- Email: [email protected]
Good reproductive preparation method of Li-intercalated hexagonal boron nitride and transmission electron microscopy – Electron energy loss spectroscopy analysis
Solid State Sciences xxx (2015) 1e5
Contents lists available at ScienceDirect
Solid State Sciences journal homepage: www.elsevier.com/locate/ssscie
Good reproductive preparation method of Li-intercalated hexagonal boron nitride and transmission electron microscopy e Electron energy loss spectroscopy analysis A. Sumiyoshi a, H. Hyodo b, Y. Sato b, M. Terauchi b, K. Kimura a, * a b
Department of Advanced Materials Science, The University of Tokyo, Japan Institute for Multidisciplinary Research for Advanced Materials, Tohoku University, Japan
a r t i c l e i n f o
a b s t r a c t
Article history: Received 5 November 2014 Received in revised form 25 April 2015 Accepted 27 April 2015 Available online xxx
A Li-intercalated hexagonal boron nitride (Li-h-BNIC) phase was synthesized using a highly reproducible method that involves annealing an Li3N and h-BN mixture at 1220 K. Powder X-ray diffraction, electrical conductivity measurements, transmission electron microscopy (TEM) and electron energy loss spectroscopy were performed. The stacking of BN atomic layers in the Li-h-BNIC phase is not the same as the two-layer stacking periodicity of h-BN. TEM observation suggests the existence of incommensurate periodicity along the intralayer direction. From the low-loss and core-loss spectra, the Li-h-BNIC phase is not metal as predicted by the first-principle calculations. Satellite peaks of 1 s to p* transition in the B Kedge core-loss spectrum indicate the presence of N atom vacancies modified by O atoms in the h-BN atomic layer. © 2015 Published by Elsevier Masson SAS.
Keywords: Hexagonal boron nitride Alkali metal intercalation X-ray diffraction Electrical conductivity Transmission electron microscopy Electron energy loss spectroscopy
1. Introduction Hexagonal boron nitride (h-BN) has a stacking structure of 2D boron nitride atomic layers analogous to graphite. Intercalation into the interlayers of h-BN to change the material properties is of similar interest to graphite intercalation compounds (GICs) [1]. Recently, theoretical studies were performed on h-BN intercalation compounds (h-BNICs). The possibility of alkaliemetal h-BNICs with metallic characteristics has been predicted by first-principles calculations [2,3]. Alkali metal GIC-like structures without defects or curvature in the h-BN atomic layer were assumed in these two studies. Oba et al. reported the possibility of inducing an impurity level in the wide band gap of h-BN by the “dilute” intercalation of donor alkali metal atoms or acceptor F atoms [4]. Several experimental studies investigating h-BN intercalation compounds have been carried out. Doll et al. reported K intercalation into h-BN thin films and a (2 2)R0 super lattice structure commensurate with the h-BN parental lattice using transmission electron microscopy (TEM) analysis [5]. However, super lattice diffraction can only be
* Corresponding author. E-mail address: [email protected] (K. Kimura).
observed by TEM prior to K de-intercalation. Sakamoto et al. reported Br2 and Cs intercalation into h-BN using the two-zone method [6]. The mass fraction of intercalates is at most 2% and Cs islands were observed by dark field TEM. Neither Doll nor Sakamoto performed X-ray diffraction (XRD) analysis. Shen et al. reported bulk-scale preparation of SO3F-h-BNIC and verified the metallic characteristics of the compound using electrical conductivity measurements [7]. Kovtyukhova et al. reported the reversible intercalation of Brønsted acids into h-BN without covalent bonding [8]. Intercalation of layered materials, such as h-BN, is interesting both for the study of fundamental property changes that result from intercalation and from a materials science perspective as a wide band-gap material. However, there have been relatively few experimental reports on the preparation of h-BNIC. In particular, there have been very few studies investigating the preparation and properties of alkali metal h-BNIC. We have successfully produced XRD-scale Li-h-BNIC and reported on the differences between the crystal structure of Li-h-BNIC and the parent h-BN phase [9,10]. In our previous studies analysis of the Li-h-BNIC phase was difficult because of the low reproducibility of the phase. Here, we report on a preparation method of Li-h-BNIC with high reproducibility and present TEM and electron energy loss spectroscopy (EELS) analysis
http://dx.doi.org/10.1016/j.solidstatesciences.2015.04.011 1293-2558/© 2015 Published by Elsevier Masson SAS.
Please cite this article in press as: A. Sumiyoshi, et al., Good reproductive preparation method of Li-intercalated hexagonal boron nitride and transmission electron microscopy e Electron energy loss spectroscopy analysis, Solid State Sciences (2015), http://dx.doi.org/10.1016/ j.solidstatesciences.2015.04.011
2
A. Sumiyoshi et al. / Solid State Sciences xxx (2015) 1e5
of the electronic structure. We also report electrical conductivity measurement and reanalysis of XRD data. 2. Experimental Li-h-BNIC was prepared by mixing Li3N powder (SigmaeAldrich, 99.5%) and h-BN powder (high-purity materials, 99%up) using an agate mortar in an Ar-filled glove box with a molar ratio Li3N/BN ¼ 0.3e1.1. The starting mixture was sealed into a stainless steel tube using arc welding. The welded stainless steel tube was annealed at a temperature ranging from 700 K to 1500 K for 10 h in an electric furnace. The annealed tube was opened in an Ar-filled glove box. The different phases present in product materials were identified using X-ray powder diffraction with Cu Ka radiation (Rigaku Co., SmartLab). Electrical conductivity measurements were performed on pellets of the product powders using the van der Pauw method (TOYO corporation, Resitest 8300) at temperatures ranging from room temperature to 400 C under vacuum conditions maintained by an oil rotary pump. The sample pellets for electrical conductivity measurement were prepared by spark plasma sintering (SPS) (Fuji Electronic Industrial Co., SPS-515S) at 720 C for 30 min. TEM observation and electron diffraction measurements of Li-h-BNIC and pristine h-BN were conducted using a JEOL JEM-2010. High-energy-resolution EELS measurements were performed under a monochromator analytical TEM [11]. B K-edge core-loss and low-loss spectra were obtained. The energy resolution was 80 meV for the low-loss spectroscopy and 0.15 eV for the core-loss spectroscopy. Rietveld analysis of the powder XRD pattern was performed using RIETAN-FP software [12]. 3. Results and discussion Fig. 1 shows the XRD patterns of samples having molar ratios Li3N/BN ¼ 0.3, 0.5, 0.7, 1.1 at 1220 K. The powder XRD patterns indicate that the samples are composed of Li-h-BNIC, alfa-Li3BN2 and Li2O. The existence of the Li2O phase could be due to relatively low purity of the starting reagent and the atmosphere not being
Fig. 1. Powder XRD patterns recorded from samples with molar ratios Li3N/BN ¼ 0.3, 0.5, 0.7, 1.1 at 1220 K and reference compounds Li/BN sample [9], Li-h-BNIC [9], Li-hBNIC [13], h-BN, a-Li3BN2, and Li2O.
completely inert. The XRD pattern of the Li-h-BNIC phase observed here is similar to the pattern observed from material synthesized from a mixture of Li and h-BN in our previous reports [9,10]. The XRD pattern of the Li-h-BNIC phase is almost identical to the XRD pattern from the Li-h-BNIC phase analyzed by Yamane et al. [13]. However, the intensity of minor peaks of the Li-h-BNIC phase reported by Yamane et al. (2q ¼ 33.8 , 40.4 , 43.7, 43.9 [13]) was not observed in the present Li-h-BNIC phase. Based on the 2q values reported by Yamane et al., the minor peaks were likely derived from Li2O or a-Li3BN2. In our previous study, Li-h-BNIC was obtained in only three of 10 trials conducted under the same conditions for the Li metal and h-BN reaction [9]. In the present study, Li-h-BNIC was successfully obtained in each of the 10 trials. The reproducibility of the Li-h-BNIC has improved considerably using the method described in this work. Furthermore, the yield in Ref. [9] was only 20 wt%, while the yield is 70 wt% in the present work. The exact composition of the Li-h-BNIC is still unknown because a singlephase Li-h-BNIC sample could not be produced. TEM analysis was performed using a powder sample synthesized with an ingredient ratio of Li3N/BN ¼ 0.5. Fig. 2 shows the diffraction pattern of the Li-h-BNIC phase ([1000] direction of incidence) and the simulated diffraction pattern from an h-BN structure with an a-direction expansion of 2.48% and a c-direction expansion of 12.86%. The expansion ratios were examined in our previous report [10]. In the present study, a split spot diffraction pattern was observed in the Li-h-BNIC phase, as well as the previously reported DebyeeScherrer ring-like diffraction pattern [10]. The split direction is the intralayer direction of the h-BN atomic layer. The distance between the split spots corresponds to approximately eight times the a-direction lattice constant of the expanded h-BN structure, 2.566 Å. This indicates the presence of an intralayer direction with incommensurate structure. This incommensurate periodicity likely originates from the stacking change of the BN atomic layers and interlayer Li atoms. This split spot diffraction pattern changed into a DebyeeScherrer ring pattern by intentionally irradiating the sample with an electron beam at a higher voltage than that used for observation. The 10e11 diffraction spot in Fig. 2 (b) is not present in Fig. 2 (a). This can be interpreted as loss of the two-layer stacking periodicity of the pristine h-BN structure due to the imperfect Li intercalation into the h-BN interlayer. In this paper, we will use the abbreviations “2L model” to indicate the two-layer stacking periodicity structure of pristine hBN and “1L model” to indicate the structure without the two-layer stacking periodicity. Because the 0001 diffraction spot is not present in the 2L model, the 0002 diffraction spot in the 2L model should be reconsidered as a 0001 diffraction spot in the 1L model. Using this interpretation, the 10e12 diffraction spot of the 2L model should be reconsidered as the 10e11 diffraction spot in the 1L model, and the 10e11 diffraction spot in the 2L model disappears in the 1L model. The disappearance of the 10e11 diffraction spot in the 2L model was confirmed by XRD. We also re-conducted the previously reported Rietveld analysis of the powder XRD pattern [10] using the 1L model of Li-h-BNIC. Fig. 3 (a) shows the Rietveld fitting of the XRD pattern reported in Ref. [10] using the 1L model of Li-h-BNIC and (b) shows the Rietveld fitting of the 2L model of Li-hBNIC. The arrangement of Li atoms is assumed as a stage-1 MC6 system. The stacking sequence of BN sheets, which is described in terms of ABAB for pristine h-BN, is assumed as AaAa for the 1L model and AaBa for the 2L model. The used lattice parameters were based on an h-BN structure with an adirection expansion 2.48% and c-direction expansion 12.86% refined in our previous report [10]. The lattice parameters of the used structure models are a ¼ 4.4437 Å, c ¼ 3.7599 Å, space group 174 for the 1L model and a ¼ 4.4437 Å, c ¼ 7.5198 Å, space group 173 for the 2L model. The c value of the 2L model is twice that of the 1L
Please cite this article in press as: A. Sumiyoshi, et al., Good reproductive preparation method of Li-intercalated hexagonal boron nitride and transmission electron microscopy e Electron energy loss spectroscopy analysis, Solid State Sciences (2015), http://dx.doi.org/10.1016/ j.solidstatesciences.2015.04.011
A. Sumiyoshi et al. / Solid State Sciences xxx (2015) 1e5
3
Fig. 2. (a) The diffraction pattern of Li-h-BNIC phase recorded with a[1000]direction of incidence. (b) The simulated diffraction pattern with a[1000]direction from an h-BN structure with a-direction expansion of 2.48% and c-direction expansion of 12.86% (2L model) [10].
model. The 101 diffraction peak that appears in the simulated XRD pattern corresponds to the 2L model of Li-h-BNIC, as shown in Fig. 3 (b). However, this diffraction peak was not observed in the experimental XRD pattern. This mismatch, which was discussed in Ref. [10], was corrected using the 1L model structure (Fig. 3 (a)). This result indicates that the two-layer stacking periodicity of pristine h-BN structure was lost because of Li intercalation. Fig. 4 shows the electrical conductivity measurements for the
Fig. 3. The result of Rietveld fitting to the Li-h-BNIC structure using (a) the 1L model and (b) the 2L model. “Obs.” is the observed pattern, “Calc.” is that calculated from the 1L and 2L models, “Delta” is the difference between “Obs.” and “Calc.” and “Bkg” is the background.
three samples collected over a temperature range from room temperature to 673 K. The a-Li3BN2 sample pellet was prepared by SPS of the Li3N/BN ¼ 1.1 powder sample, and the a-Li3BN2þLi-hBNIC sample pellet was prepared by SPS of the Li3N/BN ¼ 0.5 powder sample. These are the same powders used to collect the XRD data shown in Fig. 1. The Li-pBN sample was synthesized by annealing highly oriented pyrolytic h-BN bulk and Li metal. The electrical conductivity data corresponding to the Li-pBN was reported in Ref. [9]. The XRD pattern remained the same before and after the SPS process, with the exception of a slight intensity increase of the Li2O phase. In contrast to the Li-pBN, an electrical conductivity temperature dependence could not be measured for a-Li3BN2 and a-Li3BN2þLi-h-BNIC at lower temperatures. Table 1 shows the activation energy obtained by analysis of the electrical conductivity data measured in this study and the value reported for a-Li3BN2 by Yamane [13]. The activation energy values of a-Li3BN2 and a-Li3BN2þLi-h-BNIC reported here are similar to the value reported by Yamane. These results indicate that the conduction of the a-Li3BN2þLi-h-BNIC pellet is dominated by ionic conduction of aLi3BN2. This is likely a result of the conduction pass of the more conductive Li-h-BNIC phase not being connected and the insulating Li2O phase not blocking the ionic conduction pass in the aLi3BN2þLi-h-BNIC pellet. The absolute value of electrical conductivity and the activation energy for Li-pBN is very low because Li-hBNIC is thought to form only on the surface of the bulk sample but
Fig. 4. Electrical conductivity of a-Li3BN2, a-Li3BN2þLi-h-BNIC mixture and Li-pBN samples [9].
Please cite this article in press as: A. Sumiyoshi, et al., Good reproductive preparation method of Li-intercalated hexagonal boron nitride and transmission electron microscopy e Electron energy loss spectroscopy analysis, Solid State Sciences (2015), http://dx.doi.org/10.1016/ j.solidstatesciences.2015.04.011
4
A. Sumiyoshi et al. / Solid State Sciences xxx (2015) 1e5 Table 1 The activation energy of electrical conductivity obtained from data shown in Fig. 4 and the reported value for aLi3BN2. Sample phase
E (kJ/mol)
Li-h-BNIC, a-Li3BN2 a-Li3BN2 a-Li3BN2 [13]
62 64 78
to be connected between the edges of the sample. Fig. 5 shows low-loss EELS obtained from three different crystalline fragments of Li-h-BNIC material and pristine h-BN. The intensity profile of the p plasmon peaks of Li-h-BNIC are broader than that of h-BN. The energy position of the peak of Li-h-BNIC is almost the same as with h-BN. Although new structures appear near the onset and around 10e15 eV in the Li-h-BNIC spectra, as indicated by vertical lines, there is no apparent additional intensity on the tail of each zero-loss peak, which can be attributed to carrier excitation. This means that those Li-h-BNIC crystalline fragments may not be metal as predicted by the first-principle calculations [2,3]. Fig. 6 shows a B-K core-loss EELS of Li-h-BNIC phases and pristine h-BN, which correspond to the partial density of states with p-symmetry of the conduction bands (unoccupied states). The strong peak at 192.0 eV originates from 1s to p* transition, which is a characteristic feature of sp2 bonding of B to its three neighboring N atoms in the h-BN atomic layer. Three satellite peaks are observed at 192.6, 193.2, and 193.9 eV following Li intercalation. These peaks have often been observed in XANES spectra of 2D h-BN structured materials, such as BN thin film, BN nanotube and BCN [14e21]. The energy positions of the satellite peaks in Fig. 6 show good agreement with reported values [21], where the individual peak energies are affected by the number of O atoms adjacent to B atoms, such as
Fig. 6. B K-edge core-loss EELS data measured from the Li-h-BNIC phases and pristine h-BN. The vertical bars indicate the new spectral features that appear after Li intercalation.
BeON2, BeO2N, and BeO3 [21]. This result indicates that N defects and oxidation are induced by Li intercalation at high annealing temperatures. Therefore, the alkali metal GIC-like structure assumed in the first-principle calculations in Ref. [2,3] is not observed in the present study. The bottom of the conduction band of h-BN consists mainly of p electrons from the B atom. If the Fermilevel exists in conduction bands, the onset of the core-loss spectrum should show a sudden intensity increase where the width of the intensity increase is almost equal to the energy resolution (Fermi edge). The Fermi edge was observed in the case of Li-doped alpha-rhombohedral boron, which was metalized and showed superconductivity [22]. In Fig. 6, the Fermi edge was not observed, indicating that the Li-h-BNIC in this study is not a metal. The Fermi edge was not observed at the onset of the 1s to p* peak. Conversely, the Fermi edge was observed at the onset of the 1s to p* peak in the B K-edge core-loss spectrum measured from Li-doped a-rhombohedral boron, which metalizes and shows superconductivity [22]. The absence of the Fermi edge in the present study indicates that the Li-h-BNIC measured here is not metal. Another Li intercalation method that does not produce vacancies in the parental h-BN atomic layer will be needed for further comparison with the theoretical calculations in Ref. [2,3]. 4. Conclusion
Fig. 5. Low-loss EELS data measured from the Li-h-BNIC phases and pristine h-BN. The vertical bars indicate the new spectral features that appear after Li intercalation.
A reproducible method for the synthesis of the Li-h-BNIC phase was developed by annealing Li3N and h-BN, rather than Li metal and h-BN as reported previously [9,10]. The stacking periodicity of the BN atomic layer in the Li-h-BNIC phase changes from the twolayer stacking periodicity of h-BN to a one-layer stacking because of Li intercalation. The existence of incommensurate periodicity along the intralayer direction was observed using TEM. The band gap of
Please cite this article in press as: A. Sumiyoshi, et al., Good reproductive preparation method of Li-intercalated hexagonal boron nitride and transmission electron microscopy e Electron energy loss spectroscopy analysis, Solid State Sciences (2015), http://dx.doi.org/10.1016/ j.solidstatesciences.2015.04.011
A. Sumiyoshi et al. / Solid State Sciences xxx (2015) 1e5
the Li-h-BNIC phase was confirmed using low-loss and core-loss EELS. The existence of N atom vacancies modified by O atoms in the h-BN atomic layer was suggested from core-loss EELS. Acknowledgments One of the authors (A.S.) was supported by Research Fellowships of the Japan Society for the Promotion of Science (No. 23-5556) for Young Scientists. This work was partly supported by Scientific Research on Priority Areas of New Materials Science Using Regulated Nano Spaces, KAKENHI No. 19051005 from MEXT. References [1] [2] [3] [4]
M.S. Dresselhaus, G. Dresselhaus, Adv. Phys. 51 (2002) 1. S. Okada, M. Otani, Phys. Rev. B 81 (2010) 3e6. it, Eur. Phys. J. B 79 (2011) 301e312. B. Altintas, C. Parlak, C. Bozkurt, R. Eryig F. Oba, A. Togo, I. Tanaka, K. Watanabe, T. Taniguchi, Phys. Rev. B 81 (2010) 20e23. [5] G.L. Doll, J.S. Speck, G. Dresselhaus, M.S. Dresselhaus, K. Nakamura, S.I. Tanuma, J. Appl. Phys. 66 (1989) 2554e2558. [6] M. Sakamoto, J.S. Speck, M.S. Dresselhaus, J. Mater. Res. 1 (1986) 685e692. [7] C. Shen, S.G. Mayorga, R. Biagioni, C. Piskoti, M. Ishigami, A. Zettl, N. Bartlett,
5
J. Solid State Chem. 147 (1999) 74e81. [8] N.I. Kovtyukhova, Y. Wang, R. Lv, M. Terrones, V.H. Crespi, T.E. Mallouk, J. Am. Chem. Soc. 135 (2013) 8372e8381. [9] A. Sumiyoshi, H. Hyodo, K. Kimura, J. Phys. Chem. Solids 71 (2010) 569e571. [10] A. Sumiyoshi, H. Hyodo, K. Kimura, J. Solid State Chem. 187 (2012) 208e210. [11] Y. Sato, M. Terauchi, M. Mukai, T. Kaneyama, K. Adachi, Ultramicroscopy 111 (2011) 1381e1387. [12] F. Izumi, K. Momma, Solid State Phenom. 130 (2007) 15e20. [13] H. Yamane, S. Kikkawa, M. Koizumi, J. Solid State Chem. 71 (1987) 1e11. [14] I. Jimenez, A.F. Jankowski, L.J. Terminello, D.G.J. Sutherland, J.A. Carlisle, G.L. Doll, W.M. Tong, D.K. Shuh, F.J. Himpsel, Phys. Rev. B 55 (1997) 25. [15] M. Niibe, K. Miyamoto, T. Mitamura, K. Mochiji, J. Vac. Sci. Technol. A 28 (2010) 1157. [16] M. Petravic, R. Peter, I. Kavre, L.H. Li, Y. Chen, L.-J. Fan, Y.-W. Yang, Phys. Chem. Chem. Phys. 12 (2010) 15349. [17] A.B. Preobrajenski, M.L. Ng, N.A. Vinogradov, A.S. Vinogradov, E. Lundgren, A. Mikkelsen, N. Martensson, Nano Lett. 9 (2009) 2780. [18] H.C. Choi, S.Y. Bae, W.S. Jang, J. Park, H.J. Song, H.-J. Shin, J. Phys. Chem. B 109 (2005) 7007. [19] L. Li, L.H. Li, Y. Chen, X.J. Dai, T. Xing, M. Petravic, X. Liu, Nanoscale Res. Lett. 7 (2012) 417. [20] T. Hemraj-Benny, S. Banerjee, S. Sambasivan, D.A. Fischer, W. Han, J.A. Misewich, S.S. Wong, Phys. Chem. Chem. Phys. 7 (2005) 1103. [21] I. Caretti, I. Jime'nez, J. Appl. Phys. 110 (2011) 023511. [22] T. Nagatochi, H. Hyodo, A. Sumiyoshi, K. Soga, Y. Sato, M. Terauchi, F. Esaka, K. Kimura, Phys. Rev. B 83 (2011) 1e5.
Please cite this article in press as: A. Sumiyoshi, et al., Good reproductive preparation method of Li-intercalated hexagonal boron nitride and transmission electron microscopy e Electron energy loss spectroscopy analysis, Solid State Sciences (2015), http://dx.doi.org/10.1016/ j.solidstatesciences.2015.04.011
Contents lists available at ScienceDirect
Solid State Sciences journal homepage: www.elsevier.com/locate/ssscie
Good reproductive preparation method of Li-intercalated hexagonal boron nitride and transmission electron microscopy e Electron energy loss spectroscopy analysis A. Sumiyoshi a, H. Hyodo b, Y. Sato b, M. Terauchi b, K. Kimura a, * a b
Department of Advanced Materials Science, The University of Tokyo, Japan Institute for Multidisciplinary Research for Advanced Materials, Tohoku University, Japan
a r t i c l e i n f o
a b s t r a c t
Article history: Received 5 November 2014 Received in revised form 25 April 2015 Accepted 27 April 2015 Available online xxx
A Li-intercalated hexagonal boron nitride (Li-h-BNIC) phase was synthesized using a highly reproducible method that involves annealing an Li3N and h-BN mixture at 1220 K. Powder X-ray diffraction, electrical conductivity measurements, transmission electron microscopy (TEM) and electron energy loss spectroscopy were performed. The stacking of BN atomic layers in the Li-h-BNIC phase is not the same as the two-layer stacking periodicity of h-BN. TEM observation suggests the existence of incommensurate periodicity along the intralayer direction. From the low-loss and core-loss spectra, the Li-h-BNIC phase is not metal as predicted by the first-principle calculations. Satellite peaks of 1 s to p* transition in the B Kedge core-loss spectrum indicate the presence of N atom vacancies modified by O atoms in the h-BN atomic layer. © 2015 Published by Elsevier Masson SAS.
Keywords: Hexagonal boron nitride Alkali metal intercalation X-ray diffraction Electrical conductivity Transmission electron microscopy Electron energy loss spectroscopy
1. Introduction Hexagonal boron nitride (h-BN) has a stacking structure of 2D boron nitride atomic layers analogous to graphite. Intercalation into the interlayers of h-BN to change the material properties is of similar interest to graphite intercalation compounds (GICs) [1]. Recently, theoretical studies were performed on h-BN intercalation compounds (h-BNICs). The possibility of alkaliemetal h-BNICs with metallic characteristics has been predicted by first-principles calculations [2,3]. Alkali metal GIC-like structures without defects or curvature in the h-BN atomic layer were assumed in these two studies. Oba et al. reported the possibility of inducing an impurity level in the wide band gap of h-BN by the “dilute” intercalation of donor alkali metal atoms or acceptor F atoms [4]. Several experimental studies investigating h-BN intercalation compounds have been carried out. Doll et al. reported K intercalation into h-BN thin films and a (2 2)R0 super lattice structure commensurate with the h-BN parental lattice using transmission electron microscopy (TEM) analysis [5]. However, super lattice diffraction can only be
* Corresponding author. E-mail address: [email protected] (K. Kimura).
observed by TEM prior to K de-intercalation. Sakamoto et al. reported Br2 and Cs intercalation into h-BN using the two-zone method [6]. The mass fraction of intercalates is at most 2% and Cs islands were observed by dark field TEM. Neither Doll nor Sakamoto performed X-ray diffraction (XRD) analysis. Shen et al. reported bulk-scale preparation of SO3F-h-BNIC and verified the metallic characteristics of the compound using electrical conductivity measurements [7]. Kovtyukhova et al. reported the reversible intercalation of Brønsted acids into h-BN without covalent bonding [8]. Intercalation of layered materials, such as h-BN, is interesting both for the study of fundamental property changes that result from intercalation and from a materials science perspective as a wide band-gap material. However, there have been relatively few experimental reports on the preparation of h-BNIC. In particular, there have been very few studies investigating the preparation and properties of alkali metal h-BNIC. We have successfully produced XRD-scale Li-h-BNIC and reported on the differences between the crystal structure of Li-h-BNIC and the parent h-BN phase [9,10]. In our previous studies analysis of the Li-h-BNIC phase was difficult because of the low reproducibility of the phase. Here, we report on a preparation method of Li-h-BNIC with high reproducibility and present TEM and electron energy loss spectroscopy (EELS) analysis
http://dx.doi.org/10.1016/j.solidstatesciences.2015.04.011 1293-2558/© 2015 Published by Elsevier Masson SAS.
Please cite this article in press as: A. Sumiyoshi, et al., Good reproductive preparation method of Li-intercalated hexagonal boron nitride and transmission electron microscopy e Electron energy loss spectroscopy analysis, Solid State Sciences (2015), http://dx.doi.org/10.1016/ j.solidstatesciences.2015.04.011
2
A. Sumiyoshi et al. / Solid State Sciences xxx (2015) 1e5
of the electronic structure. We also report electrical conductivity measurement and reanalysis of XRD data. 2. Experimental Li-h-BNIC was prepared by mixing Li3N powder (SigmaeAldrich, 99.5%) and h-BN powder (high-purity materials, 99%up) using an agate mortar in an Ar-filled glove box with a molar ratio Li3N/BN ¼ 0.3e1.1. The starting mixture was sealed into a stainless steel tube using arc welding. The welded stainless steel tube was annealed at a temperature ranging from 700 K to 1500 K for 10 h in an electric furnace. The annealed tube was opened in an Ar-filled glove box. The different phases present in product materials were identified using X-ray powder diffraction with Cu Ka radiation (Rigaku Co., SmartLab). Electrical conductivity measurements were performed on pellets of the product powders using the van der Pauw method (TOYO corporation, Resitest 8300) at temperatures ranging from room temperature to 400 C under vacuum conditions maintained by an oil rotary pump. The sample pellets for electrical conductivity measurement were prepared by spark plasma sintering (SPS) (Fuji Electronic Industrial Co., SPS-515S) at 720 C for 30 min. TEM observation and electron diffraction measurements of Li-h-BNIC and pristine h-BN were conducted using a JEOL JEM-2010. High-energy-resolution EELS measurements were performed under a monochromator analytical TEM [11]. B K-edge core-loss and low-loss spectra were obtained. The energy resolution was 80 meV for the low-loss spectroscopy and 0.15 eV for the core-loss spectroscopy. Rietveld analysis of the powder XRD pattern was performed using RIETAN-FP software [12]. 3. Results and discussion Fig. 1 shows the XRD patterns of samples having molar ratios Li3N/BN ¼ 0.3, 0.5, 0.7, 1.1 at 1220 K. The powder XRD patterns indicate that the samples are composed of Li-h-BNIC, alfa-Li3BN2 and Li2O. The existence of the Li2O phase could be due to relatively low purity of the starting reagent and the atmosphere not being
Fig. 1. Powder XRD patterns recorded from samples with molar ratios Li3N/BN ¼ 0.3, 0.5, 0.7, 1.1 at 1220 K and reference compounds Li/BN sample [9], Li-h-BNIC [9], Li-hBNIC [13], h-BN, a-Li3BN2, and Li2O.
completely inert. The XRD pattern of the Li-h-BNIC phase observed here is similar to the pattern observed from material synthesized from a mixture of Li and h-BN in our previous reports [9,10]. The XRD pattern of the Li-h-BNIC phase is almost identical to the XRD pattern from the Li-h-BNIC phase analyzed by Yamane et al. [13]. However, the intensity of minor peaks of the Li-h-BNIC phase reported by Yamane et al. (2q ¼ 33.8 , 40.4 , 43.7, 43.9 [13]) was not observed in the present Li-h-BNIC phase. Based on the 2q values reported by Yamane et al., the minor peaks were likely derived from Li2O or a-Li3BN2. In our previous study, Li-h-BNIC was obtained in only three of 10 trials conducted under the same conditions for the Li metal and h-BN reaction [9]. In the present study, Li-h-BNIC was successfully obtained in each of the 10 trials. The reproducibility of the Li-h-BNIC has improved considerably using the method described in this work. Furthermore, the yield in Ref. [9] was only 20 wt%, while the yield is 70 wt% in the present work. The exact composition of the Li-h-BNIC is still unknown because a singlephase Li-h-BNIC sample could not be produced. TEM analysis was performed using a powder sample synthesized with an ingredient ratio of Li3N/BN ¼ 0.5. Fig. 2 shows the diffraction pattern of the Li-h-BNIC phase ([1000] direction of incidence) and the simulated diffraction pattern from an h-BN structure with an a-direction expansion of 2.48% and a c-direction expansion of 12.86%. The expansion ratios were examined in our previous report [10]. In the present study, a split spot diffraction pattern was observed in the Li-h-BNIC phase, as well as the previously reported DebyeeScherrer ring-like diffraction pattern [10]. The split direction is the intralayer direction of the h-BN atomic layer. The distance between the split spots corresponds to approximately eight times the a-direction lattice constant of the expanded h-BN structure, 2.566 Å. This indicates the presence of an intralayer direction with incommensurate structure. This incommensurate periodicity likely originates from the stacking change of the BN atomic layers and interlayer Li atoms. This split spot diffraction pattern changed into a DebyeeScherrer ring pattern by intentionally irradiating the sample with an electron beam at a higher voltage than that used for observation. The 10e11 diffraction spot in Fig. 2 (b) is not present in Fig. 2 (a). This can be interpreted as loss of the two-layer stacking periodicity of the pristine h-BN structure due to the imperfect Li intercalation into the h-BN interlayer. In this paper, we will use the abbreviations “2L model” to indicate the two-layer stacking periodicity structure of pristine hBN and “1L model” to indicate the structure without the two-layer stacking periodicity. Because the 0001 diffraction spot is not present in the 2L model, the 0002 diffraction spot in the 2L model should be reconsidered as a 0001 diffraction spot in the 1L model. Using this interpretation, the 10e12 diffraction spot of the 2L model should be reconsidered as the 10e11 diffraction spot in the 1L model, and the 10e11 diffraction spot in the 2L model disappears in the 1L model. The disappearance of the 10e11 diffraction spot in the 2L model was confirmed by XRD. We also re-conducted the previously reported Rietveld analysis of the powder XRD pattern [10] using the 1L model of Li-h-BNIC. Fig. 3 (a) shows the Rietveld fitting of the XRD pattern reported in Ref. [10] using the 1L model of Li-h-BNIC and (b) shows the Rietveld fitting of the 2L model of Li-hBNIC. The arrangement of Li atoms is assumed as a stage-1 MC6 system. The stacking sequence of BN sheets, which is described in terms of ABAB for pristine h-BN, is assumed as AaAa for the 1L model and AaBa for the 2L model. The used lattice parameters were based on an h-BN structure with an adirection expansion 2.48% and c-direction expansion 12.86% refined in our previous report [10]. The lattice parameters of the used structure models are a ¼ 4.4437 Å, c ¼ 3.7599 Å, space group 174 for the 1L model and a ¼ 4.4437 Å, c ¼ 7.5198 Å, space group 173 for the 2L model. The c value of the 2L model is twice that of the 1L
Please cite this article in press as: A. Sumiyoshi, et al., Good reproductive preparation method of Li-intercalated hexagonal boron nitride and transmission electron microscopy e Electron energy loss spectroscopy analysis, Solid State Sciences (2015), http://dx.doi.org/10.1016/ j.solidstatesciences.2015.04.011
A. Sumiyoshi et al. / Solid State Sciences xxx (2015) 1e5
3
Fig. 2. (a) The diffraction pattern of Li-h-BNIC phase recorded with a[1000]direction of incidence. (b) The simulated diffraction pattern with a[1000]direction from an h-BN structure with a-direction expansion of 2.48% and c-direction expansion of 12.86% (2L model) [10].
model. The 101 diffraction peak that appears in the simulated XRD pattern corresponds to the 2L model of Li-h-BNIC, as shown in Fig. 3 (b). However, this diffraction peak was not observed in the experimental XRD pattern. This mismatch, which was discussed in Ref. [10], was corrected using the 1L model structure (Fig. 3 (a)). This result indicates that the two-layer stacking periodicity of pristine h-BN structure was lost because of Li intercalation. Fig. 4 shows the electrical conductivity measurements for the
Fig. 3. The result of Rietveld fitting to the Li-h-BNIC structure using (a) the 1L model and (b) the 2L model. “Obs.” is the observed pattern, “Calc.” is that calculated from the 1L and 2L models, “Delta” is the difference between “Obs.” and “Calc.” and “Bkg” is the background.
three samples collected over a temperature range from room temperature to 673 K. The a-Li3BN2 sample pellet was prepared by SPS of the Li3N/BN ¼ 1.1 powder sample, and the a-Li3BN2þLi-hBNIC sample pellet was prepared by SPS of the Li3N/BN ¼ 0.5 powder sample. These are the same powders used to collect the XRD data shown in Fig. 1. The Li-pBN sample was synthesized by annealing highly oriented pyrolytic h-BN bulk and Li metal. The electrical conductivity data corresponding to the Li-pBN was reported in Ref. [9]. The XRD pattern remained the same before and after the SPS process, with the exception of a slight intensity increase of the Li2O phase. In contrast to the Li-pBN, an electrical conductivity temperature dependence could not be measured for a-Li3BN2 and a-Li3BN2þLi-h-BNIC at lower temperatures. Table 1 shows the activation energy obtained by analysis of the electrical conductivity data measured in this study and the value reported for a-Li3BN2 by Yamane [13]. The activation energy values of a-Li3BN2 and a-Li3BN2þLi-h-BNIC reported here are similar to the value reported by Yamane. These results indicate that the conduction of the a-Li3BN2þLi-h-BNIC pellet is dominated by ionic conduction of aLi3BN2. This is likely a result of the conduction pass of the more conductive Li-h-BNIC phase not being connected and the insulating Li2O phase not blocking the ionic conduction pass in the aLi3BN2þLi-h-BNIC pellet. The absolute value of electrical conductivity and the activation energy for Li-pBN is very low because Li-hBNIC is thought to form only on the surface of the bulk sample but
Fig. 4. Electrical conductivity of a-Li3BN2, a-Li3BN2þLi-h-BNIC mixture and Li-pBN samples [9].
Please cite this article in press as: A. Sumiyoshi, et al., Good reproductive preparation method of Li-intercalated hexagonal boron nitride and transmission electron microscopy e Electron energy loss spectroscopy analysis, Solid State Sciences (2015), http://dx.doi.org/10.1016/ j.solidstatesciences.2015.04.011
4
A. Sumiyoshi et al. / Solid State Sciences xxx (2015) 1e5 Table 1 The activation energy of electrical conductivity obtained from data shown in Fig. 4 and the reported value for aLi3BN2. Sample phase
E (kJ/mol)
Li-h-BNIC, a-Li3BN2 a-Li3BN2 a-Li3BN2 [13]
62 64 78
to be connected between the edges of the sample. Fig. 5 shows low-loss EELS obtained from three different crystalline fragments of Li-h-BNIC material and pristine h-BN. The intensity profile of the p plasmon peaks of Li-h-BNIC are broader than that of h-BN. The energy position of the peak of Li-h-BNIC is almost the same as with h-BN. Although new structures appear near the onset and around 10e15 eV in the Li-h-BNIC spectra, as indicated by vertical lines, there is no apparent additional intensity on the tail of each zero-loss peak, which can be attributed to carrier excitation. This means that those Li-h-BNIC crystalline fragments may not be metal as predicted by the first-principle calculations [2,3]. Fig. 6 shows a B-K core-loss EELS of Li-h-BNIC phases and pristine h-BN, which correspond to the partial density of states with p-symmetry of the conduction bands (unoccupied states). The strong peak at 192.0 eV originates from 1s to p* transition, which is a characteristic feature of sp2 bonding of B to its three neighboring N atoms in the h-BN atomic layer. Three satellite peaks are observed at 192.6, 193.2, and 193.9 eV following Li intercalation. These peaks have often been observed in XANES spectra of 2D h-BN structured materials, such as BN thin film, BN nanotube and BCN [14e21]. The energy positions of the satellite peaks in Fig. 6 show good agreement with reported values [21], where the individual peak energies are affected by the number of O atoms adjacent to B atoms, such as
Fig. 6. B K-edge core-loss EELS data measured from the Li-h-BNIC phases and pristine h-BN. The vertical bars indicate the new spectral features that appear after Li intercalation.
BeON2, BeO2N, and BeO3 [21]. This result indicates that N defects and oxidation are induced by Li intercalation at high annealing temperatures. Therefore, the alkali metal GIC-like structure assumed in the first-principle calculations in Ref. [2,3] is not observed in the present study. The bottom of the conduction band of h-BN consists mainly of p electrons from the B atom. If the Fermilevel exists in conduction bands, the onset of the core-loss spectrum should show a sudden intensity increase where the width of the intensity increase is almost equal to the energy resolution (Fermi edge). The Fermi edge was observed in the case of Li-doped alpha-rhombohedral boron, which was metalized and showed superconductivity [22]. In Fig. 6, the Fermi edge was not observed, indicating that the Li-h-BNIC in this study is not a metal. The Fermi edge was not observed at the onset of the 1s to p* peak. Conversely, the Fermi edge was observed at the onset of the 1s to p* peak in the B K-edge core-loss spectrum measured from Li-doped a-rhombohedral boron, which metalizes and shows superconductivity [22]. The absence of the Fermi edge in the present study indicates that the Li-h-BNIC measured here is not metal. Another Li intercalation method that does not produce vacancies in the parental h-BN atomic layer will be needed for further comparison with the theoretical calculations in Ref. [2,3]. 4. Conclusion
Fig. 5. Low-loss EELS data measured from the Li-h-BNIC phases and pristine h-BN. The vertical bars indicate the new spectral features that appear after Li intercalation.
A reproducible method for the synthesis of the Li-h-BNIC phase was developed by annealing Li3N and h-BN, rather than Li metal and h-BN as reported previously [9,10]. The stacking periodicity of the BN atomic layer in the Li-h-BNIC phase changes from the twolayer stacking periodicity of h-BN to a one-layer stacking because of Li intercalation. The existence of incommensurate periodicity along the intralayer direction was observed using TEM. The band gap of
Please cite this article in press as: A. Sumiyoshi, et al., Good reproductive preparation method of Li-intercalated hexagonal boron nitride and transmission electron microscopy e Electron energy loss spectroscopy analysis, Solid State Sciences (2015), http://dx.doi.org/10.1016/ j.solidstatesciences.2015.04.011
A. Sumiyoshi et al. / Solid State Sciences xxx (2015) 1e5
the Li-h-BNIC phase was confirmed using low-loss and core-loss EELS. The existence of N atom vacancies modified by O atoms in the h-BN atomic layer was suggested from core-loss EELS. Acknowledgments One of the authors (A.S.) was supported by Research Fellowships of the Japan Society for the Promotion of Science (No. 23-5556) for Young Scientists. This work was partly supported by Scientific Research on Priority Areas of New Materials Science Using Regulated Nano Spaces, KAKENHI No. 19051005 from MEXT. References [1] [2] [3] [4]
M.S. Dresselhaus, G. Dresselhaus, Adv. Phys. 51 (2002) 1. S. Okada, M. Otani, Phys. Rev. B 81 (2010) 3e6. it, Eur. Phys. J. B 79 (2011) 301e312. B. Altintas, C. Parlak, C. Bozkurt, R. Eryig F. Oba, A. Togo, I. Tanaka, K. Watanabe, T. Taniguchi, Phys. Rev. B 81 (2010) 20e23. [5] G.L. Doll, J.S. Speck, G. Dresselhaus, M.S. Dresselhaus, K. Nakamura, S.I. Tanuma, J. Appl. Phys. 66 (1989) 2554e2558. [6] M. Sakamoto, J.S. Speck, M.S. Dresselhaus, J. Mater. Res. 1 (1986) 685e692. [7] C. Shen, S.G. Mayorga, R. Biagioni, C. Piskoti, M. Ishigami, A. Zettl, N. Bartlett,
5
J. Solid State Chem. 147 (1999) 74e81. [8] N.I. Kovtyukhova, Y. Wang, R. Lv, M. Terrones, V.H. Crespi, T.E. Mallouk, J. Am. Chem. Soc. 135 (2013) 8372e8381. [9] A. Sumiyoshi, H. Hyodo, K. Kimura, J. Phys. Chem. Solids 71 (2010) 569e571. [10] A. Sumiyoshi, H. Hyodo, K. Kimura, J. Solid State Chem. 187 (2012) 208e210. [11] Y. Sato, M. Terauchi, M. Mukai, T. Kaneyama, K. Adachi, Ultramicroscopy 111 (2011) 1381e1387. [12] F. Izumi, K. Momma, Solid State Phenom. 130 (2007) 15e20. [13] H. Yamane, S. Kikkawa, M. Koizumi, J. Solid State Chem. 71 (1987) 1e11. [14] I. Jimenez, A.F. Jankowski, L.J. Terminello, D.G.J. Sutherland, J.A. Carlisle, G.L. Doll, W.M. Tong, D.K. Shuh, F.J. Himpsel, Phys. Rev. B 55 (1997) 25. [15] M. Niibe, K. Miyamoto, T. Mitamura, K. Mochiji, J. Vac. Sci. Technol. A 28 (2010) 1157. [16] M. Petravic, R. Peter, I. Kavre, L.H. Li, Y. Chen, L.-J. Fan, Y.-W. Yang, Phys. Chem. Chem. Phys. 12 (2010) 15349. [17] A.B. Preobrajenski, M.L. Ng, N.A. Vinogradov, A.S. Vinogradov, E. Lundgren, A. Mikkelsen, N. Martensson, Nano Lett. 9 (2009) 2780. [18] H.C. Choi, S.Y. Bae, W.S. Jang, J. Park, H.J. Song, H.-J. Shin, J. Phys. Chem. B 109 (2005) 7007. [19] L. Li, L.H. Li, Y. Chen, X.J. Dai, T. Xing, M. Petravic, X. Liu, Nanoscale Res. Lett. 7 (2012) 417. [20] T. Hemraj-Benny, S. Banerjee, S. Sambasivan, D.A. Fischer, W. Han, J.A. Misewich, S.S. Wong, Phys. Chem. Chem. Phys. 7 (2005) 1103. [21] I. Caretti, I. Jime'nez, J. Appl. Phys. 110 (2011) 023511. [22] T. Nagatochi, H. Hyodo, A. Sumiyoshi, K. Soga, Y. Sato, M. Terauchi, F. Esaka, K. Kimura, Phys. Rev. B 83 (2011) 1e5.
Please cite this article in press as: A. Sumiyoshi, et al., Good reproductive preparation method of Li-intercalated hexagonal boron nitride and transmission electron microscopy e Electron energy loss spectroscopy analysis, Solid State Sciences (2015), http://dx.doi.org/10.1016/ j.solidstatesciences.2015.04.011