Realising ultrafast perovskite photodetectors via 2D synergy for optical communication and sensitive light detection

Abstract

The realization of high-speed optical communication and sensitive light detection with self-powered perovskite photodetectors (PDs) is imperative for advancing AI-driven optoelectronic systems, including visible light communication (VLC), surveillance, and environmental monitoring, necessitating exceptional detectivity and an ultrafast response. Herein, we introduce a 2D synergy strategy employing Ti₃C₂T_x MXene to simultaneously enhance charge carrier transport, suppress defect states, and improve film morphology in perovskite films via percolation networks and surface passivation. The resultant self-powered perovskite PD achieves a dark current density of 1.46 × 10⁻¹⁰ A cm⁻², a detectivity of 6.59 × 10¹² Jones, a linear dynamic range exceeding 181 dB, and ultrafast response times (rise: 1.07 μs, fall: 0.74 μs), alongside robust stability with negligible degradation over continuous cycles. This work demonstrates integration of the optimized PD into an optical communication system, enabling ASCII-encoded signal reception across 460–730 nm and a sensitive light detection setup for low-light applications. Significantly, the optimized PD propels the advent of energy-efficient, high-speed optoelectronics, fostering unprecedented opportunities in AI-driven technologies.

---

1 Introduction

Photodetectors (PDs) significantly influence optoelectronic devices, converting light signals into electrical signals for applications comprising optical communication, image sensing, biosensing, surveillance, neuromorphic devices and environmental monitoring.^2–7 In the era of artificial intelligence (AI), where high-speed data transmission and sophisticated sensing systems are increasingly critical, the demand for advanced PDs with low dark current, high detectivity (D*), broad spectral response, rapid response times, and exceptional stability has become paramount.^8–11 However, conventional PDs often face limitations, including restricted spectral range, slow response times, poor environmental stability, high trap density of states, and low carrier mobility, hindering their suitability for next-generation applications.^12–15 Consequently, a logical transition to designing innovative PDs with enhanced versatility and ultrafast dynamics is imperative for realizing next-generation technologies, particularly in the rapidly evolving field of artificial intelligence.

Halide perovskites have arisen as a transformative class of materials for PDs, driving extensive research work owing to their unparalleled optoelectronic properties, including high absorption coefficients, tunable bandgaps, and long carrier diffusion lengths.^{3,8,12,16–18} Strategies such as additive engineering, post-deposition treatments, advanced film deposition techniques, and using tailored charge transport layer designs have been extensively investigated to elevate perovskite PD performance to meet stringent application demands.^2,16,19–23 Among these, additive engineering is a powerful approach, enhancing light harvesting, charge carrier dynamics, and surface passivation, thereby significantly suppressing non-radiative recombination losses.^24–27 Despite these advancements, practical applications such as optical communication and sensitive light detection still face challenges, including high dark current, low linear dynamic range (LDR), and limited photoresponse.

Recent advancements in device architecture and fabrication techniques have yielded substantial breakthroughs in overcoming these limitations. For instance, a CsPbBr₃-based PD achieved ultralow dark current and high D*, facilitating visible light communication (VLC) with a bit rate of 100 kbps.¹² Similarly, tin-lead perovskite PDs with azobenzene derivatives exhibited fast response speeds in the near-infrared range.¹⁰ Self-powered CsPb₂Br₅–CsPbBr₃ PDs integrated into VLC systems have also shown strong potential for voice command transmission.³¹ Additionally, photomultiplication-type PDs have demonstrated high responsivity and D* for weak-light detection.³⁶ Among next-generation materials, two-dimensional (2D) Mxenes, denoted by the general formula M_n+1X_nT_x (where T_x includes functional groups such as =O, –OH, and –F), have elicited significant interest due to their exceptional conductivity, tunable work function, and transparency.^15,25,37 MXenes have been extensively leveraged in perovskite-based devices, notably in solar cells, serving as conductive electrodes, interfacial layers, and additives to boost efficiency through enhanced charge transport and defect passivation.^{25,26,29,38–42} In contrast, this study pioneers the integration of Ti₃C₂T_x MXene as an additive into perovskite photodetectors, an area yet to be explored, focusing self-powered operation at zero bias to tap into its potential for high-speed, low-power sensing applications. Despite these strides, achieving broadband operation over the 460–730 nm spectral range and securing robust and stable performance for practical uses, such as optical communication and sensitive light detection, still present extensive challenges.

This study presents a self-powered perovskite PD with enhanced performance characteristics enabled by incorporating Ti₃C₂T_x MXene. The device achieved a low dark current density of 1.46 × 10⁻¹⁰ A cm⁻², a high D* of 6.59 × 10¹² Jones, a broad LDR exceeding 181 dB, and ultra-fast response times of 1.07 μs (rise) and 0.74 μs (fall). Therefore, this study aims to demonstrate that efficient charge transport and reduced surface defects in the optimized PD, achieved through Ti₃C₂T_x incorporation, will enable ultra-fast response times suitable for high-speed optical communication and other applications. Furthermore, the PD was proficiently integrated into a custom-built optical communication system for ASCII-based data transmission and a sensitive light detection system for wearable sensors, night vision, and surveillance applications. This study aims to contribute to the advancement of high-speed, stable perovskite PDs for next-generation optoelectronic applications.

2 Results and discussion

2.1 Structural and compositional characterization of photodetectors

The schematic device structure of the PD is shown in Fig. 1a. The hole transport layer (HTL) consists of PTAA, followed by the photoactive layer, which comprises perovskite/Ti₃C₂T_x. Next, the electron transport layer (ETL) is formed by PCBM, with BCP serving as the hole-blocking layer. Initially, it was dispersed in the organic solvent dimethylformamide (DMF), as illustrated in ESI Fig. S1.† Subsequently, varying volume fractions of this Ti₃C₂T_x dispersion were blended into the perovskite precursor solution while preserving the original precursor concentration, as shown in ESI Fig. S2.† Its optimal volume was determined by refining the I_light/I_dark ratio across all fabricated PDs. Following this optimization, all subsequent characterization studies were conducted using devices with this fixed volume unless otherwise specified. The optimized volume concentration of Ti₃C₂T_x, determined to be 20%, was employed in all further calculations, as demonstrated in ESI Fig. S3.† The addition of Ti₃C₂T_x to the perovskite solution prominently enhanced its wettability, facilitating more uniform film formation on the PTAA substrate in both centered and corner regions compared to the pristine perovskite, as verified in ESI Fig. S4.† Incorporating Ti₃C₂T_x enhances the wettability of the perovskite layer due to its hydrophilic surface, layered structure, and –OH/–F functional groups, leading to improved film formation and coverage.^44,45

Fig. 1 (a) PD device configuration incorporating MAPbI₃/Ti₃C₂T_x. (b) XRD patterns of the Ti₃C₂T_x film and perovskite film with/without an optimized amount of the Ti₃C₂T_x additive. (c) Wide scan spectra of MAPbI₃ and MAPbI₃/Ti₃C₂T_x. The high-resolution spectra of (d) Pb 4f, (e) C 1s, and (f) O 1s.

XRD analysis of the Ti₃C₂T_x layer revealed characteristic peaks at 7.4° (d ≈ 11.9 Å), corresponding to the (002) facets, confirming multilayer Ti₃C₂T_x with typical MXene interlayer spacing consistent with earlier findings.²⁶ Broad peaks at 34.7° and 40.4° reflect an in-plane Ti–C lattice with hexagonal symmetry from the MAX phase, though their broadness indicates disordered MXene powders stemming from surface terminations (e.g., –OH, –O, and –F) and etching-induced disruptions. Moreover, XRD patterns of MAPbI₃ films, both with and without Ti₃C₂T_x, exhibited a slight shift of the primary peak at around 14.6° toward a smaller angle in the film containing this additive, as demonstrated in Fig. 1b and S5.† The shift hints at an interaction between Ti₃C₂T_x and perovskite that may enhance film quality. The observed alteration in the main peak, attributed to the stress induced by integrating Ti₃C₂T_x nanosheets into the perovskite, results from homogeneous strain during the growth phase.²⁶ Additionally, slower crystal formation in films with Ti₃C₂T_x, likely due to reduced nucleation sites, combined with the rapid dispersion of 2D Ti₃C₂T_x nanosheets in the perovskite precursor, promoted uniform nucleation and larger perovskite grains, as corroborated by SEM images^47,48 A detailed analysis of grain morphology appears in the later SEM section (Fig. 3a and b).

X-ray photoelectron spectroscopy (XPS) analysis elucidated the elemental composition of Ti₃C₂T_x multilayers. The wide scan in ESI Fig. S6† identified distinct peaks for carbon (C), titanium (Ti), oxygen (O), and fluorine (F). High-resolution XPS spectra for Ti 2p, F 1s, O 1s, and C 1s (Fig. S6(b–e)†) confirmed their presence. The Ti 2p spectrum (Fig. S6b†) revealed peaks at 454.23, 458.07, 460.49, and 464.90 eV, corresponding to Ti, TiO_2−xF_x, Ti–C, and TiO⁴⁺, respectively. Similarly, the F 1s spectrum in Fig. S6c† displayed peaks at 683.51 and 684.82 eV, assigned to Ti–F and C–F bonds. Deconvolution of the O 1s peak in Fig. S6d† identified components at 528.78 and 530.66 eV, characteristic of Ti–O and O–H bonds. Lastly, the C 1s spectrum in Fig. S6e† showed 280.3 and 283.45 eV peaks, corresponding to Ti–C and C–C bonds, respectively.^47,49

XPS analysis was conducted to examine the elemental composition and chemical state of the elements in pristine MAPbI₃ and MAPbI₃/Ti₃C₂T_x films as shown in Fig. 1c. While C, lead (Pb), O, and iodine(I) were identified in both samples, the MAPbI₃/Ti₃C₂T_x spectrum additionally exhibited a Ti peak, confirming the presence of Ti₃C₂T_x. High-resolution XPS analysis detected a slight positive shift in the binding energy (BE) of the Pb 4f, C 1s, and O 1s peaks for the MAPbI₃/Ti₃C₂T_x composite compared to pristine MAPbI₃.⁵⁰ The observed slight positive shift in the BE of the Pb 4f, C 1s, and O 1s peaks in MAPbI₃/Ti₃C₂T_x relative to pristine MAPbI₃, as evidenced by high-resolution XPS analysis in Fig. 1d–f indicates strong electronic coupling between MAPbI₃ and Ti₃C₂T_x. The increase in O 1s BE, indicative of functional group incorporation, significantly influences the electronic structure and reactivity of MXenes.⁴⁹ Functionalization with oxygen and fluorine typically increases work functions due to their pronounced electronegativity, while hydroxyl groups can diminish work functions as a result of their significant surface dipole effects.^39,51,52

Moreover, to unravel the impact of Ti₃C₂T_x MXene on the electronic properties of the perovskite layer, we conducted ultraviolet photoelectron spectroscopy (UPS) measurements on both films, revealing notable changes in the work function (WF) of MAPbI₃ perovskite upon integration with Ti₃C₂T_x MXene, as depicted in Fig. 2a. Analysis of a secondary electron cut-off curve showed the x-axis intersection of its tangent decreasing from 17.29 eV to 17.01 eV for pristine MAPbI₃ and MAPbI₃/Ti₃C₂T_x films, respectively, indicating a significant change in electronic properties due to additive incorporation, which impacts charge transport and overall device performance.³⁹ As an additive, Ti₃C₂T_x effectively increased the WF of MAPbI₃ perovskite by 0.19 eV, elevating it from 3.93 eV to 4.12 eV (detailed calculations are provided in Table S1†). As depicted in Fig. 2b, the intersection points of the valence band region with the x-axis remained constant at 1.4 eV for both samples, indicating that Ti₃C₂T_x does not alter the valence band maximum (VBM) relative to the Fermi level (E_f). The energy-level diagram in Fig. 2c was derived from these UPS data, providing a visual representation of the perovskite layer's electronic structure and interfacial properties.

Fig. 2 UPS spectra are measured with a photon energy of 21.22 eV on the perovskite films with and without Ti₃C₂T_x (a) secondary electron cut-off and (b) valence band region, respectively. (c) Energy-level diagram of MAPbI₃ and MAPbI₃/Ti₃C₂T_x.

Scanning electron microscopy (SEM) analysis was performed to elucidate the role of Ti₃C₂T_x in the morphology of perovskite films. Integrating Ti₃C₂T_x enlarged the average perovskite grain size from 130 nm to 170 nm, as evidenced by the insets of Fig. 3a and b. The pristine perovskite film displayed numerous smaller grains, whereas the cross-sectional image of the perovskite film with Ti₃C₂T_x showcased a smooth, pinhole-free morphology, as depicted in Fig. 3c and d. The enhanced grain growth is ascribed to the 2D architecture of Ti₃C₂T_x MXene, as it likely facilitates heterogeneous nucleation and suppresses defect density.^21,40,53 The occurrence of larger grain sizes in perovskite films is typically advantageous, as it decreases the density of grain boundaries that perform the function of recombination centers for photogenerated charge carriers, thus improving their transport throughout the film.^25,53,54

Fig. 3 SEM and particle size distribution curves of (a) MAPbI₃ and (b) MAPbI₃/Ti₃C₂T_x, cross-sectional image of (c) MAPbI₃ and (d) MAPbI₃/Ti₃C₂T_x, and AFM image of (e) pristine MAPbI₃ and (f) MAPbI₃/Ti₃C₂T_x. (g) UV-vis-NIR absorption analysis of pristine MAPbI₃ and MAPbI₃/Ti₃C₂T_x perovskite films. (h) PL analysis of pristine MAPbI₃ and MAPbI₃/Ti₃C₂T_x perovskite films. (i) TRPL analysis of pristine MAPbI₃ and MAPbI₃/Ti₃C₂T_x perovskite films.

Furthermore, incorporating Ti₃C₂T_x into the perovskite film improved its smoothness, as demonstrated by atomic force microscopy (AFM) analysis in Fig. 3e and f. The RMS roughness of the pristine perovskite film was 13.60 nm, dropping to 11.93 nm with Ti₃C₂T_x. The increased smoothness mitigates surface irregularities that trap charges, improving charge carrier transport and reducing recombination losses. It additionally augments interfacial contact with adjacent transport layers, boosting device efficiency and operational stability, thus rendering it suitable for practical applications.^28,41,55Fig. 3g illustrates the UV-vis-NIR absorption analysis of pristine MAPbI₃ and MAPbI₃/Ti₃C₂T_x perovskite films. Both films displayed a broad absorption spectrum with a distinct band edge at approximately 780 nm, corresponding to a bandgap of 1.58 eV. The enhanced light absorption in perovskite films with Ti₃C₂T_x is likely driven by increased light scattering, extending the optical path length, improving film quality, minimizing defects, and enhancing uniformity. The intercalation of Ti₃C₂T_x within the grain structure facilitates effects, as supported by the literature on MXene applications in optoelectronics.^25,26,53

The enhanced light absorption and reduced defect density facilitated noticeably stronger photoluminescence (PL) intensity peaks in the MAPbI₃/Ti₃C₂T_x film compared to the pristine perovskite film, as shown in Fig. 3h. The improvement can be linked to reducing non-radiative recombination processes and enhancing the optoelectronic properties of MAPbI₃/Ti₃C₂T_x.²⁵ The time-resolved photoluminescence (TRPL) spectra reveal a longer average lifetime for the MAPbI₃/Ti₃C₂T_x film than the pristine perovskite film in Fig. 3i. The finding suggests that incorporating Ti₃C₂T_x into MAPbI₃ effectively reduces surface defects and suppresses non-radiative recombination losses within the active layer of PDs.^26,41 The average lifetime for the pristine MAPbI₃ and MAPbI₃/Ti₃C₂T_x films was 3.78 ns and 5.38 ns, respectively. Passivation of defects hinders the rapid trapping of charge carriers, thereby slowing down the exciton decay process. Consequently, the PL and the average lifetime (τ_avg) experience a substantial enhancement.^56,57

2.2 2D Ti₃C₂T_x synergy for charge carrier optimization

Understanding the role of defects in influencing charge carrier dynamics is pivotal for optimizing perovskite photodetector performance. Using hole-only device structures (ITO/PTAA/perovskite/Spiro-OMeTAD/Ag), we characterized defect states in both pristine and Ti₃C₂T_x-modified perovskite films, as demonstrated in Fig. 4a and b. Dark current–voltage (I–V) measurements were employed to estimate trap density by analyzing specific regions of the double-logarithmic plots in the space-charge-limited current (SCLC) regime, where charge transport is dominated by space-charge effects rather than ohmic resistance. The low-voltage regime exhibits ohmic behavior (n = 1), signifying a linear current dependence on voltage, followed by the trapped-filled limit (TFL) region (n > 2), characterized by a power-law dependence with an exponent exceeding 2. Finally, a trap-free square-law region (n = 2) emerges at high voltages, indicating a quadratic relationship between current and voltage.³⁶ The crucial parameter for trap density estimation is the trap-filled limit voltage (V_TFL), which marks the transition between the ohmic and TFL regions, as mentioned in eqn (1).

Fig. 4 A plot of space charge-limited current (SCLC) of PDs: (a) MAPbI₃ and (b) MAPbI₃/Ti₃C₂T_x and (c) capacitance vs. frequency curves at zero bias under dark conditions. (d) C⁻²vs. voltage curves. (e) curves of the PDs with and without Ti₃C₂T_x modification. (f) The t-DOS vs. E_ω plot of PDs with and without Ti₃C₂T_x modification. (g) The schematic illustration of Ti₃C₂T_x-induced grain refinement and charge transport enhancement in MAPbI₃.

The calculated values N_t are 29.42 × 10¹⁵ cm⁻³ and 26.55 × 10¹⁵ cm⁻³ for MAPbI₃ and MAPbI₃/Ti₃C₂T_x for the hole-only devices, respectively. Incorporating Ti₃C₂T_x reduces trap density, enhancing charge transport efficiency and suppressing recombination losses, a key aspect of the high performance of PDs.³⁸ Further investigation into charge transport properties was conducted by quantifying mobility of charge carriers in unipolar devices using the Mott–Gurney law. Carrier mobility (μ) is extracted by utilizing a well-defined trap-free square-law region observed in the double-logarithmic plots of the I–V curves, as exhibited in Fig. S7† for hole-only devices, using eqn (2).

The calculated hole carrier mobility was 0.786 × 10⁻³ cm² V⁻¹ s⁻¹ and 1.074 × 10⁻³ cm² V⁻¹ s⁻¹ for MAPbI₃ and MAPbI₃/Ti₃C₂T_x, respectively. Incorporating Ti₃C₂T_x diminished trap-state density and its exceptional conductivity was utilized to establish a robust charge-transport network within the perovskite layer, evidently enhancing the efficiency of carrier dynamics in PDs.^58,59

Moreover, the impact of Ti₃C₂T_x content on the performance of the PD was also elucidated by examining capacitance–frequency (C–F) and capacitance–voltage (C–V) characteristics, as displayed in Fig. 4c and S8.† The lower capacitance of the optimized PD eventually leads to a faster response time.⁶⁰ The Mott–Schottky analysis and trap density of states (t-DOS) measurements were conducted to investigate the reasons behind the enhanced performance of the Ti₃C₂T_x optimized PD. The Mott–Schottky analysis in Fig. 4d revealed a slight increase in built-in potential (V_bi) from 0.80 V to 0.83 V for devices with and without Ti₃C₂T_x modification. The increased value of V_bi suggests that the device containing Ti₃C₂T_x has a more robust driving force for separating generated photocarriers, a factor that helps minimize recombination losses and improve D* in the PDs.⁶¹ The t-DOS can be deduced from eqn (3) and (4).

where V_bi is the built-in potential, q is the elementary charge, T is the thermodynamic temperature, k_B is the Boltzmann constant, ω is the angular frequency (ω = 2πf), and W is the depletion width. The ω_o is defined as the angular frequency where the value of is a minimum point, as presented in Fig. 4e. The values of t-DOS decreased with the addition of Ti₃C₂T_x content, indicating effective passivation of surface defects in the perovskite film, as presented in Fig. 4f. The result was lower trap state energy levels and decreased trap density, minimizing recombination sites and improving device performance.^28,57 The self-powered optimized PD, with its improved built-in potential and reduced trap density, achieves efficient charge separation and transport, resulting in fast response time. In addition to SCLC analysis, PL, TRPL, Mott–Schottky, and t-DOS measurement results reveal that Ti₃C₂T_x incorporation does not introduce new interfacial trap states but rather effectively passivates existing surface and interface defects. Enhanced PL intensity, longer carrier lifetimes, increased built-in potential, and a significant reduction in t-DOS collectively demonstrate that this defect passivation mechanism plays a key role in improving device performance.

Ti₃C₂T_x, leveraging its surface functional groups, forms robust interactions with methylammonium (MA) molecules in MAPbI₃ perovskites, modulating nucleation kinetics to yield larger, highly uniform grains within the film, as illustrated in the schematic in Fig. 4g. As evidenced by prior studies, these enlarged grains reduce the number of grain boundaries, which are primary sites for carrier trapping, thereby extending carrier lifetimes.^57,61 Furthermore, hydrogen bonding between Ti₃C₂T_x functional groups and MA⁺ ions suppresses nucleation density on PTAA, resulting in fewer initial nuclei, enhanced grain growth, and improved crystallinity, ultimately lowering surface roughness. The hydrogen-bonded network between Ti₃C₂T_x and the perovskite layer noticeably enhances stability and charge carrier transport, positioning Ti₃C₂T_x-modified perovskites as a compelling candidate for high-speed optoelectronic devices.⁵⁵

2.3 Effect of 2D Ti₃C₂T_x synergy on the photoelectric performance of PDs

Fig. 5a shows distinct rectification phenomena in J–V characteristics of both devices with and without Ti₃C₂T_x modification. Although both devices demonstrate rectification, the optimized device incorporating Ti₃C₂T_x exhibits a markedly lower dark current density of 1.46 × 10⁻¹⁰ A cm⁻², outperforming dark current density of the reference device at 3.08 × 10⁻⁸ A cm⁻². The incorporation of Ti₃C₂T_x into perovskite PDs profoundly reduces dark density current through several key mechanisms: effective passivation of surface defects, enhancement of charge carrier dynamics, reduction of interface trap states, and improvements in grain size and morphology.^26,37,55,62 These mechanisms collectively contribute to a significant reduction in dark current density, enhancing the performance metrics of the PD, particularly its sensitivity and efficiency in detecting weak optical signals. Both PDs exhibited photovoltaic behavior at 730 nm, with the optimized device demonstrating an open-circuit voltage of 1.10 V and a short-circuit current density of 3.5 mA cm⁻², outperforming reference device values of 1.05 V and 2.4 mA cm⁻², respectively. The improvements suggest enhanced charge separation and reduced recombination losses, leading to more efficient light-to-energy conversion, a fundamental requirement for applications like optical communication requiring efficient signal processing.^37,62 Fig. S9† further supports this performance by illustrating the J–V characteristics of optimized PDs at varying illumination intensities (0.1–100%) of a 730 nm laser, highlighting the corresponding open-circuit voltage and short-circuit current density.

Fig. 5 (a) PD photoresponse under illumination (730 nm, 48.3 mW cm⁻²) and dark conditions with and without Ti₃C₂T_x modification. (b) I–T curves of an optimized self-powered PD under 730 nm laser illumination with the power densities varying from 0.4 to 48.3 mW cm⁻², respectively. (c) Responsivity curves of the self-powered PDs with and without Ti₃C₂T_x modification. (d) Noise current of the self-powered PDs with and without Ti₃C₂T_x modification. (e) LDR of an optimized self-powered PD at 730 nm operating wavelength. (f) Nyquist plots of the self-powered PDs with and without Ti₃C₂T_x modification.

Fig. 5b presents the I–T curve at various incident light intensities in self-powered mode of a 730 nm laser. The photocurrent demonstrates a precise correlation with the periodic laser modulation. Consequently, a direct correlation between photocurrent and optical power density is observed, with decreasing current accompanying reduced light intensity due to a lower generation rate of electron–hole pairs.^5,14 The device maintains a low dark current, signifying excellent off-state performance. Similarly, consistent I–T behavior was observed across different wavelengths at equivalent illumination power densities as shown in Fig. S10.† The device exhibits exceptional optical switching characteristics with rapid rise and fall times that remain stable across these wavelengths, making it ideal for high-speed optical communication applications. Its consistent performance across a broad spectral range enhances its applicability in diverse optical systems, ensuring reliable operation under varying lighting conditions. Moreover, the responsivity and detectivity (D*) are also calculated using eqn (5) and (6).

The responsivity of an optimized PD was 55 mA W⁻¹, whereas, on the other hand, the reference PD has a responsivity of 29.1 mA W⁻¹ at an operating wavelength of 730 nm, as illustrated in Fig. 5c. The calculated external quantum efficiency (EQE) of an optimized PD with a light intensity of 25.3 mW cm⁻² is 9.39% at 730 nm as shown in Fig. S11.† To comprehensively evaluate device performance, we calculated the noise-equivalent power (NEP) and D*, where the NEP represents minimum detectable power for a 1 Hz bandwidth, primarily limited by 1/f noise. D* directly impacts their ability to detect weak signals under noisy conditions.

where S is the active area, B is the bandwidth, NEP is the noise-equivalent power, and i_n is the noise current, respectively. A fast Fourier transform (FFT) is used to analyze the dark current over time and obtain the frequency-dependent noise spectral density to assess PD sensitivity, particularly in the low-frequency region. The data are essential for determining how noise affects D*. The noise spectral density is divided by responsivity of a PD to determine the NEP. Weak optical signals can be detected more effectively with a lower noise current. Adding Ti₃C₂T_x improves film quality, reduces trap density, and lowers dark current, thus minimizing noise and enhancing the sensitivity of PDs to weak light for optical communication applications. The optimized PD displays a much lower noise current of 1.25 × 10⁻¹⁵ A Hz^−1/2 at 1 Hz bandwidth in self-powered mode, compared to 1.85 × 10⁻¹³ A Hz^−1/2 for the reference PD, as shown in Fig. 5d. Similarly, at a wavelength of 730 nm, the optimized PD demonstrates a significantly higher D*, with values of 6.59 × 10¹² Jones compared to 3.13 × 10¹⁰ Jones for the reference PD, surpassing previously reported values in the literature for optical communication applications.^9,10,23,46 The performance is also validated by its LDR, a parameter that describes the degree to which output current of the PD is proportional to the incident light radiation flux.²⁰ The LDR is calculated by using eqn (7).^3,63In the light intensity range from 0.06 to 48.3 mW cm⁻² in self-powered mode at a wavelength of 730 nm, the optimized device exhibits a measured LDR value of 81.4 dB. At the same time, the theoretical LDR, based on dark current, can reach 181 dB, as seen in Fig. 5e, demonstrating better performance compared to that in previously reported literature studies.^29,46 Such a performance ensures accurate detection across varying light intensities, improves signal processing, reduces noise impact, and supports high-speed data transmission for practical applications.^64,65 Likewise, the reference device exhibits a measured LDR value of 56.7 dB, while the theoretical LDR, based on dark current, can reach 126 dB, as shown in Fig. S12a and b† which demonstrates the LDR of an optimized PD at an operating wavelength of 656 nm, 520 nm, and 395 nm, highlighting its versatility and effectiveness across a broad spectral range. Additionally, Fig. 5f illustrates the Nyquist plots obtained via electrochemical impedance spectroscopy (EIS) analysis in the dark, aimed at determining series resistance (R_s) and recombination resistance (R_rec), with the corresponding values summarized in Table S2 (ESI†) and the inset displaying the equivalent circuit for fitting the data. The reference PD exhibited a lower charge R_rec of 636.51 Ω compared to the optimized PD value of 1769.03 Ω, having increased by nearly a factor of three, offering compelling evidence of enhanced suppression of carrier recombination in the MAPbI₃/Ti₃C₂T_x PDs as R_rec is inversely proportional to charge recombination losses.^26,66 The improvement of R_rec is mainly attributed to the improved quality of the perovskite films, as evidenced by passivation of defect states and a decrease in grain boundaries resulting from the larger grain size.⁶²

2.4 Influence of 2D Ti₃C₂T_x synergy on PD response and stability

Rapid response times in PDs are critical for various applications, including optical communication systems, imaging technologies, and the detection of weak light signals. The reference PD demonstrated slower response characteristics, exhibiting rise and fall times of 4.32 μs and 12.00 μs, as demonstrated in Fig. 6a. In contrast, the optimized one shown in Fig. 6b exhibited exceptionally faster response times, with rise and fall times of 1.07 μs and 0.74 μs, making it more suitable for practical applications compared to previously reported devices.^11,18,23 The high electrical conductivity and mobility of Ti₃C₂T_x lead to efficient charge carrier transport, with its ability to passivate defects and reduce non-radiative recombination, considerably improving PD response times and making it suitable for applications requiring rapid signal processing.^37,67 The current–time (I–t) photoresponse characteristics under reverse bias conditions (Fig. S13†) demonstrate excellent operational stability across different applied voltages. Moreover, our optimized PD demonstrated exceptional operational stability and reliable switching ability under self-powered conditions, with negligible degradation in light and dark current responses during a 2400-second test period under 730 nm and 48.3 mW cm⁻² illumination, as displayed in Fig. 6c.

Fig. 6 Rise and fall time of self-powered PDs: (a) reference and (b) optimized. (c) The photoresponse of the optimized PD in a test period of 40 minutes. The left and right images depict magnified views of the response at the first and last times, respectively. (d) The photoresponse of an optimized PD stored under ambient conditions, and (e) an optimized PD stored at ≈80% RH relative humidity.

Conversely, the reference PD exhibited notably lower stability under self-powered conditions, with a substantial 3.12% degradation in light and dark current responses within a shorter 470-second test period, as demonstrated in Fig. S14.† A clear contrast highlights the substantial performance enhancement achieved through our optimization efforts addressing the notoriously challenging stability of 3D perovskite-based optoelectronic devices. The PD was further tested after a month of ambient storage, with only an 8.4% reduction in performance, as presented in Fig. 6d. Similarly, the PD stored at relative humidity demonstrated a 30% decay in performance, as illustrated in Fig. 6e. The incorporation of Ti₃C₂T_x significantly enhances the stability of the perovskite film through a multifaceted approach. It forms a hydrophobic barrier against moisture and oxygen ingress, contributes to improved thermal stability, and effectively passivates chemical and surface defects, and the formation of a more robust film with larger grains may offer some degree of mechanical reinforcement, collectively mitigating degradation from environmental stressors such as moisture and heat.^27,41,55,68

Table 1 compares the performance of our PD metrics with that of recently reported PDs using other doped halide perovskite-based PDs and halide perovskite-integrated 2D materials. Our PD features remarkably lower dark current density, which is ideal for low-light detection. The LDR of 181 dB outperforms that of many state-of-the-art devices, ensuring precise light intensity conversion. With rapid response times (1.07 μs rise, 0.74 μs fall), the device excels in fast optical communications and high-speed data transfer. Integrating perovskite with MXenes boosts carrier transport and extraction efficiency, culminating in superior performance. The benchmark comparison highlights the potential of PDs in next-generation optoelectronic applications.

Table 1 The detailed parameters for comparison with those in previously reported literature studies are based on doped halide perovskite-based PDs and halide perovskite-integrated two-dimensional materials^a

Materials	Bias (V), λ (nm)	Dark current/Ion/Ioff	R (mA W^−1)	D* (Jones)	LDR (dB)	Response time	Ref.
a HS (heterostructure), NC (nanocrystal), QDs (quantum dots), MW (microwires), rGO (reduced graphene oxide), NA (not mentioned).
MAPbI3/Ti3C2Tx	0, 730	1.46 × 10^−10 A cm^−2/NA	55	6.59 × 10^12	181	1.07/0.74 μs	This work
Ti3C2Tx/MAPbI3(HS)	0, 405	2.03 × 10^−7 A cm^−2/NA	1.4	—	—	2.8/2.2 ms	1
CsPbBr3/Ti3C2Tx	10, 450	NA/2.3 × NA/10^3	44.9	6.4 × 10^8	—	48/18 ms	11
Ti3C2TxNC/MAPbI3 (MW)	2, 525	—	1700	7.0× 10^11	—	<20/66.3 ms	28
CsPbBr3QDs/Ti3C2Tx	1, 515	—	0.097	∼7× 10^8	49.9	46.2/24.6 ms	29
MAPbI3/graphene	0, 530	NA/4 × 10^6	375	4.5 × 10^11	—	5 ms	30
PTAA/MAPbI3	0, 720	—	4.7	1.69 ×10^10	110.9	15.85/1.52 ms	32
MAPbI3/MAPbBr3	0, 445	8.7 × 10^−4 (nA)/NA	23.77	4.5 ×10^13	—	70/62 ms	18
Ti3C2Tx/ZnIn2S4	10, 450	NA/10^5	1.04	1 × 10^11	—	646.8/640 ms	33
MAPbI3/rGO	0.1, 1064	—	721	0.05× 10^9	—	460 μs	34
NiO/MAPbI3	0, 290	NA/169	33.39	∼10^10	—	452/363 ms	35
MAPbI3/graphene	0.01, 785	NA/23	2.2	1.78× 10^5	—	∼68 ms	43
MAPbI3/CuSCN	0, 685	—	370	1.06× 10^12	101	5.02/5.50 μs	23
(BA)2FAPb2I7: FACl/C8BTBT	2, 405	—	2300	3.2× 10^12	—	9.74/8.91 μs	9
Ti3C2Br/CsFAMA	1, 520	NA/10^4	3930	2.6× 10^12	98	29.9 μs	46

---

Hence, the optimized PD converts light into electrical signals efficiently, maintains remarkable stability under self-powered conditions with minimal degradation, and features excellent optical switching with rapid rise and fall times for high-speed data transmission. Its consistent performance across a broad spectral range enhances versatility under varying light conditions, making it ideal for optical communication and sensitive light detection applications. In the subsequent section, the optimized PD will be referred to simply as PD. The ability of a device to efficiently process and detect light signals with minimal signal distortion and loss makes it a valuable asset in these fields. Integrated into an optical communication system as a light signal receiver, it operates within the visible-NIR range at wavelengths of 460 nm (blue), 520 nm (green), 645 nm (red), and 730 nm (deep red). These wavelengths are critical for visible light communication (VLC), optical fiber communication, surveillance systems, specialized military engineering applications, and a wide array of imaging and sensing technologies vital to medical diagnostics and environmental monitoring.

2.5 PD-driven optical communication strategy

Fig. 7a illustrates a homemade optical communication system, where the PD receives the signal from the driver, and the corresponding output is displayed on an oscillograph. Initially, in a human-readable format (ASCII), information from the computer was converted into a series of high and low-intensity levels (binary) using a driver, which modulated an LED to generate light in on-and-off states. A microcontroller converts computer data into analog signals 1 and 0, corresponding to high and low intensities, respectively. Moreover, a PD at the receiving end captured these light pulses and converted them into corresponding electrical voltages of varying strengths (high and low).

Fig. 7 (a) Photograph of a homemade optical communication system with the optimized PD. (b) The waveform of NBU binary data from the PD. Signal response of the PD at various frequencies: (c) 50 Hz, (d) 100 Hz, (e) 500 Hz, and (f) 1000 Hz. (g) The schematic circuit diagram of the light detection system. (h) PCB board for the light detection system.

A microcontroller transmitted the binary data representing NBU, corresponding to high and low voltages, to the PD, and the resulting output was displayed on an oscilloscope before being transferred to a computer for final analysis. Fig. 7b demonstrates the PD's accurate and precise conversion of NBU characters into binary data at 10 Hz, showcasing its exceptional signal transfer capabilities with no signal loss. As shown in Video 1,† this process spanned a broad spectral range across all four LEDs (460–730 nm), rendering the PD versatile for diverse optical communication applications. The rapid response time screenshot from the oscillograph (rise time: 2.03 μs, fall time: 5.37 μs) ensured seamless and unaffected encoding, as presented in Fig. S15 of the ESI.†Fig. 7c–f exhibit superior signal generation across a frequency spectrum of 50 Hz to 1000 Hz without signal deterioration. ESI Fig. S16† reveals that the PD's rise and fall times remain stable with increasing frequency, indicating its capacity to manage rapid signal changes effectively.⁶⁹ The LED illumination intensity was modulated to evaluate the PD's suitability for low-light applications, such as night vision and surveillance. Although the oscilloscope output exhibited a pronounced on/off ratio at maximum LED intensity, this ratio decreased under low-light conditions, as evidenced in Video 2.† This demonstrates its efficient operation under fluctuating light intensity and reliable transfer of undistorted signals. Additionally, the effect of distance between the LED and PD was examined, with optimal performance observed within a 10 cm range, beyond which performance declined. The findings underscore the PD's suitability for applications requiring reliable light detection within practical operational limits, as illustrated in Video 3.† The optimized PD's accelerated charge transport (enabling fast response), effective defect suppression (minimizing noise), and stable operation under varying light conditions make it ideal for real-world applications. A brief overview of the device structure and operating conditions of PDs for optical communication applications is presented in Table S3.†

2.6 PD-driven sensitive light detection strategy

Sensitive light detection systems operating in the 460–730 nm wavelength range with a broad spectral response are crucial for various applications. Here, commercial LEDs were used as light targets in a detection system with a PD that converted LED light signals into electrical signals, which were then amplified by using an LM358AD operational amplifier and a 15C02MH-TL-E NPN bipolar junction transistor. The LM358AD, configured as a negative feedback amplifier, amplified the input voltage V_i, with gain A_v = (1 + R₃/R₄) and the output end V_o = A_vV_i, realizing the amplification function (R₃ and R₄ can be changed appropriately, here R₃ = 50 kΩ and R₄ = 1 kΩ). The 15C02MH-TL-E transistor acted as a switch, controlling the current flow and ensuring that the output LED was driven by the amplified signal from the PD. The system's sensitivity to specific wavelengths within the visible spectrum allowed it to generate and amplify small electrical currents in response to light. The amplified signals regulated the output LED, demonstrating the effectiveness of the system in optical detection applications. This study highlights the potential of perovskite PDs to integrate seamlessly with other circuits, paving the way for new advancements in this field. The complete schematic circuit diagram and the light detection system PCB board are displayed in Fig. 7g and h, respectively.

When the illumination strikes the PD, it generates a small electrical current. A photograph of the experimental setup of the sensitive light detection system is presented in ESI Fig. S17a.† The current is amplified by using a 15C02MH-TL-E bipolar junction transistor, significantly increasing its magnitude, which regulates the output LED, causing it to light up in response to the detected light, as demonstrated in Video 4.† The successful operation of this system underscores the effectiveness of the devices for optical detection applications, as shown in the photographs presented in ESI Fig. S17b.† The impressive outcomes of the improved device application were achieved through the enhanced perovskite PD with MAPbI₃/Ti₃C₂T_x. The PD demonstrated outstanding performance characteristics that directly aligned with its potential uses. The lower dark current density ensures robust low-light detection capabilities that are imperative for surveillance and environmental monitoring. The large LDR of 181 dB enables precise light intensity conversion, making it highly effective for applications requiring wide dynamic ranges. Moreover, the rapid response times (rise time: 1.07 μs, fall time: 0.74 μs) facilitate fast optical communications, essential for high-speed data transfer and real-time signal processing. These advancements underscore the remarkable potential of the improved perovskite PD in various practical applications, highlighting its versatility and efficiency.

3 Conclusions

This study demonstrates significant advancements in perovskite PDs by integrating 2D Ti₃C₂T_x MXene. The addition of MXenes greatly enhanced the stability, uniformity, and charge carrier transport of perovskite films, leading to remarkable improvements in photoelectric performance. The optimized self-powered PD achieved a low dark current density of 1.46 × 10⁻¹⁰ A cm⁻², an outstanding D* of 6.59 × 10¹² Jones, a wide LDR exceeding 181 dB, and ultra-fast response times of 1.07 μs (rise) and 0.74 μs (fall). These results underscore the potential of Ti₃C₂T_x PDs for high-speed optical communication and low-light detection. The synergy between perovskite materials and MXenes unveils compelling opportunities for crafting rapid, high-performance, and stable PDs, establishing them as formidable candidates for future optoelectronic breakthroughs. Ultimately, these self-powered flexible PDs excelled in a broad-spectrum optical communication system and a highly sensitive light detection strategy, further solidifying their versatility. This work lays a solid foundation for further exploration of MXene-integrated PDs, establishing them as pivotal components for advancing next-generation optical communication and sensing technologies across various applications.

4 Experimental section

4.1 Materials

Methylammonium iodide (MAI, 99.5%), lead iodide (PbI₂, 99.99%), poly[bis(4-phenyl) (2,4,6-trimethylphenyl)amine] (PTAA), [6,6]-phenyl-C61-butyric acid methyl ester (PCBM, 99%), bathocuproine (BCP, 98%), bis(trifluoromethane) sulfonimide lithium salt (Li-TFSI), 4-tert-butyl pyridine (tBP), and 2,2′,7,7′-Tetrakis (N,N-di-p-methoxyphenyl-amine)-9,9′-spirobifluorene (Spiro-OMeTAD) were purchased from Xi'an Polymer Light Technology Corp and used without further purification. Dimethylformamide (DMF), dimethyl sulfoxide (DMSO), acetonitrile (ACN), and chlorobenzene (CB) were purchased from Aladdin. Ti₃C₂T_x was purchased from Foshan Xinxi Technology Co., Ltd. All chemicals were used without further purification.

4.2 Device fabrication

ITO substrates were ultrasonically cleaned in deionized water and ethanol alternately for 15 minutes each, and then dried. Prior to the deposition of the HTL, the substrates were subjected to a 15-minute UV/ozone treatment to enhance cleanliness and adhesion. A 2 mg mL⁻¹ PTAA solution was spin-coated onto an ITO substrate at 4500 rpm for 30 seconds, thereafter undergoing annealing on a hotplate at 100 °C for 10 minutes. A one-step fabrication method involved dissolving 461 mg of PbI₂ and 159 mg of MAI in 1 mL of DMF and 79 μl of DMSO to form the perovskite precursor solution, which was then deposited on the substrates. Ti₃C₂T_x (0.5 mg mL⁻¹ in DMF) supernatant solution (to obtain single smaller layers) was added into the perovskite precursor, always retaining the original concentration of the perovskite and solvent ratio by reducing the original volume of DMF. The perovskite solution was spin-coated in two steps: 1000 rpm for 5 seconds, and then 4500 rpm for 45 seconds, with 200 μL of CB added as an antisolvent during the final 35 seconds, followed by annealing the films at 100 °C for 10 minutes. A 35 μL PCBM solution (20 mg mL⁻¹ in CB) was spin-coated onto the perovskite film (3000 rpm, 30 s) and annealed (70 °C, 5 min). The supersaturated BCP solution in isopropanol was dynamically coated on the PCBM layer (5500 rpm, 30 s). Finally, a 60 nm silver (Ag) electrode (through a shadow mask) was deposited by thermal evaporation at 3.8 × 10⁻⁴ Pa.

4.3 Characterization

The morphology of the film surface, grain size, and thickness were investigated by using a Hitachi SU-70 scanning electron microscope (SEM). X-ray diffraction patterns (XRD) were obtained using a Bruker instrument using Cu Kα radiation (D8 Advance, Germany). The surface roughness of the films was analyzed using a Veeco atomic force microscope (AFM). UV-vis absorption spectra of the deposited films on ITO were obtained on a UV-vis spectrophotometer (Agilent Cary 5000, USA). A photoluminescence system with a 532 nm excitation laser (NT-MDT Spectrum Instruments) was used to measure PL spectra. Time-resolved photoluminescence mapping (TRPL mapping) was conducted using a confocal scanning fluorescence lifetime (intensity) imaging system (PicoQuant, Micro Time 100). X-ray photoelectron spectroscopy (XPS) and ultraviolet photoelectron spectroscopy (UPS) were conducted using a Thermo Scientific ESCALAB QXI. All the tests on the PDs were conducted using the Paios measurement setup, which includes a built-in laser system in conjunction with the Fluxim Characterization Suite software. Capacitance–frequency (C–F) and capacitance–voltage (C–V) curves were measured using a Paios setup. The noise signal was measured using a Keithley 2450 source meter in a dark environment with no external bias voltage applied. An Agilent MSO9254A mixed-signal oscilloscope was employed for the custom-built optical communication system to precisely capture and display the output waveforms. A microcontroller unit (MCU: STC89C516RD+) generated the binary signals, encoded as sequences of 0s and 1s, to drive the system.

Additionally, a RIGOL DG4062 signal generator produced signals across a range of frequencies, enabling dynamic testing. Commercially sourced LEDs emitting at 460 nm, 520 nm, 645 nm, and 730 nm served as illumination sources, facilitating multi-wavelength performance evaluation. These LEDs were driven at a uniform intensity of 1 mW cm⁻² to maintain consistent illumination conditions across both experiments. For the sensitive light detection system, a printed circuit board (PCB) was engineered based on the circuit schematic, with commercially sourced LEDs employed as illumination and detection sources. All device characterization studies, performance measurements, and application evaluations were carried out at room temperature under ambient air conditions.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (Grant No. 62074088, 62174094, and 62304116), the Foundation of Zhejiang Educational Commission (Grant No. Y201737090), and the Natural Science Foundation of Ningbo City (Grant No. 2017A610018). The author Z. H. would like to acknowledge the sponsorship by K. C. Wong Magna Fund in Ningbo University.

References

1. G. Han, X.-F. Li, A. Berbille, Y. Zhang, X. Luo, L. Liu, L. Li, Z. L. Wang and L. Zhu, Adv. Mater., 2024, 36, 2313288.
2. M. Sulaman, S. Y. Yang, Z. H. Zhang, A. Imran, A. Bukhtiar, Z. H. Ge, Y. Tang, Y. R. Jiang, L. B. Tang and B. S. Zou, Mater. Today Phys., 2022, 27, 100829.
3. Y. Miao, L. Sheng, J. Wu, X. Qi, L. Yang, C. Yang, X. Wang, F. Zheng, S. Shafique, F. Zhao and Z. Hu, Laser Photonics Rev., 2024, 19, 2401096.
4. S. Shafique, S. Yang, T. Iqbal, B. Cheng, Y. Wang, H. Sarwar, Y. T. Woldu and P. Ji, Sens. Actuators, A, 2021, 332, 113073.
5. K. Abbas, P. Ji, N. Ullah, S. Shafique, Z. Zhang, M. F. Ameer, S. Qin and S. Yang, Microsyst. Nanoeng., 2024, 10, 81.
6. Z. Xiao, X. Xiao, L. B. Kong, H. Dong, X. Li, B. He, S. Ruan, J. Zhai, K. Zhou, Q. Huang and L. Chu, Int. J. Extreme Manuf., 2024, 6, 022006.
7. Y. Shi, L. Chu, H. Wei, Z. Li, F. Qin, Q. Zhang and W. Li, Cell Rep. Phys. Sci., 2024, 5, 101903.
8. W. Wu, H. Lu, X. Han, C. Wang, Z. Xu, S.-T. Han and C. Pan, Small Methods, 2023, 7, 2201499.
9. T. Wang, D. Zheng, J. Zhang, J. Qiao, C. Min, X. Yuan, M. Somekh and F. Feng, Adv. Funct. Mater., 2022, 32, 2208694.
10. N. Ma, J. Jiang, Y. Zhao, L. He, Y. Ma, H. Wang, L. Zhang, C. Shan, L. Shen and W. Hu, Nano Energy, 2021, 86, 106113.
11. W. Deng, H. Huang, H. Jin, W. Li, X. Chu, D. Xiong, W. Yan, F. Chun, M. Xie, C. Luo, L. Jin, C. Liu, H. Zhang, W. Deng and W. Yang, Adv. Opt. Mater., 2019, 7, 1801521.
12. G. Cen, Y. Liu, C. Zhao, G. Wang, Y. Fu, G. Yan, Y. Yuan, C. Su, Z. Zhao and W. Mai, Small, 2019, 15, 1902135.
13. H. Gu, S.-C. Chen and Q. Zheng, Adv. Opt. Mater., 2021, 9, 2001637.
14. A. Qadir, S. Shafique, T. Iqbal, H. Ali, L. Xin, S. Ruibing, T. Shi, H. Xu, Y. Wang and Z. Hu, Sens. Actuators, A, 2024, 370, 115267.
15. M. Long, P. Wang, H. Fang and W. Hu, Adv. Funct. Mater., 2019, 29, 1803807.
16. G. Li, Y. Wang, L. Huang and W. Sun, ACS Appl. Electron. Mater., 2022, 4, 1485–1505.
17. S. Shafique, A. Qadir, T. Iqbal, M. Sulaman, L. Yang, Y. Hou, Y. Miao, J. Wu, Y. Wang, F. Zheng, X. Wang and Z. Hu, J. Alloys Compd., 2024, 1004, 175903.
18. C. Liu, R. Wang, J. Liu, W. Xu, H. Zhang, X. Liu, X. Huo, X. Cui, H. Sun, R. Liu, H. Zhu, W. Zhang and J. Ding, Adv. Opt. Mater., 2025, 13, 2403123.
19. J. Wu, Y. Miao, X. Qi, L. Yang, X. Wang, F. Zheng, F. Zhao, S. Shafique and Z. Hu, ACS Photonics, 2023, 10, 4437–4446.
20. Y. Miao, J. Wu, X. Qi, L. Yang, X. Wang, F. Zheng, F. Zhao, S. Shafique, Y. Zhu and Z. Hu, Nano Energy, 2023, 113, 108605.
21. D. Saranin, S. Pescetelli, A. Pazniak, D. Rossi, A. Liedl, A. Yakusheva, L. Luchnikov, D. Podgorny, P. Gostischev, S. Didenko, A. Tameev, D. Lizzit, M. Angelucci, R. Cimino, R. Larciprete, A. Agresti and A. Di Carlo, Nano Energy, 2021, 82, 105771.
22. T. Wu, Z. Qin, Y. Wang, Y. Wu, W. Chen, S. Zhang, M. Cai, S. Dai, J. Zhang, J. Liu, Z. Zhou, X. Liu, H. Segawa, H. Tan, Q. Tang, J. Fang, Y. Li, L. Ding, Z. Ning, Y. Qi, Y. Zhang and L. Han, Nano-Micro Lett., 2021, 13, 152.
23. Z. Liu, X. Liu, B. Sun, X. Tan, H. Ye, J. Zhou, Z. Tang, T. Shi and G. Liao, Adv. Mater. Technol., 2020, 5, 2000260.
24. C. Yang, W. Hu, J. Liu, C. Han, Q. Gao, A. Mei, Y. Zhou, F. Guo and H. Han, Light:Sci. Appl., 2024, 13, 227.
25. S. Qamar, K. Fatima, N. Ullah, Z. Akhter, A. Waseem and M. Sultan, Nanoscale, 2022, 14, 13018–13039.
26. Z. Guo, L. Gao, Z. Xu, S. Teo, C. Zhang, Y. Kamata, S. Hayase and T. Ma, Small, 2018, 14, 1802738.
27. J. Zhang, C. Huang, Y. Sun and H. Yu, Adv. Funct. Mater., 2022, 32, 2113367.
28. H. Guo, H. Wang, H. Fan, Q. Ye, J. Zhang, F. Cao, L. Li, Y. Tong and H. Wang, Small, 2021, 17, 2101954.
29. H. Li, Z. Li, S. Liu, M. Li, X. Wen, J. Lee, S. Lin, M.-Y. Li and H. Lu, J. Alloys Compd., 2022, 895, 162570.
30. J. Li, S. Yuan, G. Tang, G. Li, D. Liu, J. Li, X. Hu, Y. Liu, J. Li, Z. Yang, S. F. Liu, Z. Liu, F. Gao and F. Yan, ACS Appl. Mater. Interfaces, 2017, 9, 42779–42787.
31. R. Liu, J. Zhang, H. Zhou, Z. Song, Z. Song, C. R. Grice, D. Wu, L. Shen and H. Wang, Adv. Opt. Mater., 2020, 8, 1901735.
32. R. Huang, K. Wu, W. Li, X. Huang, S. Zhou, S. Jiang, Y. Fu, Z. Zhao, W. Mai and C. Zhao, Adv. Opt. Mater., 2023, 11, 2301177.
33. S. Hou, C. Xu, X. Ju and Y. Jin, Adv. Sci., 2022, 9, 2204687.
34. M. Li, Y. Zhang, X. Tang, J. Li, S. Wang, T. Li, H. Zhao, Q. Li, Q. Wang and J. Yao, J. Alloys Compd., 2022, 895, 162496.
35. V. Manjunath, P. K. Mishra, R. Dobhal, S. Bimli, P. M. Shirage, S. Sen, P. A. Shaikh and R. S. Devan, ACS Appl. Electron. Mater., 2021, 3, 4548–4557.
36. G. Ma, Y. Hou, Z. Luo, J. Wu, Y. Miao, C. Yang, X. Wang, F. Zheng, S. Shafique, F. Zhao and Z. Hu, Small, 2024, 20, 2405820.
37. L. Gao, Y. Zhao, X. Chang, J. Zhang, Y. Li, S. Wageh, O. A. Al-Hartomy, A. G. Al-Sehemi, H. Zhang and H. Ågren, Mater. Today, 2022, 61, 169–190.
38. S. Palei, G. Murali, C.-H. Kim, I. In, S.-Y. Lee and S.-J. Park, Nano-Micro Lett., 2023, 15, 123.
39. A. Agresti, A. Pazniak, S. Pescetelli, A. Di Vito, D. Rossi, A. Pecchia, M. Auf der Maur, A. Liedl, R. Larciprete, D. V. Kuznetsov, D. Saranin and A. Di Carlo, Nat. Mater., 2019, 18, 1228–1234.
40. S. Cao, L. G. Gutsev, Z. Bi, Y. Zheng, X. Xu, Y. Zhu, L. Zhong, J. Zheng, G. Xu, P. A. Troshin, S. Liu, K. Wang, C. Gonzales, A. Guerrero, Z. Ren and G. Li, Chem. Eng. J., 2023, 474, 145707.
41. J. H. Heo, F. Zhang, J. K. Park, H. Joon Lee, D. S. Lee, S. J. Heo, J. M. Luther, J. J. Berry, K. Zhu and S. H. Im, Joule, 2022, 6, 1672–1688.
42. C. Hu, Y. Bai, P. Qiu, D. Wu, Y. Zhang, J. Liu, C. Fu and G. Shen, Small Struct., 2025, 2400649,  DOI:10.1002/sstr.202400649.
43. J. Ding, H. Fang, Z. Lian, Q. Lv, J.-L. Sun and Q. Yan, Nanoscale, 2018, 10, 10538–10544.
44. M. Malaki and R. S. Varma, Nano-Micro Lett., 2023, 15, 116.
45. A. Bhat, S. Anwer, K. S. Bhat, M. I. H. Mohideen, K. Liao and A. Qurashi, npj 2D Mater. Appl., 2021, 5, 61.
46. X. Zhou, L. Luo, Y. Huang, S. Wei, J. Zou, A. He, B. Huang, X. Li, J. Zhao, K. Shen, D. Zhao and J. Wu, J. Mater. Chem. C, 2022, 10, 5970–5980.
47. A. Dehingia, U. Das, H. P. Gogoi, K. K. Borgohain, S. Patra, B. Paul and A. Roy, Small, 2024, 20, 2401179.
48. L. Yang, P. Hou, B. Wang, C. Dall'Agnese, Y. Dall'Agnese, G. Chen, Y. Gogotsi, X. Meng and X.-F. Wang, Chem. Eng. J., 2022, 446, 136963.
49. J. Yoon, S. Kim, K. H. Park, S. Lee, S. J. Kim, H. Lee, T. Oh and C. M. Koo, Small Methods, 2023, 7, 2201579.
50. H. Li, X. Lv, R. Li, X. Tao and Y. Zheng, J. Power Sources, 2022, 522, 231006.
51. S. W. Koh, L. Rekhi, Arramel, M. D. Birowosuto, Q. T. Trinh, J. Ge, W. Yu, A. T. S. Wee, T. S. Choksi and H. Li, ACS Appl. Mater. Interfaces, 2023, 16, 66826–66836.
52. J. L. Hart, K. Hantanasirisakul, A. C. Lang, B. Anasori, D. Pinto, Y. Pivak, J. T. van Omme, S. J. May, Y. Gogotsi and M. L. Taheri, Nat. Commun., 2019, 10, 522.
53. C. Wu, W. Fang, Q. Cheng, J. Wan, R. Wen, Y. Wang, Y. Song and M. Li, Angew. Chem., Int. Ed., 2022, 61, e202210970.
54. L. Yang, Y. Wang, X. Wang, S. Shafique, F. Zheng, L. Huang, X. Liu, J. Zhang, Y. Zhu, C. Xiao and Z. Hu, Small, 2024, 20, 2304362.
55. X. Jin, L. Yang and X.-F. Wang, Nano-Micro Lett., 2021, 13, 68.
56. T. J. Kim, D. Kwon, J. Y. Kim, A. K. Harit, E. Lee, H. Y. Woo, J. Kim and J. Joo, ACS Appl. Electron. Mater., 2022, 4, 4560–4568.
57. W. Tian, H. Zhou and L. Li, Small, 2017, 13, 1702107.
58. L. M. Herz, ACS Energy Lett., 2017, 2, 1539–1548.
59. J. Lim, M. Kober-Czerny, Y.-H. Lin, J. M. Ball, N. Sakai, E. A. Duijnstee, M. J. Hong, J. G. Labram, B. Wenger and H. J. Snaith, Nat. Commun., 2022, 13, 4201.
60. C. Li, Y. Ma, Y. Xiao, L. Shen and L. Ding, InfoMat, 2020, 2, 1247–1256.
61. K. Wang, Z. Xu, Z. Guo, H. Wang, S. M. H. Qaid, K. Yang and Z. Zang, Adv. Energy Mater., 2024, 14, 2402249.
62. Y. Zhao, X. Zhang, X. Han, C. Hou, H. Wang, J. Qi, Y. Li and Q. Zhang, Chem. Eng. J., 2021, 417, 127912.
63. J. Wu, Y. Miao, X. Qi, L. Yang, X. Wang, F. Zheng, F. Zhao, Z. Zhao, S. Shafique, H. Zhang and Z. Hu, J. Mater. Chem. A, 2024, 12, 20425–20436.
64. C. Li, H. Wang, F. Wang, T. Li, M. Xu, H. Wang, Z. Wang, X. Zhan, W. Hu and L. Shen, Light:Sci. Appl., 2020, 9, 31.
65. Y. Tang, P. Jin, Y. Wang, D. Li, Y. Chen, P. Ran, W. Fan, K. Liang, H. Ren, X. Xu, R. Wang, Y. Yang and B. Zhu, Nat. Commun., 2023, 14, 4961.
66. N. Liu, Z. Liu, J. Wang, Q. Ye, Y. Wang, W. Han, W. Xu, J. Zhang, L. Huang, Z. Hu, Y. Zhu and X. Liu, Chem. Eng. J., 2024, 485, 149590.
67. J. Kublitski, A. Hofacker, B. K. Boroujeni, J. Benduhn, V. C. Nikolis, C. Kaiser, D. Spoltore, H. Kleemann, A. Fischer, F. Ellinger, K. Vandewal and K. Leo, Nat. Commun., 2021, 12, 551.
68. M. Karimipour, A. Paingott Parambil, K. Tabah Tanko, T. Zhang, F. Gao and M. Lira-Cantu, Adv. Energy Mater., 2023, 13, 2301959.
69. P. Wang, A. Cavallini, G. C. Montanari and G. Wu, IEEE Trans. Dielectr. Electr. Insul., 2013, 20, 245–254.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5ta02548f

---