Development of Sensitive, High‐Performance Organic Gas Sensors via Dimensionally Controlled Perovskites

ABSTRACT

Organic semiconductors are considered promising alternatives to rigid inorganic materials for next‐generation smart gas sensors. However, their intrinsic limitations such as low surface reactivity, insufficient charge density, and small surface area hinder their performance. To address these challenges, we engineer composite sensing layers by incorporating perovskite nanocrystals into poly(3‐hexylthiophene) (P3HT) thin films and systematically analyze how perovskite structure and defect formation modulate the electronic and chemical properties of the hybrid films. The semiconducting nature of perovskites and their favorable band alignment with P3HT enhance charge transport in organic field‐effect transistors (OFETs) without compromising charge transfer efficiency, which is difficult to achieve using conventional non‐conductive inorganic porous materials. Perovskite quantum dots exhibit uniform dispersion and improved percolation pathways, resulting in the significant enhancement of OFET performance. In contrast, perovskite nanowires show higher crystallinity and abundant Br‐vacancies, partially accompanied by Cs₄PbBr₆‐domains, which serve as highly reactive adsorption sites for electron‐withdrawing gas molecules. Consequently, NW‐30 wt.% blended devices demonstrate the highest responsivity toward NO₂ gas, achieving a sensitivity of 3.05%/ppm and a limit of detection of 0.0055 ppm. This work demonstrates the potential of dimensionally controlled perovskites with halide‐vacancy engineering as a versatile strategy for designing high‐performance, room‐temperature, and flexible organic gas sensors.

P3HT/perovskite hybrid gas sensors were engineered to optimize charge transport and sensing performance. While 0D quantum dots enhance carrier mobility via efficient percolation, 1D nanowires provide abundant Br vacancies as active adsorption sites. Consequently, nanowire‐blended films achieve a superior sensitivity of 3.05 %/ppm and a 0.548 ppm detection limit, demonstrating effective halide‐vacancy engineering for high‐performance sensors.

1. Introduction

With the rapid expansion of industrialization, the early detection of hazardous air pollutants such as nitrogen oxides, carbon compounds, sulfur‐based gases, and volatile organic compounds (VOCs) has become increasingly critical [1, 2]. Smart gas sensors capable of detecting these molecules are therefore garnering significant interest research attention [3, 4, 5, 6, 7, 8]. Conventional inorganic gas sensors based on metal oxides exhibit high sensitivity and stability; however, their inherent limitations such as high operating temperatures and poor mechanical flexibility make them unsuitable for integration into wearable or portable devices. Organic semiconductors have emerged as a promising alternatives owing to their lightweight nature, mechanical flexibility, and ability to operate at room temperature [9, 10, 11, 12]. In particular, conjugated polymers are regarded as strong candidates for next‐generation gas sensors because of their low power consumption, process compatibility, and tunable molecular design. Nevertheless, their relatively poor stability against oxygen and moisture, along with inherently low surface reactivity, still hinder their practical commercialization [9, 13]. To overcome these challenges, composites of organic semiconductors and inorganic materials have been extensively investigated as a strategy to simultaneously improve sensitivity and selectivity [14, 15].

Among inorganic materials, metal halide perovskites have recently emerged as highly attractive candidates for electronic applications [16, 17]. Perovskites with an ABX₃ structure offer easily tunable electronic structure, bandgap, and binding energy. Due to their outstanding optical and electrical properties and solution processability, they are widely used in various applications such as solar cells, light‐emitting diodes and photodetectors [18, 19, 20, 21, 22]. These characteristics not only enable perovskites to provide high charge carrier mobility, but also to exhibit tunable structural dimensions with surface defect states, which serve as active sites for gas adsorption. Notably, the valence band maximum of lead halide perovskites (∼−5.5 eV) is well aligned with the highest occupied molecular orbital (HOMO) level of conjugated polymers such as poly(3‐hexylthiophene) (P3HT) [20, 23]. This band alignment facilitates efficient charge transfer across composite interfaces. Unlike other non‐conductive nano‐porous materials, perovskite nanocrystals can therefore be combined with conjugated polymers without compromising their electrical properties, providing a unique advantage for hybrid gas sensors.

Building on these advantages, this study explores the fabrication of organic‐inorganic composite films by blending P3HT with perovskite nanocrystals and systematically investigates how structural evolution and defect engineering in perovskites influence gas sensing performance. Through a comparative analysis of perovskite quantum dots and nanowires, we reveal the critical role of crystallinity and halide vacancies in modulating both the electronic and chemical interactions within the sensing film. This work highlights not only the potential of perovskite‐polymer composites as room‐temperature gas sensors but also the broader opportunity to harness structural and defect control in perovskites for the rational design of next‐generation flexible and wearable sensing technologies.

2. Results and Discussion

2.1. Structural and Morphological Characterization of QDs and NWs

We synthesized perovskite QDs and NWs using a hot‐injection method. While conventional ligands, such as oleic acid and oleylamine, result in the formation of QDs, mixed ligands, including octylamine and oleylamine, promote the anisotropic growth of perovskite NWs. The surface ligand information of the perovskite nanocrystals confirmed by Fourier‐transform infrared spectroscopy analysis is provided in Figure S1. To investigate the crystal morphology of the synthesized perovskite QDs and NWs, transmission electron microscopy (TEM) and scanning electron microscopy (SEM) images were acquired (Figure 1). The TEM images of QDs show a CsPbBr₃ (100) crystal plane of the CsPbBr₃ cubic structure, with an average width of 10.45 ± 2.31 nm (Figure 1d). The SEM analysis of the spin‐coated QDs on a glass substrate showed that the QDs formed a uniform island‐like morphology (Figure 1c). In contrast, the TEM analysis confirmed that the NWs possess both CsPbBr₃ (100) and Cs₄PbBr₆ (110) crystal planes, which were arranged in a uniform 1D structure with an average short‐axis width of 7.81 ± 1.62 nm (Figure 1h). The SEM image of the spin‐coated NWs on a glass substrate showed an agglomerated wire structure.

FIGURE 1.

(a) 100 k and (b) 500 k magnification TEM images of QDs and (c) SEM image of QDs. (d) Average width distribution of QDs. (e) 100 k and (f) 500 k magnification TEM image of NWs and (g) SEM image of NWs. (h) Average short‐axis width distribution of NWs.

X‐ray diffraction (XRD) patterns of QDs and NWs were obtained to analyze their crystal structure (Figure 2a,b). The XRD patterns of the QDs correspond to the cubic CsPbBr₃ structure. Conversely, the XRD profile of the NWs exhibited both CsPbBr₃ and Cs₄PbBr₆ structures. The schematic crystal structures of CsPbBr₃ and Cs₄PbBr₆ are presented in Figure 2c.

FIGURE 2.

XRD patterns of the (a) QDs and (b) NWs. (c) Crystal structure of CsPbBr₃ and Cs₄PbBr₆. (d) Cs 3d, (e) Pb 4f, and (f) Br 3d XPS spectra of QDs and NWs. (g) Atomic percentage of QDs and NWs calculated from SEM‐EDS data. UV‐vis. and PL spectrum of (h) QDs and (i) NWs.

X‐ray photoelectron spectroscopy (XPS) analysis was performed to determine how the morphological transition from QD to NW influenced the chemical state of the perovskite components (Figure 2d–f). Both samples showed almost identical peak positions and shapes, indicating that the oxidation states of Cs⁺ and Pb²⁺ are the same regardless of morphology. These results suggest that the fundamental CsPbBr₃ lattice and the Pb‐Br coordination framework remain chemically intact and that any structural differences between QDs and NWs are not associated with changes in Cs or Pb chemistry. However, a shift of the Br 3d peak toward lower binding energies was observed in the NW samples relative to the QD samples. Such a low‐energy shift indicates that the bromine atoms within the NW samples experience a more electron‐rich local environment. This observation is consistent with the structural reconfiguration in which portions of the connected PbBr₆ network are reorganized into more isolated PbBr₆ ⁴⁻ units (partial Cs₄PbBr₆ formation) during the anisotropic growth of the NWs. The breakup or isolation of the halide framework produces bromine vacancies (Br‐vacancy) and leaves nearby Pb centers under‐coordinated; these under‐coordinated Pb sites can accumulate electron density (partial reduction or enhanced electron localization), thereby increasing the local electron density seen by neighboring Br atoms and shifting their Br 3d peaks to lower binding energies.

An additional and diagnostic feature of the NW spectra is the decrease in Br 3d intensity observed during the XPS measurements (i.e., signal attenuation over the measurement time), which was not seen to the same extent in the QD spectra. This time‐dependent loss of the Br signal is interpreted as evidence of labile halogen species at defect‐rich sites: defective or under‐coordinated halides are more susceptible to beam‐induced desorption or photochemical removal under the XPS probing conditions, leading to signal decays during acquisition [24, 25]. In other words, the NWs contain a higher density of fragile halogen sites (Br‐vacancy precursors) that contribute to both produce the lower binding energy shift and show greater propensity for signal loss under X‐ray exposure, thereby providing strong experimental evidence for increased Br‐vacancy concentration in the NW samples. To quantify the Br‐vacancy ratio, we compared the atomic percentages through energy‐dispersive X‐ray spectroscopy (EDS) mapping (Figure 2g). The ratio of Pb (9.99%) to Br (81.07%) in QDs was approximately 8.11, while the ratio of Pb (14.26%) to Br (71.56%) in NWs was approximately 5.01 (Figures S2 and S3). This confirmed that NWs had a larger amount of Br‐vacancy than QDs.

The optical properties of the QDs and NWs were investigated (Figure 2h,i). Under polarized optical microscopy (POM) with crossed polarizers, the perovskite NWs exhibited bright contrast, whereas the QD films appeared dark (Figure S4). This can be attributed to the strong optical anisotropy and birefringence inherent to the 1D morphology of the NWs. While the isotropic cubic symmetry and random orientation of the 0D QDs maintain the polarization state of incident light, the high aspect ratio of the NWs induces form birefringence, rotating the polarization vector and allowing light transmission through the analyzer. UV‐vis. absorbance spectra exhibited absorption peaks at 512 and 515 nm for the QDs and NWs. The NWs demonstrated background scattering above the absorption edge owing to light scattering effects between formed aggregates. The photoluminescence (PL) spectra of QDs and NWs were investigated at 517 and 519 nm. A red‐shift was observed in the NWs relative to the QDs, attributed to the defect formation during the lattice reconstruction process [26]. The distinct brightness of the NWs in the POM images confirms their strong optical anisotropy, originating from this 1D growth‐driven reconstruction. Specifically, the emergence of Cs₄PbBr₆ regions implies a local reconfiguration of the Pb‐Br network from a fully connected framework to isolated PbBr₆ ⁴⁻ octahedra. This structural transition inherently disrupts the halide sublattice, serving as the primary driving force for the formation of bromine vacancies. These vacancies are more than simple lattice vacancies; they substantially modify the local composition and electronic environment by effectively increasing the relative concentration of Cs and Pb atoms near the defect sites. Consequently, two important chemical implications follow:

(1) Surface electronic structure modulation. The loss of coordinated Br alters local charge distribution: Cs⁺‐rich areas strengthen electrostatic interactions with polar or electron‐rich gas molecules, while under‐coordinated Pb sites act as Lewis acidic centers that can bind electron‐donating analytes. These combined effects increase the density and chemical activity of adsorption sites at the perovskite surface. (2) Opto‐electronic effects and defect states. Br‐vacancies introduce localized electronic states (trap levels) within or near the band edges. Experimentally, this is consistent with the optical data (UV‐vis./PL) showing band‐edge red‐shifts and enhanced defect‐related emission in NW samples. Functionally, these defect states can (i) reduce the optical bandgap, (ii) facilitate charge transfer between adsorbed gas molecules and the perovskite lattice, and (iii) increase hydrophilicity at defect‐rich surfaces each of these effects being favorable for improved gas adsorption and sensor response [25, 26, 27, 28, 29, 30].

2.2. Morphology of Composite Film

Figure 3a,b shows the surface morphologies of pristine P3HT, QD‐blended, and NW‐blended films observed by SEM. The P3HT thin films were well coated on the Si/SiO₂ substrates, however, perovskite nanocrystals tended to form aggregates within the polymer matrix, and the aggregation became more pronounced at higher loading contents. Interestingly, in optical microscopy (OM) images of each film, differences in photoluminescence intensities and emission wavelengths between QDs and NWs resulted in visibly distinct film colors, indicating the preservation of their structural and optical characteristics after film fabrication (Figure S5). Both QDs and NWs maintained their crystal structures upon incorporation into the thin films. The normalized UV‐vis. absorption spectra of the composite films reveal that the overall P3HT crystallinity remained comparable regardless of the QD and NW content (Figure 3c–f). In contrast, the film thickness showed a slight increase with highest perovskite content for both QD‐ and NW‐blended films.

FIGURE 3.

(a) SEM images of pristine P3HT, QD‐blended, (b) and NW‐blended thin films. Inset: high‐resolution SEM image highlighting individual QDs and NWs embedded within the polymer matrix. (c) Normalized UV‐vis. absorption spectra of QD‐blended films. (d) Film thickness and crystallinity data of QD‐blended films. (e) Normalized UV‐vis. absorption spectra of NW‐blended films. (f) Film thickness and crystallinity data of NW‐blended films.

2.3. OFET Performance

The electrical characteristics of the OFETs incorporating QDs and NWs were examined to evaluate how the distinct perovskite dimension influenced charge transport in the P3HT matrix. Pristine‐P3HT film exhibited a low field‐effect mobility of approximately 3.02 × 10⁻⁴ cm² V⁻¹ s⁻¹. In the QD‐blended devices, the drain current and field‐effect mobility gradually increased with QD content, reaching the maximum mobility of 2.39 × 10⁻³ cm² V⁻¹ s⁻¹ at 30 wt.% (Figure 4a,b). CsPbBr₃ perovskite nanocrystals function as direct contributors to electrical charge transport in the P3HT matrix. Although perovskites are not classical conductive materials, their semiconducting nature, characterized by relatively high carrier mobility and moderate intrinsic conductivity, plays a key role in enhancing the drain current. The energy levels of CsPbBr₃ (valence band ≈ −5.6 eV, conduction band ≈ −3.3 eV) align well with the HOMO/LUMO levels of P3HT (HOMO ≈ −5.0 eV). This favorable band alignment minimizes the interfacial injection barrier for holes and facilitates efficient charge transfer across the P3HT/perovskite interfaces. Similar mechanisms are widely utilized in perovskite solar cells, where CsPbBr₃ acts as an efficient hole‐transport or electron‐blocking layer, improving overall device conductivity [20, 23, 31]. NW‐blended devices exhibited smaller enhancement, up to 9.27 × 10⁻⁴ cm² V⁻¹ s⁻¹, than that of the QD blends (Figure 4c,d). This behavior is consistent with the structural characteristics of NWs. Although NWs have higher crystallinity, anisotropic growth simultaneously induces abundant Br‐vacancies and partially formed Cs₄PbBr₆ domains. In contrast to semiconductor CsPbBr₃, Cs₄PbBr₆ is generally regarded as a wide‐bandgap (∼3.8–4.0 eV) and relatively insulating phase with less favorable energy level alignment for charge transport. Therefore, the Cs₄PbBr₆ domains can interrupt efficient interfacial charge transfer and act as interfacial trap sites, leading to increased percolation noise in the transfer curves [32, 33, 34]. The schematic energy level alignment between P3HT, CsPbBr₃, and Cs₄PbBr₆ is illustrated in Figure S6. Consequently, despite the higher crystallinity of the NW system, the coexistence of Br‐vacancy‐related defect states and Cs₄PbBr₆ heterodomains partially limits the charge transport enhancement compared to the uniformly dispersed QD‐based devices. Both QD‐ and NW‐blended devices showed a decrease in drain current at 50 wt.%, which is attributed to excessive aggregation and the subsequent disruption of the continuous P3HT charge transport pathways.

FIGURE 4.

(a) Transfer curve of OFET based on QD‐blended P3HT film (plot of the drain current vs. gate voltage at drain voltage of −60 V). (b) Field‐effect mobility and I_on/I_off ratio of QD‐blended P3HT film. (c) Transfer curve of OFET based on NW‐blended P3HT film (plot of the drain current vs. gate voltage at drain voltage of −60 V). (d) Field‐effect mobility and I_on/I_off ratio of NW‐blended P3HT film. Repeated gas sensing curves for (e) QD‐blended and (g) NW‐blended P3HT films upon exposure to 10 ppm NO₂ followed by air. (f,h) Gas sensing parameters, responsivity, response rate, and recovery rate upon exposure to NO₂ calculated from the dynamic responses in (e) and (g).

2.4. Gas Sensing Performances

To examine the gas‐sensing performance, the gas‐sensor devices were exposed to 10 ppm NO₂ for 10 s and purged with air for 200 s. The drain current (I_D) was measured at −10 V gate voltage (V_G) and drain voltage (V_D) (Figure 4e–h). The responsivity, response rate, and recovery rate were calculated as ΔI_D/I_D0, ΔR/Δt, and ΔR/Δt, respectively, from the first‐exposure cycle (Figure 4f,h). The NW‐blended devices demonstrated significantly superior gas‐sensing characteristics compared to both pristine and QD‐blended devices. Upon exposure to 10 ppm NO₂ repeatedly, both composite systems exhibited enhanced responsivity, rapid response, and partial recovery due to the increased gas adsorption capacity and interfacial interaction sites provided by the embedded perovskites. In QD‐blended P3HT films, the highest responsivity and response rate were observed at 30 wt.%, consistent with the optimal charge‐transport behavior and uniform perovskite dispersion. The sensing enhancement arises from the increased interfacial area between QDs and P3HT, which facilitates efficient charge transfer during NO₂ adsorption.

Conversely, the NW‐blended sensors showed the highest overall performance, with the NW‐30 wt.% device achieving the highest responsivity and the fastest reaction kinetics. This significant improvement stems from the dual contribution of NWs: (1) The high crystallinity and elongated 1D structure improve long‐range electronic coupling and charge transport during sensing transitions. (2) Intrinsic Br‐vacancies serve as chemically active sites that strongly interact with electron‐accepting NO₂ molecules, amplifying the charge modulation in the OFET channel. Even though NW‐blended devices exhibit slightly lower field‐effect mobility due to trap‐induced fluctuations, these specific defects become chemically advantageous for gas sensing by providing a dense network of reactive adsorption sites. Thus, NWs outperform QDs in sensing performance while QDs outperform NWs in charge‐transport enhancement, highlighting the distinct roles of crystallinity, defect states, and dimensionality in OFET‐based gas sensors. The anisotropic growth process of the perovskite nanowires simultaneously induces substantial changes in surface chemical characteristics, including increased Br‐vacancy formation, partial Cs₄PbBr₆ heterodomain generation, and modified surface coordination environments. Therefore, the superior sensing behavior of the NW system is more closely associated with defect‐mediated surface chemical activity and interfacial adsorption characteristics than with dimensionality alone. In addition, the optimal sensing performance observed at 30 wt.% can be attributed to the balanced interplay between charge transport and gas adsorption. At lower perovskite contents, the density of reactive adsorption sites remains insufficient to induce substantial charge modulation. Large amounts of perovskite loading enhances the number of Br‐vacancy‐related adsorption sites and the interfacial area within the composite film, thereby improving gas responsivity. However, excessive loading (50 wt.%) likely promotes nanocrystal aggregation and disrupts the continuous percolation pathways within the P3HT matrix, resulting in reduced carrier mobility and charge transport. Consequently, 30 wt.% represents an optimal balance between adsorption‐site density and charge‐transport continuity.

Furthermore, the gas sensing characteristics of the QD‐ and NW‐blended P3HT devices were evaluated toward NO₂, SO₂, and CO₂ at concentrations ranging from 10 to 40 ppm. Figure 5a,e,i show the dynamic responses of the QD‐blended devices, while Figure 5c,g,k present those of the NW‐blended devices. Upon exposure to each gas (blue‐highlighted region), all devices exhibited an increase in drain current (I_D), confirming the effective charge‐transfer interaction between CsPbBr₃‐P3HT hybrid films and the target gas molecules. Among the three gases, NO₂ produced the strongest current enhancement. This trend originates from the high electron‐withdrawing ability and strong oxidizing nature of NO₂, which efficiently extracts electrons from the P3HT backbone and increases the hole carrier density. SO₂ induced a moderate response due to its weaker oxidizing ability, while CO₂ being nonpolar and chemically inert under ambient conditions exhibited the smallest I_D change. This selectivity is consistent with the expected interaction strength between p‐type semiconductors and electron‐withdrawing gas molecules [35, 36]. The responsivity as a function of gas concentration was extracted from the linear fits shown in Figure 5b,f,j for the QD‐blended devices and Figure 5d,h,l for the NW‐blended devices. All devices demonstrated linear concentration‐dependent responses, confirming reliable sensing in the tested concentration range. Notably, the NW‐blended sensors exhibited a substantially higher responsivity than that of the QD‐blended sensors for all three gases, particularly for NO₂.

FIGURE 5.

Source‐drain current (I_D) vs. time curve of gas sensors based on QD‐blended P3HT thin films when the sensors were exposed to various concentrations of (a) NO₂, (e) SO₂, and (i) CO₂ from 10 to 40 ppm. The blue rectangular region denotes exposure time to NO₂, SO₂, and CO₂ (10 s). Linear fit shows the gas responsivity of QD‐blended devices as a function of the (b) NO₂, (f) SO₂, and (j) CO₂ concentration. Source‐drain current (I_D) vs. time curve of gas sensors based on NW‐blended P3HT thin films when the sensors were exposed to various concentrations of c) NO₂, (g) SO₂, and (k) CO₂ from 10 to 40 ppm. The blue rectangular region denotes exposure time to NO₂, SO₂, and CO₂ (10 s). Linear fit shows the gas responsivity of NW‐blended devices as a function of the (d) NO₂, (h) SO₂, and (l) CO₂ concentration.

This superior performance is attributed to the intrinsic structural characteristics of CsPbBr₃ QDs and Cs₄PbBr₆ NWs. The elongated 1D morphology provides a larger accessible surface area and facilitates more efficient charge transport pathways for gas‐induced modulation. More importantly, the NW growth process inherently generates Br‐vacancies, which serve as highly reactive adsorption sites. These Br‐vacancy sites possess unsaturated Pb²⁺ centers that strongly bind electron‐withdrawing molecules such as NO₂ through Lewis acid‐base interactions. As a result, gas adsorption on the NWs produces greater charge perturbations within the P3HT matrix, yielding enhanced I_D modulation. In contrast, although QD‐blended devices also display improved sensitivity compared to pristine P3HT, their spherical geometry and comparatively lower density of defect sites limit the extent of gas adsorption and the overall charge‐transfer efficiency. The QDs maintain a more stoichiometric CsPbBr₃ composition, resulting in weaker gas‐surface interactions relative to the NWs. In this regard, the observed sensing differences should be interpreted not only in terms of structural dimensionality, but also in terms of the evolution of surface chemical properties accompanying the nanowire growth process. Consequently, the NW‐30 wt.% device exhibited the most pronounced current changes and the steepest responsivity slopes across all gases, particularly toward NO₂. These results highlight the critical role of defect‐engineered perovskites in amplifying gas‐semiconductor interactions and demonstrate that Br‐vacancy‐rich perovskite nanostructures serve as highly effective sensitizers for organic gas sensing platforms.

The overall gas‐sensing characteristics of the composite film are summarized in Figure 6a–d. Both QD‐ and NW‐blended P3HT films exhibited enhanced sensitivity toward NO₂, SO₂, and CO₂ compared to the pristine device; however, a clear selectivity toward NO₂ was observed across all compositions. NO₂ induced the highest modulation of the drain current, consistent with its strong electron‐withdrawing ability and high affinity toward halide‐deficient perovskite surfaces. In contrast, SO₂ and CO₂ being less oxidizing and, in the case of CO₂, nonpolar produced substantially weaker responses, highlighting the intrinsic NO₂ selectivity of the hybrid sensing platform. Among the two types of perovskite nanostructures, the NW‐blended films, particularly the NW‐30 wt.% device, demonstrated the greatest improvement in responsivity, with a sensitivity of 3.05%/ppm for NO₂. This enhancement originates from the higher density of Br‐vacancy formed during the NW growth process, which acts as highly reactive adsorption sites for polar gas molecules. These defect‐mediated interactions lead to stronger charge transfer into the P3HT channel, yielding a larger current modulation compared to that of the QD‐blended devices. The limit of detection (LOD), shown in Figure 6c,d further confirms this trend: the NW‐30 wt.% device achieved the lowest detection limits for all three gases, with a LOD of 0.0055 ppm for NO₂, 0.0228 ppm for SO₂, and 0.0349 ppm for CO₂.

FIGURE 6.

Sensitivity of gas sensors based on (a) QD‐blended and (b) NW‐blended P3HT thin film, where the devices were exposed to NO₂, SO₂, and CO₂ gases. LOD of the (c) QD‐blended and (d) NW‐blended device exposed to NO₂, SO₂, and CO₂.

In addition to the enhanced sensing performance, the perovskite composite devices also exhibited improved environmental stability compared to pristine P3HT devices under moisture‐ and oxygen‐containing ambient conditions. To investigate the influence of environmental factors on gas sensing, humidity‐dependent NO₂ sensing measurements were additionally performed under various relative humidity (RH) conditions (20%, 40%, and 60%) (Figure S7). As the RH increased, the sensing response gradually decreased for all devices because competitively adsorbed H₂O molecules reduced the number of available adsorption sites for NO₂ interaction. Nevertheless, the perovskite composite devices retained relatively higher sensing performance than pristine P3HT under humid conditions. In particular, the NW‐based devices containing Cs₄PbBr₆‐related heterodomains exhibited the most stable sensing characteristics with increasing humidity. This behavior is attributed to the highly polar and Br‐vacancy‐rich surface characteristics of the NWs, which preserve sufficient chemically active adsorption sites for NO₂ interaction even under humid environments.

Consistent with the humidity‐dependent sensing results, the perovskite‐blended OFET devices also demonstrated improved long‐term air stability during prolonged exposure to ambient conditions (Figure S8). Although all devices exhibited gradual electrical degradation due to moisture‐ and oxygen‐induced oxidation of the P3HT layer, the composite devices retained relatively stable electrical characteristics and partial recovery behavior after vacuum treatment. Notably, the NW‐based devices showed the highest stability and recovery efficiency among all devices. The defect‐rich Cs₄PbBr₆‐related surfaces preferentially adsorb environmental moisture species, thereby partially suppressing direct oxidative degradation of the P3HT matrix. Furthermore, moisture adsorbed on the polar perovskite surfaces can be more reversibly desorbed during vacuum treatment, leading to improved recovery characteristics. These results collectively support the important role of Br‐vacancy‐related surface chemistry in determining not only the sensing performance but also the environmental stability of the OFET devices.

Figure 7 schematically illustrates the theoretical sensing mechanism. In the QD‐blended film, gas adsorption primarily occurs on intact CsPbBr₃ surfaces, producing moderate charge transfer. In contrast, the NW‐blended films contain partial Cs₄PbBr₆ domains and Br‐vacancy‐rich regions, where undercoordinated Pb²⁺ sites and halide vacancies significantly increase gas‐binding energies. These defect sites exhibit a strong affinity for electron‐withdrawing gases such as NO₂, resulting in enhanced molecular adsorption and more pronounced modulation of the OFET channel current. The synergy between the defect‐rich Cs₄PbBr₆‐NW structure and the p‐type P3HT matrix ultimately explains both the enhanced sensitivity and the superior NO₂ selectivity of the NW‐based organic sensors.

FIGURE 7.

Schematic of gas sensor based on (a) QD‐blended and (c) NW‐blended P3HT thin film structure. (b) Adsorption configuration of NO₂ molecules on Pb‐terminal CsPbBr₃, Cs‐terminal CsPbBr₃, (d) Pb‐terminal Cs₄PbBr₆ and Cs‐terminal Cs₄PbBr₆. Red, gray, light‐blue, brown and cyan spheres denote O, Pb, N, Br and Cs atoms, respectively.

To investigate the adsorption of NO₂ onto the perovskite surfaces, we compared the binding energies of gas molecules using density functional theory (DFT) calculations. Based on the experimental XRD peak data, the (100) plane of CsPbBr₃ and the (10‐1) plane of Cs₄PbBr₆ were selected and considered as the Pb‐ and Cs‐terminated planes. NO₂ gas was attached to each surface, and the binding energy was calculated using Equation (1) [37, 38]:

$$\mathrm{\Delta }{\mathrm{E}}_{\mathrm{binding}}=\mathrm{E}\left(\mathrm{pero}\mathrm{vski}\mathrm{te}/{\mathrm{NO}}_2\right)-\mathrm{E}\left(\mathrm{perovskite}\right)-\mathrm{E}\left({\mathrm{NO}}_2\right)$$ (1)

where E(perovskite/NO₂) is the total energy of an adsorbed NO₂ on perovskite surface, E(perovskite) is the total energy of the perovskite surface, and E(NO₂) is the total energy of the NO₂ molecule. The calculated binding energy values were confirmed to be stronger for Cs₄PbBr₆ than for CsPbBr₃, which is a Pb‐ and Cs‐terminated planes, at −2.96 and −1.11 eV, respectively. It was suggested that NO₂ binds more strongly to the (10‐1) plane of Cs₄PbBr₆ in NWs than in QDs with only CsPbBr₃, resulting in a more powerful gas sensor performance. Furthermore, the adsorption configurations and binding energies for SO₂ and CO₂ on both CsPbBr₃ (100) and Cs₄PbBr₆ (10‐1) surfaces were additionally calculated and are provided in Figures S9 and S10.

Compared with previously reported OFET‐ and hybrid‐type gas sensors, this perovskite/P3HT composite system exhibits superior room‐temperature sensing performance together with improved environmental stability (Table S1) [39, 40, 41, 42]. In particular, the NW composite device achieved a sensitivity of 3.05%/ppm and a low LOD of 0.0055 ppm toward NO₂ under room‐temperature operation without external heating. Notably, compared with the recently reported P3HT/CsPbBr₃ hybrid gas sensor, this NW‐based device exhibited approximately three‐times higher sensitivity and approximately two‐times lower LOD, demonstrating the effectiveness of the defect‐engineered nanowire structure for enhancing gas‐sensing performance.

More importantly, this work systematically elucidates how perovskite structural evolution and defect chemistry influence gas‐sensing behavior. The enhanced sensing performance of the NW devices was attributed to Br‐vacancy‐rich Cs₄PbBr₆‐related heterodomains generated during anisotropic nanowire growth, which improve gas adsorption capability, charge‐transfer interaction, and environmental stability. Therefore, this work provides important mechanistic insights for designing future organic–perovskite hybrid gas sensors.

3. Conclusions

This work demonstrated high‐performance organic gas sensors based on perovskite QD‐ and NW‐blended P3HT composite thin films. Compared with the QD‐based devices, the NW‐based composite sensors exhibited significantly enhanced gas‐sensing performance, particularly toward NO₂, owing to the formation of Cs₄PbBr₆‐related heterodomains and Br‐vacancy‐rich surfaces generated during anisotropic nanowire growth. These defect‐mediated surface characteristics enhance Lewis acidity, electrostatic interactions, and charge‐transfer processes, thereby promoting stronger adsorption of electron‐withdrawing gas molecules and larger modulation of the OFET channel current.

Among the investigated devices, the NW‐30 wt.% composite sensor exhibited the highest NO₂ responsivity and selectivity, achieving a sensitivity of 3.05%/ppm and a low LOD of 0.0055 ppm under room‐temperature operation without external heating. In addition to the enhanced sensing performance, the NW‐based composite devices also demonstrated improved humidity tolerance and long‐term environmental stability under ambient conditions. The defect‐rich Cs₄PbBr₆‐related surfaces effectively preserve chemically active adsorption sites while partially suppressing moisture‐ and oxygen‐induced degradation of the P3HT matrix. Through direct comparison between QD‐ and NW‐based composite systems, this study systematically elucidates how perovskite structural evolution, dimensionality, and defect chemistry influence gas‐sensing behavior.

Therefore, the significance of this work lies not only in achieving competitive room‐temperature sensing performance, but also in proposing defect‐engineered perovskite nanostructures as versatile functional materials for organic thin‐film‐based hybrid gas sensors. Considering the tunability of perovskite morphology, dimensionality, and defect states through various synthetic approaches, the proposed strategy provides important material‐design guidelines for future organic–perovskite hybrid sensing platforms.

4. Experimental Section

4.1. Materials

All the mentioned chemicals can be purchased from commercial suppliers. The materials used in this work include cesium carbonate (Cs₂CO₃, 99.995%, Sigma–Aldrich), lead bromide (PbBr₂, 98%, TCI), zinc bromide (ZnBr₂, 99.999%, Sigma–Aldrich), 1‐octadecene (ODE, 90%, Alfa Aesar), hexane (anhydrous, 95%, Sigma–Aldrich), octane (anhydrous, 99%, Sigma–Aldrich), toluene (anhydrous, 99.8%, Sigma–Aldrich), ethyl acetate (anhydrous, 99.8%, Sigma–Aldrich), oleic acid (OA, 90%, Sigma–Aldrich), oleylamine (OAm, 70%, Sigma–Aldrich), octylamine (OTAm, 99%, Sigma–Aldrich). The OA, OAm, and OTAm were purified before use.

4.2. Synthesis of Perovskite QDs and NWs

Perovskite QDs and NWs were produced using hot‐injection methods. Based on the reported synthesis method of CsPbBr₃ QDs in the literature, this study optimized the procedure [43, 44]. To synthesize Cs‐oleate of CsPbBr₃ QDs, Cs₂CO₃ (0.3909 g), OA (1.2 mL) and ODE (15 mL) were added to a three‐neck flask and dried under vacuum at 120°C for 10 min. The flask was then filled with N₂ gas for 10 min. This procedure was repeated. Finally, the temperature was increased to 150°C, until a clear solution was obtained. The obtained Cs‐oleate of CsPbBr₃ QDs was stored in a glove box and heated to 150°C before use. QDs precursor was prepared by adding PbBr₂ (0.1321 g), ODE (10 mL), OAm (1.1 mL) and OA (2 mL) into a three‐neck flask and degassed under vacuum for 10 min at 120°C, and the flask was then filled with N₂ for 10 min. This procedure was repeated. Finally, the temperature was increased to 170°C, and 0.6 mL of the heated Cs‐oleate of CsPbBr₃ QDs solution at 150°C was quickly injected. After 5 s of reaction, the mixture was cooled in an ice‐water bath. The crude solution was centrifuged at 8000 rpm for 5 min. The precipitate was redispersed in hexane. The solution was centrifuged at 5000 rpm for 5 min, and the supernatant was collected. Subsequently, ethyl acetate (20 mL) was added, and the solution was centrifuged at 9000 rpm for 5 min. Finally, we obtained CsPbBr₃ QDs by redispersing the precipitate in hexane.

NWs were prepared by modifying a previously reported approach [32]. To synthesize Cs‐oleate of CsPbBr₃/Cs₄PbBr₆ NWs, Cs₂CO₃ (0.325 g), OA (3.16 mL) and ODE (1.84 mL) were added to a three‐neck flask and stirred under N₂ at 110°C for 30 min. The flask was maintained under vacuum at 110°C for 3 h. The obtained Cs‐oleate of CsPbBr₃/Cs₄PbBr₆ NWs was stored in a glove box and heated to 100°C before use. NWs precursor was prepared by adding PbBr₂ (0.2202 g), ZnBr₂ (0.0135 g) and ODE (15 mL) into a three‐neck flask and stirred at room temperature for 10 min. The flask was then degassed under vacuum for 20 min at 120°C. OTAm (2.4 mL) and OAm (2.4 mL) were added to the flask under N₂. The temperature was increased to 135°C and the solution was stirred for 25 min. Next, 0.78 mL of the Cs‐oleate of CsPbBr₃/Cs₄PbBr₆ NWs solution was quickly injected into the flask and stirred for 1 h. The reactant was cooled in an ice‐water bath. The reactant was mixed with hexane and ethyl acetate and centrifuged at 6000 rpm for 5 min. The precipitates were redispersed in toluene for surface treatment in a glove box. PbBr₂ (0.1982 g), toluene (15 mL), OA (1.5 mL), and OAm (1.95 mL) were combined in a vial and stirred at 100°C until PbBr₂ was completely dissolved. The solution was then mixed with the reactant prepared in the previous step and stirred at 85°C for 30 min. The stirred solution was centrifuged at 6000 rpm for 5 min. Finally, we obtained CsPbBr₃/Cs₄PbBr₆ NWs by redispersing the precipitate in hexane. The prepared QDs and NWs solutions were stored at 4°C for further device experiments.

4.3. Device Fabrication

P3HT (Mw = 58 kDa) obtained from Rieke Metals, Inc., was dissolved in chloroform and stirred at 50°C, 550 rpm for 1 h. In parallel, CsPbBr₃ QDs and NWs were dispersed in chloroform and sonicated at 50°C for 2 h. The P3HT and CsPbBr₃ dispersions were then mixed and stirred at 50°C, 550 rpm for 6 h to obtain homogeneous hybrid solutions. The total P3HT concentration was fixed at 10 mg mL⁻¹, while the CsPbBr₃ loading was varied at 10, 20, 30, and 50 wt.%. The blended solutions were spin‐coated onto Si‐SiO₂ substrates (300 nm oxide thickness) at 2000 rpm for 60 s to form thin films. Prior to coating, the substrates were treated with hexamethyldisilazane to improve wettability and film uniformity. For device fabrication, OFETs were constructed in a top‐contact, bottom‐gate configuration. Gold source/drain electrodes (∼400 nm) were deposited by thermal evaporation through a shadow mask, defining channels with length of 100 µm and width of 2000 µm [45].

4.4. Computational Details

The responsivity (R) of the gas sensor was determined as the relative change in drain current upon gas exposure, defined by R = ΔI_D/I_D0 = (I_D – I_D0)/I_D0 where I_D denotes the drain current measured in the presence of the target gas and I_D0 corresponds to the baseline current recorded under ambient air conditions. The response and recovery kinetics were evaluated by calculating the rate of change in responsivity (ΔR/Δt), which reflects the speed at which the sensor signal increases to its maximum value or returns to the baseline after gas removal.

The sensor sensitivity was assessed by analyzing the dependence of the responsivity on the gas concentration (C_t). Specifically, sensitivity was extracted from the slope of the linear fitting of responsivity as a function of gas concentration, and is expressed as follows:

$$\mathrm{S}=\frac{\partial \left(\mathrm{\Delta }{I}_D/I_{D0}\right)}{\partial C_t}$$

The LOD is an important metric for evaluating gas‐sensing performance, as it represents the minimum concentration of analyte that can be reliably distinguished from the background signal. In this work, the LOD was estimated based on the root mean square (RMS) noise of the baseline current. The RMS noise quantifies the inherent fluctuation of the sensor signal in the absence of gas, while the sensitivity describes the magnitude of the sensor response per unit concentration. Accordingly, the LOD was calculated using the following relation:

$$\mathrm{LOD}=\frac{3\times \mathrm{RMS}\mathrm{noise}}{slope\left(sensitivity\right)}$$

To quantitatively evaluate the air stability of the OFETs, the recovery rate was calculated using the following equation:

$$recoveryrate\left(\%\right)=\frac{{\left(I_{off}^{\ast }\right)}_{recovery}-{\left(I_{off}^{\ast }\right)}_{14day}}{{\left(I_{off}^{\ast }\right)}_{0day}-{\left(I_{off}^{\ast }\right)}_{14day}}\times 100$$

Here, I^* _off denotes log₁₀ (I_off), which is used to account for the wide dynamic range of the off‐state current (10⁻⁷–10⁻¹² A) and to enable a clearer comparison of the recovery rates.

All DFT calculations were carried out using the Vienna Ab initio Simulation Package, utilizing the projector augmented wave method. The exchange‐correlation interactions were described by the Perdew‐Burke‐Ernzerhof functional within the generalized gradient approximation. A plane‐wave kinetic energy cutoff was set to 520 eV. For the geometry optimization of slabs, we employed a k‐point mesh of 2×3×1. The structures were relaxed until the residual forces on the atoms converged to less than 0.02 eV/Å. VESTA software was used for all structural visualizations. All structural and electronic visualizations was performed using VESTA software. Surface slabs were constructed from CsPbBr₃ (100) and Cs₄PbBr₆ (10‐1) orientations, incorporating a NO₂ molecule and 20 Å vacuum spacing to eliminate spurious periodic interactions.

4.5. Characterization

The crystalline structures of the QDs and NWs were examined using XRD (MAX‐2500, Rigaku, 40 kV) patterns, which were recorded with a Cu Kα source. QDs and NWs morphologies were observed by TEM (TEM‐ARM200F, JEOL, 200 kV) and SEM (SUPRA40VP, Carl Zeiss, 2 kV). Elemental compositions were analyzed by SEM‐EDS, and surface chemical states were investigated by XPS (ThermoFisher NEXSA G2, calibrated to C 1s = 285 eV). Optical absorption properties were studied using an UV‐vis. spectroscopy (OPTIZEN POP, KLAB). PL (FluoroMax Plus, Horiba Instrument Incorporated, USA) was measured using a spectrofluorometer. For hybrid films, surface morphologies were further inspected by OM (BX51, OLYMPUS) and SEM (JSM‐7800F, 10 kV), while their optical adsorption characteristics were analyzed by UV‐vis. spectroscopy (Lambda 365, PerkinElmer) to evaluate crystallinity and band structure modifications.

The electrical characteristics of the OFET devices were measured under vacuum at room temperature using a semiconductor analyzer (Keithley 4200‐SCS). Gas sensing properties were evaluated at room temperature using a precision gas sensing system (GASENTEST). In the pulse‐cycle tests, NO₂ gas (10 ppm) was injected for 20 s with both gate and drain voltages set to −10 V. For dynamic‐cycle sensing, NO₂, SO₂, and CO₂ gases were tested at concentrations ranging from 5 to 40 ppm, supplied from calibrated gas cylinders premixed with air.

To evaluate the environmental stability of the OFET devices, long‐term air‐exposure experiments were additionally conducted under ambient conditions (average temperature: 21 ± 2°C, relative humidity: 40 ± 10%). The electrical characteristics of the devices were initially measured immediately after device fabrication (0 day), and were subsequently monitored during storage under ambient air conditions for 14 days, followed by additional measurements after vacuum treatment. For the composite devices, the air‐stability measurements were conducted using the optimized 30 wt.% QD‐ and NW‐blended devices. The recovery behavior was analyzed by comparing the transfer characteristics before and after vacuum exposure. The threshold‐voltage shift (ΔV_th), and off‐current variation were monitored to evaluate the degree of oxidative degradation of the P3HT semiconductor layer.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: smtd70781‐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.