π-π stacking interactions: Non-negligible forces for stabilizing porous supramolecular frameworks.

Revealing the contribution of π-π stacking interactions in supramolecular assembly is important for understanding the intrinsic nature of molecular assembly fundamentally. However, because they are much weaker than covalent bonds, π-π stacking interactions are usually ignored in the construction of porous materials. Obtaining stable porous materials that are only dependent on π-π stacking interactions, despite being very challenging, could address this concern. Here, we present a porous supramolecular framework (π-1) stabilized only by intermolecular π-π stacking interactions. π-1 shows good thermal and chemical stability not only in various organic solvents but also in aqueous solution in a broad pH range. Furthermore, featuring one-dimensional channels with dangling thiolate groups, π-1 exhibits excellent Hg2+ removal performance, with adsorption capacity as high as 786.67 mg g-1 and an adsorption ratio as high as 99.998%. In addition, π-1 also shows high adsorption selectivity to Hg2+ in the presence of a series of interfering ions.

INTRODUCTION

Pioneered by Pedersen, Lehn, and Cram who were awarded the 1987 Nobel Prize in chemistry (–), supramolecular chemistry has been greatly developed in the past several decades. By Lehn’s definition, supramolecular chemistry refers to the domain of chemistry beyond that of molecules (). It focuses on molecular assemblies assembled with a number of discrete molecular components through intermolecular weak and reversible noncovalent interactions. These interactions involve hydrogen bond, π-π stacking, hydrophobic association, electrostatic interactions, van der Waals forces, and so on. In contrast to a single covalent bond with an energy of more than 200 kJ/mol, noncovalent interactions were assumed to be much weaker (). A hydrogen bond usually has an energy of 25 to 40 kJ/mol (), and other noncovalent interactions have energies lower than 10 kJ/mol (). In supramolecular compounds, the hydrogen bond is the main and most common intermolecular noncovalent interaction. It usually cooperates with other noncovalent interactions to stabilize the supramolecular networks. Because of the coexistence of these noncovalent interactions, it is difficult to differentiate their contributions to the stability of supramolecular assembly. The clear differentiation helps in understanding the intrinsic nature of molecular assembly fundamentally, particularly for macromolecules in biological systems (). Therefore, despite the great challenge of not having a related example reported to the best of our knowledge, the assembly of supramolecular networks being merely dependent on one type of noncovalent interactions is very meaningful.

Hydrogen-bonded organic frameworks (HOFs) are a class of newly emerged crystalline porous materials (CPMs) (–). In contrast to classic CPMs such as inorganic zeolites (), metal-organic frameworks (MOFs) (), and covalent organic frameworks (COFs) (), HOFs are obviously out of the ordinary, as they are classic supramolecular compounds, where hydrogen bonds contribute to their framework stability, as can be derived from the term itself. Although called HOFs, all reported HOFs are stabilized not merely by hydrogen bonds. Other noncovalent interactions, especially π-π stacking interactions, also give assistance to support porous structures. In some HOFs, π-π stacking even serves as the main intermolecular interactions instead of hydrogen bonds to stabilize the frameworks (–). Therefore, besides hydrogen bonds, π-π stacks are also important intermolecular interactions in supramolecular assembly despite the fact that they are often ignored, as they are relatively weaker.

To demonstrate the contribution of π-π stacks in supramolecular networks well, the emergence of stable CPMs that are only dependent on π-π stacking interactions, namely, π frameworks, is convincing. Undoubtedly, the assembly of a π framework is much more challenging than those of MOFs, COFs, and even HOFs in which diverse noncovalent interactions cooperatively contribute to their stability. Therefore, it is expected that no π framework with permanent porosity has been reported to date. Here, we present a π framework that may identify the contribution of π-π stacking interactions in supramolecular assembly. This π framework is based on a Zn(II) mononuclear complex, [Zn(phen)2L]·(3CH3OH·6H2O) [π-1, H2L = 1-(4-carboxyphenyl)-5-mercapto-1H-tetrazole]. Thorough structural analyses revealed that the framework of π-1 is only stabilized by intermolecular π-π stacking interactions. More impressively, π-1 shows good thermal and chemical stability, as well as excellent Hg2+ removal performance in aqueous solution due to its thiolate group–modified pore surface.

RESULTS

Synthesis and structure determination

Self-assembly of Zn(Ac)2·2H2O, phen, and H2L in a molar ratio of 2:1:1 at 105°C for 72 hours under solvothermal conditions resulted in a clear solution. This solution was allowed to stand at room temperature for 2 to 3 days, leading to the formation of rod-like crystals of π-1 in 73% yield (fig. S1, A to C). The infrared (IR) spectra of H2L π-1 and activated π-1 were measured in attenuated total refraction (ATR) mode and transmittance mode, respectively. As shown in fig. S1 (D to G), a peak at 2542 cm−1 assigned to the stretching vibration of the ─SH group is obviously observed in the IR spectra of H2L. After combining with Zn(II) to form π-1, this peak disappears, indicating the deprotonation of ─SH (). In addition, the ─COOH characteristic absorption at 1693 cm−1 red shifts to 1628 cm−1, demonstrating that the ─COOH of H2L deprotonates and participates in the coordination to Zn(II). These observations indicate that in π-1, both carboxylic and thiol groups of H2L are deprotonated. The generated L2− binds with Zn(II) to form a neutral coordination unit. This result was further confirmed by ion chromatography analysis, as no acetate (Ac−) was detected in the water solution where π-1 was immersed for 24 hours (fig. S1H).

Single-crystal x-ray diffraction (XRD) analysis showed that π-1 crystallizes in the trigonal R-3 space group (table S1; see the Supplementary Materials). The central metal Zn(II) ion coordinates with two carboxylic O atoms from one L2− ligand and four N atoms from two phen, generating a six-coordinated mononuclear Zn(II) structural unit (Fig. 1A). Two phen rings of the structural unit respectively stack with two phen rings of two adjacent structural units by strong π-π interactions to form a one-dimensional (1D) supramolecular chain building block along the c axis (Fig. 1B). Each supramolecular chain building block further connects with four other equivalent chain building blocks through strong π-π interactions between a tetrazolyl ring and a phen ring, as well as between an aromatic ring and a phen ring (Fig. 1C), resulting in a 3D porous supramolecular framework with 1D thiolate group–modified hexagonal channels (Fig. 1D and fig. S2A). The window size of the channel is about 7.5 Å by 7.5 Å, which is filled with lattice CH3OH and H2O molecules, as inferred by elemental analysis and thermogravimetric (TG) analysis (presented below).

Fig. 1

Crystal structures of π-1.

(A) Mononuclear Zn(II) structural unit in π-1. (B) The building block of π-1 viewed along the b axis, showing the strong π-π stacking interactions between phen rings. The selected distances (D) between ring centroids are as follows: DCg2-Cg5 = 3.476(5) Å, DCg3-Cg6 = 3.753(7) Å, DCg6-Cg6 = 3.577(7) Å, and DCg6-Cg10 = 3.465(6) Å. (C) π-π stacking interactions between building blocks. The selected distances (D) between ring centroids are as follows: DCg1-Cg3 = 3.621(6) Å, DCg1-Cg5 = 3.645(5) Å, DCg1-Cg10 = 3.710(6) Å, DCg4-Cg6 = 3.753(7) Å, and DCg4-Cg9 = 3.577(7) Å. (D) 3D porous supramolecular structure of π-1 viewed along the c axis, where one building block (yellow) surrounded by four other building blocks (blue and purple) is clearly demonstrated.

On the basis of the above structural analyses, it is clearly seen that the stability of π-1 is dependent on a number of π-π stacking interactions between conjugated organic groups of the mononuclear Zn(II) structural units. There are as many as 33 types of π-π stacking modes observed in π-1 (Table 1 and fig. S3). First, two phen rings of the structural unit respectively stack with two phen rings of two adjacent structural units to form the chain building block. The π-π interactions include Cg2-Cg5 [D = 3.476(5) Å], Cg2-Cg7 [D = 3.784(5) Å], Cg2-Cg8 [D = 3.721(5) Å], Cg2-Cg11 [D = 3.700(4) Å], Cg3-Cg6 [D = 3.753(7) Å], Cg3-Cg9 [D = 3.866(6) Å], Cg6-Cg6 [D = 3.577(7) Å], Cg6-Cg10 [D = 3.465(6) Å], Cg6-Cg12 [D = 3.577(7) Å], and so on (Fig. 1B; fig. S3, B, C, and E to L; and Table 1). These multiple strong π-π interactions endow the supramolecular chain building block with high stability. Then, the stable chain building block further connects with four equivalent chain building blocks through π-π interactions to weave a 3D porous supramolecular framework. The key π-π interactions herein involve those between the tetrazolyl ring and the phen ring—Cg1-Cg3 [D = 3.621(6) Å], Cg1-Cg5 [D = 3.645(5) Å], and Cg1-Cg10 [D = 3.710(6) Å] (Fig. 1C, fig. S3A, and Table 1)—and those between the aromatic ring and the phen ring, Cg4-Cg6 [D = 3.753(7) Å] and Cg4-Cg9 [D = 3.577(7) Å] (Fig. 1C, fig. S3D, and Table 1). Dependent on the synergistic effect of multiple π-π interactions, a 3D porous supramolecular framework of π-1 is assembled. Thus, many π-π stacking interactions in π-1 should be attributed to the fact that all the conjugated organic groups of the structural unit take part in the formation of π-π stacking interactions (Table 1 and fig. S3). Usually, the ring normal and the vector between the ring centroids forming an angle of about 20° up to Cg-Cg distances of 3.8 Å indicate a strong π-π stacking interaction (). As most of the π-π interactions in π-1 have shorter centroid distances [3.465(6) to 3.784(5) Å; Table 1] and the angle between the ring normal and the vector Cg-Cg is in the range of 18.6° to 22.9° (Table 1), the π-π interactions in π-1 belong to strong π-π interactions. Such strong and multiple π-π stacking interactions significantly contribute to the stability of π-1. Note that with the exception of π-π stacking interactions, there are no other covalent/noncovalent interactions between Zn(II) structural units. Therefore, π-1 represents a typical porous supramolecular π framework, which has not been reported yet to the best of our knowledge.

Table 1

π-π stacking interactions in π-1.

Cg1, N1 → N2 → N3 → N4 → C8→; Cg2, N6 → C18 → C17 → C16 → C15 → C19→; Cg3, N8 → C30 → C29 → C28 → C27 → C31→; Cg4, C2 → C3 → C4 → C5 → C6 → C7→; Cg5, C12 → C13 → C14 → C15 → C19 → C20→; Cg6, C24 → C25 → C26 → C27 → C31 → C32→; Cg7, N5 → C9 → C10 → C11 → C12 → C13 → C14 → C15 → C19 → C20→; Cg8, N6 → C18 → C17 → C16 → C15 → C14 → C13 → C12 → C20 → C19→; Cg9, N7 → C21 → C22 → C23 → C24 → C25 → C26 → C27 → C31 → C32→; Cg10, N8 → C30 → C29 → C28 → C27 → C26 → C25 → C24 → C32 → C31→; Cg11, N5 → C9 → C10 → C11 → C12 → C13 → C14 → C15 → C16 → C17 → C18 → N6 → C19 → C20→; Cg12, N7 → C21 → C22 → C23 → C24 → C25 → C26 → C27 → C28 → C29 → C30 → N8 → C31 → C32→.

Entry	π-π interactions	Cg-Cg (Å)*	α (°)†	β (°)‡	Cg-plane (Å)§	Slippage (Å)||	Symmetry operation on Cg
1	Cg1-Cg3	3.621(6)	0.0(5)	22.9	3.335(3)	1.410	x − y, −1 + x, 2 − z
2	Cg1-Cg5	3.645(5)	0.0(4)	25.3	3.015(8)/3.295(3)	1.557	1/3 + y, 2/3 – x + y, 5/3 − z
3	Cg1-Cg10	3.710(6)	0.0(4)	25.8	3.340(3)	1.615	x − y, −1 + x, 2 − z
4	Cg2-Cg5	3.475(5)	0.6(4)	8.8	3.434(3)	0.534	5/3 − x, 1/3 − y, 4/3 − z
5	Cg2-Cg7	3.784(5)	1.1(3)	25.5	3.432(3)/3.415(3)	1.629	5/3 − x, 1/3 − y, 4/3 − z
6	Cg2-Cg8	3.721(5)	0.3(3)	22.4	3.434(3)/3.440(2)	1.417	5/3 − x, 1/3 − y, 4/3 − z
7	Cg2-Cg11	3.700(4)	0.7(3)	22.2	3.427(3)3.425(2)	1.399	5/3 − x, 1/3 − y, 4/3 − z
8	Cg3-Cg6	3.753(7)	1.4(6)	24.4	3.388(5)/3.419(5)	1.547	5/3 − x, 1/3 − y, 7/3 − z
9	Cg3-Cg9	3.866(6)	2.2(5)	29.0	3.373(5)/3.380(4)	1.877	5/3 − x, 1/3 − y, 7/3 − z
10	Cg4-Cg6	3.920(6)	0.0(5)	24.9	3.557(3)	1.648	x − y, −1 + x, 2 − z
11	Cg4-Cg9	3.914(5)	0.0(3)	23.9	3.578(3)	1.587	x − y, −1 + x, 2 − z
12	Cg5-Cg2	3.476(5)	0.6(4)	8.9	3.434(3)	0.536	5/3 − x, 1/3 − y, 4/3 − z
13	Cg5-Cg8	3.627(4)	0.3(3)	18.6	3.433(2)/3.438(2)	1.154	5/3 − x, 1/3 − y, 4/3 − z
14	Cg6-Cg3	3.753(7)	1.4(6)	25.5	3.419(5)/3.387(5)	1.616	5/3 − x, 1/3 − y, 7/3 − z
15	Cg6-Cg6	3.577(7)	0.0(6)	17.7	3.408(5)	1.087	5/3 − x, 1/3 − y, 7/3 − z
16	Cg6-Cg10	3.465(6)	0.6(5)	11.5	3.398(5)/3.395(4)	0.689	5/3 − x, 1/3 − y, 7/3 − z
17	Cg6-Cg12	3.674(6)	0.6(5)	22.9	3.397(5)/3.383(3)	1.431	5/3 − x, 1/3 − y, 7/3 − z
18	Cg7-Cg2	3.784(5)	1.1(3)	24.9	3.416(3)/3.432(3)	1.594	5/3 − x, 1/3 − y, 4/3 − z
19	Cg8-Cg2	3.721(5)	0.3(3)	22.7	3.441(2)/3.434(3)	1.433	5/3 − x, 1/3 − y, 4/3 − z
20	Cg8-Cg5	3.627(4)	0.3(3)	18.8	3.438(2)/3.433(3)	1.169	5/3 − x, 1/3 − y, 4/3 − z
21	Cg8-Cg8	3.476(4)	0.0(2)	9.0	3.433(2)	0.542	5/3 − x, 1/3 − y, 4/3 − z
22	Cg8-Cg11	3.700(4)	0.4(2)	22.3	3.432(2)/3.423(2)	1.405	5/3 − x, 1/3 − y, 4/3 − z
23	Cg9-Cg3	3.866(6)	2.2(5)	29.2	3.380(4)/3.373(5)	1.889	5/3 − x, 1/3 − y, 7/3 − z
24	Cg9-Cg10	3.768(5)	1.4(4)	25.7	3.369(4)/3.394(4)	1.637	5/3 − x, 1/3 − y, 7/3 − z
25	Cg10-Cg6	3.464(6)	0.6(5)	11.3	3.395(4)/3.397(5)	0.678	5/3 − x, 1/3 − y, 7/3 − z
26	Cg10-Cg9	3.768(5)	1.4(4)	26.6	3.394(4)/3.369(4)	1.689	5/3 − x, 1/3 − y, 7/3 − z
27	Cg10-Cg10	3.749(5)	0.0(4)	25.1	3.396(4)/3.396(4)	1.588	5/3 − x, 1/3 − y, 7/3 − z
28	Cg10-Cg12	3.714(5)	0.8(3)	24.7	3.386(4)/3.376(3)	1.549	5/3 − x, 1/3 − y, 7/3 − z
29	Cg11-Cg2	3.701(4)	0.7(3)	22.2	3.426(2)/3.427(3)	1.396	5/3 − x, 1/3 − y, 4/3 − z
30	Cg11-Cg8	3.700(4)	0.4(2)	22.0	3.423(2)/3.432(2)	1.383	5/3 − x, 1/3 − y, 4/3 − z
31	Cg12-Cg6	3.673(6)	0.6(5)	22.4	3.383(3)/3.396(5)	1.398	5/3 − x, 1/3 − y, 7/3 − z
32	Cg12-Cg10	3.714(5)	0.8(3)	24.3	3.376(3)/3.386(4)	1.527	5/3 − x, 1/3 − y, 7/3 − z
33	Cg12-Cg12	3.965(4)	0.0(2)	31.5	3.380(3)/3.381(3)	2.072	5/3 − x, 1/3 − y, 7/3 − z

*Distance between ring centroids.

†Dihedral angle between planes I and J.

‡Angle between Cg(I)-Cg(J) vector and normal to plane I.

§Perpendicular distance of Cg(I) on ring J.

||Distance between Cg(I) and perpendicular projection of Cg(J) on ring I.

Thermal stability

The powder XRD measured at room temperature showed that all the peaks displayed in the pattern closely match those in the simulated one generated from the single-crystal diffraction data, which indicates that a single phase of π-1 was formed (fig. S4A). TG analysis revealed that the guest solvent molecules in π-1 can be released when heated to 75°C. The weight loss of 23.7% corresponds to three CH3OH and six H2O molecules per molecular unit in π-1 (calculated: 24.1%; fig. S4B). With the loss of the lattice guest molecules, the color of π-1 changes from light brown to dark green. Once the activated π-1 is exposed to air, the color of the activated π-1 quickly returns to light brown, exhibiting a reversible process (fig. S4C). Differential scanning calorimetry measurements showed that for π-1, a small endothermic peak at 85.11°C and a big endothermic peak at 128.27°C were observed (fig. S4D). For activated π-1, only a big endothermic peak at 127.73°C was detected (fig. S4E). These endothermic peaks correspond to the loss of guest solvents. No other endothermic peak was observed after the guest loss, showing that no structural transformations/distortions/rearrangements occurred after the guest loss. Variable-temperature powder XRD measurements demonstrated that the desolvated framework of π-1 can remain stable, as the main diffraction pattern of π-1 does not change from room temperature to at least 80°C (fig. S4F), at which the guest molecules in the pores are removed, as indicated by the TG result (fig. S4B).

The stability of the desolvated π-1 was further confirmed by gas sorption measurements. As shown in fig. S4F, the pores of π-1 cannot adsorb N2/H2 at 77 K and Ar at 87 K, while it can adsorb CO2 at 195 K, exhibiting selective adsorption behavior. Considering that the pore size of π-1 is 7.5 Å by 7.5 Å—much larger than the dynamic diameters of H2 (2.86 Å), Ar (3.4 Å), and N2 (3.64 Å)—the selective adsorption of π-1 to CO2 (3.3 Å) may be attributed to the special quadrupole moment of CO2 (−1.4 × 10−39 Cm2), which can induce the framework to interact with CO2 molecules and thus increase the binding to CO2 (). The adsorption isotherm shows a type I curve, indicating the microporous characteristic of π-1. The maximum CO2 uptake reaches 36.3 cm3/g (standard temperature and pressure) at normal pressure, confirming the real existence of permanent porosity in π-1 after the removal of the lattice guest molecules. To the best of our knowledge, π-1 represents the first example of a porous supramolecular framework stabilized merely by intermolecular π-π stacking interactions. Although several supramolecular compounds that seem to be π frameworks have been reported, there are other noncovalent interactions, such as nonclassical hydrogen bonds, that contribute to their stability besides the π-π stacking interactions. In addition, their permanent porosities have not been investigated and verified (–).

Chemical stability

Besides thermal stability, the chemical stability of π-1 was also studied (see the Supplementary Materials). As shown in fig. S5A, the powder XRD patterns of π-1 after being respectively immersed in H2O, methanol, ethanol, acetonitrile, benzene, etc. for 3 days are similar to those of the simulated one, demonstrating that the framework of π-1 can remain stable in H2O and common organic solvents. These exciting results encouraged us to further investigate its stability in aqueous solution at different pH values. The results show that the powder XRD patterns of π-1 after the immersion in aqueous solution at pH values ranging from 3 to 12 for 3 days are similar to those of the simulated one (fig. S5B), suggesting that π-1 is resistant to acid and base in the pH range of 3 to 12. The above results show that π-1 has good chemical stability. This good chemical stability may be attributed to its unique structural feature. As shown in fig. S2B, the coordination portion in π-1, the core of complex that directly determines the compound stability, is enwrapped by hydrophobic organic moieties, avoiding the attack of solvents or acid/base to the coordination units. Therefore, π-1 can remain stable in aqueous solution and common organic solvents, even in acidic/basic aqueous solutions.

Hg2+ removal

Considering the good stability and the thiolate group–modified pores, the performance of π-1 for removing Hg2+ in aqueous solution was evaluated. Typically, the Hg2+ adsorption experiments of π-1 were carried out by immersing 10 mg of desolvated π-1 into 10-ml HgCl2 aqueous solutions with different concentrations. As shown in Fig. 2A and fig. S6 (A to C), at a low initial concentration, π-1 can capture almost all the Hg2+ in the solution. With the increase of the initial Hg2+ concentration, the adsorption capacity of π-1 gradually increases. When the initial concentration of Hg2+ reaches 1200 mg/liter, the adsorption capacity of π-1 to Hg2+ reaches a maximum value of 786.67 mg g−1, corresponding to the capture of 0.927 Hg2+ per thiolate group in π-1. This result suggests that most of the thiolate groups in π-1 bind to Hg2+ ions. The maximum Hg2+ uptake capacity of π-1 is higher than that of most reported thiol/thio-functionalized porous materials (–) and comparable to that of benchmark porous materials COF-S-SH (), Bio-MOF (), PAF-1-SH (), and POP-SH (). The high Hg2+ uptake of π-1 should be attributed to the high porosity and pore surface area of π-1, together with a large number of highly accessible thiolate groups that are well dispersed and exposed on the pore surface of π-1. The good performance of π-1 for Hg2+ removal evidently indicates that crystalline porous supramolecular materials have great potential in environmental applications.

Fig. 2

Adsorption results of π-1 to Hg2+.

(A) Adsorption capacity of π-1 to Hg2+ at different initial concentrations. (B) Adsorption process of π-1 to Hg2+ at an initial concentration of 100 mg/liter. (C) Plot of t/q versus t. (D) Adsorption ratio of π-1 to metal ions in a mixed solution containing Hg2+, Pb2+, Cd2+, Cu2+, Zn2+, Ca2+, and Mg2+, with respective initial concentrations of 100 mg/liter.

To investigate the adsorption dynamics of π-1 to Hg2+, duplicate samples with equal amounts of desolvated π-1 (10 mg) were soaked and stirred in a series of Hg2+ solutions (100 mg/liter) for 0.5, 1, 2, 3, 4, 6, 8, 10, 12, and 16 hours, respectively. After filtration, the residual Hg2+ contents in the filtrate were analyzed by inductively coupled plasma mass spectrometry (ICP-MS). As shown in Fig. 2B and fig. S6 (D to F), π-1 rapidly captures Hg2+ at the very beginning. After 3 hours, more than 90% of Hg2+ in the solution was captured, and after 12 hours, 99.98% of Hg2+ was removed from the solution. This means that the concentration of the residual Hg2+ in the solution is 0.02% (0.02 mg/liter). Note that only 10 mg of π-1 was used for the capture of 25 ml of Hg2+ solution (100 mg/liter). When the amount of π-1 used was increased to 25 mg, the concentration of residual Hg2+ can be reduced to 1.6 μg/liter, reaching the U.S. standard of drinking water (2 μg/liter). These results demonstrate that the Hg2+ removal performance of π-1 is comparable to most benchmark sorbents (–).

The pseudo–second-order model shown in Eq. 1 was used to fit the experimental data to further study the Hg2+ adsorption kinetics, where k1 (g·mg−1·min−1) is the rate constant of pseudo–second-order adsorption, and q and q (mg·g−1) are the amount of Hg2+ adsorbed at time t (min) and at equilibrium, respectively (). The results shown in Fig. 2C reveal the correlation coefficient of 0.9999 and the rate constant k1 of 1.091 mg·g−1 min−1. Such a high correlation is sufficient to prove that the adsorption of π-1 to Hg2+ belongs to the pseudo–second-order model (). The k1 value is also higher than those of most reported adsorbents for Hg2+ adsorption under similar conditions (), suggesting an extraordinarily fast adsorption kinetics of π-1 to Hg2+.

The distribution coefficient (Kd) is another important indicator to evaluate the adsorption performance of adsorbents. Kd is defined and calculated using Eq. 2 (), where Ci and Cf represent the initial and the final equilibrium Hg2+ concentration, respectively, V is the volume of the treated solution (in milliliters), and m is the quantity of the adsorbent used (in grams). Generally, a Kd value reaching 1.0 × 105 ml·g−1 indicates an excellent sorbent (, ). The Kd of π-1 for Hg2+ adsorption was calculated to be 4.99 × 106 ml g−1. This value is more than 50 times larger than the standard value and only smaller than that of PAF-1-SH (). This result reflects the superabsorbent property of π-1 for Hg2+ solutions.

Besides Hg2+, π-1 also exhibits adsorption ability for other metal ions such as Pb2+, Cd2+, Cu2+, Zn2+, Ca2+, and Mg2+ under the same conditions, with adsorption ratios of 83.32, 46.49, 62.48, 26.89, 38.42, and 49.64%, respectively (fig. S7, A to C). To investigate the adsorption selectivity of π-1, we prepared and used a mixture containing HgCl2, PbCl2, CdCl2, CuCl2, ZnCl2, CaCl2, and MgCl2 instead of a single–metal ion solution. As shown in Fig. 2D and fig. S7 (G to I), π-1 still shows strong capture performance to Hg2+, with an adsorption ratio as high as 96.78%. In contrast, π-1 exhibits low adsorption capacity to Pb2+, Cd2+, Cu2+, Zn2+, Ca2+, and Mg2+, with adsorption ratios of 17.15, 11.25, 9.63, 8.68, 13.15, and 13.84%, respectively. These observations demonstrated that other ions do not interfere with the adsorption capacity of π-1 for Hg2+. To the best of our knowledge, π-1 is the first supramolecular sorbent that shows such high adsorption selectivity to Hg2+.

To investigate the possible influence of metal salt solubilities on the adsorption property of π-1, we carried out the ion adsorption experiments by using nitrate salts [Hg(NO3)2, Pb(NO3)2, Cd(NO3)2, Cu(NO3)2, Zn(NO3)2, Ca(NO3)2, and Mg(NO3)2] instead of the corresponding chloride salts. As shown in fig. S7 (D to F), the adsorption amounts of π-1 to single metal ion by using nitrate salts is similar to that using chloride salts (fig. S7, A to C). The adsorption selectivity of π-1 to the mixed metal ion by using nitrate salts is also similar to that by using chloride salts (fig. S7, G to I). These results indicate that the solubility differences of these tested metal salts have limited influence on the adsorption properties of π-1, which may be ascribed to the very low concentration of each metal salt. At a concentration of 100 mg/liter, all the metal salts tested can completely dissolve and dissociate.

The recyclability of π-1 for removing Hg2+ from aqueous solution was further examined. As shown in fig. S8A, the framework of π-1 remained stable after adsorbing Hg2+, which was also evidenced by the scanning electron microscopy (SEM) results, where the same morphology as the one without Hg2+ uptake was observed (fig. S8, C and D). After four runs of the adsorption-desorption cycle, the adsorption performance of π-1 on Hg2+ removal barely decreases (fig. S8E), and the framework of π-1 also remained intact (fig. S8B). The CO2 sorption isotherms of π-1 after the Hg2+ adsorption test are also similar to those before the Hg2+ adsorption test (fig. S8F), which further confirms the framework stability of π-1. These results illustrate that π-1 has good stability and recyclability. The good reusability of π-1 in Hg2+ removal can be attributed to its unique structural feature; that is, the coordination portions of the structure are enwrapped by hydrophobic organic moieties (Fig. 2B). As the capture of Hg2+ occurs on the pore surface of π-1 and because there are no direct contacts and interactions with the coordination portions that directly determine the stability of the whole framework, π-1 can retain its framework stability during the Hg2+ adsorption-desorption process and its recyclability during Hg2+ removal without any decrease in adsorption performance and efficiency. The good recycle performance of π-1 during Hg2+ removal is comparable to most porous absorbent materials with a state-of-the-art Hg2+ adsorption function (, , ). The high Hg2+ uptake, selective adsorption ability against interference, and good recycle performance would enable π-1 to become a potential sorbent in water purification.

DISCUSSION

In summary, we have assembled a porous crystalline framework stabilized merely by π-π stacking interactions. Without other covalent/noncovalent interactions, this stable π framework could serve as concrete evidence that π-π stacking interactions are non-negligible forces for stabilizing porous supramolecular frameworks, especially in HOFs. Notably, the thiolate-functionalized π-1 shows exceptional chemical stability not only in water and common organic solvents but also in acidic and basic aqueous solutions. As a result, π-1 exhibits excellent Hg2+ removal performance in terms of uptake capacity, selectivity, and recyclability comparable to reported pioneering Hg2+ sorbents.

MATERIALS AND METHODS

Materials and measurements

All the chemicals were commercially available and used without further purification. Fourier transform IR (FTIR) spectra were recorded on a PerkinElmer Frontier Mid-IR-FTIR apparatus in ATR or transmittance mode. Powder XRD measurements were performed on a Bruker D8 Focus diffractometer using the CuKα radiation. TG analyses were carried out on a BOIF WCT-2C Thermogravimetry Analyzer at N2 atmosphere. Gas sorption measurements were performed on a BELSORP-max automatic volumetric adsorption apparatus. ICP-MS tests were performed on a HORIBA Ultima Expert ICP Optical Emission Spectrometer and an Agilent 7900 ICP-MS instrument. Atomic fluorescence spectroscopy (AFS) measurements were carried out on a Jitian AFS-922 double-channel atomic fluorescence spectrometer. Examination of the acetate (Ac−) was performed using Metrohm Eco IC ion chromatography.

Preparation of π-1

A mixture of 0.2 mmol of Zn(Ac)2·2H2O (0.0439 g), 0.1 mmol of 1,10-phenanthroline monohydrate (phen; 0.018 g), 0.1 mmol of 1-(4-carboxyphenyl)-5-mercapto-1H-tetrazole (H2L; 0.022 g), and 12 ml of CH3OH/H2O mixture (v:v = 1:1) was stirred at room temperature for 10 min. After adjusting the pH to 7.6 by adding triethylamine, the mixture was transferred to a Teflon-lined stainless steel autoclave and heated at 105°C for 72 hours. The autoclave was cooled to room temperature at a rate of 10°C·hour−1. The mixture was filtered, and the resulting colorless clear filtrate was allowed to stand at room temperature for 2 to 3 days; rod crystals of π-1 were obtained. The crystals were collected by filtration and dried in air. Yield: 73% based on H2L. Elemental analyses calcd (%) for C35H44N8O11SZn (π-1): C, 49.44; H, 5.22; N, 13.18. Found: C, 49.12; H, 5.05; N, 13.27. IR (KBr, cm−1): 3421(s), 3057(w), 2364(m), 2341(m), 1612(vs), 1558(s), 1516(vs), 1432(vs), 1377(vs), 1342(s), 1274(m), 1222(w), 1142(w), 1102(s), 1011(w), 848(vs), 785(m), 725(vs), 642(w), 472(w).

Hg2+ removal measurements

Ten milligrams of pretreated π-1 was added to a 25-ml glass vial containing 10 ml of HgCl2 solution with a concentration of 100 mg/liter. After stirring for a certain period, the solution was filtered. The resulting filtrate was tested by ICP-MS/AFS to check the residual amount of Hg2+. The resulting crystals were tested by SEM and powder XRD. The procedures for other ion removal measurements were similar to that for Hg2+ removal by using PbCl2, CdCl2, CuCl2, ZnCl2, CaCl2, or MgCl2 solution (100 mg/liter) instead of Hg2+ solution (100 mg/liter). After stirring for 12 hours, the solution was filtered and the residual ion concentration in the filtrate was tested by ICP-MS. The procedure for metal ion removal experiments of π-1 using nitrate salts is similar to that using chloride salts.

Maximum Hg2+ uptake measurements

Ten milligrams aliquots of pretreated π-1 were added to 25-ml glass vials containing 10 ml of HgCl2 solution with concentrations of 100, 200, 400, 800, 1000, 1200, 1600, and 2000 mg/liter. After stirring for 12 hours, the solution was filtered. The resulting filtrate was tested by ICP-MS to check the residual amount of Hg2+.

Selective adsorption tests

The procedure was similar to that described above by using a 10-ml mixture containing HgCl2, PbCl2, CdCl2, CuCl2, ZnCl2, CaCl2, and MgCl2 with concentrations of 100 mg/liter instead of 10 ml of HgCl2 solution (100 mg/liter). After stirring for 12 hours, the mixture was filtered, and the filtrate was measured by ICP-MS to check the concentration of each ion in the filtrate. In addition, a 10-ml mixture containing Hg(NO3)2, Pb(NO3)2, Cd(NO3)2, Cu(NO3)2, Zn(NO3)2, Ca(NO3)2, and Mg(NO3)2 with concentrations of 100 mg/liter instead of the corresponding chloride salts was prepared and used to further confirm the influence of the solubilities of metal salts to the selective adsorption properties of π-1.

Recycle tests

After the Hg2+ adsorption experiment, the crystalline solid was isolated by centrifugation. The obtained solid was soaked in a HCl solution (pH 5.6) for 6 hours, then isolated by centrifugation and dried in a vacuum oven at 50°C, and used for the next run of the Hg2+ adsorption experiment.