Improvement of the perovskite photodiodes performance via advanced interface engineering with polymer dielectric

In this research, we developed a novel modification of the microcrystalline perovskite absorber (Cs_0.2FA_0.8PbI_2.93Cl_0.07) interfaces by utilizing P(VDF – TrFe). The copolymers of this material possess a combination of low dielectric constants with relatively high glass transition temperature, transparency, structural flexibility, ease of solution processing, and chemical stability ⁴⁹. We employed ultrathin P(VDF-TrFE) dielectric interlayers at the grain boundaries within the perovskite film and at the absorber/electron transport layer interface. The schematics of HP based photodiodes (PDs) developed in this work and molecular structure of P(VDF-TrFE) presented in the Fig.1.

Fig. 1  Photodiode scheme a with P(VDF-TrFE) structure visualization and formula b.

The devices were fabricated with a p-i-n architecture: glass (1.1 mm)/ITO (anode, 330 nm)/NiO (p-type, 30 nm)/perovskite absorber Cs_0.2FA_0.8PbI_2.93Cl_0.07 (450 nm)/PCBM (n-type, 30 nm)/BCP (hole blocking interlayer, 10 nm)/Copper (cathode, 100 nm). The solubility of P(VDF-TrFE) in solvents like dimethylformamide and N-methyl pyrrolidone allowed its integration into the perovskite solution for subsequent deposition and crystallization. Through this technological approach, we achieved a bulk distribution of the dielectric within the microcrystalline absorber. P(VDF-TrFE) deposition on the back surface of Cs_0.2FA_0.8PbI_2.93Cl_0.07, was also carried out by solution-processing. A detailed description of the experimental section presented in the Electronic Supplement Information file (ESI). Fabrication of the different types of samples was associated with modifying the perovskite absorber interfaces. To simplify the titles of the different configurations, we designated the bare perovskite films as “control”. The sample with an P(VDF-TrFE) distributed into the bulk of Cs_0.2FA_0.8PbI_2.93Cl_0.07 was designated as “Bulk”, and the sample with dielectric incorporated to the bulk and electron collection junction was referred as “Bulk/n-side”.

We examined the optical properties of perovskite layers with dielectric interlayers using Tauc plots and photoluminescence spectra (Fig. S1 in ESI). The calculation was realized using equations S1 (ESI). The analysis of absorption properties analyses for thin-films demonstrated edge shifts (Fig. S1a, ESI). The band-gap energy (E_g) was extracted using linear approximation, revealing that the E_g of the control sample was 1.573 eV. Integrating dielectric interlayers led to a decrease of the E_g. The band-gap value for Bulk sample was 1.571 eV, while for Bulk/n-side the E_g reduced to 1.570 eV. The analysis of peak positions in the photoluminescence spectra (Fig. S1b, ESI) correlated with the absorption data. Thin films containing dielectric interlayers exhibited a ‘red shift.’ The peak positions were 1.592 eV (779 nm) for the control, 1.583 eV (783 nm) for Bulk, and 1.581 eV (784 nm) for Bulk/n-side.

The influence of P(VDF-TrFE) on perovskite films’ morphology and phase composition was investigated through AFM (Fig. 2a-c) and XRD (Fig. 2d). Analysis revealed that bare perovskite absorber and P(VDF-TrFE) containing samples predominantly exhibited the β-Cs_0.2FA_0.8PbI_2.93Cl_0.07 phase, characterized by specific peaks at 20.6°, 29.3°, 36.14°, 41.9°, and 47.15°. Furthermore, distinct peaks corresponding to the ITO substrate were observed, along with peaks signifying lead iodide at 18.5° (001) and 39.2° (101). No signs of peak shifting or broadening were detected. So, integration of the P(VDF-TrFE) hadn’t discernible quantitative or qualitative effect on the phase composition of the perovskite layer. To examine the impact of polymers on the optoelectronic properties of perovskite, KPFM measurements were conducted using an NT-MDT Ntegra AFM equipped with a NSG03/Au tip. The work function (W_f, Fig. 2e) was determined with a fresh HOPG surface serving as the energy baseline (see details of the calculations in the Eq. S2, S3 in ESI). The Fermi energy of the control sample is 4.85 eV. Incorporating PVDF into the perovskite volume shifts the Fermi level to 4.16 eV. Applying PVDF to the bulk sample increased the work function from 4.16 eV to 4.28 eV. Both bulk and bulk/n-side samples displayed a significant decrease in work function, up to 0.7 eV for the bulk sample, compared to the control perovskite film.

Fig. 2  Morphology a-c, X-Ray diffraction d and surface work function of perovskite films e; scale bar 500 nm.

The incorporation of P(VDF-TrFE) into the bulk and at the interfaces of perovskite can induce the reconfiguration of the trap states. P(VDF-TrFE), a ferroelectric polymer, has been shown to influence the electronic properties of perovskite materials⁵⁰ by reconfiguring intrinsic defects and trap states, leading to improved charge separation and reduced recombination. The shift in work function from 4.85 eV (intrinsic conditions) to 4.28 eV for the Bulk/n-side sample indicates an alignment favorable for electron transport, which could be attributed to the modification of trap states at the interface and within the bulk of the perovskite. The observed Fermi level pinning (FLP)^51–53 effect may be caused by the ferroelectric properties of P(VDF-TrFE), potentially through the compensation of charged states⁵⁴. The enhancement of local fields and the formation of dipoles at intergranular boundaries can screen or compensate for states induced by ionic defects, whose concentration in Cs_0.2FA_0.8PbI_2.93Cl_0.07 thin-films can reach up to 10¹⁴ cm^{−3 55}. The work-function of thin-films based on hybrid perovskites is highly sensitive to the influence of defects and doping additives. Even small concentrations of nanomaterials introduced during solution processing can alter the work function by ~0.1 eV. In our previous work on modifying perovskite absorbers with functionalized titanium carbides (Ti₃C₂)^56,57, we found that solution-processed nanomaterials at a concentration of 10⁻² mg/mL could shift the perovskite energy levels by approximately 0.4 eV, mainly due to the influence of terminal groups (O⁻, Cl⁻, F⁻)⁵⁸.

To evaluate the recombination dynamics in thin films with dielectric interlayers, we used high time-resolved luminescence (TRPL) measurements (Fig. S2 in ESI and Table 1). Photo-carrier lifetime values were extracted by fitting eq.: I(t)=A₁exp(-x/τ₁)+A₂exp(-x/τ₂) (double exponent fit). For relevant data, we collected batch statistics from at least six samples for the control, bulk, and bulk/n-side samples. The median lifetime (τ) of the control CsFAPbI₃ was 72.6 ns. The integration of P(VDF-TrFE) into the intergranular boundaries minimally changed τ to 74.6 ns. For the Bulk/n-side configuration, a more substantial increase in τ to 82.6 ns (+14% compared to the control) was observed. Interpreting TRPL data is a complex task. For the fast component, the τ₁ value decreased from 8.9 ns to 6.9 ns with P(VDF-TrFE) integration, potentially indicating a reduced contribution of trap-assisted recombination^59,60. For the slow component, τ₂, which may be influenced by exciton states^61,62, we observed inverse correlations. We assume that this trend could result from the local polarization of P(VDF-TrFe) during the photo-injection of electron-hole pairs and the resulting accumulation effects in the enhanced electric field.

To investigate the impact of the interface modification on diode properties, we measured the dark volt-ampere characteristics (JVs) for the fabricated PPDs with bias range from -0.1 V to +1.1 V, as shown in Fig. 3. The dark JV curves of the fabricated photodiodes exhibited typical diode behavior, characterized by relevant rectification. Generally, the dark JV curve can be divided into four main regions related to the shunt current (I), recombination current (II), diffusion current (III), and contact resistance (IV). We observed that integrating the dielectric both within the bulk and at the n-type interface altered the diode properties, indicating notable changes in the charge carrier transport. The minimum value of the dark current density (J_min) measured at zero bias was 1 × 10⁻⁷ A/cm² for the reference device. In contrast, the values for the Bulk and Bulk/n-side configurations were significantly reduced by an order of magnitude, up to 2 × 10⁻⁸ A/cm² and 7 × 10⁻¹⁰ A/cm², respectively. The use of P(VDF-TrFE) further contributed to reducing the dark leakage current (J_leakage, bias= −0.1 V). The control and bulk PPDs exhibited comparable J_leakage values ~2 × 10⁻⁶ A/cm², whereas the bulk/n-side configuration demonstrated a significant reduction to 9.7 × 10⁻⁸ A/cm². The box—charts for data of J_min for PPDs presented in the Fig. S3 (ESI), which represents clear correlation between J_min value and integration of P(VDF-TrFE). Calculating the shunt resistance (R_sh) values showed an increase for devices with integrated dielectrics at the interfaces. The reference device had an R_sh = 4.2 kOhm·cm², which increased to 6.8 kOhm·cm² for the bulk configuration and reached 1.2 MOhm·cm² for the Bulk/n-side device. P(VDF-TrFE) was used to modify the perovskite layer interfaces at low solution concentrations (~10⁻³ mg/mL). Relatively low concentration was selected to avoid the negative impact that higher P(VDF-TrFE) concentrations had on the diode properties of the devices (Fig. S4, ESI). The increase of the concentration to 10⁻¹ mg/mL resulted in decrease of R_sh from ~10⁴ Ohm∙cm² (P(VDF-TrFE) concentration 10⁻³ mg/mL) to ~10³ Ohm∙cm².

Fig. 3  Dark JV curves of p-i-n structures of the PPDs with dielectric interlayers.

For a comprehensive analysis of the changes in the dark JV curve in the II and III regions, we performed a fitting using a 2-diode model to extract the appropriate parameters (Table 2 with values of the non-ideality factors (m₁, m₂); reverse saturation currents (J₀₁, J₀₂), and shunt resistance).

Generally, the total saturation dark current (J₀) is governed by recombination processes within the solar cell and can be utilized to estimate the efficiency of charge carrier transport. In hetero-structured devices, J₀ depends on the number and quality of interfaces. In our study, we modified the interfaces within the perovskite absorber volume and at the junction for electron collection. Calculations for dark JV were performed according to the equations S4-S11 (ESI), the equivalent circuit of the 2-diode model presented in ESI (Fig. S5, ESI). The calculated values of J₀ indicated that the overall value is primarily determined by J₀₂. The significant reduction of J₀ from 10⁻⁴ to 10⁻⁸ A/cm² in devices with P(VDF-TrFE) integration compared to control highlights the reduced contribution of recombination to charge carrier transport. The non-ideality factor of a photodiode is typically used to analyze the dominant recombination mechanisms involving free carriers and traps both in the volume (bulk) and at interfaces. Interpreting the calculated values of the non-ideality factor for photovoltaic semiconductor devices is a complex task and varies greatly depending on the photodiode architecture. In a classical p-n junction, the non-ideality factor (m) equals 1 when diffusion current predominates over recombination current. The parameter can reach a value of 2 when recombination current dominates in the space charge region. A lower non-ideality factor indicates reduced recombination dynamics and increased efficiency of the junction for charge carrier transport. A lower non-ideality factor points to reduced recombination dynamics and increased junction efficiency for charge carrier transport. P-i-n diode devices based on halide perovskites are more complex, involving two hetero-junctions. Therefore, recombination processes in PPDs are not described by the m range of 1-2 and require modification of standard equivalent electrical circuits. In our work, we used a 2-diode model with diodes in series, calculating the non-ideality factor as the sum of the values of each diode. For the control sample, the calculated non-ideality factor was 2.811. In the Bulk configuration, the value significantly decreased to 1.792, demonstrating notable changes in recombination processes associated with the interfaces of the microcrystalline perovskite film. Interestingly, the non-ideality factor for the Bulk/n-type configuration, while lower than the control at 2.610, exceeded the Bulk device values. Potential losses in charge carrier transport efficiency could be attributed to the non-optimized P(VDF-TrFE) deposition process on the surface.

The photo-response of PPDs in short-circuit and open-circuit modes was assessed over a wide range of illumination power densities (P₀, 540 nm, LED source, range: 10⁻³ to 10¹ mW/cm², Fig. 4). For the short-circuit current (J_sc), linear trends were observed, as the photocurrent ∝ charge generation rate. The calculation of the responsivity (R, eq.S12 in ESI), representing photocurrent dependence on P₀ at a wavelength of 540 nm, resulted in values of 0.32 A/W for the control device, 0.38 A/W for the Bulk, and 0.44 A/W for the Bulk/n-side configuration. The dependence of open-circuit voltage (V_oc) vs. P₀ exhibited a logarithmic nature. The Bulk/n-side modification of the device increased V_oc across the full range of light intensities by 0.02 to 0.05 V compared to control PPDs. On the other hand, Bulk devices showed the highest V_oc values at low-light intensities (<10⁻¹ mW/cm²), indicating changes in the separation and collection of the charge carriers.

Fig. 4  The linearity plot of J_sc to P₀ for the various perovskites modifications on a logarithmic scale a; the dependence of V_oc vs. P₀ on a semi-logarithmic scale b.

For the illumination power density range used, we calculated the linear dynamic range (LDR, eq. S13 in ESI), which defines the range of incident light power over which the photodiode output current depends linearly on the input optical power. This critical characteristic determines the photodiode’s ability to accurately measure light intensity over a wide range of power levels. Calculations of LDR values under 540 nm LED illumination indicated an increase from 76.6 dB to 99.5 dB for PPDs with P(VDF-TrFE) interlayers (Table 3). This originated from the reduction in dark current and the improvement in the signal-to-noise ratio for the Bulk and Bulk/n-side configurations. Across the P₀ range from 10⁻³ to 10¹ mW/cm², we observed no deviation from linearity in the photocurrent, suggesting that the LDR values for the designed PPDs may potentially exceed 100 dB. The spectral performance of the devices was estimated via measurements of the external quantum efficiency (Fig. S6 in ESI). All PPD configurations exhibited a high level of photoelectric conversion, between 82% and 87%, in the visible part of the spectrum. This correlates with the data obtained under high optical pumping conditions (540 nm).

Based on the assumption that shot noise is dominant in PPDs, we estimated the key figures of merit parameters—specific detectivity (D*) and noise equivalent power (NEP)—for the short-circuit mode (zero bias). The data calculated using Eq. S14, S15 (ESI) are presented in Table 3. Reducing current leakage in devices with modified interfaces increased D* by almost an order of magnitude, approaching the value of 10¹² Jones for the Bulk/n-side configuration. Strengthening the shunt properties and increasing R led to a reduction in the NEP for the P(VDF-TrFE)-containing devices. While the NEP for the control sample was around ~10⁻¹² W·Hz^−1/2, we obtained 6.6 × 10⁻¹³ and 4.00 × 10⁻¹³ W·Hz^−1/2 for the Bulk and Bulk/n-side configurations, respectively.

We compared the performance of PPDs from our work with state-of-the-art devices (Table S1 in ESI). The D* parameters for the Bulk and Bulk/n-side configurations are relevant when compared to other perovskite photodiode modifications, where the typical range is between 10¹⁰ and 10¹⁴ Jones. Moreover, our work demonstrates competitive values for NEP and response time.

Perovskite devices with diode structure are known to exhibit hysteresis^19,63,64 caused by ion migration and accumulation at the interfaces, resulting in dynamic capacitance changes, gating, and zone bending. We conducted hysteresis measurements under monochromatic LED illumination (540 nm, 0.62 mW/cm²) and cycled IV scans from open-circuit to 0 V (short-circuit). Five cycles were performed for each device type, with the IV curves shown in (Fig. S7, ESI). The hysteresis index (HI, Table S2, ESI), was calculated using formula S16 (ESI), was used to assess the effect by shifts of the diode’s maximum power point. Notably, for control devices we obtained enhanced shifts of the working point (maximum power) and V_oc, while Bulk and Bulk/n-side devices these effects were slightly suppressed. For the first cycle of measurements Control device showed the highest value of HI (14%), while Bulk and Bulk/n-side devices exhibited reduced values of 8-11%. During the cycles PPDs showed a similar trend for increase of HI. The integration of ferroelectric material at low concentrations did not contribute to accumulation effects or additional ion migration.

To assess the response dynamics of PPDs, we measured the transient characteristics of ON/OFF modes during optical pumping (Fig. 5). A square pulse of light (LED, 50 kHz) was used to evaluate changes in the rise and fall profiles of the photocurrent over time. Specifically, we analyzed the rise and fall times (t_r and t_f) of the current response at amplitudes corresponding to 10% and 90% of signal saturation. The devices exhibited a characteristic π-shaped waveform. For the control device, the t_r was 6.3 µs, and the t_f was 10.9 µs. Bulk PPDs demonstrated enhanced performance with decreased transient times: t_r = 4.9 µs (−22.6%) and t_f = 6.7 µs (−15.6%). The Bulk/n-side configurations showed the most rapid signal growth profile with t_r = 4.6 µs (−32.6%), while decay was 6.5 µs. The box—charts for data of t_r and t_f for PPDs presented in the Fig. S8 (ESI), which represents clear correlation between time of the response and configuration of the devices.

Fig. 5  Response dynamics in rise a and fall b modes for the control device; PPDs of Bulk and Bulk/n-side configurations.

A crucial parameter of the photodiode related to photo-response dynamics is the cut-off frequency, defined as the frequency at which the photodiode’s response diminishes to 3 dB below its maximum value (f_3dB). This parameter determines the speed at which the photodiode can respond to the relevant changes in the light signal. The 3 dB point corresponds to a power reduction of half of its maximum value or a decline in current to 0.707 from the maximum. Essentially, f_3dB sets the upper limit on the frequency of the light signal that the photodiode can efficiently detect. For f_3dB measurements, the amplitude of the photodiode photocurrent was measured when optically pumped with a square pulse of a green LED (540 nm) at different frequencies of the LED glow. Measurements were carried out for the frequency range from 10 to 90 kHz. The illumination was carried at a power density of 2 mW/cm². The measured data is presented in Fig. 6. Enhanced speed and transport performance in the PPDs, containing P(VDF-TrFE), increased the f_3db from 64.8 kHz for the control device to 71.9 kHz and 76.8 kHz for the Bulk and Bulk/n-side configurations, respectively. To evaluate the stability of the devices, we measured the transient characteristics after 250 hours of thermal stress (70°С) (Fig. S9 in ESI). For all PPDs configurations, the decrease in response speed was negligible. Also, we estimated the stability of the devices by monitoring changes in dark current during prolonged exposure to high humidity conditions (60-80%), as well as during ON/OFF cycling. After 36 million ON/OFF cycles, we observed negligible changes in the rise and fall signals. The storage in the high-humidity conditions also didn’t affected the dynamics of the photo-response (Fig. S10-S12, ESI).

Fig. 6  F_3db bandwidth for the fabricated PPDs with dielectric interlayers.

We utilized Photo-Induced Voltage Spectroscopy (PIVTS)⁵⁵ and Admittance Spectroscopy (AS) to conduct an in-depth evaluation of Fermi level shifts and trap state contributions. The principle of PIVTS involves monitoring V_oc under impulse light pumping (square LED pulses, 470 nm) and analyzing relaxation kinetics during signal decay over a wide temperature range (190-360K). Upon illumination, the perovskite photodiode reaches a steady-state V_oc and saturates after turn-on. Upon turning off the light, non-equilibrium charge carriers recombine over their characteristic lifetimes (typically within nanoseconds to tens of nanoseconds), leading to a decay in V_oc. Perovskite devices exhibit unique behavior due to the slow migration of ionic defects and their associated electric fields^65,66. By analyzing the PIVTS relaxation spectra, we can reveal the kinetics of trapping processes across different temperature intervals. We conducted combined PIVTS and AS measurements for the PPDs, with the data presented in Fig. S13 (ESI).

PIVTS analysis demonstrated that the introduction of P(VDF-TrFE) led to a reconfiguration of defect states. In the Control device, high amplitude signals were observed within the temperature range of 275-320 K, near room temperature and within the typical operating range for photodiodes. The contribution of ionic species can be linked to defect activation energies reported in the literature^67–69. However, identifying the precise nature of these defects is a complex task, as various factors—such as crystallization methods¹⁷ and doping⁷⁰—affect their numerical parameters. In the Control device, a defect center α, with an activation energy of 0.72 eV, was observed. This defect may correspond to the position of the iodine anion in the perovskite structure¹⁶, such as iodine in an interstitial site (I_i) or an antisite with lead (I-Pb)⁶⁹. Similar energies were observed for the Bulk and Bulk/n-side devices, but at reduced temperatures (250-300 K). The amplitude of the α-defect signal was diminished in PPDs containing P(VDF-TrFE). At higher temperatures, a deep β-center signal emerged, which is uncommon for CsFAPbI₃ and might correspond to an anti-site state of the organic cation (A-site)⁷¹. Admittance spectroscopy analysis showed that ionic species concentrations for all PPD configurations were approximately 10¹⁶ cm⁻³. These findings indicate that the introduction of P(VDF-TrFE) reconfigures defect states, reducing the α-center (E_a~0.7-0.8 eV) and introducing β-centers. Given the improved performance of P(VDF-TrFE)-based PPDs, it can be interpreted that α-defects play a critical role for CsFAPbI₃ properties in photodiodes application.

To investigate the potential interactions between PVDF-TrFE and the perovskite material, Fourier-transform infrared spectroscopy (FTIR) studies were conducted to demonstrate the interactions between the fluorinated fragments of the polymer and formamidinium iodide (see Fig. S14, ESI and a brief discussion therein). Previous studies have observed similar interactions between the fluorinated additive and FAI, resulting in a reduction in defects in the resulting formamidinium lead halides⁷².

Our data on stability performance indicated no substantial degradation impact for soft exploitation conditions of PPDs. In this study, we employed double-cation perovskite CsFAPbI₃, which demonstrated relevant stability⁷³ for several thousand hours under conditions of constant light-soaking (100 mW/cm²) and heating (65°C). However, full stabilization of such perovskites compositions and device structures has not yet been confirmed. We analyzed the PIVTS results to evaluate the potential influence of P(VDF-TrFE) on the stabilization of PPDs against intrinsic factors. The degradation of halide perovskite-based devices is largely attributed to the decomposition of the perovskite absorber and the migration of the ionic defects^74,75. Different nature of the defects could trigger specific degradation pathway. For FA-cation perovskites, I_i induces the transition from the cubic photoactive phase to the tetragonal inactive phase and accelerates the formation of neutral I₂ (filling of two traps)⁷⁶. Charged antisites (such as I-Pb) could act as recombination centers and migrate to the metal electrode under electric field^21,77 The accumulation of the ionic defects at the interface typically induces corrosion processes associated with electrochemical oxidation⁷⁸. PIVTS results revealed a decreased amplitude of the α-defect (0.7-0.8 eV) signals for PPDs with P(VDF-TrFE). Thus, we can conclude that the integration of P(VDF-TrFE) mitigates the corrosion dynamics driven by iodine-related defects and could enhance the long-term stabilization of PPDs. The obtained data on defect activation energies suggest that the Fermi level pinning may be caused by complex factors, including point defects related to iodine or organic cations.

This work presents a solution-processed approach for integrating an organic dielectric into the volume and interfaces of a perovskite absorber. The complementary solubility and selected dielectric concentrations have a significant impact on device performance. However, the method is not fully universal, as it requires careful consideration of surface state modifications and energy level alignment specific to different materials. Furthermore, we conducted an additional experiment to integrate P(VDF-co-CTFE)-graft-PEMA dielectric using a process similar to that employed for P(VDF-TrFE). P(VDF-co-CTFE)-g-PEMA represents an analogue of P(VDF-TrFE), comprising grafted chains of polyethyl methacrylate (PEMA). The synthesis of P(VDF-co-CTFE)-g-PEMA has been previously described by our group⁷⁹. P(VDF-CTFE)-g-PEMA has a high melting point of 160°C and a decomposition temperature above 390°C, whether in air or inert atmospheres⁷⁹. It has a permittivity of approximately 7, conductivity around 10⁻¹² S/cm, and a dielectric loss factor of about 10⁷⁹. The incorporation of PEMA chains into the copolymer structure enhances the solubility of PVDF-based polymeric dielectrics in the solvents typically employed for perovskite layer preparation. Additionally, non-covalent interactions, including those between the perovskite and the carbonyl groups of PEMA, may potentially occur as compared to P(VDF-TrFE).

We evaluated the impact of P(VDF-co-CTFE)-g-PEMA integration on the transport and spectral properties of PPDs. Dark JV measurements indicated that P(VDF-co-CTFE)-g-PEMA integration in Bulk and Bulk/n-side configurations weakened shunt properties and increased leakage current, with J_leakage values around ~9x10⁻⁵ A/cm² for the Bulk configuration and rising to ~3x10⁻² A/cm² for Bulk/n-side (Fig. S15a, ESI). The calculated R_sh values for P(VDF-co-CTFE)-g-PEMA remained under 1 kOhm-cm². Despite this degradation in diode properties, the alternative dielectric affected the EQE (Fig. S15b, ESI). Compared to PPDs utilizing P(VDF-TrFE), the alternative configurations exhibited a slight reduction in photoelectric conversion between 300-550 nm, while significantly raising the spectra shoulder in the 750-800 nm region. In perovskite-based p-i-n diodes, high-energy photons are absorbed in the frontal regions near the HTL/absorber interface, with conversion determined by EQE at shorter wavelengths. The increased EQE in the infrared region suggests improved electron collection efficiency, a distinctive feature of P(VDF-co-CTFE)-g-PEMA integration. However, the strong increase in leakage current prevented any notable improvement in output performance (Fig. S16 in ESI) for J_sc vs P_o and V_oc vs P_o dependencies. These approaches reveal critical dielectric properties for modifying perovskite layer interfaces at low concentrations. The molecular design of the dielectric must accommodate energy level alignment, neutralize surface states, and ensure layer continuity after crystallization. Through this comparative experiment between P(VDF-co-CTFE)-g-PEMA and P(VDF-TrFE), we have demonstrated that materials with similar molecular structures can produce significantly different outcomes in diode and spectral performance.

The integration of P(VDF-TrFE) dielectric materials has been demonstrated to be effective for perovskite-based diode structures. In early works ^80,81, researchers utilized P(VDF-TrFE) to enhance the built-in electric field within perovskite absorbers, thereby improving the collection efficiency of the free charge carriers. However, the previously presented approaches relied on poling effects, which potentially initiate ion migration processes under polarization conditions. Polling and migration effects can adversely affect device performance. In our work, we have demonstrated an approach involving complex modification of interfaces within the bulk and at the electron collection junction, wherein P(VDF-TrFE) primarily compensates for charge carrier traps. PPDs-containing incorporating P(VDF-TrFE) in the bulk absorber and at bulk/n-type interfaces operated in regimes similar to reference devices and didn’t require pre-conditioning. We observed a relevant increase in photo-carrier lifetime with the integration of P(VDF-TrFE) on the perovskite film surface, indicating a influence of the dielectric interlayer on recombination processes and charge carrier splitting. A comprehensive analysis of diode transport characteristics also shows a reduced contribution of non-radiative recombination at the interfaces for the devices with dielectric interlayers. P(VDF-TrFE) plays a significant role in enhancing the shunt properties of PPDs and reducing dark currents. The presence of a thin dielectric layer assists in mitigating potential defects in morphology, such as micro-pinholes. This correlates with the measured photo-response of PPDs. The increase in R is linked to better photo-carrier collection efficiency (see Eq. S17 in ESI)⁸². The rise in V_oc in low light intensity regions results from improved energy level alignment and reduced trap-states in the band-gap, which negatively impact quasi-Fermi level splitting⁸³ during photo-injection. Improved dynamic performance is a key result of this work. The photo-response profile directly correlates with the cut-off frequency. Different capacitive effects according to multiple models^19,84,85 describe variations in transition time in perovskite diode architectures. The appearance of slow components in the transient photocurrent (as observed in the PPD control) is typically associated with the accumulation of ionic species at interfaces^86–88, which in a perovskite photodiode can be represented by iodine vacancies, uncompensated organic cation ions, iodine in interstitials, etc. The photodiode junction capacitance, combined with other capacitances in the circuit, forms an RC constant with the load resistance, which limits the device’s bandwidth. The charge-carrier transit time and RC constant determine the cut-off frequency (Eq. 1)^89–91:

Where t_CCT - charge-carrier transit time; R - total series resistance, including the device resistance, contact resistances, and load resistances; C - the sum of the capacitance of the device.

The improvements in dynamic response for the Bulk/n-side configuration result from a complex interplay of suppressed trapping, reduced accumulation effects, and minimized current leakage. Compared to similar studies on integrating dielectrics into the absorber volume and interfaces of a perovskite photodiode^81,92–94, our results demonstrate superior f₃dB and improved transition times.

To demonstrate the practical value of interface engineering using organic dielectrics, we conducted measurements using a setup comprising an X-ray tube (RAD-160 system, tungsten tube) and a scintillator (CsI bar, Fig. 7a). The photoelectric conversion of visible scintillation light by photodiodes is a common application for string and array devices⁹⁵. Non-direct detector systems⁹⁶ operates by integrating photodiodes with scintillators, such as CsI, which emit green light (~520 nm) upon absorbing ionizing radiation⁹⁷. We employed a tungsten tube-based X-ray source with a commercial CsI bar as the scintillator and placed a substrate with photodiode pixels behind it (photo-image of setup presented in Fig. S17, ESI). To assess performance differences, we measured the amplitude of the transient current signal under X-ray exposure at 160 kV tube voltage and 1 mA tube current. The transient signal profiles for the PPDs are displayed in Fig. 7b. The signal buildup over a few milliseconds is attributed to the slow emission buildup of the X-ray tube, while the CsI luminescence time is in the nanosecond range⁹⁸, and the fast PPD performance was measured in microseconds. Analysis of the signal rise profiles from the scintillator light revealed that the photo-response amplitude increased significantly for Bulk and Bulk/n-side samples compared to the Control. The signal-to-noise ratios (SNR) were 27 for Control, 46 for Bulk, and 57 for Bulk/n-side. So, the improvements in the specific detectivity with enhanced diode properties, validate the efficiency of interface engineering in practical applications.

Fig. 7  Photo-image of the PPDs fabricated as a stroke of 10 pixels a; SNR for PPDs measured for conversion of scintillating CsI light induced by X-ray flux (tungsten tube source) b. The signal profiles correspond to the switch-on regime of X-Ray source.

Our method for manufacturing photodiodes can be scaled up to large-area arrays, with dimensions extending to tens of square centimeters, for example, in the development of flat-detector panels⁹⁹. A layer-by-layer coating can be achieved using slot-die cycles¹⁰⁰, a process well-established in solar module fabrication. The patterning of large-area photodiode arrays can be completed with multistep laser scribing¹⁰¹. In the Bulk configuration, P(VDF-TrFE) is combined with absorber deposition/crystallization. In the case of the n-side configuration, up-scaling requires the selection of a complementary solvent that prevents interaction with the absorber, or alternatively, ultrafast deposition followed by rapid drying. We assume that Bulk device configuration with P(VDF-TrFE) could also be effectively utilized in n-i-p device architectures¹⁰² or even tandems¹⁰³ without major up-grades, as the solvents for the dielectric and perovskite solutions are complementary. On the other hand, adapting the n-side modification for n-i-p structures should account for possible changes in the wetting properties of the ETL. F-atom-rich P(VDF-TrFE) on the ETL’s surface could increase the surface wetting angle and result in void formation¹⁰⁴ after absorber crystallization.

In this work, we performed a comprehensive study of P(VDF-TrFE) integration in bulk of the absorber and electron transport surfaces for PPDs. We demonstrated that dielectric incorporation provides Fermi level pinning, shifting it from 4.85 eV for bare CsFAPbI₃ to 4.16-4.28 eV. This adjustment reconfigures the energy level alignment that potentially allows more efficient electron collection. The Bulk/n-side configuration exhibited beneficial improvements in the entire spectra of photodiode performance parameters. This underscores the benefits of a complex modification of both the microcrystalline absorber and hetero boundaries with standard transport layers like C₆₀ fullerenes. Integration at low concentrations compensated charged ionic defects linked to the molecular position of iodine, leading to improved diode properties. Furthermore, the ferroelectric nature of P(VDF-TrFE) doesn’t result in undesirable hysteresis or accumulation effects. The dielectric interface layers reduced the non-ideality factor and dark leakage currents. The responsivity of the fabricated PPDs was 0.32 A/W for the control device, 0.38 A/W for the Bulk, and 0.44 A/W for the Bulk/n-side configuration. Reducing the impact of non-radiative recombination processes at the interfaces allowed to decrease the NEP to ~10⁻¹³ W·Hz^−1/2 and achieve a relevant D* of ~10¹² Jones. Despite the meaningful differences in the photo response (I_sc vs P_o, V_oc vs P₀) between bulk and Bulk/n-side devices, the dynamic response showed close behavior. The cut-off frequency was increased from 64.8 (control) to 71.9 (Bulk) and 76.8 kHz (Bulk/n-side). Photodiodes with f₃dB values between 50-100 kHz are considered for use in applications that don't require extra fast response (GHz range), such as light measurement and sensing in industrial or scientific equipment, optical switches and encoders, low-speed optical communication systems, some types of the optical remote controls. This paper demonstrates the importance of complex interface processing in perovskite photodiodes, with a focus on the unique bulk modification and heterojunction with electron-transport layers. The developed method of using dielectrics combines the efficient tuning of transport properties and fast performance. Our work provides new insights into interface engineering for perovskite optoelectronics, aiming to comprehensively enhance the output performance.

The authors gratefully acknowledge the financial support from Russian Science Foundation with project № 22-19-00812.