Solid State Communications 152 (2012) 100–103
Contents lists available at SciVerse ScienceDirect
Solid State Communications journal homepage: www.elsevier.com/locate/ssc
Local atomic structure in molten Si3 Sb2 Te3 phase change material Xuelai Li a,c,∗ , Zhimei Sun b , Feng Rao a , Zhitang Song a , Weili Liu a , Baisheng Sa b a
State Key Laboratory of Functional Materials for Informatics, Laboratory of Nanotechnology, Shanghai Institute of Micro-system and Information Technology, Chinese Academy of Sciences, 200050 Shanghai, China b
Department of Materials Science and Engineering, College of Materials, Xiamen University, 361005 Xiamen, China
Graduate School of the Chinese Academy of Sciences, 100080 Beijing, China
Article history: Received 7 December 2010 Received in revised form 8 August 2011 Accepted 24 October 2011 by Xincheng Xie Available online 28 October 2011
abstract By means of ab initio molecular dynamics calculations, we have studied the local structures of liquid and amorphous Si3 Sb2 Te3 . The results show that all the constitute elements in liquid Si3 Sb2 Te3 are octahedrally coordinated. While in amorphous state, Sb and Te atoms are mainly octahedrally coordinated and Si atoms are mainly tetrahedrally coordinated. In both states, Si is mainly homo-bonded by Si. Finally, we proposed a phase separation model for liquid and amorphous Si3 Sb2 Te3 , which is responsible for the good performance of Si3 Sb2 Te3 alloy as a phase change material. © 2011 Elsevier Ltd. All rights reserved.
Keywords: A. Phase change materials A. Semiconductors D. Ab initio molecular dynamics
1. Introduction Phase change memory (PCM) has been considered as a promising technology for next-generation non-volatile data storage. The essential element of the PCM is the phase change material, which exists in at least two phases with remarkably different electrical properties and can be rapidly cycled between these phases [1–3]. In the past decade, Ge2 Sb2 Te5 alloy as a leading candidate for phase change (PC) materials has been extensively studied [4–6]. However, for practical applications of Ge2 Sb2 Te5 in PCM, it is still necessary to decrease its melting temperature (Tm ) and increase the resistance of crystalline Ge2 Sb2 Te5 . Therefore, searching new PC materials or tuning the properties of Ge2 Sb2 Te5 to meet the above criteria is a hot topic in the phase change community. In our previous work, we have demonstrated that SiSbTe alloys are new type of phase change materials [7–9], which exhibit a lower threshold current, quicker programming speed, and better data retention in comparison to Ge2 Sb2 Te5 alloy. However, the mechanism of such outstanding behaviors are not clear, although we have investigated the crystalline structure of SiSbTe alloy, which is a nano-composite material consisting of separated phases  of amorphous Si, crystalline Sb2 Te3 and/or crystalline Te. In
addition, there are still two questions deserving further study: (1) the liquid and amorphous structures of SiSbTe, and (2) the whole phase change process. In these respects, ab initio molecular dynamics calculations (AIMD), which have been used to understand the liquid and amorphous states of GeSbTe alloys [6,11–13], can provide a good approach. In this report, we used AIMD to investigate the structures of liquid Si3 Sb2 Te3 (l-Si3 Sb2 Te3 ) and amorphous Si3 Sb2 Te3 (a-Si3 Sb2 Te3 ) and to understand the phase change process of melt–quench–crystallize of Si3 Sb2 Te3 . 2. Calculation methods Ab initio molecular dynamics (AIMD) implemented in the Vienna ab initio simulation package [14,15] was used in the present work. The density functional theory based calculations were performed in the generalized gradient approximation (GGA)  using the Perdew–Burke–Ernzerhof (PBE)  functional and the projected augmented wave (PAW)  potentials as implemented in the VASP code. The energy cutoff of 184 eV was chosen. The Gamma point was used for the electronic structure calculations and the Gaussian smearing was applied. The initial structure contained 216 atoms (81 Si, 54 Sb, and 81 Te) distributed randomly in a cubic cell with the calculated equi3
Correspondence to: Shanghai Institute of Micro-system and Information Technology, Chinese Academy of Sciences, 865 Changning Road, Shanghai 200050, China. Tel.: +86 021 62511070 8408; fax: +86 021 62134404. E-mail address: [email protected]
(X. Li). 0038-1098/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.ssc.2011.10.033
librium volume of 5736.77 Å (density of 5.55 g/cm3 , close to the experimental value of 5.76 g/cm3 for amorphous Si3 Sb2 Te3 which we measured by the X-ray reflection (XRR) method). The system was first melted and thermalized at 2300 K, and
X. Li et al. / Solid State Communications 152 (2012) 100–103
Fig. 1. (a) The calculated total structure factors and (b) calculated total pair correlation functions of liquid and amorphous Si3 Sb2 Te3 .
Fig. 2. Bond angle distributions of (a) l-Si3 Sb2 Te3 and (b) a-Si3 Sb2 Te3 .
then gradually quenched down to 1100 K (about 200 K above the melting temperature [19,20]) at a quenching rate of 5.0 × 1013 K/s. The liquid structure was further equilibrated for 9 ps at 1100 K. In order to generate a model of a-Si3 Sb2 Te3 , the liquid was continued quenching to 300 K at the same rate, and finally rethermalized at 300 K for 9 ps. 3. Results and discussion Fig. 1 shows the total structure factor (SF) and total paircorrelation functions (PCF) of l-Si3 Sb2 Te3 as well as a-Si3 Sb2 Te3 . It was proposed  that the ratio between the heights of the first two peaks of S(Q) is an indicator of the local bonding geometry. Namely, the value of the ratio R = S (Q1 )/S (Q2 ), where Q1 and Q2 are the positions of the first two peaks of the structure factor, is greater than 1 for an octahedral liquid and smaller than 1 for a tetrahedral liquid. The values of R for l-Si3 Sb2 Te3 and a-Si3 Sb2 Te3 are 1.36 and 1.12 (see Fig. 1(a)), respectively. Therefore, l-Si3 Sb2 Te3 is an octahedral-like geometry, in agreement with those of liquid GeSbTe and Sb2 Te3 [22,23]. Since the value of R for the a-Si3 Sb2 Te3 is (equal to 1.12) close to 1, it should be a defectively octahedral structure as a whole. Therefore, we proposed it as a mixture of octahedral (mainly) and tetrahedral (secondarily) structures. The calculated total PCFs for l-Si3 Sb2 Te3 and a-Si3 Sb2 Te3 are shown in Fig. 1(b). The a-Si3 Sb2 Te3 shows medium to long-ranged order up to 7.63 Å, with peaks at 2.63, 3.93 and 6.03 Å and minima at 3.13, 4.98 and 6.98 Å. The weaker features in the total PCF for l-Si3 Sb2 Te3 indicate much reduced order, with first
two peaks at 2.88 and 3.68 Å and minima at 3.43 and 4.98 Å, and there is no structure after the third maximum (5.98 Å). These positions of peaks and minima are smaller than those of Ge2 Sb2 Te5 , indicating smaller covalent bond lengths in l-Si3 Sb2 Te3 and a-Si3 Sb2 Te3 . Fig. 2(a) and (b) show bond angle distributions around Si, Sb and Te atoms in l-Si3 Sb2 Te3 and a-Si3 Sb2 Te3 , respectively. As seen in Fig. 2(a), maximal peaks are observed at 96°, 88°, and 87° around Si, Sb and Te atoms, respectively. The results demonstrate an octahedral-like geometry, similar to the case of liquid GeSbTe and Sb2 Te3 [12,13,23]. In addition, the small broad peaks at around 150° indicate that the octahedrally coordinated l-Si3 Sb2 Te3 is distorted. For a-Si3 Sb2 Te3 as seen in Fig. 2(b), the maximal peaks are observed at 87° and 88° around Sb and Te atoms, respectively. The results indicate the octahedrally coordinated Sb and Te atoms, consistent with the Sb and Te atoms in amorphous GeSbTe and amorphous Sb2 Te3 [12,22]. As occurs in amorphous GeSbTe and Sb2 Te3 , the small broad peaks at around 150° indicate that the Sb and Te atoms are always in the defective octahedral environment. While the maximum peak around Si is shifted to 108°, indicating tetrahedrally coordinated Si atoms, as that of amorphous Si . The fractional distribution of coordination numbers for various species of l-Si3 Sb2 Te3 , as shown in Fig. 3(a), agrees with the above pictures of bond angle distributions. It is seen that Si atoms are mainly a mixture of fourfold, fivefold and sixfold coordinated, indicating that most Si atoms are octahedrally coordinated. Sb and Te atoms are mainly threefold, fourfold and fivefold coordinated. For a-Si3 Sb2 Te3 , Si atoms are mainly
X. Li et al. / Solid State Communications 152 (2012) 100–103
Fig. 3. Distributions of coordination number for different species in (a) l-Si3 Sb2 Te3 and (b) a-Si3 Sb2 Te3 .
fourfold coordinated with some contribution from fivefold and sixfold coordinations, indicating a tetrahedral geometry. Sb and Te atoms are mainly a mixture of threefold, fourfold and fivefold coordinations, similar to that of amorphous GeSbTe and Sb2 Te3 system [12,25]. The partial and total coordination numbers (Z ) for each element are given in Table 1. It is clear that, in both l-Si3 Sb2 Te3 and a-Si3 Sb2 Te3 , the Si atoms are more likely coordinated with Si atoms than Sb and Te atoms. In total, the average coordination number for Si atoms in l-Si3 Sb2 Te3 is equal to 5.28, which suggests octahedrally coordinated Si atoms in l-Si3 Sb2 Te3 . While for Si atoms in a-Si3 Sb2 Te3 , the average coordination number is 4.19, which indicates tetrahedrally coordinated Si atoms in a-Si3 Sb2 Te3 . In both l-Si3 Sb2 Te3 and a-Si3 Sb2 Te3 states, the average coordination number of Sb and Te atoms is lower than the idea octahedral value of six, resulting in neighboring vacancies, which may be important for the phase change materials [11,26]. As seen in Fig. 4, the Si concentrations in and out of the circles are different, higher in the circles and lower out of the circles. So we proposed a phase separation model for l-Si3 Sb2 Te3 and a-Si3 Sb2 Te3 , which consists of Si rich part (high concentration of Si atoms in the circles) and Sb2 Te3 rich part (low concentration of Si atoms out of the circles). In order to clearly see the structure of Si atoms, we also draw the Si–Si homobonds in Fig. 4. In l-Si3 Sb2 Te3 phase, both parts are in the same geometry, defectively octahedral geometry. While in a-Si3 Sb2 Te3 phase, Si rich part is tetrahedral geometry, like that of amorphous Si, and Sb2 Te3 rich part is octahedral geometry. When crystallizes from a-Si3 Sb2 Te3 to crystalline Si3 Sb2 Te3 (c-Si3 Sb2 Te3 ) (set process), just the Sb2 Te3 rich part crystallizes, while the Si rich part keeps amorphous state. When melts from c-Si3 Sb2 Te3 to l-Si3 Sb2 Te3 (reset process), the local structure changes, while the two parts still exist. Because the crystallization does not happen in the whole material, the set speed may be faster than those whole crystallized PC materials. Since the Si rich part keeps amorphous state, the electrical resistivity of c-Si3 Sb2 Te3 may be higher than crystalline Sb2 Te3 , and hence reducing the power consumption of the set process. In the whole cycled phase change process, melt–quench–crystallize, the Si3 Sb2 Te3 is inclined to keep phase separated into two parts, hence the data reliability should be good. 4. Conclusion To summarize, we have provided a model of l-Si3 Sb2 Te3 and a-Si3 Sb2 Te3 by quenching from the melt within AIMD simulations. Based on the analysis of bond angle distributions and coordination number distributions of each species of l-Si3 Sb2 Te3 and a-Si3 Sb2 Te3 , we found that the local geometry of Si, Sb and
Fig. 4. Atomic configurations of (a) l-Si3 Sb2 Te3 and (b) a-Si3 Sb2 Te3 structures. The blue, yellow and red balls represent Si, Sb and Te atoms, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Table 1 The estimated average coordination number (Z ) for various pairs of atoms in the l-Si3 Sb2 Te3 . Data for a-Si3 Sb2 Te3 are given in parenthesis.
Si Sb Te
3.01 (2.27) 1.28 (1.11) 1.42 (1.18)
0.85 (0.74) 1.44 (1.30) 1.38 (1.21)
1.42 (1.18) 2.07 (1.81) 0.99 (0.54)
5.28 (4.19) 4.69 (4.22) 3.79 (2.93)
Te atoms in l-Si3 Sb2 Te3 are all octahedrally coordinated, while in a-Si3 Sb2 Te3 Si atoms are mainly tetrahedrally coordinated and Sb and Te atoms are mainly octahedrally coordinated. Si atoms are mainly homobonded with Si atoms in both states, and we proposed a phase separation model for liquid and amorphous Si3 Sb2 Te3 . Such a unique structure of liquid and amorphous Si3 Sb2 Te3 may play an important role on its outstanding performance in the application of phase change memory. Acknowledgments This work is partially supported by National Integrate Circuit Research Program of China (2009ZX02023-003), National Key Basic Research Program of China (2010CB934300, 2011CBA00602, 2011CB932800), National Natural Science Foundation of China (60906004, 60906003, 61006087, 61076121), Science and Technology Council of Shanghai ( 1052nm07000), and Supercomputing Center, CNIC, CAS. References  S.R. Ovshinsky, Phys. Rev. Lett. 21 (1968) 1450.  G.I. Meijer, Science 319 (2008) 1625.  S. Raoux, R.M. Shelby, J. Jordan-Sweet, et al., Microelectron. Eng. 85 (2008) 2330.
X. Li et al. / Solid State Communications 152 (2012) 100–103            
S.H. Lee, Y. Jung, R. Agarwal, Nat. Nano. 2 (2007) 626. Z. Sun, J. Zhou, R. Ahuja, Phys. Rev. Lett. 96 (2006) 55507. Z. Sun, J. Zhou, R. Ahuja, Phys. Rev. Lett. 98 (2007) 55505. T. Zhang, Z. Song, F. Rao, et al., Japan J. Appl. Phys. Part 2 46 (2007) L247. T. Zhang, Z. Song, B. Liu, et al., Solid-State Electron. 51 (2007) 950. T. Zhang, Z. Song, M. Sun, et al., Appl. Phys. A 90 (2008) 451. Y. Cheng, N. Yan, X. Han, et al., J. Non-Cryst. Solids 18–19 (2010) 884. S. Caravati, M. Bernasconi, T.D. Kuhne, et al., Appl. Phys. Lett. 91 (2007) 171906. S. Caravati, M. Bernasconi, M. Parrinello, Phys. Rev. B 81 (2010) 014201. C. Bichara, M. Johnson, J.P. Gaspard, Phys. Rev. B 75 (2007) 060201. G. Kresse, J. Hafner, Phys. Rev. B 48 (1993) 13115. J. Hafner, J. Comput. Chem. 29 (2008) 2044.
         
J.P. Perdew, Y. Wang, Phys. Rev. B 45 (1992) 13244. J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865. G. Kresse, D. Joubert, Phys. Rev. B 59 (1999) 1758. T. Zhang, Z. Song, B. Liu, et al., Chin. Phys. 16 (2007) 2475. B. Qiao, J. Feng, Y. Lai, et al., Semicond. Sci. Technol. 21 (2006) 1073. C. Steimer, M.V. Coulet, W. Welnic, et al., Adv. Mater. 20 (2008) 4535. J. Akola, R.O. Jones, J. Phys.: Condens. Matter. 20 (2008) 465103. Z. Sun, J. Zhou, A. Blomqvist, et al., J. Phys.: Condens. Matter 20 (2008) 205102. I. Štich, R. Car, M. Parrinello, Phys. Rev. B 44 (1991) 11092. S. Caravati, M. Bernasconi, T.D. Kühne, et al., J. Phys.: Condens. Matter. 21 (2009) 255501.  Z. Sun, J. Zhou, A. Blomqvist, et al., Phys. Rev. Lett. 102 (2009) 075504.