Electrochemical Reactions of Iodine Molecules Encapsulated in Single-Walled Carbon Nanotubes.

We prepared iodine molecules encapsulated in single-walled carbon nanotubes (I@SWCNTs) by electro-oxidation of iodide ions with empty SWCNT electrode. Li-ion battery electrode properties of I@SWCNTs were investigated. It was found that the I@SWCNT sample can catch and release Li ions reversibly. We performed Raman measurements to reveal the Li-ion storage mechanism of I@SWCNT. It is plausible that chemical reactions of I2 from/into LiI in SWCNTs occur during Li-ion charging/discharging of I@SWCNT. We also prepared the CsI@SWCNT sample to verify that alkali metal ions can be extracted from alkali metal halide in SWCNTs. The extraction of cesium ions from CsI@SWCNT was confirmed by Raman measurements. It was also found that I@SWCNT can work as a Li-ion battery electrode in solid electrolyte as well.

Introduction

Lithium-ion batteries (LIBs) have been widely used in portable electronic devices such as cell phones and laptop computers, owing to their high energy density.[1,2] The last decade also witnessed the use of LIBs for large machines (e.g., electric vehicles). However, of course, LIBs have some problems to be solved to ensure the wider spread of electric vehicles. One of the problems is safety, as seen from the frequently reported fire and explosion accidents of LIBs.[3] Since the accidents are mainly caused by the use of flammable organic electrolytes, it is expected that the development of all-solid-state secondary batteries using solid electrolytes would contribute to solving the safety problem of LIBs.[4] On the other hand, an all-solid-state primary battery (lithium–iodine battery) is already commercially available.[5] However, it is very difficult to recharge the lithium–iodine battery because lithium iodide produced by the discharge of the battery is electrochemically inactive due to its low electric conductivity.

We have previously investigated the electrode properties of several kinds of molecules encapsulated in single-walled carbon nanotubes (SWCNTs).[6−8] Some of the molecules we tested do not show good electrode properties originally because of their low electric conductivities. However, by encapsulating them inside SWCNTs, they can work as electrode-active materials because SWCNTs provide good electron conductive path from the current collector to the encapsulated molecules.[9]

Iodine molecules (I2) can be encapsulated inside SWCNTs. We have developed a very easy encapsulation method using the electro-oxidation of iodine ions in SWCNTs.[9] We inserted I2 molecules in several kinds of SWCNTs having different mean tube diameters by the electro-oxidation method.[10] We have also reported the detailed structural properties of the encapsulated I2 molecules in SWCNTs at several temperatures.[10,11]

If iodine molecules encapsulated in SWCNTs (I@SWCNT) are used as electrode-active materials in LIB, the electrode reaction product (LiI) should be formed inside SWCNT hollow cores. However, it is very difficult to grow a three-dimensional rock-salt crystal of LiI in an SWCNT core because the diameter of SWCNT is very small. Therefore, unusual crystal or molecule-like forms of LiI are more likely to form inside SWCNT. Such unusual forms of LiI should be unstable and reactive. The reactive LiI inside SWCNT also has a good electron-transfer path from the host SWCNT. So, it should be possible to extract Li ions from such LiI entity encapsulated in SWCNT by electrochemical methods, namely, we would be able to charge and discharge I@SWCNTs. If that is possible, we can design a new all-solid-state battery using I@SWCNTs.

In this work, we found by electrochemical measurements that I@SWCNTs can charge and discharge Li ions reversibly. To discuss the electrode reaction mechanism, we also performed ex situ Raman measurements of I@SWCNT during the charge–discharge process.

Results and Discussion

We prepared the I@SWCNR sample by the electro-oxidation method. The encapsulation of iodine molecules in SWCNTs was confirmed by X-ray absorption near edge structure (XANES), Raman, X-ray diffraction (XRD), and other measurements. As previously reported, iodine molecules are encapsulated mainly as a form of I2 (Figure S3).[10,11] However, some iodine molecules are transformed to polyiodide ions such as I3– and I5– by charge transfer from SWCNTs to the encapsulated iodine molecules.[10] This is confirmed by the Raman spectrum of the prepared I@SWCNT sample (see the inset of Figure ).

Figure 1

Raman spectra of (a) pristine SWCNT and (b) I@SWCNT samples.

It was reported that the composition ratio of several kinds of polyiodide ions depends on the tube diameter of the host SWCNT and temperature. As shown in Figure , I5– is the dominant polyiodide ion in the present sample. If SWCNTs are well crystallized and the tube diameter distribution is very small, SWCNTs are gathered by van der Waals interaction and form a two-dimensional quasi-hexagonal bundle structure. In the case of the present SWCNT sample, we could observe clear diffractions of the bundle structure. It is well known that the most intense 100 diffraction peak on the observed diffraction pattern loses its intensity if guest molecules are inserted in SWCNTs. The intensity decrease of 100 diffraction peak clearly indicated the encapsulation of iodine molecules.[9] It is possible to estimate the amount of the encapsulated iodine molecules qualitatively from the decrease in the 100 diffraction peak intensity. However, to evaluate the amount quantitatively, thermogravimetric (TG) measurement is the necessary technique. TG measurement showed that the present I@SWCNT sample includes 39 wt % iodine molecules (Figure S4).

Here, we discuss the volumetric density of the encapsulated iodine molecules of the present I@SWCNT sample. The mean tube diameter of the present SWCNT sample is about 2.5 nm. For simplification, we calculated the volumetric density of the encapsulated iodine molecules of 39 wt %, assuming them to be encapsulated in an SWCNT of chirality (18, 18) that has a diameter close to the mean tube diameter in our study. We obtained the value of 0.8 g/cm3, which is much less than the iodine crystal density of 4.93 g/cm3, indicating that iodine molecules of the present I@SWCNT are not densely packed inside SWCNTs.

We performed Li-ion charge and discharge measurements of I@SWCNT (Figure ). Since potential plateaus are seen on both the charge and discharge curves, reversible adsorption and extraction of Li ions can occur at the I@SWCNT electrode. Assuming that I2 molecule reacts with Li ion in the manner shown below, the theoretical capacity of I2 molecule is calculated to be 211 mAh/g[12]As shown in Figure , the observed discharge capacity of about 150 mAh/g, which indicates that about three-fourths of the encapsulated I2 molecules are converted to LiI.

Figure 2

Charge and discharge curves of I@SWCNT observed at constant charge density of 40 mA/g, where “g” means weight of iodine in the electrode.

Figure A shows the cyclic voltammograms of the I@SWCNT electrode. Clear redox peaks were observed at a scan rate of 1 mV/s. Redox pairs marked with (a) and (b) are attributed to the Li insertion/extraction reactions shown in eqs and 2, respectively. Half-wave potential E1/2 can be obtained by averaging oxidation and reduction peak potential values. The E1/2 values of reactions (a) and (b) were about 3.6 and 2.9 V, respectively. These two values are a little higher than the observed two-step plateau potentials (Figure ). However, we should note that the plateau potential observed by charge–discharge measurement does not correspond to the equilibrium state. To know the two-step plateau potentials of equilibrium states of reactions (a) and (b), we have performed galvanostatic intermittent titration technique (GITT) measurement. The observed open-circuit potentials by GITT measurement are in fairly good accordance with the E1/2 values determined by CV measurement. As shown in Figure A, the peak intensity of (b) is about 2 times higher than that of (a). This can be explained by the difference of reaction electron numbers of eqs and 2. On the other hand, redox peaks are not clearly seen at higher scan rate (Figure B) probably because the redox reaction of the I@SWCNT electrode is very slow. Li+ diffusion length of the I@SWCNT electrode is very long (μm order) because Li+ cannot pass through the side wall of SWCNT. We also observed very large electrochemical impedance for both the lithiated and delithiated states of I@SWCNT electrode (Figure S5). These electrochemical observations support that the electrode reactions occur inside the carbon nanotubes.

Figure 3

Cyclic voltammograms of I@SWCNT: scan rates of (A) 1 mV/s and (B) 1–10 mV/s.

The above-mentioned redox reactions were more clearly seen in the differential chrono-potentiogram. Figure S6 shows the differential chrono-potentiometry graph for the discharge process (Li insertion process) of I@SWCNT. As shown in Figure S7, two peaks at about 3.35 and 2.85 V are clearly seen. These two peaks correspond to the two-step electrochemical reactions mentioned above. The discharge capacities for the former and latter reactions are about 30 and 70 mAh/g, respectively (Figure S8). This capacity ratio (30:70) is very close to the theoretical ratio (1:2) of the electric quantity required for the two-step reactions.

Figure S9 shows I 3d X-ray photoelectron spectroscopy (XPS) image observed for the discharged (Li-inserted) I@SWCNT sample. Although the observed spectrum is very noisy, two peaks of I 3d5/2 and I 3d3/2 are clearly seen. The binding energies of I 3d5/2 are 619.3 and 620.2 eV for I– (LiI) and I0 (I2), respectively.[13] The observed binding energy of the discharged sample is between the two values (Figure S9). This can be explained that about three-fourths of the encapsulated I2 molecules are converted to LiI, as mentioned above.

The capacity value did not decrease much at least within a few cycles (Figure S11). This means that most iodine molecules are kept in SWCNT hollow cores during charge–discharge cycles. We previously reported that several kinds of molecules (e.g., anthraquinone, sulfur, phosphorus) encapsulated in SWCNTs work as Li-ion battery electrode.[6−8] As with other molecules, the encapsulation of iodine molecules provides them with a good electrically conductive path from SWCNTs and good stability against dissolution in the electrolyte. The present charge–discharge experiment revealed that Li ions caught by iodine molecules in SWCNTs can be reextracted reversibly. On the other hand, the practical Li–iodine all-solid-state battery is a primary battery. We cannot recharge the discharged Li–iodine cell mainly because it is very hard to extract Li ions from LiI produced in the discharge process. However, and as we wrote above, I@SWCNT can charge and discharge Li ions reversibly, which indicates that the LiI formed inside the nanotubes is electrochemically active probably because of the good electron-transfer path from SWCNTs to LiI molecule-like products in SWCNTs. Usually, LiI is a rock-salt-type three-dimensional crystal. However, it would be very difficult to form three-dimensional crystals inside the nanotube, although rock-salt LiI having a density of 3.49 g/cm3 would only occupy 24% of the tube inner space, assuming that all iodine molecules of the present I@SWCNT sample are converted to LiI. Instead of a rock-salt-type three-dimensional crystal, it is probable that an electrochemically active “LiI molecule-like product” was produced inside the tube.

To confirm that iodine molecules encapsulated in SWCNTs can catch and release Li ions, we performed Raman measurements of I@SWCNT during the charge–discharge process. Figure A shows the low-wavenumber region of the observed Raman spectra. For an SWCNT sample, radial breathing modes (RBMs) are usually observed in the low-wavenumber region. It is well known that the peak position of RBM is inversely proportional to the tube diameter of SWCNT and that some semiempirical equations are used to describe the relationship between the tube diameter d and RBM peak position ωRBM (e.g., d = 248/ωRBM).[14−17] In the present study, the mean tube diameter of the SWCNT sample is about 2.5 nm and the expected RBM peak position is less than 100 cm–1, which is the experimental lower limit due to the edge filter used in the present Raman measurement. Therefore, only the tail of the RBM was observed in the range of 100–150 cm–1 of the spectrum (a) of the empty SWCNT sample in Figure A. On the other hand, two peaks at about 110 and 160 cm–1 were observed on the spectrum (b) of the I@SWCNT sample. The peaks at about 110 and 160 cm–1 correspond to I3– and I5–.[10] As discussed in the previous report, the majority of the encapsulated iodine molecules exist as I2. However, a small portion of I2 is transformed to polyiodide ions by charge-transfer reaction between SWCNT and I2. The observed Raman peaks of polyiodide ions are the indicator of the existence of iodine molecules in the I@SWCNT sample. In the previous section, we mentioned that iodine molecules are transformed to LiI when Li ions are inserted in I@SWCNT. This transformation is confirmed by the disappearance of the Raman peaks of polyiodide ions in the spectrum (c) of the discharged I@SWCNT sample. Strictly speaking, Raman peaks of polyiodide ions did not disappear completely because one-fourth of I2 molecules encapsulated in SWCNTs should remain as discussed in Figure . Interestingly, upon recharging the discharged I@SWCNT sample, the Raman peaks of polyiodide ions were again observed on the spectrum (d). This means that LiI produced by the discharge of I@SWCNT is converted to I2 by the recharging process.

Figure 4

(A) Low-wavenumber region and (B) G-band-region Raman spectra of (a) empty SWCNT, (b) I@SWCNT, (c) discharged (Li-ion-inserted) I@SWCNT, and (d) charged I@SWCNT samples.

Figure B shows the change in Raman G-band spectrum of I@SWCNT during the Li-ion charge–discharge process. The G-band peak position of SWCNTs shifts toward the higher-wavenumber side by encapsulating electron acceptors. It was observed in Figure B that the G-band peak position of empty SWCNT (spectrum (a)) shifts to 1590 cm–1 by iodine insertion (spectrum (b)). The G-band peak position of I@SWCNT shifted back to the lower-wavenumber side with Li-ion insertion (spectrum (c)). By Li-ion insertion to I@SWCNT, iodine molecules having high electron negativity encapsulated in SWCNT are converted to LiI having low electron negativity. This is the reason why the G-band peak position of the discharged I@SWCNT sample went back to a lower wavenumber. By recharging the discharged I@SWCNT sample, the peak position moved again to the higher-wavenumber side (spectrum (d)). This clearly shows that it is possible to extract Li ions from the LiI molecule-like products inside SWCNTs, which was also shown by the fact that the G-band shifts during the charge–discharge process are quite consistent with the appearance and disappearance of the polyiodide ion Raman peaks in the low-wavenumber region.

As discussed above, it was revealed that I@SWCNT can catch and release Li ions reversibly. Unfortunately, the electrode reactions inside the nanotubes were not fully clarified by spectroscopic measurements. We supposed that iodine molecules encapsulated in SWCNTs are converted to LiI molecule-like products by Li-ion insertion. If we prepare a sample of LiI encapsulated in SWCNT and achieve Li-ion extraction from the LiI@SWCNT sample, we can support the hypothesis that LiI is formed inside the nanotube by Li-ion insertion to I@SWCNT. To that end, we tried to prepare LiI@SWCNT by heating the reaction glass tube, in which SWCNTs and LiI powder samples were inserted under vacuum. However, the reaction glass tube was repeatedly ruptured during the heat treatment, probably due to the high reactivity of LiI. Therefore, we decided to replace LiI with CsI because the preparation of CsI@SWCNT (CsI encapsulated in SWCNT) was already done according to a previous work.[18] The encapsulation of CsI was confirmed by several characterization techniques such as transmission electron microscopy (TEM), scanning electron microscopy (SEM), XRD, and Raman spectroscopy as done in the previous work. The amount of the encapsulated CsI was determined to be 44 wt % from the sample weight increase with CsI insertion. We prepared a test cell consisting of the CsI@SWCNT sample as working electrode and Li metal as counter electrode and performed a charge experiment corresponding to Cs ion extraction from CsI@SWCNT (Figure ). Figure shows the observed charge curve that is quite similar to the charge curve of I@SWCNT in Figure . The redox potential of Cs (E0 = −2.92 V) is the same as that of Li (E0 = −3.05 V). This results in similar charge plateau potentials of CsI@SWCNT (Figure ) and I@SWCNT (Figure ).

Figure 5

Li-ion charge (Li-ion insertion) curve of CsI@SWCNT observed at constant charge density of 40 mA/g.

Figure 7

Charge–discharge curves of I@SWCNT observed with PEO solid electrolyte observed at constant charge density of 15 mA/g at 60 °C.

We also performed Raman measurements for CsI@SWCNT before and after electrochemical measurements as we did for I@SWCNT. As shown in Figure A,B, no significant change was observed in the low-wavenumber and G-band regions by encapsulating CsI in empty SWCNTs. On the other hand, we observed after the charge experiment the appearance of the Raman peaks of polyiodide ions and also a shift in the G-band peak (spectra (c) in Figure A,B). In the present charge experiment of Figure , we extracted Cs ions corresponding to the capacity of about 70 mAh/g. Assuming that I2 molecules are produced by the extraction of Cs ions from CsI@SWCNT, we estimate that about 70% of CsI in SWCNTs has transformed to I2 because the theoretical capacity of CsI is calculated to be 103.2 mAh/g.

Figure 6

(A) Low-wavenumber-region and (B) G-band-region Raman spectra of (a) empty SWCNT, (b) CsI@SWCNT, and (c) the charged (Cs ion extracted) CsI@SWCNT samples.

So far, we have discussed the charge–discharge experiments of I@SWCNT and CsI@SWCNT using 1 M lithium bis(trifluoro methanesulfonyl)imide (LiTFSI)/1,3-dioxolane (DOL) and dimethyl ether (DME) liquid electrolyte. Since the liquid electrolyte restrains the shuttle reactions of iodide-relating species, we can perform charge–discharge experiments of I@SWCNT in a relatively short period, as short as the present experiments. However, it is essentially very difficult to use iodine species as a battery electrode using liquid electrolyte because iodine species are easily dissolved in liquid electrolyte. We therefore investigated the electrode behavior of I@SWCNT using a solid electrolyte. Poly(ethylene oxide) (PEO)-containing LiClO4 ([O](PEO)/[Li] = 8:1) was used as the solid electrolyte. A test cell consisting of I@SWCNT as the working electrode, Li metal as the counter electrode, and PEO solid electrolyte was operated at 60 °C to achieve good ionic conductivity of Li ion in PEO. Figure shows the observed charge–discharge curves. We observed a relatively large reversible capacity, albeit smaller than that observed in liquid electrolyte. These results show that I@SWCNT can catch and release Li ions reversibly even in solid electrolyte. The reversibility was obtained probably due to the observed fact that the LiI molecule-like products in SWCNTs produced by Li-ion insertion to I@SWCNT are very active as we have discussed.

Conclusions

We prepared the I@SWCNT sample by the electro-oxidation of iodide ions with empty SWCNT electrode. Li-ion battery electrode properties of I@SWCNT were investigated, showing that the I@SWCNT sample can catch and release Li ions reversibly. We performed Raman measurements to reveal the Li-ion storage mechanism of I@SWCNT. It is plausible that chemical reactions of I2 from/into LiI in SWCNTs occur during the process of Li-ion charge–discharge performed on the I@SWCNT electrode. We also prepared the CsI@SWCNT sample to verify that alkali metal ions can be extracted from alkali metal halide in SWCNTs, which we confirmed through Raman measurements performed on CsI@SWCNT. It was also found that I@SWCNT can catch and release Li ions reversibly even in a solid electrolyte.

Experimental Methods

Encapsulation of Iodine Molecules in SWCNTs

We purchased an SWCNT sample (EC type) from Meijo NanoCarbon Co. First, we performed purification treatment using acid to remove metal impurity. The detailed purification procedure is described in our previous papers.[6−8,10,19,20] The removal of metal impurity was confirmed by thermogravimetric (TG) analysis (Figure S1). The sample was annealed at 1250 °C under vacuum to improve its crystallinity. The high-temperature annealing not only decreases the surface defects of SWCNTs but also closes the tube ends by forming half-fullerene caps at those ends. Subsequently, we performed air oxidation treatments to open the closed ends by burning the half-fullerene caps. The tube opening was confirmed by N2 adsorption measurements.[19] 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 configuration cell using paper-form SWCNT (Figure S2) 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 SWCNT electrode for 15 min.[9] After the encapsulation treatments, we washed the SWCNT 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.[10] We also performed Raman measurements to check the crystallinity and synchrotron XRD measurements at BL-18C at the Photon Factory (PF—High Energy Accelerator Research Organization, Tsukuba, Japan) to determine the mean tube diameter. The nanostructure of the obtained samples was observed using a transmission electron microscope (JEOL JEM-z2500) operated at 200 kV.

Li-Ion Charge–Discharge Experiments of I@SWCNT with Liquid Electrolyte

The paper-form I@SWCNT sample on a Cu foil was used for Li-ion charge–discharge experiments. Li metal foil was used as the counter electrode. Lithium bis(trifluoro methanesulfonyl)imide (LiTFSI)/1,3-dioxolane (DOL) + dimethyl ether (DME) (1 mol/L) was used as electrolyte. The charge–discharge measurements using a conventional two-electrode-type cell (Hohsen, HS-Cell) were conducted using a galvanostat (Toyo System, TOSCAT-3200). The cell assembly was performed in an Ar-filled glovebox to avoid exposure to air and contamination. Charge–discharge experiments were done at room temperature. Cyclic voltammetry measurements were performed using a three-electrode-type cell (working electrode: I@SWCNT; counter electrode: Li metal foil; reference electrode: Li metal foil). Galvanostatic intermittent titration technique (GITT) measurement with an intermittent current rate of 100 mA/g and periodic intermission for 30 min was performed to measure open-circuit potential at a different state of charge. Alternating current impedance spectroscopy measurement was performed using an electrochemical impedance analyzer (Autolab PGSTAT128N). To understand the structural changes of iodine molecules encapsulated in SWCNTs, Raman measurements were performed using a JASCO NRS-3300 spectrometer. To analyze the discharged sample, we performed XPS measurements on ULVAC-PHI PHI5000.

Encapsulation of CsI in SWCNTs

CsI powder sample and SWCNTs were heat-treated at 580 °C in an evacuated glass tube for 14 h. The encapsulation of CsI was confirmed by TEM, SEM, Raman, and XRD measurements.

Li-Ion Charge–Discharge Experiments of I@SWCNT with Solid Electrolyte

Poly(ethylene oxide) (PEO) including LiClO4 ([O](PEO)/[Li] = 8:1) was used as a solid electrolyte. A test cell consisting of the I@SWCNT working electrode and Li metal counter electrode with PEO solid electrolyte was used. Charge–discharge experiments were done in a temperature chamber (Espec SU-241) to control the temperature of the test cell at 60 °C to obtain good ion conductivity.