2022 Volume 62 Issue 12 Pages 2476-2482
Oxalic acid is an attractive chemical platform potentially available from CO2 due to its established applications and chemical characteristics enabling it to serve as a mediator in hydrometallurgy including iron-making. However, a method for synthesizing oxalic acid from CO2 has yet to be established. In the present work, the formation of oxalate scaffold during heating of cesium carbonate (Cs2CO3) in the presence of CO2 and H2 as reactants was experimentally investigated with a particular focus on the influence of supporting Cs2CO3 over porous materials. Among the support materials examined, activated carbon (AC) had a notable effect in improving the reaction rate and yield of total carboxylates (formate and oxalate) during experiments with an autoclave. An important problem was the dominant presence of formate, the intermediate between carbonate and oxalate, accounting for over 90% of the carboxylates. Changing the reaction conditions, including temperature, reaction time, partial pressure of gas components, and amount of loaded Cs2CO3, did not alter the situation. Alternatively, re-heating of the formate-rich salts over AC under CO and CO2 enhanced the oxalate fraction while maintaining the total carboxylates yield. Benefiting from the employment of support material, the two-step conversion was carried out using a gas-flow type reactor with a packed bed of Cs2CO3 supported over AC. In this reaction system, because water, acting as a promoter, was absent, the total carboxylates yield was lower than that in the autoclave, while the oxalate fraction was higher, being 71.8% with a yield of 43.2% on a Cs2CO3-carbon basis.
A massive amount of carbon dioxide is emitted into the atmosphere in industries such as energy and transport, as well as manufacturing, including iron-making.1,2) To abate growing greenhouse gas emissions while considering the quality of life of people, the development of utilization technologies using CO2 as feedstock for chemical platforms is of utmost importance.3) Most chemical platforms, such as methane, alcohols, and alkanes, need H2 as a co-reactant, the green production of which is at an early development stage. Oxalic acid (H2C2O4) has an advantage in this regard as its formation from CO2 requires only one hydrogen atom per carbon with no removal of oxygen in stoichiometry. Also, being solid at ambient temperatures, it is stable and easy to store as feedstock or as a type of carbon fixation. Oxalic acid is widely available in the market and is used in a variety of applications, mainly for extraction, bleaching, dyeing, and purification processes.4,5) An emerging focus is its use as feedstock for polymers.6,7) Furthermore, we have recently shown that oxalic acid could be used for iron-making as a mediator. Iron was extracted from iron ores as iron (III) oxalate, photochemically reduced to iron (II) oxalate, and then converted to metallic iron by pyrolytic reduction.8,9) Because of the unique properties of oxalic acid and oxalates, the metallic iron was produced even at a low temperature of 500°C from low-grade iron ores, and it had high purity. However, the current market for oxalic acid is not large enough to support iron-making. The total production in China, the largest producer, was only 200 kilotons as of 2016.6)
The industrial production of oxalic acid has employed several types of feedstocks including biomass (e.g., sugars), ethylene glycol, carbon monoxide, alkali formates, and propylene.10) No industrial process uses CO2 as the feedstock. Meanwhile, the green aspect of oxalic acid is found in some feedstocks: Biomass absorbs CO2 when they grow, CO is generated from CO2 by a reverse water-gas shift, and alkali formates are available from CO2. Research groups in the EU are aiming at producing oxalic acid and oxalic acid-derived polymers from CO2 in their European Horizon 2020 project “OCEAN”.6,7) In their research, CO2 was electrochemically reduced to potassium formate, and the thermal coupling of formate produced oxalate. The latter reaction has been extensively studied for a long time and has already been commercialized.7,11,12) The current research focus is the promotion of coupling by adding base catalysts such as alkali hydrides.13)
The direct or one-pot synthesis of oxalate from CO2 has attracted research interests in recent years. Most studies employ transition metal complexes as the catalyst to electrochemically transform CO2 into the oxalate.14,15,16,17,18) The reaction is believed to occur as either a two-step one-electron reduction or one-step two-electron reduction.19) Facile synthesis is attractive, but the drawback is the creation of an elaborate metal complex and organic solvents, which are not viable for use in industrial processes for mass production. An alternative method is the thermal conversion of CO2 under high pressure with Cs2CO3 as the promoter. When Cs2CO3 was heated at 380°C for 2 h in an autoclave with CO and CO2, cesium oxalate was obtained at a yield of up to 90.1% on a charged carbonate basis.20) However, the reaction needed high partial pressure of 110 atm and 50 atm for CO2 and CO, respectively. Recently, Banerjee and Kanan21) reported that the presence of H2 promoted the conversion of CO2 with Cs2CO3. Under mild conditions of 320°C and 60 atm, the carbonate was converted to multicarbon carboxylates, in which formate and oxalate were dominant. In both studies, 13C-labbeling experiments revealed that CO2 was consumed as the feedstock. The reaction hardly occurred when sodium or potassium carbonate was used. The same research group22) applied salt chemistry to the insertion of CO2 into aromatic hydrocarbons. Dispersing alkali (potassium or cesium) carbonates over mesoporous titania engendered their strong base reactivity, enabling the reaction of aromatic hydrocarbons, such as benzene, with CO2 and methanol to form methyl benzoate.
Prompted by these studies, we investigated the synthesis of oxalate from CO2 with Cs2CO3 supported over porous carbon in this work. The dispersion of carbonate altered the product distribution as well as reaction rate, resulting in the proposal and experiments of two-step synthesis of the oxalate.
Cs2CO3 was purchased from Wako Pure Chemical Industries. Activated carbon (AC) was mainly used as support for dispersing Cs2CO3. AC (Granular charcoal) was purchased from Wako Pure Chemical Industries and used after crushing and sieving to sizes of 500–1000 μm.23) The specific surface area, measured by N2 ad/desorption at −196°C, was 1250 m2/g. For comparison purposes, nano powders of carbon nanofiber (CNF) (Wako Pure Chemical Industries; 24 m2/g, d 100 nm × l 20–200 μm), ZrO2 (Sigma-Aldrich; 25≥ m2/g, <100 nm), Al2O3 (Sigma-Aldrich; >40 m2/g, <50 nm), and TiO2 (45–55 m2/g, <25 nm) were also used as the support materials.
The supported Cs2CO2 was prepared by an impregnation method. In a typical experiment, 1.0 g of the support and 2 mL of water containing 2.0 mmol of Cs2CO3 with a small amount of methanol or ethanol for better dispersing the support in water were loaded in a glass bottle, followed by the removal of solvent in a rotary evaporator operated at 60–70°C and 0.05–0.20 atm. The impregnation conditions were adjusted, depending on the nature of materials, for the complete loading of Cs2CO3.
2.2. Oxalate SynthesisThe synthesis of oxalate was carried out in an autoclave (Parr, model 4561). A glass insert, which was designed to fit inside diameter of the autoclave, was loaded typically with 2.0 mmol of Cs2CO3 and 1 mL of water, and then heated at 150°C for 2 h to hydrate the salt.21) The glass insert with hydrated Cs2CO3 (supported or unsupported) was set to the autoclave, loaded with CO2 and H2, and heated to a prescribed temperature at 5°C/min. The zero time of reaction was defined as the time when the desired temperature was reached. For example, in a typical experiment at 320°C, the heating time of 60 min from room temperature (20°C) was not considered in the reaction time. After the prescribed reaction time, the autoclave was naturally cooled and purged with N2. Carboxylates in the salt were recovered as organic acids by adding 5 mM H2SO4 to the glass insert, aging for two days, and conducting filtration under reduced pressure using PTFE membrane filter with a pore size of 0.45 μm. The experiments with different gas compositions and pressures were carried out in a similar manner. The gas pressure, mentioned hereafter, is determined on a room temperature basis.
The oxalate synthesis experiment was also carried out using a gas-flow type reactor. 2.0 mmol of Cs2CO3 supported over 1.0 g of AC was packed in a 19.0 mm tube reactor made of a stainless steel. The reactor was heated to a prescribed temperature (320°C or 380°C) under the flow of CO2/H2 (50/50%) or CO2/CO (50/50%) at 100 mL/min. The temperature was controlled with a thermocouple located beside the packed bed. The total pressure was controlled at 32 atm by a back pressure regulator located at the reactor downstream. After the reaction, AC was recovered from the reactor and washed with 5 mM of H2SO4. When the experiment involved two-step conversion to oxalate, the reactor was cooled to a room temperature after the first-step conversion under CO2/H2 flow and then directly used for the second-step conversion after replacing the gas with CO2/CO.
The 5 mM H2SO4 aqueous solution containing carboxylic acid was analyzed using high-performance liquid chromatography for the quantification of oxalic acid and formic acid concentrations. The analysis employed 5 mM of H2SO4 (0.6 mL/min) as the mobile phase, a Bio-rad HPX-87H column, and a column temperature of 40°C. The oxalate and formate yields were presented on a Cs2CO3-carbon basis (%, mol-C/mol-Cs2CO3). For example, when 0.5 mmol of oxalic acid was obtained in the H2SO4 aqueous solution from 2.0 mmol of Cs2CO3, the yield of oxalate was 50%.
The oxalate synthesis from Cs2CO3 with CO2 was carried out in an autoclave. The experimental results are shown in Figs. 1 and 2. The formation of oxalate was confirmed. The reaction also produced formate, which serves as an intermediate between carbonate and oxalate.21) During the heating of Cs2CO3, H2 is deprotonated to form hydride (H–), which reacts with CO2 to form formate. Subsequently, formate is deprotonated to carbonite (CO22–) and react with CO2 to form oxalate. They reported the formation of other carboxylates such as acetate, propionate, and succinate in relatively small amounts.21) Also in the present study, the liquid chromatography showed peaks derived from compounds other than formic acid and oxalic acid. However, the other carboxylates are not discussed in this study because the focus is on oxalate. Formate is important to quantify as its generation and reaction remarkably affect the oxalate formation.
Synthesis of oxalate with unsupported Cs2CO3. Conditions: Cs2CO3 2.0 mmol, CO2:H2 = 16:16 atm, 320°C.
Influence of water addition on synthesis of oxalate with unsupported Cs2CO3. Water in the experiment of leftmost plots came from atmospheric moisture. Conditions: Cs2CO3 2.0 mmol, CO2:H2 = 16:16 atm, 320°C, 30 min.
In the comparison with their report,21) the present result showed lower yields of formate and oxalate at a short reaction time. For example, the total yields at 5 min were 111% (recalculated to the yield defined in this work) and 29.1% in the report and this work, respectively, despite apparently similar reaction conditions. The time taken for heating the reactants to 320°C (1.15 h and 1.0 h, respectively) could be one of the reasons for this difference, although the exact reason was unclear. It should be noted that the total yields of over 100% on the carbonate-carbon basis are reasonable because CO2 is involved in the product carboxylates. On the other hand, these studies showed similar trends. Regardless of the long heating-up time, a reaction time of 5 min at 320°C was not sufficient to reach the available maximum yields, indicating a slow reaction rate or the influence of mass transfer. The yield of formate was higher than that of oxalate.
In addition, the influence of water addition was confirmed (Fig. 2). In this experiment, water was present as hydrate water (as explained in the experimental section) or added further to the glass insert before the reaction. Banerjee and Kanan21) explained the reason for the change in carboxylates yield with the improvement of ion mobility by the presence of water as hydrated water or vapor. Because 200 mg of water addition likely sufficed for the yield improvement while avoiding undesired reactions such as carboxylates decomposition, this amount was used in all experiments in this study unless otherwise stated.
Nevertheless, the synthesis of oxalate, based on this method, indicated some important drawbacks. Cs2CO3 before the reaction was a fine powder, but the salt obtained after the reaction adhered to the bottom of the glass insert as a solid layer, which was difficult to recover even by scratching. As a result, it was necessary for recovery to soak the salt layer in H2SO4 aqueous solution for a long time. This was caused by the low melting point of formate (263°C).24) In other words, upon the formation from carbonate, formate was liquefied, dissolving unreacted carbonate and generated oxalate. In such a case, the contact interface between gas and salt was limited to the liquid surface. This physical effect could be the reason for the slow increase in formate and oxalate yields. Therefore, we investigated the influence of dispersing Cs2CO3 over the porous support on the reaction. The dispersion was thought to be maintained even when the salt melted. The supporting of salt over materials was beneficial to handling of the salt after the reaction and avoiding potential corrosion of reactor materials. It also enabled the employment of the gas-flow packed bed reactor for this reaction.
3.2. Influence of Supporting Cs2CO3 Over Porous CarbonAs a preliminary experiment, several types of support materials were compared at a reaction time of 5 min. The reaction conditions were the same as those used in the experiment depicted in Fig. 1, except for the support of Cs2CO3. The result is shown in Fig. 3. Significant carboxylates yields were achieved by the carbon materials, AC and CNF. The yields were 101.8% and 98.4%, respectively, which were much higher than that in the use of unsupported Cs2CO3 (29.1%). In contrast, supporting over metal oxides rather decreased the yield. During the impregnation process, TiO2 and ZrO2 particles stuck to one another easily, forming a pseudo-agglomerate. The low dispersion of Cs2CO3 was apparent. SiO2 could react with Cs2CO3 as it formed a solid agglomerate after the reaction. The employment of metal oxides as the support material, thus, should consider the presence of acid or base sites that potentially react with alkali salts. Al2O3 did not show such agglomeration trends, and, as a result, the carboxylate yield was the highest among metal oxide supports. However, the yield was at a level similar to that of unsupported Cs2CO3, indicating the low dispersion of Cs2CO3. The porous structure of the support was likely important for the good dispersion of Cs2CO3. Compared to the high surface area of AC (1250 m2/g), the surface area of metal oxides was below 50 m2/g. Because the loaded 2.0 g of Cs2CO3 occupied 32% of the pore of AC (total pore volume: 0.5 cm3/g), the pore volume of those metal oxides was clearly insufficient for holding the salt. An exception was CNF, which showed a high carboxylates yield despite its low surface area (24 m2/g). Supposedly, the fibrous nature provided a large void space between particles, allowing for the dispersion of Cs2CO3. The above-mentioned work by Xiao et al.,22) where mesoporous titania-supported Cs2CO3 worked well for methyl benzoate synthesis, confirms the positive influence of porous support even when the material is metal oxides.
Synthesis of oxalate with Cs2CO3 supported over different support materials. Conditions: Cs2CO3 2.0 mmol, support 1.0 g, CO2:H2 = 16:16 atm, 320°C, 5 min.
Supporting Cs2CO3 over porous material was, thus, effective for promoting the conversion to carboxylates. AC was employed as the support because it showed the best yield. The most important problem with this reaction was that carboxylates were dominated by formate, but not by the oxalate desired in this work. The fraction of oxalate in carboxylates (formate and oxalate) in the 5 min-long experiment was decreased from 38.8% to 4.7% by supporting over AC. Several potential factors were examined to improve the oxalate yield as follows.
First, we concerned about the possible side effect from oxygen-containing functional groups over AC. The AC was prepared by steam-activation of a palm shell at the manufacturer. Though not quantified in this work, the material was supposed to have functional groups. To remove the functional groups, AC was treated under a flow of H2, according to a reported method.25) The treated AC is denoted by AC-r. However, as shown in Fig. 3, the treatment had little influence on the carboxylates yields and distribution. The result showed the oxygen-containing functional groups were not the main factor determining the predominant formate presence. This is confirmed by the dominant presence of formate in the product obtained with CNF (Fig. 3) consisting of graphitic carbons with scarce functional groups.
The influence of reaction conditions was investigated. The results are summarized in Fig. 4. The reaction time and the amount of Cs2CO3 loaded over AC had little influence in the range examined. The small decrease and increase in the total carboxylates and oxalate yields, respectively, with loading amount show that the reaction conditions approach to those of unsupported Cs2CO3 because the loaded amount of 5 mmol is close to the maximum holding capacity of pores in AC. The temperature significantly affected the total carboxylates yield, resulting in 114.1% at 400°C in 5 min, but failed to promote the oxalate formation. The most effective factor for improving the fraction of oxalate in carboxylates in this parametric study was the CO2/H2 ratio, where the total pressure was fixed at 32 atm. H2 is necessary for formate formation from carbonate while unnecessary for the formate conversion to oxalate. As a result, the oxalate fraction increased by up to 27.2% at CO2:H2 = 30:2, in return for a large decrease in the total carboxylates yield. It was, thus, likely difficult to improve both of the fraction and yield of oxalate with these reaction conditions.
Influence of reaction conditions on synthesis of oxalate with Cs2CO3 supported over AC. General conditions: Cs2CO3 2.0 mmol, AC 1.0 g, CO2:H2 = 16:16 atm, 320°C, 5 min.
The influence of the type of alkali metals in carbonate salt was investigated (Fig. 4(e)). In the order of increasing the atomic number of alkali metals, the total carboxylates yield was increased. Only Rb2CO3 and Cs2CO3 showed acceptable yields. This trend agreed with that reported in the work by Banerjee and Kanan,21) where unsupported Rb2CO3 and Cs2CO3 showed much higher carboxylates yield than K2CO3. Of particular importance is that the thermal coupling of formate to oxalate preferentially occurs with lighter alkali metals, Li–K.19) In other words, the low conversion from cesium formate to the oxalate was rather reasonable when only the thermal coupling was considered in the reaction of the present study. The mechanism of oxalate formation in the Cs2CO3/H2/CO2 system was explained by the deprotonation of formate anion (HCO2–) with carbonate anion (CO32–) to form carbonite (CO22–) that reacted with CO2 to form oxalate anion (C2O42–).21) With this in consideration, we suspected the possibility that a lack of CO32– during formate conversion to oxalate caused the low oxalate yield. The acceleration of carbonate conversion to formate by supporting over AC was evident. When this reaction was much faster than formate conversion to oxalate, it might cause a shortage of CO32– during the latter reaction. To investigate this possibility, 2.0 mmol of Cs2CO3 was added to the AC-supported salt, rich in formate, recovered from the reaction under the conditions Cs2CO3 2.0 mmol, 320°C, 5 min, and CO2:H2 = 16:16 atm, followed by re-heating under the same conditions. However, the experiment re-produced the carbonate-based formate and oxalate yields. Another attempt was adding Na2CO3 to the recovered AC-supported salt and reheating it in the same way. We expected the dissolution of salts with melted formate caused the ion exchange of cations, leading to an acceleration in the overall reaction. Sodium formate and Cs2CO3 are prone to forming oxalate and formate, respectively. In contrast to expectations, Na2CO3 behaved as inert, and formate and oxalate were obtained only from Cs2CO3.
3.3. Two-step Conversion of Supported Cs2CO3The failure to improve the fraction and yield of oxalate indicated that the reaction conditions, especially the gas composition and presence of carbon support, were unfavorable to the oxalate formation, though a detailed mechanistic study of the thermochemistry was not pursued in this work. We sought to investigate the possibility of a stepwise conversion to oxalate, where AC-supported formate-rich salt, produced in the first step from Cs2CO3, was subjected to re-heating under different gas composition. As mentioned above, cesium formate hardly serves as precursor to oxalate. Shishido et al.12) reported that the main reaction of cesium formate during the thermal treatment under atmospheric N2 was its decomposition to carbonate with the release of H2 and CO. To control the reaction, the thermal coupling of formate was carried out under pressurized CO and CO2. In Step 1, Cs2CO3 supported over AC was treated with the conditions Cs2CO3 2.0 mmol, 320°C, 5 min, and CO2:H2 = 16:16 atm. After cooling to room temperature and replacing the gas with CO and CO2, the formate-rich salt supported over AC was reheated as Step 2. Figure 5 shows the carboxylates yield after Step 2. Generally, the total carboxylates yield was not altered by reheating. Meanwhile, the fraction of oxalate significantly increased, reaching 54.9% with a yield of 55.0%. The yield was higher than that obtained in the experiment with unsupported Cs2CO3. The result clearly showed the occurrence of thermal formate coupling. As seen from the low carboxylates yield at CO2:CO = 16:0, the presence of CO contributed to suppression of the formate decomposition to carbonate. When Step 2 was carried out only with CO (16 atm), considerable carbon deposition over the sample and glass insert occurred. This was caused by the decomposition of CO into CO2 and carbon.12) CO2 added in this study successfully suppressed this reaction.
Reheating of AC-supported salt obtained from Step 1 under CO2 and CO. General conditions: CO2:CO = 16:16 atm, 390°C, 60 min. Step 1 was performed under conditions of Cs2CO3 2.0 mmol, AC 1.0 g, CO2:H2 = 16:16 atm, 320°C, 5 min.
The two-step conversion was carried out in a gas-flow mode with a packed bed of Cs2CO3 supported over AC. An ideal system utilizing the stepwise conversion with the packed bed is presented in Fig. 6. The system consists of three steps. The Step 1 and Step 2 are the conversion of carbonate to oxalate. CO and CO2 were assumed to be supplied only for controlling the equilibrium and unchanged by passing through the packed bed, although the possibility of their consumption was not necessarily eliminated by the present experimental results. In Step 3, C2O42– in the salt leaves the reactor as alkyl ester, such as methyl oxalate, after the reaction with alcohol. Simultaneously, Cs2CO3 is regenerated by the presence of CO2 for the reuse in Step 1. The possibility of these reactions occurring in Step 3 is indicated by the report on methyl benzoate synthesis22) and a recent patent,26) where cesium oxalate was esterified in methanol under pressurized CO2. Hydrolysis of the ester finally produces oxalic acid. The overall reaction is represented by 2CO2 + H2 → H2C2O4. By switching the gas supply in the order of Step 1–3, oxalic acid is continuously produced, although there are technical challenges such as tuning the reaction time.
Proposed reaction system for continuously producing oxalic acid using packed bed of supported Cs2CO3. (Online version in color.)
Step 3 and gas analysis fell outside the scope of this work. The carboxylates yields in Step 1 and Step 2 are shown in Fig. 7. In Step 1, the total carboxylates yield (64.0%) was lower than that in the autoclave (101.8%). The yield decrease was possibly caused by an absence of water in either of hydrate or vapor forms because water was purged from the reactor upon vaporization under the gas flow. This was a critical difference between the autoclave and gas-flow reactor. Meanwhile, the fraction of oxalate in carboxylates (20.6%) was higher than that in the autoclave (4.7%). Figure 2 suggests that the relatively high oxalate fraction was also caused by the absence of water. Notably, the positive influence of a water-free atmosphere likely contributed to Step 2, resulting in the highest oxalate fraction achieved in this work (71.8%). During Step 2, the total carboxylates yield was maintained. An oxalate yield of 43.2% was higher than that in the experiment with unsupported Cs2CO3. The results indicate the additional supply of water vapor only during Step 1 will increase the overall oxalate yield. Future work will focus on demonstrating this as well as other parts of the proposed system (Fig. 6) experimentally.
Stepwise conversion of Cs2CO3 supported over AC in gas-flow type reactor. Conditions of Step 1: Cs2CO3 2.0 mmol, AC 1.0 g, CO2:H2 = 16:16 atm (100 mL/min in total), 320°C, 5 min. Conditions of Step 2: CO2:CO = 16:16 atm (100 mL/min in total), 380°C, 60 min.
The synthesis of oxalate by heating of Cs2CO3 in the presence of CO2 and H2 as reactants was experimentally investigated with a particular focus on the influence of supporting Cs2CO3 over porous materials. Based on the experiments conducted, the following conclusions were drawn:
• The experiment with unsupported Cs2CO3 in the autoclave confirmed the formation of oxalate scaffold in the salt but posed some problems including a slow reaction rate, relatively low oxalate yield, and difficulty in handling of the salt after the reaction, which was caused by the melting of intermediate formate.
• Among the tested support materials, carbon materials, AC in particular, generated a significant carboxylates yield due to the high surface area and pore volume. Dispersion of Cs2CO3 over AC contributed to the acceleration of the reaction, resulting in a yield increase from 29.1% to 101.8% in 5 min at 320°C. However, the fraction of oxalate in carboxylates was only 4.7%. The low oxalate fraction was unchanged by reaction conditions such as temperature, reaction time, amount of loaded Cs2CO3, and partial pressure of CO2 and H2.
• The two-step conversion was effective for improving the oxalate fraction while maintaining total carboxylates yield. In Step 2, CO effectively suppressed the decomposition of formate to CO and carbonate, and the carbon deposition from CO could be avoided using CO2. The highest oxalate fraction and yield were 54.9% and 55.0%, respectively.
• By employing supported Cs2CO3, the reaction in the gas-flow type reactor was successfully carried out. Due to the quick release of water from the reactor, the total carboxylates yield decreased to 64.0%. Meanwhile, it resulted in an increase in the oxalate fraction during Step 2. The oxalate fraction and yield were 71.8% and 43.2%, respectively.
This work was financially supported by New Energy and Industrial Technology Development Organization (NEDO), Japan, for the Feasibility Study Program on Uncharted Territory Challenge 2050. A part of this work was financially supported by The Iron and Steel Institute of Japan for 2020–2024. The authors are also grateful to the Cooperative Research Program of Network Joint Research Center for Materials and Devices, supported by the Ministry of Education, Culture, Sports, Science, and Technology (MEXT), Japan.
