Structural Distortion in MnO2 Nanosheets and Its Suppression by Cobalt Substitution

Co–Mn oxide nanosheets with the chemical composition H0.23Co0.23Mn0.77O2 (C23M77NS) and MnO2 nanosheets (M100NS) were prepared by exfoliation of layer-structured oxides via chemical processing in an aqueous medium. The optical properties of C23M77NS and M100NS were compared using UV-Vis spectroscopy, and the valence states of Mn and Co and local structures around them were examined using X-ray absorption spectroscopy. M100NS with an average Mn valence of 3.6 exhibits large structural distortion, whereas C23M77NS with an average Mn valence of 4.0 does not exhibit structural distortion. Spontaneous oxidization of Mn occurs during ion-exchange and/or exfoliation into nanosheets. These results have originated the hypothesis that structural distortion determines the valence state of Mn in compounds with CdI2-type-structured MnO2 layers.

Adding functionalities to the above nanosheets is mainly carried out by elemental substitution, often with the help of computational chemistry. For example, Cheng et al. have reported that two-dimensional diluted magnetic semiconductors based on MoS 2 nanosheets can be obtained by doping with transition metals [17], while Wang et al. predicted that oxide nanosheets with half-metallic electronic structures can be prepared by introducing oxygen vacancies to MnO 2 nanosheets with semiconductor-like electronic structures [18].
MnO 2 nanosheets exhibit various functionalities and have been studied as electrodes of lithium-ion secondary batteries [5], those of supercapacitors [19], catalyst [20], biosensing materials [21], and so on [22]. MnO 2 nanosheets have been reported to have a hexagonal CdI 2 -type structure [23], and density functional theory (DFT) calculations have been conducted using such a CdI 2 -type structure as the initial structure [18]. We recently reported that Mn-Ni oxide nanosheets have a distorted crystal structure due to the cooperative Jahn-Teller effect of Mn 3+ [24], which suggests the possibility that Nanomaterials 2017, 7, 295 2 of 9 MnO 2 nanosheets also exhibit a distorted crystal structure. The results of DFT calculations are greatly affected by the crystal structure used, since fundamental information on the structures of these systems is important for rationalizing and eventually predicting their physical properties. In the present work, we therefore compared the structural properties of MnO 2 and Co-Mn oxide nanosheets in detail using X-ray absorption spectroscopy (XAS). Figure 1a shows the powder X-ray diffraction (XRD) patterns of the prepared oxide materials, with a nominal composition of Na 0.50 (Co x Mn 1−x )O 2 (x = 0.15, 0.20 or 0.25); the actual chemical compositions were determined as Na 0.551 (Co 0.157 Mn 0.843 )O 2 , Na 0.520 (Co 0.211 Mn 0.789 )O 2 and Na 0.508 (Co 0.261 Mn 0.739 )O 2 . All XRD peaks were indexed to a hexagonal P2-type layered structure (space group P6 3 /mmc) with the only exceptions a few weak peaks originating from impurities. The lattice parameters were refined by Rietveld analysis using the RIETAN-FP code [25]. The powder XRD and Rietveld refinement results are shown in Figure 1a,b, respectively. The lattice parameter a decreased almost linearly with an increase in Co content x, which can be interpreted in terms of the smaller size of Co 3+ (whose ionic radius in the low-spin, hexacoordinated state is 54.5 pm [26]) compared to that of Mn 3+ (whose high-spin state radius is 64.5 pm [26]). This result thus confirms that Co substitution was successful. The lattice parameter c, on the other hand, did not change linearly with the Co content, possibly as a result of the strong influence of the interlayer Na content on this parameter. Figure 1c shows the XRD patterns of Na 0.50 MnO 2 , in which all peaks can be indexed to the orthorhombic P2-type layered structure (space group Cmcm), except for a few weak peaks associated with impurities. Despite the known challenges associated with the preparation of pure Na 0.50 MnO 2 , the oxide obtained here was almost pure. Figure 1d,e shows models of the crystal structure of Na 0.50 MnO 2 . Rietveld refinement of the lattice parameters yielded a = 0.2832 nm, b 0.5203 nm and c = 1.1303 nm, corresponding to a lattice orthorhombicity (defined as b/ √ 3a) of 1.06.

Results
Nanomaterials 2017, 7, 295 2 of 9 possibility that MnO2 nanosheets also exhibit a distorted crystal structure. The results of DFT calculations are greatly affected by the crystal structure used, since fundamental information on the structures of these systems is important for rationalizing and eventually predicting their physical properties. In the present work, we therefore compared the structural properties of MnO2 and Co-Mn oxide nanosheets in detail using X-ray absorption spectroscopy (XAS). Figure 1a shows the powder X-ray diffraction (XRD) patterns of the prepared oxide materials, with a nominal composition of Na0.50(CoxMn1−x)O2 (x = 0.15, 0.20 or 0.25); the actual chemical compositions were determined as Na0.551(Co0.157Mn0.843)O2, Na0.520(Co0.211Mn0.789)O2 and Na0.508(Co0.261Mn0.739)O2. All XRD peaks were indexed to a hexagonal P2-type layered structure (space group P63/mmc) with the only exceptions a few weak peaks originating from impurities. The lattice parameters were refined by Rietveld analysis using the RIETAN-FP code [25]. The powder XRD and Rietveld refinement results are shown in Figure 1a,b, respectively. The lattice parameter a decreased almost linearly with an increase in Co content x, which can be interpreted in terms of the smaller size of Co 3+ (whose ionic radius in the low-spin, hexacoordinated state is 54.5 pm [26]) compared to that of Mn 3+ (whose high-spin state radius is 64.5 pm [26]). This result thus confirms that Co substitution was successful. The lattice parameter c, on the other hand, did not change linearly with the Co content, possibly as a result of the strong influence of the interlayer Na content on this parameter. Figure 1c shows the XRD patterns of Na0.50MnO2, in which all peaks can be indexed to the orthorhombic P2-type layered structure (space group Cmcm), except for a few weak peaks associated with impurities. Despite the known challenges associated with the preparation of pure Na0.50MnO2, the oxide obtained here was almost pure. Figure 1d,e shows models of the crystal structure of Na0.50MnO2. Rietveld refinement of the lattice parameters yielded a = 0.2832 nm, b = 0.5203 nm and c = 1.1303 nm, corresponding to a lattice orthorhombicity (defined as b/√3a) of 1.06.  Na0.508(Co0.261Mn0.739)O2 powders were reacted with nitric acid to form the proton-exchanged form of the Na-Co-Mn oxide. A greenish-brown dispersion of Co-Mn oxide nanosheets was obtained, with a yield >60%, by exfoliation of the proton-exchanged Na-Co-Mn oxide. The chemical composition of these Co-Mn oxide nanosheets, determined by inductively-coupled plasma atomic emission spectroscopy (ICP-AES) and XAS, was H0.23Co0.23Mn0.77O2·nH2O, indicating that the Co/Mn ratio had changed upon exfoliation. Since the XRD analysis discussed above indicated small amounts of impurities in the Na-Co-Mn oxides, centrifugal separation was used to remove them from the nanosheets. Therefore, the Co/Mn ratio measured for the Co-Mn oxide nanosheets is the actual one. The obtained nanosheet dispersion is referred to as C23M77NS hereafter. Similar to the above, an orange dispersion of MnO2 nanosheets (hereafter M100NS) was also prepared from Na0.50MnO2, with a yield of approximately 30%.   Na 0.508 (Co 0.261 Mn 0.739 )O 2 powders were reacted with nitric acid to form the proton-exchanged form of the Na-Co-Mn oxide. A greenish-brown dispersion of Co-Mn oxide nanosheets was obtained, with a yield >60%, by exfoliation of the proton-exchanged Na-Co-Mn oxide. The chemical composition of these Co-Mn oxide nanosheets, determined by inductively-coupled plasma atomic emission spectroscopy (ICP-AES) and XAS, was H 0.23 Co 0.23 Mn 0.77 O 2 ·nH 2 O, indicating that the Co/Mn ratio had changed upon exfoliation. Since the XRD analysis discussed above indicated small amounts of impurities in the Na-Co-Mn oxides, centrifugal separation was used to remove them from the nanosheets. Therefore, the Co/Mn ratio measured for the Co-Mn oxide nanosheets is the actual one. The obtained nanosheet dispersion is referred to as C23M77NS hereafter. Similar to the above, an orange dispersion of MnO 2 nanosheets (hereafter M100NS) was also prepared from Na 0.50 MnO 2 , with a yield of approximately 30%.  Na0.508(Co0.261Mn0.739)O2 powders were reacted with nitric acid to form the proton-exchanged form of the Na-Co-Mn oxide. A greenish-brown dispersion of Co-Mn oxide nanosheets was obtained, with a yield >60%, by exfoliation of the proton-exchanged Na-Co-Mn oxide. The chemical composition of these Co-Mn oxide nanosheets, determined by inductively-coupled plasma atomic emission spectroscopy (ICP-AES) and XAS, was H0.23Co0.23Mn0.77O2·nH2O, indicating that the Co/Mn ratio had changed upon exfoliation. Since the XRD analysis discussed above indicated small amounts of impurities in the Na-Co-Mn oxides, centrifugal separation was used to remove them from the nanosheets. Therefore, the Co/Mn ratio measured for the Co-Mn oxide nanosheets is the actual one. The obtained nanosheet dispersion is referred to as C23M77NS hereafter. Similar to the above, an orange dispersion of MnO2 nanosheets (hereafter M100NS) was also prepared from Na0.50MnO2, with a yield of approximately 30%.    Figure 3c,d shows the cross-sectional AFM profiles corresponding to the regions indicated by the white lines in Figure 3a,b respectively. The structures of the nanosheets are clearly visible in these figures, and nanosheets of a 0.7-nm thickness (one oxide layer [27]) were observed for both C23M77NS and M100NS. Monolayer nanosheets of Co-Mn oxide and MnO 2 were thus successfully obtained. white lines in Figure 3a,b respectively. The structures of the nanosheets are clearly visible in these figures, and nanosheets of a 0.7-nm thickness (one oxide layer [27]) were observed for both C23M77NS and M100NS. Monolayer nanosheets of Co-Mn oxide and MnO2 were thus successfully obtained. Digital photographs of the C23M77NS and M100NS dispersions are shown in Figure 4a, whereas Figure 4b shows the corresponding UV-Vis absorption spectra. C23M77NS and M100NS exhibited broad absorption peaks at 366 and 386 nm, respectively. Omomo et al. previously reported similar optical properties, with an absorption peak centered around 374 nm for MnO2 nanosheets prepared by exfoliation of layer-structured H0.13MnO2 [5]. Replotting the UV-Vis spectra in Tauc form allowed estimation of the direct allowed band gaps of C23M77NS and M100NS as 2.75 and 2.62 eV, respectively. We recently reported a direct allowed band gap of 2.66 eV for H0.46Mn0.81Ni0.19O2 nanosheets in which the valence state of Mn is 3.9 [24]. The observed absorption bands here are due to the d-d transition of Mn 4+ , and any difference in the corresponding peak wavelength would be mainly caused by structural distortion of the MnO6 octahedra, as further discussed below. The C23M77NS sample also showed a shoulder peak around 600 nm, which can be attributed to the d-d transition of Co 3+ . Digital photographs of the C23M77NS and M100NS dispersions are shown in Figure 4a, whereas Figure 4b shows the corresponding UV-Vis absorption spectra. C23M77NS and M100NS exhibited broad absorption peaks at 366 and 386 nm, respectively. Omomo et al. previously reported similar optical properties, with an absorption peak centered around 374 nm for MnO 2 nanosheets prepared by exfoliation of layer-structured H 0.13 MnO 2 [5]. Replotting the UV-Vis spectra in Tauc form allowed estimation of the direct allowed band gaps of C23M77NS and M100NS as 2.75 and 2.62 eV, respectively. We recently reported a direct allowed band gap of 2.66 eV for H 0.46 Mn 0.81 Ni 0.19 O 2 nanosheets in which the valence state of Mn is 3.9 [24]. The observed absorption bands here are due to the d-d transition of Mn 4+ , and any difference in the corresponding peak wavelength would be mainly caused by structural distortion of the MnO 6 octahedra, as further discussed below. The C23M77NS sample also showed a shoulder peak around 600 nm, which can be attributed to the d-d transition of Co 3+ .

Results
C23M77NS and M100NS were restacked by reacting them with a LiOH aqueous solution, and the restacked products were also subjected to XAS measurements. Specifically, the valence states of Mn and Co were determined by XAS measurements at the L 2,3 -edge. Figure 5a,b shows X-ray absorption near-edge structure (XANES) spectra measured in total electron yield mode at the Mn L 2,3 -edge of restacked C23M77NS and M100NS, respectively. The average valence values for Mn in restacked C23M77NS and M100NS were 4.0 and 3.6, respectively, whereas the valence of Co in restacked C23M77NS, also determined by XAS, was 3.0 (results not shown).
2.62 eV, respectively. We recently reported a direct allowed band gap of 2.66 eV for H0.46Mn0.81Ni0.19O2 nanosheets in which the valence state of Mn is 3.9 [24]. The observed absorption bands here are due to the d-d transition of Mn 4+ , and any difference in the corresponding peak wavelength would be mainly caused by structural distortion of the MnO6 octahedra, as further discussed below. The C23M77NS sample also showed a shoulder peak around 600 nm, which can be attributed to the d-d transition of Co 3+ . C23M77NS and M100NS were restacked by reacting them with a LiOH aqueous solution, and the restacked products were also subjected to XAS measurements. Specifically, the valence states of Mn and Co were determined by XAS measurements at the L2,3-edge. Figure 5a,b shows X-ray absorption near-edge structure (XANES) spectra measured in total electron yield mode at the Mn L2,3-edge of restacked C23M77NS and M100NS, respectively. The average valence values for Mn in restacked C23M77NS and M100NS were 4.0 and 3.6, respectively, whereas the valence of Co in restacked C23M77NS, also determined by XAS, was 3.0 (results not shown). The local Mn structure in both the restacked C23M77NS and M100NS was studied by extended X-ray absorption fine structure (EXAFS) analysis. The Fourier transforms (FTs) of the Mn K-edge EXAFS spectra of restacked C23M77NS and M100NS are shown in Figure 6. The first peak around 0.15 nm is due to the Mn-O contacts, whereas the one around 0.25 nm corresponds to Mn-Mn or Mn-Co interactions. The areas and intensities of these two main peaks were smaller for restacked M100NS than for C23M77NS, showing that the structural distortion of MnO6 units is larger in M100NS than in C23M77NS. The Fourier-transformed EXAFS spectra previously recorded for MnO2 nanosheets obtained from K0.45MnO2 [23,28] were almost identical to those recorded in this study for the nanosheets obtained from Na0.50MnO2, which suggests that the crystal structure of MnO2 nanosheets remains the same, regardless of the starting material used for their synthesis. In the previous studies above, MnO2 nanosheets were considered to have hexagonal CdI2-type structures without distortion. Here, we have shown for the first time that MnO6 units are distorted in MnO2 nanosheets. The local Mn structure in both the restacked C23M77NS and M100NS was studied by extended X-ray absorption fine structure (EXAFS) analysis. The Fourier transforms (FTs) of the Mn K-edge EXAFS spectra of restacked C23M77NS and M100NS are shown in Figure 6. The first peak around 0.15 nm is due to the Mn-O contacts, whereas the one around 0.25 nm corresponds to Mn-Mn or Mn-Co interactions. The areas and intensities of these two main peaks were smaller for restacked M100NS than for C23M77NS, showing that the structural distortion of MnO 6 units is larger in M100NS than in C23M77NS. The Fourier-transformed EXAFS spectra previously recorded for MnO 2 nanosheets obtained from K 0.45 MnO 2 [23,28] were almost identical to those recorded in this study for the nanosheets obtained from Na 0.50 MnO 2 , which suggests that the crystal structure of MnO 2 nanosheets remains the same, regardless of the starting material used for their synthesis. In the previous studies above, MnO 2 nanosheets were considered to have hexagonal CdI 2 -type structures without distortion. Here, we have shown for the first time that MnO 6 units are distorted in MnO 2 nanosheets.
M100NS than in C23M77NS. The Fourier-transformed EXAFS spectra previously recorded for MnO2 nanosheets obtained from K0.45MnO2 [23,28] were almost identical to those recorded in this study for the nanosheets obtained from Na0.50MnO2, which suggests that the crystal structure of MnO2 nanosheets remains the same, regardless of the starting material used for their synthesis. In the previous studies above, MnO2 nanosheets were considered to have hexagonal CdI2-type structures without distortion. Here, we have shown for the first time that MnO6 units are distorted in MnO2 nanosheets. The local structure around Mn atoms in C23M77NS and M100NS was determined under the following two assumptions: (1) MnO 6 units in C23M77NS are not distorted; (2) MnO 6 distortion in M100NS is of the same type as that present in the starting Na 0.5 MnO 2 material with the orthorhombic P2-type layered structure (space group Cmcm). The first assumption is certainly reasonable, since C23M77NS is free from Jahn-Teller Mn 3+ ions. Mn-O and Mn-Mn distances were determined by nonlinear curve-fitting analysis of the inverse FT to k space with a two-shell model composed of six-fold coordination for both Mn-O and Mn-Mn [23,28]. The structural parameters obtained by fitting of the EXAFS spectra are summarized in Table 1; the estimated Mn-O and Mn-Mn distances of 0.1899 and 0.2908 nm, respectively, are roughly consistent with previously reported values for MnO 2 nanosheets, which were also determined under the assumption of six-fold coordination for both Mn-O and Mn-Mn [23,28]. In order to assess the validity of the local structure, the oxidation numbers of the Mn ions in C23M77NS and M100NS were numerically evaluated by bond valence sum (BVS) analysis [29], in which the effective valence is calculated as: where R 0 is the bond valence parameter for Mn 4+ (0.1753 nm) derived from Brown's table [29], and R is the Mn-O interatomic distance listed in Table 1. The calculated BVS values for Mn in C23M77NS and M100NS are 4.0 ± 0.1 and 3.3 ± 0.4, respectively, estimates that are in good agreement with the observed oxidation number for Mn shown in Figure 5; thus, the second assumption discussed above is also reasonable. The calculated orthorhombicity parameter R(Mn-Mn(1))/R(Mn-Mn(2)) of 1.19 was significantly larger than that of the starting Na 0.5 MnO 2 , showing that exfoliation into nanosheets promoted structural distortion.

Discussion
The optical absorption of M100NS showed a red shift compared to that of C23M77NS, which may be ascribed to the split of both the unoccupied e g and occupied t 2g levels caused by the reduction in symmetry of MnO 6 octahedra due to the cooperative Jahn-Teller effect of Mn 3+ contained in M100NS. The Co substitution into MnO 2 nanosheets resulted in the disappearance of the structural distortion.
The nominal Mn valences were 3.50 and 3.67 for the starting Na 0.50 MnO 2 and Na 0.508 (Co 0.261 Mn 0.739 )O 2 , respectively. Thus, Mn is spontaneously oxidized during ion-exchange and/or exfoliation. This spontaneous oxidation/reduction of Mn has also been observed during the preparation of MnO 2 nanosheets from K 0.45 MnO 2 [23]; Mn is oxidized during ion-exchange of the interlayer cations and reduced by exfoliation into nanosheets. The cause of oxidation is not clear, but the results above suggest that structural distortion determines the valence state of Mn for CdI 2 -type-structured MnO 2 layers. That is: (1) Mn is spontaneously oxidized during ion-exchange of interlayer cations from Na + or K + to H 3 O + , since MnO 6 octahedra are stable in their slightly distorted state in the H 3 O + -exchanged form of Na 0.50 MnO 2 or K 0.45 MnO 2 ; (2) MnO 6 octahedra are stable in their regular state, especially in the H 3 O + -exchanged form of Na 0.508 (Co 0.261 Mn 0.739 )O 2 , due to the effect of substituted Co 3+ , and Mn is spontaneously oxidized to Mn 4+ (free from Mn 3+ Jahn-Teller ions) to enable this; and (3) MnO 6 octahedra in Mn100NS are stable in their largely distorted state and Mn 4+ is spontaneously reduced partially to Mn 3+ Jahn-Teller ions. If the above hypothesis is true, then CdI 2 -type-structured Co-Mn oxide nanosheets with rather small Co contents would contain Mn 3+ and Co 2+ Jahn-Teller ions.

Materials and Methods
Layer-structured Na-Co-Mn and Na-Mn oxides were synthesized by a conventional solid-state reaction. Reagent-grade Na 2 CO 3 , Mn 2 O 3 , MnO 2 and Co 3 O 4 were mixed and ball-milled in acetone. Each dried mixture was pressed to form pellets, which were then heated at 1100 • C for 20 h. A 5% excess of Na 2 CO 3 was used to compensate for the loss due to volatilization upon heating. The resulting pellets were ground and used as starting powder materials. The XRD measurements were conducted using a D8 Advance diffractometer (Bruker AXS, Karlsruhe, Germany) equipped with a VANTEC-1 position-sensitive detector (Bruker AXS, Karlsruhe, Germany) and a Cu target X-ray tube. The SEM measurements were performed using a JSM-7000F microscope (JEOL, Tokyo, Japan).
The proton-exchanged form of the Na-Mn oxide was prepared by reacting the powder form of the oxide with 1 M HNO 3 at room temperature for 5 d in order to allow ion exchange, with the HNO 3 solution replaced daily. MnO 2 nanosheets were then prepared by exfoliating the proton-exchanged Na-Mn oxide by reacting with a tetrabutylammonium hydroxide (TBAOH) aqueous solution (TBAOH/Na-Mn oxide molar ratio = 2:1) for 10 d at room temperature. After 10 d, the unreacted particles were separated by centrifugation at 5000 rpm, and a colloidal dispersion of nanosheets was obtained as the supernatant. The same procedure as above was used to prepare Co-Mn oxide nanosheets from Na-Co-Mn oxide. The chemical compositions of the different nanosheet types were determined by ICP-AES using a SPS3100 analyzer (Hitachi High-Tech Science, Tokyo, Japan). The size and shape of each nanosheet type were determined by AFM, using a Nanocute/NanoNavi-II instrument (Hitachi High-Technologies, Tokyo, Japan). The optical absorption properties of the nanosheets were examined using a V-570 spectrophotometer (JASCO, Tokyo, Japan).
The prepared MnO 2 and Co-Mn oxide nanosheets were restacked by reacting them with 1 M LiOH aqueous solution, which was followed by washing with purified water and drying at room temperature. XAS measurements of the restacked nanosheets were conducted at both Mn K-and L 2,3 -edges. The K-edge measurements were conducted using a transmission method at the BL-12C beamline of the Photon Factory of the High Energy Accelerator Research Organization (Tsukuba, Japan), whereas the L 2,3 -edge ones were conducted in total electron yield mode at the BL-11 beamline of the Synchrotron Radiation (SR) Center of Ritsumeikan University (Kusatsu, Japan). The FTs of the Mn K-edge XAS spectra were obtained with k 3 weighting in a k range of 2.8-13.2 Å −1 . The structural parameters were determined by curve-fitting procedures using the Athena-Artemis software [30]. The models of the crystal structures were prepared using VESTA software [31].

Conclusions
Mn and Co-Mn oxide nanosheets were prepared by aqueous exfoliation from Na 0.50 MnO 2 and Na 0.508 (Co 0.261 Mn 0.739 )O 2 , respectively. Extended X-ray absorption fine structure analysis at the Mn K-edge revealed that MnO 2 nanosheets (M100NS) exhibit structural distortion due to the Jahn-Teller effect associated with Mn 3+ ions. The structural distortion in M100NS was suppressed by Co substitution. H 0.23 Co 0.23 Mn 0.77 O 2 and M100NS showed optical absorption peaks at 366 and 386 nm, respectively; the difference is mainly due to the structural distortion in M100NS. The structural information obtained in this study will contribute to achieve a better understanding of the physical properties of nanosheets, which will in turn allow the exploration of novel applications of these systems.