π–π Electronic Coupling and Gap‐Plasmonic Enhancement via Fe@C_x Nanoparticles: Synergistic Hole‐Transport Engineering for High‐Responsivity Perovskite Photodetectors

ABSTRACT

We report a multifunctional interface‐engineering strategy in which carbon‐encapsulated iron nanoparticles (Fe@C_x NPs) are blended into poly(3,4‐ethylenedioxythiophene): polystyrene sulfonate (PEDOT:PSS) to co‐optimize energetics, charge transport, and optical field confinement. Organic‐inorganic hybrid optoelectronic interfaces suffer from energetic misalignment and charge transport limitations. This study addresses these challenges through carbon‐encapsulated iron nanoparticles (Fe@Cx NPs) derived from carbon nanotube synthesis byproducts, which modulate the work function of poly(3,4‐ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) from 5.03 to 5.24 eV via π–π electronic coupling, representing a 0.21 eV improvement, with optimized energy level alignment of the highest occupied molecular orbital. When the concentration of incorporated iron nanoparticles is optimal at 2.5 v%, the dark current of the devices reduces by 87% from 2.12 × 10^{− 8} to 2.71 × 10^{− 9} A/cm², and a 53% decrease in defect‐state energy from 2.31 to 1.08 meV. Under self‐powered conditions (0 V), the responsivity increases by 13% from 0.38 to 0.43 A/W, whereas shot‐noise‐limited detectivity improves 3.2‐fold from 4.65 × 10^{1 2} to 1.49 × 10^{1 3} Jones. Frequency response analysis demonstrates a 6.6‐dB signal‐to‐noise ratio enhancement from 59.3 to 65.9 dB with stable operation beyond 300 kHz. This synergistic electronic‐plus‐plasmonic approach provides a scalable route to the development of high‐performance optoelectronic devices.

Upcycled carbon‐encapsulated iron nanoparticles (Fe@C_x NPs) from CNT synthesis byproducts demonstrate dual‐functional interface engineering through π–π interaction‐driven PEDOT:PSS reorganization and gap‐plasmonic field enhancement. This sustainable approach achieves 0.21 eV work function modulation and 87% dark current suppression in perovskite photodetectors (D = 1.49×10^{1 3} Jones), establishing generalizable design principles for organic‐inorganic hybrid optoelectronics.

1. Introduction

Hybrid organic‐inorganic optoelectronic devices represent a frontier in next‐generation electronics, yet their performance is fundamentally limited by interfacial energetic barriers and charge transport inefficiencies at heterointerfaces. Especially, metal–halide perovskite applications have advanced rapidly through interface‐engineering tactics that tune charge‐transport layers and leverage plasmonic field enhancement. Poly(3,4‐ethylenedioxythiophene): polystyrene sulfonate (PEDOT:PSS) remains the most widely used hole‐transport layer (HTL) in perovskite devices [1, 2]. The widely used PEDOT:PSS hole transport layer, despite its processability advantages, exhibits inherent work function misalignment with typical perovskite valence bands, modest conductivity and transparency, and interfacial losses, resulting in charge extraction bottlenecks that constrain device performance across multiple applications, including photodetectors, and solar cells [1, 2, 3, 4, 5, 6, 7, 8, 9].

Several alternative strategies have been explored, each with trade‐offs. Self‐assembled monolayers (SAMs) based on phosphonic acids improve energy alignment and interface quality but exhibit incomplete coverage and thermal fragility, limiting long‐term reliability [10, 11]. Plasmonic nanoparticles (e.g., Au, Ag) locally enhance optical fields, yet direct metal–perovskite contact promotes ion migration and interfacial degradation under bias and illumination [12, 13]. Metal‐oxide HTLs such as NiO_x and TiO₂ enhance selectivity and suppress dark current but may introduce parasitic absorption and additional interfacial resistance that reduce overall efficiency [14, 15]. These limitations underscore the need for a unified approach that simultaneously optimizes electronic alignment, charge transport, and optical field confinement while preserving stability.

Metal nanoparticles are distinctive additives because they can couple localized surface plasmon resonance (LSPR) with host–guest electronic interactions. Core–shell architectures such as Au@TiO₂ and Ag@Pt have shown that protective shells not only prevent core oxidation but also tailor interfacial chemistry with the surrounding matrix [16, 17, 18]. Recently, carbon‐encapsulated metal nanoparticles have emerged as promising alternatives: their atomically thin graphitic shells are electrically conductive, chemically robust, and compatible with π‐conjugated polymers, offering enhanced plasmonic effects and environmental stability. In particular, carbon‐encapsulated iron nanoparticles derived from carbon‐nanotube (CNT) synthesis byproducts have improved perovskite solar‐cell performance via gap‐plasmon effects without triggering deleterious ion migration [19].

Here we introduce carbon‐encapsulated iron nanoparticles (Fe@C_x NPs) as a multifunctional HTL modifier that addresses the above challenges through two synergistic mechanisms. First, interfacial electronic coupling, mediated by π–π interactions between the graphitic shells (1–2 nm) and PEDOT thiophene rings, raises the PEDOT:PSS work function by 0.21 eV (from 5.03 to 5.24 eV), achieving improved HOMO alignment with the valence band of FA_0.9Cs_0.1PbI₃ perovskite. Fourier‐transform infrared (FTIR) spectroscopy reveals characteristic changes in the 1300–1600 cm⁻¹ region, corroborating molecular‐level π–π interactions at the interface. Second, time‐controlled shell–shell interactions induce nanoscale clustering of Fe@C_x NPs that generates gap‐plasmonic enhancement, compensating the limited transparency of pristine PEDOT:PSS and strengthening the optical field within the active layer. By optimizing dispersion time, we maximize the gap‐plasmon effect and enable voltage‐dependent photoamplification, attaining responsivities up to 10³ A/W under low‐light conditions while maintaining an 87% dark current reduction from 2.12 × 10⁻⁸ to 2.71 × 10⁻⁹ A/cm². The carbon shells suppress metal‐ion migration, supporting device stability. Collectively, this electronic‐plus‐plasmonic interface‐engineering strategy yields perovskite photodetectors with detectivity up to 1.49 × 10^{1 3} Jones and 6.6 dB improvement in SNR, establishing a practical pathway for simultaneously optimizing energetics, transport, and optical confinement in high‐performance, stable perovskite photodetectors.

2. Results and Discussion

2.1. Structural and Electronic Characterizations of the Fe@Cx NPs in PEDOT:PSS

2.1.1. Morphological Verification and Synthesis Methodology

Figure 1a displays comprehensive TEM images combined with Fe elemental mapping, confirming the uniform distribution of Fe cores throughout the carbon matrix [13]. Elemental mapping analysis revealed systematically distributed, discrete Fe‐rich regions within the nanoparticle structure, confirming the successful synthesis of Fe‐containing NPs with controlled spatial distribution. The controlled synthesis methodology enabling this uniform distribution is presented in Figure S1, illustrating the apparatus, catalytic growth mechanism, and sample preparation procedures, including dispersion, sonication, filtering, and purification steps, used for synthesizing Fe@C_x/CNT. This systematic approach, employing catalytic chemical vapor deposition using ferrocene precursors, ensures controlled nanostructure synthesis and non‐aggregated NP extraction [11, 12]. The multiscale morphological verification conducted is shown in Figure S2, including the scanning electron microscopy (SEM) images of the Fe@C_x/CNT hybrid structures under different magnifications, revealing interconnected CNTs decorated with Fe@C_x NPs appearing as bright clusters. Figure S3 displays the biological‐TEM (bio‐TEM) characterization results, displaying hybrid Fe@C_x/CNT structures and isolated Fe@C_x NPs and revealing the presence of structural uniformity that yields the controlled distribution observed in Figure 1a.

FIGURE 1.

Structural and electronic characterizations of Fe@C_x NPs in PEDOT:PSS. (a) Transmission electron microscopy (TEM) image of Fe@C_x NPs with corresponding Fe elemental mapping (inset, green). (b) High‐resolution TEM (HRTEM) image. (c) Schematic illustration of the core‐shell structure. (d) Secondary electron cutoff spectra obtained from UPS for PEDOT:PSS without (W/O, blue) and with (W/, red) Fe@C_x NPs. (e) Valence band spectra from UPS measurements showing HOMO onset differences between PEDOT:PSS W/O (blue) and W/ (red) Fe@C_x NPs. (f) Energy level diagram of pristine PEDOT:PSS. (g) CV curves comparing electrochemical behaviors of PEDOT:PSS W/O (blue) and W/ (red) Fe@C_x NPs. (h) Magnified reduction region of the CV curves (potential range: −1.75 to −0.50 V vs. Ag/Ag⁺). (i) Energy level diagram of Fe@C_x NPs‐incorporated PEDOT:PSS. (j) Hybrid HTL/perovskite Fe@C_x plasmonic architecture. (k) Device cross‐section. (l) Energy‐level alignment of the full ITO/HTL/perovskite/ETL/Ag stack comparing pristine and Fe@C_x‐incorporated PEDOT:PSS.

2.1.2. High‐Resolution Structural Analysis and Crystallinity Verification

HRTEM imaging (Figure 1b) revealed the presence of crystalline cores with a characteristic lattice spacing of about 0.215 nm. This value is noticeably larger than the typical interplanar spacing of α‐Fe (110) (∼0.203 nm). Such lattice expansion is consistent with the presence of iron carbide phases, suggesting that the cores consist of metallic Fe with incorporated Fe₃C nanodomains, likely formed at the interface with the carbon shell [20]. The surrounding graphitic carbon shell serves as a passivation layer, providing both chemical stability and electronic compatibility with the matrix. This core‐shell architecture maintains the metallic properties of Fe necessary for plasmonic enhancement while ensuring optimal interfacial electronic coupling (π–π mediated) with PEDOT:PSS through the carbon shell interface [14]. These Fe cores are systematically encapsulated in carbon shells with 1−2‐nm thicknesses, as displayed in the schematic diagram shown in Figure 1c. The atomic‐scale structural integrity displayed in Figure 1b is corroborated by the HRTEM characterization results obtained from the Fe@C_x NPs on CNTs (Figure S4), revealing the presence of carbon shells and Fe‐based cores with atomic‐scale precision. The core‐shell architecture consists of iron‐dominant cores (∼10 nm diameter) encapsulated by 1−2 nm graphitic carbon shells. These shells serve dual roles: (1) Chemical protection, preventing reactive iron cores from direct contact with perovskite halide ions; (2) Molecular templating, providing graphitic surfaces for π‐π interactions with PEDOT aromatic rings. The interparticle spacing of 1−2 nm falls within the theoretical regime for gap‐plasmonic coupling, as this spacing arises from the balance between electrostatic repulsion (PSS polyanions) and van der Waals attraction [15, 16, 17, 18, 19].

Figure S5 provides additional evidence via TEM characterization of the dispersed Fe@C_x NPs and CNTs, confirming the morphologies and dispersion states of the separated materials used for subsequent PEDOT:PSS integration, whereas the distinct 1−2‐nm‐thick carbon shells ensure optimal balance between electronic coupling and chemical isolation. The size‐selective separation (0.20 µm filtration) recovers the Fe@C_x fraction from the CNT byproduct at an isolated yield of ≈ 20 wt% (Table S7), with CNT‐sized residues (>200 nm) removed below the DLS detection limit for this identical process [19]—supporting the scalability of the upcycling route. This core‐shell configuration possessed crucial design advantages, preventing direct metal‐perovskite contact while maintaining electronic connectivity via the conductive carbon shell, thereby addressing the ion‐migration issues typically associated with direct metal‐semiconductor interfaces present in perovskite devices [21]. Comprehensive characterization revealed that incorporating Fe@C_x NPs induced multiple synergistic mechanisms that collectively enhanced photodetector performance, enabling the surrounding carbon shell to provide efficient hole transport pathways via π‐π interactions with PEDOT thiophene rings [22, 23].

2.1.3. Electronic Structure Modification Verified via UPS Analysis

UPS analysis was used to quantitatively characterize the interfacial electronic changes induced by the incorporation of the Fe@C_x NPs. Figure 1d displays the secondary electron cutoff spectra, revealing systematic energy shifts owing to Fe@C_x integration. Pristine PEDOT:PSS exhibited a cutoff energy of 16.19 eV, which shifted to 16.09 eV after Fe@C_x integration, representing a significant 0.1 eV change in secondary electron cutoff [24]. This shift directly correlates with modifications in the work function and establishes the foundation for improved energy level alignment. Figure 1e presents valence band spectra that further clarify these electronic changes, with HOMO onset shifting from +0.16 to +0.37 eV relative to the Fermi level in the cases of the pristine and Fe@C_x‐incorporated samples, respectively. This 0.21 eV shift suggests substantial electronic structure reorganization upon Fe@C_x integration, consistent with π−π interaction‐mediated charge transfer complex formation between graphitic carbon shells and PEDOT thiophene rings [19]. The magnitude and direction of this shift align with theoretical predictions for interfacial dipole formation at organic‐ graphitic interfaces [25, 26], indicating that electronic coupling—rather than simple doping—governs the observed work function modulation.

2.1.4. Energy Level Alignment and Quantitative Analysis

Figure 1f,i displays comprehensive energy level diagrams to visualize the changes in the electronic structures of pristine and Fe@C_x‐incorporated PEDOT:PSS, respectively. These diagrams demonstrate that the modifications induce a favorable energy environment for enhanced charge transport, whereas the systematic shifts observed in Figure 1d,e indicate improved energy alignment with the perovskite valence band. Table 1 lists the UPS analysis results, establishing the fundamental electronic changes that enhance photodetector performance. The incorporation of Fe@C_x increases the work function from 5.03 to 5.24 eV, representing a significant 0.21 eV enhancement that directly correlates with the cutoff shift observed in Figure 1f [27]. This systematic electronic structure modification induces favorable energy level alignment with the perovskite valence band, establishing the foundation for enhanced charge extraction efficiency through optimized interfacial energy barriers. The interfacial dipole enhancement from 0.16 to 0.37 eV, corresponding precisely to the HOMO onset shift shown in Figure 1e, induces a favorable electrostatic environment that facilitates efficient hole extraction through improved energy level alignment while maintaining proper energy barriers to suppress unwanted charge injection from the indium tin oxide (ITO) electrode. Upon Fe@C_x incorporation, the PEDOT:PSS HOMO deepens from 5.03 to 5.24 eV (Table 1), bringing the HTL frontier level into closer alignment with the MAPbI₃ valence‐band edge and thereby promoting efficient unidirectional hole extraction across the full ITO/HTL/perovskite/ETL/Ag stack (Figure 1j–l).

TABLE 1.

Data obtained from UPS spectra: Cutoff, interfacial dipole, hole injection barrier, and HOMO.

Condition	Cutoff	Δ Dipole	Φh (HOMO_onset)	HOMO
Indium tin oxide (eV)	16.35	—	—	4.87 (Wf)
W/O Fe@Cx (eV)	16.19	0.16	0.16	5.03
W/ Fe@Cx (eV)	16.09	0.26	0.37	5.24

2.1.5. Electrochemical Properties and Determining the Lowest Unoccupied Molecular Orbital (LUMO)

Complementary electrochemical characterization via cyclic voltammetry (CV) analysis further clarifies the electronic properties of Fe@C_x‐modified PEDOT:PSS. Figure 1g displays comparative electrochemical behavior, showing that Fe@C_x incorporation shifts the reduction potential from –1.44 to –1.49 V (vs. Ag/Ag⁺, Table S1), indicating enhanced electron‐accepting capability. This 50 mV shift demonstrates improved charge transfer kinetics, supporting the UPS‐observed electronic structure modifications and confirming the synergistic effect of Fe@C_x incorporation. Figure 1h presents the magnified reduction region, showing additional electrochemical activity between −1.75 and −0.5 V, thereby indicating the presence of electrochemically active Fe centers within the carbon‐shell structure and directly supporting the presence of the core‐shell architecture confirmed in Figure 1b,c and Table S1 [28].

2.1.6. Surface Characteristics and Processing Compatibility

This structural verification is further supported by the bright‐field TEM images showing Fe@C_x morphology and the corresponding elemental maps displaying Fe (green), C (red), and O (blue) distributions in Figure S6, directly confirming the presence of a core‐shell structure based on chemical composition. Elemental mapping confirmed that the cores are Fe‐rich with negligible carbon content. The surface characteristics are verified via atomic force microscopy (AFM) conducted on the Fe@C_x NPs dispersed on a substrate (Figure S7), with 2D topographic imaging over a 5 µm × 5 µm scan area displaying NPs as bright spots with heights primarily in the 0–2 nm range [29]. The corresponding 3D visualization emphasizes individual particle height profiles and spatial arrangements, confirming that controlled deposition enabled uniform modifications in the electronic structure, demonstrated in Figure 1d–i. Table S2 lists the AFM surface roughness parameters, including an arithmetic mean roughness (R_a) of 0.344 nm and root mean square roughness (R_q) of 0.717 nm, and surface texture parameters showing negative skewness (R_sk: −6.929), indicating valley‐dominated surface profiles and high kurtosis (R_ku: 74.004) that suggests the presence of sharp peaks consistent with discrete NP features.

2.1.7. Chemical Integration and Composition Verification

The successful incorporation of the Fe@C_x NPs into the PEDOT:PSS matrix, which underlies all electronic modifications, is confirmed by the energy dispersive spectroscopy (EDS) results and Fe elemental mapping shown in Figure S8. This figure displays characteristic C, O, and Fe peaks and spatial Fe distribution over the sample area. Table S3 lists the elemental compositions obtained via EDS after N₂, revealing the presence of C (56.94 wt%, 64.21 at%), O (41.94 wt%, 35.52 at%), and Fe (1.13 wt%, 0.27 at%). The presence of Fe (1.13 wt%, 0.27 at%) within the PEDOT:PSS matrix provides direct chemical evidence for successful NP incorporation, establishing the chemical foundation for the electronic property changes listed in Table 1 and the work function modifications displayed in Figure 1d.

2.2. Fe@Cx NP‐PEDOT:PSS Interaction Mechanisms and Performance Enhancement

2.2.1. Molecular‐Level Interaction Pathways and Processing Optimization

Figure 2a,b illustrates a comprehensive schematic of the three distinct, synergistic molecular‐level chemical interactions that collectively enhance the PEDOT:PSS HTL performance; π‐π stacking interactions between the aromatic carbon shells of the Fe@C_x NPs and the thiophene rings of PEDOT represent the primary mechanism for enhancing charge transport [30]. Recent studies have demonstrated that π–π interactions in conducting polymers are fundamental to enhancing electrical conductivity, with optimal π‐π stacking distances of approximately 3.38 Å yielding conductivities of > 0.2 S cm^{− 1} in PEDOT:PSS systems [31]. The 3.38 Å optimum for π–π stacking refers to polymer–polymer, that is PEDOT–PEDOT interactions in general, whereas in our hybrid the interfacial π–π interaction between PEDOT and the carbon shell further enhances charge transport. The carbon shell provides an extended π‐conjugated system that yields continuous conductive pathways, directly correlating with the work function modifications observed in Figure 1d–i [32]. Electrostatic interactions between negatively charged PSS chains and the Fe@C_x NP surfaces provided crucial stabilization for uniform dispersion [33]. These interactions work synergistically with the energy level modifications displayed in Figure 1d–i, in which the 0.21 eV enhancement in work function induces optimal conditions for charge extraction and injection [33].

FIGURE 2.

Schematic illustration: (a) (1) π–π interactions between graphitic carbon shells and PEDOT thiophene rings, (2) electrostatic interactions between PSS chains and Fe@C_x NP surfaces, and (b) van der Waals force‐mediated nanoparticle clustering and shell–shell sp² interactions for gap‐plasmonic enhancement. (c) Dark current density−voltage characteristics, (d) results of trap‐assisted tunneling analysis (ln J vs. 1/E), (e) injection behavior (ln J vs. V), (f) EQE and normalized integrated EQE spectra, (g) spectral and normalized responsivity, (h) detectivity, and normalized D_sh ^* of the devices with various Fe@C_x concentrations.

The critical importance of processing conditions in achieving optimal molecular‐level interactions was demonstrated using the optical absorption spectra of the PEDOT:PSS films with Fe@Cx NPs at different dispersion times (Figure S9). The 1‐h dispersion condition exhibited enhanced absorption in the 400–800 nm region, which—rather than indicating detrimental optical loss—serves as spectroscopic evidence for charge‐transfer complex formation between the graphitic carbon shells and PEDOT chains [34, 35]. Given the thin nature of the HTL (∼30–40 nm) and established literature demonstrating >87% transmittance for comparable PEDOT:PSS films [36], the observed absorption modifications correlate directly with the 23% conductivity enhancement and improved device performance, rather than compromising optical transparency loss. This processing optimization directly supports the π–π interaction mechanisms illustrated in Figure 2a.

The 1–2 nm thick graphitic carbon shells provide extended π‐conjugated systems that interact with PEDOT thiophene rings through π–π stacking interactions. Figures S10 and S11 provide direct spectroscopic evidence for the molecular‐level interactions proposed in Figure 2a. Figure S10 demonstrates systematic changes in Fourier transform infrared (FTIR) transmittance spectra; the transmittance of the Fe@C_x‐incorporated samples was lower than that of pristine PEDOT:PSS, indicating successful NP integration [37, 38].

The normalized FTIR spectra displayed in Figure S11 reveal that the characteristic modifications in the C–C/C = C and C–S–C vibrational regions (1300–1600 cm^{− 1}) providing spectroscopic evidence of π–π interactions and structural modifications of the PEDOT:PSS matrix [11, 39, 40, 41]. It suggests structural changes in PEDOT chains upon Fe@C_x incorporation, which may indicate enhanced conjugation or chain extension [19, 42, 43]. While these spectroscopic signatures are consistent with π‐π stacking‐induced chain reorganization reported for graphitic additives [19], further structural characterization would strengthen this interpretation. The π–π electronic coupling is confirmed by high‐resolution XPS (Figure S22, Table S6), showing sulfonate‐S FWHM broadening of + 0.37 eV (+ 19%) and thiophene‐S/sulfonate‐S area ratio increase by × 4.38 upon Fe@C_x incorporation. These spectroscopic changes validate that the performance improvements observed in subsequent device measurements are attributed to genuine molecular‐level interactions rather than simple physical mixing effects [44]. To validate these proposed interaction mechanisms, electrical characterization was performed on simplified test structures. The I‐V characteristics of ITO/PEDOT:PSS(Fe@C_x)/Ag devices demonstrated enhanced charge transport, with conductance increasing from 0.161 to 0.198 S upon Fe@C_x incorporation (Figure S12). This electrical evidence directly supports the π–π interaction mechanism depicted in Figure 2a, confirming that the molecular‐level structural modifications translate into measurable improvements in charge carrier mobility.

2.2.2. Van Der Waals Force‐Mediated Aggregation and Network Formation

Figure 2b illustrates the aggregation mechanisms governed by van der Waals forces between individual Fe@C_x NPs, yielding controlled network formation with enhanced conductivity while maintaining device integrity [45]. The shell‐shell sp² interactions between adjacent NPs facilitate the formation of percolation pathways, building upon the core‐shell architecture confirmed in Figure 1b [46]. Theoretical studies on NP aggregation have demonstrated that van der Waals interactions are critical for determining assembly morphology, with forces ranging from weak (0.06 kJ/mol) to strong (32 kJ/mol) depending on particle polarizability [47]. In the case of Fe@C_x NPs with 10‐nm core diameter and carbon shell thicknesses between 1–2 nm, established in Figure 1c, van der Waals forces are sufficient to promote controlled aggregation while maintaining discrete characteristics [48]. Especially, this clustering mechanism enables gap‐plasmonic enhancement through localized electromagnetic field amplification. The optimal balance between dispersion and controlled aggregation directly influences the concentration‐dependent behavior observed in subsequent device measurements, in which excessive aggregation induces detrimental leakage pathways, whereas insufficient interaction limits performance enhancement.

2.2.3. Perovskite Photodetector Characterizations

Figure 2c presents systematic dark current density‐voltage (J–V) characteristics, demonstrating clear concentration‐dependent optimization based on the device structure (Figure 1j). The pristine device exhibited a dark current density of 2.12 × 10^{− 8} A/cm² at 0 V bias, progressively improving with increasing NP concentration. When 1.0 v% Fe@C_x was incorporated, the dark current reduced to 1.57 × 10^{− 8} A/cm² (26% improvement), whereas the optimal 2.5 v% loading achieved a minimum dark current of 2.71 × 10^{− 9} A/cm² (87% reduction). However, 5.0 v% concentration degrades the dark current to 3.52 × 10^{− 8} A/cm², exceeding the pristine device performance owing to strong aggregation effects, as illustrated in Figure 2b. This systematic concentration optimization demonstrates that 2.5 v% represents the optimal balance between plasmonic enhancement and film conductivity preservation, with higher concentrations leading to detrimental aggregation effects and lower concentrations providing insufficient enhancement. Cross‐plane conductance confirms this degradation is morphological rather than bulk‐transport in origin: G/G_Ref peaks at + 10.3% (2.5 v%) and remains + 8.0% at 5.0 v% (revised Figure S12). Consistently, AFM higher‐order statistics localize the 5.0 v% loss to surface inhomogeneity—R_q + 22% with a leptokurtic R_ku = 1.10 versus the Gaussian‐like R_ku = 0.46 at the 2.5 v% optimum (Figures S26, S27)—quantitatively replacing the previous qualitative aggregation attribution.

The fundamental dark current characteristics are further analyzed using the ohmic shunt J‐V behavior under different Fe@C_x concentrations (Figure S13). Shunt resistance analysis reveals that optimal NP incorporation improves interface quality and reduces leakage pathways, consistent with the dark current improvements shown in Figure 2c. The correlation between reduced shunt currents and optimal NP loading confirms the beneficial effects of controlled aggregation, highlighting the detrimental impact of high NP concentration. Batch reproducibility across 11 devices per condition is confirmed by the statistical distributions of the key figures‐of‐merit (dark current and shot‐noise‐limited detectivity at 0 V) in Figure S14, where Fe@C_x incorporation eliminates the high‐leakage device tail and raises the detectivity population.

Figure 2d presents the ln J vs. 1/E plots obtained from trap‐assisted tunneling analysis, providing quantitative insights into defect state distributions and the modulations owing to Fe@C_x incorporation [49]. Table 2 lists the calculated energy levels of the defect state, which progressively reduce with optimal NP loading. The pristine device exhibits a defect state energy of 2.31 meV, decreasing to 2.06 meV (1.0 v%), reaching a minimum of 1.08 meV (2.5 v%), and finally, increasing to 2.55 meV (5.0 v%). Recent ultrafast photocurrent spectroscopy studies have demonstrated that trapped carriers in perovskite films follow phonon‐assisted tunneling mechanisms in ultra‐shallow trap states (1.72–10.15 meV) [49]. The 53% reduction in defect state energy at optimal loading aligns with these findings, indicating the effective passivation of shallow trap energy states that dominate recombination processes. The correlation between the energy level modifications observed in Figure 1d–i (work function enhancement and HOMO optimization), and the defect state reductions listed in Table 2 demonstrate the synergistic effects of electronic structure optimization mechanism, which directly relates to the molecular‐level interactions established in Figure 2a, that is, π–π interactions improve PEDOT structural ordering. At high concentrations, aggregation‐induced structural disorder introduced new defect states, explaining the performance degradation observed using multiple characterization techniques.

TABLE 2.

Characterization of the perovskite photodetector: Energy level of the defect state using trap‐assisted tunneling current and ideality factor.

#Fe@Cx amount (v%)	Defect state [meV]	n (Ideality factor)
W/O	2.31	2.42
W/ 1.0 v%	2.06	2.11
W/ 2.5 v%	1.08	1.68
W/ 5.0 v%	2.55	2.45

Figure 2e displays the results of injection behavior analysis using ln J vs. V measurements, enabling the ideality factors, which quantify deviation from ideal diode characteristics, to be determined [50]. Table 2 lists the systematic ideality factors, which progressively improve with optimal NP incorporation. The pristine device exhibits an ideality factor of 2.42, indicating a significant departure from ideal behavior (n = 1) owing to the presence of trap states and recombination centers. The ideality factors progressively improve with the increase in Fe@C_x concentration, that is, 2.11 (1.0 v%), 1.68 (2.5 v%), and degradation to 2.45 (5.0 v%). The 31% improvement at optimal loading represents enhanced charge injection efficiency and reduced trap‐assisted recombination. Ideality factors approaching unity indicate optimized charge transport and minimized interface defects [50]. The correlation between the reduced defect state energies (Table 2) and improved ideality factors confirms the effectiveness of the charge flow. This improvement directly reflects the optimized energy level alignment demonstrated in Figure 1d–i, which reveals that enhanced work function matching and reduced hole injection barriers facilitate ideal charge injection characteristics. This systematic improvement validates the synergistic effects of electronic structure modification and interface optimization owing to Fe@C_x incorporation.

Figure 2f displays the EQE and normalized integrated EQE measurements obtained over the visible spectrum (420–780 nm), revealing systematic enhancements owing to optimal NP incorporation [51]. The optimal loading of 2.5 v% yielded superior performance across all wavelengths, with the normalized integrated EQE reaching 100% for maximum photon‐to‐current conversion efficiency. The spectral response maintained characteristic perovskite absorption features while exhibiting overall enhancement. The wavelength‐dependent performance was further analyzed using the responsivity and detectivity characteristics even at a −0.1 V bias across different wavelengths (Figure S15). The performance of 2.5 v% Fe@C_x‐incorporated device was better than the pristine (W/O) ones across the visible spectrum, with particularly significant improvements in the 600–700 nm region in which perovskite absorption was the strongest. This wavelength‐dependent analysis supports the EQE improvements displayed in Figure 2f while validating the photoelectronic enhancement mechanisms. Figure S16 displays the results of this analysis extended across different applied bias voltages, revealing that Fe@C_x incorporation offers benefits under both reverse and forward bias conditions. Voltage‐dependent responsivity and detectivity measurements demonstrated that the enhancement mechanisms remained effective across various operating conditions, indicating robust improvement that extended beyond single‐bias characterization. The correlation between bias‐independent improvements and the energy level optimizations displayed in Figure 1d–i confirms that fundamental electronic modifications are responsible for performance enhancements.

Figure 2g presents the results of spectral responsivity measurements that quantify current response per unit optical power, demonstrating a peak responsivity of 0.43 A/W (0.38 A/W of the pristine device) at 680 nm for the optimal 2.5 v% loading (Table 3) [52]. This 13% improvement directly correlates with the enhanced charge injection characteristics displayed in Figure 2c,e, respectively. Comprehensive performance parameters at self‐powered (0 V) and reverse‐bias conditions are listed in Tables 3 and S4. The responsivity enhancement reflects the combined effects of improved charge extraction (demonstrated in the energy level analysis shown in Figure 1d–i), reduced recombination (quantified via trap state analysis, as shown in Figure 2d), and optimized charge transport (validated through ideality factor improvements, as shown in Figure 2e). Figure 2h presents the detectivity (D_sh ^*) measurements obtained over the visible spectrum, comprehensively evaluating photodetector performance by combining signal enhancement and noise reduction effects [52]. The optimal 2.5 v% loading achieved a peak shot noise limited‐detectivity of 1.49 × 10^{1 3} Jones at 680 nm, representing a remarkable 3.2‐fold improvement with respect to the pristine device [53]. This enhancement reflects the synergistic effects of improved responsivity (Figure 2g) and the considerably reduced dark current (Figure 2c).

TABLE 3.

Performance parameters of the perovskite photodetector under the self‐powered condition.

#Fe@Cx amount (v%)	Jd at 0 V (A/cm2)	R at 0 V and 680 nm (A/W)	Dsh * at 0 V and 680 nm (Jones)
W/O	2.12 × 10−8	0.38	4.65 × 1012
W/ 1.0 v%	1.57 × 10−8	0.35	4.91 × 1012
W/ 2.5 v%	2.71 × 10−9	0.43	1.49 × 1013
W/ 5.0 v%	3.52 × 10−8	0.42	3.95 × 1012

2.3. Enhanced Temporal Response Dynamics and Charge Transport Kinetics

2.3.1. Light On/Off‐Based Photo Response Characteristics

The temporal response measurements presented in Figure 3a reveal significant improvements in the light‐on rise‐time characteristics of the Fe@C_x‐incorporated devices in relation to those of the pristine PEDOT:PSS photodetector [54]. Quantitative analysis demonstrates that the incorporation of Fe@C_x NPs reduced the light‐on response time from 1.021 to 0.999 µs, representing a 2.2% improvement in photocarrier generation and collection speed [55]. This exceptional speed enhancement aligns with the optimization of the energy levels shown in Figure 1, which shows that the improved HOMO level alignment from 5.03 to 5.24 eV generates favorable conditions for rapid hole extraction [56]. The complementary fall‐time characteristics shown in Figure 3b further provide substantial evidence of enhanced charge transport dynamics achieved through Fe@C_x incorporation. The light‐off response time remarkably improved from 1.961 to 1.391 µs, representing a substantial 29.1% enhancement in decay response characteristics [57]. The drastic reduction in fall‐time indicates considerably reduced carrier recombination lifetime and improved charge collection efficiency, directly supporting the trap state reduction displayed in Figure 2d, which shows that the defect state energy levels decreased from 2.31 to 1.08 meV at optimal loading [58, 59, 60].

FIGURE 3.

Temporal and frequency response characteristics of the Fe@C_x‐enhanced PEDOT:PSS photodetectors. (a) Light‐on rise‐time measurements comparing pristine (W/O) and Fe@C_x‐incorporated (W/) devices. (b) Light‐off fall‐time characteristics showing response decay. (c) Normalized current response at 1 kHz with indicated noise levels. (d) Frequency response power spectra comparing the W/O and W/ devices with SNRs of 59.3 and 65.9 dB, respectively. (e,f) Multi‐frequency temporal response analysis of pristine PEDOT:PSS across various frequencies from 10 Hz to 2 MHz. (g) Comparison of the normalized frequency responses of the W/O and W/ devices. (h,i) Multi‐frequency temporal response analysis of the Fe@C_x‐enhanced devices, showing improved response characteristics across the frequency range.

2.3.2. Frequency Response Enhancement and Signal‐to‐Noise Optimization

Figure 3c displays the normalized current response characteristics at 1 kHz frequency, revealing the high signal stability and reduced noise levels of the Fe@C_x‐enhanced devices. The normalized response maintains excellent amplitude stability with considerably small baseline fluctuations, indicating improved charge transport consistency across the frequency domain. This enhancement directly relates to the conductivity improvements achieved owing to the π‐π interactions between the Fe@C_x carbon shells and the PEDOT thiophene rings, according to the molecular interaction mechanisms shown in Figure 2a [61, 62]. The comprehensive frequency‐response power spectra analysis presented in Figure 3d quantitatively validates these improvements via SNR measurements. Fe@C_x incorporation achieved remarkable SNR enhancement from 59.3 to 65.9 dB at 1 kHz, representing a 6.6 dB improvement that translates to more than 4.57‐fold signal quality enhancement [63]. This substantial improvement stems from the combined effects of a reduced dark current (87% reduction at optimal loading, as shown in Figure 2c) and enhanced photocurrent generation owing to improved charge extraction efficiency. The optimization of energy level alignment, shown in Figure 1j–l, provides the fundamental electronic basis underlying these frequency response improvements. The systematic modification of work function from 5.03 to 5.24 eV and the corresponding LUMO level adjustment generated optimal injection barriers that minimized parasitic current fluctuations while maximizing signal current stability [64].

2.3.3. Multi‐Frequency Analysis and Bandwidth Characterization

The analysis of the multi‐frequency temporal response of the pristine PEDOT:PSS devices (Figure 3e,f) establishes the baseline performance characteristics over the 10 to 2 MHz frequency range. These measurements reveal the inherent limitations of unmodified PEDOT:PSS in maintaining stable response amplitude and phase relationships at high frequencies, with noticeable degradation above 100 kHz. The temporal waveforms showed increasing distortion and amplitude fluctuation owing to variations in frequency, indicating fundamental limitations in charge transport kinetics and capacitive effects within the pristine device architecture. Figure 3g compares the normalized frequency responses, demonstrating systematic improvements across the frequency spectrum owing to Fe@C_x incorporation [39]. The enhanced devices maintained high amplitude stability and low phase distortion up to 700 kHz, representing a significant bandwidth extension with respect to the pristine devices [65, 66]. This improvement correlates directly with the ideality factor enhancement from 2.42 to 1.68, demonstrated in Figure 2e, indicating ideal charge injection and transport characteristics that translate to excellent high‐frequency performance [67]. The detailed transient photocurrent measurements presented in Figure S17, obtained over a range of frequencies (100–700 kHz), comprehensively validate the bandwidth enhancement mechanisms. The measurements reveal that the Fe@C_x‐enhanced devices maintained consistent rise and fall times over the frequency range, whereas the temporal response characteristics of the pristine devices progressively degraded. This consistency in temporal performance directly relates to the reduced energy levels of the defect states and improved interfacial charge transfer, as demonstrated in the electronic characterization results displayed in Figures 1 and 2. The multi‐frequency temporal responses of the Fe@C_x‐enhanced devices (Figure 3h,i) confirmed sustained performance improvements across the complete frequency spectrum. The enhanced devices demonstrate remarkable temporal stability even under 1 MHz operation, with well‐defined square‐wave response characteristics that indicate minimal capacitive distortion and rapid charge carrier dynamics [68, 69, 70].

2.4. Power Spectral Density (PSD) Analysis and Noise‐Corrected SNR Enhancement

A two‐stage methodology combining direct experimental measurements (Figure 4a–g) and subsequent ideal simulation analysis to enhance noise‐corrected SNR (Figure 4h–i) was employed for obtaining the power spectral densities [71]. This approach follows established protocols for advanced photodetector characterization, in which raw frequency‐domain measurements are processed using ideal simulation algorithms to extract pure signal characteristics and quantify the intrinsic improvements in device performance [71].

FIGURE 4.

Analysis of the frequency‐dependent power spectral densities of the perovskite photodetectors with Fe@C_x NPs. (a−g) PSD measurements at 10 Hz, 100 Hz, 1 kHz, 10 kHz, 100 kHz, 300 kHz, and 700 kHz for pristine (W/O, gray) and Fe@C_x‐incorporated (W/, red) devices, showing signal + noise and noise floor levels. (h) Comparison of the SNRs over a frequency range. (i) SNR improvement factors of the Fe@C_x‐enhanced devices.

2.4.1. Experimental Methodology and Measurement Protocol

The experimental framework employed frequency‐domain characterization over the 10–700 kHz frequency range and measured the signal + noise and noise floor components at each frequency [72]. This direct measurement approach captured the actual photodetector response under operational conditions, generating raw spectral data that reflects the signal generation capability and inherent noise characteristics of each device configuration. The measurement protocol follows established methodologies for analyzing the frequency responses of photodetectors, ensuring consistent and reliable data acquisition across the bandwidth. Representative measurement procedures at 10 Hz are detailed in Figure S18.

Following the experimental measurements, the ideal simulation methodology was applied to extract noise‐corrected SNRs by mathematically processing raw spectral data. This approach uses the relationship SNR = 20 × log₁₀[((S + N) − N)/N] to quantitatively assess the pure signal enhancement achieved owing to the incorporation of Fe@C_x NPs while eliminating the influence of measurement‐related noise sources [73].

2.4.2. Raw PSD Characteristics

Direct PSD measurements conducted on the pristine and Fe@C_x‐enhanced photodetectors revealed distinct frequency‐dependent behavioral patterns. At low frequencies (10 Hz, Figure 4a), both device types exhibited strong signal + noise responses; these responses of the Fe@C_x‐enhanced and pristine devices were clearly differentiated. The Fe@C_x‐enhanced devices consistently demonstrated higher signal + noise power densities than the pristine devices, indicating excellent photo‐response generation capability. As measurement frequency increased to the intermediate range (100 Hz–1 kHz, Figure 4b–c), the signal + noise characteristics of both device types gradually attenuated while maintaining relative performance differences. The Fe@C_x devices sustain high signal + noise levels throughout the practically relevant frequency range, demonstrating consistent enhancement across the operational bandwidths typical for photodetector applications. The measurements obtained at high frequencies (10–100 kHz, Figure 4d–e) reveal progressive signal attenuation in both device types, reflecting the fundamental bandwidth limitations inherent to these device architectures. The relative performance advantages of Fe@C_x devices persist but degrade as both configurations approach the intrinsic frequency response limits determined by RC time constants and charge transport kinetics [74].

Measurements conducted at the highest frequency range (300–700 kHz, Figure 4f,g) demonstrated signal convergence behavior in which both device types approached similar response levels. This convergence reflects the dominance of fundamental physical limitations, including RC circuit constraints and carrier transit time effects, over materials‐level improvements at frequencies approaching the bandwidth boundaries corresponding to both device configurations. The noise floor characteristics of both device types remained relatively stable throughout the frequency measurements, confirming consistent measurement conditions and validating the thermal noise limitations predicted by the Johnson‐Nyquist theory [75]. This stability provides the foundation for reliable noise‐corrected analysis and ensures that observed signal differences reflect genuine device performance variations, rather than the presence of measurement artifacts.

2.4.3. Noise‐Corrected SNR Analysis

Figure 4h displays the true signal quality enhancement owing to Fe@C_x incorporation obtained via noise‐corrected SNR analysis across the measured frequency spectrum. Ideal simulation processing revealed that the pristine PEDOT:PSS devices maintained SNRs between 30–32 dB over the 10−300 kHz frequency range, with a slight degradation observed at the highest measured frequencies approaching the bandwidth limits of the devices. The SNRs of the Fe@C_x‐incorporated devices were superior, that is, between 32–35 dB, over the measured frequency spectrum. This represents a consistent 2–3 dB improvement over that of the pristine devices, corresponding to a 25%−50% enhancement in signal quality based on the logarithmic SNR scale [76]. The sustained SNR enhancement across a broad frequency range validates the effectiveness of Fe@C_x incorporation for improving fundamental signal generation capabilities, and these fundamental electronic modifications provide robust enhancement under varying operational conditions [77].

2.4.4. Analysis of the SNR Improvement Factor

The frequency‐dependent SNR improvement factors presented in Figure 4i reveal a remarkable, counterintuitive phenomenon, that is, the relative performance enhancement increases with the increase in frequency despite the reduced absolute signal differences observed in the raw PSD measurements. The improvement factors progressed from approximately 1 dB at low frequencies (10–100 Hz) to approaching 5 dB at high frequencies (300–700 kHz), representing more than a threefold enhancement in relative signal quality. This frequency‐dependent improvement pattern is attributed to the differential bandwidth degradation characteristics between the pristine and Fe@C_x devices. Although both device types experience signal attenuation at high frequencies owing to fundamental RC limitations, the Fe@C_x devices exhibit more gradual performance decline than the pristine devices. This differential degradation amplifies the relative performance advantage, as quantified using the noise‐corrected SNR calculation methodology [78]. This enables the better maintenance of signal integrity as operational frequencies approach bandwidth limitations. The correlation with photoluminescence quenching demonstrated in Figure S19 provides additional mechanistic support, that is, reduced radiative recombination directly improves the signal‐to‐noise characteristics [79, 80].

2.5. Gap‐plasmonic Enhanced Device Architecture and Voltage‐Dependent Photoamplification Under Different Illumination Conditions

Figure 5 introduces Fe@C_x nanoparticles as a novel HTL additive strategy for perovskite photodetectors. The Fe@C_x system demonstrates voltage‐dependent dual‐mechanism photoenhancement: (1) π–π interaction‐mediated electronic coupling between carbon shells and PEDOT:PSS providing broadband responsivity enhancement, and (2) electric field‐enhanced plasmonic coupling at higher bias voltages enabling selective wavelength amplification. This approach achieves unprecedented bias‐programmable photoamplification spanning 5 orders of magnitude (10^{− 2} –10³ A/W) while maintaining operational stability under both high‐light and low‐light conditions.

FIGURE 5.

Gap‐plasmonic and π−π Interaction‐Driven Photoresponse Architecture for Perovskite Photodetectors. (a) 3D structure of the core device components, showing the perovskite active layer with plasmonic enhancement based on the van der Waals force‐mediated Fe@C_x clustering within PEDOT:PSS matrix. (b) Cross‐sectional schematic of the key device layer (ITO/PEDOT:PSS with Fe@C_x NPs/perovskite (FA₀.₉Cs₀.₁PbI₃)) interface. (c) Localized electromagnetic field enhancement in Fe@C_x NPs nanoparticle clusters in PEDOT:PSS. (d) Comparison between the voltage‐dependent responsivity values of the pristine (W/O) and Fe@C_x‐enhanced (W/) devices under AM 1.5G illumination (1 sun). (e−i) Spectral responsivity measurements over a 300–900 nm wavelength range at bias voltages of −1.0, 0.0, +0.5, +1.0, and +1.5 V under low‐light conditions (approximately 0.0065 sun intensity), demonstrating enhanced photoamplification with Fe@C_x incorporation under forward bias.

Figure 5a shows a three‐dimensional view of the core photodetector components, illustrating the complex interplay between the perovskite active layer and the Fe@C_x‐enhanced PEDOT:PSS HTL. The architectural design showcases the key functional layers in which gap‐plasmonic enhancement occurs at the nanoscale interface, where the Fe@C_x NPs induce localized electromagnetic field enhancement regions that amplify light‐matter interactions. It directly correlates with the observed photocurrent amplification. The cross‐sectional schematic presented in Figure 5b illustrates the core composition stack, that is, ITO/PEDOT:PSS (w/ Fe@C_x NPs)/perovskite (FA₀.₉Cs₀.₁PbI₃), in detail. Figure 5c illustrates the fundamental gap‐plasmonic mechanism between Fe@C_x NPs and PEDOT:PSS, showing localized field enhancement via electromagnetic coupling between adjacent NPs [81]. The gap‐plasmonic effect occurs when Fe@C_x NPs are positioned at optimal separation distances, generating “hot spots” of increased electromagnetic field intensities that significantly amplify light absorption and photocurrent generation.

2.5.1. Illumination Condition‐Dependent Photoamplification Characteristics

Figure 5d presents systematic voltage‐dependent responsivity measurements under AM 1.5 G illumination (1 sun), revealing the photoamplification behavior under high‐intensity illumination conditions. Under these conditions, both the pristine (W/O) and Fe@C_x‐enhanced (W/) devices (with the structure ITO/PEDOT:PSS (w/ Fe@C_x NPs)/perovskite (FA₀.₉Cs₀.₁PbI₃)/PC₆₀BM/BCP/Ag)) demonstrated exponential responsivity enhancement under forward bias, with the latter device exhibiting superior performance. The high photon flux under AM 1.5 G conditions enables the observation of dramatic photoamplification effects in the Fe@C_x‐enhanced device, reaching responsivity values of > 2.5×10^{− 4} A/W at +1.2 V. In contrast, Figure 5e–i presents spectral responsivity measurements under low‐light conditions (approximately 0.0065 sun intensity), in which the reduced photon flux induces fundamentally different operating conditions that reveal bias‐dependent photoamplification mechanisms with distinct characteristics. Recent studies on perovskite photodetectors operated under low‐light conditions demonstrated that the relationship between photon flux and bias‐dependent enhancement increases in complexity, with different voltage regimes showing varying effectiveness [82].

2.5.2. High Responsivity Under Reverse Bias and Low‐Light Conditions

At a reverse bias of −1.0 V (Figure 5e), low‐light conditions reveal dramatically different behavior from that under forward bias. The pristine device (W/O) showed responsivity values of approximately 10² A/W, whereas the Fe@C_x‐enhanced device (W/) demonstrated significantly increased responsivity, reaching approximately 10³ A/W over the 300–900 nm range. This substantial enhancement under reverse bias and low‐light conditions indicates that the electronic structure optimizations established in Figure 1 become particularly effective when combined with the built‐in electric field enhancement under reverse bias. This responsivity enhancement under reverse bias aligns with the results included in recent studies on high‐performance perovskite photodetectors, in which reverse‐bias operations yield responsivity values of > 1 A/W owing to enhanced charge collection efficiency and reduced recombination losses [7, 83]. The order of magnitude improvement in the performance of the Fe@C_x‐enhanced devices validates the synergistic effects of energy level optimization and gap‐plasmonic enhancement under low‐photon‐flux conditions.

2.5.3. Self‐Powered and Low‐Forward‐Bias Operations

The self‐powered operations (Figure 5f) of the pristine and Fe@C_x‐enhanced devices under low‐light conditions at 0.0 V demonstrated minimal differences, with both devices showing responsivity values of approximately 10^{− 2} A/W. This observation aligns with the recent reports on perovskite photodetectors, revealing that significant photoamplification requires the applied bias to overcome the thermal noise limitations under low‐light conditions. The 13% responsivity improvement listed in Table 3 under the self‐powered condition primarily originates from electronic structure optimizations rather than gap‐plasmonic field enhancements, which require sufficient bias‐induced electric fields to manifest effectively. Similarly, at a +0.5 V forward bias (Figure 5g), the performance of the Fe@C_x‐enhanced device was comparable to that of the pristine device under low‐light conditions, with responsivity values remaining at 10^{− 2} A/W. This behavior is consistent with that obtained from the theoretical models of photoamplification in perovskite devices, which predict that significant photoamplification begins at bias voltages exceeding +0.7 V owing to the energy barriers associated with carrier injection processes [19].

2.5.4. High‐Forward‐Bias Photoamplification Enhancement

Drastic photoamplification enhancement became apparent at +1.0 V (Figure 5h) and under low‐light conditions, under which the Fe@C_x‐enhanced device achieved responsivity values approaching 10³ A/W, representing significant improvement with respect to low‐bias conditions. The predominant effect is robust, broadband responsivity enhancement (300–900 nm), attributed to chemically driven π–π interactions between the graphitic carbon shell of Fe@C_x NPs and the thiophene rings of PEDOT. This π–π coupling promotes electronic delocalization, lowers the interfacial barrier, and uniformly improves carrier transport across the spectrum. This regime matches our previous findings [84], which confirmed electronic (non‐plasmonic) enhancement in the 300–900 nm range due to improved hole extraction and reduced trap‐mediated recombination.

As the applied bias increases (+1.5 V, Figure 5i), an additional, distinctly selective enhancement (attributed to gap‐mediated near‐field coupling) is observed—most prominent in the short‐wavelength region (300–600 nm) by up to 100×, which is a pattern that closely tracks electric‐field‐driven plasmonic resonance models. The responsivity increase exceeds predictions based solely on electronic work function tuning, consistent with an additional photoamplification contribution. The closely spaced Fe cores (1–2 nm interparticle gaps observed in TEM, Figure 2) present a geometry consistent with gap‐plasmonic coupling conditions. While rigorous electromagnetic simulations are beyond the scope of this study, the observed voltage‐dependent responsivity enhancement aligns with theoretical models of localized field enhancement in similar core‐shell nanostructures [17, 79, 80, 85, 86], suggesting a potential contribution from plasmonic effects. This voltage‐ tunable dual regime—passive energetic alignment at low bias and potential field‐enhanced photoamplification at high bias—represents a unique operating characteristic not observed in conventional interface modifiers.

This selective amplification is consistent with electric field‐induced gap‐mediated near‐field coupling, as also supported by optical simulations [11], which predict field‐confinement and resonance red‐shifting. Unlike conventional gap‐plasmonic nanostructures, which often show extremely narrow spectral enhancements, our system exhibits field‐driven plasmonic gain superimposed on a broad, π–π‐coupling background. This dual mechanism enables up to 100‐fold amplification in the short‐wavelength regime while preserving enhanced performance across the entire visible spectrum. System‐specific full‐wave electromagnetic calculations (single‐particle Mie + retarded coupled‐dipole) independently verify this gap‐mediated near‐field coupling: isolated NPs are optically inactive, whereas clusters exhibit a gap‐monotonic per‐particle enhancement σ^N/(N·σ¹) > 1 (1.022 at g = 0.5 nm), and the predicted film absorption reproduces the measured UV‐Vis enhancement (Figures S23–S25 and Table S8).

The supporting morphological evidence included in Figure S20 demonstrates that incorporating Fe@C_x enhances the grain size of the perovskite, creating large crystalline domains that reduce grain boundary defects and support highly efficient photogeneration, particularly crucial under low‐light operation. SEM analysis revealed systematic improvements in the grain sizes of the perovskite films grown on Fe@C_x‐incorporated PEDOT:PSS substrates with respect to those on pristine substrates, providing the morphological foundation for the enhanced photoamplification observed under different bias conditions [87, 88, 89]. This morphological and interfacial optimization also translates into long‐term operational stability: over 432 h of dark storage, the 2.5 v% device retains its responsivity with a post‐burn‐in instability of only 1.03% (R_max)/1.75% (∫R dλ), versus 7.39%/10.46% for the pristine reference—a > 80% improvement (Figure S28). Benchmarked against recent plasmonic‐NP and interface‐engineered perovskite photodetectors (Table 4), the present device uniquely combines sub‐nA‐cm^{− 2} dark current, bias‐tunable responsivity (0.43 → 1,527 A W^{− 1}), high detectivity (1.49 × 10^{1 3} Jones), and µs‐scale response —a pairing of bias‐tunable photo‐amplification and gap‐mediated near‐field enhancement, realized from an upcycled material, not jointly attained by the cited strategies. Comprehensive benchmarking (Figure S21, Table S5) confirms this unprecedented voltage‐tunable device with 5‐order‐magnitude control.

TABLE 4.

Key‐parameter comparison of the present device with recent plasmonic‐NP and interface‐engineered perovskite photodetectors.

Device (strategy)	Jd (A cm− 2)	R (A W− 1)	D* (Jones)	Response (rise/fall)	Refs.
This work — Fe@Cx: gap‐plasmon + bias photo‐amplification (upcycled)	2.71 × 10− 9 (0 V)	0.43 (0 V) → 1,527 (+1 V)	1.49 × 101 3	0.999/1.391 µs	—
α‐FAPbI3 + modified ETL (interface, flexible)	1.5 × 10− 7	0.93	1.55 × 101 2	< 1 ms	[87]
MAPbI2Br/PDPP3T bipolar (interface, flexible)	N.R.	0.125 (bipolar)	N.R.	N.R.	[88]
Au‐NR/PEIE (plasmonic + interface)	low	0.360	1.81 × 1010	N.R.	[89]

3. Conclusions

This work establishes π‐π electronic coupling as a versatile interface engineering mechanism for hybrid optoelectronics. By utilizing carbon‐ encapsulated iron nanoparticles—upcycled from industrial synthesis byproducts—we achieve simultaneous work function modulation (0.21 eV), charge transport enhancement (23% conductivity increase), and trap state suppression (53% reduction) in PEDOT:PSS. Demonstrated in perovskite photodetectors with 87% dark current reduction (2.12 × 10^{− 8} → 2.71 × 10^{− 9} A/cm²) and 3.2‐fold enhanced detectivity (4.65 × 10^{1 2} → 1.49 × 10^{1 3} Jones) with improved SNR by 6.6 dB (59.3 → 65.9 dB), this approach transcends device‐specific optimization. The fundamental principles—graphitic shell‐induced polymer chain reorganization and core‐shell structural protection—apply directly to nanostructure engineering for interface optimization as an effective approach to fundamental performance enhancement in perovskite optoelectronic devices. Looking ahead, integrating the present upcycled‐Fe@C_x HTL strategy with emerging device architectures—such as flexible/stretchable form factors and wavelength‐selective bipolar photodetection—represents a promising direction for next‐generation perovskite photodetectors [87, 88].

4. Experimental Section

All materials used and characterization techniques followed were based on standardized protocols. Detailed descriptions of the specialized equipment, chemical sources, and procedural steps are included in the Supporting Information.

Author Contributions

Byung Gi Kim: conceptualization, methodology, data curation, investigation, validation, formal analysis, visualization, writing – original draft, writing – review and editing. Jiye Han: conceptualization, methodology, software, investigation, writing – review and editing, visualization. Jihyun Lim: investigation, conceptualization, methodology, validation, visualization, formal analysis. Woongsik Jang: validation. Dong Hwan Wang: conceptualization, methodology, investigation, project administration, supervision, writing – review and editing, resources, funding acquisition, writing – original draft. Min Soo Kim: investigation. Yu Min Lee: investigation, formal analysis, validation. Il Jeon: conceptualization, supervision, writing – review and editing, writing – original draft, resources, formal analysis, validation, investigation, data curation, funding acquisition.

Conflicts of Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Supporting information

Supporting File: smll74645‐sup‐0001‐SuppMat.docx.

Data Availability Statement

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