Bacterial Cellulose Templated p-Co₃O₄/n-ZnO Nanocomposite with Excellent VOCs Response Performance

Ⅰ.   INTRODUCTION

Nowadays, monitoring of volatile organic compounds (VOCs) is one of the urgent demands on human's health and safety. Rapid popularity of mobile platforms such as smartphones and smart bracelets offers new opportunities for VOCs detection, meanwhile, mobile applications also pose several requirements such as high sensitivity, low price, low power consumption and small dimension [1-5]. Metal oxide semiconducting gas sensors which perfectly meet the requirements have attracted enormous attentions [6-8]. ZnO, as an important n-type semiconductor, has been widely utilized for gas sensor due to its low cost, nontoxicity, high carrier mobility, outstanding physical and chemical stability [9-11]. However, ZnO based gas sensors typically need to operate at high temperature to achieve high response [12, 13], which is a big obstacle in practical application because of increased power consumption, low reliability, and even a safety risk in presence of combustible gases. Several strategies have been proposed to explore high performance VOCs sensor operated at lower temperature, such as morphology control [14-17], defect engineering [9, 18, 19], noble metal decoration [20-24], and heterojunction formation [25, 26]. Attentive to expensive and rare noble metals, the introduction of metal oxides to form composite is an economical method to improve the sensing performance. Appropriate combination of p-type oxides with n-type ZnO has been demonstrated as an efficient approach to achieve high response at lower temperature because of the formation of a deeper extended depletion layer [27-29]. Co$_3$O$_4$ has been regarded as one of the favorable materials to combine with ZnO for gas sensor application because of its high catalytic activity toward VOCs and oxygen adsorption capability [30-32]. Weak interference of sensor responses under humidity and high temperature condition is also reported in p-type oxide sensors [33-36]. Thus, Co$_3$O$_4$ may be an outstanding additive material to improve the performance of ZnO sensors.

Because the performance of oxide nanocomposite is sensitively dependent on the homogeneity of heterostructural nanocomposite, sacrificial templates were widely used to precisely control the homogeneity of grain size, especially the heterogeneous interfaces. Various soft or hard templates, such as zeolitic imidazolate frameworks (ZIFs) [26, 37], metallic organic framework (MOFs) [27] and SiO$_2$ spheres [38], have been used to synthesize porous/hollow Co$_3$O$_4$/ZnO nanocomposites, which have shown improved gas response in comparison to the ZnO counterpart. In this study, bacterial cellulose (BC), which is low-cost and bio-compatible, has been introduced as a self-sacrificial template. BC has a porous three-dimensional (3D) network composed of filamentous nanofibers [39-42]. The surfaces of the nanofibers (diameter 10-100 nm) intrinsically possess abundant hydroxyl groups which serve as adsorption sites for metal cations in precursor [42-44]. Once the surface available hydroxyl groups were occupied or combined by metal cations via strong electrostatic attraction, followed by a rinsing process, a relatively homogeneous and thin metal cation layer could be formed on BC template [45, 46]. The formation of (3D network) metal oxide nanocomposites, as well as removal of BC soft scaffold, could be simultaneously completed by an appropriate calcination process [47]. Moreover, the generation of gaseous CO$_2$ and H$_2$O, during combustion of BC (C$_x$H$_y$O$_z$) may facilitate the formation of nanopores in oxide nanocomposites, which is favorable for gas permeation in gas sensing application [48]. As a sacrifice template for porous metal oxide preparation, BC was often modified and carbonized to enhance structure stability and prevent collapse of template during calcination. In Liu's report, BC was oxidized by periodate and further oxided by chlorite to prepare Cu$_2$O microtube [42]. Yang and co-authors explored the carbon nanofibers derived from BC as a substrate, they prepared spinel CoFe$_2$O$_4$ via an electrostatic assemble method [49]. In this work, BC without any modification was directly used as a template to prepare the Co$_3$O$_4$/ZnO nanocomposite. A freeze drying process was used to prevent the collapse of BC. As a result, the Co$_3$O$_4$/ZnO nanocomposite synthesized here shows excellent VOCs gas response (63.7 to 100 ppm acetone) as compared with those reported in literatures (listed in Table S1 in supplementary materials). BC was proposed as a versatile template for designing heterostructural oxide nanocomposites for high-performance gas sensor as well as other applications.

Ⅱ.   EXPERIMENTS

A   Preparation and characterization of material

Co$_3$O$_4$/ZnO nanocomposite was synthesized using BC as a sacrifice temeplate. Cobalt nitrate hexahydrate (Co(NO$_3$)$_2$$\cdot$6H$_2$O) and zinc nitrate hexahydrate (Zn(NO$_3$)$_2$$\cdot$6H$_2$O) were purchased from Shanghai Aladdin Biochemical Technology Co., Ltd. and Sinopharm Chemical Reagent Co., Ltd. respectively. All of the chemical reagents were analytical grade and used as received without further purification. The BC pellicles produced by Hainan Guangyu Biotechnology Co., Ltd. were cut into small rectangular pieces ($\sim$2 cm $\times$2 cm) and rinsed with deionized water for several times before use. Co(NO$_3$)$_2$ solution was prepared by dissolving 10 mmol Co(NO$_3$)$_2$$\cdot$6H$_2$O in 100 mL deionized water and Zn(NO$_3$)$_2$ solution was prepared by dissolving 10 mmol Zn(NO$_3$)$_2$$\cdot$6H$_2$O in 100 mL deionized water, respectively. At first, ten BC slices were immersed into the Co(NO$_3$)$_2$ solution for 24 h and then thoroughly rinsed with deionized water to obtain BC/Co$^{2+}$ hybrid hydrogels. Afterwards, the BC/Co$^{2+}$ hybrid hydrogels were immersed into the Zn(NO$_3$)$_2$ solution for 24 h, followed by rinsing with deionized water to obtain BC/Co$^{2+}$/Zn$^{2+}$ hybrid hydrogels. For comparison, BC/Zn$^{2+}$ hybrid hydrogels were synthesized using the above mentioned method. Finally, these hybrid hydrogels were dried in a freeze-dryer and calcinated at 480 ℃ in air for 3 h. The obtained ZnO and Co$_3$O$_4$/ZnO powders were collected and ground for use. The schematic diagram of the synthesis is shown in FIG. 1.

Figure  1.  The schematic diagram of the synthesis process of the Co$_3$O$_4$/ZnO nanocomposite.

DownLoad: Full-Size Img PowerPoint

The powder X-ray diffraction (XRD) patterns were recorded on a Rigaku Smartlab (Rigaku, Japan) X-ray diffractometer. Scanning electron microscopy (SEM) images were characterized by a VGA3 SBH SEM (TESCAN Brno, Czech Republic) at 20 kV. Energy-disperive X-ray spectroscopy (EDS) analysis was performed by a GeminiSEM 500 (Zeiss, American). Transmission electron microscopy (TEM) and high-resolution TEM (HRTEM) images were obtained by Tecnai G2 F20 (FEI, American). A Thermo Scientific ESCALAB250Xi system (Thermo Scientific, American) was used for X-ray photoelectron spectroscopy (XPS) measurements. The electron paramagnetic resonance (EPR) spectroscopy was performed at A300-10/12 (Bruker, Germany). UV-visible diffuse reflectance spectrum (UV-Vis DRS) was recorded on a SOLID3700 (Shimadzu, Japan), which was used to determine the direct band gap. Brunner-Emmet-Teller (BET) measurenents were taken by an ASAP2460 (Micromeritics, American).

B   Sensor fabrication and gas sensing tests

Two kinds of sensors were fabricated for the convenience of comparison: ZnO and Co$_3$O$_4$/ZnO sensors. The slurries were prepared by ultrasonic dispersion of 0.1 g corresponding powders in 1 mL ethanol. Then, the slurries were brushed onto ceramic substrates (1.5 mm $\times$1.5 mm$\times$0.25 mm, with predefined gold testing and heating electrodes) to form sensing layers. Finally, the sensors were aged at 220 ℃ (4.5 V) for 10 days to ensure a good stability.

VOCs sensing tests were performed in four channel gas sensing measurement system (SD101, Huachuang Ruike Science and Technology Wuhan Co., Ltd) with a dynamic flowing system. The working temperature was adjusted by the heating voltage. The relationship between working temperature and heating voltage was shown in FIG. S1 (supplementary materials). The concentration of the target gas was adjusted by the flow rate of the standard gas (1000 ppm in dry air) and dry air via mass flow controllers. The total flow rate was fixed at 1000 sccm. The sensor resistances in target gas and in dry air were expressed as $R_\textrm{g}$ and $R_\textrm{a}$, respectively. Sensor response is defined by $R_\textrm{a}$/$R_\textrm{g}$.

Ⅲ.   RESULTS AND DISCUSSION

A   Materials characterizations

The typical XRD patterns of ZnO and Co$_3$O$_4$/ZnO samples are shown in FIG. 2. All the diffraction peaks of ZnO are in good agreement with the standard card of ZnO (PDF card No.36-1451). Absence of noticeable impurity peaks suggests that BC template has been completely burned away during the calcining process. Under the similar synthesis procedure, pure Co$_3$O$_4$ (PDF card No.78-1969)) is formed, as envidenced by XRD in FIG. S2 (supplementary materials). For Co$_3$O$_4$/ZnO nanocomposite, no obvious peak corresponding to Co$_3$O$_4$ phase is detected, which may be due to the low content of Co$_3$O$_4$.

Figure  2.  XRD patterns of ZnO and Co$_3$O$_4$/ZnO nanocomposite.

DownLoad: Full-Size Img PowerPoint

The SEM images of BC, BC/Zn$^{2+}$ and BC/Co$^{2+}$/Zn$^{2+}$ aerogels before calcination are shown in FIG. S3 (a-c) (supplementary materials). All of the aerogels show 3D network structure with smooth nanofiber surface. As seen in FIG. 3(a), after calcination the ZnO sample is composed of nanoparticles which are connected in fibrous strings, indicating a partly preserved 3D interconnected porous nanofibrous network structure in comparison to BC, while Co$_3$O$_4$/ZnO nanocomposite (FIG. 3(b)) shows cross-linked granular nanostructure with evident interconnected pores. It is well known that BC is rich in hydroxyl groups which are capable of trapping metal ions (Zn$^{2+}$, Co$^{2+}$). By conjugating the metal ions to the surface hydroxyl groups in diluted metallic nitrate solutions and then rinsing off the unbound metal ions, metal ions are uniformly adsorbed on the surface of nanofibers. During the calcination process, BC starts to decompose at 240 ℃ and decomposes completely at 361 ℃ [47] while ZnO and Co$_3$O$_4$ are formed, and the shape of template is partly retained. These BC-derived porous structure characteristics are conducive to the diffusion of gas molecules in the sensors and beneficial to improve gas sensing properties. It can be seen in FIG. 3(c) that Co$_3$O$_4$/ZnO nanocomposite has an average particle size of about 15 nm. The HRTEM image in FIG. 3(d) shows the characteristic spacings of 0.232 nm and 0.260 nm corresponding to the (222) and (002) lattice planes of Co$_3$O$_4$ and ZnO phases, respectively, indicating the good crystallinity of the synthesized nanocomposite. The element compositions of Co$_3$O$_4$/ZnO nanocomposite were determined by EDS as shown in FIG. 3(e). The peaks of O, Co, Zn and C elements were detected, and C element comes from the conductive carbon adhesive. The atomic ratio of Co/Zn calculated by EDS results is 1:15, which is close to the lower detection limit of XRD, and that may be the reason for absence of Co$_3$O$_4$ XRD peaks in Co$_3$O$_4$/ZnO nanocomposite. The EDS elemental mapping images (FIG. S3 (d-f) in supplementary materials) demonstrate that the elements O, Zn and Co were uniformly distributed, indicating well dispersing of Co in the composite.

Figure  3.  SEM images of (a) ZnO and (b) Co$_3$O$_4$/ZnO, (c) TEM and (d) HRTEM images of Co$_3$O$_4$/ZnO, (e) EDS spectrum of Co$_3$O$_4$/ZnO.

DownLoad: Full-Size Img PowerPoint

As the grain size plays an important role in the response of oxide sensor, the nitrogen adsorption-desorption isotherms were measured to compare the specific surface area and pore size distribution of ZnO and Co$_3$O$_4$/ZnO samples. The results are shown in FIG. 4. The calculated BET surface areas of ZnO and Co$_3$O$_4$/ZnO samples are 19.6 and 20.2 m$^2$/g, respectively. For both samples, the pore sizes show wide distribution from 5 nm to 114 nm with the center peak at 41 nm.

Figure  4.  (a) N$_2$ adsorption-desorption isotherms, (b) pore size distribution curves of the ZnO and Co$_3$O$_4$/ZnO.

DownLoad: Full-Size Img PowerPoint

The chemical compositions of the samples were further identified by XPS. The full spectra of ZnO and Co$_3$O$_4$/ZnO nanocomposite are shown in FIG. S4(a) (supplementary materials). The peak positions of the ZnO and Co$_3$O$_4$/ZnO are roughly the same, both of them have characteristic peaks of O and Zn elements, but there are weak Co peaks in the Co$_3$O$_4$/ZnO. FIG. 5(a) shows the Zn 2p spectra of ZnO and Co$_3$O$_4$/ZnO. For pure ZnO, two peaks at 1044.4 eV and 1021.2 eV with an energy separation of 23.2 eV correspond to Zn 2p$_{1/2}$ and Zn 2p$_{3/2}$, respectively. For Co$_3$O$_4$/ZnO nanocomposite, both Zn 2p$_{1/2}$ and Zn 2p$_{3/2}$ peaks show an energy shift of 1.2 eV toward higher binding energy, which arises from the interaction and electron transformation between ZnO and Co$_3$O$_4$ [38, 50]. FIG. 5(b) shows the fine-scan of Co 2p peaks in the Co$_3$O$_4$/ZnO, two major peaks at 780.4 eV and 796.2 eV were consistent with Co 2p$_{3/2}$ and Co 2p$_{1/2}$, respectively. Beyond that, two weak peaks at 784.1 eV and 804.7 eV were considered to be vibrating satellite peaks. The Co 2p$_{3/2}$ peak could be divided into Co$^{2+}$ and Co$^{3+}$ peaks centered at 784.5 eV and 780.3 eV, and similarly the Co 2p$_{1/2}$ peak could be divided into 796.9 eV and 804.1 eV, which further confirms the existence of Co$_3$O$_4$ crystal phase in the composite. FIG. 5(c, d) show the spectra of O 1s of ZnO and Co$_3$O$_4$/ZnO samples, respectively. Both of them could be divided into three oxygen peaks, indicating that there are three types of oxygen in ZnO and Co$_3$O$_4$/ZnO samples. For ZnO, the peak centered at 530.1 eV is assigned to lattice oxygen (O$_\textrm{L}$), the peak at 530.8 eV is generated by the vacancy oxygen (O$_\textrm{V}$), and the peak at 532 eV is attributed to the chemisorbed oxygen (O$_\textrm{C}$). The proportions of O$_\textrm{L}$, O$_\textrm{V}$ and O$_\textrm{C}$ are 44.13%, 28.95%, and 26.92%, respectively. For Co$_3$O$_4$/ZnO nanocomposite, the proportions of O$_\textrm{L}$ (530.5 eV), O$_\textrm{V}$ (531.1 eV) and O$_\textrm{C}$ (532.1 eV) are 44.02%, 34.08% and 21.90%, respectively. The results show that the the O$_\textrm{V}$ content is much higher in the Co$_3$O$_4$/ZnO composite than in pure ZnO, indicating the greater abundance of O$_\textrm{V}$ in the Co$_3$O$_4$/ZnO composite, which may improve the gas sensing performance of Co$_3$O$_4$/ZnO composite.

Figure  5.  (a) Zn 2p spectra of ZnO and Co$_3$O$_4$/ZnO, (b) Co 2p spectrum of Co$_3$O$_4$/ZnO, (c) O 1s spectra of ZnO and (d) O 1s spectra of Co$_3$O$_4$/ZnO.

DownLoad: Full-Size Img PowerPoint

EPR is a direct and effective method to detect the chemical nature of intrinsic defects, specifically impurities with unpaired electrons [51], thus EPR characterizations are further conducted to gain more information about the vacancies in nanocomposite. As shown in FIG. 6, EPR spectra of ZnO and Co$_3$O$_4$/ZnO are featured by $g$ signal at 2.003, which can be assigned to oxygen vacancies. The signal is stronger for Co$_3$O$_4$/ZnO compared with the ZnO. It indicates that a greater abundance of oxygen vacancies in Co$_3$O$_4$/ZnO composite than in pure ZnO, which is consistent with the result of XPS. The EPR spectra of Co$_3$O$_4$, ZnO and Co$_3$O$_4$/ZnO prepared under the same conditions are shown in FIG. S4(b) (supplementary materials), Co$_3$O$_4$ has more oxygen vacancies than ZnO, thus the composite is rich in oxygen vacancies than ZnO probably because of the introduction of Co$_3$O$_4$.

Figure  6.  EPR spectra of ZnO and Co$_3$O$_4$/ZnO.

DownLoad: Full-Size Img PowerPoint

Diffuse reflectance spectra of ZnO and Co$_3$O$_4$/ZnO were measured in the wavelength range of 200-900 nm, as shown in FIG. S4(c) (supplementary materials). The energy gap was calculated by Kubelka-Munk formula:

where, $F(R)$ is the Kubelka Munk function, $R$ is the reflectivity [52]. The band gap of pure ZnO was determined to be 3.32 eV, while that of Co$_3$O$_4$/ZnO nanocomposite was reduced to 2.49 eV. This may be an indirect evidence of the incorporation of Co$_3$O$_4$ in ZnO.

B   Gas-sensing properties

The performance of semiconducting sensors are greatly influenced by working temperature. FIG. 7(a) shows the responses of ZnO and Co$_3$O$_4$/ZnO composite sensors to 100 ppm acetone at different working temperature. The optimum working temperature of Co$_3$O$_4$/ZnO sensor is 180 ℃ which is much lower than that of ZnO sensor (240 ℃). Moreover, the response of Co$_3$O$_4$/ZnO sensor toward 100 ppm acetone is 63.7, which is about 26 times higher than that of ZnO sensor (2.3) at 180 ℃, demonstrating the remarkable improvement of gas sensing performance by the integration of Co$_3$O$_4$ and ZnO. The dynamic response curves of ZnO and Co$_3$O$_4$/ZnO sensors toward 100 ppm acetone at 180 ℃ are shown in FIG. 7(b). The resistances of both sensors decrease after exposure to acetone atmosphere, demonstrating n-type response behavior. FIG. 7(c) shows the repetitive response of Co$_3$O$_4$/ZnO sensor toward 100 ppm acetone at 180 ℃, which suggests good repeatability and reproducibility. As shown in FIG. 7(d), the typical response of Co$_3$O$_4$/ZnO sensor continuously increases as the acetone concentration increasing from 200 ppb to 500 ppm, with no sign of saturation. Notably, a response of 1.4 to a low acetone concentration of 200 ppb is observed, which is lower than the value in exhaled gas of a diabetic (1.8 ppm level), and even lower than that of a healthy person [53], allowing potential application of present Co$_3$O$_4$/ZnO sensor in respiratory gas detection. The selectivity of the sensor is also an important parameter, because the gas sensors are always exposed to complex gas environment. FIG. 7(e) shows the responses of ZnO and Co$_3$O$_4$/ZnO sensors to 100 ppm of VOCs gases, such as acetone, ethanol, $p$-xylene, benzene, anisole and formaldehyde. It can be clearly confirmed that Co$_3$O$_4$/ZnO sensor always shows a higher response than the ZnO sensor in all tested gases. The response to acetone is 63.7, much higher than other gases. The response to 100 ppm acetone of Co$_3$O$_4$/ZnO sensor at 180 ℃ was traced for ten weeks to study the long-term stability. As seen from FIG. 7(f), the sensitivity decreased gradually after three weeks and maintained about 86% after ten weeks.

Figure  7.  (a) Response of ZnO and Co$_3$O$_4$/ZnO to 100 ppm acetone at 120-260 ℃, (b) dynamic response of the tested sensors to 100 ppm acetone at 180 ℃, (c) cycle test data of the Co$_3$O$_4$/ZnO based sensor toward 100 ppm acetone, (d) dynamic and transient response-recover curves of the Co$_3$O$_4$/ZnO toward acetone from 200 ppb to 500 ppm at 180 ℃, (e) selectivity of the ZnO and Co$_3$O$_4$/ZnO sensors to 100 ppm of various VOCs gases at 180 ℃, (f) stability of Co$_3$O$_4$/ZnO sensor to 100 ppm acetone at 180 ℃.

DownLoad: Full-Size Img PowerPoint

C   Gas sensing mechanism

In this study, Co$_3$O$_4$/ZnO nanocomposite exhibits significantly improved VOCs sensing performances as compared to pure ZnO. Considering that the material morphology, specific surface area and pore size distribution of Co$_3$O$_4$/ZnO nanocomposite are similar to pure ZnO, the contributions of these factors to the performance enhancement in Co$_3$O$_4$/ZnO nanocomposite could be excluded. The possible reasons for improved performance in Co$_3$O$_4$/ZnO nanocomposite are proposed as follows. (i) The higher oxygen vacancy concentration in Co$_3$O$_4$/ZnO nanocomposite results in the formation of more active sites for adsorption of oxygen molecules and target gas, which is beneficial to improve the response to target gas [54-56]. (ii) The catalytic activity of p-Co$_3$O$_4$ promotes the chemical reactions during the sensing process. On one hand, holes, the majority carrier in Co$_3$O$_4$, can react with the surface adsorbed water molecules to produce highly reactive hydroxyl radicals ($\cdot$OH) which can cause complete decomposition of hydrocarbons to CO$_2$ and H$_2$O. On the other hand, unlike n-type oxides whose adsorption of oxygen is subject to rapid auto-inhibition due to their negatively polarized surface, p-type oxides tend to adsorb oxygen up to complete saturation of the surface [57, 58]. Thus, the Co$_3$O$_4$ additive improves the capability to decompose VOCs and increase the gas adsorption on the surface of Co$_3$O$_4$/ZnO nanocomposite. (iii) Most importantly, the formation of p-Co$_3$O$_4$/n-ZnO heterojunctions produces an additional depletion layer and improves the charge separation at the interface, which results in the enhanced resistance modulation upon interaction with the target gas. According to the literature, the electron affinity, bandgap, and work function of ZnO are 4.5 eV, 3.37 eV, and 5.2 eV, respectively, and those for Co$_3$O$_4$ are 4.91 eV, 2.08 eV, and 6.5 eV, respectively [35]. At the intimate contact areas of ZnO and Co$_3$O$_4$, electrons will flow from ZnO to Co$_3$O$_4$ while holes will flow along the contrary direction until Fermi level alignment, causing the band to bend and an additional depletion layer to form, as shown in FIG. 8. When the composite is exposed to air, the oxygen molecules are adsorbed on the surface of materials and electrons are captured from metal oxides to generate ionic species such as O$_2$$^-$, O$^-$, and O$^{2-}$ [59], resulting in an electrons depletion layer and a holes accumulation layer on the surface of ZnO and Co$_3$O$_4$, respectively [37, 60, 61]. Because of the small amount (Co/Zn=1:15) and homogeneous distribution of Co$_3$O$_4$ in the composite, the composite exhibits n-type response behaviour, suggesting the main conduction path of the composite is n-type ZnO and the contribution of p-Co$_3$O$_4$/p-Co$_3$O$_4$ homointerface to the gas response is negligible. Upon exposure to target gas, such as acetone, the surface reaction between the target gas and adsorbed oxygen ions produces CO$_2$ and H$_2$O. Besides, the trapped electrons are released back to the conduction band, leading to an increase in electron concentration on the surface of ZnO and a decrease of hole density on the surface of Co$_3$O$_4$. Consequently, the additional depletion layer thickness and the barrier height decrease, resulting in a magnified resistance modulation. In other words, the variation of resistance ($\Delta R$) from air to acetone atmosphere is larger in the presence of Co$_3$O$_4$. Therefore, the Co$_3$O$_4$/ZnO composite has a higher sensor response than pure ZnO. The reaction kinetics is as follows (Eqs. (2)-(5)):

Figure  8.  (a) Energy band diagram before contact of Co$_3$O$_4$ and ZnO, change of the heterojunction depletion layer for Co$_3$O$_4$/ZnO nanocomposite sensor in air (b) and in acetone (c).

DownLoad: Full-Size Img PowerPoint

Ⅳ.   CONCLUSION

In summary, porous granular Co$_3$O$_4$/ZnO nanocomposite was synthesized using BC as a self-sacrificial template. The sensor based on Co$_3$O$_4$/ZnO nanocomposite exhibits much higher response to VOCs gases and lower working temperature than pure ZnO, which can be attributed to the synergistic effect of high oxygen vacancy concentration, catalytic effect of Co$_3$O$_4$ and homogeneous distribution of p-n heterojunctions. The results demonstrate BC is a promising low-cost and bio-compatible template for the development of high-performance semiconducting metal oxide based gas sensors.

Supplementary materials: A table of sensing performance of Co$_3$O$_4$/ZnO reported in literatures to various gases (Table S1) is given. The relationship between heating voltage and working temperature (FIG. S1), XRD pattern of Co$_3$O$_4$ (FIG. S2), SEM images of bacterial cellulose, ZnO and Co$_3$O$_4$/ZnO before calcination, EDS spectrum and elemental mapping images of O, Zn and Co in Co$_3$O$_4$/ZnO (FIG. S3), XPS full spectra, EPR spectra of ZnO, Co$_3$O$_4$ and Co$_3$O$_4$/ZnO, UV-Vis absorption spectra for ZnO and Co$_3$O$_4$/ZnO, Tauc plot for the calculation of the direct band gap of the pure ZnO and Co$_3$O$_4$/ZnO nanocomposite (FIG. S4) are available.

Scheme  S1.  The relationship between heating voltage and working temperature.

DownLoad: Full-Size Img PowerPoint

Scheme  S2.  XRD pattern of Co₃O₄.

DownLoad: Full-Size Img PowerPoint

Scheme  S3.  SEM image of (a) bacterial cellulose, (b) ZnO and (c) Co₃O₄/ZnO before calcination, (d-e) EDS spectrum and elemental mapping images of O, Zn and Co in Co₃O₄/ZnO.

DownLoad: Full-Size Img PowerPoint

Scheme  S4.  (a) XPS full spectrum, (b) EPR spectra of ZnO, Co₃O₄ and Co₃O₄/ZnO, (c) UV-vis absorption spectra for ZnO and Co₃O₄/ZnO, (d) Tauc plot for the calculation of the direct band gap of the pure ZnO and Co₃O₄/ZnO nanocomposite.

DownLoad: Full-Size Img PowerPoint