Nanocrystalline Oxides Ni_xCo_3−xO₄: Sub-ppm H₂S Sensing and Humidity Effect

Abstract

In this work, p-type oxide semiconductors, Co₃O₄ and complex oxides Ni_xCo_3−xO₄ (x = 0.04, 0.07, 0.1), were studied as materials for sub-ppm H₂S sensing in the temperature range of 90–300 °C in dry and humid air. Nanocrystalline Co₃O₄ and Ni_xCo_3−xO₄ (x = 0.04, 0.07, 0.1) were prepared by coprecipitation of cobalt and nickel oxalates from nitrate solutions and further annealing at 300 °C. The surface reactivity of the obtained materials toward H₂S both in dry and humid atmosphere (relative humidity at 25 °C R.H. = 60%) was investigated using diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS). Sensor measurements showed a decrease in sensor signal toward 1 ppm H₂S with an increase in Ni content because of a decrease in chemisorbed surface oxygen species. On the other hand, sensor signal increases for all samples with increasing the relative humidity that depends on reactivity of the surface hydroxyl groups, which stimulate the decomposition of surface sulfites and provide better surface regeneration at higher temperature. This assumption was additionally confirmed by the faster saturation of the conductivity curve and a decrease in the sensor response time in humid air.

1. Introduction

Hydrogen sulfide is a highly toxic, reactive and flammable gas that is found in various technological processes. Therefore, it is extremely important to monitor its concentration in the air at a level below 100 ppm of H₂S, which is considered to be immediately dangerous to life and health (IDLH). Semiconductor gas sensors based on n-type metal oxides are one of the most commonly used commercial hydrogen sulfide detectors [1]. For this type of devices, the problem of the negative influence of ambient humidity often arises [2,3]. The effect of humidity had been actively studied for the most common oxides used in gas sensors, such as SnO₂ [4,5], WO₃ [6], ZnO [7], In₂O₃ [8]. The blocking effect of adsorbed water on WO₃ can be overcome by aging processes, leading to the appearance of surface hydroxyl groups [9].

Currently, there are different types of gas sensors with great potential for use in “e-nose” systems and “smart sensors” [10,11,12]. However, the attention of researchers is increasingly attracted by less common sensors based on p-type oxides, such as CuO, NiO, Co₃O₄. In particular, an increase in the sensor signal in the presence of moisture was observed for CuO sensors used for H₂S detection in a hydrogen atmosphere [13]. This effect was associated with the formation of hydroxyl groups due to the water dissociation on the oxide surface.

Dissociation of water molecules and the formation of strongly bonded hydroxyl groups coordinated at metal centers were also observed on well-ordered Co₃O₄ (111) film [14]. Similar processes can be expected on the surface of polycrystalline Co₃O₄ (Figure 1). In this case, hydroxyl groups can act as oxidizing agents, which can lead to additional oxidation processes in humid air.

Cobalt oxide Co₃O₄ with a spinel structure has a high catalytic activity in oxidation reactions, which is provided by the high concentration of chemisorbed oxygen on the oxide surface. O₂ molecules are adsorbed on the surface oxygen vacancy sites, forming reactive oxygen species (ROS), such as O₂⁻ and O₂²⁻ (Figure 2). It is assumed that these ROS are involved in the catalytic oxidation of hydrocarbons [15], CO [16,17] and formaldehyde [18] above Co₃O₄.

One of the main limitations of Co₃O₄ as a material for gas sensors is its high electrical resistance. The introduction of nickel into the spinel structure with the formation Ni_xCo_3−xO₄ (x ≤ 1) leads to a sharp drop in resistance of the material, which is realized due to an increase in the mobility of charge carriers at x ≤ 0.25 [19]. At the same time, a decrease in the concentration of chemisorbed oxygen is observed, correlating with a decrease in the proportion of Co³⁺ in octahedral positions. This leads to a drop in the CO sensor signal with an increase in the proportion of nickel.

The aim of the present study is to investigate the sensor properties of cobalt oxide Co₃O₄ and complex oxides Ni_xCo_3−xO₄ (x ≤ 0.1) to hydrogen sulfide in humid air. Special attention is paid to the mechanism of reactions and the role of surface hydroxyl groups in these processes.

2. Materials and Methods

Nanocrystalline powders of Co₃O₄ and Ni_xCo_3−xO₄ (x = 0.04, 0.07, 0.1) were prepared by coprecipitation of cobalt and nickel oxalates from nitrate solutions, and their further annealing at 300 °C. Elemental analysis of the samples was performed by XRF and ICP-MS methods. In order to avoid standard errors of the results, obtained by XRF analysis, the calibration was performed using ICP-MS method on the quadrupole mass spectrometer Agilent 7500C. Isotopes ⁵⁹Co, ⁶⁰Ni and ⁵⁸Ni were used for determination of Co and Ni. A detailed description of this technique can be found in Ref. [20]. The phase composition, microstructure parameters and the charge state of cations were studied by XRD, TEM, XPS, EPR, as well as BET measurements [19]. The main characteristics of the samples are presented in

Table 1. Composition and microstructure parameters of Co₃O₄ and Ni_xCo_3−xO₄ samples.

Sample	$\frac{\left[\mathbf{N}\mathbf{i}\right]}{\left[\mathbf{N}\mathbf{i}\right]+\left[\mathbf{C}\mathbf{o}\right]}, \mathbf{w}\mathbf{t} \mathbf{\%}$	Structural Type/Lattice Constant a (Å)	XRD Particles Size d (nm)	BET Specific Surface Area SBET (m2 g−1)
Co3O4	0	normal spinel 8.0575(14)	11 ± 2	48 ± 3
Ni0.04Co2.96O4	1.24 ± 0.06	normal spinel 8.0681(12)	12 ± 1	44 ± 3
Ni0.07Co2.93O4	2.4 ± 0.1	normal spinel 8.0734(17)	9 ± 1	56 ± 4
Ni0.1Co2.9O4	3.3 ± 0.2	normal spinel 8.0800(30)	10 ± 1	53 ± 3

.

The morphology of cobalt oxide was investigated by transmission electron microscopy (TEM) using a FEI Tecnai G2 30 UT microscope operated at 300 kV.

The gas sensor properties were determined by comparing the electrical resistance of thick films in dry or humid air (0, 20%, 60% R.H.) with the resistance in the presence of H₂S (0.07–1.3 ppm). The films were obtained by the following procedure: the synthesized powders were ground with a binder (a solution of α–terpineol in ethanol), and then the pastes were deposited on functional substrates with Pt contacts on the front side and a Pt-meander on the back side that acts both as a heating element and a temperature probe. After covering each layer, DC voltage of 5 V was applied for 2 min. Sensors were annealed at 300 °C for 24 h. The films obtained by this technique are uniform and continuous throughout the substrate, and their thickness is about 1 μm [21].

An automatic device that allowed controlling the temperature and the gas flows (100 mL/min in all experiments) was used for the measurements. Sensor properties were investigated in the temperature range 90–300 °C. For each temperature stage, at first pure air and then a mixture of the test gas and air was supplied into the chamber for 15 min. The cycle was repeated three times for each temperature stage. Pure air generator (Granat-Engineering Co. Ltd., Moscow, Russia) was used to produce purified carrier gas. In sensor measurements and DRIFTS investigations a relative humidity was set and measured by Humidifier P-2 (Cellkraft AB, Sweden). The sensor signal S was determined according to:

S = (R_g − R_a)/R_a,

(1)

From the averaged values of the resistance in air R_a and in the presence of the test gas R_g.

A Perkin–Elmer Frontier spectrometer provided with a DiffusIR annex and flow-through reactor with a furnace (HC900, Pike Technologies) sealed with the ZnSe window was used for in situ diffuse-reflectance infrared Fourier-transform (DRIFT) spectroscopy study of material interactions with H₂S. The spectra were recorded in the range of 4000–1000 cm⁻¹ with a resolution of 1 cm⁻¹ and averaging of 30 scans under control of the gas phase composition. The powders (50 mg) were placed in alundum crucibles 5 mm in diameter. The samples were heated in a flow of clean air at 300 °C before collecting the spectra. The sequence of recording conditions for Co₃O₄ is shown in Figure 3. A similar experiment was carried out for Ni_0.1Co_2.9O₄ sample. Hydrogen sulfide treatment was carried out at a temperature of 220 °C in a flow of H₂S (20 ppm, 100 mL/min) in dry air and in a humidity of 60% R.H. After 50 min of measurements, the temperature was raised to 300 °C and a flow of pure air was started.

3. Results and Discussion

3.1. Characteristics of Nanocrystalline Co₃O₄ and Ni_xCo_3−xO₄ Samples

The microstructure of Co₃O₄ was investigated by TEM (Figure 4). The sample consists of aggregated particles (20–100 nm) and has a porous structure due to the release of gases (CO, CO₂) during the decomposition of the oxalate precursor.

3.2. Sensor Properties for H₂S Detection

Initially the Co₃O₄ and Ni_xCo_3−xO₄ (x = 0.04, 0.07, 0.1) sensors were exposed to H₂S in dry air. In order to find the optimum conditions, the operating temperature was varied from 300 to 90 °C. Figure 5a shows a dynamic response of Co₃O₄ and Ni_xCo_3−xO₄ based gas sensors to the periodical change of gas phase composition from 1 ppm H₂S to dry air at 210 °C. There are 3 periodical changes in gas phase composition for every operating temperature, the flow of H₂S and dry air was supplied to the test chamber for strictly 15 min. All samples showed a typical response of p-type oxides towards a reducing gas such as hydrogen sulfide: the resistance growth in the presence of H₂S and returns back in dry air. The same type of response to H₂S was observed in the humid conditions at 300–90 °C (Figure 5b). All samples showed stable and well reproducible resistance changes and, accordingly, sensor signal.

The temperature dependencies of the sensor signal S (Figure 6a) with a maximum value in the region of 210–240 °C are very similar for all the samples. Therefore, this temperature range is the most favorable for H₂S detection. A maximum value of S = 1.1 was observed for Co₃O₄ at 210 °C.

The effect of hydrogen sulfide concentration was also studied at this temperature (210 °C). The lowest-observed-adverse-effect level of H₂S is 10 ppm, when the irritation of eyes, nose, and throat is caused. In order to prevent serious eye damage, a relatively high protection factor of 100 is used, which leads to a target value of 0.1 ppm with an averaging time of 24 h [22]. On the other hand, the human olfactory organ is able to sense the presence of hydrogen sulfide at a rather low concentration limit of about 0.008–0.13 ppm. Therefore, in this work, the range of target concentrations for Co₃O₄ and Ni_xCo_3−xO₄ samples were selected between 0.07 and 1.3 ppm, which indicates a high potential for use as an “electronic nose” system.

The dependences of the sensor signal S on the H₂S concentration were linearized in double logarithmic coordinates corresponding to the power law S~C(H₂S)ⁿ (Figure 6b). From these calibration curves the values of the lowest detectable concentration (LDC) for H₂S were determined (

Table 2. Lowest detectable concentration (LDC) of H₂S for sensors based on Co₃O₄ and Ni_xCo_3−xO₄ materials measured at 210 °C.

Sample	LDC, ppb	Sensitivity (Signal per Concentration), ppm−1
Co3O4	120	0.74
Ni0.04Co2.96O4	100	0.60
Ni0.07Co2.93O4	150	0.60
Ni0.1Co2.9O4	120	0.51

). The value $\frac{3\sigma }{R_{av}}$, where R_av is the average resistance value and σ is the standard deviation of the resistance in the air, was used as the minimum measurable response. A decrease in the sensitivity of Ni_xCo_3−xO₄ samples with increasing Ni content is accompanied by a decrease in LDC from 120 to 100 ppb if a minimum amount of nickel is introduced.

Further tests were carried out in humid air. As the R.H. value increases, the overall temperature dependence of the sensor signal is retained, but its value has increased. This effect was especially noticeable at temperatures above 210 °C (Figure 7a). The rise in the signal value was sharper for Co₃O₄ than for Ni_xCo_3−xO₄ samples (Figure 7b).

Figure 8a,b shows a comparison of the sensor signal value divided by gas concentration for H₂S, NH₃, and CO at 210 °C in dry air (0% R.H.) and humid atmosphere (30% R.H.), respectively. All samples show high sensitivity to H₂S, despite the fact that its concentration is 20 times less than the concentration of other gases. Further, it can be noticed that the sensitivity of H₂S increases in humid air, unlike ammonia and carbon monoxide. Most likely, the decisive contribution to the formation of an electric signal towards NH₃ and CO in dry and humid air is made only by the process of their oxidation with the participation of various chemisorbed oxygen species on the surface of complex oxides. During interaction with hydrogen sulfide in humid air, surface-adsorbed hydroxyl groups also make an additional contribution to the oxidation of H₂S molecules.

3.3. DRIFT Spectroscopy Investigation in a Hydrogen Sulfide Atmosphere

3.3.1. Interaction of Co₃O₄ with H₂S in Dry Air

During adsorption of the hydrogen sulfide on the Co₃O₄ surface in dry air at the temperature of the maximum sensor signal, there was no absorption bands in the region of 2560–2580 cm⁻¹ related to characteristic ν(SH) bands of adsorbed hydrogen sulfide molecules on the DRIFT spectra (Figure 9a) [23]. This fact indicates rapid oxidation of H₂S on the cobalt oxide surface, which can occur with the participation of chemisorbed oxygen or hydroxyl groups:

2H₂S_(g) + 3O₂²⁻ → 2SO_2(g) + 2H₂O + 6e⁻,

(2)

H₂S_(g) + 6OH⁻ → SO_2(g) + 4H₂O + 6e⁻.

(3)

The products of H₂S oxidation appeared in the form of a complex absorption band at 1380–1000 cm⁻¹, which can correspond to S–O vibrations in sulfates, sulfites or adsorbed SO₂ [24,25,26,27,28]. Oxidation products, as well as adsorbed water molecules (1616 cm⁻¹) and hydroxyl groups (broad band at 3400 cm⁻¹) accumulated on the oxide surface during the experiment. The decreasing signal at 1530 cm⁻¹ corresponds to surface nitrates (from the precursor [26]) consumed in the oxidation of hydrogen sulfide.

A more detailed analysis of changes in spectra over time, as well as varying temperature and humidity, allowed the determination of the composition of the oxidation products. Firstly, the bands at 1330, 1290, 1154 cm⁻¹ and the vibrational mode of hydroxyl groups were observed. Then, there was a sharp increase in the intensity of the bands at 1290 and 1154 cm⁻¹ and the appearance of the water signal and bands at 1124 and 1030 cm⁻¹. With an increase in temperature to 300 °C in a flow of dry air (Figure 9b), a new signal appeared at 1340 cm⁻¹, and the intensity of the remaining bands decreased. However, complete desorption of the products was not observed. In the presence of moisture (Figure 9c), the band at 1340 cm⁻¹ disappeared, but the intensities of signals at 1290, 1154 cm⁻¹ increased.

The low-frequency signal at 1030 cm⁻¹ can be attributed to sulfite groups, which are formed through the SO₂ adsorption on the basic O₂⁻ centers (Figure 10a) on the oxides surface [23,28]. Under humid conditions, the adsorption of SO₂ also occurs on hydroxyl groups, resulting in the formation of HSO₃⁻ groups (Figure 10b). Thus, the bands at 1290 and 1154 cm⁻¹ should be attributed to hydrosulfite groups, which correlates with the frequencies reported for HSO₃⁻ on zeolites [23,29]. The absorption at 1330 cm⁻¹, which is close to the asymmetric vibrations (ν_as) O=S=O, is probably associated with weakly chemisorbed SO₂ [27].

Temperature resistant bands at 1340 and 1124 cm⁻¹ probably correspond to surface sulfate species [24,25], which were formed upon further oxidation of SO₂ or sulfites:

SO₃²⁻ + O₂²⁻ → SO₄²⁻ + H₂O + 2e⁻.

(4)

3.3.2. Interaction of Co₃O₄ with H₂S in Humid Air

DRIFT spectra of Co₃O₄ in a flow of H₂S in humid air (Figure 11a) are characterized by intense absorption bands of water (1621 cm⁻¹) and broad bands of hydroxyl groups (3430 cm⁻¹).

In the flow of hydrogen sulfide, there was a rapid accumulation of hydrosulfite groups (1270 and 1154 cm⁻¹), as well as sulfites (1050 and 1030 cm⁻¹). The shoulder at 1306 cm⁻¹ corresponds to weakly bounded SO₂ molecules [27].

Upon further annealing at 300 °C in humid air (Figure 11b), a rapid disappearance of the oxidation products signals was observed. The negative absorption bands (1306, 1270, and 1154 cm⁻¹) indicate decomposition of the sulfite groups remained on the cobalt oxide surface after supplying H₂S in dry air. Thus, the catalytic effect of water on the decomposition of surface sulfite groups takes place (Figure 12).

3.3.3. Interaction of Ni_0.1Co_2.9O₄ with H₂S in Dry and Humid Air

In order to investigate the influence of Ni addition on the reactivity of Co₃O₄ with H₂S, the Ni_0.1Co_2.9O₄ sample has been chosen for DRIFT investigation. This sample has the maximum nickel content, which can significantly affect the surface active centers. DRIFT spectra of Ni_0.1Co_2.9O₄ (Figure 13) obtained at the same conditions are similar to those for Co₃O₄.

Comparing the absorption modes of Ni_0.1Co_2.9O₄ with Co₃O₄ spectra (

Table 3. Absorption bands on Co₃O₄ and Ni_0.1Co_2.9O₄ DRIFT spectra.

Surface Groups	Adsorption Bands, cm−1
	Co3O4		Ni0.1Co2.9O4
	0% R.H.	60% R.H.	0% R.H.	60% R.H.
OH	3400 (broad.)	3430 (broad.)	3440 (broad.)	3430 (broad.)
H2O	1616	1621	1621	1618
SO42−	1340, 1124	−	1346	−
SO32−	1030	1030, 1050	1030, 1050	1030, 1050
HSO3−	1290, 1154	1270, 1154	1287, 1145	1275, 1154
SO2(ads)	1330	1306	1327	1306

), a shift to the low-frequency region of a number of bands was noticed. In general, the introduction of nickel into the structure did not distinctly affect the H₂S oxidation process and the adsorption of products on the sample surface.

3.3.4. DRIFT Spectra Baseline Absorption and Sensor Signal

For semiconductor compounds, part of the infrared absorption occurs on free carriers. From the Drude-Zener theory it can be shown [30,31] that the absorption at a given wavelength A(λ) is directly related to the electrical conductivity of the material:

A(λ) = σ(λ)z/ε₀ c n,

(5)

where σ (λ) is the electrical conductivity depending on the wavelength, z is the thickness of the sample, ε₀ is the permittivity of the vacuum, c is the velocity of light, and n is the refractive index of the material.

Since the DRIFT spectra were recorded on the same sample, the change in the baseline intensity is associated with a change in the electrical conductivity of the sample, and hence its resistance. This allows us to relate the change in the absorption intensity of the baseline (Figure 14) and the resistance of the samples in the sensor experiments. Moreover, the change in resistance in both cases was caused by the processes of oxidation of hydrogen sulfide and adsorption of products on the surface of oxides.

As follows from the change in the absorption value at 2000 cm⁻¹, for both Co₃O₄ and Ni_0.1Co_2.9O₄ in a flow of hydrogen sulfide in dry air, the saturation electrical conductivity was lower (the resistance was higher) than in humid conditions. This means that for a long time of H₂S supply, the sensor signal should be higher in dry air. However, the conductivity curve for Co₃O₄ in humidity reaches saturation faster especially at short times (less than 10 min). The observed sensor signal (measurement for 15 min) turns out to be higher in humid air. It should be noted that the concentration of hydrogen sulfide for recording DRIFT spectra (20 ppm) is significantly higher than for sensor measurements (1 ppm) due to the different sensitivity of the methods. This should lead to an increase in the rate of all processes associated with H₂S in the DRIFT experiment.

A decrease in the sensor response time τ₉₀ also indicates the processes of the sensor signal formation for Co₃O₄ in humid air are faster (Figure 15a).

In the case of Ni_0.1Co_2.9O₄, the response time to H₂S τ₉₀ is the same in both dry and humid air, and coincides with that for Co₃O₄ in humidity. At the early times, the sensor signal of Ni_0.1Co_2.9O₄ in dry air grows even a little faster (Figure 15b). Similar trends can be seen in the DRIFT spectra (Figure 14b).

For both samples, the baseline of DRIFT spectra in dry air after the annealing cycle does not return to the initial value (Figure 14). That indicates the accumulation of sulfite and sulfate groups on the sensor surface during the detection cycles. Whereas, the sensor completely restores its original resistance in humid conditions. The observed increase in conductivity (decrease in resistance) is caused by the remove of sulfite groups during the annealing in dry air and by the increase in temperature.

4. Conclusions

In dry air, the surface of Co₃O₄ and Ni_0.1Co_2.9O₄ is partially covered by chemisorbed oxygen species and a relatively small number of hydroxyl groups, which can act as oxidants. Therefore, during H₂S interaction with the oxide surface, the predominant process of the oxidation is due to chemisorbed oxygen according to reaction (2), leading to the formation of SO₂ and water. This process increases the number of electrons (minority charge carriers) in the conduction band, and the resistance grows. The resulting water adsorbs on the surface and dissociates with the formation of new hydroxyl groups. On the one hand, hydroxyl groups are oxidizing agents and, on the other hand, they are more favorable as SO₂ adsorption centers than oxygen. Further oxidation of surface sulfites to sulfates by reaction (4) serves as an additional source of electrons, which increases the sensor signal at long times. In our previous work [19], it was shown that the concentration of chemisorbed oxygen on the surface of Ni_0.1Co_2.9O₄ is lower than on Co₃O₄. This leads to a slowdown in the rate of reaction (2) and, as a result, led to a decrease in sensor signal toward 1 ppm H₂S with increase in Ni content.

In humid air, during dissociative adsorption of water molecules, the concentration of hydroxyl groups increases on the surface of oxides. Reactivity of the surface hydroxyl groups leads to an increase of the sensor signal with increasing of the relative humidity. The decrease in the sensor response time τ₉₀ for Co₃O₄ in humid air indicates the acceleration of H₂S oxidation due to a change in the reaction mechanism, which is most likely realized with the participation of hydroxyl groups (3). The saturation state is reached when all surface hydroxyl groups are converted to hydrosulfites by reacting with SO₂. In this case, no further oxidation to sulfates was observed.

The sensor response time τ₉₀ for Ni_0.1Co_2.9O₄ in both dry and humid air coincides in value with that for Co₃O₄ in humid air. This may indicate the participation of hydroxyl groups in the oxidation of hydrogen sulfide in both cases and the predominant of the reaction (3). In this case, an increase in the sensor signal in humidity is most likely caused by an increase in the concentration of hydroxyl groups. The signals of sulfate groups in the DRIFT spectra for Ni_0.1Co_2.9O₄ in dry air are less pronounced than for Co₃O₄, which also indicates that the oxidation of sulfites to sulfates occurs due to chemisorbed oxygen rather than hydroxyl groups.