High‐Performance Optoelectronic Gas Sensing Based on All‐Inorganic Mixed‐Halide Perovskite Nanocrystals with Halide Engineering

Gas sensors are of great interest to portable and miniaturized sensing technologies with applications ranging from air quality monitoring to explosive detection and medical diagnostics, but the existing chemiresistive NO2 sensors still suffer from issues such as poor sensitivity, high operating temperature, and slow recovery. Herein, a high‐performance NO2 sensors based on all‐inorganic perovskite nanocrystals (PNCs) is reported, achieving room temperature operation with ultra‐fast response and recovery time. After tailoring the halide composition, superior sensitivity of ≈67 at 8 ppm NO2 is obtained in CsPbI2Br PNC sensors with a detection level down to 2 ppb, which outperforms other nanomaterial‐based NO2 sensors. Furthermore, the remarkable optoelectronic properties of such PNCs enable dual‐mode operation, i.e., chemiresistive and chemioptical sensing, presenting a new and versatile platform for advancing high‐performance, point‐of‐care NO2 detection technologies.


Introduction
[3][4][5] To further enhance their optical and electronic properties, perovskite nanocrystals (PNCs) have been integrated into devices using various techniques, such as facile manipulation of ionic exchange, [6] surface-defect modification, [7] surface functionalization, [8] and ligand engineering, [9] which provides a versatile platform for advanced applications.Moreover, the solution-based fabrication of PNCs with nano-scale crystallinity provides a significant advantage by offering high binding sites, which is the prerequisite for realizing the high performance of miniaturized gas sensing devices with improved sensitivity and a low detection limit for monitoring atmospheric gases at room temperature. [10,11]n the current era with increasing human health awareness, NO 2 sensors play a vital role in monitoring air pollution levels, providing an early-warning system, ensuring compliance with regulations, and improving air quality and public health in a cost-effective manner.Despite significant progress in the development of chemiresistive sensors using metal oxides, conducting polymers, and 2D materials, NO 2 sensors still suffer from poor sensitivity and reliability at room temperature, particularly in detecting low ppb levels of the gas.Perovskite nanocrystals, as promising sensing materials, have also been considered for NO 2 detection because of their high specific surface area and unique chemical interaction phenomenon such as ion exchange, [12,13] trap passivation, [14,15] and doping effects. [16]However, the low conductivity of PNCs thin films is an obstacle that hinders their performance for chemiresistive sensing.Casanova-Chafer et al. have dispersed PNCs on graphene to improve the charge transport for NO 2 sensing, [17] but the incorporation of highly conductive channel mediums may reduce the signal-to-noise ratio, thus restricting the detection limit.Our recent work showed that the p-type charge transport properties of CsPbBr 3 PNCs NO 2 sensor can be significantly strengthened through Ostwald ripening, however, the moderate sensitivity still needs to be improved for advanced sensing applications. [18]Furthermore, while numerous works have been done to investigate perovskite-based gas sensing, the effect of halide composition on the sensing performance is still unclear.It should be noted that composition engineering, especially the tuning of halide anions, has been extensively explored to tune the optoelectronic properties of halide perovskites. [19]Additionally, although halide perovskites are an extraordinary class of optoelectronic materials, there have been limited studies on PNC-based gas sensors that leverage their optical properties, such as photoluminescence (PL) or fluorescence (FL) change [20,21] and phase transitions, [22,23] for transduction.Therefore, fundamental researches are imperative to further develop PNCs-based sensing technology with improved performance for real-time monitoring across different types of NO 2 sensors.
In this work, we systematically investigated the performance of mono-halide and mixed-halide CsPbI x Br 3-x (x = 0, 1, 1.5, 2, and 3) PNCs for NO 2 sensing, which revealed a strong compositionperformance correlation.Leveraging on the halide tuning strategy, CsPbI 2 Br PNCs are identified as an ideal candidate for chemiresisitve sensing of NO 2 with enhanced sensitivity and stability.Furthermore, two gas sensing modes, chemiresistive and chemioptical sensing, are realized using CsPbI 2 Br PNCs, which exhibit exceptionally high sensitivity toward NO 2 compared favorably to other reported NO 2 sensors.Additionally, CsPbI 2 Br PNCs demonstrate rapid and efficient NO 2 detection, with the lowest detection limit of 2 ppb, which meets the minimum requirement for outdoor concentration of NO 2 (40 ppb) specified by the European Environmental Agency.The facile and versatile sensing capabilities make such mixed-halide PNCs a promising new nanomaterial platform for high-performance and versatile NO 2 detection in room-temperature environments.

Results and Discussion
In this study, the PNCs were synthesized through the widely used hot-injection method followed by a purification process, in which we modified the ratio of PbX 2 (X = Br or I) precursors to prepare CsPbI x Br 3-x PNCs with different halide compositions (see Experimental Section for details).The synthesized inorganic PNCs have a cubic ABX 3 lattice structure, in which Cs + occupies the A site, Pb 2+ B site, and halide ions X site, and their transmission electron microscopy (TEM) morphology is shown in Figure 1a.By carefully adjusting the halide compositions, the luminescence of resulting mixed-halide PNCs can cover the entire visible spectrum, as shown in the fluorescence images and the PL spectra in Figure 1a,b.The exciton peak wavelength of CsPbI x Br 3-x PNCs is red-shifted as the I/Br ratios increase due to the enhanced energy of the X sites np 6 orbitals (n = 4 and 5 for Br and I, respectively) that move the valence band maximum upward and shrink the electronic bandgap. [24]Further optical properties of CsPbI x Br 3-x nanocrystals were analyzed via ultravioletvisible (UV-vis) absorption spectrum (Figure 1c), which demonstrates an excitonic energy modulation from 525 to 675 nm as the I/Br ratios increase in PNCs.The corresponding Tauc optical bandgap of CsPbI x Br 3-x PNCs are determined as 2.33, 2.22, 2.12, 1.92, and 1.81 eV for samples with x = 0, 1, 1.5, 2, and 3 (Figure S1, Supporting Information), respectively, showing good agreement with reported values. [25]The X-ray diffraction (XRD) analysis was employed to confirm the crystal structure of assynthesized CsPbI x Br 3-x nanocrystals, in which all PNCs reveal perovskite phases with dominant (100) and (110) orientations, as illustrated in Figure 1d, which is in good agreement with other previous works. [26,27]Noteworthily, when the iodine composition increases, the corresponding diffraction peaks gradually shift to smaller angles, which can be attributed to the lattice expansion as the ionic radius of iodine cation (2.20 Å) is larger than that of bromine cation (1.96 Å). [28] Schematic illustration in Figure 2a demonstrates the fabrication process of CsPbI x Br 3-x PNCs NO 2 sensors, which were prepared by layer-by-layer spin coating method using a hexanesuspended solution of PNCs deposited onto glass substrates with interdigitated Pt electrode area (0.8 × 0.7 cm). Figure 2b displays an optical image of the as-prepared PNCs NO 2 sensor, showing an electrode spacing of 5 μm.Cross-sectional scanning electron microscope (SEM) measurement was undertaken to identify the thickness of the PNCs sensing layer, as shown in Figure 2c, indicating a nano-scale thickness of PNCs film of approximately 80 nm.
Insulating ligands such as oleic acid (OA) and oleylamine (OAm) are indispensable for synthesizing high-quality PNCs with uniform size distribution and high photoluminescence yield (PLQY), but they would impede the carrier transport in optoelectronic devices. [29]To enhance electronic coupling and charge transport, we exploited a post-treatment approach using a mild polar solvent, methyl acetate (MeOAc), to partially remove the long-chain ligands and reduce interparticle spacing. [30]It is further elucidated by I-V characteristic measurements on three different devices that were subjected to different numbers of MeOAc post-treatments (0, 1, and 3 times).Our results demonstrate that the conductivity of the PNCs device improves as the number of post-treatments increases, as shown in Figure S2 (Supporting Information).To further evaluate the efficacy of the surface ligand removal, Fourier-transform infrared spectroscopy (FTIR) measurement was carried out.As depicted in Figure 2d, the signal strength of N-H at 1631 cm −1 , as well as the C-H stretches at 2851 and 2921 cm −1 , significantly weakened, indicating a substantial reduction of both OA and OAm surface ligands following the post-treatment.After ligand engineering, the topography of the resulting PNCs sensors was analyzed by atomic force microscope (AFM), confirming the uniform distribution of PNCs films with low root-mean-square (RMS) roughness of ≈5 nm (Figures S3 and S4, Supporting Information).It is worth noting that all CsPbI x Br 3-x PNCs, irrespective of their halide composition, exhibited consistent nanocrystal sizes and structural parameters, which exclude the potential impact of surface morphology on sensing performance.
Since the effect of halide composition on gas sensing performance has remained unexplored so far, we first fabricated different CsPbI x Br 3-x (x = 0, 1, 1.5, 2, and 3) PNCs devices to analyze their chemiresistive characteristics for NO 2 gas detection.Before evaluating the performance of PNCs NO 2 sensors, it is essential to verify the contact behavior between the PNCs and Pt electrode since the perovskite/electrode interface plays a critical role in the device characteristics. [31]The I-V curves of all the CsPbI x Br 3-x PNCs sensors in Figure 3a demonstrate a near-linear current-voltage response, indicating that the PNCs generate a low barrier contact with the Pt electrode.32] In this case, the density of charge carriers in PNCs layer predominantly influences the resistance change, providing an ideal platform for chemiresistive gas sensing.Moreover, it was observed that the base current of CsPbI x Br 3-x PNCs gradually increases as the iodide ratio increases, which might be attributed to the higher carrier mobility in iodide-based perovskite.According to previous studies, the mobility of CsPbI 3 is much higher than that of CsPbBr 3 due to inherently less effective mass and lower scattering rate. [33]O 2 sensing measurement for all-inorganic CsPbI x Br 3-x PNCs was conducted at room temperature with a bias of 1 V in the dark.The sensing response is defined as S = (I gas -I air /I air ), and the response curves for all the CsPbI x Br 3-x PNCs sensors with different halide compositions are plotted in Figure 3b.For all sensors, the resistance dramatically decreased when 8 ppm of NO 2 gas was introduced, which implies that all PNCs exhibit p-type semiconductor behavior due to the strong oxidizing nature of NO 2 . [34]The sensing response enhances remarkably from 0.098 to 67 with increasing iodide ratio from 0 to 2, but decreases to 57 as the iodide ratio further increases to 3 becoming CsPbI 3 PNCs.Compared to the up-to-date perovskite and other materials-based NO 2 sensors in the literature, [16,18,[34][35][36][37][38][39][40][41][42][43][44][45][46] our sensors based on CsPbI 3 and CsPbI 2 Br PNCs demonstrate significantly higher sensing performance in terms of sensitivity and response time (Figure 3c).The CsPbI 2 Br PNCs sensor records the highest response of 67 at 8 ppm NO 2 , while the sensing response of CsPbI 3 PNCs sensor reaches 57., 48] It has been widely reported that CsPbI 3 is unstable when they contact with ionic mediums, which may lead to lattice deformation or phase transition due to the change of surface energy and strain-associated effects in the lattice. [49,50]his may also explain the unusual absorption and desorption responses of CsPbI 3 in Figure 3b, which features two consecutive transitions when the NO 2 gas was turned on and off, respectively.The instability of CsPbI 3 lattice is further corroborated by the repeatability test and the emergence of XRD peak associated with decomposed PbI 2 after NO 2 exposure (Figures S5 and S6, Supporting Information), suggesting that CsPbI 3 PNCs are not suitable for a reliable NO 2 sensing.
The NO 2 concentration dependence of sensing performance was also studied for all CsPbI x Br 3-x PNCs sensors in the low concentration range from 0.1 to 1 ppm.As shown in Figure 3d, the sensing responses of all CsPbI x Br 3-x PNCs exhibit linear relationships with increasing NO 2 concentration, which is beneficial for precise NO 2 detection.Figure 3e demonstrates the response and recovery times of all PNCs sensors, which decrease as the iodide ratio increases up to 2 in CsPbI x Br 3-x , but increases when the io-dide ratio reaches 3.This finding suggests that a higher iodide ratio generally facilitates faster adsorption and desorption kinetics upon NO 2 exposure when disregarding the labile ionic lattice of CsPbI 3 .
Figure 3f shows a radar chart that summarizes the key sensing parameters for all CsPbI x Br 3-x PNCs, including sensitivity, recovery time (t rec ), limit of detection (LOD), cycling stability, and equilibrium constant (K A ), leading to the conclusion that the CsPbI 2 Br sensor has a greater promise for serving as a reliable NO 2 sensor with high sensing output among all halideengineered PNCs (detailed calculation of sensing parameters can be found in Supporting Discussion and Table S1, Supporting Information).
Seeking to optimize the sensing performance of CsPbI 2 Br PNCs sensor, we investigated the effect of MeOAc post treatment on film morphology and carrier transport.As discussed above, the MeOAc treatment plays an important role in enhancing the conductivity of the sensing layer due to reduced insulting ligands and better coupling between adjacent PNCs.More importantly, the ligand removal process exposes surface defects such as halide vacancies and increases the number of active sites in the PNCs layer, thereby improving the interaction between PNCs and gases and enabling higher transduction ratio of chemical signals. [40,51,52]As a proof of concept, we observed a significant improvement in sensing response with 1-cycle of MeOAc treatment, which results in a 27-fold increase compared to the  [16,18,[34][35][36][37][38][39][40][41][42][43][44][45][46]  control film without the MeOAc treatment (Figure S7a, Supporting Information).However, excessive MeOAc treatment can alter the surface morphology due to the substantial loss of ligands (Figure S7b, Supporting Information), leading to a slight reduction in sensing response and cycling stability.Therefore, we selected 1-cycle of MeOAc treatment for our sensors to optimize the sensing properties.
Figure 4a shows that the resulting sensing current of optimized CsPbI 2 Br PNCs sensor can reach up to −32 nA at a bias voltage of 1 V upon 8 ppm of NO 2 , while the baseline current is ≈0.45 nA.The CsPbI 2 Br PNCs sensor features reversible sensing and the resistance returns to the initial baseline level in the absence of NO 2 gas.The response and recovery times of CsPbI 2 Br PNCs sensor are determined as 13 and 26 s, respectively (Figure 4b), indicating that it possesses fast and efficient adsorption and desorption properties.The efficient sensing output toward NO 2 observed in such halide-engineered PNCs at room temperature compared favorably to other recent works (Table S2, Supporting Information).
Figure 4c demonstrates that the response of the CsPbI 2 Br PNCs sensor increases linearly with increasing NO 2 concentration at a low concentration range between 0.1 and 1 ppm.Based on the "signal-to-noise ratio" method, [46] an ultra-low LOD of ∼2 ppb is estimated for the CsPbI 2 Br PNCs sensor.The obtained LOD meets the compliance of the European Environmental Agency for NO 2 detection (<40 ppb) and is significantly lower than most of reported NO 2 sensors based on other materials (Table S2, Supporting Information), showing great potential of such halide-engineered PNCs to serve in early-warning safety systems.
To further investigate the gas selectivity of CsPbI 2 Br sensor, we introduced other analytes, such as volatile organic compounds (propane, acetone, EtOH, and EtBz) and O 2 , to the sensor under the same experimental conditions, as shown in Figure 4d.An outstanding selectivity of CsPbI 2 Br toward NO 2 is confirmed as the sensing responses for other target gases are <0.1, far below the ultrahigh response of NO 2 (≈67).The excellent selectivity can be attributed to the highest electron affinity of NO 2 (≈2.27 eV) compared to other volatile organic compounds and oxygen. [53,54]he operational reversibility of CsPbI 2 Br PNCs sensor was tested with sequential on-off gas exposures, as shown in Figure S8 (Supporting Information), and a good reversibility was obtained showing 90% of the initial response after fifteen cycles of NO 2 exposures.
Encouraged by the exceptional chemiresistive sensing potential and superior optical properties, we also explored the optical sensing capability of CsPbI 2 Br PNCs for detecting NO 2 .As shown in Figure 5a, we measured the steady-state PL emission of CsPbI 2 Br PNCs device with and without 8 ppm of NO 2 exposure.An obvious PL intensity quenching of ≈85% in the CsPbI 2 Br PNCs film was recorded under NO 2 exposure, which was also visible as significant fluorescence quenching under UV excitation  (365 nm) to the naked eyes.These results imply that the adsorption of NO 2 on the PNCs surface can substantially alter the carrier distribution and radiative carrier recombination in the nanocrystals.The repeatability of the optical sensing response is further evaluated by continuous on-off cycles at 8 ppm NO 2 using three different batches of sensors, confirming its good reversibility in the chemioptical sensing mode (Figure 5b).It is worth noting that when the CsPbI 2 Br PNCs sensor was tested with a common reducing gas (H 2 S), the PL intensity can retain its original level under the same experimental conditions (Figure S9, Supporting Information), suggesting that the quenching of PL is primarily driven by electron transfer arisen from NO 2 as a strong electronwithdrawing gas.
The PL decay as a function of time upon NO 2 exposure is shown in Figure 5c.A rapid PL quenching was observed within the first 2 min of NO 2 exposure, followed by a slower rate of decay over the subsequent hour.Upon the initial exposure to NO 2 , a high number of binding sites on the PNCs surface facilitated quick adsorption of NO 2 molecules, causing fast quenching of PL.As the exposure time increased, the available reaction sites decreased until reached a saturation point where further PL quenching was not observed.On the other hand, our experiment revealed that the PL intensity can abruptly recover to its initial level within ≈5 s after NO 2 is purged, indicating a rapid release of captured electrons.Optical responses of PNC sensors are calculated by PL quenching ratio (P 0 -P)/P 0 , where P is the quenched PL intensity under the NO 2 gas and P 0 is the initial intensity without NO 2 exposure.The optical response of CsPbI 2 Br PNCs sensor is determined as 0.85, which is much higher than other materials based NO 2 optical sensors reported in the literature (Table S3, Supporting Information).To estimate the LOD of the sensor in the chemioptical mode, the CsPbI 2 Br PNCs sensor was tested under different NO 2 concentrations ranging from 1 to 8 ppm, as shown in Figure 5d.One can see that the optical response features a linear relationship with NO 2 concentration along with a high coefficient of determination (R 2 ) of 0.982.According to the signal-to-noise ratio and the linear slope, [20,55] the LOD is estimated to be as low as 1.34 ppm in the chemioptical sensing mode.
Regarding the PNCs sensing mechanism, chemiresistive mode is operated by the changes in resistance due to the electron transfer between the sensing layer and gas molecules.NO 2 , as a strong electron-withdrawing gas, tends to adsorb onto the PNCs surface and attracts electrons from the PNCs, which can be expressed by the equation [46] : Since the electrons are captured by NO 2 , this process would result in a higher concentration of holes in p-type PNCs, which substantially decreases the resistance of PNCs sensors.In addition to the one-step process described above, in the presence of other gases, the adsorption-charge transfer between NO 2 molecules and the perovskite surface can also follow a two-step mechanism.Specifically, when using dry air as the carrier gas, the oxygen molecule in the air (≈21%) can be adsorbed on the PNCs surface to capture electrons.Due to the large difference in electron affinity (O 2 : ≈1.29 eV, NO 2 : ≈2.27 eV), the adsorbed oxygen ions can help transferring the electrons to NO 2 to form the adsorbed NO 2 − and NO 3 − , which has been reported previously. [53,54,56,57]he two-step electron transfer can be described by the following equation and schematically illustrated in Figure 6a.
The two-step process is further supported by sensing tests using two different carrier gases, i.e., dry air and N 2 , to transport NO 2 molecules.The CsPbI 2 Br PNCs sensor exhibits a NO 2 response of 67 with dry air, which is higher than the response of 51 when using N 2 as the carrier gas (Figure S10, Supporting Information).Because the CsPbI 2 Br PNCs sensor generates only a minimal response of 0.07 when exposed to the pure oxygen alone (Figure S11, Supporting Information), the significantly enhanced sensing response in air-carried NO 2 should be attributed to oxygen-assisted two-step electron transfer that facilitates the NO 2 sensing.
In the chemioptical sensing mode, the sensing response is governed by the PL intensity, which relies on the recombination of electron-hole pairs in the PNCs.Upon NO 2 exposure, the electrons are captured by the gas molecules adsorbed on the PNC surface, resulting in a substantial reduction in carrier recombination and a consequent quenching of the PL emission.The adsorption-induced charge transfer property is further elucidated using time-resolved PL (TRPL) measurement, as shown in Figure 6b, where the carrier lifetime of PNCs is reduced from 2.8 ns in air to 0.77 ns with 8 ppm of NO 2 .[60] This result is consistent with the mechanism of PL modulation underlying the optical sensing of NO 2 by PNCs.
In this study, the varied chemoreceptive sensing results from different CsPbI x Br 3-x PNCs suggest that halide compositions play a significant role in NO 2 sensing performance.To gain a deeper insight into the charge transfer mechanism in different CsPbI x Br 3-x PNCs sensors, PL quenching ratio and adsorption/desorption kinetics of all PNCs are further studied.As shown in Figure 6c, the PL quenching effect becomes more pronounced with increasing I/Br ratio, indicating that NO 2 captures more electrons from I-rich PNCs than Br-rich counterparts under the same experimental conditions.The adsorption and desorption kinetics of different CsPbI x Br 3-x PNCs sensors are analysed to quantify their binding affinities with NO 2 .The adsorption/desorption rate constants, denoted by k ads and k des respectively, are estimated through fitting to Equations (5) and Equation (6) [38] : where R 0 represents the baseline response in the absence of NO 2 , R max corresponds to the maximum response toward NO 2 , C a is the NO 2 concentration, which is 8 ppm in this work.Figure 6d demonstrates the calculated k ads and k des for different CsPbI x Br 3-x PNCs sensors.One can see that CsPbI 2 Br exhibits the highest NO 2 adsorption tendency, while CsPbI 1.5 Br 1.5 features the fastest desorption rate to release the NO 2 molecules during the recovery process.The equilibrium constant (K A ) indicates the affinity of the target gas to the sensing layer and can be estimated based on the adsorption and desorption kinetic constants, as described by the equation: The CsPbI 2 Br PNCs sensor yields the largest K A value, indicating its superior binding affinity toward NO 2 .
In the operation of PNCs gas sensors, two main factors significantly affect the response: defects such as halide vacancies to generate active sites and strong interaction of perovskite with NO 2 gas molecules.The low formation energy of halide vacancies has been extensively reported, which causes the creation of strains in the perovskite crystals and defect sites at the surface, such as uncoordinated Pb 2+ ions. [61]The different halide ratios of CsPbI x Br 3-x in this study can lead to different formation energies in the lattice structure, which in turn results in different levels of halide vacancies.Due to the different radii between bromide and iodide ions, more intrinsic surface defects caused by halide vacancies tend to be triggered in mixed-halide PNCs, leading to lower PLQY (70-80%) than that in single-halide composition (90 and 95% for CsPbBr 3 and CsPbI 3 , respectively). [62,63]Additionally, significant halide vacancies and associated uncoordinated Pb 2+ cations can be induced after the ligand removal with MeOAc due to the rapid desorption and highly dynamic ligand binding feature. [61]Hence, appropriate MeOAc treatment not only generates more binding sites for gas molecules to be adsorbed but increases the permeability of the target gas.It is confirmed by our sensing result (Figure S7a, Supporting Information) that reducing insulating long-chain ligands bound to perovskite nanocrystals and promoting active surface defects contribute to a 27-fold higher sensing response.Previous studies proposed that target gas molecules can act as vacancy fillers in perovskite and reversibly fill the intrinsic bromide vacancies. [40,51]Lu et al. recently reported Br-based lead-free perovskite hollow nanospheres for CO detecting sensors, suggesting that the vacancies of perovskite hollow nanospheres favor the sensing of CO molecules. [52]In case of MAPbI 3 -type perovskite, it is reported that iodine anion vacancies can generate uncoordinated active sites at the surface that enhanced the gas sensing process. [14]e confirmed that more active surface sites are likely to be induced by halide vacancies for NO 2 molecules to be adsorbed and it is also important to look at the capability of perovskite to interact with NO 2 molecules.Central metal ion, which is Pb 2+ in all-inorganic halide perovskite, is the main active site for interacting with the gas molecules. [64]In one representative study, it was confirmed that more Pb cations existing on the surface can enhance the adsorption of NO 2 molecules and their interaction with the sensing layer. [36]Therefore, the oxidation-reduction www.small-methods.comproperties of Pb can make a significant impact on the CsPbI 3 -NO 2 and CsPbBr 3 -NO 2 complexes.Since Br has higher electron affinity than I (Br: ≈3.36 eV and I: ≈3.06 eV), the electronwithdrawing ability of Br is stronger than that of I when forming the Pb-X skeleton. [65]Therefore, Pb in the Pb-Br system becomes more oxidative than in the Pb-I skeleton, indicating that NO 2 has a higher adsorption ability toward Pb-I frameworks.The sensing responses of CsPbCl x Br 3-x PNCs were also tested (Figure S12, Supporting Information), showing significantly lower performances than that of CsPbI x Br 3-x PNCs.This result can be attributed to the higher electron affinity of Cl (≈3.7 eV) than that of Br and I, which further underlines the importance of the electron-withdrawing ability of halide ions that affect the oxidation property of Pb-X frameworks.Considering the higher density of surface active sites in mixed-halide perovskites and the oxidation property of Pb, the mixed-halide compositions with higher I/Br ratios, such as CsPbI 2 Br in our study, can present better sensing performance toward NO 2 .With the record high NO 2 sensing response of 8.3 per ppm observed in this work, the CsPbI 2 Br PNCs hold great promise for the development of highperformance NO 2 sensors in early-warning and real-time monitoring systems.

Conclusion
In summary, the gas sensing performances of CsPbI x Br 3-x PNCs with different halide compositions were systematically investigated and optimized for NO 2 detection.The excess long-chain ligands on PNCs surface were removed by an optimal MeOAc posttreatment approach to improve the charge carrier transport in the PNCs layer.After conducting a comprehensive analysis of various factors, including sensing performance, response time, detection limit, stability, adsorption/desorption kinetics, and equilibrium constant, the CsPbI 2 Br PNCs are determined as the most suitable candidate for NO 2 sensing when compared to other mixed halide perovskites.The CsPbI 2 Br PNCs chemiresistive sensor recorded the highest sensing response of 67 at 8 ppm of NO 2 with a short response and recovery times of 13 and 26 s, respectively, and an ultra-low LOD of ≈2 ppb.Additionally, the chemioptical sensing capability of CsPbI 2 Br PNCs was also explored, revealing a high optical response of 0.85 and a low LOD of 1.34 ppm.Overall, the remarkable optoelectronic properties and low detection limit of CsPbI 2 Br PNCs, combined with their dual-mode sensing capabilities, suggest a new and promising platform for developing next-generation high-performance NO 2 detection systems.
Synthesis of All-Inorganic Mixed Halide Perovskite CsPbI x Br 3-x (x = 0, 1, 1.5, 2, and 3) PNCs and Purification Process: CsPbI x Br 3-x PNCs were synthesized by a modified hot-injection approach.0.2 g of cesium carbonate were mixed with 10 mL of ODE and 0.75 mL of OA and heated at 120 °C for 1 h in vacuum.The temperature was then set up to 140 °C to heat the mixtures until there is no residuals in the solution.Prepared Cs precursor was kept at 100 °C before injecting into Pb precursor.0.94 mmol of PbX 2 (X = Br and I) was mixed with 2.5 mL of OA and 25 mL of ODE in 100 mL three-neck flask and degasified in vacuum at 100 °C for 1 h.Then, 2.5 mL of pre-heated OAm was added and heated at 120 °C in vacuum for 1 h, followed by heating under N 2 until the mixture yielded a clear solution.The temperature increased up to 160 °C and the previously prepared 2 mL of Cs precursor was swiftly injected to the solution under N 2 purging.The reaction was quenched in an ice bath for 20 s to obtain the PNCs.The purification process was slightly modified from the method used in our previous work. [18]As-synthesized PNCs are dispersed in MeOAc solvent in a ratio of 1:3 and centrifuged at 8000 rpm for 5 min to get rid of the organic impurities.Supernatant was discarded and PNCs pallet was again suspended in hexane/MeOAc (10 mL) to centrifuge at 7500 rpm for 5 min.The final PNCs pallet was dispersed in hexane (≈10 mL) and kept in a fridge under dark overnight and centrifuge at 4000 rpm to remove excessive residual Pb precipitates.
NO 2 Sensor Fabrication and Testing: To fabricate chemiresistive NO 2 sensor, commercial substrates comprise glass with interdigitated Pt lines and a total electrode area of 7 mm × 5 mm with 5 μm gap of the electrodes (G-IDEAU5, DropSens, Ovieo, Spain) were employed.The substrates were then rinsed with acetone, ethanol, and deionized water sequentially followed by plasma cleaning for 5 min.For sensing layer deposition, layerby-layer spin-coating method was used.CsPbI x Br 3-x PNCs suspended in hexane with the concentration of 70 mg mL −1 was spin-coated on the substrates at 2500 rpm for 30 s and this process was repeated 3 times to get a high quality and uniform film without pinholes with a ≈80 nm thickness.The ligand removal process was sequentially done with MeOAc solution dropped and soaked for 20 s and residual MeOAc solution was removed by spinning at 2500 rpm for 20 s.
The sensor measurements were performed as follows.NO 2 (10 ppm in N 2 (Coregas)), ethanol (10 ppm in N 2 (Coregas)), acetone (10 ppm in N 2 (Coregas)), propane (10 ppm in N 2 (Coregas)), and ethyl benzene (10 ppm in N 2 (Coregas)) diluted in simulated air (0.1 L min −1 O 2 + 0.4 L min −1 N 2 (BOC Ltd)) were controlled using a mass flow controller (Bronkhorst) while the total flow rate of the gas was kept at 0.5 L min −1 .The sample was loaded on a stage within a gas chamber (Linkam) and the temperature of the stage in the gas chamber (Linkam) was kept constant at 25 °C using a temperature controller to simulate room temperature conditions).For the gas sensing measurements, two Au probes were separately placed on top of the two arms of PNCs thin film coated Pt interdigitated electrode (DropSens) with an applied constant potential of 1 V.The dynamic gas sensing response of CsPbI x Br 3-x PNCs devices were recorded by a responsive electrochemical workstation (CHI 660E, USA; Biologic VMP3 electrochemical workstation).The devices were taken out for the gas sensing measurements normally for ≈1-2 h each time, and then stored in a dry desiccator for the rest of the time.
Characterization and Measurements: XRD was carried out using a Bruker D8 ADVANCE diffractometer with Cu Ka (lambda = 0.15406 nm) radiation.TEM characterization was performed using a FEI Tecnai G2 20 microscope.The surface morphology of PNCs film was examined by SEM (FEI Nova Nano 630) and AFM (Bruker Dimension ICON), respectively.The absorbance spectrum was measured using an Evolution 600 ultraviolet-visible spectrophotometer (Thermo Scientific).PL measurement was conducted at room temperature using LabRAM ARAMIS Raman microscope (Horiba Scientific) with a 325 nm laser source as excitation.

Figure 1 .
Figure 1.Optical and structural properties of CsPbI x Br 3-x (x = 0, 1, 1.5, 2, and 3) PNCs.a) Photographs of the colloidal solutions of CsPbI x Br 3-x PNCs under UV light (365 nm), along with an illustration of chemical structure and TEM image of PNCs.b) PL spectra and c) UV-vis absorption spectra of CsPbI x Br 3-x PNCs.d) XRD patterns of as-prepared CsPbI x Br 3-x PNCs.

Figure 2 .
Figure 2. a) Schematic of the preparation processes of CsPbI x Br 3-x PNCs sensors with the MeOAc post-treatment.b) Optical image of the deposited CsPbI x Br 3-x PNCs sensing layer on interdigitated Pt-electrodes on a glass substrate.Inset shows a photograph of the resulting PNCs sensor.c) Crosssectional SEM image of PNCs layer showing the thickness of ≈80 nm with one cycle of MeOAc post-treatment.d) FTIR spectra of the as-prepared CsPbI x Br 3-x PNCs sensors before and after the MeOAc post-treatment.

Figure 3 .
Figure 3. a) I-V characteristics of CsPbI x Br 3-x PNCs deposited on interdigitated Pt-electrodes.b) Dynamic sensing responses of CsPbI x Br 3-x PNCs sensors toward 8 ppm of NO 2 under a bias of 1 V in dark.c) Up-to-date progress in NO 2 sensor performance operating at room-temperature in terms of sensor response and response time in the literature.[16,18,[34][35][36][37][38][39][40][41][42][43][44][45][46]d) NO 2 concentration dependent sensor response devices, featuring a linear fitting slope for all sensors.e) Response and recovery time of CsPbI x Br 3-x PNCs sensors for detecting 8 ppm of NO 2 .f) Comparison of key gas sensor parameters, including sensitivity, stability, LOD, recovery time, and equilibrium constant (K A ), for all CsPbI x Br 3-x PNCs with different halide compositions.
Figure 3. a) I-V characteristics of CsPbI x Br 3-x PNCs deposited on interdigitated Pt-electrodes.b) Dynamic sensing responses of CsPbI x Br 3-x PNCs sensors toward 8 ppm of NO 2 under a bias of 1 V in dark.c) Up-to-date progress in NO 2 sensor performance operating at room-temperature in terms of sensor response and response time in the literature.[16,18,[34][35][36][37][38][39][40][41][42][43][44][45][46]d) NO 2 concentration dependent sensor response devices, featuring a linear fitting slope for all sensors.e) Response and recovery time of CsPbI x Br 3-x PNCs sensors for detecting 8 ppm of NO 2 .f) Comparison of key gas sensor parameters, including sensitivity, stability, LOD, recovery time, and equilibrium constant (K A ), for all CsPbI x Br 3-x PNCs with different halide compositions.

Figure 4 .
Figure 4. Sensing performance of CsPbI 2 Br PNCs gas sensor.a) Dynamic sensing response of CsPbI 2 Br PNCs sensor toward 8 ppm of NO 2 with a bias of 1 V in dark.b) Response and recovery times of CsPbI 2 Br PNCs sensor.c) NO 2 concentration dependent sensor response of CsPbI 2 Br PNCs sensor.d) Selectivity test of CsPbI 2 Br PNCs sensor for 8 ppm of various gases at room-temperature.

Figure 5 .
Figure 5. a) PL spectra of CsPbI 2 Br PNCs optical sensor with and without 8 ppm of NO 2 .Insets demonstrate the fluorescence photographs of the sensor with and without 8 ppm of NO 2 under UV excitation of 365 nm.b) PL intensity of three batches of CsPbI 2 Br PNCs sensors under four ON/OFF cycles of 8 ppm NO 2 exposure.c) PL intensity decay of CsPbI 2 Br PNCs sensor as a function of NO 2 exposure time.d) Optical response of CsPbI 2 Br PNCs sensor with respect to different concentration of NO 2 gas.

Figure 6 .
Figure 6.a) Schematic illustration of adsorption-charge transfer mechanism between PNCs and NO 2 molecules.b) Time-resolved PL of the CsPbI 2 Br PNCs sensor with and without 8 ppm of NO 2 gas.c) Normalized PL of different halide composition CsPbI x Br 3-x PNCs with and without 8 ppm of NO 2 exposure to show the PL quenching ratio.d) Adsorption and desorption kinetics of all CsPbI x Br 3-x PNCs sensors with an inset table of equilibrium constants (K A ).