Introduction

Nuclear reactors have been continuously providing ~10% of the world’s energy over the last decade1. As a sustainable and low-carbon energy supply, nuclear energy is expected to play a more important role in the future2,3,4. However, safety concerns still challenge its operation. One of the major safety issues is the volatile radioactive waste produced during the reprocessing of spent nuclear fuels, which primarily consists of radionuclides, such as 129I and 131I in the form of molecular iodine (I2) or organic iodides (e.g., methyl iodide (CH3I) and ethyl iodide)4,5,6,7,8. These compounds are harmful to the environment (129I has an extremely long half-life of approximately 1.57 × 107 years) or severely affect human metabolism by damaging the thyroid gland, and must be removed before the off-gas is discharged9,10,11.

Compared with traditional liquid scrubbing processes to capture radioactive iodine, adsorption-based processes require a simpler operation and lower maintenance costs and avoid highly corrosive solutions6. Therefore, researchers have increasingly focused on the development of various adsorbents for iodine capture, including materials containing silver (Ag)12,13,14, ceramics13,15,16, zeolites17,18, aerogels19,20,21, metal-organic frameworks8,22,23,24,25,26, and conjugated polymers27,28,29,30,31. Most of these studies have focused on the adsorption capacity of the developed adsorbent for I2, whereas only a few studies have addressed the capture of CH3I, and even fewer studies have examined the simultaneous capture of I2 and CH3I. Given that radioactive molecular iodine and organic iodides coexist in off-gas streams, it is particularly important to develop adsorbents that can capture them simultaneously and efficiently.

Various strategies have been adopted to promote the adsorption of iodine species. For I2, effective strategies include the following: (i) using adsorbents containing Ag to precipitate I2 in the form of silver iodide (AgI)2,12,13, (ii) introducing electron-rich heteroatoms (e.g., nitrogen (N), sulfur (S), and oxygen(O)) or π-donors (e.g., double/triple bonds, benzene rings, and other aromatic compounds) in adsorbents to adsorb electron-deficient I2 by forming charge-transfer complexes32,33,34,35,36,37,38,39,40,41, and (iii) modifying the adsorbent with ionic groups (e.g., [RN-(CH3)3]+·Br) to adsorb I2 via Coulomb interactions42. Compared with I2, CH3I is more difficult to capture because of the weaker intermolecular forces43. Currently, the capture of CH3I is primarily achieved either through catalytic decomposition on adsorbents containing Ag to form AgI13,44,45,46 or through an N-methylation reaction on nucleophilic N sites to form pyridinium34 or quaternary ammonium salts1,31,47,48,49. Based on these insights, we speculate that adsorbents containing abundant Ag sites or nucleophilic N sites may exhibit high capture capacity for both I2 and CH3I. Given the high cost and poor recyclability of Ag-based adsorbents, developing N-rich adsorbents is a better choice for simultaneously capturing I2 and CH3I.

In addition to the characteristics of adsorption sites that determine binding strength, the density of the adsorption sites and the textural properties (e.g., surface area, pore size, and pore volume) of the adsorbent are crucial because these factors collectively determine the number of accessible adsorption sites (i.e., the adsorption capacity) and adsorption kinetics. Therefore, an ideal adsorbent for simultaneously capturing I2 and CH3I should possess a high concentration of nucleophilic N sites along with a large surface area and a highly open porous structure.

As an emerging class of porous materials, covalent organic frameworks (COFs) provide an ideal platform for developing high-performance adsorbents because their porous structures and surface functionalities can be easily engineered to meet the requirements of specific applications50,51,52. Several COFs have been prepared as adsorbents for I2 capture, in which the binding sites are π-conjugated moieties, various N-containing functional groups, and ionic groups34,35,38,42. Although some of these COFs exhibited high I2 adsorption capacities in the measurement performed in a static closed system with high partial pressure of I2, their performance for low-concentration I2 capture under a dynamic condition was not measured. More importantly, like other recently developed adsorbents, the CH3I adsorption properties of these COFs have not been investigated.

To the best of our knowledge, there is only one COF material (SCU-COF-2) evaluated for both I2 and CH3I adsorption34. In SCU-COF-2, there are pyridine and imine moieties, which bind to I2 and CH3I through charge-transfer interaction and N-methylation reactions, respectively. At room temperature, SCU-COF-2 adsorbed 0.979 g g−1 I2 from flowing N2 (carrier gas) containing 400 ppm of I2 or 0.564 g g−1 of CH3I from flowing N2 containing 200,000 ppm of CH3I. These moderately high adsorption capacities were obtained from single-component measurements, whereas the adsorption performance of SCU-COF-2 in the coexistence of I2 and CH3I has not been explored.

In this report, we designed and synthesized two COFs, namely COF-TAPB and COF-TAPT, for the simultaneous capture of I2 and CH3I. The framework of COF-TAPB was constructed through imine linkages formed between two monomers, tris(4-formylphenyl)amine (TFPA) and 1,3,5-tris(4-aminophenyl)benzene (TAPB); COF-TAPT is a structural analog of COF-TAPB, constructed from TFPA and 2,4,6-tri(4-aminophenyl)-1,3,5-triazine (TAPT) (see Fig. 1). Through a literature search, we found that these two COF materials have previously been synthesized for CO2 adsorption and photocatalytic hydrogen evolution53,54,55, whereas the synthetic methods are not exactly the same as those used in this study. The two COFs exhibit the same crystal structure and textural properties, being different only in N content, allowing the investigation of the role of N in the adsorption of I2 and CH3I.

Fig. 1: Schematic illustration of the synthesis of COF-TAPT and COF-TAPB.
figure 1

In the structural model, different N sites are marked with different colors.

We evaluated the adsorption properties of the two COFs for I2 and CH3I under different conditions (static and dynamic adsorption at different temperatures and adsorbate concentrations) for a direct comparison with benchmark adsorbents reported in the literature. Under the static high-concentration conditions, COF-TAPB and COF-TAPT exhibited similar high I2 adsorption capacities, and their static I2 uptake values (7.94 and 8.61 g g−1, respectively) are among the highest reported for various adsorbents. Unlike the case of I2 adsorption, COF-TAPT exhibited a significantly higher CH3I uptake capacity than COF-TAPB, suggesting that the N content in the adsorbent plays a vital role in CH3I adsorption. The further systematic analysis confirmed that the CH3I adsorption capacity is positively correlated with the N content in the adsorbent. Remarkably, COF-TAPT exhibits a record-high CH3I adsorption capacity (1.53 g g−1; static conditions at 25 °C), which can be attributed to the combined effect of its high N content (16.1 wt%) and large surface area (~2300 m2 g−1). In the dynamic CH3I adsorption measurement, COF-TAPT demonstrated the highest capacity of all tested adsorbents. When used to simultaneously capture low-concentration I2 (150 ppm) and CH3I (50 ppm) from a carrier gas stream, COF-TAPT outperformed all tested adsorbents except an ionic COF in terms of total iodine capture. The adsorbed I2 and CH3I can be easily extracted by ethanol or acetone from COF-TAPT to fully restore its adsorption capacity for subsequent adsorption cycles. The density functional theory (DFT) calculations revealed that the CH3I binding energy at different N sites in COF-TAPT follows the order imine N > triazine N > sp3 N.

Results

Characterization

Powder X-ray diffraction (PXRD) indicated that COF-TAPT and COF-TAPB are highly crystalline, both exhibiting four clearly discernible peaks before 12° (2θ) and two weak peaks at around 16° and 25° (Fig. 2a, c). We performed structural modeling based on PXRD to corroborate that the synthesized COFs have the designed structures. The results revealed that for both materials, the experimental data agree well with the simulated data based on the eclipsed (AA) stacking model (Fig. 2b, d), and the observed diffraction peaks can be indexed as (100), (110), (200), (210), (310), and (001) reflections, respectively. The Pawley fitting results were reasonably good, producing a unit cell of a = b = 23.51 Å, c = 3.64 Å, α = β = 90°, and γ = 120° (RP = 2.15% and RWP = 2.89%) for COF-TAPT (Supplementary Table 1), and a unit cell of a = b = 23.61 Å, c = 3.3.65 Å, α = β = 90°, and γ = 120° (RP = 2.57% and RWP = 3.35%) for COF-TAPB (Supplementary Table 2). From the N2 sorption isotherms collected at 77 K (Fig. 2e), the Brunauer−Emmett−Teller (BET) surface areas of COF-TAPT and COF-TAPB are derived at 2348 and 2290 m2 g−1, respectively. Their pore size distribution is centered at 1.92 nm (Fig. 2e), which is consistent with the designed structure and PXRD results. The two COFs exhibit very similar PXRD patterns and N2 adsorption isotherms, indicating that they have comparable structural and textural properties (Supplementary Table 3). High-resolution transmission electron microscopy confirmed that they both possess a hexagonal structure containing one-dimensional channels (Supplementary Fig. 1).

Fig. 2: Structural characterization of COF-TAPT and COF-TAPB.
figure 2

a, c PXRD patterns of a COF-TAPT and c COF-TAPB, with the experimental profiles in black, Pawley refined profiles in red, calculated (from the refined structure model) profiles in blue, and differences between experimental and refined PXRD patterns in pink. Green lines indicate Bragg positions. b, d Refined structure model of b COF-TAPT and d COF-TAPB viewed along the c-axis (upper) and a-axis (lower). e N2 sorption isotherms of COF-TAPT and COF-TAPB. Inset shows the derived pore size distribution profiles of COF-TAPT and COF-TAPB. f FTIR spectra of the synthetic precursors (TFPA, TAPB, and TAPT) and the two COFs (COF-TAPB and COF-TAPT).

The completion of the Schiff base reaction between TFPA and TAPT/TAPB was evidenced by the disappearances of bands at 3359 to 3428 cm−1 (amino groups) and 1685 cm−1 (aldehyde groups) and the synchronous appearance of the characteristic -C=N- stretching band at ~1625 cm−1 in the Fourier transform infrared (FTIR) spectra (Fig. 2f). The successful construction of the designed COF frameworks and the presence of different N species in the frameworks were further confirmed by solid-state 13C nuclear magnetic resonance (NMR) spectroscopy and N 1s X-ray photoelectron spectroscopy (XPS) (Supplementary Fig. 2). In addition, the elemental analysis revealed that the carbon (C), hydrogen (H), and N content in COF-TAPT and COF-TAPB closely agreed with the theoretical values (Supplementary Table 4). These COFs retained high crystallinity after treatment with concentrated HNO3 aqueous solution (5 M) or β-irradiation (200 kGy), exhibiting the excellent stability required to capture radioactive iodine from the off-gas stream (Supplementary Fig. 3).

Static I2 and CH3I adsorption

In most previous studies, I2 adsorption was performed in a static closed system with saturated I2 vapor at 75 °C, and the adsorption capacity was determined based on the mass increase subsequently measured under ambient conditions at room temperature29,32,33,35,36,37,56. To directly compare with the previously developed adsorbents, we evaluated COF-TAPB and COF-TAPT using the same experimental setup (see the Experimental section in the Supporting Information for detailed methods). The results revealed that COF-TAPB and COF-TAPT adsorbed 7.94 and 8.61 g g−1 I2 within 96 h, respectively (Fig. 3a and Supplementary Table 3). These values rank high among all adsorbents tested under the same conditions (Fig. 3c and Supplementary Table 5). It is worth noting that under this commonly used evaluation condition, where the concentration of I2 is rather high (16,000 ppm), the adsorption is dominated by the intermolecular interactions of I2; consequently, the capacity is largely determined by the pore volume of the adsorbent in addition to the characteristics of the binding sites.

Fig. 3: Static I2 and CH3I adsorption performance.
figure 3

Gravimetric measurement of static a I2 and b CH3I vapor adsorption capacities of COF-TAPT and COF-TAPB materials as a function of time at 75 °C. c Comparison of the static I2 adsorption capacities of various high-performance adsorbents. The specific I2 uptake values of the reported adsorbents and corresponding references are presented in Supplementary Table 5. The error bars are the standard deviations from three parallel measurements.

We used the average adsorption rate determined at 80% of the full adsorption capacity (K80%)38 to describe the adsorption kinetics of the adsorbents. Despite their similar porous structures, COF-TAPT exhibited a higher K80% value than COF-TAPB (0.48 vs. 0.33 g g−1 h−1), which can be attributed to its higher N content promoting the initial adsorption of I2. The measured I2 adsorption kinetics of COF-TAPT and COF-TAPB are faster than those of many previously reported microporous adsorbents (Supplementary Table 5) due to their highly crystalline structures that facilitate mass transport. To investigate the effect of porosity on the adsorption capacity and adsorption kinetics, we prepared a control sample, which was synthesized using the same process as COF-TAPT, except that N,N−dimethylformamide (DMF) instead of mixed ethanol and o-dichlorobenzene was used as the solvent. The obtained material (denoted as TFPA-TAPT) has the same chemical composition as COF-TAPT but a much lower porosity (BET specific surface area: 1284 m2 g−1; total pore volume: 0.19 cm3 g−1) due to its poor crystallinity (Fig. 4a, b). Under the same conditions, TFPA-TAPT exhibited a lower I2 adsorption capacity (4.31 g g−1) and a slower adsorption rate (K80%: 0.098 g g−1 h−1) than COF-TAPT (Supplementary Fig. 4c). This result demonstrates the significant influence of textural properties of the adsorbent on its I2 adsorption behavior.

Fig. 4: Dynamic I2 adsorption and CH3I adsorption performances.
figure 4

a I2 breakthrough profiles of COF-TAPB and COF-TAPT at 25 °C. The concentration of I2 in the carrier gas is 400 ppm. b Dynamic I2 uptake values of COF-TAPB and COF-TAPT under dry and humid conditions. The water uptake under humidity is also presented. c CH3I breakthrough profiles of COF-TAPB and COF-TAPT at 25 °C. The concentration of CH3I in the carrier gas is 200,000 ppm. d Dynamic CH3I uptake values of COF-TAPB and COF-TAPT under dry and humid conditions. The water uptake under humidity is also presented.

We also evaluated the CH3I adsorption performance of COF-TAPT and COF-TAPB in a static closed system with a saturated CH3I vapor at 75 °C, as used in the previous study34, for direct comparison. The results demonstrated that COF-TAPT adsorbed 1.53 g g−1 of CH3I (Fig. 3b), exceeding the capacity of the state-of-the-art CH3I adsorbent SCU-COF-2 (1.45 g g−1)34. In contrast, COF-TAPB adsorbed only 0.81 g g−1 of CH3I under the same conditions (Fig. 3b). The control sample TFPA-TAPT exhibited a CH3I adsorption capacity of 1.37 g g−1, although its BET surface area is only half of that of COF-TAPT (Supplementary Fig. 4d). These results collectively indicate that, unlike I2 adsorption, CH3I adsorption capacity depends on the number of strong binding sites of the adsorbent rather than its surface area or pore volume. This conclusion was further verified by the plots of CH3I uptake vs. the BET surface area (Supplementary Fig. 5a) and CH3I uptake vs. N content (Supplementary Fig. 5b), based on the data for seven adsorbents, which clearly indicate that the CH3I uptake is irrelevant to the surface area but positively correlated with the N content of the adsorbent (Supplementary Table 3). In addition, at their full adsorption capacity, the CH3I/N ratios in COF-TAPT and COF-TAPB were 0.97 and 0.89, respectively, whereas the I2/N molar ratios were 3.05 and 4.89, respectively.

These results collectively indicate the essential difference between I2 adsorption and CH3I adsorption. The nearly one-to-one correspondence between CH3I and N suggests that CH3I molecules are only adsorbed on N sites, presumably by forming salts. In contrast, I2 adsorption can occur at other sites in the π-conjugated frameworks (e.g., benzene rings) or be promoted by forming polyiodide species, resulting in high I2/N ratios.

Dynamic I2 and CH3I adsorption

Given the low concentrations of molecular iodine and organic iodides (usually <200 ppm) in the off-gas stream2,57,58, it is more meaningful to test the adsorption behavior of adsorbents at low I2 (or CH3I) concentrations related to practical applications. To explore the potential of the developed COFs for practical applications, we evaluated their I2 and CH3I adsorption capacities under dynamic conditions using a fixed-bed column-breakthrough configuration, which allowed the concentration of I2 and CH3I, the temperature of the adsorbent bed, and humidity to be freely adjusted42.

To directly compare the developed COFs with benchmark adsorbents17,34,42, we conducted dynamic adsorption at 25 °C with 400 ppm of I2 in N2 flow (adsorbent: 30.0 mg; flow rate: 10.0 mL min−1). Under these conditions, COF-TAPB exhibited a steep breakthrough step after 45 h with a total I2 uptake of 2.18 g g−1, and COF-TAPT demonstrated a similar breakthrough profile, whereas its breakthrough time was 50 h, corresponding to a total I2 uptake of 2.38 g g−1 (Fig. 4a). The observed I2 adsorption capacities are significantly higher than those of most reported adsorbents under similar measurement conditions (Supplementary Fig. 6 and Supplementary Table 6). In addition, the presence of water vapor only caused a slight decrease in the I2 uptake of COF-TAPT (from 2.38 g g−1 to 2.32 g g−1) and COF-TAPB (from 2.18 to 2.13 g g−1) (Fig. 4b), indicating that they are both tolerant to moisture, which is important for practical applications.

We performed dynamic CH3I adsorption using the same column-breakthrough setup, with the CH3I concentration controlled at 200,000 ppm. The results revealed that, under dry conditions, the CH3I uptakes of COF-TAPB and COF-TAPT were 0.71 and 1.30 g g−1, respectively (Fig. 4c, d). The CH3I adsorption capacity of COF-TAPT (1.30 g g−1) is higher than that of all reported adsorbents except MIL-101-Cr-HMTA1 (Supplementary Fig. 7 and Supplementary Table 7). The ultrahigh CH3I adsorption capacity of MIL-101-Cr-HMTA is attributed to the combination of a large surface area and abundant tertiary amine groups that can strongly and specifically interact with CH3I. However, MIL-101-Cr-HMTA exhibits a limited adsorption capacity for I2 (Supplementary Fig. 8 and Supplementary Table 6) due to the lack of effective I2 binding sites other than tertiary amines. When water vapor was introduced into the feed stream (relative humidity = 50%), the CH3I adsorption capacity of COF-TAPT decreased from 1.30 to 1.06 g g−1, indicating competitive adsorption between H2O and CH3I. This result is because the adsorption of CH3I primarily occurs on N sites, which also adsorb H2O molecules. The dynamic adsorption measurements indicate that COF-TAPT has a similar I2 uptake capacity but a significantly higher CH3I uptake capacity than COF-TAPB. This outcome is consistent with the results of the static adsorption experiments and further validates the above conclusion about the difference between I2 adsorption and CH3I adsorption.

The adsorbed I2 and CH3I in COF-TAPT can be fully extracted by ethanol to regenerate its adsorption capacities. When the I2-saturated COF-TAPT (I2@COF-TAPT) or CH3I-saturated COF-TAPT (CH3I@COF-TAPT) was immersed in ethanol, the I2 or CH3I desorption process proceeded spontaneously and accelerated with the assistance of sonication. As a commonly used method for regenerating adsorbents after I2 adsorption, extraction with ethanol can also efficiently remove CH3I from COF-TAPT because the N-methylation reaction is reversible in protic solvents59. The regenerated COF-TAPT (COF-TAPT-Re) restored its original physiochemical properties, as evidenced by the FTIR, PXRD, and N2 sorption characterizations. Correspondingly, the exceptionally high adsorption capacity of COF-TAPT can be fully restored in four successive adsorption/extraction cycles under the above conditions (Supplementary Figs. 9, 10).

Simultaneous capture of low-concentration I2 and CH3I

After evaluating the adsorption performance of the developed COFs for I2 and CH3I under commonly used conditions, we explored their ability to capture I2 and CH3I at much lower concentrations (150 and 50 ppm, respectively) relevant to practical off-gas treatment applications. Several state-of-the-art adsorbents for I2 or CH3I capture, including MIL-101-Cr-HMTA1, SCU-COF-234, and iCOF-AB-5042, were evaluated under the same conditions for comparison. We started with single-component measurements, introducing only I2 (150 ppm) or CH3I (50 ppm) in the feed stream for capture. The results revealed that, for I2 capture, the order of adsorption capacity of the tested adsorbents is iCOF-AB-50 (1.52 g g−1) > COF-TAPT (1.25 g g−1) > COF-TAPB (1.12 g g−1) > MIL-101-Cr-HMTA (0.83 g g−1) > SCU-COF-2 (0.49 g g−1) > TFPA-TAPT (0.42 g g−1) (Fig. 5a and Supplementary Table 8). This order is consistent with that derived from the static adsorption measurements (Fig. 3c). The exceptionally high I2 uptake of iCOF-AB-50 is attributed to the presence of abundant ionic groups that effectively promote I2 adsorption via strong Coulomb interactions. The results of various adsorbents capturing CH3I, as summarized in Fig. 5b and Supplementary Table 8, indicate that COF-TAPT ranks second among all tested adsorbents. The specific CH3I adsorption capacities are as follows: MIL-101-Cr-HMTA (0.51 g g−1) > COF-TAPT (0.39 g g−1) > TFPA-TAPT (0.18 g g−1) > COF-TAPB ≈ iCOF-AB-50 (0.12 g g−1) > SCU-COF-2 (0.08 g g−1).

Fig. 5: Comparison of different adsorbents in dynamic adsorption capacity.
figure 5

The capacities were measured from a single-component I2 at 150 ppm, b single-component CH3I at 50 ppm, c mixed-component (150 ppm I2 and 50 ppm CH3I) breakthrough experiments performed at 25 °C. The error bars are the standard deviations from three parallel measurements.

These results further confirm that the adsorption behaviors of I2 and CH3I are different. The adsorption of I2 can be initiated through specific functional groups (e.g., ionic groups) in the adsorbent and further promoted through strong intermolecular interactions. Therefore, the type of binding sites and textural properties of the adsorbent both critically influence the I2 uptake. In contrast, the adsorption of CH3I is primarily determined by the type and number of binding sites and is not much related to the textural properties of the adsorbent. In addition, ionic groups can strongly promote the adsorption of I2 but have a little promotional effect on the adsorption of CH3I because I2 can be easily induced to form charge species, such as I3 and I5, whereas CH3I cannot.

Finally, we measured the simultaneous capture of iodine species on various adsorbents by co-feeding I2 (150 ppm) and CH3I (50 ppm). Considering the excellent capture ability of COF-TAPT for I2 and CH3I in the single-component measurements, good performance in the co-capture of these two species is expected. Indeed, COF-TAPT exhibited a significantly higher total iodine (I2 + CH3I) capture capacity (1.51 g g−1) than other tested adsorbents, except iCOF-AB-50 (Fig. 5c and Supplementary Table 8). The total uptake of iCOF-AB-50 primarily derives from the contribution of I2 adsorption; thus, it is conceivable that COF-TAPT is more suitable for feed streams containing a high fraction of organic iodide. Similar to the single-component measurement results, the high performance of the COF-TAPT in simultaneous capture of I2 and CH3I can be fully regenerated in successive tests (Supplementary Fig. 11).

Discussion

To analyze the adsorption sites of COF-TAPT, we characterized the I2-saturated COF-TAPT sample (I2@COF-TAPT) with PXRD after the static adsorption measurement. The obtained PXRD pattern did not exhibit diffraction peaks related to the original crystalline structure of COF-TAPT (Supplementary Fig. 9b), indicating the loss of structural order due to the incorporation of I2 into the porous channels. Moreover, no diffraction peaks associated with I2 were observed, ruling out the possibility that the high I2 adsorption capacity of COF-TAPT was caused by the recrystallization of I2 outside its porous structure. Solid-state 13C NMR spectra revealed that the chemical shifts of all carbon atoms in COF-TAPT changed to a certain extent after the adsorption of I2 (Fig. 6a), suggesting that I2 molecules interacted with various sites throughout the entire π-electron conjugated framework27,34. The N 1s XPS indicated that after the adsorption of I2, the peak fractions at 398.6 and 399.5 eV, assigned to imine/triazine N and sp3 N in COF-TAPT33,34, shifted to 400.7 and 401.4 eV, respectively (Fig. 6b), suggesting the formation of charge-transfer complexes between I2 and various N species in COF-TAPT. In the FTIR spectra, the adsorption of I2 caused the bands of C=N at 1625 cm−1, C=C at 1586 cm−1, C–N (in N-ph3) at 1361 cm−1, and C–N (in ph-N=C) at 1195 cm−1 to decrease in intensity or shift (Fig. 6c), indicating that all functional groups in the entire framework of COF-TAPT interact with I227,33,34. The Raman spectrum of I2@COF-TAPT exhibited characteristic bands at 107.5, 142.2, and 166.7 cm−1, which can be assigned to the symmetric stretching vibration of I3, asymmetric stretching vibration of I3, and stretching vibration of I5, respectively (Fig. 6d)60,61. These spectroscopic observations indicate that various functional groups in COF-TAPT, including phenyl rings, imine and triazine moieties, and sp3 N, were all involved in forming the charge-transfer complex with I2 and that polyiodide species were produced during the adsorption process.

Fig. 6: Characterization of I2/adsorbent and CH3I/adsorbent interactions.
figure 6

a 13C NMR spectra of (A) pristine, (B) I2-saturated, and (C) CH3I-saturated COF-TAPT. b N 1s XPS spectra of (A) pristine, (B) I2-saturated, and (C) CH3I-saturated COF-TAPT. c FTIR spectra of (A) pristine, (B) I2-saturated, and (C) CH3I-saturated COF-TAPT. d Raman spectra of pure I2, pristine COF-TAPT, and I2-saturated COF-TAPT.

The CH3I-saturated COF-TAPT sample (CH3I@COF-TAPT) was also characterized using 13C NMR, N 1s XPS, and FTIR. Compared to the original material, CH3I@COF-TAPT exhibited an intense new signal at ~58.9 ppm in the 13C NMR spectrum, originating from CH3I that formed salts at various N sites through methylation reactions (Fig. 6a)34. In addition, after the adsorption of CH3I, the N 1s electron binding energies of imine/triazine N and sp3 N in COF-TAPT increased by 1.3 and 3.7 eV, respectively, providing additional evidence for the binding of CH3I on N species (Fig. 6b)1,34. Compared with I2 adsorption, CH3I adsorption resulted in a less pronounced peak shift for imine/triazine N and a more pronounced peak shift for sp3 N, implying differences in affinity of I2 and CH3I at different N sites. In FTIR, the adsorption of CH3I on COF-TAPT also resulted in the intensity change or shift of the characteristic bands of C=N and C–N bonds, and the appearance of a band at ~939 cm−1 indicated the formation of new C–N bonds on the heterocycles in COF-TAPT (Fig. 6c)31,34. Notably, the bands of C=C at 1560–1590 cm−1 were unchanged upon the adsorption of CH3I. These spectroscopic results support the conclusion that CH3I molecules are not adsorbed on the benzene ring moieties in COF-TAPT but are specifically bonded to the nucleophilic N sites through N-methylation reactions. The CH3I-adsorbed COF-TAPT exhibits strong anion-exchange ability (Supplementary Fig. 12), confirming the generation of exchangeable iodide ions.

There are three types of nucleophilic N species (i.e., imine, triazine, and sp3 N) in COF-TAPT, all of which can bind to CH3I. To gain more insight into the preferential binding sites of CH3I, we performed DFT calculations to assess their binding energies with CH3I. The calculations were conducted at the B3LYP62,63 level of the exchange functional, using TFPA-T as the model molecule to represent COF-TAPT (Fig. 7). The calculated binding energy is −15.0 kcal mol−1 for imine, −5.4 kcal mol−1 for triazine, and −2.6 kcal mol−1 for sp3 N sites, indicating that the imine groups in the COF are the most favorable adsorption sites for CH3I (Fig. 7). This result suggests that introducing imine groups and maximizing their content in the adsorbent may be a good choice to improve the ability to capture low-concentration CH3I. In previous studies, sp3 N promoted the capture of CH3I, where N was connected with alkyl chains1,47,48. However, in TFPA-T, sp3 N is connected with three benzene rings; thus, its nucleophilicity is greatly reduced due to the conjugation effect. We note that the XPS data (Fig. 6b) suggest that CH3I interacts more strongly with sp3 N than with imine/triazine N (more pronounced peak shift). This discrepancy is not fully understood and requires further exploration.

Fig. 7: Density functional theory calculations of the binding energies of CH3I with different N sites.
figure 7

TFPA-T is the model molecule used for calculations that represent COF-TAPT. In the molecular structure of TFPA-T (left panel), sp3 N, imine N, and triazine N are labeled in pink, green, and blue, respectively.

In conclusion, the development of high-performance adsorbents for the simultaneous capture of molecular iodine and organic iodides relies on understanding the similarities and differences between these two processes. Based on previous studies, we hypothesized that N-rich carbonaceous adsorbents are conducive to simultaneously capturing I2 and CH3I. Further studies revealed that I2 could be relatively easily adsorbed on a variety of electron-donor sites, including various N species and aromatic moieties, by forming charge-transfer complexes and polyiodides. Therefore, the characteristics of binding sites and textural properties (e.g., surface area and pore volume) of the adsorbent both affect I2 uptake. The adsorption of CH3I occurs specifically on nucleophilic N sites through N-methylation reactions to form salts and is unrelated to the textural properties of the adsorbent. In addition, ionic groups can strongly promote the adsorption of I2 but have little promotional effect on the adsorption of CH3I. These findings motivated the development of a COF-based adsorbent, COF-TAPT, which combines a high surface area and numerous nucleophilic N sites, including imine, triazine, and sp3 N, thereby exhibiting excellent adsorption capacity for I2 and CH3I. We evaluated COF-TAPT for I2 and CH3I adsorption under different conditions and found that it outperforms most state-of-the-art adsorbents in all measurements, especially at low-concentration conditions relevant to practical applications. In addition, we calculated the binding energies of CH3I at different N sites, and the results revealed that imine groups might be the most preferred adsorption sites.