One-step synthesis of visible light CO2 reduction photocatalyst from carbon nanotubes encapsulating iodine molecules

We describe the synthesis and visible-light CO2 photoreduction catalytic properties of a three-component composite consisting of AgI, AgIO3, and single-walled carbon nanotubes (SWCNTs). The catalyst is synthesized by immersing SWCNTs encapsulating iodine molecules in AgNO3 aqueous solution, during which neutral iodine (I2) molecules encapsulated in SWCNTs transform disproportionately to I5+ (AgIO3) and I− (AgI), as revealed from the characterization of the composite by Raman spectroscopy, X-ray diffraction, and X-ray photoelectron spectroscopy. In addition, photoirradiation experiments using a solar-simulator (AM1.5G) showed that the obtained three-component composite works as a CO2 photoreduction catalyst under visible light despite the wide band gap of AgIO3, suggesting possible transfer of the visible light-excited electron from AgI via SWCNTs.

Many of the above-mentioned molecules are inserted into SWCNTs in a gas-phase process, in which the molecules should be sublimed and deposited in SWCNTs. However in 2013, our group reported another type of encapsulation, in which we inserted iodine molecules in SWCNTs by electro-oxidation of iodide ions 22 . This method was not only very easy and scalable, but also very effective, because we can control the amount of the encapsulated molecules just by changing the electrolysis time or electric current. We can insert iodine molecules by applying positive potential to the SWCNTs electrode in an electrolytic solution including iodide ions, then extract the inserted molecules just by changing the direction of the potential. The reversible insertion and extraction of iodine molecules can be used as battery electrode reactions 23,24 .
The investigation of the structural properties of the iodine molecules encapsulated in SWCNTs (I@SWCNT) have been investigated by many experiments such as TEM, X-ray absorption fine structure (XAFS) spectroscopy, and Raman spectroscopy [22][23][24][25] . These experiments revealed that the structure and electronic properties of the iodine molecules inserted by the electro-oxidation of iodide ions were identical to those of the molecules inserted by gas phase reaction. Iodine molecules were inserted in SWCNTs in the form of I 2 regardless of the insertion method. We also observed that after the insertion, charge transfer from SWCNTs to iodine molecules causes some of the iodine molecules to convert to polyiodide ions (I 3 − , I 5 − etc.). This was shown from the Raman spectrum of I@SWCNT, where characteristic Raman peaks of polyiodide ion were observed in the low wavenumber region. It is also known that this charge transfer reaction is strongly affected by temperature 25 .
Although the structural properties of the iodine molecules of I@SWCNT are well investigated, their chemical reaction properties are still not well understood. However, it is plausible that the iodine molecules should be reactive because they are energetically unstable in a restricted space. During our attempts to investigate the chemical reaction properties, we obtained a three-component composite consisting of AgI, AgIO 3 and SWCNTs, in which I 5+ ions in AgIO 3 and I − ions in AgI were simultaneously produced from I@SWCNT. In the present study, we describe this disproportionation reaction, and the structural and photocatalytic properties of the composite.

Experimental
We purchased an SWCNTs sample from Meijo Nanocarbon Co., Ltd (EC type). We performed a purification acid treatment using to remove metallic impurities from the sample. The detailed purification procedure is described in our previous papers 22, [24][25][26] . The removal of metal impurities was confirmed by thermogravimetric (TG) analysis. Iodine molecules were inserted into SWCNTs by electrochemical oxidation of iodide ions in electrolytic solution. To achieve electrochemical iodine insertion, we fabricated a three-electrode cell using paper-form SWCNTs as the working electrode, Pt as the counter electrode, and Ag/AgCl as the reference electrode. NaI aqueous solution (1 M) was used as the electrolyte, and we applied 1.8 V to the SWCNTs electrode for 15 min. After the encapsulation treatment, we washed the SWCNTs samples by distilled water to remove the iodine molecules deposited on the outer surfaces of SWCNTs, and dried the washed samples. The amount of the encapsulated iodine molecules was determined by TG measurements. We also performed Raman measurements to check the crystallinity. The nanostructure of the obtained samples was observed using a transmission electron microscope TEM (JEOL JEM-z2500) operated at 200 kV. The sample of SWCNTs encapsulating iodine molecules is denoted I@SWCNT.
To prepare the composite catalyst, we immersed the paper-form I@SWCNT sample in 20 mM AgNO 3 aqueous solution for 10 h. The SWCNTs paper was then taken out and washed with distilled water. The washed SWCNTs sample was dried and characterized by TEM, SEM, XRD, XPS and Raman measurements. As will be discussed in the following section, the characterization revealed that the obtained sample is a composite consisting of AgI, AgIO 3 , and SWCNTs.
We investigated the CO 2 photoreduction properties of the composite. The photocatalytic reduction of CO 2 was carried out in an airtight cell using a solar simulator (XES-40S2-CE of SAN-EI ELECTRIC Co.) with power density of 100 mW/cm 2 was used in a mode of AM1.5G. Prior to the irradiation, CO 2 gas bubbling for 30 min was performed to presaturate the 0.5 M KOH aqueous solution with CO 2 . Gaseous samples from the airtight

Results and discussion
Preparation and characterization of AgI-AgIO 3 -SWCNT composite. It was confirmed by TEM and SEM observations that the SWCNTs sample had low impurity content ( Fig. S1a-c). Raman measurement also showed that the SWCNTs sample has good quality, because the intensity of D-band relative to G-band is quite weak (Fig. 2a-(i)). The peak positions of the radial breathing modes (RBM) observed in the low wavenumber region of the Raman spectrum can be used to determine the diameter distribution of the SWCNTs in the sample. However, it should be noted that only a part of SWCNTs having the gap energy of van Hove singularities close to Raman excitation laser energy in the sample mainly contributed to the observed Raman spectrum by resonance process. Therefore, in order to obtain a reliable diameter distribution by Raman measurements, we should measure the Raman spectra using different energy excitation lasers. Judging from , and those molecules are encapsulated in SWCNTs. However, after the encapsulation, charge transfer from SWCNTs to I 2 molecules occurs and some of the I 2 molecules are converted into polyiodide ions. In fact, even if we insert I 2 molecules into SWCNTs by gas phase treatment, almost the same Raman spectrum having polyiodide ion peaks is observed (Fig. S2). As shown in Fig. 2a-(ii), the G-band peak position shifted toward a higher wavenumber upon the encapsulation of I 2 molecules. The magnitude of the shift relates to the amount of the encapsulated I 2 molecules. The iodine content in I@SWCNT was evaluated to be about 45% by TG measurement (Fig. S3). Figure 2a-(iii) shows the Raman spectrum of the sample obtained by the immersion of I@SWCNT in AgNO 3 aqueous solution. The polyiodide ion peaks of I@SWCNT are not observed in Fig. 2a-(iii). On the other hand, we can see the G-band of SWCNTs, which has almost the same peak position as the pristine sample, in addition to many peaks in the range100-800 cm −1 . That indicates that iodine molecules encapsulated in SWCNTs reacted with AgNO 3 aqueous solution. For convenience, the reacted SWCNT sample (Fig. 2a-(iii)) will be abbreviated as Ag-I-SWCNT. The inset of Fig. 2a shows the Raman peaks in the low wavenumber region of the spectrum of the reacted SWCNT sample. Compared to the reported Raman patterns of some iodine related compounds, the peaks in the range 150-800 cm −1 are found to be of AgIO 3 . The two strong Raman peaks of AgIO 3 observed around 720 and 755 cm −1 can be assigned to the A1 and E modes of IO 3 − ion. On the other hand, the peaks observed in the range 50-130 cm −1 are identical to those observed for AgI powder sample. Therefore, the observed Raman  Figure 2b shows the XPS spectra of Ag-I-SWCNT. The observed peak position of the Ag 3d 5/2 binding energy is about 367.7 eV, which indicates positively charged Ag because the value is lower than the binding energy of Ag metal. On the other hand, and as shown in Fig. 2c, two kinds of iodine species exist in Ag-I-SWCNT. The two peaks at 623.7 and 635.2 eV can be assigned as I 3d 5/2 and I 3d 3/2 of I 5+ , while the peaks at 619.2 and 630.7 eV are attributed to I − . Therefore, the I 3d binding energies also indicate the simultaneous formation of AgI and AgIO 3 . Analyzing the XPS peak intensities of I 5+ and I − , we calculated a content ratio of AgIO 3 to AgI of about 2.2. One possible reaction for the simultaneous formation of AgI and AgIO 3 is given by the equation: However, the content ratio for the reaction above should be 0.5. This discrepancy leads us to hypothesize that more complicated reactions (e.g. subsequent reaction of AgI) are involved in the synthesis process, although we currently do not have any evidence on the nature of such reactions.
If I 2 molecules exist in the Ag-I-SWCNT sample, the I 3d 5/2 peak of I-should have a shoulder peak or at least become broader, because I 3d 5/2 of neutral I should be observed at around 620 eV, which is close to that of I − . However, the peak profile of I 3d 5/2 peak of I − is sharp and symmetrical. Therefore, I 2 molecules do not exist in Ag-I-SWCNT sample, and it is plausible that all the iodine molecules reacted to AgI or AgIO 3 . This is consistent with the observation that polyiodide ion Raman peaks of I@SWCNT completely disappeared after the reaction with AgNO 3 aqueous solution.
The SEM image of Ag-I-SWCNT sample (Fig. 2d) shows some deposited material on SWCNTs, but the deposits are not very clear. As mentioned above, the iodine content of I@SWCNT was about 45 wt.%. Assuming that all the iodine molecules reacted to form AgI and AgIO 3 , and that the deposits on SWCNTs consists of only these two crystals having the molar ratio of 1:2.2, the crystal volume of the deposition should be very small compared to SWCNTs volume. We performed XRD measurements to identify the deposits. As shown in Fig. 2e, we could not observe sharp diffractions, which means that the deposited materials should be aggregations of very fine crystals or amorphous materials. Amorphous AgI and AgIO 3 are quite unlikely to be formed. In fact, in a different synthesis experiment, very weak diffraction peaks corresponding to the diffractions of AgI were observed (Fig. S4). Therefore, it is reasonable to assume that very fine crystals of AgI and AgIO 3 are formed in the synthesis of Ag-I-SWCNT sample.
Up until now, we found that fine crystals of AgI and AgIO 3 on SWCNTs can be prepared just by immersing I@ SWCNT in AgNO 3 aqueous solution. Although the detailed mechanism of the reaction is not yet clear, the disproportionation reaction from I 0 to I 5+ and I − occurs very smoothly for iodine molecules encapsulated in SWCNTs.

Photocatalytic properties of AgI-AgIO 3 -SWCNT composite.
In this section, we describe the solar light CO 2 reduction properties of AgI-AgIO 3 -SWCNT. The photocatalytic reduction of CO 2 was carried out in an airtight cell containing CO 2 -saturated 0.5 M KHCO 3 aqueous solution. Simulated solar light (AM1.5G, 100 mW/cm 2 ) was irradiated for 2-20 h. After the photoirradiation, the gas in the airtight cell was analyzed by gas-chromatography. As shown in Fig. 3b, a peak that corresponds to CO was observed in the gas-chromatogram. On the other hand, CH 4 was not detected by the chromatography. HCOOH is often produced by the reduction of CO 2 . However, in the present study, we could not detect HCOOH by NMR measurement of the aqueous solution of the airtight cell after the photoirradiation experiment. That does not mean that no HCOOH was produced by the CO 2 reduction, because our NMR can detect HCOOH only in the case that the HCOOH concentration is more than 60 μmol/L. Since we could not see any CO peaks in the chromatogram for the blank test without www.nature.com/scientificreports/ the SWCNTs composite, we conclude that the composite works as a solar light CO 2 reduction photocatalyst. Figure 3a shows the time dependent CO yields of this photocatalyst. As shown in Fig. 3a, the reduced amount is proportional with photo-irradiation time. The CO conversion efficiency is calculated to be 0.18 μmol/(g·h). We also investigated the chemical stability of AgI-AgIO 3 -SWCNT composite (Fig. S5). As shown in Fig. S5, the composite can work as photocatalyst at least up to 72 h, although the catalytic ability slightly decreased at 72 h. As mentioned in the introduction section, we think that the photoexcited electron of AgI is transferred to AgIO 3 via SWCNTs, and that the transferred electrons reduce CO 2 to CO. In order to confirm that, we performed photoreduction experiments with a UV-cut filter (HOYA L42) that absorbs light with wavelengths less than 400 nm. In this condition, direct photoexcitation of AgIO 3 is impossible because the band gap of AgIO 3 is about 3.4 eV. Even in this case, the SWCNTs composite was able to reduce CO 2 effectively (see Fig. 3b). On the other hand, in the case of only AgI sample (control experiment), only a limited amount of CO was detected (Fig. 3b). Namely, the electron transfer shown in Fig. 1 was confirmed. Furthermore, we also detected O 2 generation when CO was detected by the photo-irradiation of the AgI-AgIO 3 -SWCNT composite (Fig. S6). It means that the reduction of CO 2 to CO and oxidation of water to O 2 occur in the photocatalytic process. The corresponding chemical reactions are summarized in Fig. S7. Similar electron transfer is already reported by Zheng for an Ag-AgI-AgIO 3 system that also reduces CO 2 to CO under visible light 11 . Zheng reported that the visible light-excited electrons of AgI are transferred to AgIO 3 via Ag 11 . So, in their case, Ag worked as the electron transfer medium similarly to SWCNTs in our study. However, it is obvious that SWCNTs having fiber form are better as electron transfer medium because they effectively connect AgI and AgIO 3 .
Another merit of the SWCNTs composite is the easy fabrication of a flexible transparent conductive film of the CO 2 photoreduction catalyst (Fig. 3c). The Transparent conductive film is easily prepared just by spray coating SWCNTs on a transparent polymer film or glass sheet. The obtained conductive film can be used as the electrode for subsequent iodine insertion, which is achieved easily by the electrooxidation encapsulation technique. After the encapsulation, we simply immersed the film in AgNO 3 to prepare the three-component photocatalyst Ag-I-SWCNT as a transparent conductive film. This expands the scope of application of the CO 2 photoreduction catalyst.

Conclusion
We synthesized a three-component photocatalyst composed of AgI, AgIO 3 , and SWCNTs. The synthesis starts by preparing iodine-encapsulating SWCNTs (I@SWCNT) sample with about 45 wt.% iodine through applying 1.8 V (vs. SHE) to paper-form empty SWCNTs electrode in 1.0 M NaI aqueous solution for 15 min. We then obtained the three-component composite of AgI, AgIO 3 , and SWCNTs (AgI-AgIO 3 -SWCNT) by simply immersing I@SWCNT in AgNO 3 aqueous solution. The obtained composite was characterized by Raman measurements, revealing that AgIO 3 exists in the composite, because a characteristic Raman peak pattern of AgIO 3 was observed in the spectrum of the composite. The Raman spectrum also indicated the existence of AgI in the composite, which was also confirmed by observing weak diffraction peaks of AgI in the XRD pattern and I 3d binding energies in the XPS spectra of the composite, all clearly indicating the coexistence of I 5+ and I − . These measurements revealed that neutral iodine (I 2 ) molecules encapsulated in SWCNTs transform disproportionately to I 5+ (AgIO 3 ) and I − (AgI). We found that the obtained three-component composite works as a CO 2 photoreduction catalyst under visible light, overcoming the issue of the otherwise wide bandgap of AgIO 3 , suggesting that the visible lightexcited electron of AgI is transferred to AgIO 3 via SWCNTs, and that the transferred electrons reduce CO 2 to CO.