Gas Sensing Properties of p-Co₃O₄/n-TiO₂ Nanotube Heterostructures

Abstract

In this paper, we fabricated p-Co₃O₄/n-TiO₂ heterostructures and investigated their gas sensing properties. The structural and morphological characterization were performed by scanning electron microscopy (SEM), X-ray diffraction (XRD), and X-ray photoelectron spectroscopy analysis (XPS). The electrical properties of the heterostructure were studied within the temperature range from 293 K to 423 K. Changes in electrical properties and sensing behavior against reducing and oxidizing gases were attributed to the formation of p–n heterojunctions at the Co₃O₄ and TiO₂ interface. In comparison with sensing performed with pristine TiO₂ nanotubes (NTs), a significant improvement in H₂ sensing at 200 °C was observed, while the sensing response against NO₂ decreased for the heterostructures. Additionally, a response against toluene gas, in contrast to pristine TiO₂ NTs, appeared in the Co₃O₄/TiO₂ heterostructure samples.

1. Introduction

Metal-oxide (MOX)-based gas sensors have a wide range of applications such as medical, air quality, energy efficiency, and the detection of hazardous gases. MOXs are preferred for sensing layers due to their easy fabrication, low cost, high sensor response, and easy integration [1]. Moreover, device performance is improved when sensors are fabricated with nanostructured MOXs. However, for almost all sensor applications, there is a strong demand for better sensor properties such as lower operation temperature and higher sensitivity and selectivity [2,3]. Different techniques, such as doping, loading, or heterostructure fabrication, have been employed to meet these requirements [4,5,6,7,8]. Among them, the combination of n-type and p-type MOX materials in the form of nanostructures is one of the most promising techniques [6,9]. An extended depletion layer between two different types of MOX semiconductors can provide higher sensor properties [9].

This study is focused on the combination of p-type cobalt oxide (Co₃O₄) and n-type TiO₂. TiO₂ is a very important material, especially for the sensing of H₂ and VOCs, due to its high sensitivity and easy nanostructure fabrication [10,11,12,13,14,15]. Co₃O₄, well known as a catalytic activator in chemical processes such as oxidation reactions, has a great potential for sensor applications [16,17,18,19]. p-type Co₃O₄ can advance the sensing capabilities of TiO₂ nanostructures. In this study, different amounts of Co₃O₄ were loaded on TiO₂ nanotubes (NTs) to understand the heterostructure effect on sensing properties. The heterostructures were fabricated by a two-step electrochemical process. TiO₂ NTs were fabricated via an anodization process and Co₃O₄ was loaded onto the NTs via electrochemical deposition. Then, electrical and sensor measurements were carried out.

2. Experiment

Co₃O₄/TiO₂ heterostructures were fabricated via two-step electrochemical deposition consisting of anodization and cathodic deposition. Ti foils (20 mm in length and 10 mm in width) were provided from Sigma-Aldrich (St. Louis, MO, USA). NH₄F (ammonium fluoride, 99.99%), ethylene glycol (99.8%), and cobalt (II) nitrate hexahydrate (99.999%) were purchased from Sigma-Aldrich. Ultra-pure (UP) water was used for all experiments. In the first step, Ti foils were cleaned with acetone, isopropanol, and UP water in ultrasonic bath, respectively. Then, the foils were dried in a nitrogen stream. An electrolyte solution was prepared with NH₄F, ethylene glycol, and UP water with a molar ratio of 0.25:2:100. The anodization was carried out at 50 V with Pt foil as cathode in a thermos-stated bath at 20 °C for 1 h. Afterward, Co₃O₄ was loaded on TiO₂ NTs (set as cathode) via cathodic deposition. The electrolyte for cathodic deposition was composed of 0.01 wt % cobalt (II) nitrate hexahydrate in UP water. Cathodic deposition was performed under three different conditions (5 V for 3 min, 10 V for 1 min, and 10 V for 3 min), and the resulting samples were named and are hereafter referred to as CT-1, CT-2, and CT-3, respectively. After the fabrication of Co₃O₄/TiO₂ heterostructures, samples were rinsed with UP water and dried under a nitrogen stream. Finally, the fabricated samples were annealed at 500 °C for 3 h in an oxygen atmosphere to obtain a crystalline structure.

To investigate the crystal structure and morphological properties of the fabricated samples, XRD, SEM, and XPS methods were used.

Current–voltage (I–V) measurements were performed in the temperature range of 20–200 °C to investigate the electrical properties of fabricated samples. Au electrodes 100 nm thick were thermally evaporated on top of the fabricated samples as shown in Figure 1a,b. Because of the geometry of the devices, measurements were performed for vertical configuration (contacts are on top of the NTs and on Ti foil). A Keithley electrometer, model 6517 A, was used for electrical measurements with a Lakeshore 340 temperature controller. I–V curves were obtained by a linear voltage sweep between −1 V and +1 V with a scan rate of 0.05 V.

Pristine TiO₂ and Co₃O₄/TiO₂ heterostructure sensors were investigated toward H₂, NO₂, and VOCs (ethanol, acetone, toluene, and xylene). The sensors were mounted in a homemade test chamber with a volume of 1 L. Desired H₂ and NO₂ gas concentrations were obtained from gas tubes via mass flow controllers. VOCs were generated from cooled bubblers immersed in a thermal bath with dry air as a carrier gas [20]. Concentration of the VOCs were calculated using Antoine’s equation [21]. Baseline signal of the sensor was determined under dry air flow conditions. After the exposure to the target gases, recovery of the sensor from saturated conditions was achieved under dry airflow. Sensor measurements are presented in terms of sensor response, defined as R_s = ΔI/I₀, where ΔI is the change in current. Here, R_s is defined for reducing gases such as H₂ and VOCs, and ΔI is given as I_g − I₀. For oxidizing gases, such as NO₂, sensor response is defined slightly differently: R_s = ΔI/I_g. Here, ΔI is given as I₀ − I_g. For both formulations, I_g is the current after the sensor is exposed to the reducing or oxidizing analyte gas, and I₀ is the reference value (baseline current) of the sensor exposed to high-purity dry air.

3. Results and Discussion

Corresponding reactions for TiO₂ and Co₃O₄ formation in the two-step electrodeposition route can be given as follows [17,22,23]:

$$2{\mathrm{H}}_2\mathrm{O}\to { \mathrm{O}}_2+4{\mathrm{e}}^{-}+4{\mathrm{H}}^{+}$$

(1a)

$$\mathrm{Ti}+{\mathrm{O}}_2\to {\mathrm{TiO}}_2$$

(1b)

$${\mathrm{CO}}_3{\mathrm{O}}_4+{\mathrm{H}}_2\mathrm{O} +{\mathrm{OH}}^{-}\leftrightarrow 3\mathrm{CoOOH}+{\mathrm{e}}^{-}$$

(2a)

$$\mathrm{CoOOH}+{\mathrm{OH}}^{-}\leftrightarrow {\mathrm{CoO}}_2+{\mathrm{H}}_2\mathrm{O}+{\mathrm{e}}^{-}.$$

(2b)

Figure 2 shows SEM images of fabricated structures. Figure 2a indicates that a film of vertically aligned TiO₂ NTs homogeneously covers the surface. The average NT diameter is 80 nm as seen in Figure 2a inset. Figure 2b–d shows cathodically deposited particles on the NTs. In the cathodic process, the amount of deposited Co₃O₄ changes with bias voltage and process time. Thin film formation was observed for CT-3, which was grown with a 10 V bias voltage and a 3 min process time (Figure 2d). EDS mapping and spectrums of CT-1 and CT-3 are given in Figure 3. The presence of the Co, Ti, and O was observed. XRD patterns of pristine TiO₂ NTs and Co₃O₄/TiO₂ heterostructures are given in Figure 4. The XRD patterns reveal that all diffraction peaks could be indexed to Ti and anatase TiO₂ NTs [20]. According to the XRD results, Co₃O₄ particles on the surface most likely do not have crystalline structures. Even if there is a small degree of crystallinity for Co₃O₄ at 36.32°, the TiO₂ reflection overlaps with that of the Co₃O₄, thus preventing any clearly visible peak from the rest of the signal [24].

To identify the surface chemical compositions and oxidation states of the NTs after Co deposition, X-ray photoelectron spectroscopy (XPS) was performed. CT-2 spectra shows Ti 2p and O 1s peaks originating from the TiO₂ NTs. However, the Co 2p peak is very weak and can only be seen when the data is magnified (×10), which will be shown in the high-resolution XPS spectrum of the Co 2p region (Figure 5). This implies that the Co oxide coverage is very low in this sample. The presence of Co, in addition to the presence of Ti and O, is visible in the spectrum of CT-1, which indicates that the coverage is more pronounced in this sample. On the other hand, the CT-3 survey scan shows a more intense peak for Co 2p but not for Ti 2p, indicating that the surface is fully covered by cobalt oxide, which means that TiO₂ NTs are too far under the surface to detect. SEM images of CT-1, CT-2, and CT-3 in Figure 2 support these findings. CT-1 and CT-2 show islands of Co₃O₄ on TiO₂ NTs, whereas a full coverage is clear in CT-3.

Figure 5 shows Co 2p spectra of CT-1, CT-2 and CT-3. CT-1 and CT-2 spectra show two peaks at binding energies of 780.1 eV and 795.1 eV, corresponding to Co 2p3/2 and Co 2p1/2 peaks, implying a Co₃O₄ form [25]. The peak intensity of CT-2 is lower than CT-1 (the intensity of CT-2 is magnified 10× for clarity in Figure 4), indicating lower coverage compared to CT-1. The Co peak intensity of CT-3 is the highest, which was expected from the complete Co₃O₄ layer. The Co 2p3/2 peak of CT-3 shifted to 779.9 eV, which can be attributed to the increase in Co₃O₄ coverage from the sub-monolayer to the multilayer, resulting in a change in the local electronic properties and bonding geometries. Similar shifts have been observed in TiO₂ and Al₂O₃ films as well [26,27,28].

Figure 6 shows the high-resolution XPS spectrum of the O 1s regions for CT-1, CT-2, and CT-3, all of which can be fitted into two peaks. CT-1 and CT-2 show a primary peak at 530.5 eV, which can be attributed to the O–Ti bond in TiO₂ [25]. The smaller secondary peaks at 532.3 eV and 531.1 eV in CT-2 and CT-1, respectively, can both be attributed to the O in Co₃O₄. In CT-1, the difference in peak height between the 530.5 eV peak and the secondary peak is higher compared to that in CT-2, suggesting that the Co₃O₄ contribution is less in this sample compared to CT-1, indicative of less Co₃O₄ coverage. In CT-3, where the surface is fully covered with Co₃O₄, the peak position of the Co–O bond is shifted from 531.1 eV to 529.9 eV and becomes the primary peak, possibly due to the Co₃O₄ coverage from the sub-monolayer to the multilayer. The small changes in the geometry of the film and the changes in the local electronic properties and bonding geometries could be the main contributor to the shift. Similar shifts have been observed for TiO₂ and Al₂O₃ films as well [26,27,28]. The peak at 530.9 eV is attributed to the O–H group in the Co hydroxide precursor [25].

Electrical characterization was performed under the condition and parameters as given in our previous work [29]. Electrical resistance of the heterostructures given in Figure 7a shows an increment with the loading of Co₃O₄. This behavior is attributed to the modification of the conduction channel by the depletion of charge carriers at the heterojunction interface. This mechanism, as discussed later in this text, is also responsible for the enhanced sensor performance of the heterostructures in terms of reducing gases. The I–V curves measured at 200 °C are given in Figure 7b.

Amperometric sensor measurements of H₂, VOCs (ethanol, acetone, toluene, xylene, and chloroform), and NO₂ were performed under a constant voltage at different operation temperatures. The gas sensing responses of samples at 200 °C are shown in Figure 8. These data confirm the absence of sensitivity toward acetone and ethanol for Co₃O₄/TiO₂ heterostructures, while TiO₂ NTs could sense both VOCs. On the other hand, the sensor response to hydrogen increased with Co₃O₄/TiO₂ heterostructures. The CT-1 sample, compared to the sample of pristine TiO₂ NTs, showed an approximately nine-fold higher response to H₂. In addition, the response to toluene starts to appear after Co₃O₄ loading. Sensor responses to 1000 ppm H₂ in 50% relative humidity (RT) can be seen in Figure 8. It is well known that the presence of humidity can cause a deterioration in sensor properties [30]. Despite the relative magnitudes of heterostructure sensor responses, compared with those of pristine TiO₂, to pure H₂ and H₂/humidity mixtures, there was no significant improvement observed when humidity was eliminated (Figure 8). In addition to H₂ and VOCs, sensors responses to 50 ppm NO₂ are given in Figure 7. The sensor responses are 17, 0.02, 1.32, and 3.3 for pristine TiO₂, CT-1, CT-2, and CT-3, respectively.

Figure 9a shows sensor responses for different H₂ concentrations at 200 °C. Co₃O₄-loaded samples could detect lower H₂ concentrations than pristine TiO₂ could. The CT-1 sample exhibited the highest sensor responses at every concentration of H₂. Sensor response vs. temperature graph was given in Figure 9b. The CT-1 sample showed better sensor responses at all temperatures. Although the CT-1 sample showed the highest sensor response at 100 °C, a better response time and recovery time for CT-1 were observed at 200 °C (data not shown). Dynamic sensor responses can be seen in Figure 10. It is obvious that, to obtain the best sensor performance from these devices, operation temperature is a critical variable, and its optimum value will be different for each device. Regarding the response time and recovery time, the loaded samples exhibited lower values and thus better performance than did the pristine TiO₂ NTs, as can be seen in Figure 9c,d.

Based on the above results, Co₃O₄-loaded TiO₂ NTs heterostructures showed enhanced sensor properties compared to the pristine TiO₂ NT sensor. Sensing mechanisms of MOX sensors that are composed of only p-type or n-type materials have been studied and well explained in the literature [31,32,33,34,35]. Enhanced sensor properties can be attributed to variation in the resistance by the formation of the p–n junction [35]. Even though there is no readily accepted and developed mechanism with detailed sensing treatment for heterostructures, two considerations might be taken into account as responsible mechanisms. The first one is the catalytic effect of Co₃O₄. In this case, Co₃O₄ plays a role as a catalyst material in the reaction between analyte gas and TiO₂ [36]. If a continuous film is formed on the surface, a catalytic role of the Co₃O₄ turns into a sensing layer, so a lower sensor performance generally is observed [37]. This idea is also supported by the sensing behavior of the CT-3 sensor response (in Figure 9) and by the XPS and SEM data that indicate full coverage on the surface for CT-3. Previous works have reported the enhanced sensor properties due to the catalyst role and hydrogen sensitivity of Co₃O₄ [16,38,39].

Another possible mechanism that is responsible for the enhanced performance of the sensors is the formation of the p–n junction at the interface between p-type Co₃O₄ and n-type TiO₂ [31,39]. This formation is partially observed on electrical measurements with an increase in the resistance with Co₃O₄ loading (Figure 7). Especially, the CT-1 sensor that exhibits optimal sensor properties evidently showed diode-like rectifying behavior. Due to the depletion region at the heterojunction interface, which causes a narrower cross-sectional area for charge carriers in the nanorod, an increase in resistance is observed. A reduced cross-sectional area available for charge conduction in the nanowire will result in increased resistance. When reducing gases such as VOCs or H₂ are introduced, gas molecules can diffuse into the interface, modify the junction, and thus enhance the sensor response by decreasing the initially high resistance value because of the released electrons into the junction [5]. A relatively easy diffusion of H₂ molecules into the interface, appears as sensing selectivity against H₂ in Figure 8 [40]. Contrary to reducing gases, introduction of an oxidizing gas such as NO₂ will give an additional increase in resistance due to a further widening of the depletion region. However, a tendency for the resistance to further increase will not appear noticeably in sensor response due to the presence of exceedingly high resistance. This mechanism well explains the difference in response data of our samples with respect to reducing and oxidizing gases, given in Figure 8. Another important parameter of the electrical behavior of the p–n junction is operation temperature because of its role in carrier concentration. Its effect appears prominently in Figure 9b.

4. Conclusions

p-Co₃O₄/n-TiO₂ heterostructures were fabricated via a two-step electrochemical process involving anodization and cathodic deposition. The structure and morphology of the heterostructures were characterized by SEM, XRD, and XPS, which confirms that Co₃O₄ particles were spread uniformly over the TiO₂ NT surface. Changes in electrical resistance and I–V curves partially support the p–n junction formation. The p-Co₃O₄/n-TiO₂ heterostructures, in comparison with those of the pristine TiO₂ NTs, show significantly improved H₂ sensing response at 200 °C, which is an up to 9-fold increase. Co₃O₄ loading also provides a sensing response against toluene. In contrast, sensing response against NO₂ decreased for Co₃O₄-loaded samples. This difference in sensor behavior against reducing and oxidizing gases might be attributed to the formation of p–n heterojunctions, leading to the formation of a carrier depletion region and modification of the electrical characteristics of the device with an enhancement in gas sensor characteristics.