Effect of Preparation Condition of TiO 2 Film and Experimental Condition on CO 2 Reduction Performance of TiO 2 Photocatalyst Membrane Reactor

It was previously reported that CO2 could be reduced into CO, CH4, and so forth, which can be used as fuels, by TiO2 as the photocatalyst under UV radiation. To increase the concentration of fuel and improve CO2 reduction performance on TiO2 photocatalyst, a membrane reactor composed of TiO2 and gas separation membrane prepared by sol-gel and dip-coating method has been built. Factors such as rising speed (RS) in the dip-coating process and the timing and amount of water injected in the membrane reactor in CO2 reduction experiment have been investigated. As a result, the largest amount of TiO2 film is obtained for RS = 0.66 mm/s among various RS conditions investigated in this study. According to CO2 reduction experiment by gas circulation type reactor, too much water which cannot be consumed in CO2 reduction process would not help improving the CO2 reduction performance.

For example, for CH 4 and CO, 5.3 vol.% and 12.5 vol.% are required, respectively.Therefore, the big breakthrough in increasing the concentration level is necessary to advance the CO 2 reduction technology.
According to the calculation by the author, the mass transfer time of 10 * 5 -10 * −1 s is much slower than the photo reaction time of 10 * −9 -10 * −15 s in this process.Therefore, the mass transfer is thought to be the main factor contributing to the slow photocatalytic reaction.Another reason causes the low reduction rate is the reoxidization of the products.Namely, due to the reaction surface covered by products, the further movement of the reactants to the reaction surface is prevented and the reverse reaction, that is, re-oxidization, which produces CO 2 from CO and CH 4 , occurs.Therefore, it is desirable that the products, that is, CO and CH 4 , are removed from the reaction surface as soon as they are produced.The reactants, that is, CO 2 and water vapour, can then continue to react on the reaction surface, and the fuel production can be sustained under this nonequilibrium reaction condition.In other words, by removing the products away, the reaction is forced to head to one direction.The gas separation membrane is usually used in the gas separation processes like H 2 production from hydrocarbon, O 2 enrichment from the air, and CO 2 capture of the industrial power plants.Since the molecular diameters of reactants of CO 2 and water vapour are smaller than that of CO and CH 4 (CO 2 : 0.33 nm, water vapour: 0.28 nm, CH 4 : 0.38 nm, CO: 0.38 nm) [25], the promotion of the reaction by gas separation is thought to be possible and was attempted in this study.This is a novel approach aiming to improve CO 2 reduction performance over the TiO 2 .No similar attempts have been reported yet.Therefore, the present paper investigates the effect of preparation condition of membrane reactor consisting of TiO 2 photocatalyst and gas separation membrane and experimental condition on CO 2 reduction performance of the membrane reactor.

Experiment
2.1.Preparation of TiO 2 Film Coated on Gas Separation Membrane.Since TiO 2 photocatalyst membrane reactor is a novel approach to improve CO 2 reduction performance over the TiO 2 , it is necessary to verify the effect of combination of gas separation membrane and TiO 2 on CO 2 reduction performance.Therefore, the authors investigated the preparation procedure of TiO 2 film coated on gas separation membrane by sol-gel and dip-coating method and the experimental operation conditions to promote the CO 2 reduction performance of the TiO 2 photocatalyst membrane reactor.The rising speed (RS) of gas separation membrane from the TiO 2 sol solution in dip-coating process, which influences the thickness and physical and chemical structure of TiO 2 film coated on gas separation membrane, was investigated.The surface structure and crystallization characteristics of TiO 2 film coated on gas separation membrane, under the various RS conditions, were analysed by SEM (scanning electron microscope), EPMA (electron probe microanalyzer), and XPS (X-ray photoelectron spectroscopy) to understand the impact of difference of RS on the surface structure and crystallization characteristics of TiO 2 film, as the first step.
The CO 2 reduction and permeation performance of TiO 2 film coated on gas separation membrane was evaluated by the batch type reactor in order to select the optimal TiO 2 film coating conditions to use to prepare the membrane for the reactor of gas circulation type.In other words, the ideal TiO 2 film for this application should have large reaction surface areas and high crystallization characteristics but does not block the pores in gas separation membrane.After the suitable TiO 2 film coating conditions are known, the CO 2 reduction performance of TiO 2 film coated on porous gas separation membrane was investigated by the gas circulation type reactor.The effectiveness of gas separation and gas circulation using the gas separation membrane on CO 2 reduction performance was compared with the results obtained from the batch type reactor experiment.Sol-gel and dip-coating method was used for preparing TiO 2 film in this study.Figure 1 shows the flow chart of the sol-gel and dip-coating method.TiO 2 sol solution was made by mixing [(CH 3 ) 2 CHO] 4 Ti (purity of 95 wt.%, Nacalai Tesque Co.), anhydrous C 2 H 5 OH (purity of 99.5 wt.%, Nacalai Tesque Co.), distilled water, and HCl (purity of 35 wt.%, Nacalai Tesque Co.).
The gas separation membrane (silica-alumina gas separation membrane, Noritake Co., Ltd.), which was the porous  1 that the mean pore size of silica layer is not ideal, as it is not between the molecular diameter of reactants and that of products, as required.It is difficult to find the gas separation membrane with the ideal pore size.However, the gas separation membrane selected is capable of separating gases through both molecular sieving diffusion and socalled Knudsen diffusion mechanisms; therefore, it can be used.The Knudsen diffusion can separate the gases whose molecular diameters are smaller than the pore size of silica layer.Since the molecular diameter of reactant and that of product are actually different as described above, we have decided to adopt this gas separation membrane.)) located at 20 mm apart from the surface of gas separation membrane symmetrically.These parts are assembled with stainless plates by bolts and nuts.The center wave length and mean light intensity of UV light illuminated from energy UV lamp are 365 nm and 2.4 mW/cm 2 , respectively.This is similar to the average light intensity level of UV ray in solar radiation in the daytime.Figure 3 illustrates the whole experimental system setup, which is termed as membrane reactor.With this membrane reactor, not only batch type but also gas circulation type experiment can be conducted.When it is used for batch type experiment, the valves located at inlet and outlet of reactor are closed.The membrane reactor is composed of reactor part, CO 2 gas cylinder, mass flow controller (MODEL3660, KOFLOC), mass flow meter (CK-1A, KOFLOC), pressure gauge, gas drier, and tube pump (WM-520S/R2, Iwaki Pumps).In the CO 2 reduction experiment by the batch type reactor, CO 2 gas whose purity was 99.995 vol.% was flowed through the reactor as a purged gas for 15 min at first.After that, the valves located at inlet and outlet of reactor were closed.After confirming the gas pressure and gas temperature in the reactor were at 0.1 MPa and 298 K, respectively, the distilled water of 1.00 mL (55.6 mmol) was injected into the reactor, and UV light illumination was started at the same time.This water was vaporized after injected into the reactor.Despite the heat of UV lamp, the temperature in reactor was kept at about 343 K during the CO 2 reduction experiment.The amounts of the injected water and the CO 2 in the batch type reactor are 55.6 mmol and 13.0 mmol, respectively.The gas in reactor was sampled every 24 h in CO 2 reduction experiment.The gas samples were analysed by FID gas chromatograph (GC353B, GL Science) and methanizer (MT221, GL Science).The concentration of water vapour and the temperature in reactor was measured by dew point meter (VAISALA HUMICAP HMT330, VAISALA) and thermocouple, respectively.In this experiment, only CO was detected as the product.In the CO 2 permeation experiment, the CO 2 reduction and permeation experimental system shown in Figure 3 was arranged.

Experimental
Figure 4 illustrates the CO 2 reduction and permeation experimental system for CO 2 permeation experiment by batch type.In the CO 2 permeation experiment, the CO 2 permeation flux was measured under the condition that the absolute pressure and temperature of supply gas to the apparatus were 0.10-0.40MPa and 298 K, respectively.The flow rate of supply gas was set at 500 mL/min by mass flow controller.The flow rate of permeation gas was measured by mass flow meter.In the CO 2 reduction experiment carried out by the membrane reactor of gas circulation type, UV light was illuminated under the same condition of batch type reactor until the steady reaction state was confirmed.After that, the gas circulation by tube pump was started.The suction pressure and flow rate of permeation gas were controlled to evaluate the effect of gas separation and circulation on CO 2 reduction performance of this membrane reactor.The suction pressure and flow rate of permeation gas were set at 0.2 MPa and 0.39 mL/min, respectively.The produced CO would be removed from the reactor to outside of the system by switching the outlet valve of reactor on and off when needed.The distilled water of 1.00 mL (55.6 mmol) or 3.00 mL (166.8 mmol) was injected into reactor when the International Journal of Photoenergy (2)  CO 2 reduction experiment under the condition of batch type reactor was established.The gas samples taken every 24 h from reactor were analysed by FID gas chromatograph and methanizer.The concentration of water vapour and the temperature in the reactor were also measured. Atmosphere (2) (5)

Analysis Result of TiO 2 Film Coated on Gas Separation
Membrane by SEM. in this study at the center, and the alumina layer is exposed except for the area covered by silica layer.Then, SEM images of TiO 2 film coated were taken for the silica-covered area and the alumina area separately.From these figures, it can be seen that the number of clucks of TiO 2 film coated on alumina layer is less than that on silica layer, resulting that the amount of TiO 2 coated on alumina layer is larger than silica layer.Since the pore size of alumina layer is larger than that of silica layer as listed in Table 1, it can be thought that TiO 2 sol solution flows into the alumina layer more easily than the silica layer in dip-coating process.Therefore, it seems that TiO 2 film coated on alumina layer is fixed more strongly than that on silica layer.

Analysis Result of TiO 2 Film Coated on Gas Separation
Membrane by EPMA. Figure 7 demonstrates EPMA images of TiO 2 film prepared under various RS conditions.These EPMA images are taken by 1500 times magnification under the condition of acceleration voltage of 15 kV and current of 3.0 × 10 * −8 A. In Figure 7, the concentration distribution of Ti detected in observation area is indicated by the difference of colour.Light colours, for example, white, pink, and red, mean that the amount of Ti is large, while dark colours like black, blue, and green mean that the amount of Ti is small.EPMA detects each element whose crystallization characteristic is memorized in advance.Therefore, if the large concentration of Ti is detected, it means that the amount of crystallized TiO 2 coated on gas separation membrane is large.The average concentration of Ti in the observation area is also shown in Figure 7.According to Figure 7, the average concentration of Ti detected in observation area for alumina layer is larger than that for silica layer.It can be said that TiO 2 film is coated in the pores of alumina layer mainly.From Figure 7, it reveals that the concentration of Ti is reduced with the increase in RS.Generally speaking, the thickness of TiO 2 film becomes thick and hubbly with the increase in RS.
The thermal stress is acted on the interface between TiO 2 film and gas separation membrane in the firing process, resulting that formation of large clucks and detachment of TiO 2 film occur.Consequently, the concentration of Ti is reduced when RS is high.

Analysis Result of TiO 2 Film Coated on Gas Separation
Membrane by XPS.Figures 8 and 9 show the intensity distributions of Ti detected in silica layer and alumina layer, respectively.These XPS data were obtained under the condition of ion acceleration voltage is 4 kV and pass energy of 112 eV.The samples were sputtered by Ar ion laser whose acceleration voltage is 2 kV.The sputtering speed was 15 nm/min, which was estimated by assuming the sample as SiO 2 .The electron orbits of detected elements which were Ti, Si, and Al were set at 2p.From these figures, it is known that the sputtering time for RS = 1.1 mm/s is the shortest among various RS conditions.However, the intensity of detected Ti is over 80000 cps for RS = 1.1 mm/s.It can be said that the amount of Ti is large with RS = 1.1 mm/s, resulting that fine TiO 2 film is prepared.Regarding RS = 0.66 mm/s, it is seen that the intensity of Ti over 60000 cps can be detected up to about sputtering time of 25 min for silica layer.Though the intensity of Ti detected in alumina layer is smaller than that in silica layer, the detecting period of Ti in alumina layer is almost equal to that in silica layer.
According to Figure 7, the average concentration of Ti for RS = 0.66 mm/s is the largest among various RS conditions.In addition, when the sputtering time is longer, it can be said that TiO 2 is coated deeper to the thickness direction of gas separation membrane, resulting that the reaction surface of TiO 2 film is larger.From these results, one can conclude that the largest amount of TiO 2 film is coated on gas separation membrane with RS = 0.66 mm/s.Meanwhile, the intensity of detected Ti with RS = 1.7 mm/s is the lowest among various RS conditions.In addition, the average concentration of Ti is 0 cps as shown in Figure 7. Consequently, it is clear that the amount of TiO 2 film coated on gas separation membrane for RS = 1.7 mm/s is very small.

CO 2 Reduction by the Membrane Reactor of Batch Type.
Figure 10 shows the CO concentration change in products with illumination time of UV light for several TiO 2 film prepared under various RS conditions.According to our previous studies, the reversal of superiority or inferiority on CO 2 reduction performance of TiO 2 film among selected parameters was confirmed until UV light illumination time of 48 h.However, this reversal was not confirmed, and the superiority or inferiority among selected parameters was kept after UV light illumination time of 72 h.From this reason, in this study, the data is obtained only up to UV light illumination time of 72 h for the purpose of determining the best condition for promotion of CO 2 reduction performance.The distilled water of 1.00 mL (55.6 mmol) was injected into reactor at the beginning of the CO 2 reduction experiment.
From the data at UV light illumination time of 72 h shown in Figure 10, it is known that the concentration of CO is increased with decreasing RS values.In this experiment, only CO was detected as a product.Referring to the images of the SEM, EPMA, and XPS shown in Figures 5, 6, 7, 8, and 9, the reason of this is thought to be that the amount of TiO 2 coated on gas separation membrane becomes larger when RS decreases within the range of 0.66-1.7 mm/s.Under slow RS condition, TiO 2 sol solution is easy to remain in the pore of silica and alumina layer in the dip-coating process, and TiO 2 film coated becomes thin and even.Therefore, it is thought that the largest amount of TiO 2 film is coated on gas separation membrane under the condition of RS = 0.66 mm/s.
According to the reaction scheme in this study as shown in Figure 11 [1,5,6,11,16,[26][27][28][29], the number of electron and hydrogen ions (H + ) decides the type of product in the reaction.In this experiment, CH 4 , C 2 H 4 , and the other hydrocarbons were not detected by gas chromatograph, due to less H + in the reaction.Therefore, the amount of water vapour injected into the reactor is an important parameter to be investigated since it is the source of H + .From the reaction scheme, water vapour of 1 mol to CO 2 of 1 mol is necessary to produce CO of 1 mol.In this experiment, the amount of substance of injected water and CO 2 charged in the batch type reactor is 55.6 mmol and 13.0 mmol, respectively, resulting that the molar ratio of water vapour to CO 2 is 4.28.Although the amount of water vapour injected seemed sufficient for this reaction, the change in temperature distribution and the concentration distribution of water vapour in the batch type reactor with time need to be checked to confirm what had happened.

CO 2 Permeation by the Membrane Reactor of Batch Type.
Figure 12 shows the relationship between CO 2 permeation flux and pressure difference for several TiO 2 film prepared under various RS conditions.The pressure difference is known by subtracting the gas pressure after penetrating the gas separation membrane from the gas pressure before.The CO 2 permeation flux is calculated by the following equation: where F CO2 , V p , A p , and t p are CO 2 permeation flux [mol/(m 2 •s)], volume of permeated gas (mol), outer surface area of gas separation membrane (m 2 ), and gas separation time (s), respectively.
Comparing these results at pressure difference of 0.30 MPa, it is known that the CO 2 permeation flux for RS = 1.1 mm/s is the highest among these RS conditions.According to XPS analysis as shown in Figures 8 and 9, the sputtering time of detecting Ti for RS = 1.1 mm/s is the shortest among various RS conditions, indicating that the depth of coated TiO 2 film diffused into the silica and alumina layers of gas separation membrane is the shallowest.Consequently, the highest CO 2 permeation flux is obtained at RS = 1.1 mm/s.On the other hand, regarding RS = 0.66 mm/s and 1.7 mm/s, the sputtering time of detecting Ti is longer though the intensity of Ti detected is lower as shown International Journal of Photoenergy in Figures 8 and 9, indicating the depth of coated TiO 2 film.TiO 2 film diffused into the silica and alumina layers is deeper, thus the CO 2 permeation flux is lower.

Selection of the Optimum Coating Condition.
To select the optimum coating condition of TiO 2 film which would lead to the highest CO 2 reduction and permeation performance, the results by SEM and EPMA analysis and the results of CO 2 reduction and permeation experiment by the batch type reactor are compared and analysed.Figure 13 shows the comparison of the results between the concentration of produced CO and the CO 2 permeation flux for various RS conditions.In Figure 13, the concentration of CO at UV illumination of 72 h shown in Figure 10 and CO 2 permeation flux at pressure difference of 0.30 MPa shown in Figure 12 are demonstrated.It can be seen that the concentration of CO is decreased with increasing RS gradually.On the other hand, the CO 2 permeation flux peaks at RS = 1.1 mm/s.Therefore, the optimum RS is different from the viewpoint of CO 2 reduction and permeation performance.Since the main goal of this study is to promote the CO 2 reduction performance, we have selected the RS = 0.66 mm/s as the optimum coating condition to reduce CO 2 .

CO 2 Reduction by the Membrane Reactor of Gas Circulation Type.
According to the reaction scheme shown in Figure 11, CO is reoxidized with the O 2 , that is, a by-product in this reaction.After attaining to the steady reaction state, the concentration of CO is decreased.This is the opposite reaction toward CO 2 reduction into fuel.Moreover, since the C 2 H 4  photocatalytic reaction occurred on the reaction surface, it is easy for the reaction surface to be covered by the products, which would stop the further reaction to happen.Therefore, removing the product of CO and CH 4 from the reaction surface as well as transporting the reactants, that is, CO 2 and water vapour, to the reaction surface quickly are necessary to promote further reaction and prevent the re-oxidization of CO.In this study, a tube pump and a gas separation membrane are used to realize this desirable measure for the promotion of CO 2 reduction performance.
Figure 14 shows the concentration change of CO produced with illumination time of UV light.In this experiment, only CO was detected as a product.The distilled water of 1.00 mL (55.6 mmol) was injected into reactor when the CO 2 reduction experiment by batch type reactor started.To show the effect of gas separation and circulation on the CO 2 reduction performance, the gas circulation by tube pump only starts after the steady reaction state is reached.The steady reaction state was defined as the state at which the concentration of CO no longer increases along the time.Since the concentration of CO is diluted with the CO 2 in the pipe lines of the gas circulation type reactor after starting gas circulation, the concentration of CO is corrected by the following equation: where C c , V total , C d , and V batch mean corrected concentration of CO (ppmV), total gas volume inside the experimental apparatus including the gas volume in the pipe lines (m 3 ), detected concentration of CO (ppmV), and total gas volume inside the experimental apparatus in the case of batch type reactor (m 3 ), respectively.The experiment by batch type and gas circulation type was carried out during the period from 0 h to 216 h and from 216 h to 480 h, respectively.It is observed that the concentration of CO in the reactor keeps increasing until 72 h and starts to be steady after   15, is calculated by where R CO and t int mean production rate of CO (ppmV) and gas sampling interval (h), respectively.The R CO used for calculating is 24 h.In the experiment by batch type reactor, it can be seen that the production rate of CO peaks at 3.58 ppmV/h at the UV light illumination time of 24 h and is decreased afterwards gradually.The production rate of CO which is 0 means that the reaction steady state is reached.The negative value of the production rate means the inverse reaction, that is, re-oxidization occurs.In the experiment by gas circulation type reactor, the production rate of CO after starting the gas circulation peaks at the highest value of 0.89 ppmV/h in the period from 240 h to 264 h.However, the production rate of CO became smaller after the passage time of 48 h, that is, after the total illumination time of UV light of 264 h.Comparing the production rate of CO after starting the gas circulation and that at steady state of batch type reactor except for the period from 0 h to 24 h, the effect of gas International Journal of Photoenergy separation and circulation on CO 2 reduction can be verified.However, the production rate of CO and the concentration of CO are still lower than the target value levels set for this study.Figure 16 shows the concentration of the water vapour during the CO 2 reduction experiment by batch type and gas circulation type reactor with illumination time of UV light.Since the concentration of water vapour is diluted with the gas in the pipe lines of the gas circulation type reactor after starting gas circulation, the concentration of water vapour is corrected by the following equation: where C c-H2O , V total , C d-H2O , and V batch mean corrected concentration of water vapour (ppmV), total gas volume inside the experimental apparatus including the gas volume in the pipe lines (m 3 ), measured concentration of water vapour (ppmV), and total gas volume inside the experimental apparatus in the case of batch type reactor (m 3 ), respectively.From this figure, it is known that the highest concentration of water vapour in the experiment by batch type reactor and gas circulation reactor is 32400 ppmV and 65387 ppmV, respectively.According to saturated steam table, if the water vapour is saturated in the reactor at 343 K, it should have the concentration of 307545 ppmV.The 1.00 mL water injected into batch type reactor and gas circulation reactor, if all evaporated, could make the vapour concentration of 53040 ppmV and 112968 ppmV, theoretically.As not sure the lower water concentration measured in caused by water vapour was absorbed by gas separation membrane or not all of water injected was vaporized, more water was injected in order to evaluate the effect of amount and timing of water injection.
The further experiment plan was based on the assumption that at the steady and inverse reaction states, there was  not sufficient water in the system.Therefore, the water was injected into reactor when the steady and inverse reaction state in the CO 2 reduction was confirmed for not only batch type but also gas circulation type experiment.The amount of water injected was 1.00 mL at every time in this experiment.If the steady state was maintained after the injection of water in the batch type experiment by reactor, the gas circulation experiment then starts.
Figure 17 shows the concentration change in produced CO with illumination time of UV light.The initial distilled water of 1.00 mL (55.6 mmol) was injected into reactor when the CO 2 reduction experiment by batch type reactor started.It had shown that the concentration of CO in the batch type reactor kept increasing until 168 h, and the concentration of CO reached 136 ppmV.After the water of 1.00 mL was added into reactor at 216 h when the steady state was confirmed, the concentration of CO increased again and attained to 186 ppmV at 480 h.Since the steady state was confirmed again at 504 h, another 1.00 mL of water was added into reactor again.However, the concentration of CO did not increase any more, indicating the steady state maintained.After gas circulation started from 576 h, the concentration of CO started to increase again and peaked at 179 ppmV at total UV light illumination of 624 h.Since the steady state in the experiment by gas circulation type reactor was confirmed at 648 h, the further 1.00 mL of water was injected into reactor.However, the concentration of CO did not increase further, indicating the water inside system was sufficient, and its effect was peaked.
Figure 18 shows the concentration change in water vapour inside the system in the water adding experiment described above.From this figure, the measured concentration of water vapour obtained in CO 2 reduction experiment by batch type reactor is almost 25000 ppmV with total 3.00 mL water injected.Figure 19 shows the change in production rate of CO with illumination time of UV light.From this figure, it can be seen that there are two peaks of production rate of CO in CO 2 reduction experiment by batch type reactor, and there is one peak of production rate of CO in CO 2 reduction experiment by gas circulation type.The first peak was obtained at UV light illumination of 24 h with total of 1.00 mL water in the system, which was injected at the start of CO 2 reduction experiment.According to Figure 20, the production rate of CO decreases, after peaking at 24 h, gradually and reaches negative value at UV light illumination of 216 h.As mentioned above, since the steady reaction state was confirmed at UV light illumination of 216 h, the other 1.00 mL water was added into reactor.As a result, the second peak was obtained at UV light illumination of 264 h.After that, the production rate of CO decreases again.Although further 1.00 mL water was added into reactor again at UV light illumination of 504 h, the production rate of CO remains negative value.After gas circulation from total UV light illumination of 576 h, the production rate of CO increases and peaks at total UV light illumination of 600 h.From these results, the effect of water injection on the promotion of CO 2 reduction performance is verified.However, both the highest concentration and the highest production rate of CO in this CO 2 reduction experiment are lower than those in the case of only total 1.00 mL water injected.In addition, compared to the case of total amount of water injected of 1.00 mL, the effect of switching batch type reactor to gas circulation type reactor on the promotion of CO 2 reduction performance is not confirmed in this experiment.
Nevertheless, the above-described results seem to reveal that the optimum timing of water injection is the very beginning of CO 2 reduction experiment.Therefore, one more experiment was conducted, that is, the distilled water of 3.00 mL (166.8 mmol) was injected into reactor at the very beginning of the experiment.Figure 20 shows the concentration change in produced CO with illumination time of UV light.The experiment by batch type and gas circulation type reactor was carried out during the period from 0 h to 168 h and from 168 h to 264 h, respectively.The highest concentration of CO which is 126 ppmV is obtained at UV light illumination of 144 h.Since the concentration of CO is not increased over 144 h, indicating that the experiment by batch type reactor attains to the steady reaction state, the gas circulation was started.After gas circulation, the concentration of CO starts to increase again and peaks at 171 ppmV at UV light illumination of 240 h.time of UV light for the amount of water injected of 1.00 mL and that of 3.00 mL, respectively.The change rate of water vapour is calculated by where R H2O , ΔC H2O , and t int mean change rate of water vapour (ppmV/hour), the amount of increase in concentration of water vapour (ppmV), and measurement interval of water vapour (h), respectively.The used for calculating R H2O is 24 h.From these figures, it can be seen that the concentration of water vapour in both cases decreases rapidly from 0 h to 48 h.Since the highest production rates of CO for the amount of water injected of 1.00 mL and 3.00 mL are obtained from 0 h to 48 h, the CO 2 reduction is carried out well in the period.While the concentration of water vapour increases with illumination time of UV light from 0 h to 24 h due to temperature rise in reactor, the change rate of water vapour closes to 0 in the period from 48 h to 96 h irrespective of the amount of water injected.Therefore, it can be thought that the amount of water consumed by photocatalytic reaction is balanced out by the amount of water vaporized due to the heat of UV lamp in the period from 48 h to 96 h.After starting gas circulation, the change rate of water vapour keeps at low level and decreases gradually irrespective of the amount of water injected, while the production rate of CO rises.Since the CO 2 reduction performance is promoted by gas separation and circulation operation, the water vapour is consumed by the oxidization reaction shown in Figure 11.In addition, the water vapour is also adsorbed by drier, which is installed to protect the mass flow meter, in pipeline of gas circulation type reactor during gas circulation.The water concentration increase due to temperature increase balances out by consumption of both oxidization reaction and adsorption by drier.Therefore, the change rate of water vapour keeps low and decreases gradually after starting gas circulation.Consequently, it reveals that CO 2 reduction performance of gas circulation type reactor is declined by increasing the amount of water injected due to decreasing the concentration of water vapour.
Therefore, it can be concluded that too much water in that system, no matter when it was added, would not help improving the CO 2 reduction performance.
Finally, the conversion and the quantum yield are calculated for the best result, which was obtained under the condition that the distilled water of 1.00 mL was injected into reactor when the CO 2 reduction experiment by batch type reactor started, in gas circulation experiment of this study.In this study, the conversion is calculated by the following equation: where α is the conversion (%), C CO2-0 is the concentration of CO 2 in the reactor and pipelines at the start of CO 2 reduction experiment (ppmV).In this study, CO of 1 mol is produced from CO 2 of 1 mol according to reduction reaction in the reaction scheme shown in Figure 11.Then, it is thought that C C is equivalent to the amount of concentration of CO 2 decreased by reduction reaction.From (6), the conversion under the best result condition is calculated as follows: α = (234) (99.995 × 10000) × 100 = 0.0234 (%).
On the other hand, the quantum yield is calculated by the following equation in this study: where η is the quantum yield (%), N input is the number of photon absorbed by TiO 2 photocatalyst (-), N output is the number of photon used for photocatalytic reaction (-), I is the light intensity of UV light (W/cm 2 ), t is the UV light International Journal of Photoenergy 13 illumination time (s), λ is the wave length limit of light which TiO 2 absorbs for photocatalytic reaction (m), A re is the reaction surface area of TiO 2 film which is thought to be almost equal to outer surface area of gas separation membrane (cm 2 ), h is Plank's constant (J•s), c is light speed (m/s), N CO is the electron number which is necessary for producing CO of a molecular (-), M CO is the amount of substance of produced CO (mol), N A is Avogadro's number (1/mol), V total is the total gas volume inside the experimental apparatus including the gas volume in the pipe lines (m 3 ), V CO-mol is the volume of CO of a mol at initial temperature in experimental apparatus (m 3 ).From these equations, the quantum yield under the best result condition is calculated as follows: Since the conversion and the quantum yield are low, further investigation, for instance, controlling the TiO 2 film structure coated on gas separation membrane and metal doping which lead promotion of reduction reaction are necessary.They are future subjects in this study.

Conclusions
Based on the above-experimental results and discussion, the following conclusions can be drawn from this experimental study.
According to characterization by SEM, EPMA, and XPS, the amount of TiO 2 film coated on gas separation membrane is reduced with increasing RS, and the largest amount of TiO 2 film is obtained for RS = 0.66 mm/s among various RS conditions investigated in this study.
According to CO 2 reduction experiment by batch type reactor, the concentration of CO is decreased with increasing RS gradually.On the other hand, the CO 2 permeation flux peaks at RS = 1.1 mm/s.Since the main goal of this study is to promote the CO 2 reduction performance, the RS = 0.66 mm/s is selected as the optimum coating condition in this study.
According to CO 2 reduction experiment by gas circulation type reactor, the positive effect of gas separation and circulation on CO 2 reduction performance is confirmed.However, too much water in that system which cannot be consumed in CO 2 reduction process, no matter when it was added, would not help improving the CO 2 reduction performance.

Figure 2 :
Figure 2: CO 2 reduction reactor composed of TiO 2 film coated on gas separation membrane (UV lamp is removed from this figure for understanding the inside of CO 2 reduction reactor).

Figure 4 :
Figure 4: Arranged CO 2 reduction and permeation experimental system for CO 2 permeation experiment.

Figure 5 :Figure 6 :
Figure 5: SEM images of TiO 2 film coated on silica layer prepared under various RS conditions.

Figure 8 :
Figure 8: Intensity distributions of detected Ti in silica layer.

Figure 9 :
Figure 9: Intensity distributions of detected Ti in alumina layer.

Figure 10 :
Figure 10: Concentration change in produced CO with illumination time of UV light for several TiO 2 film prepared under conditions of different RS.

Figure 12 :Figure 13 :Figure 14 :
Figure 12: Relationship between CO 2 permeation flux and pressure difference for several TiO 2 film prepared under various RS conditions.

Figure 16 :
Figure 16: Concentration change in water vapour during the CO 2 reduction experiment by batch type and gas circulation type reactor with illumination time of UV light (injection of water of 1.00 mL at the start of CO 2 reduction experiment).

Figure 17 :
Figure 17: Concentration change in produced CO with illumination time of UV light (injection of water of 1.00 mL many times).

Figure 18 :Figure 19 :
Figure 18: Concentration change in water vapour during the CO 2 reduction experiment by batch type and gas circulation type reactor with illumination time of UV light (injection of water of 1.00 mL many times).

Figure 20 :
Figure 20: Concentration change in produced CO with illumination time of UV light (injection of water of 3.00 mL at the start of CO 2 reduction experiment).

Figure 21 :Figure 22 :
Figure 21: Change in production rate of CO and change rate of water vapour with illumination time of UV light for the amount of water injected of 1.00 mL.