Abstract
Nanosized tin dioxide with an average particle size of 5.3 nm was synthesized by a sol-gel method and characterized by IR spectroscopy, TEM, X-ray, and electron diffraction. The obtained SnO2 can be used as initial material for creation of gas-sensitive layers of adsorption semiconductor sensors. Addition of palladium into the initial nanomaterial allows to improve response to hydrogen of such sensors in comparison with sensors based on undoped SnO2 and provides fast response and recovery time, a wide measuring range of hydrogen content in air ambient, and good repeatability of the sensor signal. Such promising properties could make useful the sensors based on these nanomaterials for devices intended to determine hydrogen in air.
1. Introduction
Nowadays, development of nanosized oxides is actual for obtaining functional materials with required properties. In particular, such nanomaterials can be useful for creating adsorption semiconductor gas sensors intended to determine the presence of combustible gases, e.g., hydrogen, in air. Different semiconductor oxides, such as SnO2, TiO2, ZnO, ZrO2, and WO3 [1–5], can be used as initial materials for gas-sensitive layers of the sensors. Tin dioxide among them is the most popular due to its chemical stability, band structure, and extreme sensitivity of its conductivity to the surface state where the molecules adsorbed on the SnO2 surface are actively engaged in chemical reactions occurred in the temperature range 20–500°C [6–11]. It is known that a decrease in the particle size of semiconductor material can lead to increase in the ratio of the atoms fraction on the surface of particles to the atoms fraction in their volume that makes impact of the surface processes into the material properties significant [12]. Thus, the sensitivity of the adsorption semiconductor sensors should depend greatly on the morphology of the materials of the gas-sensitive layers: a decrease in the semiconductor particle size of the gas-sensitive material could cause an increase in the sensor response as it was experimentally observed by other authors [13–16].
Increase in the sensor response can also be achieved by introducing catalytically active additives into the semiconductor materials of the gas-sensitive layers [13]. It leads to an increase in the rate of the catalytic reaction of the analyzed gas with oxygen chemisorbed on the sensor surface that in turn increases the sensor response [13]. Palladium can be such an additive for hydrogen sensors due to its high catalytic activity in the hydrogen oxidation reaction [17].
Thus, usage of tin dioxide nanomaterial doped with the palladium additive is promising to provide high sensitivities of the semiconductor sensors to hydrogen.
The aim of this work is to synthesize oxide nanomaterials based on tin dioxide with the addition of palladium for creation of highly sensitive adsorption semiconductor sensors to hydrogen.
2. Experimental
Initial nanosized tin dioxide was synthesized by a sol-gel process through chemical oxidation of tin (II) oxalate by 35% solution of H2O2 [18]. Tin (II) oxalate (3 g) was gradually added under stirring to 10 ml of 35% solution of hydrogen peroxide in water. After two hours, a resulting sol was quickly heated to decompose the excess of the hydrogen peroxide and evaporate the water. As a result, a transparent water-based gel was obtained, and then it was dried at 90°C for 24 hours until the gel turned into a yellowish translucent xerogel [19]. The xerogel was calcinated at different temperatures and temperature holding times for obtaining crystalline nanoscale SnO2 particles with minimal sizes.
In order to optimize the temperature treatment conditions of the obtained xerogel, its thermal decomposition was studied in air with a heating rate of 10°C/min on a DTG-60H derivatograph (Shimadzu, Japan).
Gas-sensitive nanomaterials based on unmodified SnO2 and tin dioxide doped with 0.24 wt.% palladium were obtained through applying a paste formed by mixing the initial nanosized SnO2 with aqueous solution of carboxymethylcellulose (CMC) onto ceramic plates with subsequent drying at 90°C. Impregnation of the dried paste with a solution of palladium (II) chloride in dilute hydrochloric acid was performed with subsequent drying at 90°C in order to obtain palladium-doped gas-sensitive material. The coated ceramic plates were then heated at temperatures up to 620°C in order to provide decomposition of PdCl2 and CMC.
The amount of palladium introduced into the gas-sensitive material was determined by X-ray fluorescence analysis (ElvaX EXS–01, Elvatech, Ukraine).
The morphology of the synthesized nanomaterial was studied by transmission electron microscopy on a SELMI PEM-125K device (Ukraine) with an accelerating voltage of 100 kV.
The phase composition of the materials was studied using a LabX XRD-6000 diffractometer, Shimadzu (Japan) (CuKα radiation). Particle sizes of the nanomaterials were estimated using the Scherrer equation [20]:where D is XRD particle size; k is a constant that depends on crystallite shape and is close to unity (for our calculation k was equal to 0.9); λ is the wavelength of CuKα radiation (λ = 1,5418 Å); β is a true broadening of diffraction peak (β = Δ − b, where Δ is an experimental broadening of the diffraction peak; b is an instrumental broadening); and θ is a Bragg angle. The lattice parameters of the materials were calculated using a program UnitCell.
The specific surface areas of the xerogel temperature treatment products were determined by the argon thermal desorption method using Al2O3 as a standard sample (Ssp = 22 m2/g). Before measurements, the investigated samples were degassed in a helium stream at 300–350°C for two hours.
The infrared spectra of the samples were recorded on a PerkinElmer BX spectrophotometer (USA).
The sensors were made on the basis of planar ceramic plates by the same preparation method as the gas-sensitive nanomaterials. The paste (mixture of the initial nanosized SnO2 with the 3% solution of CMC in water) was deposited between measuring electrodes on the one side of the ceramic sensor plate. The other side of the plate contained a platinum heater for controlling the operational temperature of the sensor. Dimensions of the sensors plates were 1.8 × 1.8 × 0.3 mm. Design of the sensor in more detail is presented in [21].
The sensor parameters were determined in a special electric stand. The electrical circuit of the stand was presented in [22]. A hydrogen-air mixture with 44 ppm H2 was used to measure the sensor responses at different operational temperatures in order to find optimal hydrogen sensing conditions.
Stabilization of the sensor electrical conductivities was achieved by pretreatment of the sensors for three days at the operational temperature of 405°C with a periodic supply of the hydrogen-air mixture (935 ppm H2) to the measuring chambers where the sensors were placed.
The ratio of the sensor electrical conductivity in hydrogen-air mixture (σg) to the electrical conductivity in clean air (σ0) was taken as a measure of the sensor response to this hydrogen content in air ambient (γ): .
Two parameters (response time (τ0.9) and recovery time (τrelax)) were used to estimate the dynamic properties of the sensors. A value τ0.9 was estimated as the time required for conductivity signal of the sensor to attain 90% of its equilibrium value after an injection of the hydrogen-air mixture to the measuring chamber of the sensor. The recovery time (τrelax) was estimated as the time required for the conductivity signal of the sensor to attain 10% of its equilibrium value in the hydrogen-air mixture after applying clean air to the measuring chamber.
3. Results and Discussion
Study by the DTA-DTG method of the thermal decomposition process of the obtained xerogel (Figure 1) showed that a weight loss occurs in several steps. The first step (up to 100°C) corresponds to loss of physically sorbed water (∼10 wt.%) with endoeffect at 43°C on the DTA curve. Further increase in the temperature up to 600°C leads to a weight loss (∼8 wt.%) corresponding to the removal of the strongly bounded water. It should be noted that in this temperature range, partial crystallization of the tin dioxide can also occur and a presence of a broadened peak of exoeffect on the DTA curve (Figure 1) could indicate this process.

Several absorption bands can be observed in the IR spectra of the xerogel and the materials obtained by its temperature treatment up to 400°C with isothermal exposure at this temperature during 1 hour 20 minutes and 2 hours 20 minutes: the absorption band at 1632 cm−1 that refers to deformation vibrations of the adsorbed water, a wide absorption band in the region at 3000–3600 cm−1 corresponding to the total contribution of stretching vibrations of surface hydroxyl groups and the water adsorbed on the surface, and two intense absorption bands in the region at 550–600 cm−1 and 650–680 cm−1 corresponding to vibrations of the bond between tin atoms and oxygen for the terminal and bridge fragments, respectively [23, 24]. In the last case, the absorption band at 660 cm−1 is the characteristic for the O-Sn-O fragment [24]. It should be noted that an increase in the duration of the temperature treatment of the xerogel from 1 hour 20 minutes to 2 hours 20 minutes leads to a slight shift (5 cm−1) of the absorption band at 650–680 cm−1 to the larger wave numbers region. In these conditions, the absorption band at 590 cm−1 for the xerogel shifts to 616 cm−1 and to 631 cm−1 after 1 hour 20 minutes and 2 hour 20 minutes at 400°C, respectively, due to the formation of the crystalline structure of the nanomaterial [24, 25].
Additional formation of the SnO2 structure accompanied with rearrangement of tin and oxygen atoms after the temperature treatment of the xerogel is also evidenced by a change in the ratio between the intensity of the absorption bands corresponded to the terminal Sn-O groups and to the bridge fragments. Higher band intensity of the terminal fragments in comparison with the bridge ones in the material heated at 400°C for 2 hours 20 minutes may be attributed to the formation of the SnO2 crystal structure as it was observed for the materials obtained in [25].
According to the XRD analysis (Figure 2), the diffraction patterns of the xerogel and the materials obtained after xerogel thermal treatment at the temperature range of 400–600°C are broadened that indicates the nanoscale nature of the samples regardless of their isothermal processing time (1 hour 20 min and 2 hours 20 min) at different temperatures (Table 1). It was established that all obtained materials have the cassiterite structure (ICDD PDF-2 Version 2.0602 (2006), card no. 00-041-1445). Calculated lattice parameters are listed in Table 1. Estimation of the materials XRD particle sizes using the Scherrer equation has shown that the SnO2 particle sizes increase from 4.8 to 12.1 nm (Table 1) with an increase in the isothermal processing temperature of the xerogel. Such increase should lead to a decrease in the specific surface area (Ssp) of the material that was experimentally observed: the value of Ssp decreased from 110 to 37 m2/g when isothermal processing temperature increased from 400 to 600°C (Table 1).

It should be noted that changes in isothermal processing time do not make any significant influence on the specific surface area: the value of Ssp decreases from 85 to 81 m2/g at 450°C and from 60 to 58 m2/g at 500°C when the processing time increases from 1 hour 20 minutes to 2 hour 20 minutes, respectively. Thus, it can be assumed that exposure temperature makes a greater impact on the particles sizes than the processing time. For the samples obtained through the temperature treatment of the xerogel at 400°C during 1 hour 20 minutes, the specific surface area was not measured because the preparation of the sample required high-temperature degasation in the argon flow (the temperature range 300–350°C) that can make changes in the incompletely formed crystal structure of this material. Thus, the obtained value of the specific surface area will not be objective and representative.
The TEM study of the material obtained by the xerogel thermal treatment at 400°C for 2 hours 20 minutes showed that it consisted of individual nanoparticles with sizes from 3 to 9 nm (average size of the nanoparticles is 5–6 nm) (Figure 3(a)). The presence of the nanosized crystalline particles for this material is confirmed by the ring-shaped electron diffraction pattern presented in the inset of Figure 3(a). The specific surface area of the synthesized nanosized tin dioxide is 110 m2/g (Table 1).

(a)

(b)

(c)
Thus, the thermal treatment of the xerogel up to 400°C during 2 hours 20 minutes is sufficient to provide formation of the nanoscale crystalline SnO2 with an average particle size 5–6 nm. This material was chosen as the initial for creating the adsorption semiconductor sensors. To increase their sensitivity to hydrogen, a small amount of palladium (0.24 wt.%) was added into the gas-sensitive layer. Both types of the obtained gas-sensitive materials (undoped and doped with 0.24 wt.% Pd) consist of the nanosized particles observed by TEM (Figures 3(b) and 3(c)). According to the XRD study, only cassiterite phase was present in the gas-sensitive material (Figure 2). The XRD sizes of SnO2 particles are 13.9 and 12.6 nm for the undoped and doped with 0.24 wt.% Pd materials, respectively, that could be explained by a stabilization role of the palladium additives [26].
As can be seen in Figure 4, the palladium additive increases the conductivities of the sensors that can be attributed to the increase in the number of defects in the tin dioxide crystal structure that were formed during the high-temperature sensor treatment process due to the introduction of palladium [27]. It should be noted that both types of the sensors (based on unmodified SnO2 and Pd/SnO2) demonstrate extreme dependences of the gas-sensitive layer conductivities on the operational temperature of the sensor. Such character of the conductivity changes can be caused by the influence of oxygen adsorption-desorption processes occurring on the sensor surface. An increase in the operational temperature of the sensors from 225 to 290°C (for the sensors based on Pd/SnO2) and up to 325–345°C (for the sensors based on unmodified SnO2) can lead to an increase in the amount of the chemisorbed oxygen that, in turn, leads to a decrease in the sensors conductivities [13, 14]. A following increase in the operational temperature of the sensor can facilitate the desorption of the chemisorbed oxygen and, thus, leads to an increase in the sensor conductivity (Figure 4).

As can be seen from Figure 4, the sensors based on the unmodified SnO2 have the highest response value to 44 ppm H2 (ca. 9.4–9.5) at the same temperature range where the minimal conductivities of the sensors were observed (325–345°C, Figure 4). Such behavior of the sensor parameter changes allowed to conclude that the formation of the sensor responses is greatly influenced by the amount of the chemisorbed oxygen available for the hydrogen oxidation. For the sensors based on the Pd/SnO2 material, the highest sensor responses to 44 ppm H2 are observed at 260°C, and it is equal to ∼26 which is much higher than the responses to H2 of the sensors without any additives (Figure 4). The discrepancy between the maximum of the γ dependence on temperature (260°C) and the minimum of the σ0 dependence on temperature (290°C) can be explained by the significant catalytic activity of palladium in the hydrogen oxidation reaction [17]. Above 260°C, the hydrogen oxidation rate on the palladium can be high enough to provide the formation of reaction products in a large amount. The products prevent or complicate the consumption of the oxygen chemisorbed on the tin dioxide surface by the hydrogen oxidation reaction that occurred on palladium, and thus, such “blocking” of the sensor surface could reduce the sensor responses to hydrogen [13, 14]. The assumption of SnO2 surface blocking by the hydrogen oxidation products and the oxygen desorption correlates well with a further decrease in the sensitivities of the Pd/SnO2-based sensors with an increase in their operational temperature. It can be seen (Figure 4) that the sensor responses in the operational temperature range of 345–400°C become even less than the responses of the sensors based on the unmodified SnO2 probably due to almost complete isolation of the tin dioxide surface from hydrogen by increasing the amount of the reaction products formed on the palladium particles.
To assess the potential usage of the Pd/SnO2-based sensors for hydrogen detection in air, the dependences of conductivity changes on time with the change of analyzed gas mixtures surrounding the sensors were studied at the different operational temperatures of the sensors (Figure 5). It was found that in the operational temperature range of 260–400°C, the sensors possess good dynamic properties: a steady-state conductivity level in the presence of 44 ppm H2 and in clean air is attained quickly. In particular, values τ0,9 are in the range of 8–29 s and τrelax falls in the range 12–28 s depending on the operational temperature of the sensor (Table 2).

As can be seen in Figure 5, the time required to achieve the steady-state conductivity level in the presence of 44 ppm H2 at the sensor operational temperature 225°C is ca. 7 minutes that is significantly bigger in comparison with the higher operational temperatures. Further increase in the operational temperature will lead to an improvement of the sensor dynamic properties, since the rates of the chemical reactions (oxygen chemisorption and catalytic reaction of hydrogen oxidation), the diffusion of the reagents into the gas-sensitive layer, and the rates of the reaction products elimination from the sensor surface increase significantly at the higher sensor temperatures. This statement is in good correspondence with observed experimental data (Figure 5 and Table 2). It is clear from Table 2 that the values of τ0,9 and τrelax decrease with the increase in the operational temperature of the sensor. Obtained values are better than those previously reported in the literature (where τ0,9 = 2 min and τrelax = 15 min at 300°C) [28]. Thus, the comparison of the sensor responses data with their dynamic properties allowed to determine the optimal operational temperature for the 0.24 wt.% Pd/SnO2-based sensors (about 260°C).
To determine a range of the hydrogen detection in air for the Pd/SnO2-based sensor, the dependences of changes in the sensor conductivities on the hydrogen content in air ambient were studied in the concentration interval 3–935 ppm H2 at the optimal sensor operational temperature (260°C) (Figure 6). It can be seen that the studied sensor can measure hydrogen in the wide range of its concentration: the response values of the sensors were found to be 5.7 and 193 for 3 and 935 ppm H2 in air ambient, respectively. Besides, the conductivity level of the sensor in the presence of 3 ppm H2 demonstrates good repeatability (inset in Figure 6). There are no evidences of a memory effect of the sensor and distortion of the conductivity value in 3 ppm H2 by influence of 935 ppm H2 applied to the sensor previously. Thus, the sensors based on Pd/SnO2 can be used for the reliable detection of hydrogen in air ambient.

The dependence of the conductivity of the sensor based on Pd/SnO2 on the hydrogen content in air is shown in Figure 7. As it can be seen, in linear scale, σg increases with increasing the H2 content in air over all the measured hydrogen concentration ranges. In logarithmic scale, the dependence is linear and its slope is equal to 0.62:

The obtained value for the slope is in good correspondence with the reported data for the typical slope of the conductivity dependence on the concentration of the reducing gases for the adsorption semiconductor sensors in the logarithmic scale [29]. The ability to linearize well the dependence in the logarithmic scale makes a periodic calibration of the sensor during its operation easier because it can be performed using at least two points of H2 concentrations only.
4. Conclusions
Nanosized tin dioxide material with an average particle size of 5.3 nm allowed to create Pd-doped gas-sensitive material for highly sensitive hydrogen sensors. The optimal sensors based on Pd/SnO2 nanomaterial possess a high response to microconcentration of H2 (44 ppm in air ambient), a wide range of hydrogen detection in air (3–935 ppm H2), good stability, lack of the sensor memory effect after exposure to a high hydrogen concentration, and good dynamic properties. These properties make the studied sensors promising for further application in creation of gas analytical devices intended to detect H2 in air ambient.
Data Availability
The data used to support the findings of this study are included within the article.
Conflicts of Interest
The authors declare that they have no conflicts of interest.