Hybrid 1D/3D-Structured Perovskite as a Highly Selective and Stable Sensor for NO2 Detection at Room Temperature

To exploit high-performance and stable sensing materials with a room working temperature is pivotal for portable and mobile sensor devices. However, the common sensors based on metal oxide semiconductors usually need a higher working temperature (usually above 300 °C) to achieve a good response toward gas detection. Currently, metal halide perovskites have begun to rise as a promising candidate for gas monitoring at room temperature but suffer phase instability. Herein, we construct 1D/3D PyPbI3/FA0.83Cs0.17PbI3 (denoted by PyPbI3/FACs) bilayer perovskite by post-processing spin-coating Pyrrolidinium hydroiodide (PyI) salt on top of 3D FACs film. Benefitting from the 1D PyPbI3 coating layer, the phase stability of 1D/3D PyPbI3/FACs significantly improves. Simultaneously, the gas sensor based on the 1D/3D PyPbI3/FACs bilayer perovskite presents a superior selectivity and sensitivity toward NO2 detection at room temperature, with a low detection limit of 220 ppb. Exposed to a 50 ± 3% relative humidity (RH) level environment for a consecutive six days, the 1D/3D PyPbI3/FACs perovskite-based sensor toward 10 ppm NO2 can still maintain a rapid response with a slight attenuation. Gas sensors based on hybrid 1D/3D-structured perovskite in this work may provide a new pathway for highly sensitive and stable gas sensors in room working temperature, accelerating its practical application and portable device.


Introduction
Environmental pollution has been a social issue and attracted extensive attention. Especially air contaminations, such as carbides, sulfides, nitrogen compounds and volatile organic compound (VOC) gases, can cause serious damage to the environment and human health [1][2][3][4]. Among them, nitrogen dioxide (NO 2 ) is an extremely common contamination in air atmosphere, predominantly originating from the massive discharge of fossil-fuel combustion and motor vehicle exhaust, but NO 2 with a ppm-level concentration can cause serious respiratory diseases, lung damage, neurasthenia and even endanger life safety [5][6][7]. Therefore, reliable and convenient detection of NO 2 in the surrounding environment is closely relevant to humankind's daily life. Currently, gas monitoring technology contains gas/liquid chromatography methods, acoustic sensors, chemical resistance sensors, electrochemical sensors and optical sensors, etc. [8]. In terms of equipment operation, cost, gas selectivity and response performance, chemical resistance sensors have been considered the most promising gas sensor. The chemical resistance sensors based on the metal oxide layer [9] have also been widely studied as NO 2 gas sensors due to their stability and low cost. However, a high operating temperature of the common metal oxides-based sensor (usually above 300 • C) increases the power consumption and security issues, which impedes further commercial application in portable and mobile sensor devices. Therefore, exploiting efficient sensing materials for NO 2 monitoring at room temperature is pivotal for practical application and is urgently desirable.
Organic-inorganic hybrid metal halide perovskites have recently attracted tremendous scientific research owing to their unique photoelectric properties and low-temperature

Structure and Characterizations
As illustrated in Figure 1a, the 1D/3D PyPbI 3-and FACs perovskite-based sensor was prepared by the post-processing spin-coating method. Firstly, 3D FA 0.83 Cs 0.17 PbI 3 (FACs) perovskite deposited on the gold interdigitated electrode was synthesized by ethyl acetate (EA) antisolvent quenching and annealing at 170 • C. Then, Pyrrolidinium hydroiodide (PyI) in isopropyl alcohol (IPA) solution was dropped on the spinning 3D FACs film. After spincoating and annealing, a 1D/3D PyPbI 3 /FACs bilayer perovskite film was obtained. More detailed experiment procedures are in the Supplementary Materials. As the photographs show in Figure 1b and c, both 3D FACs and 1D/3D PyPbI 3 /FACs perovskite film display a homogeneous black and seem to have no obvious difference.
superior long-term stability, which can almost maintain the original phase and response without few attenuations when exposed to high humidity conditions for a successive six days. Our finding may open up a new application pathway of 1D/3D hybrid-structured perovskite, which accelerate the development of high-stability gas sensors as NO2 monitoring under room temperature and high humidity atmosphere.

Structure and Characterizations
As illustrated in Figure 1a, the 1D/3D PyPbI3-and FACs perovskite-based sensor was prepared by the post-processing spin-coating method. Firstly, 3D FA0.83Cs0.17PbI3 (FACs) perovskite deposited on the gold interdigitated electrode was synthesized by ethyl acetate (EA) antisolvent quenching and annealing at 170 °C. Then, Pyrrolidinium hydroiodide (PyI) in isopropyl alcohol (IPA) solution was dropped on the spinning 3D FACs film. After spin-coating and annealing, a 1D/3D PyPbI3/FACs bilayer perovskite film was obtained. More detailed experiment procedures are in the Supplementary Materials. As the photographs show in Figure 1b and c, both 3D FACs and 1D/3D PyPbI3/FACs perovskite film display a homogeneous black and seem to have no obvious difference.   [50]. At the top of the 1D/3D perovskite film, the surface shares a 1D chain of [PbI3] -octahedron, and a large number of pyridine cations (Py + ) are arranged around the chain, which is the crystal structure of PyPbI3 [22]. Scanning electron microscopy (SEM) images were performed to reveal the surface morphology effect of 1D PyPbI3 on perovskite film. As the top-view SEM image displayed in Figure 2b, 3D FACs perovskite film is composed of densely packed particles with few pinholes but presents a rough surface. Simultaneously, these bare grains present a mass of differentoriented planes. Moreover, the distribution of crystallite size is relatively uniform, with an average value of 360 nm (see the inset in Figure 2b). The thickness of the 3D FACs perovskite layer is about 470 nm, as is shown in the cross-sectional SEM image ( Figure  2c). However, after coating on the 1D PyPbI3 layer, the crystal particles on 1D/3D PyPbI3/FACs perovskite film significantly enlarge, and the corresponding statistical diagram (inset in Figure 2d) shows the crystallite size is about 545 nm. According to the previous reports [45,48], the N-H in Py + cation ions can bind with I in [PbI3] − octahedra cage   [50]. At the top of the 1D/3D perovskite film, the surface shares a 1D chain of [PbI 3 ]octahedron, and a large number of pyridine cations (Py + ) are arranged around the chain, which is the crystal structure of PyPbI 3 [22]. Scanning electron microscopy (SEM) images were performed to reveal the surface morphology effect of 1D PyPbI 3 on perovskite film. As the top-view SEM image displayed in Figure 2b, 3D FACs perovskite film is composed of densely packed particles with few pinholes but presents a rough surface. Simultaneously, these bare grains present a mass of different-oriented planes. Moreover, the distribution of crystallite size is relatively uniform, with an average value of 360 nm (see the inset in Figure 2b). The thickness of the 3D FACs perovskite layer is about 470 nm, as is shown in the cross-sectional SEM image (Figure 2c). However, after coating on the 1D PyPbI 3 layer, the crystal particles on 1D/3D PyPbI 3 /FACs perovskite film significantly enlarge, and the corresponding statistical diagram (inset in Figure 2d) shows the crystallite size is about 545 nm. According to the previous reports [45,48], the N-H in Py + cation ions can bind with I in [PbI 3 ] − octahedra cage via hydrogen bonding. Thus, the formed PyPbI 3 is coated on the top of different-oriented planes of FACs grain, which seems to prompt the grain growth. Moreover, the grain size is heterogeneous, and adjacent grains become loose with vast grain boundaries, availing the penetration and contact of gas molecules. The cross-sectional SEM image verifies that the thickness of the 1D/3D hybrid-structured PyPbI 3 /FACs film is larger than that of 3D FACs (Figure 2e), indicating that the 1D layer successfully covers 3D perovskite film. via hydrogen bonding. Thus, the formed PyPbI3 is coated on the top of different-oriente planes of FACs grain, which seems to prompt the grain growth. Moreover, the grain siz is heterogeneous, and adjacent grains become loose with vast grain boundaries, availin the penetration and contact of gas molecules. The cross-sectional SEM image verifies tha the thickness of the 1D/3D hybrid-structured PyPbI3/FACs film is larger than that of 3D FACs (Figure 2e), indicating that the 1D layer successfully covers 3D perovskite film. X-ray diffraction (XRD) patterns were carried out to confirm the crystal structure o the 1D/3D hybrid-structured and 3D perovskite films. As shown in Figure 2f, compare with 3D perovskite (black curve), the XRD pattern of the 1D/3D hybrid-structured perov skite film (red curve) contains all diffraction peaks of 3D FACs perovskite, including 12.7 14.1°, 19.9°, 24.5°, 28.3°, 31.7° and 34.9° [51]. Meanwhile, the extra characteristic peaks a 11.2°, 25.1° and 29.6° are well corresponding to the (100), , (201), and (31 ̅ 0) planes of 1D PyPbI3 [22], which suggests the successful construction of the 1D/3D PyPbI3/FACs bilaye perovskite film. It is worth mentioning that the diffraction peak of PbI2 at 12.7° appears i 3D FACs but disappears in 1D/3D PyPbI3/FACs, which may be due to the PyI precurso solution reacting with the residual PbI2 in 3D FACs perovskite to form a PyPbI3 coatin layer. The peak intensity of 1D PyPbI3 is relatively weaker in comparison with that of 3D FACs, which is due to the low content of PyI species. This result is consistent with th X-ray diffraction (XRD) patterns were carried out to confirm the crystal structure of the 1D/3D hybrid-structured and 3D perovskite films. As shown in Figure 2f, compared with 3D perovskite (black curve), the XRD pattern of the 1D/3D hybrid-structured perovskite film (red curve) contains all diffraction peaks of 3D FACs perovskite, including 12.7 • , 14.1 • , 19.9 • , 24.5 • , 28.3 • , 31.7 • and 34.9 • [51]. Meanwhile, the extra characteristic peaks at 11.2 • , 25.1 • and 29.6 • are well corresponding to the (100), (201), and (310) planes of 1D PyPbI 3 [22], which suggests the successful construction of the 1D/3D PyPbI 3 /FACs bilayer perovskite film. It is worth mentioning that the diffraction peak of PbI 2 at 12.7 • appears in 3D FACs but disappears in 1D/3D PyPbI 3 /FACs, which may be due to the PyI precursor solution reacting with the residual PbI 2 in 3D FACs perovskite to form a PyPbI 3 coating layer. The peak intensity of 1D PyPbI 3 is relatively weaker in comparison with that of 3D FACs, which is due to the low content of PyI species. This result is consistent with the thinner thickness of PyPbI 3 in the 1D/3D bilayer perovskite film. Optical properties are important characteristics of perovskite materials. Ultraviolet and visible spectrophotometry (UV-vis) and photoluminescence (PL) of the 1D/3D hybrid-structured perovskite film were tested. In Figure 2g, the optical absorbance spectrum of 1D/3D PyPbI 3 /FACs perovskite shows an onset of the absorbance at about 800 nm with a corresponding bandgap of 1.546 eV, and the strong emission peak in the PL spectrum appears at about 815 nm. Compared with the 3D FACs film ( Figure S1), the 1D/3D PyPbI 3 /FACs bilayer film displays similar adsorption in the whole visible wavelength range, but the PL intensity is slightly weakened, which may Molecules 2023, 28, 2615 5 of 13 be due to the increment of defects introduced by PyPbI 3 . These results also suggest that the 1D PyPbI 3 layer is successfully coated on the 3D FACs perovskite film with negligible optical properties changes.

Gas-Sensing Performances
All detection gas-sensing performances based on 3D FACs or 1D/3D PyPbI 3 /FACs were carried out in a sealed container, which is characterized by recording the resistance signals on a Keithley 2401 source meter. The response of the 3D FACs and 1D/3D PyPbI 3 /FACs bilayer perovskite exposed to 10 ppm NO 2 was detected under ambient conditions (temperature 25 ± 2 • C and humidity 50 ± 3%).
Figures 3a and S2a exhibit the transient resistance curve of sensor-based 1D/3D PyPbI 3 /FACs bilayer perovskite and 3D FACs perovskite film toward 10 ppm NO 2 at a constant applied bias voltage of 5 V. Once exposed to NO 2 atmosphere (Figures 3a and S2a), the resistance of both the 1D/3D PyPbI 3 /FACs and 3D FACs perovskite-based sensor decrease quickly, and then recovers when NO 2 is removed, which indicating a reversible interaction between NO 2 detection gas and perovskite sensor material [29]. In general, the gas response can be obtained by the equation S = R a /R g , and the corresponding response/recovery time (τ res /τ res ) is defined as the required time to reach 90% of the response value during the response/recovery process [52].
thinner thickness of PyPbI3 in the 1D/3D bilayer perovskite film. Optical properties are important characteristics of perovskite materials. Ultraviolet and visible spectrophotometry (UV-vis) and photoluminescence (PL) of the 1D/3D hybrid-structured perovskite film were tested. In Figure 2g, the optical absorbance spectrum of 1D/3D PyPbI3/FACs perovskite shows an onset of the absorbance at about 800 nm with a corresponding bandgap of 1.546 eV, and the strong emission peak in the PL spectrum appears at about 815 nm. Compared with the 3D FACs film ( Figure S1), the 1D/3D PyPbI3/FACs bilayer film displays similar adsorption in the whole visible wavelength range, but the PL intensity is slightly weakened, which may be due to the increment of defects introduced by PyPbI3. These results also suggest that the 1D PyPbI3 layer is successfully coated on the 3D FACs perovskite film with negligible optical properties changes.

Gas-Sensing Performances
All detection gas-sensing performances based on 3D FACs or 1D/3D PyPbI3/FACs were carried out in a sealed container, which is characterized by recording the resistance signals on a Keithley 2401 source meter. The response of the 3D FACs and 1D/3D PyPbI3/FACs bilayer perovskite exposed to 10 ppm NO2 was detected under ambient conditions (temperature 25 ± 2 °C and humidity 50 ± 3%).
Figures 3a and S2a exhibit the transient resistance curve of sensor-based 1D/3D PyPbI3/FACs bilayer perovskite and 3D FACs perovskite film toward 10 ppm NO2 at a constant applied bias voltage of 5 V. Once exposed to NO2 atmosphere (Figures 3a and S2a), the resistance of both the 1D/3D PyPbI3/FACs and 3D FACs perovskite-based sensor decrease quickly, and then recovers when NO2 is removed, which indicating a reversible interaction between NO2 detection gas and perovskite sensor material [29]. In general, the gas response can be obtained by the equation S = Ra/Rg, and the corresponding response/recovery time (τres/τres) is defined as the required time to reach 90% of the response value during the response/recovery process [52]. As displayed in Figure 3b, the response and recovery time of the 1D/3D PyPbI3/FACs perovskite-based sensor toward 10 ppm NO2 is 13.75 s and 82.76 s, respectively, which is slower than that of 3D FACs (seen Supplementary Materials Figure S2b), but the time is faster than most of metal oxide-based sensors [53][54][55][56][57]. A longer recovery time suggests that the desorption process of NO2 is relatively difficult, while the response curve at room temperature can automatically return to the initial baseline, indicating the excellent desorption properties of the 1D/3D PyPbI3/FACs perovskite-based sensor. Moreover, the 1D/3D PyPbI3/FACs perovskite-based sensor to NO2 gas displays a stronger response than 3D FACs-based sensors and compared with common metal oxide semiconductor sensors, the operating temperature is room temperature rather than high temperature, which As displayed in Figure 3b, the response and recovery time of the 1D/3D PyPbI 3 /FACs perovskite-based sensor toward 10 ppm NO 2 is 13.75 s and 82.76 s, respectively, which is slower than that of 3D FACs (seen Supplementary Materials Figure S2b), but the time is faster than most of metal oxide-based sensors [53][54][55][56][57]. A longer recovery time suggests that the desorption process of NO 2 is relatively difficult, while the response curve at room temperature can automatically return to the initial baseline, indicating the excellent desorption properties of the 1D/3D PyPbI 3 /FACs perovskite-based sensor. Moreover, the 1D/3D PyPbI 3 /FACs perovskite-based sensor to NO 2 gas displays a stronger response than 3D FACs-based sensors and compared with common metal oxide semiconductor sensors, the operating temperature is room temperature rather than high temperature, which indicates that the 1D/3D PyPbI 3 /FACs perovskite is a promising candidate in the field of low-energy consumption and portable gas sensors.
The reversibility and stability of gas sensors are the pivotal parameters that determine their further practical application. Figures 4a and S3 depict the dynamic transient response of sensors based on the 1D/3D PyPbI 3 /FACs and 3D FACs perovskite toward 10 ppm NO 2 gas at room temperature. After four consecutive cycles, the response of both 1D/3D PyPbI 3 /FACs and 3D FACs perovskite-based sensors toward 10 ppm NO 2 almost remains at the initial value with negligible attenuation, indicating that the 1D/3D PyPbI 3 /FACs and the perovskite-based sensor has excellent repeatability and reversibility. mine their further practical application. Figures 4a and S3 depict the dynamic transient response of sensors based on the 1D/3D PyPbI3/FACs and 3D FACs perovskite toward 10 ppm NO2 gas at room temperature. After four consecutive cycles, the response of both 1D/3D PyPbI3/FACs and 3D FACs perovskite-based sensors toward 10 ppm NO2 almost remains at the initial value with negligible attenuation, indicating that the 1D/3D PyPbI3/FACs and the perovskite-based sensor has excellent repeatability and reversibility.  Figure S4. Similarly, the resistance values of the 1D/3D perovskite-based sensor will decrease when the contact concentration of NO2 increases and subsequently recover to baseline after removing NO2 gas. Figure 4c reveals the fitted sensor response of the 1D/3D hybrid-structured PyPbI3/FACs perovskite as a function of NO2 gas. As shown in Figure 4c, the 1D/3D PyPbI3/FACs perovskite-based sensor depicts good linearity in response to low-concentration NO2 gas, with a corresponding coefficient of 0.99. The fitted equation derived from The dynamic sensing transients of the 1D/3D PyPbI 3 /FACs perovskite-based sensors toward different concentrations of NO 2 gas ranging from 0.5 to 10 ppm were manifested in Figure 4b. The transient response of the 1D/3D PyPbI 3 /FACs perovskite-based sensor toward NO 2 gas rapidly increases with a continuous increase of NO 2 and then returns to the original baseline state once NO 2 is removed. The response values of the 1D/3D PyPbI 3 /FACs perovskite-based sensor are 1.13, 1.2, 1.78 and 2.47 toward 0.5, 1, 2, 5 and 10 ppm NO 2 , respectively. The corresponding dynamic resistance variation of the 1D/3D perovskite-based sensor toward different concentrations of NO 2 is manifested in Figure S4. Similarly, the resistance values of the 1D/3D perovskite-based sensor will decrease when the contact concentration of NO 2 increases and subsequently recover to baseline after removing NO 2 gas. Figure 4c reveals the fitted sensor response of the 1D/3D hybrid-structured PyPbI 3 /FACs perovskite as a function of NO 2 gas. As shown in Figure 4c, the 1D/3D PyPbI 3 /FACs perovskite-based sensor depicts good linearity in response to low-concentration NO 2 gas, with a corresponding coefficient of 0.99. The fitted equation derived from the response/recovery curve in Figure 4b is y = 1.07 + 0.14x, which can deduce that the average sensitivity is 0.14 ppm −1 . The limit of detection (LOD) is defined by the lowest detectable concentration, according to the equation: LOD = 3 × rms/S, wherein rms refers to the root-mean-squared noise of the sensor, and S refers to sensitivity. The calculated LOD of the NO 2 gas sensor based on the 1D/3D hybrid-structured PyPbI 3 /FACs perovskite can be as low as 220 ppb, suggesting its high sensitivity.
The selectivity of a sensor device is another critical factor owing to the complicated gas compositions in the atmosphere. To prove the selectivity of the device, the response of 1D/3D PyPbI 3 /FACs perovskite sensors to various gases with 10 ppm, including NH 3 , methanol, triethylamine, n-butanol, H 2 , ethanol, acetone and formaldehyde, are investigated under the same environmental conditions. In Figure S5, except for NO 2 , and NH 3 , the resistances of the 1D/3D PyPbI 3 /FACs perovskite-based sensor toward other gases depict no signals. In Figure 4d, the 1D/3D hybrid-structured PyPbI 3 /FACs perovskite-based sensor presents the most remarkable response toward NO 2 . Although the 1D/3D PyPbI 3 /FACs perovskite-based sensor to NH 3 presents a slight response of 1.12, it is easy to distinguish NH 3 and NO 2 . It is due to the transient resistance changes of the 1D/3D PyPbI 3 /FACs perovskite-based sensor toward NO 2 and NH 3 that it showed the opposite response (Figures 3a and S6) when gas was injected or removed. These results suggest that the 1D/3D PyPbI 3 /FACs perovskite-based sensor shows a superior selectivity toward NO 2 gas.
The long-term stability of a gas sensor is another important factor in determining a large-scale commercial application. Previous perovskite devices are usually stored in a glove box or sealed. Here, to reveal the stability, sensor-based 1D/3D PyPbI 3 /FACs hybridstructured perovskite material is stored without any encapsulation under high humidity conditions (50 ± 3%). As the photographs manifest in Figure 5a, the 1D/3D PyPbI 3 /FACs hybrid-structured perovskite film still maintains the original macroscopic characteristics well in the successive six days, and no obvious degradation occurs on the surface. The structure and phase changes of the 1D/3D PyPbI 3 /FACs perovskite film are traced by an XRD test. In Figure 5b, no new diffraction peaks appear, and the position and strength of the characteristic peaks in XRD patterns almost remain the same for the consecutive days, but a 3D FACs film without 1D PyPbI 3 coating layer presents decomposition after three days ( Figure S7). These results confirm that 1D PyPbI 3 coating film significantly enhances the phase stability of 3D FACs perovskite, which may be due to the fact that PyPbI 3 with a one-dimensional structure as a barrier layer can effectively restrain the intrusion of water molecules into the vulnerable FACs perovskite film to improve the phase stability. In addition, gas-sensing tests in Figure 6 reveal the response of the 1D/3D PyPbI3/FACs and 3D FACs perovskite-based sensor to 10 ppm NO2 gas in the consecutive six days, and the sensor device without encapsulation is kept at room temperature in the ambient environment (temperature 25 ± 2 °C and relative humidity 50 ± 3%). As the transient response curves depicted in Figure 6a-c, the 1D/3D PyPbI3/FACs perovskite-based sensor toward 10 ppm NO2 still maintains a superior sensing performance with fast response/recovery time and reversibility after being successively exposed to high humidity conditions (50 ± 3%) for 6 days. The initial 3D FACs perovskite-based sensor shows good reversibility ( Figure S3), but the 3D FACs perovskite-based sensor exhibits poor stability ( Figure 6). The time-dependent response of the 3D FACs perovskite-based sensor toward 10 ppm NO2 shows a sharp attenuation. On the 3rd day, the response of the 3D FACs perovskite-based sensor decreased from 2.10 to 1.20 (Figure 6b), suggesting that a sensor device based on 3D FACs perovskite scarcely works. An obvious comparison of the timedependent stability of 3D FACs and 1D/3D PyPbI3/FACs is depicted in Figure 6d, suggest- In addition, gas-sensing tests in Figure 6 reveal the response of the 1D/3D PyPbI 3 /FACs and 3D FACs perovskite-based sensor to 10 ppm NO 2 gas in the consecutive six days, and the sensor device without encapsulation is kept at room temperature in the ambient environment (temperature 25 ± 2 • C and relative humidity 50 ± 3%). As the transient response curves depicted in Figure 6a-c, the 1D/3D PyPbI 3 /FACs perovskite-based sensor toward 10 ppm NO 2 still maintains a superior sensing performance with fast response/recovery time and reversibility after being successively exposed to high humidity conditions (50 ± 3%) for 6 days. The initial 3D FACs perovskite-based sensor shows good reversibility ( Figure S3), but the 3D FACs perovskite-based sensor exhibits poor stability ( Figure 6). The time-dependent response of the 3D FACs perovskite-based sensor toward 10 ppm NO 2 shows a sharp attenuation. On the 3rd day, the response of the 3D FACs perovskite-based sensor decreased from 2.10 to 1.20 (Figure 6b), suggesting that a sensor device based on 3D FACs perovskite scarcely works. An obvious comparison of the time-dependent stability of 3D FACs and 1D/3D PyPbI 3 /FACs is depicted in Figure 6d, suggesting the superior stability of the 1D/3D PyPbI 3 /FACs perovskite-based sensor. In addition, the responding response of the 1D/3D PyPbI 3 /FACs bilayer perovskite-based sensor is 2.54, 2.21, and 2.12 on the 1st, 3rd and 6th day, respectively, presenting a slight reduction in the successive six days. Although the 1D PyPbI 3 layer can block the water molecules infiltrating into 3D FACs perovskite, the water molecules in high-humidity conditions inevitably absorb and occupy active sites, which decreases adsorption sites for NO 2 along with a decline of the response value. More interestingly, the 1D/3D PyPbI 3 /FACs bilayer perovskite-based sensor also displays good stability in a higher humidity environment. The 1D/3D PyPbI 3 /FACs sensor is placed in a higher humidity environment (relative humidity 70 ± 1%) for six successive days. As shown in Figure S8, the 1D/3D PyPbI 3 /FACs perovskite-based sensor in relative humidity of 70 ± 1% displays a similar response tendency to that in 50 ± 3%. The time-dependent response in Figure S8 manifests a good stability of the 1D/3D PyPbI 3 /FACs perovskite-based sensor in a higher humidity condition. The response value in RH of 70 ± 1% is slightly inferior to that in RH of 50 ± 3%, which can ascribe to the decreased sites for NO 2 due to more water molecules in higher humidity adsorbed on the sites of the perovskite material. Fortunately, the response attenuation of the gas sensor based on a 1D/3D PyPbI 3 /FACs bilayer perovskite can be effectively suppressed by device encapsulation or prepositive dehydration equipment. Thus, the 1D/3D PyPbI 3 /FACs hybrid-structured perovskite-based sensor exhibits a promising application toward low-concentration NO 2 detection at room temperature. can ascribe to the decreased sites for NO2 due to more water molecules in higher humidity adsorbed on the sites of the perovskite material. Fortunately, the response attenuation of the gas sensor based on a 1D/3D PyPbI3/FACs bilayer perovskite can be effectively suppressed by device encapsulation or prepositive dehydration equipment. Thus, the 1D/3D PyPbI3/FACs hybrid-structured perovskite-based sensor exhibits a promising application toward low-concentration NO2 detection at room temperature.

Fabrication Process of the 1D/3D Perovskite-Based Sensor
An interdigital electrode (electrode spacing: 100 µm, size: 10 × 10 mm 2 , substrate: glass) was prepared by a thermal evaporation method. The detailed procedure is as the following: Firstly, the glass substrate was successively washed with distilled water, ethanol, acetone and 2-propanol for 20 min. Then, clean N 2 gas was used to blow away the residual liquid. The glass substrate was treated by Ar plasma with a power of 250 W for 5 min. After that, Au as an evaporated source is deposited on the top of a clean glass substrate under a pressure of 5 × 10 −5 Pa in a vacuum chamber to obtain an interdigital electrode.
The 1D/3D perovskite film on the interdigital electrode was prepared via a two-step spin-coating process. Firstly, for the mixed A cation FA 0.83 Cs 0.17 PbI 3 (abbreviated as FACs in the main text) metal halide perovskite layer, a one-step spin-coating method was adopted as previously reported [51]. In detail, to prepare FACs film, the precursor solution was first prepared by dissolving 1 mmol FAI (170 mg), 0.2 mmol CsI (56 mg) and 1.2 mmol PbI 2 (552 mg) in 1 mL mixed solvent containing DMF (700 µL) and DMSO (300 µL). The prepared solution was stirred for 2 h at room temperature to obtain a clear and transparent liquid. Before the preparation of the FACs thin film, the precursor solution was filtered through a 0.22 µm poly tetra fluoroethylene (PTFE) filter. Then, 50 µL precursor solutions were spin-coated at 1000 rpm for 10 s and then at 6000 rpm for 30 s via a one-step method. Additionally, 110 µL of EA was dripped on the film to extract the solvent at 12th s prior to the end of the process. Then, the film was annealed at 170 • C for 10 min to obtain a 3D FACs film.
For the 1D/3D PyPbI 3 /FACs hybrid-structured perovskite film, 50 µL of PyI (10 mg PyI in 1 mL IPA) solutions were spin-coated on top of the FACs perovskite layer at 3000 rpm for 30 s, then the film was annealed at 120 • C for 10 min.
The whole process was carried out under ambient conditions, and the devices were stored under ambient conditions (temperature 25 ± 2 • C; humidity 50 ± 3%) without any encapsulations.

Characterization
The surface morphology of 3D FACs and 1D/3D PyPbI 3 /FACs perovskite was characterized by using scanning electron microscopy (SEM, Regulus-8100, HITACHI, Tokyo, Japan) operating at 20 kV. The crystal structure was investigated by X-ray diffraction (XRD, DX2700, Dandong Kemait NDT C., Ltd, Liaoning, China) with a 2θ ranging from 5 to 37 • in a step of 5 • min −1 , using a monochromatic Cu Kα radiation source. Before the 3D FACs and 1D/3D PyPbI 3 /FACs perovskite were exposed to gases, the XRD measurement was held on a glass substrate different from the deposited material on electrodes in order to avoid any interaction of sensing elements from the X-ray beam. The absorbance spectrum of the 3D FACs and PyPbI 3 /FACs perovskite were measured by UV-visible spectroscopy (T9, PERSEE, Beijing, China) in the wavelength ranging from 400 to 870 nm on a quartz glass substrate. The photoluminescence (PL) spectrum of 3D FACs and 1D/3D PyPbI 3 /FACs hybrid-structured perovskite were investigated with a Hitachi F-4600 spectrofluorometer, Tokyo, Japan.

Gas-Sensing Test Parameters
The gas-sensing test was collected by a Keithley 2401 source meter. The environmental test conditions were as follows: the pressure was 101,325 Pa, the working temperature was about 25 ± 2 • C and the relative humidity was about 50 ± 3%. A sealed chamber with a volume of 18 L is used for gas sensor testing, and an air inlet and outlet are located in the chamber. Dry NH 3 , H 2 and NO 2 were supplied by a sealed gas bag. After injecting the detection gas into the closed box, a built-in fan was used to diffuse the detection gas. When the target source is gas, the volume of injected gas is calculated by the formula Vg = cV 0 /c 0 where c is the measured concentration in the test chamber, c 0 is the concentration of the target gas and V is the volume of the sealed chamber (18 L).
While the target source was liquid, such as methanol, ethanol, n-butanol, acetone, triethylamine and formaldehyde, the detection gases were obtained by evaporating the corresponding injected pure liquids. The required amount of target liquid was dropped onto the micro-heating plate via a microsyringe, and the required liquid volume was calculated by the formula V L = PV 0 cM/ρRT where P is the standard atmospheric pressure (P = 101.325 kPa), V 0 is the volume of the sealed chamber (18 L), c is the gas concentration, M is the molar mass, R is the gas constant (R = 8.314 J/(mol·K)), T the ambient temperature and ρ is the density of the liquid.

Conclusions
In this work, a 1D/3D PyPbI 3 /FACs bilayer perovskite-based sensor is constructed by post-spin-coating a PyI precursor on the top of a 3D FA 0.83 Cs 0.17 PbI 3 perovskite film under an ambient atmosphere. The introduced 1D PyPbI 3 coating layer can not only stabilize the 3D FACs perovskite phase but also produce loose grain boundaries to avail the penetration and contact of gas molecules. Consequently, the 1D/3D PyPbI 3 /FACs bilayer perovskite-based sensor shows a superior selectivity and sensitivity to NO 2 at room temperature, with a low detection limit of 220 ppb. Moreover, compared with the 3D FACsbased sensor, the response and stability of the 1D/3D PyPbI 3 /FACs hybrid-structured perovskite-based sensor have been significantly improved. Exposed in a high humidity environment (50 ± 3%) for six consecutive days, the 1D/3D PyPbI 3 /FACs perovskite-based sensor toward 10 ppm NO 2 can rapidly respond with a slight attenuation. This work opens up a new approach to designing and constructing high-stability sensors to detect NO 2 at room temperature, accelerating its further practical application.