Near‐Infrared and Short‐Wavelength Infrared Photodiodes Based on Dye–Perovskite Composites

Organohalide perovskites have emerged as promising light‐sensing materials because of their superior optoelectronic properties and low‐cost processing methods. Recently, perovskite‐based photodetectors have successfully been demonstrated as both broadband and narrowband varieties. However, the photodetection bandwidth in perovskite‐based photodetectors has so far been limited to the near‐infrared regime owing to the relatively wide band gap of hybrid organohalide perovskites. In particular, short‐wavelength infrared photodiodes operating beyond 1 µm have not yet been realized with organohalide perovskites. In this study, narrow band gap organic dyes are combined with hybrid perovskites to form composite films as active photoresponsive layers. Tuning the dye loading allows for optimization of the spectral response characteristics and excellent charge‐carrier mobilities near 11 cm2 V−1 s−1, suggesting that these composites combine the light‐absorbing properties or IR dyes with the outstanding charge‐extraction characteristics of the perovskite. This study demonstrates the first perovskite photodiodes with deep near‐infrared and short‐wavelength infrared response that extends as far as 1.6 µm. All devices are solution‐processed and exhibit relatively high responsivity, low dark current, and fast response at room temperature, making this approach highly attractive for next‐generation light‐detection techniques.


Experimental Section
Preparation of perovskite and dye precursors: (1) Formamidinium iodide synthesis: HC(NH2)2I (Formamidinium iodide, FAI) was synthesized by dissolving formamidine acetate (Sigma-Aldrich) powder in a 1.5 molar excess of hydroiodic acid (HI; Sigma-Aldrich), 57 wt% in H2O. After addition of acid, the solution was left stirring for 10 minutes at 50°C. Upon drying at 100°C for approximately 2h, a yellow-white powder was formed. This powder was then washed three times with diethyl ether to remove excess I2. The powder was later dissolved in ethanol heated at 80°C to obtain a supersaturated solution. Once fully dissolved, the solution was then placed in a refrigerator for overnight recrystallization. The recrystallization process resulted in formation of white flake-like crystals. The FAI flakes were later washed with diethyl ether three times. Finally, the FAI flakes were dried overnight in a vacuum oven at 50°C.
(3) CyPF6 synthesis: CyPF6 was prepared according to previously reported methods. [S1] Equimolar amounts of 2- (CyI) and sodium hexafluorophosphate (Sigma Aldrich) were dissolved in 5:1 MeOH:DCM and stirred for 15 minutes at room temperature. The crude product was collected using vacuum filtration with methyl hydroxideto wash and re-dissolved in dichloromethane (DCM) to filter through a plug of silica gel with DCM eluent to filter out unreacted precursors. Product purity was performed using a Waters Xevo G2-XS QToF mass spectrometer coupled to a Waters Acquity ultrahigh pressure liquid chromatography (LC) system in both positive and negative modes and found to be >95%.
(4) Preparation of organic dye-perovskite mixed solution: indolium tetrafluoroborate (Cy1BF4, FEW Chemicals) was dissolved in DMF with a concentration of 30 mg mL -1 . The solution was made in air and sonicated for 30 min at room temperature. The obtained dye solution was directly mixed with 1.4 M perovskite precursor with various volume ratio, e.g., 1:1, 2:1, 3:1 and 4:1.

Device fabrication
(1) Hole transport layer: The received PEDOT:PSS solution (Clevios P VP.Al4083) was first diluted with methanol (1:2, v:v). The films were spin-coated in air on a pre-cleaned indium tin oxide (ITO) substrate at a spin-coating speed of 4000 rpm for 40s. The films were then annealing at 150 °C for 10 min.
(2) Dye-perovskite active layer: The dye-perovskite precursor solutions were mixed with desired ratios right before disposition, and were deposited through a two-step spin coating program (10 s at 1000 rpm and 32 s at 6000 rpm) with dripping of chlorobenzene as anti-solvent during the second step, 8 s before the end. All the films were annealed at 100 °C for 2 min.
(3) Electron transport layer: [6,6]-phenyl-C61-butyric acid methyl ester (PC61BM, SOLENNEBV) was dissolved in chlorobenzene to obtain 20 mg/mL solution in a glovebox. The PCBM solution was stirred at 4 50 °C overnight and filtered before use with 0.22 μm polytetrafluoroethylene (PTFE) filter to remove possibly present large particles. The PC61BM layer was spin-coated on top of perovskite films at a speed of 1500 rpm for 40 s.
(3) Electrode deposition: 80 nm silver electrodes were thermally evaporated under vacuum of ~5×10 -6 Torr, at a rate of ~0.1 nm s -1 . Note that the temperature of the vacuum chamber was controlled under 40 °C during the evaporation of metal electrodes, as a higher temperature will cause possible degradation of the perovskite films.
Characterization: All the thin film samples used for spectroscopy characterization were spincoated on z-cut quartz substrates with similar spin-coating parameters as described above for device fabrication. UV-vis-NIR and short-wavelength infrared absorption spectra were collected using a Bruker Vertex 80v Fourier transform infrared (FTIR) spectrometer fitted with a reflection/transmission accessory. PL spectra were collected with an intensified charge coupled device (iCCD, PI-MAX4, Princeton Instruments), with each sample photoexcited by a 400 nm picosecond pulsed diode laser (PicoHarp, LDH-D-C-405M). Time-resolved photoluminescence was collected by means of time-correlated single photon counting (TCSPC, PicoHarp300). The obtained PL decay curves were fitted with a single exponential function. The crystallinity of the films was characterized by X-ray Diffraction (XRD). All XRD Spectra were obtained on a Rigaku SmartLab X-ray diffractometer with CuK 1 (1.54060 Å) and a HyPix-3000 2D hybrid pixel array detector, and operated at 40 kV with a 2 scan range of 10~80 o . The surface morphology of the perovskite films was imaged using a SEM (Hitachi S-4300) at an accelerating voltage of 4 kV. Film thicknesses were determined using a surface profilometer (Veeco Dektak 150).
The optical pump/THz probe measurements utilized the output of an amplified laser system (Spectra Physics -Tsunami, Empower, Spitfire), which had a center wavelength of 800 nm, repetition rate of 1 kHz, and pulse duration of 35 fs. THz radiation was generated using 5 optical rectification in a 450-µm thick GaP(110) crystal and detected via free-space electrooptic sampling in a ZnTe chip with 0.2 mm ZnTe(110) on 3 mm ZnTe(100). The 400-nm pump beam was generated by focusing the 800-nm laser beam onto a beta-barium borate (BBO) crystal. To record time-dependent decay traces, the change in peak THz amplitude was measured as a function of pump-probe delay time. In conjunction with the THz measurements, in-situ PL spectra were collected using a UV-visible mini spectrometer (Ocean Optics, USB200+) fitted with an optical fiber, collimating lens, and 435-nm long-pass filter.
All measurements were taken under vacuum using thin films deposited on z-cut quartz substrates.

Charge-carrier mobility calculation:
The charge carrier mobility μ is given by where ΔS is the sheet conductivity of the perovskite thin film, Aeff is the effective area of the overlap of optical pump and THz probe pulse, N is the number of photoexcited charge carriers, and e is the elementary charge. 6 Assuming that the film thickness is much smaller than the THz wavelength, the sheet photoconductivity ΔS of a thin film between two media of refractive indices, nA and nB, can be expressed as [S2,S3] where ΔT/T is the experimentally determined change in transmitted THz electric field amplitude. In our experiment, nA = 1 for vacuum and nB = 2.13 for the z-cut quartz substrate.
The number of photo-excited charge carriers N can be determined using the following equation: where E is the energy contained in an optical excitation pulse of wavelength λ, Rpump is the reflectivity of the sample at normal incidence of the excitation beam, Tpump transmittance of the pump beam, and φ is the ratio of free charge carriers created per photons absorbed (the photon-to-charge branching ratio).
Substituting Equations S2 and S3 into Equation S1, the following equation is obtained: Because 0 ≤ φ ≤ 1, the effective mobility φµ represents a lower limit, which becomes identical to the actual mobility for full photon to free carrier conversion. To allow accurate determination of φμ, we ensured that excitation conditions were in the linear regime. It should also be noted that the determined charge carrier mobility arises from the contributions of both electrons and holes and that these contributions cannot be separated.

Fits to THz photoconductivity transients:
The overall recombination dynamics can be described by the following equation: where n is the charge-carrier density, k1 is the monomolecular rate constant, k2 is the bimolecular rate constant, and k3 is the Auger rate constant.
As the experimentally observed quantity in optical pump-THz probe measurements, , is proportional to the photoconductivity, it is also proportional to the carrier density.
The photon-to-charge branching ratio  indicates the fraction of absorbed photons which are converted to charge carriers. The proportionality factor C = n0 / x0 is the ratio of the absorbed photon density n0 to the initial THz response at time zero The absorbed photon density is a function of the absorption coefficient α and reflectance  and A2 is determined via a global fit to a fluence dependent set of THz transients. The 8 coefficient A3 was set to zero as the maximum charge-carrier density achievable experimentally was insufficiently high to obtain a reliable value. Similar to the mobility calculation, we can only determine the values φk2 from our fits. These equal the actual decay rate constants k3, k2 and k1 in case the material exhibits full photon-to-free-charge conversion and are lower limits when φ < 1.
To account for the spatially varying charge density profile, the fit routine takes into account the exponential charge density profile created by the pump beam by dividing the sample into 30 equally thick slabs and computing the decay function for all of these individually.