Reversible Redox Property of Co(III) in Amorphous Co-Doped SiO2/γ-Al2O3 Layered Composites

This paper reports on a unique reversible reducing and oxidizing (redox) property of Co(III) in Co-doped amorphous SiO2/γ-Al2O3 composites. The Fenton reaction during the H2O2-catalyzed sol–gel synthesis utilized in this study lead to the partial formation of Co(III) in addition to Co(II) within the composites. High-resolution transmission electron microscopy (HRTEM) and high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) analyses for the composite powder sample with a composition of Al:Si:Co = 85:10:5 showed the amorphous state of the Co-doped SiO2 that modified γ-Al2O3 nanocrystalline surfaces. In situ X-ray absorption fine structure (XAFS) spectroscopic analysis suggested reversible redox reactions of Co species in the composite powder sample during heat-treatment under H2 at 500 °C followed by subsequent cooling to RT under Ar. Further analyses by in situ IR spectroscopy combined with cyclic temperature programmed reduction/desorption (TPR/TPD) measurements and X-ray photoelectron spectroscopic (XPS) analysis revealed that the alternating Co(III)/(II) redox reactions were associated with OH formation (hydrogenation)-deformation (dehydrogenation) of the amorphous aluminosilicate matrix formed in situ at the SiO2/γ-Al2O3 hetero interface, and the redox reactions were governed by the H2 partial pressure at 250–500 °C. As a result, a supported mesoporous γ-Al2O3/Co-doped amorphous SiO2/mesoporous γ-Al2O3 three-layered composite membrane exhibited an H2-triggered chemical valve property: mesopores under H2 flow (open) and micropores under He flow (closure) at 300–500 °C.


Introduction
Microporous amorphous silica (SiO 2 )-based membranes exhibit excellent hydrogen (H 2 )-selectivity and higher thermal stability compared to organic polymer membranes [1][2][3][4][5]. The SiO 2 membrane component is also attractive for developing novel mixed matrix membranes combined with a polymer membrane component such as polysulfone [6], which shows better membrane performance with improved mechanical properties. Moreover, SiO 2 membranes can be further modified as SiO 2 -based organic-inorganic hybrid membranes for other applications such as liquid-phase separation [7].
It has been recognized that high H 2 -selectivity can be achieved by the molecular sieving property of amorphous SiO 2 with micro-pores of approximately 0.3 nm in size [3][4][5]. The microporosity can be formed in situ and is related to the highly branched fractal dimensions having a large amount of hydroxyl (Si-O-H) groups [8]. Accordingly, microporous SiO 2 membranes are moisture sensitive [9] and lack hydrothermal stability [10].
Recently, it has been reported that doping amorphous SiO 2 with metal (M) cations (M = Zr [11], Ni [12], Co [13][14][15]) improved the thermal or hydrothermal stability of sol-gel-derived SiO 2 membranes. Among the dopants, Co was found to be effective at enhancing H 2 permeance at 500 • C [16]. Nanostructure observation of the Co-doped SiO 2 -based membranes revealed that CoO and Co 3 O 4 nanoparticles of approximately 5-20 nm in size were distributed within the SiO 2 matrix [16]. More recently, SiO 2 -based membranes synthesized from Co-modified ethoxy group-functionalized polysiloxane have been reported to show a unique reversible gas molecular sieving property above 300 • C [17]. The reversible gas molecular sieving property was suggested to be governed by alternating volume shrinkage/expansion due to the reversible change in the reducing/oxidizing (redox) state of the Co oxide particles (Co(OH) 2 and CoO/Co 3 O 4 ) finely distributed within the polymer-derived SiO 2 matrix [17]. These results reveal that the divalent(II) or trivalent(III) Co cations that remained under H 2 atmosphere contributed to the enhanced H 2 permeance at 300-500 • C.
Prior to the composite sample syntheses, the Co/Si ratio of the precursor-derived Co-doped SiO 2 was examined as follows: The Co-doped SiO 2 sols with different Co/Si ratios were heated with stirring to 80 • C and kept at this temperature overnight to give dried Co-doped SiO 2 gel powder samples. The dried gel powder samples were placed on a quartz boat in a quartz tube furnace (Model ARF-50KC, Asahi Rika Seisakusho Co. Ltd., Chiba, Japan), and heat-treated in air at 600 • C for 3 h with a heating/cooling rate of 20 • C·min −1 to afford Co-doped SiO 2 powder samples. Then, elemental analyses were performed for Si and Co (ICP spectrometry, Model ICPS-7510, Shimazu Co., Kyoto, Japan), and carbon (C, non-dispersive infrared method, Model CS844, LECO Co., St Joseph, MI, USA). The residual carbon in the powder samples were negligible (<0.1%), and as shown in Table S1, we confirmed that the Co/Si ratios in the range of 1/8 to 1/2 were well controlled through the present sol-gel route using H 2 O 2 .
Boehmite (γ-AlOOH) coating solution was prepared using the previously reported procedure [28] and spin-coated on a SiO 2 substrate (10 mm × 10 mm × 3 mm, Meijo science Co. Inc., Nagoya, Japan) at 3000 rpm using a spin coater (Model, MS-A100, Mikasa Co. Ltd., Tokyo, Japan). The γ-AlOOH-coated SiO 2 substrate was placed on an Al 2 O 3 plate and loaded in an electric furnace (Model FUW220PA, Advantec Toyo Kaisha Ltd., Tokyo, Japan) then heat-treated in air at 600 • C for 3 h with a heating/cooling rate of 100 • C·h −1 to afford the formation of a thin layer of γ-Al 2 O 3 over the SiO 2 substrate. Subsequently, Co-doped SiO 2 sol with Co/Si = 1/8 was spin-coated on the γ-Al 2 O 3 layer on the SiO 2 substrate by using the same manner as described above. The spin-coated sample substrate was placed on a quartz boat and loaded in a quartz tube furnace (Model ARF-50KC, Asahi Rika Seisakusho Co. Ltd., Chiba, Japan) then heat-treated in air at 600 • C for 20 h with a heating/cooling rate of 20 • C·min −1 . The sample was labelled CoSiOAlcoat.
2.1.2. Co-Doped SiO 2 /γ-Al 2 O 3 Composite Powder Sample γ-Al 2 O 3 powder was prepared from γ-AlOOH sol using a previously reported procedure [28]. γ-Al 2 O 3 powder was mixed with Co-doped SiO 2 sol with Co/Si = 1/2 at the Al:Si:Co ratio of 85:10:5. The resulting mixture was heated with stirring to 80 • C and kept at this temperature overnight to give a dried Co-doped SiO 2 gel/γ-Al 2 O 3 powder sample. The dried powders were placed on a quartz boat and set in a quartz tube furnace (Model ARF-50KC, Asahi Rika Seisakusho Co. Ltd., Chiba, Japan) then heat-treated in air at 600 • C for 20 h with a heating/cooling rate of 20 • C·min −1 . The sample was labelled CoSiOAlpow.

Homogeneous Co-Doped Aluminosilicate Powder Sample
To study the redox behavior of the Co cation in an amorphous aluminosilicate matrix, Co-doped aluminosilicate sol was prepared according to an atomic ratio of Al (in Al(NO 3 4 and H 2 O 2 (2.3 mL, 30% aq) were added and maintained at RT with stirring for 1 h, then heated at 60 • C for 2 h to give a Co-doped aluminosilicate sol. Then, the resulting sol was dried at 80 • C overnight to give the dried gel powder sample. The samples were placed on a quartz boat in a quartz tube furnace (Model ARF-50KC, Asahi Rika Seisakusho Co. Ltd., Chiba, Japan) and heat-treated in air at 600 • C for 3 h with a heating/cooling rate of 20 • C·min −1 . This sample was labelled CoSiOAlpow2.
As reference data, the textural properties were characterized for the composite powder samples. The surface morphology and the pore size distribution (PSD) of the CoSiOAlpow and CoSiOAlpow2 samples are shown in Figures S1 and S2, respectively. The CoSiOAlpow sample kept the morphology of the γ-Al 2 O 3 polycrystallites ( Figure S1a), while the CoSiOAlpow2 sample exhibited a featureless glassy surface ( Figure S1b). The N 2 adsorption and desorption isotherms of the CoSiOAlpow sample exhibited Type-IV isotherms according to the IUPAC classifications [29] and formed a hysteresis loop indicating the presence of mesoporosity, while those of CoSiOAlpow2 sample presented Type-II isotherms [29] without a distinct hysteresis loop ( Figure S2a). The PSD curve characterized by the BJH method [30] for each powder sample was consistent with the N 2 adsorption-desorption behavior: The CoSiOAlpow sample exhibited a PSD curve with a dominant peak at 8 nm, while the CoSiOAlpow2 sample showed a broad and weak peak ranging from 20 to 100 nm ( Figure S2b). The resulting Brunauer-Emmett-Teller (BET) surface areas of the CoSiOAlpow and CoSiOAlpow2 samples were measured to be 249 m 2 g -1 and 94 m 2 g -1 , respectively.
A mesoporous γ-Al 2 O 3 layer was formed on an outer surface of a macroporous α-Al 2 O 3 support (tubular-type, 6 mm outer diameter, 2 mm thickness, and 60 mm length, porosity 40%, Noritake Co., Ltd., Aichi, Japan) according to a published procedure [28]. The mesoporous γ-Al 2 O 3 -modified tubular support was dipped into Co-doped SiO 2 sol with Co/Si = 1/4 at 10 mm·s −1 , kept in the sol for 30 s, then pulled out at 1 mm·s −1 before heat treatment at 600 • C for 20 h under the same manner as that for the CoSiOAlpow sample synthesis. This dip-coating-heat treatment sequence was repeated once to form a Co-doped SiO 2 thin film on the supported mesoporous γ-Al 2 O 3 layer. Then, the Co-doped SiO 2 thin film was coated with γ-AlOOH sol (diluted as 0.5 mol·L −1 ) and heat-treated in air at 600 • C for 3 h by using the same manner as described above to give γ-Al 2 O 3 /Co-doped SiO 2 /γ-Al 2 O 3 three-layered composite membrane. The sample was labelled CoSiOAlmemb.

Characterization Techniques
X-ray diffraction (XRD) measurements were performed on powder samples (Model X'pert Pro 1, Philips Ltd., Amsterdam, The Nederlands).
X-ray photoelectron spectroscopic (XPS) analysis was performed on the CoSiOAlpow2 sample with an Al Kα X-ray source operated at 14 kV and 14 mA (Model PHI-5000, Ulvac-phi, Kanagawa, Japan). An alignment on the C 1s peak (at 284.8 eV) was performed before survey scans. To investigate the reversible redox property of the Co cation in the aluminosilicate, the CoSiOAlpow2 sample was heat-treated at 350 • C for 1 h with a heating/cooling rate of 5 • C·min −1 under H 2 flow. Then, under Ar flow, the sample was heat-treated at 350 • C for 10 h with a heating/cooling rate of 5 • C·min −1 . The heat treatments were cyclically performed under flowing of H 2 -Ar-H 2 or H 2 -Ar-H 2 -Ar using a catalyst analyzer (BELCAT-A, MicrotracBEL Corp., Osaka, Japan). Then, Co 2p spectrum was recorded for the heat-treated sample.
To investigate the distribution of Co cations within the CoSiOAlcoat sample, the depth-profiling XPS analysis was carried out by combining Ar ion gun etching cycles at 3 kV. The etching rate was calibrated as 3.73 nm min −1 by measuring that of the SiO 2 substrate.
Surface morphology of CoSiOAlpow and CoSiOAlpow2 samples and the cross-section of the CoSiOAlmemb sample were observed by a scanning electron microscope (SEM, model JSM-6010LA, JEOL Ltd., Tokyo, Japan) operated at a voltage of 15 kV.
The BET surface area of the CoSiOAlpow and CoSiOAlpow2 samples was evaluated by measuring N 2 adsorption and desorption isotherms at −196 • C under relative pressure ranging from 0 to 0.99 (Model Belsorp Max, BEL Japan Inc., Osaka, Japan).
Nanostructure of the CoSiOAlpow sample was characterized by transmission electron microscopy (TEM) using a high-angle annular dark-field-scanning transmission electron microscope (HAADF-STEM, Model JEM-ARM200F, JEOL Ltd., Tokyo, Japan, operated at an accelerating voltage of 200 kV). The distribution of the constituent elements of Si, Al, Co, and O was measured and analyzed by the energy dispersive X-ray spectrometer (EDS, Model JED-2300, JEOL Ltd., Tokyo, Japan) mounted on a JEM-ARM200F.
Local structure of γ-Al 2 O 3 powder sample was studied by 27 Al solid-state magic angle spinning nuclear magnetic resonance (MAS NMR) spectroscopic analysis with a 600 MHz NMR spectrometer (Model JNM-ECA600II, JEOL Ltd., Tokyo, Japan) operating at a static magnetic field of 14.01 T (155.4 MHz). Single gas permeances of the CoSiOAlmemb sample were evaluated for He and H 2 by the constant-volume manometric method [31]. The evaluations were performed at 100-500 • C according to the procedure as reported elsewhere [32,33]. The pressure on the gas feed side in this study was maintained at 120 kPa. On the gas permeate side, the gas line and the buffer tank with the total volume of 3.11 × 10 −4 ·m 3 were completely evacuated by using an oil rotary vacuum pump. After the evacuation was terminated, the pressure increase rate at the inside of the buffer tank was measured three times. The average of the three measurements was used for calculation of the permeance of gas-i (Q i ).

In Situ Characterizations
The hydrogenation/dehydrogenation reactions were monitored in situ for the CoSiOAlpow sample by diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS, Model Spectrum 100, Perkin Elmer, Waltham, MA, USA). The Co-doped SiO 2 /γ-Al 2 O 3 composite powder sample was loaded in a diffuse reflection cell (Model STJ900C Diffuse IR Heat Cham, S.T. JAPAN Inc., Tokyo, Japan), and IR spectrum was recorded after each treatment: • heat-treatment to remove adsorbed water at 500 • C for 8 h under a flow of Ar (4 mL·min −1 ). • subsequent heat treatment at 500 • C for 0.5 h under a flow of H 2 (4 mL·min −1 ). • final heat treatment at 500 • C for 8 h under a flow of Ar (4 mL·min −1 ).
Temperature-programmed reduction (TPR) and desorption (TPD) experiments were performed on the CoSiOAlpow sample using a catalyst analyzer (BELCAT-A, MicrotracBEL Corp., Osaka, Japan) fixed with a quadrupole mass spectrometer (BEL Mass, MicrotracBEL Corp., Osaka, Japan). The TPR profile was measured during heating from 100 • C to 500 • C (5 • C·min −1 ) under a flow of 10% H 2 /Ar (50 mL·min −1 ). Then, the sample powder was maintained under pure-H 2 flow (50 mL·min −1 ) for 1 h and cooled down to room temperature. The subsequent TPD measurement was carried out by heating up to 500 • C (5 • C·min −1 ) under He flow (50 mL·min −1 ). This cyclic TPR/TPD measurement was repeated 3 times.
In situ X-ray adsorption fine structure (XAFS) spectroscopic analyses were conducted at the BL5S1 of Aichi Synchrotron Radiation Center (AichiSR, Aichi Science & Technology Foundation, Seto, Aichi, Japan). The experimental set up is shown in Figure 1. The absorption spectrum can be obtained based on Equation (4), where µ and t are absorption coefficient and sample thickness, respectively.
Materials 2020, 13, x 5 of 15 Local structure of γ-Al2O3 powder sample was studied by 27 Al solid-state magic angle spinning nuclear magnetic resonance (MAS NMR) spectroscopic analysis with a 600 MHz NMR spectrometer (Model JNM-ECA600II, JEOL Ltd., Tokyo, Japan) operating at a static magnetic field of 14.01 T (155.4 MHz). Single gas permeances of the CoSiOAlmemb sample were evaluated for He and H2 by the constant-volume manometric method [31]. The evaluations were performed at 100-500 °C according to the procedure as reported elsewhere [32,33]. The pressure on the gas feed side in this study was maintained at 120 kPa. On the gas permeate side, the gas line and the buffer tank with the total volume of 3.11 × 10 −4 ·m 3 were completely evacuated by using an oil rotary vacuum pump. After the evacuation was terminated, the pressure increase rate at the inside of the buffer tank was measured three times. The average of the three measurements was used for calculation of the permeance of gas-i (Qi).

In Situ Characterizations
The hydrogenation/dehydrogenation reactions were monitored in situ for the CoSiOAlpow sample by diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS, Model Spectrum 100, Perkin Elmer, Waltham, MA, USA). The Co-doped SiO2/γ-Al2O3 composite powder sample was loaded in a diffuse reflection cell (Model STJ900C Diffuse IR Heat Cham, S.T. JAPAN Inc., Tokyo, Japan), and IR spectrum was recorded after each treatment: • heat-treatment to remove adsorbed water at 500 °C for 8 h under a flow of Ar (4 mL·min −1 ). • subsequent heat treatment at 500 °C for 0.5 h under a flow of H2 (4 mL·min −1 ). • final heat treatment at 500 °C for 8 h under a flow of Ar (4 mL·min −1 ). Temperature-programmed reduction (TPR) and desorption (TPD) experiments were performed on the CoSiOAlpow sample using a catalyst analyzer (BELCAT-A, MicrotracBEL Corp., Osaka, Japan) fixed with a quadrupole mass spectrometer (BEL Mass, MicrotracBEL Corp., Osaka, Japan). The TPR profile was measured during heating from 100 °C to 500 °C (5 °C·min −1 ) under a flow of 10% H2/Ar (50 mL·min −1 ). Then, the sample powder was maintained under pure-H2 flow (50 mL·min −1 ) for 1 h and cooled down to room temperature. The subsequent TPD measurement was carried out by heating up to 500 °C (5 °C·min −1 ) under He flow (50 mL·min −1 ). This cyclic TPR/TPD measurement was repeated 3 times.
In situ X-ray adsorption fine structure (XAFS) spectroscopic analyses were conducted at the BL5S1 of Aichi Synchrotron Radiation Center (AichiSR, Aichi Science & Technology Foundation, Seto, Aichi, Japan). The experimental set up is shown in Figure 1a. The absorption spectrum can be obtained based on Equation (4), where μ and t are absorption coefficient and sample thickness, respectively. The CoSiOAlpow sample (3.1 mg) was diluted with hexagonal boron nitride powder (100 mg, provided by the AichiSR, Seto, Aichi, Japan) using an agate mortar and pestle, then it was uniaxially pressed into a disk with a diameter of 7 mm. The disk sample was loaded in a quartz cell equipped The CoSiOAlpow sample (3.1 mg) was diluted with hexagonal boron nitride powder (100 mg, provided by the AichiSR, Seto, Aichi, Japan) using an agate mortar and pestle, then it was uniaxially pressed into a disk with a diameter of 7 mm. The disk sample was loaded in a quartz cell equipped with an electric heater and gas lines (Model 2000-1431, Makuharikagaku Garasu Seisakusho Inc., Chiba, Chiba, Japan) and maintained for 30 min under Ar flow. Then, under H 2 flow, the disk sample was heat-treated at 500 • C for 30 min with a heating rate of 8 • C·min −1 . Subsequently, under Ar flow, the furnace was cooled down to RT with a cooling rate of 8 • C·min −1 then maintained at RT for additional 1 h. During this treatment, the spectrum was recorded stepwise at each measurement as follows: (1) at RT under initial Ar flow, (2) after H 2 treatment at 500 • C for 30 min, (3) at RT after being cooled down from 500 • C under Ar flow, and (4) at RT after being maintained under Ar flow for 1 h. In this study, the measurements were also performed on Co 3 O 4 (4N, Kojundo Chemical Laboratory Co., Ltd., Saitama, Japan), CoO, and Co-foil (stored at AichiSR) at RT as reference samples. Then, to investigate the reversible redox property of the CoSiOAlpow sample, X-ray absorption near edge structure (XANES) spectroscopic analyses were performed using software (ATHENA, ver. 0.9.24/Demeter [34]).

Distribution of Co Cations within the Co-Doped SiO 2 /γ-Al 2 O 3 Layered Composite Sample
To investigate the chemical composition at the hetero interface between the Co-doped SiO 2 and γ-Al 2 O 3 , a Co-doped SiO 2 thin layer was formed on a mesoporous γ-Al 2 O 3 layer supported on a SiO 2 substrate. This sample labelled CoSiOAlcoat was deposited by spin-coating of the Co-doped SiO 2 sol with Co/Si = 1/8 followed by heat treatment in air at 600 • C for 20 h as detailed in the experimental section. Figure 2 presents the results of depth-profiling XPS analysis for the constituent elements of Si, Co, and Al. In this study, the measurements were also performed on Co3O4 (4N, Kojundo Chemical Laboratory Co., Ltd., Saitama, Japan), CoO, and Co-foil (stored at AichiSR) at RT as reference samples. Then, to investigate the reversible redox property of the CoSiOAlpow sample, X-ray absorption near edge structure (XANES) spectroscopic analyses were performed using software (ATHENA, ver. 0.9.24/Demeter [34]).

Distribution of Co Cations within the Co-Doped SiO2/ γ-Al2O3 Layered Composite Sample
To investigate the chemical composition at the hetero interface between the Co-doped SiO2 and γ-Al2O3, a Co-doped SiO2 thin layer was formed on a mesoporous γ-Al2O3 layer supported on a SiO2 substrate. This sample labelled CoSiOAlcoat was deposited by spin-coating of the Co-doped SiO2 sol with Co/Si = 1/8 followed by heat treatment in air at 600 °C for 20 h as detailed in the experimental section. Figure 2 presents the results of depth-profiling XPS analysis for the constituent elements of Si, Co, and Al. The thickness of the Co-doped SiO2 top layer was approximately 37 nm, estimated based on the Ar sputtering rate determined for the SiO2 substrate. Then, almost all of the mesopore channels of the γ-Al2O3 layer with a thickness of about 85 nm were infiltrated by the Co-doped SiO2. During the spincoating process, Co-doped SiO2 sol solution easily penetrated into the mesoporous γ-Al2O3 layer. However, the depth profile of Co revealed that the Co/Si ratio of the top layer was approximately 1/37, much lower than the nominal composition of Co-doped SiO2 coating sol (1/8). On the other hand, around the Co-doped SiO2/γ-Al2O3 hetero interface (around 50 nm in depth), the Co content increased to approximately 3 at.%, then, within the mesoporous γ-Al2O3 layer at 60-100 nm in depth, the Co content was shown to be constant at a value of ca. 2 at.% and the resulting Al:Si:Co ratio was 85:10:5.
The high Co concentration at the interface of the γ-Al2O3 layer was consistent with the previous report on the synthesis of Co-doped SiO2 membrane on a porous Al2O3 support by Diniz Da Costa et al. [14]. The observed high concentration of Co around the hetero interface revealed preferential diffusion of Co cations from the SiO2 matrix to the γ-Al2O3 surface. As shown in Figure S3, 27 Al solidstate MAS-NMR spectroscopic analysis reveals that the γ-Al2O3 prepared in this study was composed of AlO4 and AlO6 units with the AlO6/AlO4 unit ratio of about 2.4, which is consistent with reported data [35]. Moreover, our previous study on mesoporous γ-Al2O3 revealed that a considerable amount The thickness of the Co-doped SiO 2 top layer was approximately 37 nm, estimated based on the Ar sputtering rate determined for the SiO 2 substrate. Then, almost all of the mesopore channels of the γ-Al 2 O 3 layer with a thickness of about 85 nm were infiltrated by the Co-doped SiO 2 . During the spin-coating process, Co-doped SiO 2 sol solution easily penetrated into the mesoporous γ-Al 2 O 3 layer. However, the depth profile of Co revealed that the Co/Si ratio of the top layer was approximately 1/37, much lower than the nominal composition of Co-doped SiO 2 coating sol (1/8). On the other hand, around the Co-doped SiO 2 /γ-Al 2 O 3 hetero interface (around 50 nm in depth), the Co content increased to approximately 3 at.%, then, within the mesoporous γ-Al 2 O 3 layer at 60-100 nm in depth, the Co content was shown to be constant at a value of ca. 2 at.% and the resulting Al:Si:Co ratio was 85:10:5.
The high Co concentration at the interface of the γ-Al 2 O 3 layer was consistent with the previous report on the synthesis of Co-doped SiO 2 membrane on a porous Al 2 O 3 support by Diniz Da Costa et al. [14]. The observed high concentration of Co around the hetero interface revealed preferential diffusion of Co cations from the SiO 2 matrix to the γ-Al 2 O 3 surface. As shown in Figure S3, 27 Al solid-state MAS-NMR spectroscopic analysis reveals that the γ-Al 2 O 3 prepared in this study was composed of AlO 4 and AlO 6 units with the AlO 6 /AlO 4 unit ratio of about 2.4, which is consistent with reported data [35]. Moreover, our previous study on mesoporous γ-Al 2 O 3 revealed that a considerable amount of the Al-OH group existed at the γ-Al 2 O 3 surface, which led to the highly hydrophilic property of γ-Al 2 O 3 [28]. Accordingly, the driving force for the Co diffusion is due to the charge compensation for the AlO 4 site within the amorphous aluminosilicate formed in situ at the hetero interface, m[(SiO 4 )(Al − O 4 )]Co m+ (m = 2 or 3). Based on these results, the Al:Si:Co ratio of 85:10:5 was selected as an equilibrium chemical composition to investigate the redox property of Co cations at the SiO 2 /γ-Al 2 O 3 hetero interface.

Properties of Co-Doped SiO 2 /γ-Al 2 O 3 Composites
To investigate the later redox properties of Co cations at the SiO 2 /γ-Al 2 O 3 hetero interface, we prepared powders labelled CoSiOAlpow with the equilibrium chemical composition (Al:Si:Co = 85:10:5). We first investigated the structural and microstructural changes of the material after H 2 treatment (the reductive condition) at 500 • C. As shown in Figure 3, the X-ray diffraction pattern of the CoSiOAlpow sample is composed of XRD peaks of γ-Al 2 O 3 , indicating that the Co-doped SiO 2 layer is X-ray amorphous. Then, after the H 2 treatment at 500 • C, it retains the X-ray amorphous state of the Co-doped SiO 2 phase.
Materials 2020, 13, x 7 of 15 of the Al-OH group existed at the γ-Al2O3 surface, which led to the highly hydrophilic property of γ-Al2O3 [28]. Accordingly, the driving force for the Co diffusion is due to the charge compensation for the AlO4 site within the amorphous aluminosilicate formed in situ at the hetero interface, m[(SiO4)(Al − O4)]Co m+ (m = 2 or 3). Based on these results, the Al:Si:Co ratio of 85:10:5 was selected as an equilibrium chemical composition to investigate the redox property of Co cations at the SiO2/γ-Al2O3 hetero interface.

Properties of Co-Doped SiO2/γ-Al2O3 Composites
To investigate the later redox properties of Co cations at the SiO2/γ-Al2O3 hetero interface, we prepared powders labelled CoSiOAlpow with the equilibrium chemical composition (Al:Si:Co = 85:10:5). We first investigated the structural and microstructural changes of the material after H2 treatment (the reductive condition) at 500 °C. As shown in Figure 3, the X-ray diffraction pattern of the CoSiOAlpow sample is composed of XRD peaks of γ-Al2O3, indicating that the Co-doped SiO2 layer is X-ray amorphous. Then, after the H2 treatment at 500 °C, it retains the X-ray amorphous state of the Co-doped SiO2 phase.   of the Al-OH group existed at the γ-Al2O3 surface, which led to the highly hydrophilic property of γ-Al2O3 [28]. Accordingly, the driving force for the Co diffusion is due to the charge compensation for the AlO4 site within the amorphous aluminosilicate formed in situ at the hetero interface, m[(SiO4)(Al − O4)]Co m+ (m = 2 or 3). Based on these results, the Al:Si:Co ratio of 85:10:5 was selected as an equilibrium chemical composition to investigate the redox property of Co cations at the SiO2/γ-Al2O3 hetero interface.

Properties of Co-Doped SiO2/γ-Al2O3 Composites
To investigate the later redox properties of Co cations at the SiO2/γ-Al2O3 hetero interface, we prepared powders labelled CoSiOAlpow with the equilibrium chemical composition (Al:Si:Co = 85:10:5). We first investigated the structural and microstructural changes of the material after H2 treatment (the reductive condition) at 500 °C. As shown in Figure 3, the X-ray diffraction pattern of the CoSiOAlpow sample is composed of XRD peaks of γ-Al2O3, indicating that the Co-doped SiO2 layer is X-ray amorphous. Then, after the H2 treatment at 500 °C, it retains the X-ray amorphous state of the Co-doped SiO2 phase.

Redox Behavior of Co Species
Redox properties of the Co cation in the CoSiOAlpow sample were assessed by in situ using XAFS spectroscopic analyses. The CoSiOAlpow sample was heat-treated at 500 °C under H2 flow followed by cooling to RT under Ar flow (Figure 6a). The spectra recorded at the measurement points shown in Figure 6a are shown in Figure 6b. The CoSiOAlpow sample at RT under the initial Ar flow (Ar(0) in Figure 6a) exhibited a spectrum with a peak at 7726 eV, and this spectrum was different from those of references, Co3O4, CoO, and Co-foil recorded at RT. The unique behavior is that the peak intensity at 7726 eV changes reversibly (Figure 6c): After the 500 °C H2-treatment (H2 in Figure 6a), the peak intensity decreases, i.e., approaches the intensity level of the Co-foil at 7726 eV, then increases in a stepwise manner under

Redox Behavior of Co Species
Redox properties of the Co cation in the CoSiOAlpow sample were assessed by in situ using XAFS spectroscopic analyses. The CoSiOAlpow sample was heat-treated at 500 • C under H 2 flow followed by cooling to RT under Ar flow (Figure 6a). The spectra recorded at the measurement points shown in Figure 6a are shown in Figure 6b.

Redox Behavior of Co Species
Redox properties of the Co cation in the CoSiOAlpow sample were assessed by in situ using XAFS spectroscopic analyses. The CoSiOAlpow sample was heat-treated at 500 °C under H2 flow followed by cooling to RT under Ar flow (Figure 6a). The spectra recorded at the measurement points shown in Figure 6a are shown in Figure 6b. The CoSiOAlpow sample at RT under the initial Ar flow (Ar(0) in Figure 6a) exhibited a spectrum with a peak at 7726 eV, and this spectrum was different from those of references, Co3O4, CoO, and Co-foil recorded at RT. The unique behavior is that the peak intensity at 7726 eV changes reversibly (Figure 6c): After the 500 °C H2-treatment (H2 in Figure 6a), the peak intensity decreases, i.e., approaches the intensity level of the Co-foil at 7726 eV, then increases in a stepwise manner under The CoSiOAlpow sample at RT under the initial Ar flow (Ar(0) in Figure 6a) exhibited a spectrum with a peak at 7726 eV, and this spectrum was different from those of references, Co 3 O 4 , CoO, and Co-foil recorded at RT. The unique behavior is that the peak intensity at 7726 eV changes reversibly (Figure 6c): After the 500 • C H 2 -treatment (H 2 in Figure 6a), the peak intensity decreases, i.e., approaches the intensity level of the Co-foil at 7726 eV, then increases in a stepwise manner under Ar flow after cooling down to RT (Ar(1) in Figure 6a) followed by being maintained at RT for 1 h (Ar(2) in Figure 6a).
As discussed above, the metallic Co was not detected by our intensive HRTEM-STEM analyses. Thus, the spectra shown in Figure 6b suggested that the Co species in the CoSiOAlpow sample were Co(III) and Co(II) with the Co(III)/Co(II) ratio lower than that in Co 3 O 4 (Co(III)/Co(II) = 2/1). Moreover, the peak intensity change shown in Figure 6c suggests the reversible redox property of the Co cation species in the CoSiOAlpow sample under the present cyclic condition of Ar (RT)-H 2 (500 • C)-Ar (RT) treatment.
To investigate the nature of the Co-doped SiO 2 /γ-Al 2 O 3 hetero interface during the redox reactions of Co species, in situ DRIFT spectroscopic analysis was performed on the CoSiOAlpow sample under a cyclic heat-treatment condition under Ar flow, followed by 10% H 2 /Ar flow and subsequent Ar flow at 500 • C. The results are shown in Figure S4. Due to the heterogeneous structure with the Co-doped SiO 2 /γ-Al 2 O 3 hetero interface, the CoSiOAlpow sample exhibited broad spectra of OH groups in the range of 3800-3200 cm −1 : The initial DRIFT spectrum under Ar flow at 500 • C exhibited a broad peak around 3736 cm −1 attributed to free Si-OH groups [36,37] together with a broader one from 3640-3300 cm −1 assigned to hydrogen-bonded X-OH (X = Si or Al) group [23,38] ( Figure S4a). After heat treatment under 10% H 2 /Ar flow for 5 h at 500 • C, the intensity of the broader peak at 3640-3300 cm −1 increased ( Figure S4b), then decreased to be close to the initial intensity level after the subsequent Ar treatment at 500 • C ( Figure S4c). Accordingly, it is suggested that the reduction-oxidation of Co cation species in the CoSiOAlpow sample is associated with OH formation-deformation, i.e., hydrogenation-dehydrogenation.
To identity the hydrogenation-dehydrogenation reactions, cyclic TPR/TPD analysis was performed on the CoSiOAlpow sample. In this attempt, the first TPR/TPD was conducted to initialize the oxidation state of the composite powder sample, then the TPR/TPD profiles were recorded for the 2nd and 3rd cycles. As shown in Figure S5, both the 2nd and 3rd TPR profiles showed a weak broad peak for H 2 up-take at 250-300 • C, and a larger one at 400-500 • C, then the H 2 -uptake continued to some extent under the isothermal H 2 -treatment at 500 • C. Both the 2nd and 3rd TPD profiles showed appropriate desorption peaks, a very weak peak at 250-350 • C, and another larger one during the isothermal Ar-treatment at 500 • C. The simultaneous MS analysis resulted in the detection of a trace amount of H 2 as a single desorption component at the two temperature regions appropriate to the desorption peaks.
These results suggest the following: • The reduction-oxidation of Co cation species in the CoSiOAlpow sample is associated with the formation-deformation of OH groups.

•
The redox reactions begin to proceed at relatively low temperatures around 250-350 • C.

•
The redox reactions are reversible and governed by the H 2 partial pressure at T ≥ 250 • C.
For further investigation on the redox reactions of the Co cation species that proceeded around 250-350 • C, we prepared another powder sample labelled CoSiOAlpow2 with homogeneous structure having almost same Co content but a higher Co/Al of 1/8 compared with that of CoSiOAlpow (1/17).
The redox behaviors of the Co cation species were investigated by cyclic heat treatment up to 350 • C. The CoSiOAlpow2 sample kept its X-ray amorphous state after the cyclic heat treatment at 350 • C under flow of H 2 -Ar-H 2 or H 2 -Ar-H 2 -Ar ( Figure 7A). As shown in Figure 7B, the XPS spectrum of as-synthesized CoSiOAlpow2 sample (Green line) exhibited a broad peak centered at 782 eV attributed to Co 2+ together with another broad peak around 788 eV corresponding to a Co 2+ satellite peak (labeled * in the graph) [39]. In addition, on the right side of the Co 2+ peak, there was a peak shoulder at 779.9 eV assigned to Co 3+ [39]. The Co 3+ peak intensity at 779.9 eV apparently decreased after the second H 2 treatment of the H 2 -Ar-H 2 run (blue line), while that due to metallic Co (777.8 eV) [39] was not observed. Then, after the subsequent Ar treatment (H 2 -Ar-H 2 -Ar run), the Co 3+ peak intensity increased to some extent (orange line).
The oxidation state of the Co cation in the starting Co(NO 3 ) 2 ·6H 2 O was divalent (II). Moreover, metallic Co was not formed under the present heat treatment conditions, and thus formation of Co(III) via the disproportionation reaction (3Co 2+ → 2Co 3+ + Co) was excluded. It should be also noted that the oxidation state of the Co cation in the polymer-derived Co-doped amorphous SiO 2 was completely divalent [40]. It is well known that the Fenton reaction is the Fe(III/II)-catalyzed oxidation of organic compounds (RH) via generation of highly reactive radical species (•OH) in the presence of H 2 O 2 [41,42]: Recently, Ling reported [43] that Co(II) analogously generates •OH in the presence of H 2 O 2 : In this study, H 2 O 2 was used to prepare the precursor solutions for the Co-doped SiO 2 system, which resulted in the partial formation of Co(III) cations within the sample powders via the Fenton reaction. The results obtained in the present study reveal that reversible Co(III)/Co(II) redox reactions are associated with OH formation-deformation, which is governed by the H 2 partial pressure at 250-500 • C: The H 2 up-take/desorption of the CoSiOAlpow sample was mainly detected at higher temperatures around 500 • C ( Figure S2). This could be explained by the kinetic factor necessary for H 2 diffusion through the amorphous SiO 2 network between the top surface and the SiO 2 /γ-Al 2 O 3 hetero interface where Co cations mainly existed.

Gas Permeation Properties of γ-Al
The cross-sectional SEM image of the layered composite membrane sample labelled CoSiOAlmemb is shown in Figure 8. In this study, the initial Co content of the SiO 2 layer was reduced from Co/Si = 1/2 to 1/4, then the Co-doped SiO 2 layer (Co/Si = 1/4) was placed between the two mesoporous γ-Al 2 O 3 layers to investigate the relationship between the Co(III)/(II) redox reactions at the SiO 2 /γ-Al 2 O 3 hetero interface and gas permeation properties. Initially, He permeance was measured at 100-500 • C. As shown in Figure 9a, initial He permeance (Q He -1st) at 100 • C was 3.1 × 10 −6 mol·m −2 ·s −1 ·Pa −1 , then it slightly increased with the permeation temperature. On other hand, after H 2 -treatment at 500 • C, the H 2 permeance (Q H2 ) at 500 • C was slightly higher than that of Q He -1st, and the Q H2 showed an opposite temperature dependency, with decreasing temperature, the Q H2 increased. Finally, after the subsequent He-treatment at 500 • C for 8 h, the He permeance (Q He -2nd) at 500 • C was close to that of Q He -1st, then showed a positive temperature dependency similar to that observed for Q He -1st.  The temperature dependency of the QHe-1 st and QHe -2 nd reveals that He dominantly permeates through micropore channels with a pore size close to the kinetic diameter of He (0.26 nm) [44], and both QHe-1 st and QHe -2 nd are governed by activated diffusion, theoretically expressed as, where Qi is the permeance of permeate gas-i, Q0 is the pre-exponential factor (mol·m −2 ·s −1 ·Pa −1 ), Ea is the activation energy (J·mol −1 ), R is the gas constant (8.314 J·mol −1 ·K −1 ), and T is the temperature (K) [45,46]. On the other hand, as shown in Figure 9a, the QH2 is proportional to the inverse square root of permeation temperature (T −0.5 ), and is thus suggested to be governed by the Knudsen's diffusion mechanism. The theoretical gas permeance based on the Knudsen diffusion through a membrane having pore radius (dp), porosity (ε), tortuosity (τ), and thickness (L) is formulated using the molecular weight of the permeating gas-i (Mi) and the permeation temperature (T) as follows, where ρ is the probability of diffusion through the pore channel, ρi, εi, dpi, τi, and Li are membrane structural factors due to possible dependence on the size of permeate gas-i [47]. Knudsen diffusion occurs when the mean free path of the permeate gas molecule is relatively longer than the pore diameter, so the permeate gas molecules collide frequently with the pore wall. Generally, Knudsen diffusion is dominant for mesopores (2-50 nm) [28,47]. Since the kinetic diameter of He (0.26 nm) is smaller than that of H2 (0.289 nm) [44], prior to the activated diffusion through micropores, He molecules should exhibit Knudsen diffusion characteristics by permeating through   The temperature dependency of the QHe-1 st and QHe -2 nd reveals that He dominantly permeates through micropore channels with a pore size close to the kinetic diameter of He (0.26 nm) [44], and both QHe-1 st and QHe -2 nd are governed by activated diffusion, theoretically expressed as, where Qi is the permeance of permeate gas-i, Q0 is the pre-exponential factor (mol·m −2 ·s −1 ·Pa −1 ), Ea is the activation energy (J·mol −1 ), R is the gas constant (8.314 J·mol −1 ·K −1 ), and T is the temperature (K) [45,46]. On the other hand, as shown in Figure 9a, the QH2 is proportional to the inverse square root of permeation temperature (T −0.5 ), and is thus suggested to be governed by the Knudsen's diffusion mechanism. The theoretical gas permeance based on the Knudsen diffusion through a membrane having pore radius (dp), porosity (ε), tortuosity (τ), and thickness (L) is formulated using the molecular weight of the permeating gas-i (Mi) and the permeation temperature (T) as follows, where ρ is the probability of diffusion through the pore channel, ρi, εi, dpi, τi, and Li are membrane structural factors due to possible dependence on the size of permeate gas-i [47]. Knudsen diffusion occurs when the mean free path of the permeate gas molecule is relatively longer than the pore diameter, so the permeate gas molecules collide frequently with the pore wall. Generally, Knudsen diffusion is dominant for mesopores (2-50 nm) [28,47]. Since the kinetic diameter of He (0.26 nm) is smaller than that of H2 (0.289 nm) [44], prior to the activated diffusion through micropores, He molecules should exhibit Knudsen diffusion characteristics by permeating through The temperature dependency of the Q He -1st and Q He -2nd reveals that He dominantly permeates through micropore channels with a pore size close to the kinetic diameter of He (0.26 nm) [44], and both Q He -1st and Q He -2nd are governed by activated diffusion, theoretically expressed as, where Q i is the permeance of permeate gas-i, Q 0 is the pre-exponential factor (mol·m −2 ·s −1 ·Pa −1 ), Ea is the activation energy (J·mol −1 ), R is the gas constant (8.314 J·mol −1 ·K −1 ), and T is the temperature (K) [45,46]. On the other hand, as shown in Figure 9a, the Q H2 is proportional to the inverse square root of permeation temperature (T −0.5 ), and is thus suggested to be governed by the Knudsen's diffusion mechanism. The theoretical gas permeance based on the Knudsen diffusion through a membrane having pore radius (d p ), porosity (ε), tortuosity (τ), and thickness (L) is formulated using the molecular weight of the permeating gas-i (M i ) and the permeation temperature (T) as follows, where ρ is the probability of diffusion through the pore channel, ρ i , ε i , dp i , τ i , and L i are membrane structural factors due to possible dependence on the size of permeate gas-i [47]. Knudsen diffusion occurs when the mean free path of the permeate gas molecule is relatively longer than the pore diameter, so the permeate gas molecules collide frequently with the pore wall. Generally, Knudsen diffusion is dominant for mesopores (2-50 nm) [28,47]. Since the kinetic diameter of He (0.26 nm) is smaller than that of H 2 (0.289 nm) [44], prior to the activated diffusion through micropores, He molecules should exhibit Knudsen diffusion characteristics by permeating through the mesopore channels where H 2 molecules dominantly permeate. These results suggest that the size of the gas permeable channel pores alternatively change from micropores under He flow to mesopores under H 2 flow, then back to micropores under He flow.
Time dependence of the Q He was monitored after maintaining the CoSiOAlmemb sample under H 2 flow for 4 h. In this attempt, H 2 treatment and subsequent Q He measurements were performed at the lower temperature of 300 • C to highlight the time dependence. As shown in Figure 9b, the initial Q He of 3.53 × 10 −6 mol·m −2 ·s −1 ·Pa −1 slightly decreased with time and reached 3.45 × 10 −6 mol·m −2 ·s −1 ·Pa −1 after 24 h. Then, after the second H 2 treatment for 4 h, the Q He regained close to its initial level. Accordingly, the size of pore channels under the H 2 flow became larger, and then reduced to be close to the initial size. This reversible change in pore size could be explained by the reversible redox of Co(III)/(II) associated with the OH group formation/deformation (Equation (8)), i.e., bond cleavage-regeneration of the Si-O-Al amorphous network governed by the H 2 partial pressure at 300-500 • C.

Conclusions
In this study, Co-doped SiO 2 /γ-Al 2 O 3 composite samples were synthesized by conventional sol-gel methods using H 2 O 2 . The redox behaviors of the Co cations in the composite samples were studied and the results can be summarized as follows: 1. XPS depth profile analyses for the Co-doped SiO 2 /mesoporous γ-Al 2 O 3 layered composite sample revealed preferential Co diffusion from the SiO 2 top-layer matrix to the γ-Al 2 O 3 surface, and the equilibrium chemical composition of the Al, Si, and Co within the mesoporous γ-Al 2 O 3 layer was determined as Al:Si:Co = 85:10:5. 2.
XRD, HRTEM, and HAADF-STEM analyses revealed that the Co-doped SiO 2 /γ-Al 2 O 3 composite powder sample with the equilibrium composition of Al:Si:Co = 85:10:5 kept its amorphous state without crystallization of Co oxides after H 2 treatment at 500 • C.

3.
In situ XANES spectroscopic analyses during the 500 • C heat treatment under H 2 flow and subsequent cooling to RT under Ar flow exhibited reversible redox properties of Co cations in the Co-doped SiO 2 /γ-Al 2 O 3 composite powder sample. 4.
XPS and DRIFT spectroscopic analyses and cyclic TPR/TPD measurements concluded the reversible Co(III)/Co(II) redox reactions were associated with OH formation (hydrogenation)-deformation (dehydrogenation) within amorphous aluminosilicate, which was governed by the H 2 partial pressure at 250-500 • C. 5.
The Co(III) cations in the present composite samples formed via the Fenton reaction in the presence of H 2 O 2 during the sample syntheses. The observed H 2 -triggered chemical valve response needs a long time to function, especially for closure that takes approximately 8 h at 500 • C under Ar flow. The relations between the relative amount of Co(III) in the Si-O-Al amorphous network, recyclability of the redox reactions, and the valve response remain as future subjects.