Hydrothermal growth and liquid-phase exfoliated VS₂ nanoflakes towards ammonia gas sensing performance at room-temperature

Abstract

Two-dimensional transition metal dichalcogenides (TMDs) have attracted significant interest in gas sensing applications due to their large surface area and favorable electronic properties. In this work, bulk VS₂ was prepared hydrothermally and subsequently exfoliated into thin nanoflakes, as evidenced by structural and morphological analyses that showed preserved crystallinity and sheet-like morphology. The exfoliated VS₂ nanoflakes were employed as an active sensing material for room-temperature (28 °C) ammonia (NH₃) detection. The fabricated sensor exhibited strong selectivity toward NH₃ compared with other interfering gases and demonstrated reliable detection over a concentration range of 300 ppb to 2 ppm under ambient conditions. The device showed a strong response (%), excellent repeatability, and fast response/recovery speed. The sensors showed a strong sensitivity of 89% to 2 ppm NH₃ within 100 s, with prompt recovery in 160 s. The enhanced sensing performance was attributed to adsorption-induced charge transfer between NH₃ molecules and the highly conductive exfoliated VS₂ nanoflakes. These results indicate that exfoliated VS₂ is a promising candidate for low-power, room-temperature (28 °C) gas sensing applications.

Response (%)–recovery turn of NH₃ gas sensing with different ranges.

1. Introduction

Ammonia (NH₃) is a colorless, highly pungent, and toxic gas that humans can exhale under both healthy and unhealthy conditions.¹ In healthy individuals, breath NH₃ levels usually range from 50 to 2000 ppb, which can vary depending on the severity of chronic kidney diseases and gastric disorders.² Additionally, exposure to high concentrations of NH₃ can cause serious health problems, including irritation of the eyes, nose, skin, and respiratory system. Prolonged exposure may lead to severe lung damage, pulmonary edema, or even death.³ Moreover, NH₃ is widely used in refrigeration, chemical manufacturing, and fertilizer production, from which it can be emitted at higher concentrations, causing acid rain and accelerating eutrophication in aquatic systems, both of which negatively impact ecosystems and human health.⁴ Therefore, monitoring breath NH₃ levels around 100 ppb and detecting leaks in pipelines and storage systems are essential to prevent explosions and exposure to dangerous gases. Although Transition Metal Dichalcogenides (TMDs) and metal-oxide-based sensors can detect roughly 100 ppb, most require higher operating temperatures.⁵ Room-temperature NH₃ sensors are highly desirable to reduce power consumption and enable portable sensing devices.⁶ Therefore, developing reliable gas sensors that are easy to prepare and operate and that can detect low levels of NH₃ with straightforward instrumentation remains a key focus of ongoing research.

Recent years have experienced a rise in research on NH₃ gas sensors, driven by advances in nanostructured materials, simple fabrication methods, and breakthroughs in nanoelectronics.^4,7–9 These developments have enabled the deployment of NH₃ sensors in various practical fields, including healthcare monitoring, environmental assessment, food safety, and industrial process control.^7,8,10 Various sensing materials have been explored for NH₃ detection, including metal oxide semiconductors, conducting polymers, carbon-based materials, and two-dimensional (2D) layered nanomaterials.^9,11 Although metal oxide sensors often exhibit good sensitivity, they typically require elevated operating temperatures, which increases power consumption and fabrication complexity.^12,13 Conducting polymers can operate at room-temperature but often suffer from limited stability.^14,15 Carbon nanomaterials offer high surface area and electrical conductivity; however, challenges with selectivity and reproducibility remain.^13–16 Therefore, there is a high demand for sensing materials that function efficiently at room temperature.

Vanadium disulfide (VS₂) belongs to the layered TMDs family, which has attracted significant attention for gas-sensing applications due to its large surface-to-volume ratio and tunable electronic properties.^11–15 Additionally, VS₂ demonstrates metallic conductivity, a high carrier density, and numerous surface-active sites that enable efficient charge transfer during gas adsorption, making it a promising material for gas sensing.^16–18 Structurally, VS₂ consists of a layered sulfur–vanadium–sulfur configuration in which individual layers are strongly bonded within the plane and stacked via weak interlayer van der Waals interactions.^19,20 This layered structure allows bulk material to be split into a thin nanostructure, increasing surface area and improving interactions with gas molecules, which is beneficial for sensing applications.²¹ Although VS₂ powder is commercially available, it was prepared hydrothermally in this study to control the growth of crystalline nanostructures at relatively low temperatures (100–200 °C) and to reduce impurity levels.^22,23 Additionally, previous reports indicate that hydrothermally synthesized TMDs have large accessible surface areas, which enhance their electrochemical performance, a key requirement for gas sensing applications.^22,24,25

Previously, sensing materials based on VS₂ have been studied for detecting gases such as nitrogen dioxide (NO₂) and hydrogen (H₂), humidity,^21,26 and specific volatile organic compounds.^4,14 In contrast, research on NH₃ detection using liquid-phase exfoliated VS₂ nanoflakes, especially at room-temperature and at low concentrations, remains limited. The metallic nature of VS₂ and its strong surface activity suggest that exfoliated nanoflakes will provide improved sensing performance for NH₃ gas.^23,27,28

In this work, liquid-phase exfoliation was employed to synthesize metallic VS₂ nanoflakes by dispersing bulk VS₂ powder in a suitable solvent and then subjecting the dispersion to ultrasonic treatment.¹⁷ The structural and morphological characteristics of the exfoliated nanoflakes were analyzed using various characterization techniques, and the resulting nanoflakes were used as the active sensing material for a chemiresistive NH₃ sensor operating at room-temperature (28 °C). The fabricated sensor demonstrated detection capabilities in the low-concentration range of 300 ppb to 2 ppm, offering improved sensitivity, rapid response, and stability. The chamber temperature was stabilized before each test to maintain thermal equilibrium. In addition, the sensing characteristics were evaluated under ambient air at 28 °C and 65 ± 5% relative humidity, which served as the baseline condition, and under optimized humidity conditions to simulate practical operating environments. These findings suggest that exfoliated VS₂ nanoflakes are promising for low-power, room-temperature (28 °C) NH₃ detection.

2. Experimental

2.1. Materials

Ammonium metavanadate (NH₄VO₃) and thioacetamide (C₂H₅NS), used as a precursor for vanadium (V) and sulfur (S), were procured from Sigma-Aldrich, India. Ammonium hydroxide (NH₄OH) was purchased from Sisco Research Laboratories (SRL), India. High-purity carbon dioxide (CO₂), hydrogen (H₂), and ammonia (NH₃) gases were supplied in compressed cylinders (purity ≥ 99.99%) by Linde Bangladesh Limited. Hydrogen sulfide (H₂S) was used as a certified low-concentration gas mixture (ppm level, balanced with air or N₂). Commercial liquefied petroleum gas (LPG) was used as received from a standard domestic cylinder. Ethanol (C₂H₅OH), methanol (CH₃OH), and acetone were used as liquid sources (analytical grade, ≥99% purity). All chemicals were of analytical grade (98% purity) and used without further purification.

2.2. Synthesis of VS₂ powder

VS₂ powder was synthesized through a hydrothermal process using analytical-grade reagents. A stoichiometric amount of 1.052 g (8.9 mmol) of ammonium metavanadate (NH₄VO₃) was dissolved in 80 mL of an aqueous ammonium hydroxide (NH₄OH) solution under magnetic stirring at 2400 rpm for 2 h to ensure homogeneous mixing. 3.756 g (0.05 mol) of thioacetamide (TAA) was added, and stirring continued for 30 min until the reddish-brown solution gradually turned black, indicating the formation of VS₂ species. The resulting mixture was transferred to a Teflon-lined stainless-steel autoclave and heated to 200 °C for 24 h. After hydrothermal treatment, the autoclave was allowed to cool naturally to room-temperature. The black precipitate was collected, washed thoroughly 3 times with ethanol and DI water to remove residual reactants or impurities, and then dried in an oven set at 60 °C for 3 h to obtain black VS₂ powder.

2.3. Liquid-phase exfoliation of VS₂ powder

The synthesized VS₂ powder was subjected to Liquid-Phase Exfoliation (LPE) to produce few-layer nanoflakes. The powder was dispersed in a binary mixture of ethanol and DI water (BMED, 1 : 1 by volume) and ultrasonicated in a bath sonicator for 3, 5, and 7 h at room temperature (28 °C).¹⁷ Ultrasonic cavitation separated weakly bonded van der Waals layers without disrupting the crystal framework. The resulting dispersions were centrifuged at 3000 rpm for 30 min to separate exfoliated nanoflakes from unexfoliated residues. The upper one-third of the centrifuged solution was collected as the supernatant, which contained exfoliated VS₂ nanoflakes, which were then drop-cast onto clean glass substrates and dried at room-temperature for further structural, optical, and morphological characterization. A schematic diagram of the VS₂ exfoliation process is shown in Fig. 1.

Fig. 1. LPE steps for the formation of VS₂ nanoflakes.

2.4. Characterization techniques

The crystal structure and phase identification of bulk and exfoliated VS₂ nanoflakes were examined by X-ray diffraction (XRD) using CuKα₁ radiation (λ = 1.5406 Å). Raman spectra of the samples were measured with a Raman spectrometer (Horiba Scientific, Macro Ram, λ = 532 nm) using the maximum step size. Atomic force microscopy (AFM) was used to analyze the surface morphology and thickness of the samples. Transmission electron microscopy (TEM) (Thermo Fisher, Czech Republic) was employed to obtain low-magnification images, high-resolution TEM (HRTEM) micrographs, and selected area electron diffraction (SAED) patterns of the exfoliated samples. An X-ray photoelectron spectrometer (XPS, ULVAC-PHI 5000) was used to analyze the elemental composition of exfoliated VS₂. I–V characteristics of the samples were measured using a standard four-point probe technique. A Hall Effect Measurement System (Ecopia HMS-5000, South Korea) was used to measure the carrier concentration, mobility, and Hall coefficient of the nanoflakes by applying a magnetic field of 0.34 T and a current of 10 mA at room temperature (300 K).

2.5. Test gas preparation and concentration estimation

Ethanol (C₂H₅OH), methanol (CH₃OH), and acetone (C₃H₆O) vapors were generated from their respective liquid sources using a controlled evaporation method. A known volume of each liquid was introduced into a sealed gas-sensing chamber and allowed to evaporate completely, producing the corresponding vapor.

Gas sensing experiments were conducted in a cylindrical chamber with an internal volume (V_chamber) of 2.0 L, determined from the internal dimensions of the chamber. Prior to measurement, sufficient time was allowed to ensure uniform mixing of the gas or vapor inside the chamber.

The vapor concentration shown in eqn (1) or vapor generated from liquid sources was calculated using the ideal gas relation:

1

where n is the number of moles of the injected liquid, R is the universal gas constant (0.082057 L atm mol⁻¹ K⁻¹), T is the absolute temperature (303 K), P is the atmospheric pressure (1 atm), and V_chamber is the chamber volume (2.0 L). The number of moles was determined from the injected volume using the liquid's density and molar mass. For example, for methanol, the molar mass (M = 32.04 g mol⁻¹) was used to convert the injected mass into moles. The calculated concentration assumes complete evaporation and uniform distribution throughout the chamber.

Table 1 lists the equivalent liquid volume required to generate the target vapor concentration, which was calculated using the following relation:where V_l is the liquid volume (mL), C is the gas concentration (ppm), V_c is the chamber volume (L), M is the molecular weight (g mol⁻¹), and ρ is the liquid density (g mL⁻¹). The value 24.45 represents the molar gas volume at room temperature.

Table 1. Physical properties of the volatile compounds used in the gas sensing experiments.

Analyte	Molecular weight (g mol−1)	Density (g mL−1)	Liquid volume for 2 ppm (µL)
CH3OH	32.04	0.792	0.0066
C2H5OH	46.07	0.789	0.0096
C3H6O	58.08	0.791	0.0120

Using this calculation for the 2.0 L sensing chamber, the estimated liquid volumes corresponding to a 2 ppm vapor concentration were determined. Since these theoretical values are extremely small and below the practical injection limit of conventional micropipettes, the Volatile Organic Compound (VOC) vapors were generated within the sealed sensing chamber using the serial dilution method, and their concentrations were estimated using the ideal gas law. Therefore, the reported concentrations represent estimated equilibrium vapor concentrations rather than directly injected microliter liquid volumes.

For gases supplied from compressed cylinders (NH₃, H₂S, CO₂, and H₂), the concentration inside the chamber was controlled by injecting a known volume of gas and estimated using a volumetric dilution method shown in eqn (3).

3

where V_gas is the injected gas volume and V_chamber is the total chamber volume (2.0 L). The concentration values were calculated assuming uniform gas distribution inside the chamber.

2.6. Gas sensor fabrication and measurement

The sensing device was fabricated by depositing an exfoliated VS₂ nanoflake dispersion onto gold (Au) contact electrodes patterned on a glass substrate, as illustrated in Fig. 2. As a control experiment, a bare Au interdigitated electrode was exposed to 2 ppm NH₃ under identical conditions. No significant change in current was observed, confirming that the sensing response originates from the VS₂ sensing layer rather than the Au substrate. The deposited film was naturally dried at room temperature (28 °C) to form the active sensing layer. The device was then configured in a two-probe arrangement for electrical measurements.

Fig. 2. Representation of the NH₃ sensing system and the interaction between NH₃ molecules and exfoliated VS₂ nanoflakes.

Gas sensing experiments were carried out in a sealed glass dome chamber under ambient laboratory conditions (28 ± 2 °C). Before introducing the target gas, the chamber was evacuated using a vacuum system to eliminate residual gases until a stable baseline current was obtained. A fixed volume (mL) of analyte was introduced using a liquid-based vapor generation method. The liquid analyte was heated under control conditions to ensure complete evaporation, and the generated vapor was subsequently transferred to the sensing chamber for interaction with the VS₂ sensor. A constant bias voltage of 5 V was applied using an electrometer, which simultaneously supplied voltage and recorded the resulting current. The temporal variation of current was continuously monitored during gas exposure and recovery cycles. Upon exposure to gas, an increase in current was observed, attributed to charge transfer between the adsorbed gas molecules and the metallic VS₂ sensing layer. When the gas supply was stopped, the current gradually returned to its baseline value, indicating good recovery behavior.

The sensor response, S (%) was calculated using eqn (4)

4

where I_g and I_a represent the current in the presence of the target gas and in air, respectively. The response (%) and recovery times were defined as the time required to reach 90% of the total current change during adsorption and desorption, respectively. Each measurement was repeated five times to ensure reproducibility.

The same experimental procedure was followed to evaluate the sensor's selectivity toward other gases, including carbon dioxide (CO₂), ethanol (C₂H₅OH), methanol (CH₃OH), acetone (C₃H₆O), liquefied petroleum gas (LPG), hydrogen sulfide (H₂S), and hydrogen (H₂). All analytes were introduced under identical experimental conditions, using the same vapor-generation approach to ensure consistent comparison.

3. Results and discussion

3.1. Growth mechanism

VS₂ nanoflakes form from NH₄VO₃, NH₄OH, and TAA through precursor decomposition and sulfurization. First, NH₄VO₃ dissolves in the reaction solution, releasing vanadium ions, as shown in eqn (5). NH₄OH stabilizes the vanadium species and adjusts the solution's pH shown in eqn (6). During heating, thioacetamide slowly decomposes shown in eqn (7) and (8), releasing sulfide ions (S²⁻) as the sulfur source shown in eqn (9).^29,30 These S²⁻ ions react with the vanadium species to form VS₂ nuclei shown in eqn (10). As the reaction continues, the nuclei grow and form layered structures because VS₂ naturally prefers flake-like growth, with strong bonding within the layers and weak interactions between layers. After synthesis, the stacked layers are exfoliated, producing ultrathin VS₂ nanoflakes with a larger surface area, as shown in eqn (11).

NH₄VO₃  (aq) → NH₄⁺ + VO₃⁻ [dissolution] 5

NH₄OH  (aq) ⇌ NH₄⁺ + OH⁻ [alkaline solution] 6

7

H₂S + OH⁻ ⇌ HS⁻ + H₂O 8

HS⁻ + OH⁻ ⇌ S²⁻ + H₂O [formation of sulfide] 9

Vanadium species react with sulfide to form VS₂ nuclei:

V (aq species) + 2S²⁻ → VS₂ (stacked layered flakes) [sulfurization] 10

Subsequently, the powder was sonicated, generating shear forces on the stacked VS₂ layers, which disrupt the van der Waals interactions between layers, leading to exfoliation, as confirmed by the reduction in layer size observed in XRD results.

VS₂ (stacked layers) → VS₂ (2D nanoflakes) [exfoliation] 11

3.2. Structural and morphological analysis

The crystalline structure and phase purity of the prepared bulk VS₂ powder and exfoliated VS₂ nanoflakes were investigated using XRD, as shown in Fig. 3(a). The bulk VS₂ sample exhibits sharp, intense diffraction peaks at 2θ = 15.3°, 31.9°, 35.7°, 45.5°, 59°, and 65°, which can be indexed to the (001), (100), (011), (012), (110), and (004) planes, respectively. These reflections confirm the formation of hexagonal layered VS₂, and the peak positions agree well with the JCPDS standard reference data (file no. 01-089-1640), indicating good crystallinity of the bulk material.^24,25 After exfoliation, the diffraction peaks remain at nearly the same 2θ positions, indicating that the intrinsic crystal phase is preserved during ultrasonication-assisted exfoliation. The exfoliated sample exhibits a pronounced decrease in diffraction intensity compared with the bulk material. This reduction in intensity can be attributed to a decrease in the coherent scattering volume due to reduced stacking thickness and the formation of few-layer nanoflakes, which lower long-range stacking order.²⁶ Notably, the intensity of two peaks (012) and (011) that are prominent in the bulk sample is suppressed in the exfoliated sample, indicating that exfoliation reduces the bulk-like stacked component and weakens higher-index reflections. However, the (011) peak remains visible in the exfoliated sample, confirming that the in-plane crystalline structure of VS₂ is retained after exfoliation. Peak broadening observed after exfoliation validates the effectiveness of the process. The FWHM (β) of the major peak (001) increases from 0.195° for bulk to 0.228° for exfoliated VS₂, as shown in Fig. 3(b), indicating a reduction in crystallite size and stacking thickness due to the shear forces generated during ultrasonication. The disappearance of the peak (110) also indicates the exfoliation of the layer from its bulk counterpart.¹¹ The average crystallite size was estimated from the (001) peak using the Scherrer equation.^26,31,32 The calculated crystallite sizes are approximately 41 nm for bulk and 35 nm for exfoliated VS₂, confirming a reduction in nanoscale thickness after exfoliation.

Fig. 3. XRD patterns of (a) bulk and exfoliated VS₂ nanoflakes; (b) enlarged view of (001) peak broadening.

The Raman spectra of bulk and exfoliated VS₂ samples are shown in Fig. 4(a). Two characteristic vibrational modes observed near 145 cm⁻¹ and 278 cm⁻¹ correspond to the E_1g and A_1g phonon modes of layered VS₂. After exfoliation, both peaks exhibit a noticeable shift toward lower wavenumbers (A_1g: 278 → 272 cm⁻¹ and E_1g: 145 → 140 cm⁻¹), as shown in Fig. 4(b), accompanied by reduced intensity and slight broadening. These changes indicate weakened interlayer coupling and lattice softening associated with thickness reduction during the exfoliation process.^33,34 Additional weak features observed at higher Raman shifts may originate from structural disorder introduced during ultrasonic treatment. The observed Raman modifications provide clear evidence of the successful exfoliation of bulk VS₂ into thinner nanoflakes, induced by ultrasonic cavitation and shear forces during liquid-phase processing.

Fig. 4. (a) Raman spectra of bulk and exfoliated VS₂ nanoflakes; (b) enlarged view of A_1g and E_1g modes showing peak shift.

The AFM topography reveals the presence of flakes-like nanostructures distributed on the substrate surface after the exfoliation process, as shown in Fig. 5. The thickness of the VS₂ nanoflakes, as determined from the height profile and 3D image, is approximately 8 nm. The thickness of a single VS₂ layer is reported to be about 0.6–0.7 nm, indicating that multilayer (∼12) VS₂ nanoflakes form via the LPE process in our case. These results provide clear evidence of the successful thinning of bulk VS₂ into nanoscale flakes after exfoliation.³⁵

Fig. 5. AFM image, height profile, and 3D image of the clearly seen VS₂ nanoflakes.

Fig. 6(a) shows a low-magnification TEM image of exfoliated VS₂, revealing thin, flake-like structures with lateral dimensions in the nanoscale range. The semi-transparent contrast of the flakes further confirms the reduced thickness of the nanoflakes.³⁶

Fig. 6. (a) Low magnification TEM, (b) high resolution TEM images, and (c) SAED pattern of exfoliated VS₂ nanoflakes.

HRTEM analysis (Fig. 6(b)) shows clear lattice fringes with an interplanar spacing of d = 0.30 nm, which is associated with the (001) plane of VS₂. The corresponding SAED pattern, shown in Fig. 6(c), exhibits distinct diffraction rings that can be indexed to the crystalline planes of VS₂, confirming phase preservation after exfoliation. The SAED pattern is consistent with those observed in the XRD analysis, demonstrating the hexagonal crystalline nature of the exfoliated VS₂.^11,37

3.3. Chemical state analysis

XPS analysis was performed to examine the chemical state of the component, the valence-band position, and the binding energies of the exfoliated VS₂ nanoflakes. The survey of the exfoliated VS₂ sample clearly detects signals of the V and S elements in the respective binding energy regions, as shown in Fig. 7(a).³⁸ No impurities were found in the samples, aside from adventitious carbon (C 1s peak at 285 eV). Fig. 7(b) shows that the V 2p spectrum exhibits four characteristic peaks at 516 eV, 518 eV, 524 eV, and 525 eV corresponding to V 2p_3/2 and V 2p_1/2, respectively, confirming the presence of V⁴⁺ and V⁵⁺ in the sample.²⁶ The S 2p spectrum in Fig. 7(c) shows a doublet with peaks around 168.54 eV (S 2p₃) and 169 eV (S 2p₁), verifying the presence of oxidized sulfur in the sample.²⁶ The relatively higher binding energy of the sulfur peaks may be attributed to surface oxidation arising from surrounding conditions.³⁶ According to literature, a doublet of V 2p is usually found between 510 eV (V 2p_3/2) and 520 eV (V 2p_1/2), and the S 2p doublet appears between 160–170 eV, resulting from S 2p_3/2 and S 2p_1/2 spin–orbit splitting.³⁷ Here, upon exfoliation, the S 2p peaks shift slightly to higher binding energies, consistent with increased levels of organic sulfur, which are typically generated through exfoliation. This change in sulfur oxidation plays a crucial role in breaking bulk material into 2D nanosheets and often influences the resulting defect structure, which serves as an active adsorption scattering center for gas sensing.³⁸ The O 1s spectrum shown in Fig. 7(d) displays triplet peaks at 530.28, 531.31, and 533.89 eV. The lowest binding energy (530.28 eV) corresponds to the stoichiometric metal–oxygen bonds in the material, while the peak at 531.31 eV results from surface defects caused by oxygen vacancies and the reaction of lattice oxygen with atmospheric moisture.²⁶ The small peak at higher binding energy (532.89 eV) is attributed to adsorbed water or surface contamination.³⁹

Fig. 7. (a) XPS survey of exfoliated VS₂ nanoflakes, (b) V 2p, (c) S 2p region, and (d) O 1s.

3.4. Electrical transport properties of exfoliated VS₂ nanoflakes

The room-temperature I–V characteristics of both bulk and exfoliated VS₂ were measured to explore their electrical behavior. As illustrated in Fig. 8(a), both samples exhibit nearly linear I–V curves, indicating good electrical contact between the VS₂ film and the Au electrodes. The exfoliated VS₂ nanoflakes conduct a higher current than the bulk sample at the same voltage, indicating enhanced charge transport after exfoliation. The resistance drops from approximately 833 MΩ for bulk VS₂ to 641 MΩ for exfoliated VS₂. This improvement is attributed to the thinner nanosheets and more efficient charge-carrier mobility in the exfoliated VS₂ network.

Fig. 8. (a) I–V characteristics graph of VS₂, (b) average Hall coefficient at different conditions, and (c) resistivity decrease with increasing carrier concentration.

Hall-effect measurements were conducted to investigate the electrical properties of the exfoliated VS₂ nanoflakes shown in Fig. 8(b). The Hall coefficient was negative for all samples, ranging from −50.43 to −26.38 cm³ C⁻¹, confirming the material's n-type conductivity. The carrier concentration was on the order of 10¹⁷ cm⁻³, indicating a high density of charge carriers in the sensing film. The resistivity remained in the range of 10⁻² Ω cm, indicating good electrical conductivity. Small differences in the Hall parameters may be related to slight variations in film thickness and nanosheet distribution on the substrate.

The n-type conductivity of VS₂ aligns well with the surface-controlled chemiresistive gas-sensing behavior of exfoliated VS₂ nanoflakes, as later confirmed by a gas-sensing experiment. The influx of n-type majority-carrier electrons narrows the electron-depletion layer near the surface, increasing conductivity and strengthening the current response.

3.5. Gas sensing performance

The gas-sensing characteristics of the exfoliated VS₂ nanoflakes were evaluated at room temperature (28 °C) using synthetic air as the reference atmosphere.^40,41 The sensor response was determined by monitoring the change in electrical current upon exposure to different target gases. To maintain consistent environmental conditions, all measurements were conducted within a fixed time between 1:10–1:20 p.m., during which the ambient relative humidity was at its maximum (∼65 ± 5%) shown on Fig. 9. The selectivity was examined by exposing the device to various gases, including NH₃, CO₂, C₂H₅OH, acetone, LPG, H₂S, CH₃OH, and H₂ at a fixed concentration of 3 ppm.^42,43 The relatively lower response to other gases confirms the preferential interaction of NH₃ with the VS₂ surface.

Fig. 9. Time-dependent relative humidity in the presence of TMDs samples in the test chamber under identical test conditions.

The gas-sensing mechanism of exfoliated VS₂ nanosheets at room temperature is attributed to surface adsorption and charge-transfer interactions between the gas molecules and the sensing material. When the target gas is introduced, gas molecules absorb on the surface of the VS₂ nanoflakes, increasing the charge-carrier density and electrical conductivity under a constant bias, thereby increasing current. Since VS₂ exhibits metallic or highly conductive behavior, the sensing response (%) is primarily governed by adsorption-induced modulation of the carrier density rather than by depletion-layer formation. Ultrasonication-assisted exfoliation generates oxidized sulfur compounds, surface defects such as sulfur vacancies and unsaturated V atoms (V⁴⁺/S²⁻ defects), along with dangling bonds and distortions at edge and corner atoms, serving as effective adsorption centers. Since NH₃ is a Lewis base containing a lone pair of electrons, it can interact strongly with these active sites. When NH₃ molecules adsorb on the surface and interact with the nanoflake defect sites, which serve as low-energy absorption centers. This process enhances charge transport through the sensing film, leading to an increase in current during NH₃ exposure, as shown in Fig. 10. The high density of accessible adsorption sites in the exfoliated nanoflakes contributes to the improved sensing performance compared with bulk VS₂. The signal becomes stable when adsorption reaches saturation.⁴² The sensing performance of exfoliated 2D-VS₂ compared to bulk can be attributed to the larger surface area and the increased number of active adsorption sites generated during exfoliation.

Fig. 10. Illustration of the charge transfer mechanism between NH₃ gas and VS₂ nanoflakes.

In our case, it is conjectured that when the exfoliated VS₂ surface is exposed to dry air, oxygen molecules might adsorb onto it and trap free electrons, forming an electron-depletion layer that reduces the baseline conductivity. During NH₃ gas purging, the gas molecules react with the negatively charged, adsorbed oxygen ions (O_2(ads)⁻) on the VS₂ surface, releasing the trapped electrons back into the conduction band of the VS₂ nanoflakes. This results in a change in electrical current, as shown in eqn (12) and (13). When gas flow stops, and synthetic air is introduced, NH₃ molecules start desorbing from the sensor surface, causing the current to decrease baseline.

O_2(air) + e⁻ → O_2(ads)⁻ 12

4NH₃ + 3O_2(ads)⁻ → 2N₂ + 6H₂O + 3e⁻ 13

The process is reversible and stable at room temperature. Fig. 11(a) shows the transient current responses to different gases, and the corresponding histogram is summarized in Fig. 11(b). Due to the high electrical conductivity of VS₂, the sensing response (%) is mainly governed by modulation of the charge-carrier density upon gas adsorption. Reducing gases such as NH₃, H₂, CH₃OH, and LPG donate electrons to the sensing surface, thereby increasing conductivity. In contrast, CO₂ shows a comparatively stronger interaction with the VS₂ surface, resulting in a noticeable sensing response. The enhanced sensing performance is attributed to the high surface-to-volume ratio and the increased number of active sites in the exfoliated nanoflakes, which promote effective gas adsorption and charge transfer.

Fig. 11. (a) Selectivity of the sensors based on the 2D VS₂ nanoflakes exposure to various gases, and (b) histogram of the VS₂ flakes-based device to various gases.

Based on the strong selectivity toward NH₃, the concentration-dependent sensing behavior was investigated over the range of 300 ppb to 2 ppm, as shown in Fig. 12(a). A pronounced response (%) is observed in exfoliated VS₂ nanoflakes, whereas the bulk VS₂ sample shows no response (%), demonstrating the weak interaction between bulk and gas molecules.⁴³ Exfoliation induced defects in the samples, thereby enhancing their interaction with gas molecules.⁴³ The calibration curve in Fig. 12(b) shows that the response (%) rises from about 18% at 300 ppb to 88% at 2 ppm, indicating reliable detection at low concentrations. To the best of our knowledge, there are no reports demonstrating a significant response (%) at these low concentrations. The repeatability of the device was examined at 2 ppm NH₃, as shown in Fig. 12(c), where consistent current changes are observed across five successive cycles, confirming stable, reproducible sensing performance at room-temperature.

Fig. 12. (a) Response (%)–recovery turn of NH₃ gas sensing with different ranges, (b) response (%) as a function of NH₃ concentration, and (c) repeatability test of the gas sensor for 5 cycles with 2 ppm NH₃ at room temperature.

The response (%) and recovery characteristics of the exfoliated VS₂ sensor were analyzed from the transient curves, as shown in Fig. 13(a). The response time (τ_res) and recovery time (τ_rec) were estimated as the times required to reach 90% of the total current change during adsorption and desorption, respectively.^11,44 The sensor exhibits a response time of approximately 100 s and a recovery time of 160 s at 2 ppm NH₃. The longer recovery time compared to the response time can be attributed to stronger adsorption of NH₃ molecules on the active sites of the exfoliated nanoflakes, which requires additional time for desorption.⁴⁵

Fig. 13. (a) Calculation of response time (τ_res) and recovery time (τ_rec) for NH₃ gas at 2 ppm, and (b) response times (τ_res) and recovery times (τ_rec) in response (%) to NH₃ concentration.

Fig. 13(b) shows the variation in response and recovery times with NH₃ concentration, where both parameters increase with increasing gas concentration due to enhanced surface interaction at higher molecular coverage. This occurs because a higher concentration of NH₃ increases the number of molecules that interact with active sites, prolonging the desorption process.⁴⁵ Therefore, the response (%)–recovery behavior is affected by the surface adsorption energy and the gas diffusion rate into the sensing layer.

Table 2 compares previously reported VS₂-based gas sensors, indicating that most studies focus on NO₂ detection at relatively high gas concentrations using complex synthetic methods.^46–49 Kang et al. exclusively reported that the VS₂ nanosheet responded to 10 ppm NH₃ at 40 °C.²⁶ In contrast, the sensor developed in this work can detect NH₃ at a lower concentration of 2 ppm and operates at room temperature (28 °C) using a simple LPE approach. The ability to detect such a small amount of NH₃ at room-temperature with high stability is particularly important for practical applications, as it enables low-power operation and improves device safety.⁵⁰ Therefore, the achieved sensing performance, combined with the low-cost, scalable fabrication process and room-temperature (28 °C) operation, underscores the strong potential of the developed sensor for real-world gas-detection applications.

Table 2. A comparison study of VS₂-based sensor materials with this work.

Materials	Synthesis method	Operating temperature (°C)	Gas purge	Gas concentration (ppm)	Response time (s)	Recovery time (s)	Ref.
VS2–octylamine	Hydrothermal method	120	NO2	50	100	15	46
VS2–NH4 +	Solvo-thermal NH3 intercalation	100	NO2	10	177	10	47
VS2 nano-composites	Hydrothermal and annealing technique	180	NO2	50	26	15	48
VS2 on CNFs	Liquid-assisted exfoliation	60	NO2	23	9	10	49
VS2 nanosheets	Facile LPE	40	NH3	10	—	—	26
VS2 nanoflakes	LPE	Room temp (28)	NH3	2	100	160	This work

4. Conclusions

In this study, a few-layer VS₂ nanoflake was successfully produced via LPE assisted by bath sonication. The formation of thin 2D nanoflakes with improved surface features compared to bulk VS₂ was confirmed by XRD and Raman spectroscopy, although the prepared 2D VS₂ consists of 11–12 layers, as shown by the AFM image. The exfoliated VS₂ nanoflakes exhibited n-type semiconducting behavior and enhanced charge-carrier transport, demonstrating their suitability for efficient room-temperature NH₃ sensing. The prepared VS₂ gas sensor exhibited a strong, selective 89% response to NH₃ at room-temperature (28 °C). The sensing signal increased with increasing NH₃ concentration from 300 ppb to 2 ppm, confirming clear concentration-dependent behavior. The sensor also showed stability over 5 repeatable recovery cycles, with response–recovery times of 100 s and 160 s, demonstrating reliable sensing performance. This low-level NH₃ gas sensing can be designed for early leak detection in hospitals and industrial environments to protect personnel from respiratory hazards; refrigeration; food spoilage detection; and fertilizer safety in cold storage, meeting safety standards before risks become severe. These results support exfoliated VS₂ nanoflakes as a promising, low-cost material for practical NH₃-sensing devices.

Author contributions

Rutaba Jania: investigation, formal analysis, methodology, graphics, original draft preparation. Mehnaz Sharmin: editing, reviewing, and supervision. Kazi Hanium Maria: conceptualization, original draft preparation, writing, editing, reviewing, and supervision.

Conflicts of interest

There are no conflicts to declare.

Data availability

On behalf of all authors, the corresponding author would like to confirm that most of the data used in this research is presented in the article. All the graphs have been drawn using Origin software.