Breathing is a vital physiological process in living organisms. For humans, this process involves inhaling air containing oxygen into the lungs, where gas exchange occurs across the alveolar-capillary membrane. Carbon dioxide is excreted during exhalation, released through the nose or mouth. The entire process, from inhalation to exhalation, is known as the breathing or respiration cycle. Respiratory rate, a key vital sign, is used to monitor the progression of illness, with abnormal rates serving as critical markers of serious conditions.

Continuous monitoring of respiratory rate is crucial in various medical settings. Substantial evidence indicates that alterations in respiratory rate can predict potentially serious clinical events, such as cardiac arrest or admission to intensive care units. Studies have shown that respiratory rate is a more reliable indicator than other vital measurements, such as pulse and blood pressure, in differentiating between stable patients and those at risk^{[1, 2]}. Changes in respiratory rate can identify high-risk patients up to 24 h before an event, with a specificity of 95%. Additionally, in the context of COVID-19, respiratory monitoring is critical for evaluating pulmonary function.

Another important area of study is sleep apnea-hypopnea syndrome (SAHS), a prevalent and potentially serious disorder characterized by at least 30 apneas during a typical 7-h sleep period, where each apnea lasts at least 10 s. SAHS leads to sleep interruptions, resulting in various pathophysiological changes, including complications such as hypertension and stroke. SAHS has become the second leading cause of stroke. Therefore, respiratory monitoring of individuals with SAHS is particularly important and urgent^[3].

One effective strategy for respiratory monitoring involves using the moisture content in exhaled human breath, where the relative humidity ranges from 60% to 88%^[4]. In this context, devices with active materials at the nanoscale have emerged as promising candidates for gas sensing applications due to their high surface-to-volume ratio and small physical dimensions, which are comparable to the charge screening length. A good gas sensor must exhibit high sensitivity and selectivity toward specific gases. Additionally, the sensor should offer long-term stability, repeatability, a low operating temperature, and consequently low power consumption. Furthermore, a cost-effective fabrication process is essential from an industrial perspective.

In this context nanostructures, with their high surface area-to-volume ratio, play a crucial role in advancing sensor performance. Their enhanced surface area provides more active sites for gas adsorption, directly boosting sensor sensitivity. In nanostructured sensors, the small scale of the materials provides stronger interactions with gas molecules, allowing for detectable changes in electrical properties when specific gases are present. For example, one-dimensional nanostructures with particle sizes near twice the Debye length, exhibit enhanced responses due to quantum confinement and the formation of space-charge regions^[5]. These unique properties of nanomaterials enable numerous advantages, such as low power consumption, exceptional sensitivity at trace gas concentrations, high accuracy, rapid response times, abundant surface-active sites, and effective operation at room temperature—leading to more efficient and sensitive gas detection compared to their bulk counterparts^{[6, 7]}.

Nanostructures, such as nanotubes^[8], nanoparticles^{[9, 10]}, nanosheets^{[11, 12]}, and nanowires^{[1, 13, 14]}, have demonstrated good sensitivity to different gases. Among these nanostructures, silicon nanowires (SiNWs) have been shown to have substantial advantages^[1]. SiNWs can be processed using relatively standard techniques, allowing for integration with complementary metal−oxide−semiconductor (CMOS) processes for large-scale production. Additionally, SiNWs offer flexible doping concentrations and can be chemically functionalized for the selective detection of gas-phase molecules, making them particularly suitable for advanced gas sensing applications. Of several methods to synthesize SiNWs, metal assisted chemical etching provides a simple, cost effective means to fabricate SiNWs while giving control over various parameters for instance length, shape, orientation and doping level^[15−17].

Incorporating semiconductor metal oxides (MO_x) provides high electron mobility, high sensitivity, fast response/recovery times, long-term stability, and stable chemical properties^[18−20]. Additionally, these materials allow easy adjustment of surface properties. In recent years, n-type semiconducting metal oxides, such as iron oxide, zinc oxide, titanium oxide, tin oxide, tungsten oxide, and indium oxide^{[9, 19−24]}, have become pivotal in gas sensing due to their low cost, high sensitivity, simplicity, and ease of integration into electronics^{[18, 20]}. However, gas sensors based on metal oxides (MO_x) face drawbacks such as high resistance, high operating temperatures, and low selectivity^{[18−20, 25]}. The operation of these devices at high temperature tends to alter the properties of nanoparticles over time, thereby causing reduced stability and shorter lifetime. Further the requiring temperature controlled system for these devices requires additional power consumption induces additional cost and complexity. Several work were proposed to improve the metal oxide based sensors performance, such as surface structuring to increase the number of adsorption sites, for instance using porous Si^[26].

Another vastly employed approach^{[27, 28]} to address these issues, is the use of conducting polymers such as polyacetylene, polypyrrole, polyaniline (PANI), and poly-diacetylene which have been explored^[29−32]. These materials offer high conductivity, low-temperature operation, and low power consumption. Among them, PANI has been extensively studied and widely applied^[18] due to its efficiency in gas sensing, low cost, easy synthesis, room-temperature operation, and low power consumption. However, conducting polymer-based gas sensors suffer from poor stability, selectivity, and long response times^{[24, 25, 30, 33]}.

Hybrid nanocomposites of metal oxides and conducting polymers present a promising solution to enhance the gas-sensing properties of the individual components while maintaining their unique desirable properties. Recent advancements have led to the successful development of PANI-based hybrid nanocomposites by several research groups for detecting various target gases^{[18, 20−22, 33, 34]}.

Therefore, we propose an innovative approach: the synthesis of SiNW-based PANI:metaloxide hybrid nanocomposite structures for respiratory sensing. In this work, SiNWs were synthesized using metal-assisted chemical etching, followed by coating with an organic−inorganic sensing layer comprising PANI and MO_xnanoparticles (MO_x−NPs). The breath sensor operates on a chemisorption-based mechanism, where the target gases (such as those in exhaled breath) interact with the sensor surface, causing a measurable change in resistance. This chemisorption involves the chemical interaction of gas molecules with the active surface of the sensor, leading to electron exchange, which modifies the overall conductivity of the material. When exposed to exhaled gases, each component of the sensor (metal oxide, PANI, and SiNWs) plays a unique role in enhancing the sensitivity, stability, and selectivity of the sensor. The metal oxides nanoparticles provides high sensitivity to gas molecules, particularly oxidizing and reducing gases, due to its natural tendency to interact with oxygen and other gas species. When exposed to gases like CO₂, O₂, or volatile organic gases (VOCs) in the breath, the MO_xsurface adsorbs these molecules, leading to electron transfer reactions that alter the material’s resistance. The incorporated PANI, which is known for its high sensitivity to changes in the environment, particularly to acidic and basic gases enhances sensitivity and provides additional active sites for gas interaction^[27]. The SiNWs contributuion is a high surface-to-volume ratio^[26], increasing the overall surface area available for gas interactions. The SiNW provides structural support and stable electrical conductivity, ensuring a robust signal even under varied respiratory conditions. This approach aims to leverage the combined advantages of both materials, potentially leading to significant improvements in respiratory sensing capabilities.

Synthesis of arrays of random SiNWs were carried out by applying metal (silver, Ag) assisted chemical etching (MACE) on p-type 10 × 10 mm² single-side polished Si-substrate of 525 µm thick, with the resistivity ρ of 0.1−0.5 and 0.009 Ω∙cm. The process steps of the synthesis (shown schematically in Fig. 1) are as follows:

• Deposition of Ag-NPs by immersing Si-substrates in a solution of 3 M HF and 1.5 mM AgNO₃ for 60 s, followed by rinsing in DI-water.

• Etching the resulting sample from the previous step in HF:H₂O₂ (5M:0.4M) solution for 20 min to obtain vertically aligned SiNWs. The etching was abrupted by immersing the sample in DI-water.

• The resulting structure is immersed for few seconds in 60% of nitric acid to remove residual Ag-NPs, and rinsed afterwards with DI-water.

Various schemes of structures are considered in this study (see Figs. 2(a)−2(c) for schematics), one in which SiNWs are coated with PANI, polyaniline MO_xhybrid nanocomposite and spin-caoted MO_x−NPs, respectively.

TiO₂ (a mixture of anatase and rutile phases, with particle size <100 nm) and WO₃ (particle size of ~100 nm) nanopowders from Sigma Aldrich were used for deposition of MO_x−NPs onto SiNWs structures. Initially the nanopowders were dispersed in dimethylformamide (DMF) solvent (0.01 g/ml) and then the solution was sonicated for 1 h. A spin coater was used to coat the SiNWs with prepared nanoparticles suspension. The coated sample was placed on a hot-plate at 90 °C for 10 min to evaporate the solvent.

PANI and PANI:MO_xnanocomposites were deposited on SiNWs by suspending the substrate in the polymerization solution, during chemical oxidative polymerization of aniline with ammonium persulfate (APS) in acidic medium^[35]. For PANI deposition, SiNWs were immersed in the solution containing 0.1 M aniline in 0.1 M H₂SO₄ and kept for 30 min under continuous stirring conditions. A pre-cooled solution of 0.1 M APS in 0.1 M H₂SO₄ was added, drop by drop, to the above obtained previous solution containing the monomer. The polymerization solution was allowed to react for 24 h at room temperature under continuous stirring condition. The resulting precipitate, and SiNWs, were collected and washed several times with deionized water and methanol. The collected powder was dried at 60 °C overnight while the SiNWs decorated with PANI was dried at room atmosphere. Similar procedure was used for deposition of PANI:MO_xnanocomposites on SiNWs with the addition of 0.28 g of TiO₂ or 2.31 g of WO₃ in the monomer solution. An additional glass substrate was placed alongside the sample, for Raman spectroscopy and XRD analysis.

Structural and elemental characterization of the films was conducted using X-ray diffraction Raman spectroscopy (Horiba LabRam Evolution) and scanning electron microscopy (SEM) (Zeiss Supra 35). Empyrean diffractometer by Panalytical was utilised for XRD in a parallel beam geometry with a line-focused copper anode source operating at 45 kV and 40 mA with radiation Cu-Kα (wavelength of 1.54 Å). A parabolic X-ray mirror was used with a 1/2° divergence slit to limit the X-ray spot size on the sample. A parallel plate collimator slit (0.27°) was used in the diffracted beam path followed by a PIXcel detector operating in open-detector mode for XRD and in frame-based mode for reciprocal space mapping (RSM). The X-ray diffraction measurement was performed for 2θ selected scanning range i.e., 10°−50° for PANI coated structures and 20°−80° for PANI:MO_xstructures. Where as, ω : 2θ with scanning range of 4° was performed for RSM measurements. Raman spectroscopy analysis was made using an air-cooled frequency-doubled Nd:Yag (100 mW/1 MHz) with 532 nm excitation line along with an 1800 mm high resolution grating. Each Raman measurement was composed of 20 integrated spectra with an acquisition time of 10 s each. The laser power setting was adjusted to 10% in order to avoid any heating effects during measurements. For sensing characterization the samples were coated with Au-contacts (120 nm thick) in coplanar configuration of 2 × 10 mm² using electron beam deposition method. A schematic representation of structure with metal contacts is shown in Fig. 2(d). The electrical characterization setup contains a source meter (Keithley 2400), a controlled sample stage with a micro-manipulator, vacuum suction and stainless-steel contact arms connected to source meter.

Prior to deposition, MO_xnanoparticles (Sigma Aldrich) were characterized by Raman spectroscopy (see Fig. 3). The Raman spectra of WO₃ nanoparticles showed well-resolved peaks at 248 and 310 cm⁻¹, ascribed to O−W−O bending, and at 695.8 and 789.5 cm⁻¹, attributed to O−W−O stretching^[36]. As expected, the Raman spectra of TiO₂ nanoparticles consist of a mixture of rutile and anatase phases, as indicated by the solid and dashed lines, respectively. Three characteristic active modes, i.e., E_g, B_1G, and A_1G, for both a-TiO₂ and r-TiO₂, were observed and are in line with other studies^{[37, 38]}.

A room-temperature Raman spectra of PANI, PANI:WO₃, and PANI:TiO₂ structures are presented in Fig. 4. The representative Raman spectra of the structures were deconvoluted using Gaussian fitting to determine the characteristic Raman shifts. For PANI, the vibration modes observed at 1574 cm⁻¹ correspond to C=C stretching vibration of the quinoid ring, while 1494 cm⁻¹ is assigned to the C−N stretching vibration, and 1330 cm⁻¹ is associated with C−N stretching vibration of semiquinone radicals, indicating the synthesis of the conductive form of PANI^{[35, 39]}. The room-temperature Raman spectra of PANI:MO_xstructures are shown in the same figure. The characteristic peaks associated with WO₃ and TiO₂ were clearly observed, in relation to those shown in Fig. 3, along with the simultaneous presence of peaks attributed to PANI. Raman spectra of the obtained hybrid composite showed slight alteration in peaks corresponding to vibration modes in PANI, with no evident formation of new phase, due to the interaction of PANI with MO_xnanoparticles, confirming the polymerization and successful formation of composite structure. Further, an overlap of the PANI vibration modes to that from WO₃ can be identified in the figures.

The structural analysis of SiNWs/PANI, SiNWs/PANI:WO₃, and SiNWs/PANI:TiO₂ was further conducted using grazing incidence X-ray diffraction (GIXRD), and the results are shown in Fig. 5. Due to the perpendicular and parallel periodicity of PANI chains, the XRD pattern of PANI exhibits two broad peaks located at 19.4° and 25.47° 2θ values, attributed to (020) and (200) crystallographic planes of PANI^[35]. For WO₃ and r- and a-TiO₂, the peaks are positioned at standard tabulated values according to JCPDS Nos. 98-007-1692, 00-021-1276, and 00-021-1272, respectively^{[37, 38, 40]}, with higher peak intensities indicating the preferred crystallographic orientations. As an example, for WO₃, the peak around 2θ ~29° corresponding to (120) plane is attributed to monoclinic phase. The broadening of peaks observed in case of MO_xnanoparticles, is attributed to the presence of small nanoparticles with mixed crystal grains and the presence of defects in the nanoparticles^{[41, 42]}.

Additionally, to investigate the structural features of SiNWs, X-ray reciprocal space maps (RSMs) around the Si (004) reciprocal lattice point were performed, providing information regarding the out-of-plane lattice, strain, and crystal imperfections^[43]. The X-ray RSMs along (q_z, q_x) coordinates for SiNWs alone and those coated with PANI are presented in Figs. 6(a) and 6(b). The q_xand q_zcoordinates are the projections of the scattering vector along [100] and [001] directions, given by:

1qx=2sin(ω−θ)/λandqz=2sin(θ)/λ.

For SiNWs, an intense peak is located around q_z∈ (0.7360−0.7365) Å⁻¹. For a cubic crystal, the lattice constant a can be expressed as a = 4/q_z^[44], and is calculated to be 5.43 Å, which corresponds to the lattice parameter of bulk Si. This confirms that the MACE process used to obtain SiNWs does not affect the lattice parameter. For SiNWs coated with PANI, the spot broadening increases in both the q_zand q_xdirections, which can be ascribed to bending and torsion acting on the SiNWs due to higher surface energy. Furthermore, the elongation of the q_xarea in the RSM for these structures, which is related to diffuse scattering, is observed and can be associated with crystal imperfections. A wider angular dispersion for SiNWs/PANI is also observed.

Fig. 7 shows the top and cross-sectional views of the SiNWs structures, both uncoated and those coated with PANI, PANI:MO_x, and spin-coated with MO_xnanoparticles (MO_x−NPs). The length of the nanowires was measured to be approximately 6.5 µm, with diameters ranging between 20 and 50 nm (Fig. 7(a)), depending on the size distribution of Ag particles obtained during deposition using AgNO₃. For SiNWs (Figs. 7(a) and 7(b)), the SEM images revealed vertically aligned, randomly distributed nanowires bundled together due to the capillary effect induced by wetting, as is evident in Fig. 7(b). For structures coated with PANI (Fig. 7(c)), the top view showed a cauliflower-like morphology of PANI, similar to that obtained in the study by Lascu et al.^[35].

Figs. 7(d)–7(f) show cross-sectional and top views of PANI:TiO₂ and PANI:WO₃ structures. It was observed that compared to PANI:WO₃, SiNWs coated with PANI:TiO₂ showed agglomeration of TiO₂ nanoparticles over the surface of PANI. Furthermore, the deposition of PANI:MO_xon SiNWs resulted in increased surface coverage, i.e., a possible increase in nanowire bundling, as supported by the RSM plot showing an increased q_xvalue.

For SiNWs structures spin-coated with WO₃ nanoparticles, SEM images are shown in Figs. 7(g) and 7(h). Agglomerated WO₃ nanoparticles can be seen on the SiNWs surface, while the top view reveals smaller particles that infiltrated the nanowires.

Human respiration involves a complex mixture of gases, including nitrogen, oxygen, carbon dioxide, and water vapor, which can cause minor variations in temperature and pressure^{[45, 46]}. Silicon nanowires (SiNWs) have demonstrated a significant piezoresistive (PZR) effect, particularly in low-pressure environments, making them highly suitable for bio-compatible applications like breath sensors^[47−49]. This suggests that the sensing mechanism in breath sensors is likely influenced by both piezoresistive responses and humidity effects. For instance, Gosh et al.^[50] developed a breath sensor using arrays of n-type silicon nanorods, attributing its functionality to the PZR effect. According to PZR theory, pressure changes, such as those caused by airflow over the SiNW surface, can deform the nanowires, leading to changes in their electrical resistance^[47]. In this study, however, the focus is on analyzing the breath-sensing mechanism with respect to chemisorption on the sensing materials, without considering the effects of pressure changes. Additionally, while the structure conceptually targets respiratory gases like oxygen, carbon dioxide, and VOCs, the primary objective was not to examine the effect of these individual gases or VOCs on sensor performance. However, the primary parameter monitored and controlled during measurement and profiling of breath includes resistance which correlated the interaction between the sensor surface and the fluctuating composition of the inhaled and exhaled breath, i.e., the sensor’s ability to detect variations in respiratory patterns (inhalation, exhalation, and pauses in breathing) and establish the sensor’s response to different breathing modes. Other parameters considered are the humidity level and the operation temperature. Thus, during the measurements, the humidity level was kept constant to ensure that variations in resistance could be attributed solely to breath detection rather than fluctuations in environmental humidity as PANI and MO_x-based sensors are often highly sensitive to moisture and can introduce noise in the resistance measurements. Further, by maintaining room temperature(~23.5 °C), the study excludes temperature fluctuations as a variable, ensuring that any changes in resistance are due to respiratory interactions and not temperature shifts. It should be mentioned here that the effect of different gases and VOC’s are underway to determine the effect and sensitivity of sensor towards specific gases during respiration. For instance in our recent work, we have investigated the detection of NH₃, NO₂ gases and the humidity using SiNWs with further exploration aim to study the effect of similar structure presented in this study on selectivity and sensitivity towards VOCs and gases. To facilitate the understanding and effect of humidity, supplementary information provided show effect of varying relative humidity and sensor response for case of SiNWs/WO₃ structure.

Before we delve into the electrical characterization of SiNWs, it is important to understand the sensing mechanism of interconnected silicon nanowires (SiNWs), which operates as follows: to detect respiration, there must be an interaction between moisture and the SiNWs, leading to moisture absorption onto the surface of the SiNWs, facilitated by their highly hydrophilic nature^{[1, 51]}. This phenomenon induces changes in the width of the SiNWs’ hole accumulation layer (HAL) and surface potential, consequently altering the conductance of the SiNWs (Fig. 8). Since moisture itself is not inherently oxidizing or reducing in this context, in the presence of an oxidizing gas, the gas extracts electrons from the SiNWs’ conduction band, narrowing the HAL width^{[52, 53]}. Conversely, in the presence of a reducing agent, electrons are released and trapped by oxygen molecules, resulting in a widening of the HAL width^[52].

Expanding on this, moisture in the breath ionizes into hydronium and hydroxide ions, establishing an electric double layer between the moisture and the SiNWs^{[1, 51]}. As moisture permeates the nanochannels, it induces electron accumulation in the silicon, causing the electrons to migrate towards the anode, in association with holes generated at the cathode. The unique forest-like morphology of SiNWs, coupled with their large surface area, significantly influences the sensor’s response^[1]. In our previous work, we studied respiratory sensing using SiNWs obtained by MACE, which demonstrated a reasonable response to human breath. In this study, we aimed to enhance respiratory sensing properties by using SiNWs decorated with PANI and PANI:MO_xhybrid nanocomposites.

The electrical characterization of the structures was carried out using a custom-built setup. For respiratory sensing, the distance between the sample and the human nose was ~3 mm. Three common respiratory profiles—normal breathing (NB), tachypnea (TB, usually more than 20 breaths per minute^[54]), and deep breathing (DB)—were monitored. The data acquisition speed was set to ~1 ms, limited by the equipment. Fig. 9 shows the sensitivity of the structures by the change in resistance under various respiratory patterns for SiNWs structures coated with PANI, PANI:WO₃, and PANI:TiO₂. The sensitivity and response time of the structures increased with the introduction of MO_xnanoparticles into PANI, following the order PANI < PANI:TiO₂ < PANI:WO₃. The response time R_tis defined as:

2Rt=t90%−t10%,

where t_90% and t_10% are the times when the change due to external stimuli reaches 90% and 10%, respectively. The calculated R_tvalues are tabulated in Table 1 and represent average values calculated over several waveform recordings. Moreover, compared to SiNWs structures coated with PANI:WO₃, the structures coated with PANI and PANI:TiO₂ exhibited distorted features and significant baseline drift, indicating that SiNWs coated with PANI:WO₃ possess enhanced sensitivity for respiratory sensing. Notably, the profiles observed for SiNWs coated with PANI exhibited a decrease in resistance during exhalation, contrary to the increase in resistance observed for PANI:MO_xcoated structures. This behavior can be understood by the different sensing mechanisms in each structure, which will be discussed later. For better visualization, the exhalation and inhalation phases are highlighted in Fig. 9.

The study of breath waveform visualization determines the respiration rate/frequency, however the effect of flow rate variation during different breath modes might also effect the response of these sensor. Since respiratory rate and flow rate are closely connected because different breathing modes generally involve varying lung activity, which affects both the frequency and flow rate of exhaled gases. In general, the respiratory frequency refers to the number of breaths taken per minute. In our study, the three breathing modes—normal breathing (NB), tidal breathing (TB), and deep breathing (DB)—are characterized by distinct frequencies, with DB generally having a slower frequency than NB and TB, which are higher. For instance in case of SiNW/PANI:WO₃ the frequencies obtained by fast fourier transform(FFT) for NB, TB, and DB is ~0.41, 1.43, and 0.23 Hz, which is in agreement with previous reports^{[2, 55, 56]}. Whereas considering flow rate, as the breathing mode changes the lung expansion and contraction dynamics also change, affecting how much air is inhaled and exhaled per breath. For example, deep breathing typically involves a higher tidal volume (volume per breath), which often translates into a higher flow rate during exhalation, even if the respiratory frequency is slower. In case of NB and TB they involve smaller lung volumes and, consequently, lower flow rates per exhalation compared to deep breathing. As the flow rates are not take into account in this study, it is challenging to separate the exact contribution of respiration frequency and flow rate to sensor response. Therefore, the context of their relationship can provide a useful understanding to changes in flow rate likely contributing to differences in sensor response time observed in Table 1. For instance, a higher flow rate generally increases the concentration of exhaled gases that reach the sensor per unit time, potentially causing a quicker and stronger resistance change. In contrast, higher respiratory frequency may increase the frequency of resistance fluctuations in response to the breath cycle.

In contrast, a different sensing mechanism occurs in hybrid PANI/MO_x coated SiNWs structures. In this case, the sensing is predominantly based on moisture trapping within the hybrid nanocomposite, leading to changes in surface resistance when molecular species adsorb and react with the material. While pure metal oxide-coated SiNWs exhibit high sensitivity, PANI/MO_xcoated structures are more porous due to the presence of MO_xnanoparticles within the PANI matrix. This porosity provides active centers that enable localized donor and acceptor states^[20]. An interesting aspect of these structures is the formation of p−n heterojunctions between p-type PANI and n-type MO_xparticles^{[20, 24]}. When exposed to moisture, electron and hole transfer between PANI and WO₃ or TiO₂^[57] forms a depletion layer at the PANI/WO₃ interface, resulting in the creation of a heterojunction barrier^{[18, 20]} (illustrated in Fig. 10). PANI, being a p-type material, has holes as majority charge carriers. Upon moisture exposure, free electrons neutralize the holes, reducing the hole concentration in PANI that increases the heterojunction potential barrier, leading to a rise in resistance in the hybrid structure^{[20, 24]}. As a consequence, this is increasing the overall resistance of the structure. Further, since the conduction band of WO₃ is at higher energy level than that of PANI and Si, the electron will transport form high energy to lower energy region, that is the p-type Si will interact p-type PANI, while the hybrid sensing layer, potentially influencing carrier redistribution and overall device characteristics.

The increased sensitivity observed in PANI:WO₃ can be attributed to WO₃’s high sensitivity to gas molecules, such as oxygen, ammonia, nitrogen oxides (NO_x), and other volatile organic compounds (VOCs) present in exhaled breath. WO₃ ability to detect low concentrations of these gases enhances its effectiveness for respiratory sensing applications. In contrast, while TiO₂ is also sensitive to gas molecules, it generally performs better under UV light due to its photocatalytic properties^[57−60]. WO₃ has a larger surface area and more active sites for gas interaction compared to TiO₂, which has fewer active sites for gas interaction under room temperature and normal light conditions^{[59, 61]}.

For comparison, a room-temperature dR/dt vs time plot for SiNWs/PANI:WO₃ and bare SiNWs is shown in Fig. 11(a). The hybrid structures demonstrate a higher rate of change over time, suggesting greater sensitivity to breath profile variations than bare SiNWs. In addition, bare SiNWs show baseline drift (inset), implying reduced stability compared to hybrid structures, which display minimal drift. The response times are tabulated in Table 1. It is important to note that for bare SiNWs, physical deformation due to inhalation or exhalation pressure could influence resistance through the piezoresistive effect^{[47, 50]}. In contrast, the hybrid layer stabilizes the nanowires, preventing deformation and ensuring that changes in resistance are mainly due to chemisorption rather than physical pressure effects, unlike in the bare SiNWs where both factors may influence the resistance.

A further comparison between SiNWs decorated with PANI:WO₃ and those spin-coated with WO₃ nanoparticles (Fig. 11(b)) reveals that the introduction of a semiconducting polymer composite with WO₃ leads to lower resistance and faster response times. The R_tvalues for SiNWs structures decorated with PANI:WO₃ and WO₃-NPs are approximately 0.37, 0.15, and 0.45 s for NB, TB, and DB respiratory patterns, respectively, compared to 0.43, 0.30, and 1.18 s for the spin-coated WO₃ structure. The enhanced response time is attributed to the uniform distribution and adherence of PANI:WO₃ between interconnected nanowires, unlike the agglomerated WO₃ particles seen in spin-coated structures, which have less adherence and functionalization.

To explain the sensing mechanism in n-type WO₃ and TiO₂ metal oxides, the adsorption and desorption of molecules on the sensor surface lead to changes in resistance. Oxygen molecules in the air react with surface electrons, decreasing electron concentration on the WO₃ surface and increasing the width of the depletion regions^{[18, 19]}, thereby increasing sensor resistance (Fig. 10).

In the case of SiNWs/MO_xstructures, the sensing mechanism involves the formation of a junction between p-type SiNWs and n-type MO_xnanoparticles, modulating the sensing behavior of the nanowires. As documented by Ref. [1], the dominant conductive path is the interconnected SiNWs, with charge carrier diffusion occurring at the interface between the SiNWs and MO_xnanoparticles due to differences in the Fermi levels^[62]. This leads to the formation of a depletion region with an internal electric field at the interface. When O₂ molecules are adsorbed on the surface, electrons are extracted, increasing the hole concentration in p-type SiNWs and disrupting the balance in the depletion layer. This results in a positive charge layer that impedes hole diffusion from p-type SiNWs to MO_x, thus reducing the thickness of the depletion layer^{[62, 63]}.

Returning to the comparison between the hybrid and MO_xSiNWs structures, as observed in Fig. 11, the hybrid structure, which combines cauliflower-like PANI with WO₃ nanoparticles, demonstrated an increased response with more detailed features. This improvement can be attributed to the larger surface area of the PANI:WO₃ composite compared to pure WO₃, allowing for greater molecule adsorption and enhanced sensitivity. Additionally, the acidified PANI, a conductive polymer, has a wider conduction channel, leading to lower resistance in the hybrid structure compared to SiNWs/MO_x, as illustrated by the plots in Fig. 11 and consistent with observations from He et al.^[18]. The hybrid structure also contains more oxygen vacancies than pure WO₃, which facilitates the adsorption of gas molecules. Moreover, as noted by Staerz et al.^[64], nanoparticle agglomeration (as seen in Fig. 7) can reduce sensitivity. This issue has been addressed through the use of templates^[65] and will be further explored by adjusting the concentration of nanoparticles in solution, the length and density of SiNWs, and the parameters of the spin-coating process. The structure presented in the current study, especially SiNWs coated with hybrid PANI:WO₃, showed increased rate of change in resistance compared to those observed in other works^{[55, 66]}. The encapsulation of MO_xnanoparticles in PANI resulted in much lower resistance for instance as observed for hybrid structure i.e., SiNW/ZnO/rGO^[55], and to mention that the voltage applied i.e, translating to the power consumption is much lower than those observed in previous works^{[2, 55, 56]}.

To investigate the stability of the sensor, the structure was stored for more than 2 months in an ambient atmosphere with a variation in relative humidity between 36 ± 3%, while the temperature during that period did not change significantly i.e., 23.5 °C. It is well acknowledged that PANI is chemically stable^[66−68] and that the incorporation of WO₃ nanoparticle along with PANI (both of which posses minimal toxicity^{[68, 69]}) on to SiNWs forming a immobilized solid state device ensures sensing materials remain in place, providing stable and reliable sensing performance while reducing health risks associated with airborne particles or degradation products. After this period, measurements showed negligible changes in resistance or deterioration in the observed respiratory profile (see Fig. 12). In efforts to develop practical solutions for real-time monitoring of breathing patterns, we designed a breath sensor (see Fig. 13). This device features a mask equipped with the sensor and shielded by a grid for optimal performance and protection against potential damage. A photograph of the circuit and the circuit diagram are included in the figure. Data acquisition is conducted using an oscilloscope and an Arduino microcontroller board (not shown here) for recording the data. Additionally, the sensors will be encapsulated and/or coated for instance with polyamide coating to prevent possibility of (if any) airborne particles or flakes from the structures, as also conducted in work by Ghosh et al.^[50].

We have successfully demonstrated the synthesis of a hybrid PANI:MO_xstructure decorated on SiNWs, using a cost-effective and straightforward chemical oxidative polymerization method for respiratory sensing applications. The structure was characterized using Raman spectroscopy, X-ray diffraction (XRD), and reflection spectroscopy microscopy (RSM), which confirmed the presence of PANI encapsulating metal oxide nanoparticles, specifically WO₃ and TiO₂. SEM micrographs revealed the formation of vertically aligned, randomly distributed silicon nanowires covered with a porous PANI:MO_xstructure.

Comparative analysis of SiNWs structures coated with PANI or WO₃ alone showed that those coated with PANI:MO_xhybrid composite exhibited superior performance in respiratory sensing across various breathing profiles at room temperature. In particular, the PANI:WO₃ coated structures demonstrated enhanced sensitivity and response time with minimal to no baseline drift. These findings underscore the promising potential of hybrid PANI-based sensors for both research and industrial applications. Their low fabrication cost, scalability, and ease of manufacture make them a viable option for developing efficient respiratory sensing technologies.