Optical properties of polymer precursors. In this contribution, two fluorescence materials TCz and TCzP composed of carbazole as donor and dibenzo [a, c] phenazine (DPPZ) as acceptor were designed (Supplementary Fig. 1). Compared with TCz, an additional phenyl ring is inserted between carbazole and DPPZ group for TCzP, the design strategy for extended π-conjugated structure is to increase locally emissive (LE) component in excited state and to optimize the size of the micropores in CMP obtained by EP. Then, the ultraviolet-visible (UV-Vis) absorption and photoluminescence (PL) properties of TCz and TCzP were investigated to understand the basic photophysical properties (Fig. 2a). TCzP and TCz show almost the same absorption peaks about 317 nm, which are ascribed to the π-π* of carbazole. As a comparison, the TCzP shows a narrower absorption band around 429 nm than that of TCz (445 nm), indicating to a more π-π* like character. For the PL spectra, TCz and TCzP give a yellow emission with λmax at 553 nm and 548 nm, respectively, and the abnormal 5 nm blue-shift of TCzP compared to TCz in PL spectrum can be attributed to the enhanced LE component in the emissive state of TCzP30.
To better understand the excited state properties, the solvatochromic effect of TCzP and TCz were further investigated. The PL spectra of TCzP and TCz show significant red-shift as the solvent polarity increases, which is the typical feature of the charge-transfer (CT) state. Moreover, the PL peaks of TCzP exhibit a larger red-shift (185 nm) than TCz (130 nm) from hexane (HEX), ethyl ether (ETE), dichloromethane (DCM) to acetonitrile (ACN), indicating that TCzP is more susceptible to external stimulation than TCz. Obviously, these features are expected to greatly help construct rapid and sensitive sensors (Fig. 2b). Generally, it is well-known that the photoluminescence quantum yield (PLQY) of CT molecules should be decreased with the increase in the solvent polarity because the high-polarity solvents could induce the stronger CT state and result in low PLQY. Unlike the common CT molecules, the two molecules show greatly enhanced PLQY with the increase of the solvent polarity, which is the result of the HLCT formation. Nevertheless, the CT-dominated HLCT state could occur in the large-polarity solvents, resulting in a decrease in PLQY (Supplementary Table. 1)31.
In order to further clearly reveal the singlet state properties of TCz and TCzP, the linear relationships of the slope of Stokes shift (νa–νf) verse solvent polarity (f) using the Lippert-Mataga equation (Equation S3) were carried out (Fig. 2c) 32, 33. In fact, two segmental fitting lines in TCz and TCzP represent the existence of two excited states. In low-polarity solvents (f<0.12), the lesser slope fitting lines with the small dipole moments (µe) are attributed to be LE state, while in high-polarity solvents (f>0.16), the higher slope fitting lines with large µe belong to CT state. As a comparison, TCzP has a larger µe (37.11 D) related to TCz (28.72 D), resulting that TCzP shows more obvious red-shift in the solvatochromic shift. In medium-polarity solvents, the energy levels of intrinsic LE and CT excited state are relatively close, and the coupling and crossing between the LE state and the CT state promote the formation of the HLCT state34, which is consistent with solvatochromic results. In addition, the lifetime decay curves of TCz and TCzP in a medium-polarity solvent show a single exponential lifetime of 2.7 ns and 1.58 ns, respectively, indicating that the HLCT excited state is a hybrid state, rather than a single mixed state of LE and CT (Fig. 2d).
Sensing performance and sensing mechanism of polymer precursors. The fluorescence response of the two small molecules with HLCT excited state to DCP vapors was investigated first. The fluorescence change of the TCzP and TCz spin-coated films in DCP vapors was monitored as a function of exposure time (Supplementary Fig. 6). It is worth noting that the two small molecule films are highly sensitive to DCP vapors due to the de-hybridization of HLCT excited state. The actual LOD of the two films can be evaluated as 1.32 ppb, which are one of the best reported so far (Supplementary Table. 3). To further examine and confirm the detection mechanism of TCz and TCzP to DCP vapors, the titration experiments of the 1H NMR were performed. Upon addition of DCP to TCz and TCzP in CDCl3, respectively, the chemical shifts of the protons on the DPPZ move to downfield, while no significant change for the other groups (Supplementary Fig. 7). This result may be attributed to the nucleophilic substitution reaction between TCz or TCzP and DCP, and the formed intermediate quickly undergoes a hydrolysis reaction with trace water, resulting in the protonation of the N atom in the DPPZ. Furthermore, 31P NMR experiments were performed to reveal this hydrolysis process (Supplementary Fig. 8). Upon addition of TCz to DCP in CDCl3, three new chemical shift peaks appear at 1.527 ppm, 0.630 ppm and -12.599 ppm, which correspond to the intermediate, diethyl phosphate and tetraethyl pyrophosphate (TEPP), respectively. According to previous reports35, the TEPP is obtained through the nucleophilic substitution between diethyl phosphate and DCP, and diethyl phosphate is produced through the hydrolysis of intermediates with water. Therefore, the rational mechanism we speculated has been shown in Supplementary Fig. 9.
Calculated Excited State Properties. The quenching process of TCz and TCzP to DCP were confirmed by time-dependent density functional theory (TDDFT) calculations. Both the energy levels of the intermediate and protonation of two fluorescence molecules exhibits obviously decrease accompanied with satisfied quenching (Supplementary Fig. 10). The S1 energy level difference of TCz (1.21 eV) before and after protonation is significantly smaller than that of TCzP (2.28 eV), indicating that TCzP is more sensitive to DCP than TCz. In addition, the energy level and the oscillator strength (f) of phosphorylated intermediates is lower than that of protonated products, which makes it possible to conclude that the two fluorescence molecules are more inclined to generate phosphorylated intermediates and then form protonated product. Then, the essence of their excited state quenching process was also revealed by natural transition orbit (NTO) in terms of S1 excited states (Fig. 3). Before protonation, the S1 transition of the two small fluorescence molecules is mainly the LE-dominated HLCT excited state, which is ascribed to S1 transition of DPPZ. After protonation, the vertical electron effect of the TCz and TCzP is forbidden, resulting in the intra-molecular charge transfer (ICT) transitions, which correspond to the new absorption peak at 592 nm and 521 nm in the absorption spectrum, respectively (Supplementary Fig. 11). In this case, the uniform HLCT state of the two fluorescence molecules becomes a separate LE state and CT state, leading to a significant fluorescence quenching. Upon addition of DCP into the two fluorescence molecules, their fluorescence lifetime recorded by transient PL spectra enhances significantly, indicating the S1 transition is forbidden while generating a new none-emissive CT state (Supplementary Fig. 12), which is agreement with the results by the NTO calculation. However, the two fluorescence films faced serious photo-bleaching problems, which greatly shorten the working life of these films and are not suitable for device integration.
The preparation of CMP films and pore size analysis. In order to further improve the detection performance and the optical stability of these films, the CMP films were prepared by cyclic voltammetry (CV) in a standard three-electrode electrochemical cell36. In the single-cycle CV curve, the initial oxidation potentials of TCz and TCzP are observed at 1.03 V and 0.98 V, which are attributed to the oxidation of carbazole group. As the scanning potential increases to 1.11 V and 1.03 V in TCz and TCzP, respectively, indicating more carbazole groups were oxidized and produce more carbazole radical cations (Fig. 4a). Then, an oxidation potential appears at 1.12 V for TCzP, which is assigned to the oxidation of the benzene ring connected to carbazole. As the scanning potential continues to increase, TCz and TCzP generate a peak potential at 1.23 V and 1.35 V, respectively, indicating that DPPZ was oxidized. During the negative scan, two obvious reduction peaks of TCz are observed at 0.88 V and 1.09 V, which correspond to the reduction of dimeric carbazole cations and DPPZ. Correspondingly, the reduction peaks of TCzP are located at 0.74 V and 1.01 V. Accordingly, the high oxidation potentials of 1.2 V and 1.1 V were selected for the preparation of TCz-CMP and TCzP-CMP films, respectively (Supplementary Fig. 13). This oxidation potential can ensure the effective coupling reaction of the carbazole groups while avoiding the oxidation of DPPZ to increase the fluorescence intensity of the CMP films. In this case, the oxidation of carbazoles forms carbazole cation radicals which can quickly undergo a coupling reaction with a neutral carbazole to form a dimeric carbazole cation with a lower oxidation potential, and CMP films are obtained by the redox reaction of the dimeric carbazole cations. In the multicycle CV curves of TCz and TCzP, the increase of the reduction peak indicates that the CMP films are starting to be deposited on the electrode, which is the proof of the formation of CMP films (Supplementary Fig. 13). The CMP films with uniform surface morphology can be observed by high-resolution transmission electron microscope (HRTEM) (Supplementary Fig. 14). Moreover, the film thickness of TCz-CMP and TCzP-CMP can be precisely controlled by scanning cycles. It can be seen that for TCz-CMP and TCzP-CMP films, the film thickness has a good linear relationship with the scanning cycles and increases by 1.05 nm and 2.95 nm per scanning cycle, respectively (Supplementary Fig. 15). The formation of TCz-CMP and TCzP-CMP films were further confirmed by Fourier transform infrared (FT-IR) spectroscopy (Supplementary Fig. 16). TCz-CMP and TCzP-CMP show a new peak appears about 804 nm that is assigned to dimeric carbazole, whereas the absorption peaks of DPPZ at 766 nm and 775 nm were almost unchanged before and after EP, indicating that the oxidation of carbazole did not affect DPPZ. In addition, the powder X-ray diffraction (XRD) patterns of TCz-CMP and TCzP-CMP demonstrate a broad and dispersion peak within the 2θ range of 5-40° (Supplementary Fig. 17), indicating the amorphous feature of CMP films, which is expected for the preparation of the CMP films.
To further study the microporous structure of TCz-CMP and TCzP-CMP, we conducted nitrogen adsorption/desorption experiments and evaluated their porosity at 77.3 K (Fig. 3b). Both TCz-CMP and TCzP-CMP show typical i-type and iv-type nitrogen adsorption isotherms according to the IUPAC classification. The obtained curve shows that nitrogen uptake increases sharply at low relative pressure (p/p0 less than 0.05), which indicates the existence of a microporous structure. Under relative high pressure (0.4-1.0), the hysteresis loop shows the coexistence of micropores and mesopores. The Brunauer-Emmett-Teller (BET) surface area and a total pore volume were calculated to be 407 m2/g and 1.29 cm3/g for TCzP-CMP. In a comparison, TCz-CMP has a smaller BET surface area (137 m2 /g) and total pore volume (0.12 cm3/g). Correspondingly, the pore size distributions of TCzP-CMP and TCz-CMP are 0.81-1.0 nm and 0.71-0.74 nm, respectively (Fig 4c). After the optimization by TDDFT, the possible microporous structures of TCzP-CMP and TCz-CMP were obtained (Fig. 4d). However, the pore size of TCz-CMP is smaller than that of DCP (0.84 nm), which is not conducive to the diffusion of DCP into the interior. In contrast, the pore size of TCzP-CMP is suitable for the diffusion of DCP.
Adsorption performance test of CMP. The adsorption capacity of CMP to the analytes is an important factor affecting the sensing performance, so a simple self-made adsorption system was first built to evaluate the adsorption capacity of TCz-CMP and TCzP-CMP to DCP vapors (Supplementary Fig. 18). It is found that the adsorption capacity of the two CMP to DCP vapors increases rapidly within 200 min, which is attributed to the existence of sufficient action sites in CMP (Fig. 5a). Then, as the exposure time prolonged, the adsorption capacity shows a slow tendency, which is attributed to that most sites are occupied, resulting that the adsorption rate gradually decreases and finally reaches equilibrium at 1400 min. The maximum adsorption capacity of TCzP-CMP can reach 936 mg/g, while TCz-CMP is only 75 mg/g. This may be because TCzP-CMP has a larger pore structure and specific surface area than TCz-CMP, which is more conducive to the diffusion of DCP molecules into CMP. Then, in order to understand the adsorption process in depth, the adsorption kinetics were evaluated using the pseudo first-order (PFO) (Equation 1), pseudo-secondary (PSO) (Equation 2) model and intra-particle diffusion model (Equation 3)37, 38.
Comparing the two correlation coefficients (R2) of TCz-CMP and TCzP-CMP to DCP vapors, the PSO model (R2>0.99) has better correlation than the PFO model (R2 <0.96), which indicates that the PSO model is more suitable for describing the adsorption process for DCP than the PFO model (Fig. 5b-c). The results show that the adsorption of DCP by TCz-CMP and TCzP-CMP is a chemical adsorption process, which may depend on the nucleophilic substitution reaction and electronic interaction between the two CMPs and DCP37. The experimental adsorption capacity of TCz-CMP of 78 mg/g is much lower than that of TCzP-CMP (936 mg/g), once again proving that the pore size directly affects the diffusion of DCP. The calculated equilibrium adsorption capacity qe (976 mg/g) of TCzP-CMP is close to the experimental value. Furthermore, the intraparticle diffusion model is adopted to fit experimental data, revealing the rate-controlling steps in the DCP adsorption process. It can be observed that there are two linear relationships in the entire time range of the adsorption process, which indicate that surface adsorption and intraparticle diffusion synergistically affect the adsorption process (Fig. 5d). For TCzP-CMP, the fitting line of intra-particle diffusion almost passes through the origin (C1=6), which proves that the intra-particle diffusion of TCzP-CMP is a speed-controlled step in the first adsorption stage of DCP, and there may be no boundary effect. Concurrently, we were surprised to find that TCzP-CMP not only has good adsorption performance for DCP vapors, but also for other CWA simulants, such as DCNP, DMMP, TEP and 2-CEES. Similarly, the adsorption kinetics of PFO, PSO and intraparticle diffusion models were evaluated (Supplementary Fig. 19). It can be seen that the adsorption of these CWA analogs by TCzP-CMP is more consistent with the PSO model, which may depend on the point-to-dipole π interaction37, 39 (Supplementary Table. 4-5). More importantly, it has been proved that the adsorption capacity of these five CAW simulants by TCzP-CMP is 2-3 times that of activated carbon (Supplementary Table. 6). Therefore, this TCzP-CMP as the adsorption material for CWA has potential application in military protection.
Optical properties of CMP films. Compared with the spin-coated films of TCz and TCzP monomers, the emission peaks of TCz-CMP and TCzP-CMP films are red-shifted by 60 nm and 41 nm, respectively, which are ascribed to the extended π-conjugate structure (Supplementary Fig. 20). To study the fluorescence stability of these two CMP films, the TCz-CMP films and TCzP-CMP films were continuously irradiated under the excitation light with the maximum excitation wavelength at 450 nm and 440 nm for 90 s. The fluorescence intensity of the TCz-CMP films nearly has no changes, while that of the TCzP-CMP films is only quenched by 7%. Both the TCz-CMP and TCzP-CMP films demonstrate much better stability than their monomers films because of their cross-linked structure, thus effectively overcoming the photobleaching problems, which is fully satisfactory for practical applications (Supplementary Fig. 21).
Sensing performance of CMP films. The fluorescence response of these two CMP films to DCP vapors was investigated (Fig. 6a-b). The actual LOD of TCz-CMP can be determined to be 132 ppt, which is an order of magnitude lower than monomer spin-coated films. The excellent can be attributed to the susceptible "on-off" effect of hybridization and dehybridization of HLCT materials, and the "molecular wire effect". As a comparison, the LOD of TCzP-CMP (13.2 ppt) is an order of magnitude lower than that of TCz-CMP. This enhanced detection performance can be ascribed the diffusion of DCP into the CMP films. The coefficient of determination (R2) of TCz-CMP (0.9996) and TCzP-CMP (0.9850) films shows that the stability of these films can effectively improve the detection accuracy (Fig. 6c-d). Importantly, TCz-CMP and TCzP-CMP films have a rapid response to the low concentration of DCP vapors within 2.1 s and 5 s (Fig. 6e-f), which fully meets the requirements of real-time detection. Furthermore, it is found that the two quenched CMP films by high-concentration of DCP vapors (1.32 ppm) were degassed in a vacuum oven at 50 ℃ for 2 h, the fluorescence intensity of the two CMP films can be restored to the original level, respectively. Even though recycled six times, the two CMP films still show a good response to DCP vapors (Fig. 6g-h). In a comparison, TCzP-CMP films exhibit better self-recovery than TCz-CMP films, which may be due to the fact that the pore size of the TCzP-CMP films is larger than that of the TCz-CMP films and DCP molecules, which facilitate the escape of DCP molecules under vacuum conditions. Further，the fluorescence response of the two CMP films to the possible interferences was investigated. Upon exposure to DCP (1.32 ppm), ethanol (780 ppm), water (32000 ppm), trifluoroacetic acid (TFA 13 ppm), DCNP (2 ppm), DMMP (200 ppm), triethylamine (1090 ppm), 2-CEES (38 ppm), pyridine (25 ppm), acetone (816 ppm), n-hexane (21 ppm) vapors and toluene (140 ppm), respectively, the TCz-CMP and TCzP-CMP films exhibit higher selectivity to DCP vapors than other interferences (Supplementary Fig. 22).
System assembly and testing. To verify the practical applicability of CMP films, a portable fluorescent detection systems consisted of intake pump, TCzP-CMP films, LED (365 nm), optical recognition detection system, Bluetooth transceiver, and an optimized sealed cavity was built (Fig. 7a, c). The working principle of the device is as follows: using STM32 as the Microcontroller Unit (MCU), controlling the switch of the air pump and driving the linear array CCD sensor to collect signals, and then send the signals to the upper computer (PC) through the Bluetooth transceiver (Fig. 7b). The switch of the LED light is synchronized with the CCD signal collection. Then, the various level of DCP vapors were injected into the sealed cavity for 10 seconds through the intake pump and the fluorescence signal was analyzed by operating software. It can be observed that the changes of fluorescence intensity to DCP vapors show a good linearity (R2=0.98), and the LOD can be determined as 13.2 ppt (Fig. 7d). In addition, the portable fluorescence detection systems with a Bluetooth device can realize remote and wireless monitoring the DCP vapors, thereby avoiding the risk of direct detection by personnel. The fluorescent system equipped with TCzP-CMP films show rapid and sensitive performances for DCP detection, indicating that they can fully meet the requirements of practical applications.