Development of Co(OH) x F 2 − x Nanosheets for Acetone Gas Sensor Applications: Material Characterization and Sensor Performance Evaluation

: This study reports the employment of Co(OH) x F 2 − x nanosheets, a new material in the sensor ﬁeld, for gas sensor applications. We synthesize Co(OH) x F 2 − x nanosheets via a hydrothermal route using SiO 2 sphere templates. Our material characterization conﬁrms that the material is a densely clustered Co(OH) x F 2 − x nanosheet with an amorphous microstructure with some short-range ordering. Sensors based on the nanosheets demonstrate a high response of 269% toward 4.5 ppm of acetone gas at an operation temperature of 200 ◦ C and a very low minimum detection limit of 40 ppb. It functions e ﬀ ectively up to a temperature below 300 ◦ C, above which F is found to start to evaporate. Our discussion suggests that an excellent sensor performance arises from the high catalytic function of F incorporated in a high concentration in the material as well as the high speciﬁc surface area due to the morphology of densely clustered nanosheets.


Introduction
There is a growing demand for the sensitive detection of acetone due to its possible usage as a breath marker for the diagnosis of diseases such as diabetes [1]. For breath marker applications, a very high sensitivity (sub-ppm level or lower) is necessary since the exhaled acetone concentration range is very low-from 0.3 to 1.8 ppm [2,3]. Due to the low cost and simplicity of fabrication/operation, metal oxide semiconductors, such as ZnO, SnO 2 , and Co 3 O 4 , have been actively explored as acetone sensor materials [4][5][6][7]. Among them, Co 3 O 4 , as a p-type material, is a promising candidate for acetone sensor materials due to its low and stable initial resistance and the high catalytic reactivity of Co 2+ [8][9][10][11]. In a study conducted by Vladimirova et al., for instance, the tolerance of Co 3 O 4 toward high humidity levels is discussed [12]. In experiments by Zhang et al., a Co 3 O 4 -based sensor with an optimized morphology of crossed nanosheets showed a response of 36.5 toward acetone gas of 1000 ppm. Zhang et al. also reported Co 3 O 4 nanowires being hybridized with hollow carbon spheres exhibiting a response of 3.71 toward acetone gas of 200 ppm [13]. Xu et al. synthesized metal-organic framework (MOF)-derived Co 3 O 4 /Fe 2 O 3 p-n hollow cubes to enhance acetone sensing properties (response of 3.27 to 100 ppm) [14]. However, the detection of acetone at a low concentration (sub-ppm or lower) still remains a challenge.
Doping metal oxide materials with fluorine, known to promote catalytic activity, has been considered as one of the effective routes to enhance sensing ability [15,16]. Barreca et al. employed Co(hfa) 2 ·TMEDA (hfa = 1,1,1,5,5,5-hexafluoro-2,4-pentanedionate; TMEDA = N,N,N ,N -tetramethylethylenediamine) as a precursor in chemical vapor deposition to produce F-doped Co 3 O 4 nanomaterials [15]. Nonetheless, such a doping method has a limitation in obtaining a fluorine concentration high enough for a significant sensitivity improvement. Recently, cobalt fluoride hydroxide (Co(OH) x F 2−x ), a material including F anions as one of its main elements, came into the spotlight for being applied to catalysts for water oxidation, hydrogen production, and overall water splitting, taking full advantage of the high catalytic activity of F anions in the materials [9,[17][18][19]. Nevertheless, a challenge to employ this material for gas sensor applications is its poor thermal stability: it could easily transform into Co 3 O 4 via fluorine evaporation during a high temperature process such as the aging process (200~450 • C) which is usually required for sensor fabrication [20].
In this study, we report, for the first time, Co(OH) x F 2−x nanosheets employed as a sensing material for acetone gas sensors. We synthesized Co(OH) x F 2−x nanosheets via a hydrothermal route on the template of SiO 2 spheres. Our synthesis method using the SiO 2 spheres as template helped achieve a high fluorine density and suppress fluorine evaporation to a considerable extent upon annealing at elevated temperatures. Sensors fabricated using the Co(OH) x F 2−x materials showcased a high response at a low acetone concentration (269% at 4.5 ppm) and ppb-level detection limit (~40 ppb) at an operation temperature of 200 • C. We also discussed the effects of F anions on the sensing properties of Co(OH) x F 2−x nanosheets.

Fabrication of Co(OH) x F 2−x Nanosheets
All the chemicals were used without any purification. Figure 1 shows the schematic diagram of the synthesis procedure used for the fabrication of Co(OH) x F 2−x nanosheets. SiO 2 spheres were synthesized using tetraethyl orthosilicate (TEOS) as described in another report [21]. Co(OH) x F 2−x nanosheets were synthesized as follows. CoCl 2 (0.3 mM), NH 4 F (0.3 mM), and urea (0.1 mM) were added into de-ionized (DI) water (30 mL). After stirring for 30 min, the SiO 2 spheres (60 mg) were added into the solution, followed by sonication for 30 min. Finally, the solution was transferred to a Teflon-lined steel autoclave with a capacity of 50 mL and heated at 120 • C for 12 h under autogenous pressure. The final solution was then cooled to room temperature. Powder, precipitated from the solution, was collected by filtering and was washed with water and ethanol. The powder was then dried at 80 • C overnight. It went through a subsequent aging process at 120 • C for 10 h for removal of the SiO 2 sacrificial template [22]. Chemical analysis using electron dispersive spectroscopy (EDS; EDS attached to the JEOL ARM 200F), nevertheless, disclosed that some SiO 2 had remained in the sample after the removal process (see Supplementary Materials Figure S1). For comparison, some materials were synthesized without templates via the same process.

Characterization
Scanning electron microscopy (SEM; JSM-7600F, JEOL, Tokyo, Japan) and transmission electron microscopy (TEM; JEM-ARM 200F, JEOL, Tokyo, Japan) were used to investigate the morphology of the SiO2 spheres and Co(OH)xF2−x nanosheets. The infrared spectra of materials were recorded using Fourier transform infrared spectroscopy (FTIR; IFS-66/S, Bruker, Billerica, MA, USA). For characterization of the phases of the materials, X-ray diffraction (XRD; D8 Advance diffractometer, Bruker, Billerica, MA, USA) was utilized. We used X-ray photoelectron spectroscopy (XPS; K-alpha, Thermo Scientific, Waltham, MA, USA) for the chemical analysis of the materials. The contour of the sensor device with Co(OH)xF2−x powders was scanned by a 3D profiler (Contour GT-X, Bruker, Billerica, MA, USA). The Brunauer-Emmett-Teller (BET; Micromeritics, Norcross, GA, USA) surface area of the Co(OH)xF2−x nanosheets was determined from nitrogen adsorption-desorption isotherm. The SEM equipment used in this study is located at the MEMS·Sensor Platform Center of SungKyunKwan University (SKKU), Suwon, Korea.

Fabrication of Gas Sensor and Gas-Sensing Measurements
The detailed fabrication process of interdigitated (IDT) electrodes and details of gas-sensing system are reported in our previous work [23]. Co(OH)xF2−x nanosheet powders were dissolved in DI water (0.5 wt. %) and drop-coated on the IDT electrodes on a hot plate (100 °C). After full evaporation of the solvent, these devices were used for gas-sensing measurements (see Figure S2). Besides, some of the sensors were annealed at 300, 350, 400, and 450 °C for 2 or 4 h to investigate the effects of postannealing conditions on the materials. For additional platinum (Pt) doping on the materials, a 1 nmthick Pt film was loaded on the Co(OH)xF2−x nanosheets of the devices using a magnetron sputtering system. A home-made gas-sensing system was described in our previous study. Gas-sensing characteristics were investigated for acetone (CH3COCH3), toluene (C6H5CH3), carbon monoxide (CO), hydrogen (H2), and nitrogen dioxide (NO2), which were diluted with dry air (2 L/min) using mass flow controllers (MFCs). The relative humidity (RH) of the laboratory varied from 30 to 63% and that of the sensing chamber was kept at 10% for all sensing tests. The operating temperature of gas sensors was controlled from 25 to 350 °C by Joule heating using a ceramic heater connected to a power supplier. The response of the test is defined as [(Rg − Ra)/Ra]*100 or [(Ra − Rg)/Rg]*100 for reactions with reducing gases and oxidizing gases, respectively, where Ra and Rg represent resistances in ambient and an analyte gas.

Characterization
Scanning electron microscopy (SEM; JSM-7600F, JEOL, Tokyo, Japan) and transmission electron microscopy (TEM; JEM-ARM 200F, JEOL, Tokyo, Japan) were used to investigate the morphology of the SiO 2 spheres and Co(OH) x F 2−x nanosheets. The infrared spectra of materials were recorded using Fourier transform infrared spectroscopy (FTIR; IFS-66/S, Bruker, Billerica, MA, USA). For characterization of the phases of the materials, X-ray diffraction (XRD; D8 Advance diffractometer, Bruker, Billerica, MA, USA) was utilized. We used X-ray photoelectron spectroscopy (XPS; K-alpha, Thermo Scientific, Waltham, MA, USA) for the chemical analysis of the materials. The contour of the sensor device with Co(OH) x F 2−x powders was scanned by a 3D profiler (Contour GT-X, Bruker, Billerica, MA, USA). The Brunauer-Emmett-Teller (BET; Micromeritics, Norcross, GA, USA) surface area of the Co(OH) x F 2−x nanosheets was determined from nitrogen adsorption-desorption isotherm. The SEM equipment used in this study is located at the MEMS·Sensor Platform Center of SungKyunKwan University (SKKU), Suwon, Korea.

Fabrication of Gas Sensor and Gas-Sensing Measurements
The detailed fabrication process of interdigitated (IDT) electrodes and details of gas-sensing system are reported in our previous work [23]. Co(OH) x F 2−x nanosheet powders were dissolved in DI water (0.5 wt. %) and drop-coated on the IDT electrodes on a hot plate (100 • C). After full evaporation of the solvent, these devices were used for gas-sensing measurements (see Figure S2). Besides, some of the sensors were annealed at 300, 350, 400, and 450 • C for 2 or 4 h to investigate the effects of post-annealing conditions on the materials. For additional platinum (Pt) doping on the materials, a 1 nm-thick Pt film was loaded on the Co(OH) x F 2−x nanosheets of the devices using a magnetron sputtering system. A home-made gas-sensing system was described in our previous study. Gas-sensing characteristics were investigated for acetone (CH 3 COCH 3 ), toluene (C 6 H 5 CH 3 ), carbon monoxide (CO), hydrogen (H 2 ), and nitrogen dioxide (NO 2 ), which were diluted with dry air (2 L/min) using mass flow controllers (MFCs). The relative humidity (RH) of the laboratory varied from 30 to 63% and that of the sensing chamber was kept at 10% for all sensing tests. The operating temperature of gas sensors was controlled from 25 to 350 • C by Joule heating using a ceramic heater connected to a power supplier. The response of the test is defined as [(R g − R a )/R a ]*100 or [(R a − R g )/R g ]*100 for reactions with reducing gases and oxidizing gases, respectively, where R a and R g represent resistances in ambient and an analyte gas.

Synthesis and Characterization of Co(OH) x F 2−x Nanosheets
The morphology of the templates of SiO 2 spheres and the material hydrothermally grown on the templates were characterized using SEM, as shown in Figure 2. Figure 2a,b indicates that the SiO 2 spheres were fabricated with a good dispersity, clean surface, and uniform size. The average diameter of the SiO 2 spheres is around 160 nm. Figure 2c,d shows the morphology of the material grown on the templates after removal of the templates. It consists of very thin, very curly sheets, which appear clustered around what seem to have been occupied by the SiO 2 spheres before the removal.

Synthesis and Characterization of Co(OH)xF2−x Nanosheets
The morphology of the templates of SiO2 spheres and the material hydrothermally grown on the templates were characterized using SEM, as shown in Figure 2. Figure 2a,b indicates that the SiO2 spheres were fabricated with a good dispersity, clean surface, and uniform size. The average diameter of the SiO2 spheres is around 160 nm. Figure 2c,d shows the morphology of the material grown on the templates after removal of the templates. It consists of very thin, very curly sheets, which appear clustered around what seem to have been occupied by the SiO2 spheres before the removal. We employed XPS to investigate the chemical state of the synthesized material. Figure 3a-c present XPS analysis results on the peaks of F 1s, Co 2p, and O 1s. The binding energy of the F 1s (~684.3 eV) peak is close with that reported in a previous study on Co(OH)F [9]. Co 2p subpeaks corresponding to inorganic Co 2+ species with unpaired 3d electrons and mainly appear at 781.1 ± 0.05 and 783.3 ± 0.05 (Co 2p1/2), and 796.9 ± 0.05 and 798.0 ± 0.05 eV (Co 2p3/2), which are somewhat higher than the binding energies of Co 3+ species [17,18,24]. Other subpeaks at 785.9 ± 0.05, and 803.2 ± 0.05 eV are ascribed to the Co 2+ shake-up satellite [18]. The O 1s peak is deconvoluted into two subpeaksa large one at 531.6 ± 0.05 eV corresponding to metal hydroxides and a small one at 533.5 ± 0.05 eV from organic contaminants [9]. A quantitative analysis of the XPS data discloses that the fluorine content in the sample is substantial (peak intensity ratio between F and Co peaks (F/Co): ~ 25%), much higher than that obtained via doping (F/Co: ~10%) [15]. Figure 3d shows an XRD pattern of the synthesized material, which exhibits very small peaks at positions close to those of the orthorhombic Co(OH)F phase. It is interesting to note the asymmetric bundle of peaks at 32~45°, an unusual feature reported also in other studies on Co(OH)2 materials [25,26]. These data of XRD and XPS suggest that the synthesized material could be amorphous or nanocrystalline Co(OH)xF2−x with a broad range of fluorine compositions. We employed XPS to investigate the chemical state of the synthesized material. Figure 3a-c present XPS analysis results on the peaks of F 1s, Co 2p, and O 1s. The binding energy of the F 1s (~684.3 eV) peak is close with that reported in a previous study on Co(OH)F [9]. Co 2p subpeaks corresponding to inorganic Co 2+ species with unpaired 3d electrons and mainly appear at 781.1 ± 0.05 and 783.3 ± 0.05 (Co 2p 1/2 ), and 796.9 ± 0.05 and 798.0 ± 0.05 eV (Co 2p 3/2 ), which are somewhat higher than the binding energies of Co 3+ species [17,18,24]. Other subpeaks at 785.9 ± 0.05, and 803.2 ± 0.05 eV are ascribed to the Co 2+ shake-up satellite [18]. The O 1s peak is deconvoluted into two subpeaks-a large one at 531.6 ± 0.05 eV corresponding to metal hydroxides and a small one at 533.5 ± 0.05 eV from organic contaminants [9]. A quantitative analysis of the XPS data discloses that the fluorine content in the sample is substantial (peak intensity ratio between F and Co peaks (F/Co):~25%), much higher than that obtained via doping (F/Co:~10%) [15]. Figure 3d shows an XRD pattern of the synthesized material, which exhibits very small peaks at positions close to those of the orthorhombic Co(OH)F phase. It is interesting to note the asymmetric bundle of peaks at 32~45 • , an unusual feature reported also in other studies on Co(OH) 2 materials [25,26]. These data of XRD and XPS suggest that the synthesized material could be amorphous or nanocrystalline Co(OH) x F 2−x with a broad range of fluorine compositions. In the TEM images shown in Figure 4a,b, the sheets appear highly electron-transparent, confirming that they are very thin, probably only a few atomic layers thick. The morphology of the nanosheets suggests that they have been densely grown around SiO2 spheres, the places of which now appear empty because of the removal of the spheres via aging (for example, see the box in Figure  4a). The BET analysis confirmed that the specific surface area of the material was around 193.38 m 2 g −1 , a value larger than or comparable with those of 2D materials such as graphene and MoS2 (see Figure S3 for the BET data) [27][28][29]. Figure 4c shows elemental mapping acquired from the EDS analysis for the area shown in Figure  4b, exhibiting that Co, O, and F are uniformly distributed throughout the sheets. Figure 4d,e shows high-resolution TEM (HRTEM) images of the Co(OH)xF2−x nanosheets, showing that the material is mainly amorphous, without any noticeable long-range order. Nevertheless, a close look into the image discloses that there are many areas (some marked with yellow circles) showing some shortrange ordering. Analysis on such short-range ordering areas indicates that the plane spacings are close to those of Co(OH)F, as shown in Figure 4e [19]. These results confirm that the synthesized nanosheets are amorphous Co(OH)xF2−x with some short-range ordering. In addition, it should be also noted from the EDS analysis that although the SiO2 spheres were supposedly removed as a whole through the removal process, the analysis indicates that there are some Si atoms uniformly distributed over the Co(OH)xF2−x nanosheets, implying that the removal process did not fully remove SiO2 but rather left some redistributed over the nanosheets. In the TEM images shown in Figure 4a,b, the sheets appear highly electron-transparent, confirming that they are very thin, probably only a few atomic layers thick. The morphology of the nanosheets suggests that they have been densely grown around SiO 2 spheres, the places of which now appear empty because of the removal of the spheres via aging (for example, see the box in Figure 4a). The BET analysis confirmed that the specific surface area of the material was around 193.38 m 2 g −1 , a value larger than or comparable with those of 2D materials such as graphene and MoS 2 (see Figure S3 for the BET data) [27][28][29]. Figure 4c shows elemental mapping acquired from the EDS analysis for the area shown in Figure 4b, exhibiting that Co, O, and F are uniformly distributed throughout the sheets. Figure 4d,e shows high-resolution TEM (HRTEM) images of the Co(OH) x F 2−x nanosheets, showing that the material is mainly amorphous, without any noticeable long-range order. Nevertheless, a close look into the image discloses that there are many areas (some marked with yellow circles) showing some short-range ordering. Analysis on such short-range ordering areas indicates that the plane spacings are close to those of Co(OH)F, as shown in Figure 4e [19]. These results confirm that the synthesized nanosheets are amorphous Co(OH) x F 2−x with some short-range ordering. In addition, it should be also noted from the EDS analysis that although the SiO 2 spheres were supposedly removed as a whole through the removal process, the analysis indicates that there are some Si atoms uniformly distributed over the Co(OH) x F 2−x nanosheets, implying that the removal process did not fully remove SiO 2 but rather left some redistributed over the nanosheets.

Post-Annealing Effects on Co(OH)xF2−x Nanosheets
For sensor applications, it is important to ensure the thermal stability of sensing materials since sensors are usually operated at elevated temperatures. Figure 5 exhibits SEM images of Co(OH)xF2−x nanosheets on a sensor device annealed at 350 °C for 2 h, showing that the nanosheet cluster structure remains stable. Figure 6a shows the fluorine content (F/Co*100) acquired from an XPS analysis for the samples of Co(OH)xF2−x nanosheets synthesized with SiO2 templates and without templates after annealing at various temperatures for 2 h. It is clear that the F ratio in the nanosheets decreases with the post-annealing temperature increasing, suggesting that F evaporates at high temperatures. It should be also noted that the sample grown on the template has retained fluorine more effectively at elevated temperatures than the sample without a template, implying that the SiO2 template has possibly played some role in retaining F.
Our FTIR analysis sheds some light on the possible influence of the SiO2 template on retaining fluorine, as shown in Figure 6b. In the FTIR data, the sample synthesized with SiO2 shows a strong peak at 1020 cm −1 , corresponding to bonding between Si and F, which does not decline upon annealing at elevated temperatures [30]. In contrast, the no-template sample does not show any peak at 1020 cm −1 but a large one at 650 cm −1 , which could be attributed to Co-O bonding [31]. These data indicate that the remnants of the SiO2 templates have possibly helped hold fluorine atoms (prone to evaporation) effectively by providing a strong bond with Si.
The TEM investigation further indicated no difference in the morphology of nanosheets before and after post-annealing at 350, 400, and 450 °C (see Figure S4). Figure 7 displays the high-resolution TEM images of the Co(OH)xF2−x nanosheets annealed at 350, 400, and 450 °C. A careful observation

Post-Annealing Effects on Co(OH) x F 2−x Nanosheets
For sensor applications, it is important to ensure the thermal stability of sensing materials since sensors are usually operated at elevated temperatures. Figure 5 exhibits SEM images of Co(OH) x F 2−x nanosheets on a sensor device annealed at 350 • C for 2 h, showing that the nanosheet cluster structure remains stable. Figure 6a shows the fluorine content (F/Co*100) acquired from an XPS analysis for the samples of Co(OH) x F 2−x nanosheets synthesized with SiO 2 templates and without templates after annealing at various temperatures for 2 h. It is clear that the F ratio in the nanosheets decreases with the post-annealing temperature increasing, suggesting that F evaporates at high temperatures. It should be also noted that the sample grown on the template has retained fluorine more effectively at elevated temperatures than the sample without a template, implying that the SiO 2 template has possibly played some role in retaining F.
Our FTIR analysis sheds some light on the possible influence of the SiO 2 template on retaining fluorine, as shown in Figure 6b. In the FTIR data, the sample synthesized with SiO 2 shows a strong peak at 1020 cm −1 , corresponding to bonding between Si and F, which does not decline upon annealing at elevated temperatures [30]. In contrast, the no-template sample does not show any peak at 1020 cm −1 but a large one at 650 cm −1 , which could be attributed to Co-O bonding [31]. These data indicate that the remnants of the SiO 2 templates have possibly helped hold fluorine atoms (prone to evaporation) effectively by providing a strong bond with Si.
Crystals 2020, 10, x FOR PEER REVIEW 7 of 16 of the images of Figure 7a,d, and g reveals that the nanosheets have become nanocrystalline. Some of the nanocrystals are Co(OH)F while others belong to Co3O4 according to d-spacing analysis (see the images of the center and right-side columns). These results indicate that a significant part of the material has been transformed into Co3O4 upon post-annealing.  The TEM investigation further indicated no difference in the morphology of nanosheets before and after post-annealing at 350, 400, and 450 • C (see Figure S4). Figure 7 displays the high-resolution TEM images of the Co(OH) x F 2−x nanosheets annealed at 350, 400, and 450 • C. A careful observation of the images of Figure 7a,d, and g reveals that the nanosheets have become nanocrystalline. Some of the nanocrystals are Co(OH)F while others belong to Co 3 O 4 according to d-spacing analysis (see the images of the center and right-side columns). These results indicate that a significant part of the material has been transformed into Co 3 O 4 upon post-annealing.

Gas-Sensing Characteristics of Co(OH) x F 2−x Nanosheet Sensors
We evaluated the gas-sensing ability of the sensor based on the Co(OH) x F 2−x nanosheets by examining the operating temperature, acetone gas concentration, gas species, and time elapse effects. Figure 8a shows gas responses at operating temperatures of 25,43,72,150,200, and 250 • C. The response graph exhibits a typical volcano shape with an optimized operating temperature of 200 • C. The response at 200 • C is around 269% toward acetone gas at a concentration of 4.5 ppm. For such a low acetone concentration, a response of 269% is quite high, compared with those of other reports on various cobalt-based materials for acetone gas sensor applications, as shown in Table 1.

Gas-Sensing Characteristics of Co(OH)xF2−x Nanosheet Sensors
We evaluated the gas-sensing ability of the sensor based on the Co(OH)xF2−x nanosheets by examining the operating temperature, acetone gas concentration, gas species, and time elapse effects. Figure 8a shows gas responses at operating temperatures of 25,43,72,150,200, and 250 °C. The response graph exhibits a typical volcano shape with an optimized operating temperature of 200 °C. The response at 200 °C is around 269% toward acetone gas at a concentration of 4.5 ppm. For such a low acetone concentration, a response of 269% is quite high, compared with those of other reports on various cobalt-based materials for acetone gas sensor applications, as shown in Table 1. Figure 8b shows dynamic sensing characteristics of the sensor upon the input of acetone gas at various concentrations at 200 °C. Calculated responses at each acetone concentration were also plotted (see Figure S5). The limit of detection of the sensor toward acetone gas is 0.04 ppm with a response of 17%. The sensor also shows a good selectivity toward other gases such as CH3COCH3 (2.2 ppm), C6H5CH3 (3.6 ppm), CO (0.8 ppm), H2 (0.8 ppm), and NO2 (1.1 ppm) at 200 °C, as shown in Figure 8c. Figure 8d shows responses to 4.5-ppm acetone gas as a function of time elapse for 20 days. The response of the Co(OH)xF2−x nanosheet-based sensor degrades from 269 at the first measurement to 106% after 6 days and stayed at a similar level during following measurements.   Figure S5). The limit of detection of the sensor toward acetone gas is 0.04 ppm with a response of 17%. The sensor also shows a good selectivity toward other gases such as CH 3 COCH 3 (2.2 ppm), C 6 H 5 CH 3 (3.6 ppm), CO (0.8 ppm), H 2 (0.8 ppm), and NO 2 (1.1 ppm) at 200 • C, as shown in Figure 8c. Figure 8d shows responses to 4.5-ppm acetone gas as a function of time elapse for 20 days. The response of the Co(OH) x F 2−x nanosheet-based sensor degrades from 269 at the first measurement to 106% after 6 days and stayed at a similar level during following measurements.

Gas-Sensing Characteristics of Annealed Co(OH) x F 2−x Nanosheet Sensors and Pt-Doping Effect
Next, we examined if the sensor functioned effectively at high temperatures (≤450 • C). We investigated dynamic sensing characteristics by changing the operating temperature between 200 and 300 • C toward acetone gas (4.5 ppm), as shown in Figure 9a,b. In the first cycle, the sensor shows a stable response and recovery at 200 • C, whereas it displays an unstable behavior at 250 and 300 • C [14]. We also noticed some color change of the material (from pink to dark brown) on the device during measurements at elevated temperatures (see the photographs in Figure 9a), an indication of the deterioration of the material, possibly due to fluorine evaporation [20].  We tested the sensors after post-annealing at 300 to 450 • C to further investigate the temperature effects on sensor performance. Figure 9c displays the gas responses of sensors annealed at 300, 350, 400, and 450 • C for 2 or 4 h toward 4.5 ppm acetone at 300 • C, which show much lower responses than those measured before post-annealing. It should be noted that the trend of the response, decreasing with the post-annealing temperature increasing, is similar to that of the fluorine concentration with temperature (see Figure 6a), hinting that the loss of fluorine possibly influenced the response to acetone gas. Increasing the post-annealing time from 2 to 4 h also deteriorates the gas response to 28 and 14% at 300 and 350 • C, respectively, as shown in Figure 9c.
Using the sample annealed at 350 • C for 2 h, we further performed a dynamic test, finding that although the response is low, the sensor works reliably with a full recovery for three successive gas on/off cycles at 300 • C, as shown in Figure 9d. Furthermore, additional annealing at 300 • C for 2 h does not decrease the response, as shown in Figure 9e, suggesting that once the sensing material is annealed at 350 • C, further annealing at a lower temperature would not affect the response much. The result of a long-term stability test shown in Figure 9d also exhibits stable operations with a similar response level for around 3 months. Thus, the annealed sensor at 350 • C shows stable responses although the response level is significantly lower, as compared to the performance of the non-annealed sensor.
It is well known that noble metal dopants such as Pt, Pd, and Au could enhance gas sensor responses because of their strong catalytic properties [43]. In an effort to recover the deteriorated gas response of the 350 • C sample, we loaded Pt (1 nm) on the sensor and conducted measurements for performance ( Figure 10). Figure 10a clearly shows improvements in responses at all working temperatures after loading Pt to the device. The sensor also demonstrated a good selectivity toward 4.5 ppm of acetone gas, compared to 4.5 ppm of other gases such as C 6 H 5 CH 3 , CO, H 2 , and NO 2 , as shown in Figure 10b. Figure 10c presents a good sensitivity of the sensor toward acetone at a concentration of 1.1, 2.2, 3.4, and 4.5 ppm. Dynamic sensing characteristics in Figure 10d displays responses of stepwise increase upon the injection of acetone with the concentration increasing from 1.1 to 4.5 ppm, showing a sensitive response at each step and full recovery upon stopping the injection of acetone gas. These data confirm that the Pt loading on the sensor device has enhanced the sensing response, enabling a sensor with a high sensitivity and good stability.

Gas-Sensing Mechanisms
In this study, we synthesized Co(OH)xF2−x nanosheets, a material successfully incorporating a high concentration of fluorine and keeping it to some extent through annealing at elevated temperatures (owing to the support from the bond with Si of the remnant SiO2). It is clear from our sensor data that the presence of fluorine of a high concentration helped achieve a high gas response. In comparison, a sensor based on a material fabricated without adding NH4F in the synthesis process

Gas-Sensing Mechanisms
In this study, we synthesized Co(OH) x F 2−x nanosheets, a material successfully incorporating a high concentration of fluorine and keeping it to some extent through annealing at elevated temperatures (owing to the support from the bond with Si of the remnant SiO 2 ). It is clear from our sensor data that the presence of fluorine of a high concentration helped achieve a high gas response. In comparison, a sensor based on a material fabricated without adding NH 4 F in the synthesis process showed a low response (92%) (see Figure S6), highlighting the effects of F.
It should be noted that incorporated F, with a high electronegativity, could enhance the Lewis acidity of the material, which could facilitate the adsorption of gas molecules [15,16,19]. For the present study, the main reactions occurring at the surface of the Co(OH) x F 2−x nanosheets can be expressed as shown below [1,44,45]. The high Lewis acidity of the nanosheets with a high concentration of F has possibly helped the material accept electrons easily and promoted the reaction of Equation (4).
O 2 − (ads) + e − → 2O − (ads) (100 • C < T < 300 • C) In addition, the morphology of the Co(OH) x F 2−x nanosheets with a high specific surface area (193.38 m 2 /g), brought about by the assistance of the SiO 2 template, has possibly contributed to the high sensing performance. For example, the high response of the Co(OH) x F 2−x nanosheets (269%) toward 4.5 ppm of acetone gas at 200 • C contrasts with the low response (147%) of the sensor based on Co(OH) x F 2−x fabricated without SiO 2 templates, which has a bulk-type morphology (see Figure S7). Other studies also show the strong influence of morphology on the sensitivity of sensors, attributing to the increase in oxygen-adsorption sites promoting the reaction of Equation (1) [46,47]. Thus, the adoption of an SiO 2 template in the synthesis process played a beneficial role in two ways, maintaining a high F concentration for a high catalytic property and producing a structure with a very high specific surface area.

Conclusions
We synthesized Co(OH) x F 2−x nanosheets by adopting SiO 2 sphere templates and fabricated sensor devices based on the materials. Our material characterization using such analytical tools as XPS, FTIR, and TEM confirms that the material is made up of Co(OH) x F 2−x nanosheets containing a high concentration of F with an amorphous microstructure with some short-range ordering. The sensor based on the Co(OH) x F 2−x nanosheets exhibited a very high response of 269% toward 4.5 ppm of acetone gas at an operation temperature of 200 • C, and a minimum detection limit of 40 ppb was obtained. Although the sensor showed some indications of degradation (possibly due to F evaporation) at a temperature higher than 300 • C, the Pt doping was found to help recover gas response to some extent (77% toward 4.5 ppm of acetone). We discussed the strong catalytic property of F in a high concentration as well as the high specific surface area due to the morphology of densely clustered nanosheets being responsible for the excellent sensor performance. This study highlights the possibility of further applications of a new material group of metal fluoride hydroxides to gas sensors owing to their distinct advantages compared to conventional metal oxide materials.  Figure S1. EDS spectrum of Co(OH) x F 2−x nanosheets with atomic percentages of each element. Figure S2. Profile image of the interdigitated sensor device with coated Co(OH) x F 2−x nanosheets powders. Figure S3. N2 adsorption-desorption isotherm with calculated BET surface area (193.38 m2/g). Figure S4. TEM bright-field images of the Co(OH) x F 2−x nanosheets (a) before and after annealing at (b) 350, (c) 400, and (d)