specific surface area and the amine groups adsorb CO₂. The synergism between them results in enhancement of the CO₂ adsorption capacity and selectivity. PEI is a suitable polyamine with highly dense amine groups, especially accessible primary amine sites at chain ends, which can react with CO₂ by forming carbamates.³⁴ But as a viscous liquid, PEI itself is not a suitable candidate for CO₂ adsorption. Therefore, PEI is usually supported onto porous materials such as zeolites,³⁵ microporous/mesoporous silica,³⁶ and porous polymers³⁷ as well as MOFs^23,30−33 to improve their CO₂ sorption. Because of the higher porosity and specific surface area, MOFs seems to be more suitable for PEI incorporation to reach high CO₂ adsorption capacity.

In this work, we report PEI-modified UiO-66-NH₂ in an attempt to prepare CO₂ adsorbent with high adsorption capacity, kinetics, and selectivity. UiO-66-NH₂ was postsynthetically modified with PEI in the presence of glutaraldehyde (GD) via Schiff base reaction. The adsorbents were characterized with ¹H NMR, Fourier transform infrared (FTIR), powder X-ray diffraction (PXRD), scanning electron microscopy (SEM), and thermogravimetric analysis (TGA). CO₂ adsorption was tested under different conditions to evaluate the adsorbents for potential application in CO₂ capture from flue gas.

2. Results and Discussion

2.1. Synthesis and Structural Characterization

Powdery UiO-66-NH₂ was prepared according to a previously reported procedure and then postsynthetically modified with PEI in the presence of GD, which linked UiO-66-NH₂ and PEI together through Schiff base reaction, as illustrated in Scheme 1. He et al.⁴² reported that a CO₂-imprinted solid amine adsorbent prepared by cross-linking CO₂-preabsorbed PEI with glutaraldehyde showed a 25% higher CO₂ adsorption capacity. In this study, CO₂-preabsorbed PEI was also used in the postsynthetic modification to prepare three adsorbents named as PEI^C@UiOx, in which x (48, 72, and 96%) means mass percentage of PEI based on UiO-66-NH₂. For comparison, PEI72@UiO was prepared using PEI without CO₂ preabsorption and PEI96+UiO was prepared by impregnation of PEI in the absence of glutaraldehyde. Detailed preparation conditions and structural characteristics of the adsorbents are summarized in Table 1. The PEI content in the adsorbent (PEI_exp) was calculated by the following formula

where W is the weight gain based on the unit mass of UiO-66-NH₂, N is the nitrogen content of the adsorbent measured by elemental analysis, and N(PEI) is the theoretical nitrogen content (30.74 wt %) of PEI. For the three PEI^C@UiO samples, PEI content (10.6–22.4 wt %) increased with increasing PEI feeding (48–96 wt %) while the color changes from yellow to grayish brown. The weight gain is less than the PEI feeding for all the samples. It seems to be independent of CO₂ preabsorbing PEI (PEI^C72@UiO vs PEI72@UiO) or the presence of glutaraldehyde (PEI^C96@UiO vs PEI96+UiO). But, the nitrogen content of PEI96+UiO is clearly higher than that of PEI^C96@UiO because both PEI and nitrogen-free glutaraldehyde are gained in PEI^C96@UiO but only PEI is gained in PEI96+UiO. From these results, it is concluded that only partial PEI has been incorporated onto the surface of the MOF, especially in the chemically modified PEI@UiO and PEI^C@UiO series. The incomplete chemical grafting of PEI may be attributed to the steric hindrance imposed by partial pendant −NH₂ groups, which point to the MOF cavities other than those outside.³⁸ In other words, these −NH₂ groups in the MOF did not contribute to graft PEI. Another possibility is that partial glutaraldehyde reacted with the MOF or PEI itself, resulting in intra-cross-linking of MOF or PEI instead of PEI grafting onto the MOF.

Scheme 1. Schematic Diagram for the Synthesis and Hypothesized Structure of the PEI-Modified UiO-66-NH₂ Adsorbents (PEI^C@UiO).

Table 1. Preparation Conditions and Results of Pristine and PEI-Modified UiO-66-NH₂.

sample	PEItheoa (wt %)	PEItheob (wt %)	Wc (wt %)	Nd (wt %)	PEIexpe (wt %)	Apf (m2/g)	Vpf (cm3/g)	Dpf (nm)
UiO-66-NH2	0	0	0	5.0	0	987	0.51	2.05
PEIC48@UiO	48	26	41	6.8	10.6	151	0.14	3.58
PEIC72@UiO	72	34	64	9.0	19.4	12.0	0.076	25.4
PEIC96@UiO	96	41	90	9.5	22.4	7.64	0.11	55.5
PEI72@UiO	72	34	62	8.4	17.3	14.4	0.096	26.5
PEI96+UiO	96	49	89	16.0	43.4	1.40	0.0034	9.60

^aThe mass percentage of PEI used in adsorbent synthesis, based on the mass of UiO-66-NH₂.

^bThe mass percentage of PEI used in adsorbent synthesis, based on the sum of UiO-66-NH₂, PEI, and GD fed.

^cWeight gain based on the mass of UiO-66-NH₂, calculated from the weight change before and after the modification reaction.

^dNitrogen content measured by elemental analysis.

^ePEI content in adsorbent, calculated by nitrogen content.

^fA_p: specific surface area; V_p: total pore volume; D_p: average pore size.

Figure 1 shows the ¹H NMR spectra of PEI-modified and PEI-impregnated UiO-66-NH₂ adsorbents after digestion in NaOH/D₂O solution as well as PEI itself. In UiO-66-NH₂, as expected, the three proton signals of 7.02, 7.09, and 7.53 ppm are attributed to the benzene ring structure (3 × 1H) of organic ligand in MOFs. The peak at 2.07 ppm is attributed to Zr–OH⁴³ formed during digestion. The reason of the peak at 8.27 ppm is not clear yet. It also exists in the spectrum of digested UiO-66-NH₂ reported previously.⁴⁴ After PEI modification, the multiple broad proton signals of −CH₂ in PEI and glutaraldehyde appeared at 2.3–2.6 ppm. The peak intensity shows an increasing trend while the PEI loading increased. However, −N=CH bond formatted by Schiff base reaction of glutaraldehyde was not observed in PEI@UiO & PEI^C@UiO. It was possible that −N=CH bonds were unstable under alkaline conditions and ruptured.

Figure 1.

¹H NMR spectra of PEI, UiO-66-NH₂, and PEI-modified MOFs. The samples except PEI were digested for 24 h in a 1 M solution of NaOH/D₂O before NMR measurement.

The FTIR spectra shown in Figure 2 provided more information for the products. For UiO-66-NH₂, a strong peak at 1658 cm^–1, assigned to the v(C=O) stretching of N,N-dimethylformamide (DMF) can be seen. The absence of this band in PEI^C@UiO-66 indicates the complete exchange of DMF with water.⁴⁵ The 3000–2800 and 3600–3100 cm^–1 signals are assigned to the stretching vibration of CH₂ and unreacted amino groups, respectively. They became stronger when more PEI was incorporated. The characteristic stretching vibration peaks of NH₂ on the MOF appeared at 1436 and 1258 cm^–1. The two peaks were weakened because of its Schiff base reaction with aldehyde group in glutaraldehyde to form N=C bond at 1618 cm^–1 (stretching vibration), resulting PEI grafting shown in Scheme 1. In addition to the grafting reaction, the intra-cross-linking reactions of PEI and UiO-66-NH₂ with glutaraldehyde also occurred unavoidably at the same time. In fact, all these Schiff base reactions contributed to the signal at 1618 cm^–1, but could not be identified from each other in the FTIR results. For simplicity and clarity, Scheme 1 only shows the expected grafting structure of the products.

Figure 2.

FTIR spectra of PEI, UiO-66-NH₂, and PEI-modified MOFs.

The PXRD patterns of UiO-66-NH₂ and the five PEI-modified adsorbents are shown in Figure 3. The as-synthesized UiO-66-NH₂ showed narrow diffraction peaks in accordance with previously published data,^17,38 confirming successful synthesis and high crystallinity of UiO-66-NH₂. The PXRD patterns of PEI-modified MOFs were essentially identical to that for the parent UiO-66-NH₂, indicating that the crystal structure of UiO-66-NH₂ was well maintained after the modification. The peak intensity of UiO-66-NH₂ decreased after PEI modification because of the decrease of MOF content and the enhancement of the interaction between the PEI amino groups and the UiO-66-NH₂ Zr sites.²³ Another reason that cannot be completely excluded is that the MOF framework structure could be destroyed to some extent in alkali amine solution though UiO-66-NH₂ is reported to show excellent alkali resistance.³⁸ Similar structure destruction were reported in some amine-impregnated MOFs, such as MIL-101,³¹ HKUST-1,³³ and MOF-74.⁴⁶

Figure 3.

Powder X-ray diffraction (PXRD) patterns of pristine UiO-66-NH₂ and PEI-modified adsorbents.

The nitrogen adsorption isotherms of the five PEI-modified adsorbents and UiO-66-NH₂ are shown in Figure S1. The specific surface areas calculated from N₂ adsorption isotherms are listed in Table 1. It reaches 987 m²/g for UiO-66-NH₂ but decreases to 151, 12.0, and 7.6 m²/g for PEI^C48@UiO, PEI^C72@UiO, and PEI^C96@UiO, respectively. At the same time, the average pore size gradually increases and the pore volume decreases with increasing PEI loading. The results suggested that PEI modification remarkably decreased the specific surface area of UiO-66-NH₂ due to the pore-filling effect of PEI on the macropores between MOF particles. Whether or not pretreating PEI with CO₂ preabsorption did not show clear influence on the pore characteristics (PEI^C72@UiO vs PEI72@UiO), but impregnation of UiO-66-NH₂ with large amount of PEI (PEI96+UiO) led to serious pore blockage (supported by SEM observation, see next paragraph), making the micropores in MOF crystals no longer accessible for N₂ adsorption. Therefore, PEI96+UiO only maintained a very small surface area of 1.40 m²/g.

To further understand the effect of PEI modification on the pore characteristic, the morphological structure of the samples was further observed by SEM. As shown in Figure 4, the UiO-66-NH₂ appears to be near-spherical crystals. For PEI@UiO or PEI^C@UiO series, the morphology of MOF particles did not show an observable change though there might exist PEI in the intercrystal void of UiO-66-NH₂. But for PEI96+UiO, the surface and intercrystal void of UiO-66-NH₂ crystals are almost fully occupied by PEI. In other words, the UiO-66-NH₂ crystals are embedded in the PEI layers. This phenomenon explains why the specific surface area of PEI96+UiO was very low.

Figure 4.

SEM images of pristine UiO-66-NH₂ and PEI-modified adsorbents.

CO₂ adsorbents must have enough thermal stability to endure high temperature of at least 100–150 °C of flue gas.⁴⁷ The TGA curves of UiO-66-NH₂ and the five PEI-modified adsorbents are shown in Figure 5. The TGA curves of UiO-66-NH₂ showed a two-step weight loss. The initial weight loss at 100–280 °C was ascribed to the removal of adsorbed gas and dehydroxylating of the zirconium oxoclusters.³⁸ The second-step weight loss (after 280 °C) was due to the framework decomposition.³⁸ The PEI-modified adsorbents were all stable up to 200 °C, if we ignore the initial weight loss over 100 °C, which possibly results from absorbed water and gases. At temperature higher than 200 °C, the weight loss was caused mainly by PEI decomposition. With the increase of PEI loading, the weight loss curve moves toward low temperature side. The differences in preparation methods do not show a clear influence on thermal stability. In conclusion, the PEI-modified adsorbents have good thermal stability to endure normal flue gas temperature for CO₂ separation.

Figure 5.

TGA curves of pristine UiO-66-NH₂ and PEI-modified adsorbents (N₂ atmosphere, heating rate 10 °C/min).

2.2. CO₂ Adsorption Capacity, Rate, and Selectivity over N₂

CO₂ adsorption performance of UiO-66-NH₂ and the five PEI-modified adsorbents at 25 °C and 1 atm is shown in Figure 6A. The saturated adsorption capacity of UiO-66-NH₂ reaches up to 2.67 mmol CO₂/g. It is roughly in accordance with some of the reported values (2.86–3.15 mmol CO₂/g).²¹ In comparison, the PEI^C@UiO adsorbents exhibit higher adsorption capacity ranging from 2.75 to 3.26 mmol CO₂/g (Table 2). The NH₂ groups in PEI and UiO-66-NH₂ contributed to chemical CO₂ adsorption, and the pores in UiO-66-NH₂ contributed to physical CO₂ adsorption. However, partial NH₂ groups in UiO-66-NH₂ that had converted to N=C bonds no longer contributed to CO₂ adsorption and offset part of the contribution of PEI to CO₂ adsorption. The amount of incorporated PEI is not positively correlated with the CO₂ adsorption capacity. Among them, PEI^C72@UiO shows the highest adsorption capacity, 3.26 mmol CO₂/g. In comparison with PEI72@UiO, PEI^C72@UiO shows 9% higher CO₂ adsorption capacity (3.26 vs 2.99 mmol CO₂/g). The increment is higher than the experimental error (<3%) but not as high as the value (25%) reported previously.⁴² The saturated adsorption capacity of PEI96+UiO is much lower than that of PEI^C96@UiO (1.87 vs 3.15 mmol CO₂/g) because of its much lower specific surface area.

Figure 6.

CO₂ (A) and N₂ (B) adsorption behaviors of pristine UiO-66-NH₂ and PEI-modified adsorbents.

Table 2. CO₂ Adsorption/Desorption Properties of Pristine UiO-66-NH₂ and PEI-Modified Adsorbents.

	Cs,1bara (mmol CO2/g)
adsorbent	dry	moistb	α	Cs,expc(mmol CO2 + N2/g)	Cs,0.15bard(mmol CO2/g)	Qdese(kJ/mol CO2)
UiO-66-NH2	2.67	2.13	25.5	2.06	1.98
PEIC48@UiO	2.75	2.69	30.9	2.06	2.00	63.3
PEIC72@UiO	3.26	3.43	40.6	2.17	2.12	67.3
PEI72@UiO	2.99	3.02	36.1	2.10	2.04
PEIC96@UiO	3.15	3.33	48.0	2.38	2.33	68.0
PEI96+UiO	1.87	2.69	36.0	1.59	1.55

^aExperimental CO₂ adsorption capacity.

^bThe CO₂ adsorption capacity measured with moist CO₂.

^cExperimental gas adsorption capacity measured in simulated flue gas (14.3 vol % CO₂).

^dCO₂ adsorption capacity calibrated by the separation factor α in simulated flue gas.

^eThe CO₂ desorption heat.

Furthermore, it can be seen that all the PEI@UiO adsorbents exhibit high adsorption rate, reaching saturated adsorption capacity in 15 min. However, PEI96+UiO need much longer time to reach CO₂ saturation due to low specific surface area and consequent slower adsorption rate.

CO₂/N₂ adsorption selectivity is another important indicator to evaluate a solid sorbent for practical use. For this purpose, the adsorption of pure N₂ (60 mL/min) was conducted under the same condition. The results are shown in Figure 6B and listed in Table 2. UiO-66-NH₂ adsorbed N₂ significantly (0.11 mmol/g). But, the PEI^C@UiO adsorbents had a relatively small amount of N₂ adsorption, decreasing to 0.06–0.09 mmol/g. The separation factor α (adsorption selectivity of CO₂ over N₂) was calculated by the formula

where Q_i and P_i (P_i = 1 bar here) are the adsorption capacity and the partial pressure of component i, respectively. The separation factor of UiO-66-NH₂ is 25.5, which is between the reported values in literature (13²⁰–32²¹). After PEI modification, the value of PEI^C@UiO increased to 30.9–48.0. As expected, it was obviously improved with increasing PEI loading.

To simulate flue gas that usually contains about 15 vol % CO₂, a CO₂/N₂ mixture gas (10 + 60 mL/min, CO₂ 14.3 vol %) was also used for CO₂ adsorption. The result is shown in Figure 7. The saturated gas adsorption capacity of PEI^C@UiO was 2.06–2.38 mmol CO₂ + N₂/g. The true CO₂ sorption capacity from the mixed gas (C_s,0.15bar) can be calibrated by the following formula. The adsorption rate and capacity for simulated flue gas both decreased to some extent in comparison with those for pure CO₂⁴⁸ due to lower CO₂ partial pressure. Among them, PEI^C96@UiO reaches the highest adsorption capacity (2.33 mmol CO₂/g).

The PEI^C96@UiO is compared with some other CO₂ adsorbents reported in literature. As shown in Table 3, it can be seen that this functional MOF adsorbent shows a reasonably good CO₂ adsorption performance, especially in CO₂ selectivity.

Figure 7.

CO₂ uptake of pristine UiO-66-NH₂ and PEI-modified adsorbents from CO₂ + N₂ mixture containing CO₂ 15% at 25 °C.

Table 3. CO₂ Adsorption Performance of Different Adsorbents.

		Csc (mmol/g)
adsorbentsa	Tb (K)	1 bar	0.15 bar	αd	αe	refs
PEIC96@UiO	298	3.1	2.3	48		this work
PEI (50%) + silica	348	3.1	3.0			(36)
0.70PEI@PDVB	298	3.2	2.4			(37)
PEI@PS polyHIPE	313	3.5		26		(49)
PEI@poly(GMA)HIPE	313	4.0	3.1	27		(50)
PEI + UiO-66	298	3.3	1.6			(23)
GMA-UiO-66	298	4.3	1.8	46		(21)
TEPA (50%) + NH2-MIL-101	298	3.1				(29)
PEI (2.5%) + HKUST	298	4.1	0.8	2		(32)
en-[Mg2(dobpdc)]	298	4.5	3.5		230	(16)
pyrrolic N-enriched carbon	298	3.6	1.5		115	(51)

^aPAF: porous aromatic framework; PDVB: nanoporous poly(divinylbenzene). PS: poly(styrene divinylbenzene), modified with polyacrylic acid.

^bTemperature.

^cCO₂ adsorption capacity at the indicated CO₂ partial pressure.

^dCO₂/N₂ selectivity: α = nCO₂ (1 bar)/nN₂ (1 bar).

^eCalculated by ideal adsorption solution theory (IAST) model, α = [nCO₂ (0.15 bar)/nN₂ (0.85 bar)] × (0.85/0.15).

2.3. CO₂ Desorption Enthalpy

Differential scanning calorimetry (DSC) was used to determine the CO₂ desorption heat (Q_des). As shown in Figure 8, a strong endotherm (155–191 J/g CO₂-saturated adsorbent) was observed for all the three CO₂-saturated PEI^C@UiO adsorbents due to CO₂ desorption but none for the neat PEI^C96@UiO. The CO₂ desorption occurred in a wide temperature range from 70 to 230 °C, and the endotherm peak temperature appears at 133–144 °C. From the desorption enthalpy value, the desorption heat Q_des (kJ/mol CO₂) of CO₂-saturated PEI^C@UiO can be calculated by the equation Q_des = E/(C_s/(C_sM_CO2/1000 + 1)), where E, C_s, M_CO2 are the desorption enthalpy (J/g), the CO₂ adsorption capacity (mmol/g), and the molar mass of CO₂ (44 g/mol), respectively. The value ranges from 63.3 to 68.0 kJ/mol CO₂, higher than that of pristine UiO-66-NH₂ (ca. 27 kJ/mol).²⁰ The more PEI is incorporated, the higher the CO₂ desorption heat is. The values are located in the range of 60–90 kJ/mol of typical chemsorption⁴⁷ but lower than the typical value, ca. 80 kJ/mol, of alkanolamine aqueous solution.^52,53 Therefore, this CO₂ adsorbent seems competitive in reducing the regeneration energy consumption.

Figure 8.

DSC curves of neat PEI^C96@UiO and three CO₂-saturated PEI^C@UiO adsorbents.

2.4. Effect of Moisture on CO₂ Adsorption

Moisture is a non-negligible factor that affects practical CO₂ capture from flue gas, which usually contains 8–17% water vapor.⁵⁴ Due to the hydrophilicity, the adsorbents unavoidably adsorb water vapor from the flue gas. To determine the water content in adsorbents during CO₂ capture, the water adsorption of PEI^C96@UiO at 25 °C was measured. This hygroscopic adsorbent adsorbed 1.7 wt % water in 15 min. To study the influence of moisture on CO₂ adsorption, dry CO₂ (d-CO₂) and moist CO₂ (m-CO₂) were used to pass through dry adsorbent (d-Sorb) separately. The moist CO₂ gas was obtained by feeding dry CO₂ gas through water. The results are shown in Figure 9.

Figure 9.

CO₂ adsorption of pristine UiO-66-NH₂ and PEI-modified adsorbents using dry or moist CO₂ gas (solid: d-CO₂; hollow: m-CO₂).

As hygroscopic adsorbents, UiO-66-NH₂ showed a 20.2% decrease in CO₂ adsorption capacity in the “m-CO₂ + d-Sorb” case because water vapor adsorption competed with CO₂ adsorption. The existence of water could also affect the CO₂ binding to the open metal sites in MOF.⁵⁵ The adsorption capacity of PEI^C48@UiO slightly decreased by 2.2%, which was within the allowable error range of the adsorption experiment. However, CO₂ adsorption capacity of PEI^C72@UiO, PEI^C96@UiO, and PEI96+UiO increased in the m-CO₂ + d-Sorb case by 5.2, 5.7, and 43.8%, respectively. The rise instead of descent in CO₂ adsorption is attributed to the tertiary amino groups in PEI (about 12 wt %), which might contribute to CO₂ adsorption in the presence of moisture.⁵⁶ Therefore, the PEI^C@UiO adsorbents show excellent moisture endurance in CO₂ adsorption.

2.5. Adsorption/Desorption Cycles

Finally, CO₂ adsorption/desorption cycles of PEI^C96@UiO were conducted to assess the recyclability and durability of the adsorbent. The CO₂ desorption was carried out at 120 °C. The same temperature was also used for the CO₂ desorption of aqueous monoethanolamine absorbent.⁵² Pure N₂ purging was used instead of vacuum to facilitate the desorption experiment, though the latter seemed more suitable for practical CO₂ capture. Figure 10 shows three cycles of CO₂ adsorption/desorption of PEI^C96@UiO. Because of weak chemical bonding between PEI^C96@UiO and CO₂ at high temperature, the adsorbed CO₂ was desorbed rapidly and nearly completely. In each cycle, the adsorbent effectively captured 3.0 mmol CO₂/g. A high adsorption capacity, 97.4%, was retained after three cycles. The preadsorded moisture and N₂ might account for the slight loss in adsorption capacity. In conclusion, PEI^C96@UiO, showed quite steady CO₂ adsorption and superior durability.

Figure 10.

CO₂ adsorption/desorption cycles of PEI^C96@UiO.

3. Conclusions

UiO-66-NH₂, a zirconium-based functional metal–organic framework (MOF), was postsynthetically modified with CO₂-preabsorbed polyethyleneimine (PEI) via Schiff base reaction. The resulting PEI^C@UiO adsorbents contain PEI up to 22.1 wt % or nitrogen up to 9.5 wt %. They show specific surface area of 7–150 m²/g and maintain unchanged crystal structure of the MOF. When compared with pristine UiO-66-NH₂, PEI^C96@UiO shows significantly improved CO₂/N₂ adsorption selectivity (48 vs 25) and higher CO₂ adsorption capacity (3.2 vs 2.7 mmol/g), but retains fast CO₂ adsorption/desorption in spite of its reduced specific surface area. It also displays moderate desorption heat (68 kJ/mol CO₂), superior moisture endurance, and recyclability, which are very desirable for practical applications. Further studies on raising PEI content as well as maintaining high specific surface will be reported elsewhere.

4. Experimental Section

4.1. Materials

Zirconium tetrachloride (ZrCl₄, 98%), branched polyethyleneimine (PEI, 99%, M_w = 600, viscous liquid, theoretical amino group content 32.9 wt %, primary amino group content 12.8 wt %), and glutaraldehyde (GD, 50%) were purchased from Aladdin Chemical Co. Ltd. China. 2-Aminobenzene-1,4-dicarboxylic acid (ABDC or H₂BDC-NH₂, 98%) was supplied by J&K Chemicals. N,N-dimethylformamide (DMF), anhydrous ethanol, anhydrous methanol, sodium hydroxide (NaOH), and hydrochloric acid (HCl, 36.5%) were all analytical grade reagents purchased from Sinopharm Chem. Reagent Co. Ltd. China. All chemicals were used as received. Nitrogen (N₂, ≥99.5%) and carbon dioxide (CO₂, ≥99.995%) were obtained from Jingong Specialty Gases Co., Ltd. China.

4.2. Synthesis of UiO-66-NH₂

[Zr₆O₄(OH)₄(BDC-NH₂)₆] (UiO-66-NH₂) was prepared according to a procedure previously reported,^20,38 with some modifications. In brief, ZrCl₄ (5 mmol, 1.166 g), DMF (15 mL), and HCl (5 mL) were mixed and sonicated for 20 min until all materials were dissolved fully at room temperature. ABDC (5 mmol, 0.9057 g) and the residual DMF (35 mL) were then added and the mixture was sonicated for another 20 min. The homogeneous mixture was heated at 80 °C for 12 h and then at 100 °C for 24 h. The resulting product precipitated as microcrystalline powder. The mixture was cooled down to room temperature and centrifuged at 8000 rpm for 10 min to separate the solid from the solution. After washing three times with DMF (30 mL) and ethanol (30 mL), respectively, and drying under vacuum at 100 °C for 24 h, yellow powder was finally obtained at a yield of 92%.

4.3. Synthesis of PEI-Modified UiO-66-NH₂ Adsorbent (PEI^C@UiO and PEI@UiO)

Scheme 1 depicts the preparation process. PEI (0.375, 0.563, or 0.751 g) was dissolved in water (10.0 mL). CO₂ was fed into the resulting solution at room temperature to prepare CO₂-preabsorbed PEI. After 2 h, UiO-66-NH₂ (0.5 mmol, 0.78 g) was added. After stirring at 400 rpm for 30 min, an aqueous solution of GD (0.6007 g) was added dropwise under stirring at room temperature for another 12 h. The resulting solid was isolated via centrifugation and washed with deionized water three times. After vacuum drying at 100 °C for 24 h to remove absorbed CO₂ and solvent, a blocky solid product was obtained. It was symbolized as “PEI^Cx@UiO”, where UiO and superscript C represented UiO-66-NH₂ and preabsorption of CO₂, x was the mass percentage of PEI fed based on UiO-66-NH₂. An adsorbent was also prepared without CO₂ preabsorption. In this case, the product was named as “PEIx@UiO”. The details of preparation conditions and structural characteristics of the adsorbents are listed in Table 1.

4.4. Preparation of PEI-Impregnated UiO-66-NH₂ Adsorbent (PEI + UiO)

A PEI-impregnated UiO-66-NH₂ adsorbent was also prepared using an impregnation method according to the procedure previously reported for PEI-impregnated UiO-66 adsorbent.²³ PEI (1.5 mmol, 0.716 g) was dissolved in 5 mL anhydrous methanol. UiO-66-NH₂ (0.5 mmol, 0.78 g) powder was added into the solution and the mixture was treated under ultrasonication for 1 h. Then, it was dried under nitrogen atmosphere at 40 °C for 6 h and further dried under vacuum at 100 °C for 12 h. Finally, the obtained sample was symbolized as PEI96+UiO, where UiO represented UiO-66-NH₂ and 96 was the mass percentage of PEI fed based on UiO-66-NH₂.

4.5. Characterization

The samples were characterized by elemental analysis (Flash EA1112, ThermoFinnigan Co.), ¹H NMR (Bruker AC-80 spectroscopy, 400 MHz, NaOH/D₂O solution), and FTIR (Nicolet 5700). Before ¹H NMR measurement, the PEI-modified samples (20 mg) were digested with 0.6 mL of a NaOH/D₂O solution (1 M) and allowed to stand for 24 h. After the digestion process, the organic portions of the samples (linkers in the MOF and grafted amines) were dissolved, and the inorganic component of the MOF was converted to zirconium hydroxide, which was separated from the solution by filtration.³⁸ PEI was directly measured using NaOH/D₂O solution (1 M) as the solvent. The FTIR spectra of PEI-modified samples were recorded using KBr disk samples.

Powder X-ray diffraction (PXRD) pattern was measured using a PANalytical X’Pert X-ray diffraction system (PANalytical Company) using Cu Kα radiation (1.54 Å), working at 40 kV and 40 mA.

N₂ adsorption–desorption isotherms were recorded with a Quantachrome Autosorb-1-C instrument at 77 K. The specific surface area was estimated with the Brunauer–Emmett–Teller (BET) method and the pore volume was calculated with the desorption branch of the isotherms. The samples were degassed at 100 °C for 24 h before BET analysis.

The morphology was recorded with a scanning electron microscope (SEM, Utral 55, Carl zeiss, Germany) at an acceleration voltage of 1.5 kV, using the samples coated with a thin gold layer.

Thermogravimetric analysis (TGA) was recorded with a Perkin-Elmer instrument Pyris 1 TGA analyser under N₂ flow (20 mL/min) from 50 to 650 °C at a heating rate of 10 °C min^–1.

4.6. CO₂ Capture

Before CO₂ adsorption, about 1.0 g of adsorbent was outgassed under vacuum at 100 °C. CO₂ adsorption/desorption was performed under atmospheric pressure, as previously reported.^39,40 Desorption of the adsorbent was conducted under pure N₂ (60 mL/min) at a temperature of 120 °C. Some repeated experiments of CO₂ adsorption and desorption were done. CO₂ desorption heat (Q_des) was recorded with differential scanning calorimetry (DSC, Q200, TA Instrument).⁴¹ The samples were scanned at a heating rate of 10 °C/min under nitrogen flow.

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsomega.8b02319.

• N₂ adsorption isotherms of pristine UiO-66-NH₂ and PEI-modified adsorbents (PDF)

The authors declare no competing financial interest.