Hierarchical Nanoflowers of Colloidal WS₂ and Their Potential Gas Sensing Properties for Room Temperature Detection of Ammonia

Abstract

A one-step colloidal synthesis of hierarchical nanoflowers of WS₂ is reported. The nanoflowers were used to fabricate a chemical sensor for the detection of ammonia vapors at room temperature. The gas sensing performance of the WS₂ nanoflowers was measured using an in-house custom-made gas chamber. SEM analysis revealed that the nanoflowers were made up of petals and that the nanoflowers self-assembled to form hierarchical structures. Meanwhile, TEM showed the exposed edges of the petals that make up the nanoflower. A band gap of 1.98 eV confirmed a transition from indirect-to-direct band gap as well as a reduction in the number of layers of the WS₂ nanoflowers. The formation of WS₂ was confirmed by XPS and XRD with traces of the oxide phase, WO₃. XPS analysis also confirmed the successful capping of the nanoflowers. The WS₂ nanoflowers exhibited a good response and selectivity for ammonia.

1. Introduction

Effective air quality management requires regular monitoring of both indoor and outdoor environments for the detection of harmful and toxic pollutants such as NH₃. Even though semiconducting metal oxides (SMOs) are still the front runners in the category of chemiresistive gas sensors, their high energy input requirement (300–500 °C) is a major drawback [1,2]. Therefore, an alternative that can be incorporated into low-power operating air monitoring systems and still provide a high response even at room temperature (RT) is sought [3]. As a result, the gas sensing research community is putting a lot of effort into the research on other materials such as the transition metal dichalcogenides (TMDCs). TMDCs are layered semiconducting materials with outstanding electronic, chemical and mechanical stabilities [3,4] and a large surface-to-volume ratio [3,5]. Another interesting feature about TMDCs is their layer-dependent properties such as the band gap [4], indirect-to-direct band gap transition [6,7], electronic transport [5,8] and gas sensing [5]. WS₂ is steadily catching up with its TMDC family member, MoS₂, with respect to the research dedicated to its potential to detect NH₃. Huo et al. reported on the dynamic response of exfoliated multilayer WS₂ nanoflakes upon exposure to the reducing gas NH₃ at room temperature [9]. A transfer of charge between the adsorbed NH₃ and the exfoliated multilayer WS₂ nanoflakes resulted in increased conductivity. NH₃ donated electrons to the WS₂ nanoflakes, thereby inducing an n-type behavior. High NH₃ room temperature sensitivity was also observed with thin films of WS₂ obtained via a plasma-assisted synthesis [6]. However, the thin films displayed incomplete recovery. This is a common feature with nanomaterial-based sensors operating at low temperatures; the thin films can be subjected to annealing or UV light illumination to speed up the recovery. Remarkable selectivity to interferents, good selectivity, significant response/recovery rates and a p-type response was reported by Li and co-workers when WS₂ nanoflakes produced by ball milling commercial powders were exposed to NH₃ at 20–200 °C [10]. The sensor response increased with an increase in humidity level (up to 73% relative humidity, RH).

The influence of layer numbers on the recovery rate after removal of NH₃ gas also came under scrutiny in Qin et al.’s study [11]. The WS₂ thin films in this study were fabricated from nanosheets which were obtained via lithium ion intercalation. The monolayer-based thin film had the shortest recovery time compared to its few-layered and bulk counterparts. It was proposed that the NH₃ that is in the interlayers of WS₂ is not easily desorbed, hence the slow or incomplete recovery by few-layered thin films. Results from a study by Perozzi and co-workers on the thermal stability of thin films prepared from commercially bought WS₂ revealed progressive oxidation to WO₃ within the 25–450 °C range [12]. Changes in morphology were noted at 150–250 °C and regarded as the indication of the formed WS₂/WO₃ composite. The gas sensing performance of the WS₂/WO₃ composite towards NH₃ decreased at annealing temperatures above 150 °C. Meanwhile, the 150 °C annealed composite recovered completely and displayed no cross sensitivity to water vapor at 60% RH. High sensitivity and fast recovery to room temperature NH₃ sensing and p-type character were shown by the hydrothermally prepared nanocomposite of Pt QDs/WS₂ nanosheets [13]. Other composites, such as nanosheets of WS₂|O [3] and WS₂/TiO₂ [14], also exhibited high sensitivity, excellent selectivity and fast recovery at room temperature.

The nanoflower morphology presents a larger surface area, a larger number of active sites at the edges of the WS₂ layers and many interlayer spaces, which are all desirable in gas sensing. To exploit the properties of WS₂ nanomaterials, methods that enable precise control of the morphology are required. The reported synthetic routes for WS₂ nanoflowers are hydrothermal [15,16,17,18,19,20,21,22,23] and CVD [24,25]. Colloidal synthesis also offers precise control of the reaction parameters in order to obtain desirable morphology. It is catalyst-free, template-free, one-pot, easily scalable and short one-step synthesis at relatively low temperatures. It also accommodates the use of a capping agent which offers protection of the nanoparticles against agglomeration [26], modifies the surface of the nanoparticles, may introduce new functionalities [27] and can influence the type of morphology formed [28]. Oleylamine (OLA), an N-terminated ligand, is suitable for transition metal semiconductors. OLA is low cost, and can act as a solvent and reducing agent, thereby eliminating the use of many chemicals; it can also lower the decomposition temperature of a metal precursor. Nanoparticles capped by OLA are easily dispersed in various organic solvents with improved properties and potential applications in various fields [26,29]. Colloidally synthesized WS₂ nanoflowers have not been extensively reported and their gas sensing potential has not been explored.

In this study, a colloidal method with OLA as both solvent and capping agent was followed to obtain hierarchical WS₂ nanoflowers. The nanoflowers were obtained after 45 min and were tested at room temperature for potential ammonia gas sensing properties. A good response and higher selectivity were observed towards ammonia vapors than with acetone, chloroform, ethanol and toluene. The dependence of sensing properties of a particular sensor on its method of synthesis is well known. To our knowledge, colloidal hierarchical nanoflowers of WS₂ have not been synthesized for application in gas sensing.

2. Materials and Methods

2.1. Chemicals

Analytical grade tungstic acid (H₂WO₄), thiourea (CS(NH₂)₂), oleylamine (OLA), ammonium hydroxide (NH₄OH), ethanol (CH₃CH₂OH), hexane (C₆H₁₄), chloroform (CHCl₃), toluene (C₆H₅CH₃), acetone ((CH₃)₂CO) and isopropyl alcohol (C₃H₈O) were purchased from Sigma-Aldrich, Johannesburg, South Africa and DieLab, Curitiba, Brazil. Interdigitated electrodes (IDEs) of electroless nickel immersion gold (ENIG) (18 pairs, 7.9 mm long and 0.1 mm apart) were purchased from Micropress SA, Curitiba, Brazil.

2.2. Synthesis of WS₂ Nanoflowers

Hierarchical nanoflowers of WS₂ were synthesized by mixing H₂WO₄ (0.05 mol) and CS(NH₂)₂ (0.2 mol) in 20 mL of degassed OLA. The mixture was heated at 25 °C with continuous stirring under N₂ gas flow for 15 min in a three-neck round-bottom flask. The temperature was increased rapidly to 320 °C to allow for the decomposition of both precursors and held at this temperature for 45 min. After cooling the black reaction mixture for 5 min, ethanol was added in order to separate the colloids. WS₂ nanoflowers were collected by centrifugation and washed several times with a mixture of ethanol and hexane (1:1). The WS₂ black powders were dried at room temperature.

2.3. Material Characterization

2.3.1. Optical Characterization

The WS₂ powder was dispersed in CHCl₃ and placed in quartz cuvettes (1 cm path length) for UV–Vis absorption and PL spectral measurements on the Specord 50 Analytik Jena UV–Vis spectrophotometer and Agilent Cary Eclipse fluorescence spectrometer, respectively. Raman spectroscopy (Bruker Senterra Infinity 1 software, 50× optical objective, 532 nm laser wavelength, 0.2 mV laser power and integration power of 15 s) measurements of the dry powder of WS₂ nanoflowers were used to estimate the number of layers.

2.3.2. Structural Characterization

Measurements were performed on the WS₂ powder with the Bruker MeasSrv (D2-205530)/D2-205530 diffractometer to shed light on the structure and phase of the powdered nanoflowers. X-ray photoelectron spectroscopy measurements were carried out with a PHI 5000 Versaprobe—Scanning ESCA Microprobe (100 µm 25 W 15 kV Al monochromatic X-ray beam) to determine the surface properties of the powders. Sizes and morphologies of the nanomaterials were studied using an FEI Nova NanoLab FIB/SEM and a JEOL JEM-2100 field emission gun transmission electron microscope (200 kV) Chemical composition of the crystal structure was analyzed with energy-dispersive X-ray spectroscopy (EDS) integrated into the TEM instrument.

2.4. Device Fabrication and Gas Sensing Measurements

The WS₂ nanoflower-based sensor was fabricated by drop casting 20 μL of the 5 mg/mL WS₂ toluene dispersion on clean IDEs. The sensor was dried in an oven at 130 °C for 30 min. A surface profiler (Dektak XT; Bruker) was used to measure the thickness of the film (about 600 nm). A custom-made gas chamber, depicted in Figure 1, housed the sensor. The gas chamber was grounded and stabilized for 1 h before measurements were started.

An LCR meter (Agilent 4284A 20 Hz–1 MHz Precision LCR meter) was attached to a computer interfaced with a GPIB for data acquisition and set at an operational voltage and frequency of 1000 mV and 10 kHz, respectively. A large variation in conductance and relative signal-to-noise were obtained at these parameters. The measurements were performed in the dark and under dry nitrogen to provide a controlled atmosphere. The sensor was exposed to incremental concentrations of NH₃ (NH₄OH was used as the source of NH₃) at RT (~23 °C) and 25% RH. A 50 s stabilizing period followed by introduction of 1.5 µL of NH₄OH (analyte) at every 200 s was applied per measurement. The time interval was enough to evaporate the analyte and saturate the chamber. The measurements were taken with concentrations of NH₄OH in the range of 240–958 ppm. The selective character of the sensor was established by exposing it to incremental concentrations of interferents such as acetone, chloroform, ethanol and toluene. Response and recovery measurements were carried out to determine how fast the sensor responds to NH₃ and how fast it recovers to its initial chemical status after NH₃ is removed. The effect of humidity on the sensor’s gas performance was determined by measuring its response to NH₃ under various RH levels.

3. Results

3.1. Characterization of WS₂ Nanoflowers

The crystal phase, crystallinity and composition of the WS₂ nanoflowers were determined with PXRD. The diffraction patterns, as shown in Figure 2, confirmed the nanoflowers to be the 2H-WS₂ polytype according to PDF No: 00-002-0131\JCP2.2CA. The diffraction peaks correspond to the (002), (004), (101), (103), (006), (008) and (200) planes. The prominence of the (002) plane suggested the existence of more than one layer of WS₂ with good crystallinity while its broadness is an indication of a reduction in size. Closer analysis hinted at partial oxidation based on the small peak at 29.5° (denoted by an asterisk) which was assigned to WO₃. This came as no surprise as WS₂ is known for spontaneous oxidation.

XPS studies provided detailed composition of the nanoflowers. The W and S peaks are clearly displayed on the survey spectrum, shown in Figure 3. Partial oxidation of WS₂ to WO₃ was confirmed by the presence of the O 1s peak. The oxygen peak is also attributed to the oxidation of OLA. The successful capping of the nanoflowers by the organic ligand OLA was confirmed by the presence of the strong C 1s peak. Atewolegun et al. used XPS in their studies to confirm the success of ligand exchange on colloidal quantum dots [30]. The capping of the samples is further confirmed by the large carbon composition (77.4%) of the sample, as seen in

Table 1. Summary of the atomic composition stoichiometric assignments obtained from the fitting of the XPS spectra of WS₂ nanoflowers.

Element	Atomic %	Peak Binding Energy (eV)	Assignments	Peak Area %
C	77.4	284.8	C-C	91
		286.3	C-O	6
		288.9	O-C=O	3
O	10.8	532.2	C-O	100
W	3.7	30.9	WS2	41
		33.2	WS2	38
		34.4	W5+	6
		36.3	WO3	15
S	6.2	160.5	S2W def.	9
		161.6	S2W	60
		162.8	S2W	31

.

Figure 4 shows the high-resolution core level spectra of WS₂ with the deconvoluted C-C peak by OLA. The O 1s, C-O and O-C=O peaks were attributed to the oxidation of OLA. Four components of the W4f core level spectrum were identified; W 4f_7/2 (30.9 eV), W 4f_5/2 (33.2 eV) doublet, W 5p_3/2 (34.4 eV) and WO₃ (36.3 eV). Meanwhile, the W 4f_7/2 and W 4f_5/2 are ascribed to the 2H-WS₂ polytype. W 5p_3/2 is attributed to the partially coordinated W in WO₃. The S2p spectrum has three peaks, S₂Wdef (160.5 eV), S₂W (161.6 eV) and S₂W (162.8 eV) which were observed in the deconvoluted core level spectrum.

Raman spectroscopy was used to estimate the number of the layers in the nanoflowers. The WS₂ nanoflowers showed Raman features which are characteristic of the second order longitudinal acoustic 2LA(M) and out-of-plane A_1g(Γ) at approximately 351 cm⁻¹ and 417 cm⁻¹, respectively, as seen in Figure 5. The shoulder peak at 313 cm⁻¹ belongs to the in-plane E_12g(Γ). The shift to lower frequencies from bulk for both A_1g(Γ) and 2LA(M) (356 and 421 cm⁻¹, respectively) is associated with decreasing interlayer interactions by van der Waals forces. This suggested the formation of few layers, in agreement with Varghese et al. [31] and Tan et al. [32]. The calculated frequency difference between A_1g(Γ) and 2LA(M) as well as the intensity of 2LA(M), suggesting a reduction in size from bulk to a few layers.

SEM analysis of as-synthesized WS₂ revealed nanosheets arranged as petals to form the nanoflower morphology, as seen from Figure 6A,B. These nanoflowers self-assembled to form hierarchical structures. Figure 6B gives a clearer picture of the well-ordered nanosheets with interspaces. This feature of nanoflowers translates to a larger surface area, which is ideal for gas sensing.

Meanwhile, the TEM image from Figure 7A confirmed that the nanoflowers were made up of individual nanosheets. The edges of the nanosheets are exposed and serve as active sites. The microspheres in Figure 7B are consistent with the nanoflower morphology recorded by SEM in Figure 6A,B.

UV–Vis absorption and photoluminescence spectroscopy further confirmed the reduced size of WS₂. Figure 8 shows the blue-shifted 625–630 nm and 505–515 nm peaks corresponding to the excitons A and B (636 and 525 nm for bulk WS₂, respectively) [33]. The existence of few layers was further suggested by the diminished photoluminescence peak, shown in Figure 8. A similar result was reported by Gutiérrez et al. and was said to be the result of a competition between indirect and direct transitions [34].

3.2. Gas Sensing Properties of the WS₂ Nanoflowers

The sensing principle of thin films of semiconducting TMDCs is generally based on the change in conductance due to the reaction of the gas that is adsorbed on the surface. An electric charge is transferred between the gas (target analyte) and the active material (WS₂ nanoflowers), causing changes in the electrical properties of the sensing material. Preliminary gas sensing measurements at RT under dry N₂ (25% RH) revealed a fast decrease in the conductance, G, of the WS₂ nanoflowers. The sharp decrease in conductance, G, as shown in Figure 9A (red curve), was due to the adsorption of NH₃ gas molecules on the surface of the WS₂ nanoflowers. The negative slope (sensitivity = −7.99 ppm⁻¹) as shown in Figure 9B confirmed the decrease in conductance. The response was linearly proportional to the concentration of NH₃. NH₃ is a Lewis base and therefore serves as an electron donor. In this work, the NH₃ lone pair electron was transferred to the conduction band of the WS₂ nanoflower-based sensor upon adsorption. The reduction in the electrical conductance suggested that positive holes are the main charge carriers on the surface of WS₂ nanoflowers, hence the p-type doping behavior. This is due to the depletion of positive (hole) charge carriers on the sensor surface by the negative charge carriers from NH₃. A similar p-type behavior was observed previously for MoS₂ and WS₂ sensors by Järvinen et al. [35]. The gas response of the WS₂ nanoflower-based sensor to NH₃ vapors was determined from the variation in the conductance, G, by using ΔG/G₀ in the equation below:

$$S=\frac{\Delta G}{G_0}$$

(1)

where ΔG = (G − G₀). G is the maximum conductance under NH₃ vapors while G₀ is conductance under dry N₂. The sensitivity, S (in ppm⁻¹), of the sensor was determined from the slope of the fit.

Specificity or selectivity is another parameter that determines the practical use of a sensor. It is the ability of a sensor to detect the target analyte in the presence of other contaminants. The specificity value for a sensor is between 0 and 1 with a value closer to 1 representing high selectivity of the sensor for the target analyte relative to the interferents. Consider a particular sensor and a set of n species; the specificity to the particular species i, δ_i can defined as follows:

$${\delta }_i=\frac{{\left(\raisebox{1ex}{$\Delta G$}\!\left/ \!\raisebox{-1ex}{$G_0$}\right.\right)}_i}{{{\displaystyle \sum }}_{j=1}^n{\left(\raisebox{1ex}{$\Delta G$}\!\left/ \!\raisebox{-1ex}{$G_0$}\right.\right)}_j}$$

(2)

The equation above was adapted from Llobet et al. [36] and was used to calculate the specificity value (δ) of the sensor to NH₃ and the interferents. In this work, the sensor showed higher selectivity for ammonia than to acetone, chloroform, ethanol and toluene, as shown in Figure 10. The specificity values of the vapors are summarized in

Table 2. Specificity values of the sensor to each chemical vapor.

Analyte	Concentration (ppm)	Specificity
Acetone	502	0.086
Ammonia	240	0.23
Chloroform	444	0.031
Ethanol	632	0.026
Toluene	352	0.045

. Therefore, WS₂ nanoflowers have the potential for use as elements in chemical sensor arrays.

For practical purposes, a sensor must respond fast upon exposure to the target gas and recover just as fast when the gas is removed. The characteristic sensor curve (dynamic range) shown in Figure 11 was analyzed and used to estimate the response and recovery speed at incremental concentrations of NH₃. The conductance of the sensor decreased sharply (~28 s) upon exposure to NH₃, giving a negative response, and slowly recovered (~42 s) when removed from NH₃. Slow recovery at RT is common to TMDC thin film-based NH₃ sensors [11,37] due to strong interactions between NH₃ molecules and the active layer. This leads to analyte accumulation on the surface of the sensor [38].

Theoretical calculations predicted that the analyte is physically adsorbed on the surface of a perfect 2D monolayer [39]. However, this is not the case for few-layered and bulk WS₂ sensors where the NH₃ molecule can be inserted into the inner layers and interact with the two adjacent layers, as shown in Figure 12 below [11].

Intercalation compounds such as (NH4⁺)_X(NH₃)_Y(WS₂)^x− may be formed. The intercalated NH₃ molecules are more difficult to desorb than the surface molecules. This behavior is similar to the intercalation and deintercalation processes of NH₃ with layered TiS₂ and TaS₂ which were detailed by McKelvy and Glaunsinger [40]. Based on the confirmation of the existence of a few layers of WS₂ by XRD, Raman and TEM, the argument of intercalated NH₃ molecules can be applied in this study to explain the slow recovery of the sensor. Furthermore, the nanoflower morphology has interspaces which provide more reactive sites; thus, more NH₃ molecules end up deeper inside the layers, leading to a slow recovery. It is worth mentioning that defects are introduced during colloidal synthesis, and this too provides more reactive sites than the perfect lattice of a 2D monolayer. The WS₂ nanoflower-based sensor recovered completely after irradiation with UV light for 60 min or heating in an oven for 10 min at 100 °C.

Water molecules are easily adsorbed on the surface of the sensor, leading to an increase or decrease in its performance. This is common with RT chemical sensors. As the WS₂ nanoflower-based sensor is designed for application in RT sensing of NH₃, it was imperative to measure its performance under various levels of humidity. Figure 13 illustrates the response of the sensor at different RH conditions. The effect of humidity on the gas sensing performance of WS₂ nanoflowers was found to be more pronounced at 97% RH.

The increase in NH₃ response in the presence of humidity could be attributed to the increased acidity due to the water molecules on the surface of the WS₂ nanoflower-based sensor [10]. Such an increase can be explained by the hydroxylation reaction below:

SO₄²⁻ (WS₂ surface) + H₂O ↔ H⁺ + SO₄²⁻ + OH⁻

(3)

The NH₃ molecules donated more electrons to the already acidic surface of the WS₂ sensor, resulting in the observed increase in response.

4. Conclusions

WS₂ nanoflowers were successfully synthesized after 45 min via a simple colloidal route. The colloidal nanoflowers displayed a good response and higher selectivity towards NH₃ vapors relative to acetone, ethanol, toluene and chloroform. The gas sensing performance exhibited by the nanoflowers is evidence that the partial oxidation did not adversely compromise it. Humidity interference was established and its significance was observed at very high % RH values. Evidence of incomplete recovery due to the strong interaction between NH₃ molecules and the sensor was observed. The sensor has potential for use as an active material in RT chemiresistive sensors and sensor arrays. Studies are underway to develop strategies to improve the overall gas sensing performance of the WS₂ nanoflowers by doping with metal oxides, metals, carbon materials, perovskites and other TMDCs. The focus is on reducing the recovery time, enhancing the sensitivity and selectivity to NH₃ and slowing down oxidation. The sensor will also be tested for sensitivity to NO₂ and CO.