Synthesis of Cs2PdBr6
0.426 g CsBr (2 mmol, TCI, 99% purity), 0.266 g PdBr2 (1 mmol, Aladdin, 99% purity) and 5 mL of 48% HBr (Macklin) were added to the three-necked flask, and the solution was heated to 85°C with stirring for 5 min. The solution was continued to be heated. 0.5 mL of dimethyl sulfoxide (Macklin, AR) was added to it when the solution temperature reached 120°C. In order to fully react, the solution was continued to stir for 10 min. After the solution cooled to room temperature, the solution containing the black Cs2PdBr6 crystalline was filtered and washed several times with toluene. The product was dried under reduced pressure at 100°C overnight.
Synthesis of Cs2AgBiBr6
Cs2AgBiBr6 was successfully synthesized according to the reported method. 35 0.426 g CsBr (2 mmol, TCI, 99% purity), 0.449 g BiBr3 (1 mmol, TCI, 99% purity) and 0.188 g AgBr (1 mmol, TCI, 99% purity) were mixed with 10 mL of 48% HBr in a round-bottomed flask. The solution was continuously stirred for 2 hours at 120°C. The solution was allowed to stand for 2 hours after cooling to room temperature to obtain an orange precipitate. Subsequently, the solution containing the orange Cs2AgBiBr6 crystalline was filtered and washed several times with ethanol. The product was dried under reduced pressure at 100°C overnight.
Synthesis of Cs3Bi2Br9
Cs3Bi2Br9 was successfully synthesized according to the reported method. 36 0.638 g CsBr (3 mmol), 0.897 g BiBr3 (2 mmol) and 5 mL of 48% HBr were added to the three-necked flask. The solution was continuously stirred at 80°C, heated for 1 h, and then cooled to room temperature. The solution containing yellow Cs3Bi2Br9 powder was filtered and washed several times with ethanol. The product was dried under reduced pressure at 100°C overnight.
Synthesis of Cs2SnI6
Cs2SnI6 was successfully synthesized according to the reported method.37 3.258 g Cs2CO3 (10 mmol, Aladdin, 99% purity) were mixed with 20 mL of 55% HI (Macklin) in a 100 ml beaker to afford a concentrated acidic solution of CsI. 3.132 g SnI4 (5 mmol, Aladdin, 99% purity) was dissolved in 10 mL ethanol to afford a clear orange solution. The SnI4 solution was added to the CsI solution under stirring, and black solids were continuously precipitated. In order to fully react, the solution was continued to stir for 10 min. The solution containing Black Cs2SnI6 powder was filtered and washed several times with ethanol. The product was dried under reduced pressure at 100°C overnight.
Synthesis of TpPa-1 COF
126 mg triformylphloroglucinol (Tp) (0.6 mmol, Macklin, 97% purity), 96 mg p-phenylenediamine (Pa-1) (0.9 mmol, Aladdin, 99% purity), 16.5 mL of mesitylene (TCI, 97% purity) and 16.5 mL of dioxane (Aladdin, 99% purity) were added to the vial and sonicated for 10 min. Subsequently, the mixture was transferred into a 50 mL Teflon-lined stainless steel autoclave, and 1 mL of 6 M aqueous acetic acid was added and sonicated for 10 minutes. Finally, the autoclave was heated to 120°C and kept for three days. The product was filtered and washed with N,N-dimethylacetamide (TCI, 99% purity), anhydrous tetrahydrofuran (TCI, 99% purity) and acetone. The collected powder was then dried at 120°C under vacuum for 12 hours to give TpPa-1 COF.
Synthesis of TpPa-2, TpPa-CN, TpPa-NO2 and TpPa-COOH
The method is the same as the synthesis of TpPa-1, and only the corresponding raw materials need to be replaced.
Characterizations of COFs
Five 2D COFs also were synthesized by the Schiff base reactions of 1,3,5-triformylphloroglucinol (Tp) and p-phenylenediamine (Pa-1) (2,5-dimethyl-p-phenylenediamine (Pa-2), 2,5-diaminobenzonitrile (Pa-CN), 2-nitro-1,4-phenylenediamine (Pa-NO2) and 2,5-diaminobenzoic acid (Pa-COOH), in Supplementary Fig. 3). Typically, to prepare TpPa-1, Tp and Pa-1 were dissolved in a solvent mixture (mesitylene/dioxane = 1:1) to form a precursor solution, and acetic acid was added as a catalyst. Then, the solution was transferred into the reactor and reacted at 120°C for 72 hours (Supplementary Fig. 3b). TpPa-2, TpPa-CN, TpPa-NO2 and TpPa-COOH were synthesized under the same conditions (Supplementary Figs. 3c-f). The X-ray diffraction (XRD) pattern confirmed COF’s formation as previously reported (Supplementary Figs. 3g-k).26, 38, 39, 40 It is noteworthy that all COFs have a π-π stacking (AA) structure except TpPa-NO2, which has a staggered (AB) structure. The Brunauer − Emmett − Teller (BET) surface areas of the activated COFs were found to be 59–596 m2/g (Supplementary Figs. 3l-p). The Fourier Transform Infrared (FT-IR) spectra of TpPa-1 indicated total consumption of the starting materials on the basis of the disappearance of the N − H stretching bands of Pa-1 (3100 − 3300 cm− 1) and the carbonyl stretching bands of Tp (1638 cm− 1) (Supplementary Fig. 4a).26 The peak positions represented by the C = C bond and C-N bond in TpPa-1 are shifted compared to the raw materials, which also indicates the successful synthesis of TpPa-1. The FT-IR spectra of several other COFs all have similar shifts (Supplementary Figs. 4b-e).
Preparation of TpPa-1/Cs2PdBr6
160 mg Cs2PdBr6 powder was dissolved in 1 mL mixed solvent (DMF: DMSO = 1:1) by heating. TpPa-1 (30 mg, 48.4% wt) was sonicated in 10 mL tert-butanol for 30 minutes and then stirred for 60 minutes. Subsequently, slowly drop 200 µL of Cs2PdBr6 precursor solution into the TpPa-1 suspension and stir for 30min. The mixture was left to stand for 12h.
Preparation of TpPa-1/Cs2AgBiBr6, TpPa-1/Cs3BiBr9, TpPa-1/Cs2SnI6, TpPa-2/Cs2PdBr6, TpPa-CN/Cs2PdBr6, TpPa-NO2/Cs2PdBr6 and TpPa-COOH/Cs2PdBr6
The method is similar to that for the preparation of TpPa-1/Cs2PdBr6, with only a slight adjustment of the solvent ratio.
Fabrication of sensor
The TpPa-1/Cs2PdBr6 was drop-coated on an Al2O3 substrate printed with interdigitated electrodes (channel width: 200 µm, MJ-10, Beijing Elite Technology Co. Ltd, China) and dried under an infrared drying lamp.
Gas sensing measurements
The sensor was placed in a 1200mL gas chamber, and different concentrations of NO2 gas were introduced into the gas chamber. The sensor current changes were monitored in real time using the Keithley 4200-SCS. The gas flow rate was always stabilized at 100mL/min, and the temperature was stabilized at 300K to reduce the influence of flow rate and temperature on the test. A schematic of the sensing system is presented in Supplementary Fig. 9. The other gas tests are the same as those for NO2.
Resonant microcantilever fabrication
The length, width, and thickness of the cantilever are 200, 100, and 3 µm, respectively, and its effective mass is about 33 ng. The design and fabrication of cantilever have been reported in detail.41 A small amount of COF is ultrasonically dispersed in tert-butanol, and then the suspension is deposited on the cantilever through the sample preparation device.
Resonant microcantilever sensing experiment to gas
Before gas detection, the cantilever is put into a 9.4 mL testing chamber to obtain the baseline signal. When different concentrations of NO2 are introduced, the material loaded at the free end of the cantilever absorbs the NO2, so that the vibration frequency of the cantilever changes. The gas flow rate was always stabilized at 30mL/min, and the temperature was stabilized at 300K to reduce the influence of flow rate and temperature on the test. A schematic of the sensing system is presented in Supplementary Fig. 12. The frequency change was monitored through frequency − time measurements using intelligent physicochemical parameters analyzer (IPPA). The other gas tests are the same as those for NO2.
Measurements and General Methods
The microscopic morphologies of all objects were characterized by SEM (HITACHI Japan S-4700) and TEM (FEI TECNAI G20), respectively. The structural characterization of perovskite and COF was determined by XRD (Bruker D8 Advance) and FTIR (VERTEX70). Keithley 4200-SCS was used to test the sensor performance. Resonant microcantilever gas sensing data were recorded using an intelligent physicochemical parameters analyzer (IPPA). The resonant microcantilever is produced by Xiamen High-End MEMS Technology Co., Ltd. To view a copy of this license, visit http://highend-mems.com/product_center. The SFG system was built by EKSPLA: the visible beam (incident angle 60°, 532 nm) and IR beam (incident angle 55°, around 2700–3800 cm− 1) were about 25 ps at 50 Hz. Since the energy of visible and IR beams was less than 200 mJ, the sample photodamage during the test can be ignored.
Theoretical Calculation
All the density functional theory (DFT) calculations were performed using the Vienna Ab initio Simulation Package (VASP).42, 43, 44 The exchange and correlation potentials were determined with the Perdew, Burke, and Ernzerhof within the generalized gradient approximation (PBE-GGA) functional. 45 The projector augmented-wave (PAW) method was used to describe the electron wave function.46 To accurately describe the van der Waals interaction, the DFT-D3 method with Becke-Jonson damping was used in all the calculations.47, 48 The plane wave energy cut-off was set to 520 eV, and the energy convergence was set to 1 × 10− 5 eV. The lattice supercell (3*3) with a vacuum of 15 Å is composed of the Cs2TeI6 (111) surface. The geometry optimization is performed when the Hellmann–Feynman force on each atom is under 0.02 eV·Å-1. The crystal orbital Hamilton population (COHP) analysis was performed using LOBSTER code.49 The optimized structure and the charge density difference distributions were illustrated with VESTA software.50