A High-Performance UVA Photodetector Based on Polycrystalline Perovskite MAPbCl3/TiO2 Nanorods Heterojunctions

The application of TiO2 nanorods in the field of ultraviolet (UV) photodetectors is hindered by a high dark current, which is attributed to crystal surface defects and intrinsic excitation by carrier thermal diffusion. Here, a photodetector based on polycrystalline perovskite MAPbCl3/TiO2 nanorods heterojunctions has been fabricated to overcome the shortcoming. The structure was composed of horizontal MAPbCl3 polycrystalline and vertically aligned TiO2 nanorods array. Many localized depletion regions at the MAPbCl3/TiO2 interface can reduce the dark current. The TiO2/MAPbCl3 detector shows high performance including a high ratio of light-dark current of about six orders of magnitude, which is much larger than that of the TiO2 detector. This study indicates the potential in the TiO2/MAPbCl3 heterojunction to fabricate high-performance UV detectors.

Researchers have found that an advantage of the TiO 2 nanorods array in the UV photodetector is a reduction of the recombination probability of photogenerated electron-hole pairs because of the surface trap states associated with adsorbed O 2 molecules on the surface of TiO 2 nanorods [21]. Hakan Karaagac et al. fabricated a Schottky UV photodetector based on well-aligned TiO 2 nanorod arrays, which exhibits high photosensitivity and excellent spectral selectivity, but the dark current (1.2 × 10 −7 A at reverse 1 V) was relatively high [21]. CH 3 NH 3 PbCl 3 (MAPbCl 3 ) absorbs UV light below 400 nm due to its 2.88-3.11 eV bandgap [22][23][24] and has been applied in UV detection due to its high optical absorption capacity, high carrier mobility, long carrier diffusion length and stability. UV detectors based on MAPbCl 3 single crystal show high properties, but the difficult fabrication technique places a restriction on their development in photonic crystal structures [25][26][27]. Therefore, polycrystalline MAPbCl 3 film is preferred to compose the heterojunction. Jialin Yang et al. have successfully studied how a polycrystalline CH 3 NH 3 PbCl 3 /ZnO heterojunction improved UV photodetector performance compared to single ZnO, but the dark current remained high [28]. Liu shuo et al. have fabricated a Ga 2 O 3 /polycrystalline MAPbCl 3 UVA photodetector with rapid response and recovery property, which nonetheless had a dark current of 6.8 µA at −1.5 V bias [29].
In this work, a UV photodetector based on polycrystalline perovskite MAPbCl 3 /TiO 2 nanorods array heterojunction has been successfully prepared. The UV photodetector with many localized depletion regions at the MAPbCl 3 /TiO 2 interface shows better performance than that of pure TiO 2 , with improved dark current, light-dark current ratio and a shorter response time. The results indicate that MAPbCl 3 /TiO 2 heterojunction is a promising way to improve UV photodetector performance.

Preparation of TiO 2 One-Dimensional Nanorods Array Film on FTO
TiO 2 nanorods array film was prepared on the surface of the FTO (fluorine-doped tinoxide) substrate by a low-temperature hydrothermal method. First, FTO substrate (15 Ω per square) was cleaned in acetone, ethanol and deionized water and dried in a nitrogen stream. Subsequently, 10 mL of toluene, 1 mL of tetrabutyltitanate [Ti(OC 4 H 9 ) 4 ], 0.2 mL of titanium tetrachloride and 1 mL hydrochloric acid (37%) were added in a sealed Teflon-lined stainless steel autoclave (23 mL). Then, the substrate was placed in the autoclave, heated at 150 • C for 5 h and air-cooled to room temperature (25 • C). After washing with deionized water, a uniform nanorods array was obtained. The TiO 2 growing system in this experiment is a mixed solution composed of Ti(OC 4 H 9 ) 4 , Ti(OH) 4

Preparation of TiO 2 Nanorods/MAPbCl 3 Heterojunction on FTO
Polycrystalline MAPbCl 3 film was prepared by a one-step spin-coating method with antisolvent-assisting. First, 1 mL dimethyl sulfoxide (DMSO) and 1 mL N, N-Dimethylformamide (DMF) were mixed, and then 0.135 g MACl and 0.566 g PbCl 2 were weighed and added to the above solution and stirred for 30 min until the solution became completely transparent to obtain MAPbCl 3 precursor solution (1 mol/L) [23]. Then, 75 µL solution was coated on TiO 2 film by rotation for 30 s at 3000 rpm. Next, a drop of toluene was put on the TiO 2 film and spun for 20 s. Finally, it was dried at 80 • C for 5 h. Finally, two Ag paste pads were deposited directly on the film and FTO, respectively, to make a Schottky photodiode UV detector. The active area of the electrode is about 0.25 mm 2 .

Material Characterization and Device Measurement
The morphology of the prepared films was characterized by a scanning electron microscope (SEM JEOS JSM-6700F). X-ray diffraction (XRD) patterns were performed using a Shimadzu XRD-6000 diffractometer (Shimadzu, Kyoto, Japan). A Shimadzu UV-3600 Pharma Spec UV spectrophotometer was used to obtain the UV-Vis absorption spectra. The photoelectric performances were analyzed by a program-controlled semiconductor characterization system (Keithley 2450 Source Meter, Solon, OH, USA). The light source was provided by a 30 W deuterium lamp, and a monochromatic lamp was used to provide monochromatic light.

Results and Discussion
The UVA photodetector with the structure of FTO/TiO 2 nanorods array/polycrystalline perovskite MAPbCl 3 is shown in Figure 1. TiO 2 nanorods/perovskite MAPbCl 3 heterojunctions worked as the active layer. FTO and Ag served as electrodes (A mask plate with a hollow circular pattern (about 0.25 mm 2 ) was used to mask the material. Silver paste electrodes were coated on the FTO and the material, respectively, and the leads were led out for testing. Then, the whole device was dried at 75 • C for 15 min to stabilize the electrodes. Then, two red and black wires were led out from the Keithley 2450, with the red wire (positive) connected to the Ag above the FTO and the black wire (negative) connected to the Ag above the semiconductor).

Material Characterization and Device Measurement
The morphology of the prepared films was characterized by a scanning electron microscope (SEM JEOS JSM-6700F). X-ray diffraction (XRD) patterns were performed using a Shimadzu XRD-6000 diffractometer (Shimadzu, Kyoto, Japan). A Shimadzu UV-3600 Pharma Spec UV spectrophotometer was used to obtain the UV-Vis absorption spectra. The photoelectric performances were analyzed by a program-controlled semiconductor characterization system (Keithley 2450 Source Meter, Solon, OH, USA). The light source was provided by a 30 W deuterium lamp, and a monochromatic lamp was used to provide monochromatic light.

Results and Discussion
The UVA photodetector with the structure of FTO/TiO2 nanorods array/polycrystalline perovskite MAPbCl3 is shown in Figure 1. TiO2 nanorods/perovskite MAPbCl3 heterojunctions worked as the active layer. FTO and Ag served as electrodes (A mask plate with a hollow circular pattern (about 0.25 mm 2 ) was used to mask the material. Silver paste electrodes were coated on the FTO and the material, respectively, and the leads were led out for testing. Then, the whole device was dried at 75 °C for 15 min to stabilize the electrodes. Then, two red and black wires were led out from the Keithley 2450, with the red wire (positive) connected to the Ag above the FTO and the black wire (negative) connected to the Ag above the semiconductor). The morphology of TiO2 one-dimensional nanorods array film on FTO and TiO2 nanorods/MAPbCl3 heterojunctions on FTO is shown in Figure 2. Figure 2a,b show the topview and the sectional-view SEM images of the TiO2 one-dimensional nanorods array, respectively. It can be observed that the TiO2 nanorods array is uniform, compact and perpendicular to the substrate, which is conducive to the preparation of electronic devices. Figure 2c,d show the top-view and the sectional-view SEM images of the TiO2 nanorods/MAPbCl3 heterojunctions, respectively. It can be found that the polycrystalline MAP-bCl3 film prepared by the one-step method covers the surface of TiO2 nanorods well. The thickness of MAPbCl3 film is about 0.467 μm, and the thickness of TiO2 nanorods film is about 2.178 μm. There are lots of grain boundaries in polycrystalline MAPbCl3, which could scatter carriers and thus lead to low mobility. The MAPbCl3 layer adheres to the TiO2 nanorods array layer, which contributes to the high performance of the detector. The morphology of TiO 2 one-dimensional nanorods array film on FTO and TiO 2 nanorods/MAPbCl 3 heterojunctions on FTO is shown in Figure 2. Figure 2a,b show the top-view and the sectional-view SEM images of the TiO 2 one-dimensional nanorods array, respectively. It can be observed that the TiO 2 nanorods array is uniform, compact and perpendicular to the substrate, which is conducive to the preparation of electronic devices. Figure 2c,d show the top-view and the sectional-view SEM images of the TiO 2 nanorods/MAPbCl 3 heterojunctions, respectively. It can be found that the polycrystalline MAPbCl 3 film prepared by the one-step method covers the surface of TiO 2 nanorods well. The thickness of MAPbCl 3 film is about 0.467 µm, and the thickness of TiO 2 nanorods film is about 2.178 µm. There are lots of grain boundaries in polycrystalline MAPbCl 3 , which could scatter carriers and thus lead to low mobility. The MAPbCl 3 layer adheres to the TiO 2 nanorods array layer, which contributes to the high performance of the detector. In Figure 3, which displays the XRD patterns of the TiO2 nanorods array and polycrystalline perovskite MAPbCl3, it can be observed that the crystal diffraction peaks are very high, indicating that the obtained materials have good crystallization and clear crystal particles. Figure 3a shows that the diffraction peaks of the obtained TiO2 correspond exactly to the peaks of the standard rutile type TiO2 (JCPDS Card No.76-1938). The characteristic diffraction peaks of MAPbCl3 generated by crystal planes (100), (110), (200), (210) and (211) in Figure 3b are included, indicating that the prepared MAPbCl3 perovskite shows consistency with the previously reported data of MAPbCl3 perovskite obtained through the conventional crystallization technique [31]. The optical characteristics of each layer in the photodetector were studied by absorption spectra and Tauc plots, as shown in Figure 4. Figure 4a shows the UV visible absorption spectra of FTO/TiO2 and FTO/TiO2/MAPbCl3 from 300 nm to 600 nm. When the photodetector operates, the illumination light enters from the FTO side. Radiation with a wavelength below 300 nm is completely absorbed, and only radiation with a wavelength   Figure 3, which displays the XRD patterns of the TiO2 nanorods array crystalline perovskite MAPbCl3, it can be observed that the crystal diffraction very high, indicating that the obtained materials have good crystallization and tal particles. Figure 3a shows that the diffraction peaks of the obtained TiO2 c exactly to the peaks of the standard rutile type TiO2 (JCPDS Card No.76-1938) acteristic diffraction peaks of MAPbCl3 generated by crystal planes (100), (110), and (211) in Figure 3b are included, indicating that the prepared MAPbCl3 shows consistency with the previously reported data of MAPbCl3 perovskit through the conventional crystallization technique [31]. The optical characteristics of each layer in the photodetector were studied tion spectra and Tauc plots, as shown in Figure 4. Figure 4a shows the UV visib tion spectra of FTO/TiO2 and FTO/TiO2/MAPbCl3 from 300 nm to 600 nm. Whe todetector operates, the illumination light enters from the FTO side. Radiat The optical characteristics of each layer in the photodetector were studied by absorption spectra and Tauc plots, as shown in Figure 4. Figure 4a shows the UV visible absorption spectra of FTO/TiO 2 and FTO/TiO 2 /MAPbCl 3 from 300 nm to 600 nm. When the photodetector operates, the illumination light enters from the FTO side. Radiation with a wavelength below 300 nm is completely absorbed, and only radiation with a wavelength above 300 nm can pass through the FTO substrate. Moreover, the active layer composed of TiO 2 or TiO 2 /MAPbCl 3 absorbs radiation with a wavelength below 400 nm. TiO 2 has excellent absorption, from 330 to370 nm, and the absorption decreases over 370 nm, while TiO 2 /MAPbCl 3 has excellent absorption, from 330 to400 nm, and the absorption decreases over 400 nm. One of the advantages of TiO 2 /MAPbCl 3 is that it increases the detector's response range closer to 400 nm. Therefore, the radiation whose wavelengths distribute between 300 nm and 400 nm can be collected by TiO 2 /MAPbCl 3 , as shown in the inset of Figure 4b, which corresponds to the UVA range. As calculated in Figure 4b Tauc plots, the bandgap width of rutile TiO 2 is 3.05 eV and the bandgap width becomes 2.98 eV when introducing MAPbCl 3 .
Sensors 2023, 21, x FOR PEER REVIEW 5 of 12 above 300 nm can pass through the FTO substrate. Moreover, the active layer composed of TiO2 or TiO2/MAPbCl3 absorbs radiation with a wavelength below 400 nm. TiO2 has excellent absorption, from 330 to370 nm, and the absorption decreases over 370 nm, while TiO2/MAPbCl3 has excellent absorption, from 330 to400 nm, and the absorption decreases over 400 nm. One of the advantages of TiO2/MAPbCl3 is that it increases the detector's response range closer to 400 nm. Therefore, the radiation whose wavelengths distribute between 300 nm and 400 nm can be collected by TiO2/MAPbCl3, as shown in the inset of Figure 4b, which corresponds to the UVA range. As calculated in Figure 4b Tauc plots, the bandgap width of rutile TiO2 is 3.05 eV and the bandgap width becomes 2.98 eV when introducing MAPbCl3. The absorption edge of TiO2/MAPbCl3 shows a slighter redshift than that of TiO2, indicating a narrower bandgap of TiO2/MAPbCl3, which will provide a possibility to adjust the detector's response range. The practical absorption edge of TiO2 was lower than the theoretical value of pure rutile TiO2 (1240/3.0 = 413 nm). This may be attributed to the size quantization of nanorods, which has been demonstrated by Brus [32]: where R is the radius of semiconductor particle, * and * are effective masses of the electron and hole in the semiconductor, is the permittivity of rutile TiO2, ℎ is the Planck constant, ∆ is the bandgap difference between the original bulk size and nanoscale of the same semiconductor material. According to this formula, the absorption edge will move to the short wavelength as the particle size decreases [33].
The I-V characteristics of the UV detector based on FTO/TiO2, FTO/MAPbCl3 and FTO/TiO2/MAPbCl3 in dark and under illumination are shown in Figure 5. The prepared UV detectors have the characteristics of Schottky diodes, which exhibit nonlinear and unsaturated behavior, as is shown in Figure 5d-f. Under forward bias, the dark current increases rapidly and results in large noise. Therefore, we focus on the reverse characteristics of the detector. For the UV detector based on FTO/TiO2, at −2 V bias, the dark current is 1.557 × 10 −6 A, and the photocurrent reaches 1.359 × 10 −4 A under the irradiation of 200 μW/cm 2 at a wavelength of 350 nm UV light, as shown in Figure 5a. For the UV detector based on FTO/MAPbCl3, at −2 V bias, the dark current is 1.043 × 10 −9 A, and the photocurrent reaches 1.561 × 10 −7 A under the irradiation of 580 μW/cm 2 at a wavelength of 390 nm UV light, as shown in Figure 5b. For the UV detector based on FTO/TiO2/MAPbCl3, at −2 V bias, the dark current is 2.69 × 10 −10 A and the photocurrent reaches 1.632 × 10 −4 A under the irradiation of 255 μW/cm 2 at a wavelength of 360 nm UV light, as shown in Figure 5c. The ratio of light to dark current is more than six orders of magnitude, which proves that The absorption edge of TiO 2 /MAPbCl 3 shows a slighter redshift than that of TiO 2 , indicating a narrower bandgap of TiO 2 /MAPbCl 3 , which will provide a possibility to adjust the detector's response range. The practical absorption edge of TiO 2 was lower than the theoretical value of pure rutile TiO 2 (1240/3.0 = 413 nm). This may be attributed to the size quantization of nanorods, which has been demonstrated by Brus [32]: where R is the radius of semiconductor particle, m * e and m * h are effective masses of the electron and hole in the semiconductor, ε R is the permittivity of rutile TiO 2 , h is the Planck constant, ∆E g is the bandgap difference between the original bulk size and nanoscale of the same semiconductor material. According to this formula, the absorption edge will move to the short wavelength as the particle size decreases [33].
The I-V characteristics of the UV detector based on FTO/TiO 2 , FTO/MAPbCl 3 and FTO/TiO 2 /MAPbCl 3 in dark and under illumination are shown in Figure 5. The prepared UV detectors have the characteristics of Schottky diodes, which exhibit nonlinear and unsaturated behavior, as is shown in Figure 5d-f. Under forward bias, the dark current increases rapidly and results in large noise. Therefore, we focus on the reverse characteristics of the detector. For the UV detector based on FTO/TiO 2 , at −2 V bias, the dark current is 1.557 × 10 −6 A, and the photocurrent reaches 1.359 × 10 −4 A under the irradiation of 200 µW/cm 2 at a wavelength of 350 nm UV light, as shown in Figure 5a. For the UV detector based on FTO/MAPbCl 3 , at −2 V bias, the dark current is 1.043 × 10 −9 A, and the photocurrent reaches 1.561 × 10 −7 A under the irradiation of 580 µW/cm 2 at a wavelength of 390 nm UV light, as shown in Figure 5b. For the UV detector based on FTO/TiO 2 /MAPbCl 3 , at −2 V bias, the dark current is 2.69 × 10 −10 A and the photocurrent reaches 1.632 × 10 −4 A under the irradiation of 255 µW/cm 2 at a wavelength of 360 nm UV light, as shown in Figure 5c. The ratio of light to dark current is more than six orders of magnitude, which proves that the optical properties of the TiO 2 /MAPbCl 3 heterojunction detector meet our requirements.

ER REVIEW 6 of 12
the optical properties of the TiO2/MAPbCl3 heterojunction detector meet our requirements.  Figure 6 shows I-V characteristics of FTO/TiO2/MAPbCl3 UV detector in dark and at a wavelength of 360 nm UV light with various light intensity. All data were obtained at −2 V bias. Under the irradiation of 120 μW/cm 2 , the dark current is 3.69 × 10 −12 A and the photocurrent reaches 3.621 × 10 −6 A. Under the irradiation of 185 μW/cm 2 , the dark current is 2.27 × 10 −9 A and the photocurrent reaches 1.37 × 10 −5 A. Under the irradiation of 255 μW/cm 2 , the dark current is 2.69 × 10 −10 A and the photocurrent reaches 1.632 × 10 −4 A. Under the irradiation of 345 μW/cm 2 , the dark current is 1.516 × 10 −9 A and the photocurrent reaches 2.484 × 10 −4 A. Compared with dark current of FTO/TiO2 UV detector, which is 10 −6 A, the dark current of FTO/TiO2/MAPbCl3 UV detector improved a lot, which changes from 10 −9 A to 10 −12 A. The photo current increased as irradiation increased.  Figure 6 shows I-V characteristics of FTO/TiO 2 /MAPbCl 3 UV detector in dark and at a wavelength of 360 nm UV light with various light intensity. All data were obtained at −2 V bias. Under the irradiation of 120 µW/cm 2 , the dark current is 3.69 × 10 −12 A and the photocurrent reaches 3.621 × 10 −6 A. Under the irradiation of 185 µW/cm 2 , the dark current is 2.27 × 10 −9 A and the photocurrent reaches 1.37 × 10 −5 A. Under the irradiation of 255 µW/cm 2 , the dark current is 2.69 × 10 −10 A and the photocurrent reaches 1.632 × 10 −4 A. Under the irradiation of 345 µW/cm 2 , the dark current is 1.516 × 10 −9 A and the photocurrent reaches 2.484 × 10 −4 A. Compared with dark current of FTO/TiO 2 UV detector, which is 10 −6 A, the dark current of FTO/TiO 2 /MAPbCl 3 UV detector improved a lot, which changes from 10 −9 A to 10 −12 A. The photo current increased as irradiation increased.
The introduction of TiO 2 /MAPbCl 3 heterojunction plays an important role in lightto-dark current ratio. The only difference between FTO/TiO 2 and FTO/TiO 2 /MAPbCl 3 detectors is the TiO 2 /MAPbCl 3 active layer in the latter. The energy level diagrams and the schematic band diagrams of TiO 2 /MAPbCl 3 heterojunction are shown in Figure 7a [24,34]. A built-in electric field would be formed at the TiO 2 /MAPbCl 3 interface [2,35,36]. The TiO 2 /MAPbCl 3 detector has characteristics similar to Schottky diode. That is, at forward bias, the built-in electric field is weakened and current increases. At reversed bias, the built-in electric field is enhanced and the current reduces. Compared to the TiO 2 detector, the dark current of TiO 2 /MAPbCl 3 detector was reduced due to the built-in electric field of the heterojunction. Furthermore, the TiO 2 /MAPbCl 3 contact interface between horizontal MAPbCl 3 polycrystalline and vertically aligned TiO 2 nanorods array can produce many localized depletion regions, which contribute to lower dark current. What's more, grain boundaries in polycrystalline MAPbCl 3 could scatter carriers in the dark and thus lower mobility. Therefore, the dark current of TiO 2 /MAPbCl 3 heterojunction UV detector is much lower than that of pure TiO 2 UV detector. Under illumination, photogenerated electron-hole pairs can be excited in two materials, which are separated by the built-in electric field at the TiO 2 /MAPbCl 3 interface. Electrons flow along the vertical downward direction of the TiO 2 nanorods, while holes are along the plane direction of the MAPbCl 3 film, so it is difficult for recombination and the collection efficiency of carriers improves. Halogen interstitial defects or MA on halogen antisite defects in polycrystalline MAPbCl 3 create deep level defects that can trap holes and can be recombination centers affected by the nonequilibrium carriers [37]. Therefore, under illumination, compared with a pure TiO 2 UV detector, TiO 2 /MAPbCl 3 UV detector produces more photogenerated carriers [38,39].  Figure 6 shows I-V characteristics of FTO/TiO2/MAPbCl3 UV detector in dark and at a wavelength of 360 nm UV light with various light intensity. All data were obtained at −2 V bias. Under the irradiation of 120 μW/cm 2 , the dark current is 3.69 × 10 −12 A and the photocurrent reaches 3.621 × 10 −6 A. Under the irradiation of 185 μW/cm 2 , the dark current is 2.27 × 10 −9 A and the photocurrent reaches 1.37 × 10 −5 A. Under the irradiation of 255 μW/cm 2 , the dark current is 2.69 × 10 −10 A and the photocurrent reaches 1.632 × 10 −4 A. Under the irradiation of 345 μW/cm 2 , the dark current is 1.516 × 10 −9 A and the photocurrent reaches 2.484 × 10 −4 A. Compared with dark current of FTO/TiO2 UV detector, which is 10 −6 A, the dark current of FTO/TiO2/MAPbCl3 UV detector improved a lot, which changes from 10 −9 A to 10 −12 A. The photo current increased as irradiation increased.  The introduction of TiO2/MAPbCl3 heterojunction plays an important role in lig to-dark current ratio. The only difference between FTO/TiO2 and FTO/TiO2/MAPbCl3 tectors is the TiO2/MAPbCl3 active layer in the latter. The energy level diagrams and schematic band diagrams of TiO2/MAPbCl3 heterojunction are shown in Figure 7a [24,3 A built-in electric field would be formed at the TiO2/MAPbCl3 interface [2,35,36]. T TiO2/MAPbCl3 detector has characteristics similar to Schottky diode. That is, at forwa bias, the built-in electric field is weakened and current increases. At reversed bias, built-in electric field is enhanced and the current reduces. Compared to the TiO2 detect the dark current of TiO2/MAPbCl3 detector was reduced due to the built-in electric fi of the heterojunction. Furthermore, the TiO2/MAPbCl3 contact interface between horizo tal MAPbCl3 polycrystalline and vertically aligned TiO2 nanorods array can produ many localized depletion regions, which contribute to lower dark current. What's mo grain boundaries in polycrystalline MAPbCl3 could scatter carriers in the dark and th lower mobility. Therefore, the dark current of TiO2/MAPbCl3 heterojunction UV detec is much lower than that of pure TiO2 UV detector. Under illumination, photogenera electron-hole pairs can be excited in two materials, which are separated by the built electric field at the TiO2/MAPbCl3 interface. Electrons flow along the vertical downwa direction of the TiO2 nanorods, while holes are along the plane direction of the MAPb film, so it is difficult for recombination and the collection efficiency of carriers improv Halogen interstitial defects or MA on halogen antisite defects in polycrystalline MAPb create deep level defects that can trap holes and can be recombination centers affected the nonequilibrium carriers [37]. Therefore, under illumination, compared with a pu TiO2 UV detector, TiO2/MAPbCl3 UV detector produces more photogenerated carri [38,39].    Figure 7b displays the response and recovery characteristics of the detectors at −2 V bias, obtained from measuring the voltage variation of a 1 MΩ load resistance in the test circuit. The rise times of the TiO 2 detector and TiO 2 /MAPbCl 3 detector are 1.85 s and 0.48 s, respectively, and the fall times are 1.92 s and 2.93 s, respectively. The reason why there is an improvement in response characteristic is that the photogenerated electron-hole pairs are rapidly separated by the built-in electric fields of many localized heterojunction regions, as mentioned above. As for the TiO 2 /MAPbCl 3 photodetector, more photogenerated carriers make the recovery process slower.
Another important parameter for UV photodetector is spectral responsivity and Detectivity* under the irradiation of monochromatic UV light ranging from 310 nm to 450 nm. The spectral responsivity and Detectivity* of both detectors at −2 V bias are shown in Figure 8. The responsivity R was calculated by [32]: in which E is the incident optical power, A is the effective photosensitive area of detector and I p is the photocurrent of the detector under irradiation of the corresponding incident light. The Detectivity* D * , which demonstrates the ability to detect weak signals from a noise environment, is calculated by [40]: where e is the electronic charge constant, and I d , which contributes to background noise, is dark current at −2 V bias. Both TiO 2 and TiO 2 /MAPbCl 3 detectors exhibit well spectrum selectivity for 310~450 nm and R and D * have increased when compared with the pure TiO 2 device. The response peak of TiO 2 /MAPbCl 3 detector is 17.25 A/W at 360 nm, and the corresponding D * is 9.2094 × 10 11 Jones, which is higher than 15.5 A/W and the corresponding D * 1.097 × 10 10 Jones of pure TiO 2 detector at 350 nm, respectively.
in which is the incident optical power, is the effective photosensitive area of de tor and is the photocurrent of the detector under irradiation of the corresponding cident light. The Detectivity* * , which demonstrates the ability to detect weak sig from a noise environment, is calculated by [40]: *

/
where is the electronic charge constant, and , which contributes to background no is dark current at −2 V bias. Both TiO2 and TiO2/MAPbCl3 detectors exhibit well spect selectivity for 310~450 nm and and * have increased when compared with the p TiO2 device. The response peak of TiO2/MAPbCl3 detector is 17.25 A/W at 360 nm, and corresponding * is 9.2094 × 10 11 Jones, which is higher than 15.5 A/W and the co sponding * 1.097 × 10 10 Jones of pure TiO2 detector at 350 nm, respectively. The gain G represents the number of detected charge carriers per single incident p ton, and is given by [41]: where is quantum efficiency and ℎ is the excitation energy. Taking TiO2/MAP detector for example, at −2 V bias and at 360 nm UV light, the photocurrent is 1.632 × A and the irradiation is 255 μW/cm 2 . The energy of each photon is 1240/360 = 3.44 eV. incident light energy is 255 × 0.25 × 10 −2 = 0.6375 μW. The number of photons per sec The gain G represents the number of detected charge carriers per single incident photon, and is given by [41]: where η is quantum efficiency and hυ is the excitation energy. Taking TiO 2 /MAPbCl 3 detector for example, at −2 V bias and at 360 nm UV light, the photocurrent is 1.632 × 10 −4 A Assuming that all photons are absorbed by the semiconductor surface, the photocurrent generated by these photons is supposed to be 1.157 × 10 12 × 1.602 × 10 −19 = 1.853 × 10 −7 A, so the gain of TiO 2 /MAPbCl 3 detector is 1.632 × 10 −4 /1.853 × 10 −7 = 881. The gain of TiO 2 detector is 963. Both of these detectors have a large gain. Under illumination, the defects of the metal/semiconductor interface will act as minority traps, thus mirroring an equal amount of opposite charges inside the semiconductor and reducing the Schottky barrier height. Therefore, more carriers pass through the barrier and obtain high photocurrent and responsivity, which leads to a large gain. The reason why the gain of TiO 2 /MAPbCl 3 heterojunction is less than pure TiO 2 is because the detecting wavelength (360 nm) and the corresponding irradiation power (255 µW/cm 2 ) of TiO 2 /MAPbCl 3 is larger than that of TiO 2 (wavelength (350 nm), the corresponding irradiation power (200 µW/cm 2 )). After being calculated according to Formula (4) above, the gain of heterojunction is lower. The statistical results of the photodetectors' performance parameters are given in Figure 9. The photo and dark currents at −2 V bias of TiO 2 and TiO 2 /MAPbCl 3 photodetectors are shown in box plots. We have fabricated about eight TiO 2 detectors, and two detectors have good performance like the results above. For example, the ratio of light to dark current at −2 V bias is bigger than two orders of magnitude, and the ratio of others is smaller towards one order of magnitude. The smallest dark current at −2 V bias is close to 10 −6 A. The reason why the performance of TiO 2 detectors is not stable is due to fabrication process of the devices. In this work, the FTO substrate we used is 1.5 cm × 4 cm so as to be put within the 25 mL autoclave and be faced down. Then TiO 2 nanorods array film can grow on the conductive layer of FTO during a hydrothermal process. Usually, this film covers all the area of the conductive layer, and there are lots of defects in TiO 2 nanorods film, which leads to relatively high dark and photo currents. If the TiO 2 film didn't form well on FTO substrate, the performance of UV detector would be poor. The statistical results of the photodetectors' performance parameters are given in Figure 9. The photo and dark currents at −2 V bias of TiO2 and TiO2/MAPbCl3 photodetectors are shown in box plots. We have fabricated about eight TiO2 detectors, and two detectors have good performance like the results above. For example, the ratio of light to dark current at −2 V bias is bigger than two orders of magnitude, and the ratio of others is smaller towards one order of magnitude. The smallest dark current at −2 V bias is close to 10 −6 A. The reason why the performance of TiO2 detectors is not stable is due to fabrication process of the devices. In this work, the FTO substrate we used is 1.5 cm × 4 cm so as to be put within the 25 mL autoclave and be faced down. Then TiO2 nanorods array film can grow on the conductive layer of FTO during a hydrothermal process. Usually, this film covers all the area of the conductive layer, and there are lots of defects in TiO2 nanorods film, which leads to relatively high dark and photo currents. If the TiO2 film didn't form well on FTO substrate, the performance of UV detector would be poor. We have fabricated about fifteen TiO2/MAPbCl3 detectors and four of them have good performance. That is, the ratio of light to dark current at −2 V bias is bigger than six orders of magnitude, and the ratio of others is smaller from four to two orders of magnitude. The smallest dark current at −2 V bias is close to 10 −11 A. Because the perovskite We have fabricated about fifteen TiO 2 /MAPbCl 3 detectors and four of them have good performance. That is, the ratio of light to dark current at −2 V bias is bigger than six orders of magnitude, and the ratio of others is smaller from four to two orders of magnitude. The smallest dark current at −2 V bias is close to 10 −11 A. Because the perovskite MAPbCl 3 might not be stable in the atmosphere (H 2 O, O 2 ), and the fabrication process of MAPbCl 3 might also influence the quality of film, the performance of TiO 2 /MAPbCl 3 detectors is not stable compared to our other device with different structures and fabrication procedure (such as MSM detectors with sol-gel method). We have fabricated many detectors and choose the detectors with relatively good performances. Table 1 lists the dark current and responsive performances comparison between some reported TiO 2 based UV detectors. It is clear that the TiO 2 /MAPbCl 3 -based UV detector shows a high responsivity and detectivity*, a fast response speed and a low dark current.

Conclusions
A high-performance UVA photodetector based on a polycrystalline perovskite MAPbCl 3 /TiO 2 nanorods heterojunction has been fabricated successfully. MAPbCl 3 polycrystalline perovskite film forms a good heterojunction with TiO 2 one-dimensional nanorods by onestep spin-coating method with antisolvent-assisting. The special structure makes the TiO 2 /MAPbCl 3 contact interface produce plenty of localized depletion regions. Responsivity and response properties were also improved. Therefore, we successfully improved the performance of pure TiO 2 UV photodetector by introducing a TiO 2 /MAPbCl 3 heterojunction. These results indicate that the TiO 2 /MAPbCl 3 heterojunction detector is a potential candidate for UV detection.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.