Achieving Sub‐ppm Sensitivity in SO₂ Detection with a Chemically Stable Covalent Organic Framework

Abstract

We report the inaugural experimental investigation of covalent organic frameworks (COFs) to address the formidable challenge of SO₂ detection. Specifically, an imine‐functionalized COF (SonoCOF‐9) demonstrated a modest and reversible SO₂ sorption of 3.5 mmol g⁻¹ at 1 bar and 298 K. At 0.1 bar (and 298 K), the total SO₂ uptake reached 0.91 mmol g⁻¹ with excellent reversibility for at least 50 adsorption‐desorption cycles. An isosteric enthalpy of adsorption (ΔH_ads ) for SO₂ equaled −42.3 kJ mol⁻¹, indicating a relatively strong interaction of SO₂ molecules with the COF material. Also, molecular dynamics simulations and Møller–Plesset perturbation theory calculations showed the interaction of SO₂ with π density of the rings and lone pairs of the N atoms of SonoCOF‐9. The combination of experimental data and theoretical calculations corroborated the potential use of this COF for the selective detection and sensing of SO₂ at the sub‐ppm level (0.0064 ppm of SO₂).

This study introduces the first experimental exploration of COFs for SO₂ detection. The imine‐functionalized COF exhibited reversible SO₂ sorption with over 50 cycles of reversibility. The isosteric enthalpy of adsorption was measured at −42.3 kJ/mol, reflecting a strong SO₂‐COF interaction. Molecular dynamics and Møller–Plesset calculations elucidated the interaction mechanisms, underscoring the COF′s potential for selective sub‐ppm SO₂ sensing.

Introduction

Sulfur dioxide (SO₂), an irritant, colorless, and non‐flammable gas with a pungent odor, is cataloged by the World Health Organization as a highly venomous chemical to humans. Exposures exceeding 100 ppm of SO₂ can be fatal in only a few minutes because it can be easily absorbed by dermal contact and the respiratory system.[ ¹ , ² ] Even at concentrations lower than 100 ppm, SO₂ can be responsible for bronchoconstriction and impair lung function. For example, exposure to just 1.5 ppm of SO₂ for a few minutes can induce temporary respiratory impairment. ^[3] Chronic bronchitis, laryngitis, and severe respiratory tract infections can be caused in healthy individuals at SO₂ concentrations higher than 1.5 ppm. ^[4] Thus, air quality guidelines recommend limiting human exposure to SO₂ to 500 μg m⁻³ (175 ppb) over a 10‐minute period and 20 μg m⁻³ (8 ppb) for daily averages. ^[5] SO₂ serves as a precursor to particulate matter (PM), which is toxic to humans as well. ^[6] Additionally, SO₂ is highly soluble in water, forming sulfurous acid, which can further oxidize to sulfuric acid, a major contributor to acid rain. This damages aquatic environments and hampers the growth of forests and crops. ^[7] In urban areas, acid rain can accelerate metal corrosion and deteriorate materials like limestone, marble, and mortar. ^[8] Hence, it is crucial to not only enhance air quality by diminishing emissions or augmenting SO₂ capture but also monitor potentially polluted environments with SO₂. Such measures are essential for averting severe health repercussions for individuals laboring in various urban settings. However, up to now, the vast majority of technological efforts toward SO₂ have been focused on the capture of this toxic gas from scrubbers (based on aqueous alkaline solutions and wet‐sulfuric‐acid processes), with apparent drawbacks, including huge quantities of wastewater, pipeline corrosion, and high operational and recovery costs. ^[9] Current solid‐state materials for SO₂ capture, e.g., metal oxides ^[10] and zeolites, ^[11] have considerable disadvantages like high activation temperatures (above 250 °C) and reduced porosity over multiple cycles. Consequently, there is a pressing need for materials capable of effectively detecting SO₂, exhibiting both high responsiveness and chemical stability.

Porous materials, notably metal–organic frameworks (MOFs), have garnered attention for their potential in SO₂ capture and sensing due to their high porosity. ^[12] Some promising results from selected examples demonstrate remarkable SO₂ capture, ^[13] even under humid conditions. ^[14] Nonetheless, many MOFs have relatively poor chemical stability to SO₂/water ^[15] and low detection limits to SO₂. ^[16] In contrast, covalent organic frameworks (COFs), an emerging class of porous materials synthesized through reticular chemistry with building blocks containing light elements (e.g., C, H, O, N, and B) interconnected by covalent bonds, have exhibited elevated stability. ^[17] Their intrinsic stability, rooted in the covalent nature of their bonds, imparts greater stiffness to their structures, enhancing thermal and chemical stability and rendering them resilient to adverse conditions. ^[18] Apart from high stability, the functional diversity and structural adjustability of COFs find significant applications in several areas, including gas storage and separation, ^[19] heterogeneous catalysis, ^[20] drug delivery, ^[21] and electronic devices. ^[22] However, to our knowledge, no studies have specifically addressed the use of COFs for SO₂ detection despite the considerable potential demonstrated by COFs in this field.

Here, we present the first example of a robust and crystalline COF, SonoCOF‐9 (Figure 1), ^[23] that addresses SO₂ detection and sensing effectively. SonoCOF‐9 exhibits chemical stability to SO₂, maintaining consistent capture capacity over 50 adsorption–desorption cycles without losing its crystallinity and porosity. Fluorescence experiments further revealed SonoCOF‐9 as a promising and effective selective detector and sensor for SO₂ at the ppm level.

Figure 1.

a) Structure of SonoCOF‐9. b) PXRD of SonoCOF‐9. c) HR‐TEM image of SonoCOF‐9 (insert shows the FFT pattern, where 3.7 nm is the d‐spacing of (200) reflection).

Results and Discussion

Powder X‐ray diffraction (PXRD) confirms the phase purity of SonoCOF‐9 (Figure 1b). It showed an excellent crystallinity with two strong diffraction signals at 2.16° and 3.79°, and relatively weak signals at 4.38°, 5.81°, and 7.61°, which correspond to the (200), (001), (111), (221), and (202) reflections, respectively. Structural simulation (Figure S1) reveals that sonoCOF‐9 shows a monoclinic structure with the ffc net, together with the Pawley refinement profile (blue dots) aligning closely with the experimental pattern (Figure 1b, R_wp of 4.24 % and R_p of 3.09 %). High‐resolution transmission electron microscopy (HR‐TEM) images (Figure 1c and Figure S2) depict well‐defined and highly crystalline SonoCOF‐9 crystallites. N₂ sorption test at 77 K (Figure S3) demonstrates a Brunauer–Emmet–Teller (BET) surface area of 1147 m² g⁻¹ and a pore volume of 0.52 cm³ g⁻¹. The pore size distribution of SonoCOF‐9 centers around 18 Å (Figure S4), which agrees with the pore width on the crystalline structure (19.5 Å).

Based on the SO₂ isotherm of this COF (Figure 2a), an SO₂ uptake of approximately 0.91 mmol g⁻¹ was obtained at 0.1 bar. The SO₂ adsorption isotherm exhibits an approximately linear increase from 0.1 to 0.5 bar, reaching a total uptake of ca. 2.9 mmol g⁻¹. Finally, a relatively flat trend of SO₂ sorption was observed from 0.5 to 1.0 bar, with a total SO₂ uptake of 3.5 mmol g⁻¹. This total SO₂ uptake is comparable to several typical chemically stable porous materials.[ ²⁴ , ²⁵ ] For example, CTF‐CSU41 shows a higher SO₂ uptake of 6.7 mmol g⁻¹ at 298 K and 0.11 bar under kinetic conditions over a 360‐minute period. ^[26] For an amine‐functionalized COF, PI‐COF‐m60 displays 4.74 mmol g⁻¹ at 298 K during a 60‐minute kinetic test. ^[27] At similar conditions, ICTF‐SCN exhibits 9.22 mmol g⁻¹ at 298 K and 1 bar over a 60‐minute kinetic test. ^[28] Under static conditions, TAM‐POF shows 9.45 mmol g⁻¹ at 298 K and 1 bar. ^[29]

Figure 2.

a) The SO₂ isotherm of SonoCOF‐9 at 298 K and 1 bar. b) Adsorption–desorption cycles of SO₂ in SonoCOF‐9 at 298 K and 0.1 bar. c) PXRD before and after SO₂ sorption performance. d) SO₂ interaction with SonoCOF‐9.

A slight hysteresis was observed in the SO₂ desorption isotherm of SonoCOF‐9, indicating a relatively strong SO₂/COF interaction energy. A key characteristic of a chemically stable material for SO₂ detection is a relatively high interaction of SO₂ with particular functional groups of the material. ^[30] Such interactions are most evident at the low pressures (P <0.1 bar). Since SO₂ detection typically focuses on concentrations at the ppm level, these concentrations correspond to the low‐pressure range. For instance, flue gas shows SO₂ concentrations up to 10,000 ppm, equivalent to pressures below 0.05 bar. ^[31]

Cycling experiments with SO₂ at 298 K and 0.1 bar were conducted to assess the stability of the SO₂ sorption performance. The results showed that the SO₂ capture performance remained consistent over 50 adsorption–desorption cycles with a capacity of 0.92±0.11 mmol g⁻¹ (Figure 2b). This SO₂ cyclability corroborates that SO₂ can be fully released during subsequent desorption cycles. Additionally, PXRD and Fourier‐transform infrared spectroscopy (FTIR) analyses of sonoCOF‐9 following 50 adsorption–desorption cycles verified retained crystallinity and stability (Figure 2c and Figure S5).

The host–guest interaction (SonoCOF‐9⋅⋅⋅SO₂) was estimated by calculating the isosteric enthalpy of adsorption (ΔH _ads) for SO₂ at low coverage using a fully activated sample of SonoCOF‐9 at 298 and 308 K (Figure S7). ^[32] The calculated ΔH _ads (−42.3 kJ mol⁻¹) indicates stable adsorbed SO₂ layers on the COF's walls. This ΔH _ads value is coherent with the observed SO₂ isotherm hysteresis, which can be associated with a strong interaction of SO₂ molecules with the π density of the rings and lone pairs of the N atoms from the ETTA building block, as disclosed by the classical reactive molecular dynamics simulations (ReaxFF MD) results and Møller–Plesset perturbation theory calculations (MP2) on reduced model systems of the COF. We also demonstrated that the decrease of ΔH _ads as the SO₂ adsorption increased (Figure S8) was probably due to the addition of the weaker binding species that laid on the ring plane connected to the COF structure through unconventional hydrogen bonds between the SO₂ oxygen and the hydrogen of the benzylic ring (Figure 2d). However, being only a small percentage (around 25 %), these configurations had an almost undetectable effect on the total binding values, which were estimated from the MP2‐optimized models after grouping the adsorbates into families (i.e., stacked and hydrogen bonded).

The binding energy and enthalpy are displayed in Tables S1 and S2. The binding tendencies are clearly displayed in Figure S10, where a snapshot of the RMD trajectory is depicted. There, the occupation of the various regions of the COF framework is highlighted by the spatial distribution contours (dark yellow areas). It is apparent that the higher occupation densities are around the circular ten‐ring bands of the COF, where the molecules are preferentially accommodated on top of the ring planes, interacting with the framework but also with each other. To give an idea of the typical locations, we examined the atom‐atom radial distribution functions of S and O in relation to the carbon and hydrogen atoms of the COF. The position of the peak at short distances (lower than 3 Å, Figure S11) suggests that SO₂ tends to stay connected to the framework but remains slightly further from the nitrogen sites (approximately 3.2 Å), which is most probably due to self‐intermolecular interactions that mitigated the adsorption distances. ^[33] Indeed, the evaluation of the binding energy on reduced model systems revealed that, on the one hand, these self‐interactions weakened the binding energy of each single molecule but, on the other hand, improved the stability of the formed SO₂ shells.

In the dominant adsorption mode (75 % population), where the molecules were on top of the ring systems, the strongest binding was around −42 kJ mol⁻¹, whilst the weakest was approximately −20 kJ mol⁻¹. A representative geometry (extracted from the RMDs and optimized at the MP2 level, Figure S12) used for estimating the binding energy is displayed in Figure 2d. There, both characteristic adsorptions, namely in‐plane and out‐of‐plane, are shown. The average binding energy estimated considering the populations and the correction for the enthalpic contribution was approximately −30 kJ mol⁻¹. The trend of the evolution of the adsorption enthalpy as a function of the COF uptake (Figure S8) indicates that this estimate corresponds to a scenario of intermediate coverage (Figure 2d), where the most interacting sites are occupied, and the other approaching guest molecules have a weaker interaction with the COF framework.

N₂ sorption isotherm at 77 K showed that the surface area of this COF after SO₂ sorption did not decrease significantly (from 1147 to 1053 m² g⁻¹, Figure S13). Moreover, the pore size distribution of SonoCOF‐9 centers after SO₂ exposures is around 16 Å (Figure S14), which agrees with the value of the pristine material. These results demonstrate that SO₂ can be fully released during SO₂ cycle experiments without affecting the crystallinity and porosity of SonoCOF‐9. Scanning electron microscopy (SEM) analyses of the pristine COF after SO₂ saturation and after SO₂ adsorption‐desorption cycles revealed no significant changes in the submicrometric crystallite morphology and particle size of the material, supporting its stability and recyclability in the presence of SO₂ (Figure S15).

Thus far, we have successfully demonstrated the reversible SO₂ capture by SonoCOF‐9, particularly at low pressure (0.1 bar), identified the preferential adsorption sites of SO₂, and evaluated the extraordinary chemical stability of this COF to SO₂ uptake. Both reversible adsorption and stability are indeed ideal for SO₂ detection. Selectivity is another crucial aspect of any detector. To compare the SO₂/CO₂ selectivity, CO₂ sorption isotherm was collected at 298 K up to 1 bar (Figure S16). The SO₂ uptake in SonoCOF‐9 surpassed the CO₂ uptake across the entire pressure range. Subsequently, the selectivity of SO₂ over CO₂ at varying compositions (SO₂/CO₂) was calculated using the Python package (Table S3), pyIAST. ^[34] Our result indicates a remarkable SO₂/CO₂ selectivity by SonoCOF‐9, especially at low SO₂ concentrations; e.g., a selectivity of 25.44 was estimated for a 1 : 99 SO₂/CO₂ mixture and 12.93 for a 10 : 90 mixture. These values are higher than or comparable to some representative porous materials, such as CPL‐1 (8.7 at 10 : 90 SO₂/CO₂), ^[35] Ni‐gallate (25 at 10 : 90 SO₂/CO₂), ^[36] NaX zeolite (3.2 at 10 : 90 SO₂/CO₂), ^[11] and MFM‐300 (31 at 10 : 90 SO₂/CO₂). ^[37]

Upon systematically establishing the promising SO₂ adsorption characteristics of SonoCOF‐9, we explored the potential utilization of this COF material as a fluorescent SO₂ detector. Due to the highly conjugated crystalline structure, COF displays multiple identical binding sites across the framework. Therefore, the signal can be effectively measured when the sample is stimulated, which can be related to fluorescence change. ^[38]

Initially, under visible light irradiation at λ_ex=527 nm (Figure S19), an activated sample of SonoCOF‐9 exhibited a broad photoluminescence signal centered at λ_max=622 nm (Figure 3a). Another activated sample of SonoCOF‐9 was exposed to 0.1 bar of SO₂ using our homemade in situ adsorption system (Figure S17) to assess its photoluminescence absorption properties. Remarkably, the photoluminescence was almost wholly quenched (Figure 3a). This turn‐off effect at low pressure could be associated with the high interaction between SO₂ molecules and SonoCOF‐9. A set of control experiments was carried out to evaluate the selectivity of SonoCOF‐9. The fluorescence of SonoCOF‐9 when exposed to air and CO₂ (saturated) was measured. Notably, these exposures did not induce significant changes in the shape or intensity of the photoluminescence relative to the spectrum of activated SonoCOF‐9 (Figure 3a). Thus, the lack of substantial alterations in the emission shape or intensity during exposures to air and CO₂ supports the selectivity of SonoCOF‐9 for SO₂. This suggests that the observed fluorescence change is solely attributable to SO₂ adsorption and is not influenced by the inclusion of other gases. The spectrum of a SonoCOF‐9 sample saturated with H₂S was also carried out to test the effect of another acid gas on the material (Figure S20). Although H₂S generated a change in the fluorescence of the material, this was not identical to that generated with SO₂, since the quenching was not as pronounced as with SO₂, confirming that the fluorescence response is more specific and sensitive to SO₂.

Figure 3.

a) Solid‐state photoluminescent emission of SonoCOF‐9 with air, CO_2, and SO₂. b) Time‐resolved photoluminescence (TRPL) of SonoCOF‐9 before (activated) and after SO₂ sorption in the solid state. c) Linear fit, and d) TRPL of SonoCOF‐9 before (activated) and after SO₂ sorption in THF solution.

Later, five independent exposition cycles were carried out (Figure S21). An activated SonoCOF‐9 sample was exposed to 0.1 bar of SO₂, revealing a consistent absorption spectrum with an average decrease in emission intensity of 87 (±2) %. Up to this point, the results, while not directly translatable to SO₂ sensing, consistently demonstrated a reduction in emission intensity upon re‐exposure to SO₂ and re‐activation of theSonoCOF‐1 at 0.1 bar SO₂ and air, which holds promise for the transition to the next stage: SO₂ sensing. It is worth mentioning that we tested another working temperature of detection (308 K), finding a non‐temperature dependence (Figure S22).

To probe the possible mechanism of SO₂ detection by SonoCOF‐9, we conducted time‐resolved photoluminescence (TRPL) experiments through a 375 nm picosecond‐pulsed laser for excitation (Figure 3b). TRPL analyses were performed on both an activated SonoCOF‐9 and a sample exposed to 0.1 bar of SO₂. Photoluminescence decay was recorded at 622 nm. The average decay lifetime decreased moderately upon SO₂ exposure (Table S4). Specifically, the average fluorescence lifetime of activated SonoCOF‐9 (at λ_emission=622 nm) was 0.41 ns, whereas that of SonoCOF‐9 exposed to 0.1 bar of SO₂ (at λ_emission=622 nm) was 0.24 ns. Considering that the decrease in half‐life is minimal (from 0.41 ns to 0.24 ns) and the turn‐off effect is very significant, we postulate that the drastic decrease in emission intensity (≈87 %) is due to a static quenching,[ ³⁹ , ⁴⁰ ] indicating that the electronic transition π*→π centered on the building blocks of the SonoCOF‐9 structure could be avoided due to the formation of a non‐fluorescent “complex” in the basal state due to the previously discussed intermolecular interactions occurring between SonoCOF‐9 and SO₂. ^[41]

Once the selective SO₂ detection properties of SonoCOF‐9 at 0.1 bar of SO₂ were established, we aimed to explore its potential for sensing applications at very low SO₂ concentrations (sub‐ppm level). It is worth mentioning that 0.1 bar of SO₂ equals approximately 20,000 ppm (considering our experimental setup at 0.2 L and 298 K, Figure S17). This concentration is deemed high, far exceeding the ppm regime. ^[42] Although SonoCOF‐9 consistently showed a significant fluorescence change (turn‐off effect) upon exposure to 0.1 bar of SO₂, accurately quantifying the precise amount of SO₂ remains challenging. Additionally, the most pertinent SO₂ concentration regime for a sensing material is at the sub‐ppm level. Thus, solutions of SO₂ in THF were used to investigate the sensing properties of SonoCOF‐9 by luminescence. First, we measured the photoluminescence absorption properties of 2.5 mg of pristine SonoCOF‐9 suspended in 3 mL anhydrous THF, which is equivalent to a concentration of 327.8 μM SonoCOF‐9 in THF. The shape of the fluorescence spectrum agreed with the spectrum of the solid‐state material with an emission maximum at 622 nm (Figure S23). Later, a solution with a concentration of 1×10⁻⁴ mM (0.0001 mM=0.0064 ppm=6.4 ppb) of SO₂ (in anhydrous THF) was prepared, and the fluorescence emission was measured on the resultant suspension (Figure S24). Upon immersion of the SonoCOF‐9 in the SO₂ solution, the photoluminescence emission intensity increased by 70.05 %. Six independent experiments of suspensions (1×10⁻⁴ mM) with SonoCOF‐9 were prepared and analyzed by luminescence. Consistently, the same emission spectra were observed with an approximate 70 % increase in emission intensity (Figure S25), showcasing remarkable reproducibility. It is noteworthy that the change in fluorescence response was observed immediately after the immersion of the COF material in the SO₂ solution, and the response remained stable for up to 12 hours (Figure S26). Table S6 gives a brief review of different materials that have been tested for SO₂ detection. To the best of our knowledge, this is the first study to evaluate the SO₂ detection capability of a COF material.

The high affinity of SonoCOF‐9 for SO₂ enables us to estimate the limit of detection (LOD) by measuring fluorescence emission intensity as a function of the SO₂ concentration. The LOD was calculated with the formula: LOD=3σ/m, where σ is the standard deviation of the initial intensity of pristine SonoCOF‐9, and m represents the slope of the linear fit from the experimental data (Figure 3c). ^[43] The resulting LOD value for SO₂ sensing in a THF solution was estimated to be 1.36 mM. Additionally, the Stern–Volmer plot (Figure S27) affords a quenching parameter (Ksv) of 7.6×10⁻⁶ M⁻¹. This relatively low value corroborates the high affinity of SonoCOF‐9 for SO₂. ^[44] The value agrees with the detection of acid compounds (3.12×10⁻⁷ M⁻¹). ^[45]

We performed TRPL experiments (Figure 3d) on THF suspensions of a pristine sample of SonoCOF‐9 and a sample exposed to 1×10⁻⁴ mM of SO₂, where measurements were performed at λ_em=622 nm. Herein, an apparent decrease in photoluminescence half‐life times was obtained after SO₂ exposure (Table S5), going from 5.25 ns for the pristine sample suspension to 1.23 ns for the sample exposed to 1×10⁻⁴ mM SO₂. This apparent decrease in the half‐life times suggests that a small amount of SO₂ is enough to stiffen SonoCOF‐9, which in turn results in a consistent photoluminescence turn‐on effect of ca. 70 %. These results again confirm the propensity of SO₂ molecules to interact favorably with the COF framework.

At a very low concentration of SO₂ (0.0064 ppm), there is an effective interaction between SonoCOF‐9 and SO₂ molecule, which enhances the stiffness of the structural molecular motions, hindering the non‐radiative decay pathways of the excited state.[ ⁴⁶ , ⁴⁷ ] In other words, vibrational relaxations with half‐lives of 10⁻¹² s or less lead to an increased fluorescence lifetime, and the increased number of excited species undergoing radiatively decay contributes to enhanced photoluminescence. On the other hand, at high SO₂ concentrations (20,000 ppm), after the continuous interaction of SO₂ with the COF, the preferential adsorption sites of the COF became saturated, and the interactions between SO₂ molecules, SO₂⋅⋅⋅SO₂, occurred as previously reported, ^[48] leading to a quenching effect. More specifically, our current hypothesis is that at high SO₂ concentrations, the predominant interactions and exchange dynamics within the pores of the COF are governed by SO₂⋅⋅⋅SO₂ interactions rather than COF⋅⋅⋅SO₂ interactions. This preference for SO₂⋅⋅⋅SO₂ interactions leads to minimizing the stiffening effects and additional beneficial electronic effects arising from the COF⋅⋅⋅SO₂ interaction, thereby quenching the fluorescence in favor of non‐radiative processes (vibrational relaxation).[ ⁴⁹ , ⁵⁰ ]

Conclusion

In conclusion, we report the first experimental investigation on the sorption and detection of SO₂ on a COF (SonoCOF‐9). While the SO₂ uptake of 3.5 mmol g⁻¹ (298 K and 1 bar) does not rival the leading MOF materials and polymers, SonoCOF‐9 demonstrates outstanding chemical stability toward SO₂ and maintains excellent cyclability over a minimum of 50 sorption/desorption cycles without compromising its crystalline structure. Notably, the combination of imine functionalities and aromatic moieties of SonoCOF‐9 was demonstrated to be the key for the selective detection of SO₂ with the relatively high energy of interaction (−42.3 kJ mol⁻¹), which also provided the ability to sense SO₂ at the sub‐ppm level (0.0064 ppm), making SonoCOF‐9 promising for practical SO₂ sensing.

Conflict of Interests

There are no conflicts to declare.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Zhao W., Obeso J. L., López-Cervantes V. B., Bahri M., Sánchez-González E., Amador-Sánchez Y. A., Ren J., Browning N. D., Peralta R. A., Barcaro G., Monti S., Solis-Ibarra D., Ibarra I. A., Zhao D., Angew. Chem. Int. Ed. 2025, 64, e202415088. 10.1002/anie.202415088

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.