Halide Perovskite glues activate two-dimensional covalent organic framework crystallites for selective NO2 sensing

Two-dimensional covalent organic frameworks (2D COFs) are promising for gas sensing owing to the large surface area, abundant active sites, and their semiconducting nature. However, 2D COFs are usually produced in the form of insoluble micro-crystallites. Their poor contacts between grain boundaries severely suppress the conductivity, which are too low for chemresistive gas sensing. Here, we demonstrate that halide perovskites can be employed as electric glues to bond 2D COF crystallites to improve their conductivity by two orders of magnitude, activating them to detect NO2 with high selectivity and sensitivity. Resonant microcantilever, grand canonical Monte Carlo, density functional theory and sum-frequency generation analyses prove that 2D COFs can enrich and transfer electrons to NO2 molecules, leading to increased device conductivity. This work provides a facile approach for improving the conductivity of polycrystalline 2D COF films and may expand their applications in semiconductor devices, such as sensors, resistors, memristors and field-emission transistors.

Therefore, halide perovskites can be a suitable conductive glue for electrically linking COF crystallites.
Herein, we demonstrate that halide perovskites can act as a semiconductive glue to link 2D COF crystallites (Fig. 1a) to activate them for selective NO 2 sensing.High selectivity of the sensor was demonstrated with a response to NO 2 that was 70 times higher than that to 20 other gases of the same concentration, and high sensitivity with the lowest detection limit of 40 ppb was also exhibited.Resonant microcantilever (RMC), grand canonical Monte Carlo (GCMC), density functional theory (DFT), and sum frequency generation (SFG) analyses prove that 2D COFs can enrich and transfer electrons to NO 2 molecules, leading to increased device conductivity.Halide perovskites lower the boundary resistance between the COF crystallites without short-circuiting, in sequence modulating the entire sensor's basis conductance suitable for gas sensing.Our work gives a simple and effective approach to improve the conductivity of 2D COF films and expand the application of 2D COF crystallites in resistance-related applications.

Structure and characterization
The synthesis of halide perovskite Cs 2 PdBr 6 was reported in our previous work [27][28][29] .PdBr 2 and CsBr were dissolved in hydrobromic acid (HBr) at a molar ratio, the oxidant dimethyl sulfoxide (DMSO) was added, and the mixture was heated and stirred to form the double perovskite Cs 2 PdBr 6 with a regular octahedral structure (Supplementary Fig. 1a).High-resolution transmission electron microscopy (HRTEM) and X-ray diffraction (XRD) characterizations demonstrate the synthesis of Cs 2 PdBr 6 (Supplementary Figs.1b and 2) 29 .Five 2D COFs were also synthesized by Schiff base reactions (Supplementary Figs. 3 and 4; see the "Methods" section for details) 30 .In addition, we successfully used perovskite nanospheres to link 2D COF crystallites by an antisolvent growth method.First, TpPa-1 powders are dispersed in tert-butanol by sonification, and then the Cs 2 PdBr 6 solution is added dropwise to the dispersion and fully stirred to form TpPa-1/Cs 2 PdBr 6 (Supplementary Fig. 5).The XRD pattern showed that TpPa-1/Cs 2 PdBr 6 retains the Cs 2 PdBr 6 and TpPa-1 phases after ultrasonication (Supplementary Fig. 1c).The binding between Cs 2 PdBr 6 and TpPa-1 was first investigated by XPS.In Fig. 1b-d, the Pd 4+ peak in TpPa-1/Cs 2 PdBr 6 blue-shift compared to Cs 2 PdBr 6 , while the Pd 2+ peak did not shift.In addition, both the O peak and the N peak are red-shift compared to TpPa-1.These results indicate that TpPa-1 and Cs 2 PdBr 6 form the simultaneous coordination of Pd 4+ with O on the carbonyl group and N on the imine group in TpPa-1 (Fig. 1i).Pd 4+ is formed due to the Br vacancy (V Br ) and Pd-Br antisite (Pd Br ) point defects inherent in perovskites 31 .The Fourier transform infrared (FT-IR) pattern also shows that the C-N bond in TpPa-1/Cs 2 PdBr 6 has a significant shift compared to that in TpPa-1, which also shows that the N in the imine group is coordinated with Pd (Supplementary Fig. 4f).Field emission scanning electron microscopy (FSEM) and energy dispersive X-Ray spectroscopy (EDX) show that the perovskites are distributed on the surface of the COFs in the form of nanospheres so that the COF crystallites can be well connected through the perovskites (Fig. 1e-h).To demonstrate the role of the electric glue, the current-time plots of TpPa-1 and TpPa-1/ Cs 2 PdBr 6 under constant voltage were measured.We fabricated TpPa-1/Cs 2 PdBr 6 films as the semiconducting active layer in sensor devices.The TpPa-1/Cs 2 PdBr 6 film (∼100 μm) was directly drop-coated on an Al 2 O 3 substrate printed with interdigitated electrodes (channel width: 200 μm).With the addition of electric glue, the resistivity of the material decreased from 9.86 × 10 11 to 7.06 × 10 9 Ω m when a constant voltage of 5 V was applied across the interdigitated electrodes (Supplementary Fig. 6).

Nitrogen dioxide sensing
Gas sensors play an indispensable role in various fields, such as food quality assessment, plant growth detection, and noninvasive medical diagnosis [6][7][8] .A chemresistive gas sensor with proper basis conductance is required to detect gas varying from one part per billion (ppb) to one part per million (ppm) among dozens or even hundreds of interfering gases (e.g., >800 in breath or >250 in indoor air) 32,33 .We dropped the TpPa-1/Cs 2 PdBr 6 (51.6 wt%) sample on an interdigital electrode to fabricate chemresistors and tested their performance (Supplementary Figs. 5 and 7).Under a constant bias voltage of 5 V, the sensor responds to the flushing of NO 2 flow with increasing concentration (Fig. 2a).The TpPa-1/Cs 2 PdBr 6 exhibits superiority in chemresistive detection of NO 2 with a detection limit of 40 ppb, which is lower than that of most NO 2 sensors (Supplementary Fig. 8).Even under an atmosphere with different humidity, the sensor can still work smoothly (Supplementary Fig. 9).Multiple tests of the same device have established the sensor's excellent reusability (Supplementary Fig. 10).At a concentration of 10 ppm NO 2 , the response/recovery time was 71/254 s (Supplementary Fig. 11).The voltage-current characteristics remain in the ohmic mode, indicating that the current is mainly determined by the intrinsic conductivity of the film contributed by the thermal release of the carriers 34 .With increasing NO 2 concentration, the intrinsic conductivity of the film gradually increases (Fig. 2b).A high selectivity of the sensor's response to NO 2 is 70 times or more sensitive than that to the other 20 gases (CO, HCl, NH 3 , NO, SO 2 , H 2 , acetone, 3-pentanone, ethyl acetate, butyl acetate, toluene, chlorobenzene, benzaldehyde, anisole, isopropanol, ethanol, n-heptane, nhexane, acetic acid, and formic acid.Fig 2c).Stability is also an important parameter of sensor performance.After 160 days in the atmosphere, the sensor is still able to distinguish between different NO 2 concentrations with a response that is >60 at 2 ppm NO 2 (Fig. 2d).

Universality of electric glue
To demonstrate the general nature of electric glue in the improvement of sensor performance, different COF/perovskite sensors were prepared and tested, including sensors for different Cs 2 PdBr 6 contents, different COFs, and different halide perovskites.The addition of different contents of Cs 2 PdBr 6 enhanced the response and selectivity of 2D COFs to NO 2 .The results showed that, with the decrease in the amount of doping, the response and selectivity to NO 2 showed a trend of first increasing and then decreasing, and the incorporation of 51.6 wt% Cs 2 PdBr 6 could maximize the sensor sensitivity (Fig. 3a).We incorporated 51.6% wt Cs 2 PdBr 6 into five different 2D COFs (TpPa-1, TpPa-2, TpPa-CN, TpPa-NO 2, and TpPa-COOH).All these sensors have the most significant increase in response and selectivity to NO 2 (Fig. 3b).However, sensors based on COFs and perovskites alone have poor sensing performance and cannot meet the actual requirements (Fig. 3a, b and Supplementary Fig. 12).The combination of high mobility and low initial carrier concentration is critical in the development of ultrasensitive sensors for oxidizing gases 35 .We applied this method to other halide double perovskites (Cs 2 AgBiBr 6 , Cs 3 Bi 2 Br 9 , and Cs 2 SnI 6 ; halide double perovskites have better environmental stability than other ABX 3 halide perovskites) and proved good universality for double perovskites (Fig. 3c and Supplementary Fig. 13).XPS result indicates that TpPa-1 and the halide double perovskites form the chelation of metal ions (Ag, Bi, and Sn) by O from the carbonyl group and by N on the imine group in TpPa-1 (Supplementary Fig. 14).

Mechanism
To understand the significant increase in sensor response and selectivity, we conducted an RMC test, which is able to detect the adsorbed mass of gaseous molecules (Supplementary Fig. 15). Figure 4f shows the prepared cantilever where the material is loaded onto its free end.Thereafter, the fabricated cantilever is placed inside the testing chamber for gas-sensing performance evaluation.With TpPa-1/ Cs 2 PdBr 6 deposited on resonant microcantilevers, the adsorbed NO 2 acts as an added mass to shift the cantilever resonant frequency for gravimetric sensing signal readout (Fig. 4b) 36 .The advantage of the RMC test is that the signal is only related to the adsorption of the gas, and the electron transfer between the gas molecules and the sensing material does not affect the test results, which is very important for the analysis of the sensing mechanism.As shown in Fig. 4b, the incorporation of COFs can greatly improve the gas adsorption capacity of the material, which is caused by the porous structure of COFs.For the convenience of comparison, we calculated the weight of gas adsorbed per nanogram of material (Fig. 4a).Figure 4a shows that COFs have a strong gas enrichment effect, which can partly explain the stronger response of TpPa-1/Cs 2 PdBr 6 compared to Cs 2 PdBr 6 .The GCMC simulation proved that the average excess uptake of the NO 2 /N 2 mixture was over 40 pg/ng of TpPa-1 or other COFs (Fig. 5a and Supplementary Figs.16-20).However, this enrichment has no gas selectivity.
To further understand the sensing mechanism, we used DFT and SFG analyses of the adsorption of NO 2 molecules on TpPa-1/ Cs 2 PdBr 6 , Cs 2 PdBr 6 , and TpPa-1 films.As shown in Fig. 5c, the SFG signal intensity of the three types of membranes significantly increases after treatment with NO 2 , indicating that charge transfer occurs after NO 2 adsorption.The TpPa-1/Cs 2 PdBr 6 film showed the greatest enhancement of the background signal, which indicates that NO 2 produces a stronger charge transfer between NO 2 and TpPa-1/Cs 2 PdBr 6 .Therefore, the TpPa-1/ Cs 2 PdBr 6 film has an excellent response to NO 2 .We initially propose 12 adsorption models based on the fact that 2D COFs have layered and porous structures and that NO 2 molecules typically adsorb on the surface and within the pores (Supplementary Fig. 21 and Supplementary Table 1).The DFT calculation results show that NO 2 is physically adsorbed on TpPa-1 (Supplementary Figs.22 and 23).In addition, when the NO 2 molecule is close to TpPa-1/Cs 2 PdBr 6 , it will coordinate with Pd on the perovskites (Supplementary Fig. 24).Therefore, the TpPa-1/ Cs 2 PdBr 6 film presents a distinct characteristic peak of NO 2 after adsorption, which indicates NO 2 has a relatively high orientation uniformity and selectivity at room temperature (Fig. 5c) 37,38 .Our analysis shows that charge transfer between TpPa-1 and NO 2 molecules and the bond of Cs 2 PdBr 6 glue to TpPa-1 crystallites significantly increase the base conductivity of the TpPa-1 film, which gives the sensor excellent sensing performance.

Discussion
We proposed using halide perovskites as electric glues to bond 2D COF crystallites.These electric glues significantly improve 2D COF crystallite conductivity by two orders of magnitude, activating them to detect NO 2 with high selectivity and sensitivity.The combination of high mobility and low initial carrier concentration is critical in the development of ultrasensitive NO 2 sensors.The TpPa-1/Cs 2 PdBr 6 sensor realized high selectivity (the sensor's response to NO 2 is 70 times or more sensitive than that to the other 20 gases) and high sensitivity (the lowest detection limit can reach 40 ppb) in the detection of NO 2 .Overall, this work takes a simple and effective approach and has important implications for improving the conductivity of 2D COFs and advancing their resistance-related applications.

Methods
Synthesis of Cs 2 PdBr 6 0.426 g CsBr (2 mmol, TCI, 99% purity), 0.266 g PdBr 2 (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 1.5x10 -9 2.0x10 -9 2.5x10 -9 3.0x10  cooled to room temperature, the solution containing the black Cs 2 PdBr 6 crystalline was filtered and washed several times with toluene.The product was dried under reduced pressure at 100 °C overnight.

Synthesis of Cs 2 AgBiBr 6
Cs 2 AgBiBr 6 was successfully synthesized according to the reported method 39 .0.426 g CsBr (2 mmol, TCI, 99% purity), 0.449 g BiBr 3 (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 h at 120 °C.The solution was allowed to stand for 2 h after cooling to room temperature to obtain an orange precipitate.Subsequently, the solution containing the orange Cs 2 AgBiBr 6 crystalline was filtered and washed several times with ethanol.The product was dried under reduced pressure at 100 °C overnight.

Synthesis of Cs 3 Bi 2 Br 9
Cs 3 Bi 2 Br 9 was successfully synthesized according to the reported method 40 .0.638 g CsBr (3 mmol), 0.897 g BiBr 3 (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 Cs 3 Bi 2 Br 9 powder was filtered and washed several times with ethanol.The product was dried under reduced pressure at 100 °C overnight.

Synthesis of Cs 2 SnI 6
Cs 2 SnI 6 was successfully synthesized according to the reported method 41 .3.258 g Cs 2 CO 3 (10 mmol, Aladdin, 99% purity) was mixed with 20 mL of 55% HI (Macklin) in a 100 mL beaker to afford a concentrated acidic solution of CsI.3.132 g SnI 4 (5 mmol, Aladdin, 99% purity) was dissolved in 10 mL ethanol to afford a clear orange solution.The SnI 4 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 Cs 2 SnI 6 powder was filtered and washed several times with ethanol.The product was dried under reduced pressure at 100 °C overnight.

Synthesis of TpPa-2, TpPa-CN, TpPa-NO 2 , and TpPa-COOH
The method is the same as the synthesis of TpPa-1, and only the corresponding raw materials need to be replaced.

Gas sensing measurements
The sensor was placed in a 1200 mL gas chamber, and different concentrations of NO 2 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 100 mL/min, and the temperature was stabilized at 300 K to reduce the influence of flow rate and temperature on the test.A schematic of the sensing system is presented in Supplementary Fig. 7.The other gas tests are the same as those for NO 2 .

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 the cantilever have been reported in detail 45 .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 NO 2 are introduced, the material loaded at the free end of the cantilever absorbs the NO 2 , so that the vibration frequency of the cantilever changes.The gas flow rate was always stabilized at 30 mL/min, and the temperature was stabilized at 300 K to reduce the influence of flow rate and temperature on the test.A schematic of the sensing system is presented in Supplementary Fig. 15.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 NO 2 .
Calculation method for deposited material and adsorbed gas mass on a resonant microcantilever The resonant microcantilever can convert the mass increase induced by the target analyte molecules' adsorption into a decrease in the resonant frequency of the microcantilever.The mass of the adsorbed gas is proportional to the frequency-shift signal when the gas mass is much smaller than the effective mass of the resonant microcantilever itself.
Δf ≈0:5Δm × f 0 m eff where f 0 is the initial resonant frequency before mass adsorption, m eff is effective mass of the resonant microcantilever itself.The length, width, and thickness of the cantilever are 200, 100, and 3 μm, respectively.Thus, the effective mass can be calculated as about 33 ng 45,46 .

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 <200 mJ, the sample photodamage during the test can be ignored.

Theoretical calculation
All the density functional theory calculations were performed using the Vienna Ab initio Simulation Package (VASP) [47][48][49] .The exchange and correlation potentials were determined with the Perdew, Burke, and Ernzerhof within the generalized gradient approximation (PBE-GGA) functional 50 .The projector augmentedwave (PAW) method was used to describe the electron wave function 51 .To accurately describe the van der Waals interaction, the DFT-D3 method with Becke-Jonson damping was used in all the calculations 52,53 .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 Cs 2 TeI 6 (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 54 .The optimized structure and the charge density difference distributions were illustrated with VESTA software 55 .

Fig. 1 |
Fig. 1 | Structure and characterization of electric glues bond 2D COF crystallites.a Schematic diagram of the gas sensor and perovskite glue to bond COF crystallites.b XPS spectra of the Pd 3d region of Cs 2 PdBr 6 and TpPa-1/Cs 2 PdBr 6 ; c N 1s and d O 1s region of the TpPa-1 and TpPa-1/Cs 2 PdBr 6 .e, f FSEM images and g, h EDX analysis images of TpPa-1/Cs 2 PdBr 6 .i Theoretical models of the combination of Pd exposed by surface defects of perovskites and N and O in TpPa-1.The yellow, pink, red, green, violet, brown, and gray balls refer to C, H, O, N, Br, Pd, and Cs atoms, respectively.Source data are provided as a Source Data file.

Fig. 2 |
Fig. 2 | Characterization of sensing performance.a Responses variation versus NO 2 concentrations (40 ppb to 10 ppm).b Current-voltage characteristics for sensor when exposed to different NO 2 concentrations (0-10 ppm).c Gas selectivity of the TpPa-1/Cs 2 PdBr 6 sensor.(Air is the carrier gas, and all gas concentrations are 2 ppm.) d The long-term stability of the TpPa-1/Cs 2 PdBr 6 sensor toward different concentrations of NO 2 .Source data are provided as a Source Data file.

Fig. 3 |
Fig. 3 | Comparison of sensor gas response and selectivity performance (all gas concentrations are 2 ppm).a The response of TpPa-1 to different gases after mixing different proportions of Cs 2 PdBr 6 .b The response of different COFs to different gases after mixing 51.6% wt Cs 2 PdBr 6 .c The response of TpPa-1 to different gases after mixing different perovskites.Source data are provided as a Source Data file.

Fig. 4 |
Fig. 4 | Structure and test performance of the resonant microcantilever.a The mass of gas adsorbed by 1 ng of material deposited on the microcantilever.b Frequency shift of the microcantilever Cs 2 PdBr 6 , TpPa-1/Cs 2 PdBr 6 , and TpPa-1 sensors to various types of gases (all at concentrations of 2 ppm).FSEM images of c Cs 2 PdBr 6 , d TpPa-1/Cs 2 PdBr 6 , and e TpPa-1 deposited on the microcantilever.f FSEM image of the structure of the microcantilever.Source data are provided as a Source Data file.

Fig. 5 |
Fig. 5 | Characterization of the sensing mechanism.a Grand canonical Monte Carlo simulation of NO 2 adsorption density plot on TpPa-1.The insets in a show the GCMC simulation of the average excess uptake of the NO 2 /N 2 mixture per ng of TpPa-1.b Schematic diagram of the function of electric glue.c SFG spectra of the film before and after NO 2 adsorption.Source data are provided as a Source Data file.