Ag-Loaded WO₃ Ceramic Nanomaterials: Characterization and Moisture Sensing Studies

Abstract

Pellet samples of 0–4% weight Ag-loaded WO₃ prepared through soft chemical route were sintered at 700°C and exposed to humidity. Resistance of the pellets decreased with increase in relative humidity. Sensitivity increased with increase in the % loading of Ag. Four percent Ag-loaded WO₃ showed maximum sensitivity of 2.38 MΩ/% RH in 20–90% relative humidity range. This sensing element manifests highest crystallinity as well as maximum void concentration. Hysteresis and repeatability for this sensing element after 6 months are within ±2%. A polynomial fit of the humidification data revealed a strong correlation between resistance and relative humidity.

Introduction

Research has been going on to find suitable materials that show good sensitivity over large range of relative humidity (RH), low hysteresis, and properties that are stable to thermal cycling and exposure to various chemicals likely to be present in the environment. Ceramic humidity sensors, particularly those based on porous and sintered metal oxides have been attracting attention due to their chemical and physical stability, and mechanical strength. These materials possess a unique structure consisting of grains, grain boundary surfaces, and pores, which make them suitable for this kind of sensor. Humidity sensors of thin film or pellet type having nanosize grains and nanoporous structures have drawn much interest because of the high surface exposure for adsorption of water molecules.[1-8] Jeseentharani et al. have investigated humidity sensing properties of the composites prepared by mixing 1:1 mole ratio of CuO–ZnO, CuO–NiO, and NiO–ZnO compound. The samples were sintered at 800°C for 5 h and were then subjected to resistance measurements as a function of relative humidity (RH) in the range of 5–98% RH. Among the three composites, CuO–NiO compound possessed the highest humidity sensitivity. The response and recovery times of the CuO–NiO composites were 80 s and 650 s respectively.[9] Yawale et al. have doped semiconducting materials SnO₂ and ZnO with TiO₂ and Al₂O₃ and screen printed them in the form of films. The DC electrical resistance of the films has been measured in the presence of humidity. They have found SnO₂–5 mol% Al₂O₃ and ZnO–5 mol% Al₂O₃ to be good sensing materials for humidity.[10] Yadav et al. have reported moisture sensing properties of niobium penta-oxide, neodymium oxide, and lanthanum oxide in the form of pellets for annealing temperatures 200°C and 400°C. Variation in DC electrical resistance with the relative humidity was recorded for each annealing temperature. They have reported lanthanum oxide to have the highest sensitivity and results to be reproducible within ±5% error.[11]

Just as the structure of a material plays a decisive role in its performance; the size, shape, and nanostructure of WO₃ units determine the performance of these devices to a large extent.[12] Wu et al. have prepared homogeneous nano-WO₃ grains using gas–liquid reaction.[13] Xu et al. have synthesized micro level WO₃ powder via hydrothermal method.[14] Cheng et al. have fabricated WO₃ nanotubes using templates.[15] Patil et al. have prepared the organic-inorganic composites containing Poly (2, 5-Dimethoxyaniline) as organic material and WO₃ as inorganic material and examined the possibility of using them as candidates for humidity sensing application.[16] Researchers have paid attention to WO₃ as gas-sensing material. Stankova et al. have studied influence of the annealing and operating temperatures on the gas-sensing properties of RF sputtered WO₃ thin film sensors.[17] Bittencourt et al. have studied the characterization and gas sensing properties of WO₃:Ag films.[18] Humidity sensing studies of WO₃–ZnO–TiO₂ composites, ZnO doped with Cu₂O and optical humidity sensor based on titania films have already been reported.[19-22]

This article reports characterization and humidity sensing studies of pure WO₃ and Ag-doped WO₃. Using soft chemical route WO₃ has been obtained. The aim of this article is to analyze the changes induced on the WO₃ sensing element by Ag loading. The nature of the changes analyzed is twofold. On the one hand, morphological and crystalline changes are studied and, on the other hand, changes in the response to humidity are investigated. Scanning electron microscope (SEM) has been used to investigate the morphological changes; the X-ray diffraction (XRD) has been used to look into the changes in the crystalline structure. Finally, humidity sensing studies of WO₃ and Ag-loaded WO₃ have been investigated. We have demonstrated the feasibility of the porous Ag-doped WO₃ nanocomposite, prepared using soft chemical route, as humidity sensor. The performance characteristics, such as sensitivity, reproducibility, and response time, have been recorded and correlated to the microstructures. Herein, it has been shown that humidity sensor can be made from the binary system of Ag-loaded WO₃.

Sensing Mechanism

The sensing mechanism of metal oxide-based humidity sensors lies in changes in oxide resistance resulting from physisorption, chemisorptions, and catalytic reactions of gas phase species at the surface of the semiconductor devices. The catalytic properties of WO₃ are determined by the ability of tungsten ions to change their valence state upon reduction or oxidation processes. In stoichiometric WO₃, the tungsten ions have the 6+ valence state with the 5d shell empty (d ⁰ oxides); that is, there are no cation d electrons available to transfer to adsorbates. However, if oxygen vacancies are present at the surface of d ⁰ oxides, the d electron orbitals on adjacent cations can be partially occupied. These reduced surface cations provide the active sites for much of the chemisorptions and catalytic activity that takes place in d ⁰ transition metal oxides. In other words, the oxygen vacancies that are associated with the existence of sub stoichiometric compounds WO_3−x, where x is the oxygen composition parameter (0 < x < 1), determine the sensing properties of the system.[23]

With the aim of achieving improvements in their sensing properties, small amounts of noble metals are added to metal oxide active layers. The metal additives dispersed in the oxide matrix act as activators or sensitizers.[24-28] Therefore, the use of WO₃ as active material for gas sensing is widely related to its use with catalytic metals added to the matrix. The addition of a noble metal results in changes in the electronic states of the active layer and can also modify the microstructure of the base material, alter the mechanism of the grain growth and the (local) sticking coefficient. All these strongly influence sensor performance. It is important to add that under noble metal loading, it is expected that the metal atoms will form clusters at the surface of WO₃.[28] These clusters will be in metallic or oxidized forms depending on the noble metal and the loading process. The mechanism by which a metal atom interacts with the surface of a metal oxide is extremely varied and complex, it has been shown that, in some oxides, the metal atoms localize preferentially at defect sites. If this is the case, Ag atoms migrate to these sites during the annealing process and change the electronic density near the W atom compensating the effect of an oxygen vacancy.[29] The surface of the sensing element is essentially inhomogeneous, and made up of grains and voids. The voids in the structure provide direct conduits for water molecules to flow in from the environment.

Sample Preparation and Experimental Process

The WO₃ nanocrystalline powders have been obtained by a soft chemistry route based on tungstic acid. A quantity of 10 g tungstic acid was dissolved in a 50:50 volume mixture of methanol (100 mL) and water (100 mL). To stabilize the precipitate 25 mL of water was further added gradually. This mixture was heated at 80°C for 24 h under continuous magnetic stirring in air and then dried by further heating at 110°C in air, leading to tungsten oxide hydrate. This material was annealed in a furnace at 300°C for 3 h to obtain nanocrystalline pure WO₃.

To make the samples of pure WO₃ and Ag-loaded WO₃, 10% by weight ethyl cellulose has been added as binder to both pure WO₃ and Ag-loaded WO₃.

Separately, 1, 2, 3, and 4 wt% of Ag metal powder (Loba Chemie, Mumbai, India; 99.9% pure) has been added to pure WO₃ obtained through the above process. The resultant powders have been grinded separately in mortar and pestle for 3 h to get uniform mixtures. These powder mixtures have been given pellet shape in a hydraulic pressure machine (M.B. Instruments, Delhi, India) under uniaxial pressure of 350 MPa at room temperature. These pellets have been sintered at temperatures of 700°C in a muffle furnace (Ambassador, Mumbai, India) to form ceramic samples. The ceramic pellet samples are 12 mm in diameter and 3 mm in thickness. Samples of pure WO₃ (0% Ag-doped WO₃), 1% Ag-loaded WO₃, 2% Ag-loaded WO₃, 3% Ag-loaded WO₃ and 4% Ag-loaded WO₃, have been labeled AW-0, AW-1, AW-2, AW-3, and AW-4 respectively.

After sintering, the samples have been exposed to humidity in a humidity chamber. Standard solution of potassium sulfate has been used as humidifier and potassium hydroxide as de-humidifier. Inside the humidity chamber, a thermometer (±1°C) and standard hygrometer (Huger, Baden-Württemberg Germany, ±1% RH) are placed for the purpose of calibration. Variation in dc resistance has been recorded with change in relative humidity at room temperature (27°C). Relative humidity has been measured using the standard hygrometer. Variation in resistance of the pellets has been recorded using a multimeter (±0.001 MΩ, VC-9808). Copper electrodes have been used to measure the resistance of the pellets. The resistance of the pellet have been measured normal to the cylindrical surface of the pellets. The electrical resistance at different relative humidity levels of the sensing elements in the form of pellets has been determined by two-probe method. The two electrical contacts have been made on the surface of pellet by means of two thin copper sheets. Given the high resistivity of the materials under consideration, the potential inaccuracy due to contact resistance is assumed negligible. The surface contact area of all the sensing elements with the electrodes is 113.11 mm² and the surface area that has been exposed to the humidity in the chamber is also 113.11 mm². These values have been kept constant for all the sensing elements. The stability of the sensing element has been checked by keeping the sensing element at fixed values of % RH in the chamber and the values of resistance recorded as a function of time. The values have been found to be stable within ±3% of the measured values.

Results and Discussion

Humidification Graphs

Figure 1 depicts humidification graphs for the sensing elements AW-0, AW-1, AW-2, AW-3, and AW-4 for the sintering temperature 700°C for the range of relative humidity (RH) from 20% to 90%. These are the graphs between the changing values of resistance of the sensing elements with variation in relative humidity. These measured values shown in the graph have been found to be repeatable within ±3% over repeated cycles. The maximum sensitivity of 2.38 MΩ/%RH has been found for the sensing element AW-4 and the lowest for AW-0. Here, the sensitivity of humidity sensor is defined as the change in resistance (ΔR) of sensing element per unit change in relative humidity (ΔRH%). For calculation of sensitivity, the humidity from 20% to 90% RH has been divided in equal intervals of 5% RH each. Difference in the value of the resistance for each of this interval has been calculated and then divided by 5. The average has been taken for all these calculated values. Formula for calculation of sensitivity of the sensing elements may be written as given below:

• (1)

Figure 1. Variation of resistance of sensing elements with relative humidity for increasing cycle of relative humidity.

There is an increase in the value of sensitivity for the sensing element of Ag loaded WO₃ as compared to the sensitivity of the sensing element AW-0. The average increase in the value of sensitivity for the sensing element of AW-4 over AW-0 (pure WO₃ sample) is 11.34%. We observe a continuous increase in the value of sensitivity with increase in the percentage of Ag in WO₃. Figure 2 shows the graph between the variations in the value of sensitivity with change in the value of doping of Ag in WO₃.

Figure 2. Variation of sensitivity of sensing elements with % loading of Ag in WO₃.

It can be seen from Fig. 1 that there is a continuous decrease in the value of resistance with increase in the value of relative humidity. Since WO₃ is an n-type semiconductor sensitivity to humidity is a result of electronic conduction.[30, 31] As semiconducting dry oxide of WO₃ nanocomposite is brought in contact with humid air, water molecules chemisorb on the available sites of the oxide surface. The adsorption of water molecules on the surface takes place via a dissociative chemisorptions process which may be described in a two-step process as given below:

(i) Water molecules adsorbed on grain surface react with the lattice W as

• (2)

Here, O_o is the lattice oxygen and V_o the vacancy created at the oxygen site according to the reaction:

• (3)

(ii) Doubly ionized oxygen, displaced from the lattice, reacts with the H⁺ coming from the dissociation of water molecules to form a hydroxyl group as given below:

• (4)

Tungsten Oxide has electron vacancies. Hence, because of this reaction, the electrons are accumulated at the WO₃ surface and consequently, the resistance of the sensing element decreases with increase in relative humidity.

All the graphs in Fig. 1 show that from 20% to 40% RH, there is a rapid decrease in the value of resistance although in 40–90% RH range the fall in resistance is slow. Thus, in 20–40% RH range sensitivity of the samples is high whereas in 40–90% RH range sensitivity is low. This phenomenon may be understood in the manner that the increase in the conductivity of the samples with relative humidity in the lower range (<40% RH) is due to the adsorption of the water molecules on the pellet surface with capillary nanopores. Higher porosity increases surface to volume ratio of the materials and enhances diffusion rate of water into or out-off the porous structure; and thus, helps in getting good sensitivity. At high relative humidity (>40% RH), vapor condenses in the capillary-like pores, forming a liquid layer.

Humidification and Desiccation Graphs (Hysteresis Effect)

Metal oxides and binary systems of metal oxides show deviation in their behavior in the decreasing cycle of % RH from those in increasing cycle of % RH. Minimization of this hysteresis behavior is an important condition for sensor application. To determine the hysteresis effect in the sensing elements, the humidity in the chamber has been increased from 20% to 90% RH and then cycled down to 20% RH and the values of resistance of the sensing elements recorded with change in % RH. It is observed that all sensing elements have acceptable hysteresis values. The change in the value of sensitivity in decreasing cycles over increasing cycles are ±1.42%, ±0.46%, ±0.46%, ±1.78%, and ± 1.74% for sensing elements AW-0, AW-1, AW-2, AW-3, and AW-4 respectively. Thus, the maximum change in the values of sensitivity for the decreasing cycles over increasing cycles is below ±2%. The phenomenon of hysteresis may be attributed to the initial chemisorptions on the surface of the sensing elements. This chemisorbed layer, once formed is not further affected by exposure to or removal of humidity, it can be thermally desorbed only. Hence, in the decreasing cycle of % RH, the initially adsorbed water is not removed fully leading to hysteresis. Figure 3 shows desiccation graphs for sensing elements AW-0, AW-1, AW-2, AW-3, and AW-4 sintered at temperature of 700°C.

Figure 3. Variation of resistance of sensing elements with relative humidity for decreasing cycle of relative humidity.

A regression analysis of the data was carried out on humidification graphs. Least square fit with respect to polynomials of third degree and fifth degree were established. The coefficients of the various powers of x (RH) with respect to y (Resistance) is tabulated in Table 1 (coefficients of a third degree polynomial) and Table 2 (coefficients of a fifth degree polynomial). It can be observed from the tables that a₁, the coefficients of the first power of x in the polynomial is the dominant one, indicating a strong linearity between x and y (RH and resistance). Considering the present state of instrumentation, it is not essential to go for a linear response in sensors; a single valued function should suffice. The nanocomposite qualifies to be a good ceramic humidity sensor.

Table 1. Coefficients of Various Powers of x for Polynomial of Degree 3

Samples	Polynomial coefficients
	a0	a1	a2	a3
x (relative humidity); y(resistance of the samples); data fitted to a polynomial of degree 3. y = a0 + a1x + a2x^2 + a3x^3
AW-0	318	−10.63	0.142	−0.000
AW-1	406.4	−17.98	0.287	−0.001
AW-2	463.8	−22.13	0.357	−0.001
AW-3	378.5	−13.81	0.190	−0.000
AW-4	389.1	−14.02	0.199	−0.001

Table 2. Coefficients of Various Powers of x for for Polynomial of Degree 5

Samples	Polynomial coefficients
	a0	a1	a2	a3	a4	a5
x (relative humidity); y(resistance of the samples); data fitted to a polynomial of degree 5. y = a0 + a1x + a2x^2 + a3x^3 + a4x^4 + a5x^5.
AW-0	430.5	−23.21	0.665	−0.010	9 × 10^−5	−3 × 10^−6
AW-1	1012	−84.67	2.996	−0.052	0.000	−2 × 10^−6
AW-2	1083	−122.2	4.348	−0.075	0.000	−2 × 10^−6
AW-3	740.6	−52.95	1.737	−0.029	0.000	−8 × 10^−6
AW-4	949.1	−75.99	2.718	−0.048	0.000	−1 × 10^−6

A polynomial of third degree fitted to the curve of sensitivity versus dopant % of Ag in WO₃.

• (5)

Here, x = % of Ag in WO₃ and y = sensitivity in MΩ/%RH.

It was found that in this case the higher powers of the dopant percent are prominent and it does affect the sensitivity in a manner more than linear.

Aging Effect

Aging is a significant problem in sensing devices based on metal oxides.[32] After studying humidity sensing properties, sensing elements have been kept in laboratory environment and their humidity sensing characteristics regularly monitored. To see the effect of aging, the sensing properties of these elements have been examined again in the humidity control chamber after 6 months and variation of resistance with % RH recorded. The variation of resistance of the sensing elements AW-0, AW-1, AW-2, AW-3, and AW-4 with change in % RH after 6 months have been shown in Fig. 4. After 6 months, the values of aging are found to be ±0.48%, ±1.40%, ±0.46%, ±0.45%, and ± 1.72% for sensing elements AW-0, AW-1, AW-2, AW-3, and AW-4 respectively. Here, % aging has been defined as the % deviation in the value of sensitivity over the period 6 months from the initial measured values. This may be termed repeatability over a period of time as well. Thus, the maximum aging over the period of 6 months is within ±2%. We observe that the sensing elements both of pure WO₃ (sample AW-0) and Ag-doped WO₃ (samples AW-1, AW-2, AW-3, and AW-4) prepared through soft chemical route show significantly less aging effect in their performance.

Figure 4. Variation of resistance of sensing elements with relative humidity for increasing cycle of relative humidity after 6 months.

Aging mechanisms in humidity sensors may be due to prolonged exposure of surface to high humidity, adsorption of contaminants preferentially on the cation sites, loss of surface cations due to vaporization, solubility and diffusion, annealing to a less reactive structure, or migration of cations away from the surface due to thermal diffusion. In general, the more sensitive a material is to humidity, the more it tends to be susceptible to aging.

X-Ray Diffraction Analysis

The X-Ray diffraction has been studied using the PRO-Analytical XRD system (X'pert, Almelo, The Netherlands). The wavelength of the source CuKα used is 1.54060 Å. The XRD patterns show extent of crystallization of the sensing elements in the form of powder. The average crystalline size of the samples has been calculated using Debye Scherer's formula

• (6)

Here, D is the crystallite size, K is a fixed number of 0.9, λ is the X-ray wavelength, θ is the Bragg angle, and B is the full width at half maximum of the peak. Figures 5 and 6 show X-ray patterns for the sensing elements of pure WO₃ (AW-0) and Ag-doped WO₃ (AW-1). As measured by the Scherer's formula, the crystallite size for the sensing element AW-0 is in the range of 12–72 nm. The range of the crystallite size for the sensing element AW-1 is in the range of 19–73 nm. The XRD pattern shows formation of Ag:WO₃ bronze, viz. hexagonal tungstite, and orthorhombic silver tungsten oxide (Ag₈W₄O₁₆, Ag₂W₂O₇), anorthic silver ditungstate (Ag₂WO₄) due to intercalation of silver atoms into the WO₃ tunnel. Formation of Ag:WO₃ bronze suggests that the Ag atoms cause a charge distribution in the WO₃ unit cell in such a way that the W–O bond in the bronze formed is stronger (shorter bonds) than that for pure WO₃. In this case, the WO₃ grains are tightly connected to the Ag clusters.[29] In addition to these peaks, the pattern also manifests presence of large number of peaks of hexagonal tungstite.

Figure 5. XRD for sensing element AW-0 (pure WO₃).

Figure 6. XRD for sensing element AW-1 (1% Ag doped in WO₃).

Scanning Electron Microscopic (SEM) Study

The study of surface morphology of sensing elements has been carried out using scanning electron microscope (LEO-430, Cambridge, U.K.). Figures 7-11 show SEM micrographs of sensing elements of pure WO₃ (AW-0) and Ag-doped WO₃ (AW-1, AW-2, AW-3, and AW-4) prepared through soft chemical route, sintered at temperature 700°C. The SEM micrograph of sensing element of pure WO₃ manifests beautiful mountain-like pattern whereas the SEM micrographs of sensing element of Ag-doped WO₃ manifest great distribution in grain size and increasing crystallinity. Scanning micrographs show flakes of pure WO₃ and Ag-doped WO₃ scattered throughout, forming a network of pores that are expected to provide sites for humidity adsorption. The SEM micrographs show that the sensing elements manifest porous structure having granulation and tendency to agglomerate. Tungsten Oxide mixed with Ag powder has a grain-like morphology. Upon treatment of heat, clusters of Ag are formed at the surface of the WO₃ grains. Due to diffusion of Ag in WO₃, Ag:WO₃ bronze forms at the interface between the grains and the clusters. Given the tendency of the noble metals to form clusters at the surface of the metal oxide grains, the actual increase observed in grain size distribution with increase in the % of Ag in WO₃ can be associated with increase in the size of the Ag clusters and subsequent coalescence.

Figure 7. SEM micrograph of sensing element AW-0 (pure WO₃).

Figure 8. SEM micrograph of sensing element AW-1 (1% Ag doped in WO₃).

Figure 9. SEM micrograph of sensing element AW-2 (2% Ag doped in WO₃).

Figure 10. SEM micrograph of sensing element AW-3 (3% Ag doped in WO₃).

Figure 11. SEM micrograph of sensing element AW-4 (4% Ag doped in WO₃).

As the doping percent of Ag in WO₃ is increased, we notice from the SEM micrographs that there is an evolution in the distribution of the grain size, and that the crystallinity is increasing with increase in the % of Ag in WO₃. It can be seen from the SEM micrographs that the texture and grain size distribution are highly influenced by addition of Ag in WO₃. One may notice that the distribution of grain size in AW-3 and AW-4 is much more prominent, and that the range is wide. As observed from the two micrographs of AW-3 and AW-4, it can be easily seen that due to addition of Ag in WO₃, more voids are created due to drift of Ag in the oxygen vacancies of WO₃ during the annealing process. Comparing these two micrographs with the micrograph of AW-0, viz. pure WO₃, we find, in AW-0, a continuity and uniform distribution of grains, devoid of the formation of voids; absence of inter connected voids or capillaries which are very important conditions for the adsorption of water molecules in the sensing elements as mentioned in the section 'Sensing Mechanism'. Thus, AW-0 sensing element will have a very low adsorption capacity for moisture and hence the low sensitivity.

Sensing element of AW-4 has the highest crystallinity as well as the maximum void concentration. This synergy turns out to be the reason for the affinity toward moisture and hence a higher conductivity. It may be argued that the higher sensitivity may be due to a multiplicative effect of two or more than two parameters of the resultant composite. As explained in the section 'Sensing Mechanism', the mechanism by which a metal atom interacts with the surface of a metal oxide is extremely varied and complex, it is very difficult to exactly pinpoint the real parameter affecting the sensitivity of the sensing elements. The lesser Ag percentage samples may have more grains per unit area of the surface, yet the range of grain size is so narrow that the surface of the composite appears smooth and therefore does not provide kinks as adsorption sites.

Response and Recovery Times

The time taken to accomplish 90% of the initial total resistance variation is defined as response/recovery time during the humidification and desiccation processes. The response times for the sensing elements AW-0, AW-1, AW-2, AW-3, and AW-4 are 95 s, 75 s, 80 s, 74 s, and 72 s respectively. The recovery times for the sensing elements AW-0, AW-1, AW-2, AW-3, and AW-4 are 480 s, 290 s, 300 s, 280 s, and 276 s respectively. Thus, there is a marked improvement in the value of response and recovery times when Ag is doped in WO₃. Since desorption is an endothermic process, it takes longer time to desorb the water vapor; therefore, the recovery time is always greater than the response time.[33-35]

Conclusions

The sensitivity is found to increase with increase in the % loading of Ag in WO₃. Sensing element AW-4 (4% Ag-doped WO₃) sintered at 700°C shows maximum sensitivity of 2.38 MΩ/%RH in the 20–90% relative humidity range. For this sensing element, the average hysteresis is within ±2% and the repeatability over different humidification cycles is also within ±2% after 6 months. The response and recovery times for AW-4 are the lowest of all the sensing elements at 72 s and 276 s respectively. The XRD pattern of this sensing element shows peaks of Ag:WO₃ bronze. There is an increase in the grain size distribution and crystallinity of sensing elements with increase in the % loading of Ag in WO₃. A regression analysis of the data on humidification and desiccation graphs gives a least square fit with respect to a polynomial of third degree and fifth degree, indicating a strong correlation between x and y (RH and resistance). A polynomial of third degree fitted to the curve of sensitivity versus dopant %.

Acknowledgment

The authors thank the Geological Survey of India, Lucknow, for providing XRD and Birbal Sahni Institute of Paleobotany, Lucknow, for providing SEM facilities.