Introduction

Electrochromic devices (ECDs) have emerged as a promising class of dynamic optical components capable of reversibly modulating light transmission and absorption in response to an applied electrical bias1,2,3,4. By enabling real-time control over visible and near-infrared (NIR) light, ECDs offer a compelling solution for applications that demand adaptive light regulation, including smart building exteriors5, energy-saving windows6,7,8, optical modulators3,9,10, and wearable electronics11,12. Their low power consumption, solid-state nature, and scalability further support their integration into modern multifunctional systems without the need for mechanical components3,13,14,15.

Despite their promise, the widespread implementation of ECDs remains constrained by limitations in material performance and device design. Inorganic electrochromic materials, particularly tungsten oxide (WO₃), have long served as the cornerstone of ECD technologies due to their stability and relatively well-understood switching behavior16,17. However, they are inherently limited by sluggish response speeds, restricted color tunability, high processing temperatures, and mechanical rigidity4,18. These shortcomings hinder their use in flexible, large-area, or energy-autonomous systems. Even with ongoing efforts in structural engineering1, doping strategies19, and electrolyte optimization20,21,22, state-of-the-art inorganic ECDs typically exhibit switching times on the order of seconds, moderate coloration efficiencies (< 400 cm2 C⁻1), and noticeable degradation after several thousand operational cycles23,24,25,26,27.

Organic electrochromic devices (OECDs) offer a promising alternative, leveraging the structural tunability of organic semiconductors to achieve a broad spectral tunability, high coloration efficiencies, and compatibility with low-temperature, solution-based fabrication methods28,29,30,31. Their lightweight, flexible nature also makes them well-suited for integration into conformal and portable systems. Moreover, enhanced ionic mobility and the ability to engineer symmetric charge storage layers provide pathways toward faster switching and improved cycling stability32. However, OECDs are still constrained by trade-offs among optical contrast, switching speed, operating voltage range, and long-term stability, which limit their broader functional deployment.

To address these challenges, we present a complementary OECD that integrates two electrochemically compatible conductive polymers, poly(3,4-ethylenedioxythiophene)-poly(styrenesulfonate) (PEDOT:PSS) and poly(benzodifurandione) (PBFDO)33,34,35, into a high-performance bilayer architecture. This device exhibits a unique combination of key performance metrics, including high optical contrast (51% at 570 nm), ultrafast switching speeds (0.17 s for coloration and 0.36 s for bleaching), exceptional coloration efficiency (1688 cm2 C⁻1 at 550 nm), and robust cycling stability over 10,000 cycles at ± 1 V with negligible degradation. The flexible quasi-solid-state gel through in-situ polymerization eliminates the need for liquid electrolyte and enables scalable fabrication on both large-area (20 × 20 cm2) and flexible (10 × 10 cm2) substrates. Integrated into a vertically stacked smart window (Fig. 1a), the OECD is powered by a semitransparent organic solar cell (ST-OSC) module (active area 100 cm2), enabling autonomous switching under standard solar illumination. This self-powered system reduces interior temperatures by up to 7 °C under simulated sunlight, highlighting its potential for energy-efficient, solar-responsive architecture.

Fig. 1: Design and working principles of the multifunctional electrochromic platform.
Fig. 1: Design and working principles of the multifunctional electrochromic platform.
Full size image

a Schematic of the self-powered smart window, where a ST-OSC supplies the driving voltage for autonomous light and heat modulation by the OECD. b Device architecture of the on-chip OECS, comprising a monolithically integrated electrochromic filter array stacked over a FP cavity-enhanced OPD array.

Beyond building-scale applications, the voltage-tunable optical properties of the OECD enable new possibilities in compact and reconfigurable optical systems. With the growing demand for programmable alternatives to bulky components like prisms and gratings36,37,38,39,40,41,42,43, electrochromic materials, with their low power consumption, electrical tunability, and compactness, are well-suited for this role2,3. As shown in Fig. 1b, leveraging the high performance of our OECD architecture, we develop a microscale organic electrochromic spectrometer (OECS) featuring 16 pixels, each 4 mm2 and addressable across 14 discrete redox states. This configuration enables precise spectral filtering and accurate broadband light reconstruction with a mean squared error (MSE) as low as 0.001. Integrated with a 4 × 4 static Fabry Pérot (FP) cavity array and high-sensitivity organic photodetectors (OPDs) (with peak detectivity up to 1.6 × 1012 Jones), the OECS achieves a spectral resolution of 7.2 nm and over 99% classification accuracy across diverse organic solvents.

This work presents two applications of OECDs, a self-powered smart window and a miniaturized spectrometer, both based on voltage-controlled modulation of optical properties. The smart window enables adaptive light regulation for energy-efficient buildings, while the spectrometer achieves precise light filtering and spectral detection in a compact, portable format. Together, these results demonstrate a scalable electrochromic platform that bridges large-area energy management with high-resolution optical sensing, positioning OECDs as a foundational technology for next-generation adaptive optoelectronics and real-time environmental monitoring.

Results

Self-powered smart window

Device architecture and optimization of electrochromic components

To enable efficient light and heat modulation in the self-powered smart window, we first developed and optimized the underlying OECD, adopting a five-layer configuration: ITO/PEDOT:PSS/gel electrolyte/PBFDO/ITO (Fig. 2a). The two outer ITO layers function as transparent electrodes, while PEDOT:PSS and PBFDO serve as the cathodic and anodic electrochromic layers, respectively. A UV-curable ion-conducting gel (PC + PEGDA + LiTFSI + 1173 photoinitiator) is used as the intermediate quasi-solid electrolyte, facilitating reversible redox switching. Cross-sectional scanning electron microscopy (SEM) analysis (Fig. 2a) revealed well-defined layer thicknesses of 132 nm for PEDOT:PSS, and 116 nm for PBFDO, with field emission SEM (FESEM) images confirming the formation of dense, homogeneous, and amorphous electrochromic layers (Supplementary Fig. 1). The molecular structures of PEDOT:PSS and PBFDO are shown in Fig. 2b. As shown in Fig. 2c, both materials exhibit high transmittance in their bleached states, with minimal absorption near 506 nm and 574 nm, respectively, wavelengths close to the peak sensitivity (~550 nm) of human eye44. This spectral alignment ensures effective daylight modulation while maintaining visual comfort, making them well-suited for architectural applications.

Fig. 2: Structural design and process optimization of the OECD.
Fig. 2: Structural design and process optimization of the OECD.
Full size image

a Schematic illustration and cross-sectional SEM image of the multilayer OECD structure. b Chemical structures of PEDOT:PSS and PBFDO electrochromic polymers. c Transmittance spectra of PEDOT:PSS and PBFDO films. d Dependence of PBFDO film conductivity and transmittance at 550 nm on film thickness. e Electrolyte layer conductivity and sheet resistance as a function of thickness. f Effect of UV curing time on the electrolyte conductivity and sheet resistance. g Mechanical stability test of the electrolyte layer using a 10 × 10 cm2 flexible OECD with a curvature radius of 1 cm.

To achieve uniform and defect-free films, the blade-coating process was systematically optimized. Incorporating FS-30 surfactant into the PEDOT:PSS precursor improved substrate wetting and minimized edge shrinkage during high-temperature curing (Supplementary Fig. 2). Similarly, BYK-333 was added to the PBFDO solution to enhance film uniformity by suppressing dewetting behavior (Supplementary Fig. 3). However, elevated curing temperatures (> 60 °C) led to morphological defects due to rapid solvent evaporation (Supplementary Fig. 4). Electrical measurements identified 60 °C under ambient conditions as the optimal annealing temperature, yielding the highest conductivity of 1350 S cm⁻1 (Supplementary Fig. 5). Based on these findings, a curing protocol of 60 °C for 10 min was adopted for PBFDO layer. The thickness of the PBFDO layer was also tuned to balance optical transmittance and electrical conductivity. As shown in Fig. 2d and Supplementary Fig. 6, increasing the film thickness from 40 nm to 350 nm led to a monotonic decrease in visible–NIR transmittance, while the conductivity plateaued at 122 nm. An optimized range of 80–122 nm provided 60–70% transmittance at 550 nm and maintained high conductivity (~ 1351 S cm⁻1). Electrochromic performance at a ± 2 V bias (Supplementary Fig. 7) showed that this thickness range delivered the highest optical modulation, with ΔTmax values reaching 51% at 570 nm.

The quasi-solid-state electrolyte was optimized for both ionic transport and film morphology. A cured thickness of 200 μm and a UV exposure time of 7 s resulted in high conductivity (Fig. 2e, f) and a robust, flexible ion-transport layer (Fig. 2g). The 10 × 10 cm2 flexible OECD showed only a 7% optical contrast loss after 5000 bending cycles. This composition provided a stable electrochemical environment while ensuring conformal contact with the adjacent electrochromic layers. Overall, these optimizations in material formulation, curing conditions, and structural dimensions collectively enabled high-performance electrochromic switching at low voltage, while ensuring compatibility with scalable fabrication processes and flexible substrates, laying a solid foundation for integration into the self-powered smart window system.

Electrochromic performance of the complementary device

To understand the intrinsic electrochromic behavior of each active material, we first examined PEDOT:PSS and PBFDO in half-cell configurations: PE-OECD (ITO/PEDOT:PSS/electrolyte/ITO) and PB-OECD (ITO/PBFDO/electrolyte/ITO). As shown in Supplementary Fig. 8a and Supplementary Fig. 9a, both devices exhibited moderate optical modulation, with transmittance changes (ΔT) of 14.9% at 612 nm and 13.6% at 694 nm, respectively. The PE-ECD underwent a reversible transition from doped (transparent) to undoped (light blue) states, accompanied by a lightness (L*) drop from 81.2 to 57.3 and a slight blue hue (Supplementary Fig. 8b, c, Supplementary Fig. 10a and Supplementary Table 1). In contrast, the PB-ECD displayed a neutral gray-to-dark transition with a larger L* drop (80.8 to 41.3), but minimal chromatic shift (Supplementary Fig. 9b, c, Supplementary Fig. 10b and Supplementary Table 1). The reversible intercalation/deintercalation processes can be described as following equations:

$$ {{{{\rm{PEDOT}}}}}^{n+}\cdot {x{{{\rm{PSS}}}}}^{-}\cdot \left(n-x\right){{{{\rm{TFSI}}}}}^{-}+{n{{{\rm{Li}}}}}^{+} \\ + {n{{{\rm{e}}}}}^{-}\leftrightarrow {{{{\rm{PEDOT}}}}}^{0}\cdot {x{{{\rm{PSS}}}}}^{-}\cdot \left(n-x\right){{{{\rm{TFSI}}}}}^{-}\cdot {n{{{\rm{Li}}}}}^{+}$$
(1)
$$ {{{{\rm{PBFDO}}}}}^{n-}\cdot {x{{{\rm{H}}}}}^{+}\cdot \left(n-x\right){{{{\rm{Li}}}}}^{+}+{n{{{\rm{TFSI}}}}}^{-}-{n{{{\rm{e}}}}}^{-}\\ \leftrightarrow {{{{\rm{PBFDO}}}}}^{0}\cdot {x{{{\rm{H}}}}}^{+}\cdot \left(n-x\right){{{{\rm{Li}}}}}^{+}\cdot {n{{{\rm{TFSI}}}}}^{-}$$
(2)

Both systems exhibited fast response times (~5 s–8 s) and decent coloration efficiencies of 128 and 123 cm2 C⁻1 at 550 nm (Supplementary Fig. 8d, e and Supplementary Fig. 9d, e). However, their cycling stability was poor, ΔT dropped below 2% within 200–300 cycles (Supplementary Figs. 8f, 9f), primarily due to unbalanced charge storage at the inert ITO counter electrodes, which led to irreversible electrochemical degradation20,45. In fact, even when employing extremely thin PEDOT:PSS and PBFDO films (~20 nm) as counter electrodes, the ECD performance is significantly enhanced (Supplementary Figs. 11, 12, and Supplementary Table 1).

Subsequently, we constructed full-cells, combining PEDOT:PSS (cathodic) and PBFDO (anodic) in a symmetric configuration. This complementary pairing allows for balanced ion intercalation and charge compensation during switching. As shown in Fig. 3a the optimized OECD achieves a ΔTmax of 51% at 570 nm, significantly outperforming both half-cells. The full transmittance spectra at varying bias conditions are provided in Supplementary Fig. 13. Colorimetrically, the device undergoes a striking visual transformation. The CIE L*a*b* coordinates shift from (80.12, –0.25, 3.24) in the bleached state to (9.52, – 2.56, – 55.12) in the colored state (Fig. 3b and Supplementary Table 1), indicating a substantial decrease in brightness and a pronounced blue shift. This deep black-blue transition is further confirmed in the CIE xy diagram (Fig. 3c and Supplementary Table 1) and is easily visible to the naked eye. Notably, the OECD demonstrates exceptional electrochromic dynamics, with switching times of 0.17 s for coloration and 0.36 s for bleaching (Fig. 3d). Its coloration efficiency reaches 1688 cm2 C⁻1 at 550 nm (Fig. 3e), highlighting the highly efficient utilization of charge during redox cycling. The transmission spectrum (at 550 nm), charge accumulation (Q) and optical density (ΔOD) variations over time are shown in Supplementary Fig. 14. Energy consumption was calculated based on current–time (J–t) profiles in Supplementary Fig. 15 using the following expressions46:

$$P=V{J}_{{mean}}$$
(3)
$${J}_{{mean}}=\frac{{\int }_{{t}_{0}}^{{t}_{1}}{Jdt}}{\left({t}_{1}-{t}_{0}\right)}$$
(4)
Fig. 3: Performance characteristics of the complementary OECDs.
Fig. 3: Performance characteristics of the complementary OECDs.
Full size image

a Transmittance spectra of the OECD measured at − 2 V (colored state) and 2 V (bleached state); insets show corresponding photographs of the device at applied bias. b CIE L*a*b* color coordinates across a range of applied voltages. c Corresponding CIE xy coordinates over the same voltage range. d Transmission spectrum at 550 nm for the OECD in its colored and bleached states, at applied voltages of − 2 V and 2 V, respectively. e Plot of ΔOD at 550 nm as a function of injected charge density, used to calculate coloration efficiency. f Real-time transmittance modulation at 550 nm over multiple switching cycles. g Radar plot comparing key performance metrics against representative state-of-the-art ECDs31,64,65,66. h Estimated manufacturing cost comparison among our OECDs, transition metal (TM)-based ECDs, conductive polymer (CP)-based ECDs, and conventional ECDs, with a focus on smart window applications.

The required energy per switching cycle is 256.03 μW cm⁻2 for coloration and 237.95 μW cm⁻2 for bleaching, significantly lower than those of cholesteric LCDs (~ 1 mW cm⁻2), e-paper (~2 mW cm⁻2)45,47,48, or most reported ECDs16,49,50. This low power demand reinforces the device viability for autonomous, energy-efficient applications.

Durability testing at various voltages revealed remarkable stability. After 10,000 cycles at ± 1 V, 6000 cycles at ± 1.5 V, and 3000 cycles at ± 2 V, the device retained ΔT values of > 26%, 38%, and 50%, respectively (Fig. 3f and Supplementary Fig. 16), with minimal optical degradation. Given the exceptional electrochromic performance of complementary ECDs, we investigated the intrinsic properties of EC materials. Both polymers exhibit favorable energetic offsets (symmetric open-circuit and redox potentials) and matched ionic diffusion coefficients, enabling rapid and synchronized optical switching (Supplementary Fig. 17 and Supplementary Fig. 18). In addition, their high and comparable charge capacitance not only promotes redox kinetics that yields large optical modulation and high coloration efficiency, but also ensures efficient charge balance, reduces power consumption, and enhances cycling stability (Supplementary Fig. 19)33,51. As summarized in Fig. 3g, the device simultaneously achieves high optical contrast, ultrafast switching, low energy consumption, and long-term stability, key metrics for smart window deployment. A comparative summary against other state-of-the-art ECDs is provided in Supplementary Table 2, highlighting the superior performance of our system. Moreover, the estimated fabrication cost is only 78 USD m⁻2 (Fig. 3h and Supplementary Table 3), markedly lower than typical inorganic7, organic21 or conventional ECDs52, supporting its scalability for widespread architectural integration.

Scalability and thermal management performance of the smart window

Having established the high optical contrast, ultrafast switching speed, and robust cycling stability of our OECDs at a small scale (5 × 5 cm2), we next evaluated their scalability and real-world applicability. We first scaled up the OECDs from 25 cm2 to 100 cm2 (10 × 10 cm2) to evaluate their electrochromic performance at a practical size. As shown in Supplementary Fig. 20, the large-area OECD window exhibits uniform color switching at an applied voltage, indicating excellent electrochemical homogeneity. Figure 4c presents the dynamic transmittance change at 550 nm measured at the center and edge of the device at ±2 V bias. The transmittance decreases from 57% to 15% within 2 s and further drops to 5% after 5 s, demonstrating high optical contrast and fast switching kinetics at a large scale. The overlapping curves from different spatial locations confirm minimal voltage drop and excellent electrochromic uniformity. Full transmission spectra from 400 to 2000 nm at various voltages are provided in Supplementary Fig. 21. To further demonstrate scalability, a 20 × 20 cm2 device fabricated via blade coating also showed consistent and homogeneous switching behavior (Fig. 4a). Furthermore, a flexible 10 × 10 cm2 device retained stable electrochromic response under bending stress (Fig. 4b), underscoring its mechanical robustness and compatibility with flexible substrates.

Fig. 4: Scalability and thermal shielding performance of the OECD.
Fig. 4: Scalability and thermal shielding performance of the OECD.
Full size image

a Photographs of a 20 × 20 cm2 OECD fabricated via blade coating. b A flexible 10 × 10 cm2 device under mechanical bending. c Dynamic transmittance at 550 nm measured at the center and edge of the device during a full switching cycle (from − 2 V to 2 V). d Solar irradiance spectra derived from the measured transmittance spectra at different bias voltages. e Indoor temperature setpoints used in EnergyPlus simulations to replicate real-world climate control scenarios. f Simulated annual cooling energy savings for an office building in Guangzhou using the OECD at different operational states, benchmarked against commercial low-E glazing. g Global simulation results showing absolute (MJ m⁻2) and relative (%) reductions in annual cooling energy consumption when replacing low-E glass with the OECD ( − 2 V), across diverse climate zones.

In addition to visible light modulation, a key advantage of electrochromic windows lies in their ability to regulate solar heat gain, particularly in the NIR region. We therefore evaluated the thermal shielding performance of the smart window by converting the transmittance spectra into solar-weighted irradiance profiles using AM1.5 G conditions (Fig. 4d). At a bias of −0.5 V, the device effectively blocks 92.4% of NIR solar irradiance within the 780–2000 nm range. Even in the fully bleached state (at 2 V), the device still blocks 79.3% of NIR radiation (Supplementary Table 4). Furthermore, at a bias of − 2 V, the device blocks 89.2% of total solar energy. Although the modulation of visible light reaches 40% at ± 2 V, the NIR transmittance exhibits only a slight change of ~3%, enabling a “cool mode” operation that maintains indoor illumination while minimizing solar heating. The thermal shielding efficiency (SEheat) was calculated as the percentage of NIR solar irradiance blocked, using the following equations45:

$${{SE}}_{{{{\rm{heat}}}}}=100-{TE}$$
(5)
$${I}_{{{{\rm{T}}}}}={\sum }_{780}^{2000}\varphi (\lambda )\varDelta \lambda$$
(6)
$${TE}=\frac{{I}_{{{{\rm{T}}}}}}{{I}_{{{{\rm{total}}}}}}\times 100\%$$
(7)

where IT and Itotal refer to the transmitted solar irradiance and total solar irradiance. φ(λ) is the transmitted solar irradiance spectrum at AM1.5 G conditions. TE is transmission efficiency. Detailed results are summarized in Supplementary Table 5, revealing that SEheat values range from 78.7% to 92.1% across different voltage states, with most values exceeding 90%, thereby confirming the excellent heat-blocking performance of the OECD smart window.

To quantify the potential energy-saving benefits of our OECD smart windows, we conducted simulations using EnergyPlus software, incorporating both the measured optical properties and regional climate data. A standard office building (15 × 10 × 3 m3) with eight 3 × 2 m2 window panels was modeled (Supplementary Fig. 22 and Supplementary Table 6). Indoor temperature and lighting conditions were set to reflect real-world usage. As shown in Fig. 4e, the simulation framework captures seasonal variations in solar load and cooling demand. Taking Guangzhou as a representative subtropical city, the simulated annual cooling energy savings reached 228 MJm⁻2 in the colored state (−2 V) and 171 MJ m⁻2 in the bleached state (2 V), compared to conventional low-emissivity (low-E) glass (Fig. 4f). Simulations across diverse global climates confirmed consistent energy savings, with up to 594 MJ m⁻2 savings in Brasília and a 50.9% reduction in annual cooling load in Reykjavík (Fig. 4g). The full simulation data are presented in Supplementary Table 7, collectively demonstrating the broad applicability of the OECD smart window across different environmental contexts.

ST-OSC powered smart windows

To further enhance the autonomy and practicality of the OECD system, we integrated it with an ST-OSC, creating a fully self-powered smart window. This integrated design eliminates the need for external power sources, enabling the system to dynamically regulate light and heat transmission in response to ambient sunlight. Figure 5a schematically illustrates the device architecture, where the ST-OSC serves dual roles: as a photovoltaic power supply and a spectral modulator, vertically stacked directly above the OECD in a compact monolithic structure. Upon illumination, the ST-OSC generates sufficient voltage to switch the OECD between its bleached and colored states, ensuring seamless autonomous operation.

Fig. 5: Performance of the ST-OSC powered large-area smart window.
Fig. 5: Performance of the ST-OSC powered large-area smart window.
Full size image

a Schematic illustration of the monolithically integrated ST-OSC–OECD device and its self-powered electrochromic switching mechanism. b J–V characteristics of ST-OSCs with varying Ag electrode thicknesses and MoOX antireflection layers. c Corresponding EQE and transmittance spectra for ST-OSC devices in b. d CIE xy chromaticity coordinates of ST-OSCs: red (10 nm Ag), green (15 nm Ag), blue (15 nm Ag + 30 nm MoOX), and yellow (30 nm Ag). e J–V curve of a large semitransparent module (active area 100 cm2). f Photographs of a scaled model house under three conditions: (i) no device, (ii) ST-OSC only, and (iii) integrated ST-OSC–OECD operating in the colored state.

The ST-OSC features an ITO/PEDOT:PSS/bulk heterojunction (BHJ)/PNDIT-F3N/Ag/MoOX structure, with a ternary BHJ active layer comprising PTzBI-Cl and BTR-Cl donors, and DT-Y6 acceptor53. Figure 5b, c present the current density–voltage (J–V) characteristics, external quantum efficiency (EQE), and transmittance spectra for devices with varied silver (Ag) electrode thicknesses. Increasing Ag thickness from 10 nm to 30 nm boosts short-circuit current density (JSC) from 19.2 to 24.4 mA cm⁻2 but reduces average visible transmittance (AVT) from 27.9% to 7.1% (Supplementary Table 8). This trade-off arises from increased reflectivity in thicker Ag layers, which enhances charge generation but diminishes light utilization efficiency (LUE). To mitigate this trade-off, we introduced a high-refractive-index MoOX antireflection coating atop the Ag layer. Compared to the 15 nm Ag-only device, the MoOX-coated device exhibits significantly improved visible transmittance with minimal losses in JSC (0.27 mA cm⁻2) and EQE (0.22%), representing a favorable compromise (Fig. 5b, c and Supplementary Table 8). The MoOX layer effectively suppresses reflection in the visible range while maintaining efficient charge extraction and spectral absorption. The optimized configuration, 15 nm Ag with 30 nm MoOX, achieves a maximum power conversion efficiency (PCE) of 12.97% and an LUE of 3.72. Colorimetric analysis (Fig. 5d) reveals that increasing Ag thickness shifts device appearance from white to blue, with the optimized device delivering a high color rendering index (CRI) of 86.4 and CIE chromaticity coordinates of (0.2376, 0.2700) (Supplementary Table 8). Based on the optimized small-area device, we fabricated a large-area ST-OSC module (active area 100 cm2) to power an OECD of matching size. As shown in Fig. 5e and Supplementary Table 9, the module exhibits a VOC of 10.06 V, JSC of 1.31 mA cm⁻2, PCE of 6.72%, AVT of 33.7% (Supplementary Fig. 23), and LUE of 2.26, confirming its capability to effectively drive the OECD. The integrated system transitions from a light-blue transparent state to a dark-blue colored state under illumination (Supplementary Fig. 24 and Supplementary Movie 1 and 2), demonstrating functional synergy.

Finally, a scaled-down model house experiment was conducted to evaluate the system thermal regulation capabilities under realistic conditions (Fig. 5f). The interior temperatures recorded were 33 °C without any device, 30.1 °C with the ST-OSC alone, and 26 °C with the integrated ST-OSC–OECD system, demonstrating significant improvements in thermal insulation. This cooling effect results from the combined contributions of the high infrared reflectance of the ST-OSC (IRR = 75.4%)53 and the strong suppression of NIR solar transmittance in the colored state OECD. Collectively, these results highlight the feasibility of a fully self-powered, energy-autonomous smart window system that dynamically modulates light and heat to enhance indoor comfort and reduce building energy consumption. By integrating the OECD with a high-efficiency ST-OSC, the system autonomously controls optical transmission while lowering indoor temperature, underscoring its potential for large-scale, sustainable deployment.

On-chip organic electrochromic spectrometer (OECS)

Building on the scalability, tunable optics, and self-powered operation demonstrated in OECD-based smart windows, we translate the platform into chip-scale, voltage-programmable spectrometers. By monolithically integrating electrochromic filters with high-sensitivity OPDs, these devices provide reconfigurable spectral filtering and computational reconstruction. This transition from meter-scale dynamic windows to millimeter-scale photonic sensors underscores the versatility and cross-scale adaptability of the PEDOT:PSS–PBFDO electrochromic platform.

Working principle of the OECS

The OECS operates via a hybrid dual-mode modulation scheme. Co 2 V, while fine spectral resolution is obtained by incrementally addressing 14 discrete voltages between − 2 V and 2 V. The resulting photocurrent \({I}_{m}^{n}\) at pixel m at bias voltage n follows the integral expression,

$${I}_{m}^{n}={\int }_{{\lambda }_{1}}^{{\lambda }_{{{{\rm{m}}}}}}{T}_{m}^{n}(\lambda )\times R(\lambda )\times S(\lambda )d\lambda$$
(8)

where \({T}_{m}^{n}({{{\rm{\lambda }}}})\) the voltage-dependent transmittance of the electrochromic pixel, R(λ) is the wavelength-dependent responsivity of the OPD, and S(λ) is the unknown incident light spectrum. With this forward model and a library of calibrated transmission spectra, the input spectrum can be reconstructed using computational inversion techniques. The responsivity calibration of the OPDs is shown in Supplementary Fig. 25. As illustrated in Fig. 6a, the OECS leverages a hybrid modulation strategy that combines broadband, voltage-tunable absorption from the electrochromic layer with pixel-specific static filtering from integrated FP cavities. The FP cavity is nanometer-scale, eliminating the need for additional optical components and enabling a compact all-solid-state spectral device. Each OECD pixel contains an Ag-based FP cavity with a tunable spacer layer. Figure 6b displays the 175 nm spacer layer electric field intensity and the operational range spans 50–425 nm. To validate the microcavity effect, additional TMM simulations were performed across different spacer thicknesses. The results show clear resonance peaks whose positions and intensities shift consistently with spacer variations, confirming tunable spectral selectivity of the FP cavities (Supplementary Fig. 26). The complex refractive indices of each material layer were extracted and used in transfer matrix method (TMM) simulations (Supplementary Fig. 27)54. Figure 6c presents the normalized responsivity of the OPD after FP cavities processing, as the spacer layer thickness varies linearly from 0 nm to 450 nm. FP cavities introduce sharp resonance peaks within the 380–1180 nm range, enhancing spectral discrimination and reducing crosstalk. The OPDs adopt a n–i–p architecture: ITO/ZnO/PCE-10:COTIC-4F:Y6/MoOX/Ag, optimized for broadband responsivity55. The structural relationship between the OECD and the FP cavities is illustrated in Supplementary Fig. 28. All 16 pixels in the OPD array exhibit uniform dark current densities on the order of 10⁻10 A cm⁻2 at zero bias (Fig. 6d), demonstrating excellent pixel uniformity. The measured noise current spectra (Fig. 6e) show flat noise characteristics over the 20–50 Hz range, with an average noise density of 6.9 × 10⁻15 A Hz⁻1/2. The specific detectivity (D*) was calculated using the standard formulation56:

$${D}^{*}=\,\frac{R\sqrt{A\Delta f}}{{i}_{{{{\rm{noise}}}}}}\,=\,\frac{q\lambda \sqrt{A\Delta f}{EQE}}{{hc}{i}_{{{{\rm{noise}}}}}}$$
(9)

where q is the electron charge, A is the effective area (1.1 mm2), and Δf is the electrical bandwidth (1 Hz). The OPD achieved a peak D* of 1.6 × 1012 Jones (Fig. 6f), placing it among the best-performing organic devices in this class. In addition, Supplementary Fig. 29 shows the OPDs exhibit a fast response with rise and fall times of 14.9 μs and 15.2 μs, respectively, and a − 3 dB cutoff bandwidth (f−3dB) of 101 kHz (Fig. 6g). The linear dynamic range (LDR), calculated as refs. 57,58:

$${LDR}=20{\log }_{10}\left(\frac{{I}_{\max }}{{I}_{\min }}\right)$$
(10)

where Imax is the maximum unsaturated photocurrent and Imin is the noise-limited detection threshold. The OPD achieves a broad LDR of 152 dB, indicating its ability to resolve optical signals across a wide intensity range (Fig. 6h)59,60,61. Figure 6i exhibits a distinct transmission spectrum as a function of applied voltages, providing a rich library of optical fingerprints for multi-dimensional sampling. This synergy between dynamic (EC) and static (FP cavity) filtering enables accurate spectrum reconstruction from a small number of physical pixels. Fabricated FP cavity and the monolithic OPD array are shown in Supplementary Fig. 30.

Fig. 6: Working principle of the OECS.
Fig. 6: Working principle of the OECS.
Full size image

a Schematic illustration of the OECS working principle based on voltage-dependent transmission and computational spectrum reconstruction. b Optical electric field distribution inside the FP cavity. c FP cavity-modulated responsivity, enabling wavelength-selective filtering with complementary spectral bands. d Dark J-V characteristics of the 16-pixel OPD array. e Noise spectrum of the OPD. f D* of OPDs calculated based on the measured noise. g −3 dB cutoff bandwidth and h LDR of OPD. i Normalized responsivity distribution with broadband, voltage-tunable transmittance after EC filter in OECD.

Voltage-encoded multidimensional spectral sampling and reconstruction

Different pixels measured at 14 bias voltages exhibit distinct EQE responses (Supplementary Fig. 31), enabling sufficient spectral information for reconstructive spectroscopy. Building upon the hybrid dual-mode modulation strategy and calibrated responsivity matrix \({R}_{m}^{n}(\lambda )\), we implemented computational spectral reconstruction to evaluate the practical performance of the on-chip OECS under diverse illumination conditions (Fig. 7). This assessment encompasses both narrowband and broadband light inputs, showcasing high spectral resolution, reconstruction accuracy, and adaptability of the system. We first validated the reconstruction accuracy under quasi-monochromatic illumination (3 nm bandwidth) spanning the 300–1200 nm range. As shown in Fig. 7a, the reconstructed spectra exhibit excellent agreement with those measured by a commercial grating spectrometer, demonstrating precise recovery of both peak positions and relative intensities. This confirms the effectiveness of our calibrated forward model across a broad spectral range.

Fig. 7: Spectral reconstruction capabilities of the on-chip OECS.
Fig. 7: Spectral reconstruction capabilities of the on-chip OECS.
Full size image

a Reconstructed spectra of quasi-monochromatic light using the OECS, compared with reference spectra from a commercial grating spectrometer. b Spectral resolution validation showing successful discrimination of two peaks separated by 7.2 nm. c Accuracy assessment of reconstructed peak wavelengths for 1 nm-separated NIR sources, showing minimal deviation from ground-truth values. d Reconstruction of broadband spectra with a single dominant peak. e Reconstruction result of a complex broadband spectrum using a single OECS (OECD-only). f Accurate reconstruction of a cold white LED spectrum using a 16-pixel OECS array with hybrid (OECD + FP cavity array) modulation. g MSE of spectral reconstruction as a function of the number of voltage sampling points.

To evaluate spectral resolution, we tested the ability to distinguish closely spaced wavelengths of the OECS. As shown in Fig. 7b, two peaks centered near 620 nm and separated by just 7.2 nm were clearly resolved. Further testing with 1 nm-spaced peaks in the NIR range (920–930 nm, Fig. 7c) yielded a mean deviation of only 0.29 nm between reconstructed and reference spectra, highlighting the high-resolution capability of the system even in the NIR range. We then assessed reconstruction performance for broadband light sources. For single-peak spectra such as filtered LED emissions, the OECS accurately reproduced the full profile (Fig. 7d). However, reconstructing complex broadband spectra remains a fundamental challenge for miniaturized systems62, particularly when relying on single-pixel or sequentially tuned devices due to their limited encoding diversity. To investigate this limitation, we compared the reconstruction fidelity between two architectures: a single-pixel OECD device and a multi-pixel OECS with integrated FP cavities. As shown in Fig. 7e, the single OECD (OECD-only), despite voltage tunability, exhibited poor reconstruction accuracy due to low-dimensional spectral encoding and a lack of spatial or spectral orthogonality. In contrast, the hybrid OECS (OECD + FP cavity array) substantially improved reconstruction quality, accurately recovering complex LED-filtered spectra as shown in Fig. 7f. This enhancement arises from the hybrid device capacity for parallel, multi-voltage, and spectrally selective sampling, enabling high-dimensional encoding across the array.

To quantify the effect of sampling dimensionality, we examined reconstruction error as a function of applied voltage steps. As shown in Fig. 7g, increasing the number of sampling voltages significantly reduced the MSE. With more than four voltages, the MSE dropped below 0.001, underscoring the importance of dense voltage sampling in enhancing reconstruction fidelity. The reduction in MSE with multiple voltage steps arises from enhanced spectral encoding. Each voltage slightly shifts the FP cavity resonance, causing a single OPD to produce distinct responses. Correlation analysis of a representative OPD (Supplementary Fig. 32) shows that additional voltage steps improve response distinguishability, providing richer measurement information for the reconstruction algorithm and directly lowering the MSE. These results clearly demonstrate that transitioning from single-pixel to array-based hybrid modulation not only improves resolution and accuracy but also enables robust, high-throughput spectral analysis in compact form factors. Using 14 voltage-modulated cavity states as sampling points, the spectrometer maintains <0.005 MSE in spectrum reconstruction even when 20% Gaussian white noise is added (Supplementary Fig. 33). Overall, the OECS combines active voltage-controlled filtering, static FP cavity selection, and computational inversion to deliver high-resolution, miniaturized spectrometry, a capability essential for next-generation portable sensing and imaging systems.

Application of OECS in material identification

To demonstrate the practical utility of the on-chip OECS beyond laboratory spectral reconstruction, we explored its application in real-world chemical analysis. Specifically, we focused on material composition identification via absorption spectroscopy, a standard analytical technique typically dependent on bulky, high-cost instruments. By leveraging the miniaturized form factor of OECS, reconfigurable spectral sampling, and broad operating range, we developed a compact transmission-mode setup for broadband absorption fingerprinting of organic materials.

As shown in Fig. 8a, the experimental setup comprises a broadband micro-LED light source, a cuvette containing either the sample or air (reference), and a narrow-angle engineered diffuser was incorporated to homogenize the angular distribution of the incident light, followed sequentially by the OECS. By maintaining near-normal incidence across all FP cavities, the diffuser ensures a consistent spectral baseline without significantly broadening the FP resonance peaks, thus preserving the intrinsic spectral resolution critical for accurate reconstruction. Figure 8b shows the OECS chip and the chip in a metal-encapsulated module for measurement. For each measurement, the device captured photocurrent responses corresponding to the sample spectrum Ssample(λ) and reference spectrum Sref(λ). These signals were then reconstructed into full spectra using the pre-calibrated responsivity matrix \({R}_{m}^{n}(\lambda )\). The assumption of negligible sample reflectance, analogous to conventional UV–Vis–NIR spectrometry, allows the transmittance T(λ) and absorbance A(λ) to be determined via the Beer–Lambert law63:

$$T(\lambda )=\,{S}_{{{{\rm{sample}}}}}(\lambda )/{S}_{{{{\rm{ref}}}}}(\lambda )$$
(11)
$$A(\lambda )=\,-{\log }_{10}(T)={\log }_{10}(1/T)$$
(12)
Fig. 8: Material identification using the on-chip OECS.
Fig. 8: Material identification using the on-chip OECS.
Full size image

a Schematic illustration of the transmission-mode measurement setup. b Photographs of the OECS chip and a chip in a metal-encapsulated module. c Four organic semiconductor material solutions. d Reconstructed absorption spectra obtained with the OECS, compared with reference spectra measured by a commercial spectrometer. e Classification results based on the reconstructed spectra, represented as spectral response weight vectors for each sample.

To validate the performance, we measured the absorption spectra of four common organic dye solutions (10 mg mL−1), each with similar visible appearance but distinct molecular signatures (Fig. 8c). As shown in Fig. 8d, the reconstructed absorbance spectra from the OECS closely matches those obtained using a commercial grating spectrometer, with accurate peak positions and relative intensities, demonstrating high-fidelity spectroscopic capability. As a proof-of-concept for material classification, we tested the ability of our OECS to differentiate among the four dyes, despite their partially overlapping absorption bands. Thanks to the high spectral resolution and encoding diversity, the OECS successfully resolved subtle spectral differences and enabled unambiguous identification of each material. As shown in Fig. 8e, a classification model based on the reconstructed spectra achieved prediction accuracy exceeding 99%, confirming the system potential for reliable, compact, and cost-effective optical analysis in portable settings.

Discussion

In this work, we present a versatile and scalable organic electrochromic platform that bridges macroscale energy-saving applications and microscale spectral sensing technologies. Leveraging the complementary electrochromic properties of PEDOT:PSS/PBFDO-based OECDs, we realize two structurally unified yet functionally distinct systems: large-area self-powered smart windows and miniaturized, high-resolution spectrometers. As smart windows, the OECDs demonstrate outstanding electrochromic performance, including high optical contrast (51% at 570 nm), ultrafast switching speeds (0.17/0.36 s coloration/bleaching), exceptional coloration efficiency (1688 cm2 C⁻1 at 550 nm), and robust cycling stability (over 10,000 cycles at ± 1 V). Operating at ultralow power, these windows effectively regulate indoor temperature, achieving up to a 7 °C reduction when integrated with self-powered ST-OSCs. The successful fabrication of large-area and flexible devices further confirms the platform scalability and mechanical robustness, underscoring its potential for next-generation energy-efficient architectural integration. At the microscale, we introduce a compact OECS that synergistically combines voltage-tunable electrochromic filtering and fixed FP cavity modulation integrated with OPDs. This dual-mode approach enables broadband (300–1200 nm) spectral detection with high resolution (7.2 nm peak separation) within a miniaturized 4 mm2 pixel footprint. The OECS achieves precise spectral reconstruction (mean wavelength deviation of 0.29 nm) and maintains excellent fidelity under diverse illumination conditions (MSE < 0.001 with ≥ 4 voltage sampling points), rivaling commercial spectrometers. Its practical utility is validated by high-accuracy (>99%) material identification of spectrally similar organic compounds via absorption spectroscopy. By unifying high-performance electrochromic modulation, scalable fabrication, and computational reconstruction, this platform overcomes key limitations of conventional spectroscopy, including bulkiness, high power consumption, and limited tunability.

All three devices, the standalone OECD, the ST-OSC-powered smart window, and the on-chip OECS, share the same electrochromic material system but are tailored for distinct functions. The OECD benchmarks device performance, the ST-OSC smart window demonstrates scalable, self-powered light and heat regulation, and the on-chip OECS enables precise spectral sensing. Together, they showcase the platform’s scalability, functional versatility, and potential for technology transfer from fundamental materials to integrated systems.

Methods

Materials

All chemical reagents used in this study were of analytical grade and applied without further purification. PBFDO was sourced from Jurong Optoelectronic Materials Technology Co., Ltd. PEDOT:PSS (PH1000) was obtained from Heraeus Co., Ltd. Propylene carbonate (PC), Lithium Bis(trifluoromethanesulfonyl)imide (LiTFSI) and 2-Hydroxy−2-methyl-1-phenylpropan-1-one (1173 photoinitiator) were purchased from Energy Chemical Co., Ltd. Poly(ethylene glycol) diacrylate (PEGDA) was purchased from Aladdin Co., Ltd. Polyether-modified polydimethylsiloxane (BYK-333) was purchased from BYK Co., Ltd. Fluorosurfactant (FS-30) was purchased from Chemours Co., Ltd. ITO glasses (10 Ω Sq−1) were purchased from Foshan Jingjiexin Glass Co., Ltd. The OPD array fabrication employed commercially available semiconductor materials, including the donor polymer PCE-10 and the non-fullerene acceptors COTIC-4F and Y6, which were procured from established materials suppliers (1-Material Inc., Solarmer Materials Inc., Macklin Biochemical Technology Co., and Dongguan Volt Ampere Photoelectric Technology Co., Ltd.).

Preparation of electrochromic film and electrolyte

ITO-coated glass substrates were sequentially cleaned via ultrasonication in soap solution, deionized water, acetone, and isopropanol, each for 15 min, to ensure a pristine surface. After drying under a nitrogen stream, the substrates were treated with oxygen plasma for 20 min to enhance surface hydrophilicity and adhesion. To prevent degradation of plasma-induced surface modifications, coating was performed immediately after treatment. Large-area EC films (5 cm × 5 cm, 10 cm × 10 cm, 11.5 cm × 12 cm, and 20 cm × 20 cm) were fabricated using a blade-coating and heat-curing method. For PEDOT:PSS film preparation, DMSO (5 wt%) and FS-30 (with a PEDOT:PSS-to-FS-30 dry weight ratio of 1:0.05) were sequentially added to PH1000 ink under magnetic stirring to enhance film conductivity and uniformity. The ink was then blade-coated using a four-sided applicator with a 50 μm gap. The wet films were subsequently dried in an oven at 110 °C for 5 min. To prepare PBFDO films, a commercially available PBFDO solution (12 mg mL−1 in DMAC) was diluted to 4 mg mL−1 with DMAC. BYK-33 was added (PBFDO-to-BYK-33 dry weight ratio of 1:0.1) under magnetic stirring to improve film uniformity. The mixture was blade-coated using a 100 μm applicator gap and dried at 60 °C for 10 min to yield the final film. Small-area EC films (1.5 cm × 1.5 cm) were fabricated using the same procedures, except spin coating was used instead of blade coating. Specifically, spin-coating parameters for PEDOT:PSS and PBFDO films were set to 1000 rpm for 10 s and 3000 rpm for 30 s, respectively. The electrolyte was prepared by stirring a solution containing PC (46.3 wt%), PEGDA (43.2 wt%), LiTFSI (2.2 wt%), and photoinitiator 1173 (8.3 wt%) until fully dissolved.

Material mechanism characterization

CV measurements were performed using a CHI660E electrochemical workstation. The electrochromic films (area 1.5 × 1 cm2) were characterized were carried out in a three electrode cell configuration using ITO/glass (for PEDOT:PSS films) or ITO/glass sheet (for PBFDO films) as the working electrode, Pt wire as the counter electrode, and Ag/AgCl as the reference electrode.

Fabrication of OECD

A double-sided conductive copper tape (thickness: 50 μm) was affixed to the four edges of the ITO/PBFDO glass substrate to minimize voltage drop. Notably, a segment of the copper tape was extended outward to facilitate connection with the external circuit. Subsequently, commercial 3 M polyimide tapes with thicknesses of 50 μm and 100 μm were sequentially applied around the perimeter of the substrate, forming a final spacer gap of 200 μm. The gel electrolyte was then injected into the device using a syringe. Thereafter, the ITO/PEDOT glass was gently placed on top, and any trapped air bubbles were carefully expelled. The assembled device was secured with clips and cured under UV light for 7 s. For the half-cell configuration, the assembly procedure remained the same, except that the counter electrode was replaced with bare ITO glass.

Preparation and integration of OPD and FP cavity

ITO-coated glass substrates were sequentially cleaned by ultrasonication in isopropanol, detergent solution, and deionized water, followed by a final isopropanol rinse. After drying at 80 °C for 2 h, substrates were treated with oxygen plasma (120 s) to enhance surface wettability. For the electron transport layer, a ZnO sol–gel precursor was prepared by dissolving zinc acetate (0.4 g) and ethanolamine (110 μL) in 2-methoxyethanol (4 mL), stirred at 55 °C for 10 h. The precursor was spin-coated (3300 rpm, 45 s) and annealed at 150 °C for 30 min to form a compact ZnO layer.

The photoactive layer comprised a ternary blend of PCE-10:COTIC-4F:Y6 (1:1.05:0.45 by weight) in chlorobenzene (25 mg mL⁻1) with 0.2 vol% 1-chloronaphthalene as a processing additive. After overnight stirring at 60 °C, the solution was spin-coated in two steps—800 rpm for 40 s (spreading) and 2000 rpm for 40 s (film formation)—yielding an active layer ~ 270 nm thick. A post-deposition annealing step at 100 °C for 10 min enhanced molecular ordering. Finally, a 10 nm organic hole transport layer and 100 nm Ag top electrode were thermally evaporated under high vacuum (9 × 10⁻7 Torr) to complete the device.

The FP cavity arrays were monolithically integrated with the OPD array via precision-controlled thermal evaporation. Each cavity consisted of two 13 nm Ag mirrors and an Alq₃ spacer layer, with the latter precisely tuned between 50–425 nm to define spectral response. A 7 nm MoOX interlayer was inserted to improve Ag mirror morphology and optical performance. Pixel-level patterning and nanometer-scale thickness control were achieved using custom shadow masks and real-time quartz crystal monitoring, allowing sequential, vacuum-integrated deposition without interruption. Devices were hermetically encapsulated using UV-cured epoxy and glass coverslips for environmental stability. Electrical interfacing was established through spring-loaded pogo pins, connecting the chip to a custom-designed PCB with integrated readout electronics. The complete assembly was housed in a compact mechanical enclosure, resulting in a fully portable miniature spectrometer system. For optical characterization, additional active layer solutions were prepared by dissolving PCE-10 (10 mg mL⁻1) and IEICO-4F (15 mg mL⁻1) in chlorobenzene, stirred at 60 °C for 2.5 hours, and deposited via optimized spin-coating protocols. This integrated fabrication strategy enables wafer-scale production of spectrally selective photodetector arrays, fully compatible with standard optoelectronic packaging, while eliminating alignment challenges typically associated with discrete optical components in miniaturized spectrometer systems.

Material and device characterization

The UV–Vis–NIR transmittance and absorption spectra of the OECD were measured using a UV-3600i Plus spectrophotometer. Polymer surface morphology was characterized via field emission scanning electron microscopy (FESEM, Merlin), while cross-sectional images were obtained using a SU8600 system. Cyclic stability was evaluated using an electrochromic device cyclic tester (KV-EC-7500, Kaivo, Zhuhai). Colorimetric properties in the CIE L*a*b* space were assessed with a Benchtop Grating Spectrophotometer (LS173, Linshang, Shenzhen), and CIE xy coordinates were measured using LCE Intelligent Detection Instrument (HP350C, Shuangse, Hangzhou). Electrical conductivity (σ) was calculated using the following formula:

$$\sigma=\frac{1}{{R}_{{{{\rm{s}}}}}t}$$
(13)

where σ denotes the electrical conductivity, t is the film thickness measured by a profilometer, and Rs is the sheet resistance identified by a Four Probe Tester. The coloration efficiency (CE) at a given wavelength was determined by the following formula:

$${CE}\left(\lambda \right)=\frac{\Delta {{{\rm{OD}}}}}{Q}=\frac{\log \left(\frac{{T}_{{{{\rm{b}}}}}}{{T}_{{{{\rm{c}}}}}}\right)}{Q}$$
(14)

where ΔOD represents the change in optical density, Q is the injected/extracted charge, and Tb and Tc refer to transmittance in the bleached and colored states, respectively. The gel electrolyte was UV-cured using a DH48S LED curing system (Deshengxing, Shenzhen). Chronoamperometry and chronocoulometry measurements were performed on a CHI660E electrochemical workstation. Dark J–V characteristics were recorded using a high-precision semiconductor parameter analyzer (PDA FS380 Pro, Platform Design Automation). EQE spectra were acquired using a custom-built system consisting of a stabilized halogen lamp (7IPT250B, SOFN Instruments), a low-noise current preamplifier (SR570, Stanford Research Systems), and a lock-in amplifier (SR830, Stanford Research Systems), with spectral calibration performed using reference Si and InGaAs PDs. Frequency response measurements employed a 1050 nm LED modulated by a function generator (AFG1062, Tektronix), and the transient photocurrent was captured using a 350 MHz digital oscilloscope (MDO3052, Tektronix). For LDR evaluation, the illumination intensity from the 1050 nm LED, driven by a Keithley 2450 SourceMeter, was precisely modulated and measured using a calibrated optical power meter (1919-R, Newport) with a Si photodiode (818-UV/DB), while photocurrent was simultaneously recorded by the FS380 Pro analyzer. Reference spectral data of incident light were acquired with a high-end UV–Vis–NIR spectrometer (UV-3600, SHIMADZU). Quasi-monochromatic illumination was provided by a filtered broadband source, and complex spectra were generated using either a xenon arc lamp or a cold white LED, enabling comprehensive characterization across the visible to NIR range under controlled ambient conditions with appropriate light shielding.

Optical simulation

The optical characteristics of the Fabry-Pérot (FP) cavity were numerically modeled using the transfer matrix method (TMM), a rigorous computational technique for analyzing multilayer optical systems. The wavelength-dependent complex refractive index (ñ = n + ik) of the Alq₃ spacer layer, a key input for the simulation, was experimentally obtained via spectroscopic ellipsometry using a ME-L ellipsometer (Wuhan Eoptics Technology). This method enabled accurate extraction of both the refractive index (n) and extinction coefficient (k) across the relevant spectral range, facilitating precise modeling of light propagation and interference effects within the cavity. By combining first-principles optical calculations with experimentally validated material parameters, the simulation provides reliable predictions of the FP cavity’s resonant behavior and spectral response.