J. Phys. Chem. C 2007, 111, 2421-2425
2421
One-Dimensional Phase-Change Nanostructure: Germanium Telluride Nanowire Xuhui Sun,* Bin Yu,* Garrick Ng, and M. Meyyappan Center for Nanotechnology, NASA Ames Research Center, Moffett Field, California 94035 ReceiVed: September 8, 2006; In Final Form: December 14, 2006
High-quality nanowires of germanium telluride (GeTe), a one-dimensional chalcogenide phase-change nanostructure, were synthesized via thermal evaporation method under vapor-liquid-solid mechanism. The physical morphology, chemical composition, and crystal structure of the as-synthesized GeTe nanowires were investigated by scanning electron microscopy, energy dispersive X-ray spectroscopy, high-resolution transmission electron microscopy (HR-TEM), and X-ray photoemission spectroscopy. Through real-time TEM imaging of nanowire sample heated in an incrementally controllable heating system, the melting point of a single crystalline GeTe nanowire (∼40-80 nm diameter) is found to be significantly lower than that of its bulk counterpart (46% reduction, from 725 to 390 °C). The significant reduction in melting point makes one-dimensional phase-change chalcogenide nanowire a potential material for application in low-power highdensity resistive switching nonvolatile data storage in which the thermal energy for material phase transition would be significantly reduced.
Introduction Interest in inorganic nanowires continues to grow, fuelled by the potential ability to engineer their properties for various applications in memory, logic, and sensor devices. Phase-change materials (PCMs) are among the most promising media for nonvolatile, rewritable, and highly endurable data storage. The idea to employ material structural phase changes for data storage dates back to the 1960s when Ovshinsky suggested a memory switch based on phase transition of multicomponent chalcogenides.1 This technology was developed into the mainstream optical data storage media such as CDs and DVDs that are widely used today.2-7 The concept has been later extended to develop electrically operated phase-change random access memory (PRAM).8-13 Phase-change memories rely on binary or multiple reflective/resistive states of the programmable element to represent the logic levels for data storage. The optical reflectivity or electrical resistance in the material is reversibly switched due to thermally induced transition between two stable material phases: the orderly single crystalline or polycrystalline phase (c-phase) and the less orderly amorphous phase (R-phase). In particular, resistive switching PRAM features faster write/ read, improved endurance, and much simpler fabrication as compared with the transistor-based nonvolatile memories. Because the data is stored in the form of material phases, PRAM offers soft-error- or radiation-free operation. A major issue limiting the extensive use of thin-film-based PRAM is the enormous programming current to generate the thermal energy needed for phase change. This is particularly a concern in the crystal to amorphous (c-to-R) phase transition when high current is required for melting. The Joule heating effect may cause power dissipation issues and intercell thermal interference, preventing possibilities for future scaling. The phase change behavior of chalcogenide materials in the nanoscale may overcome these limitations14-15 and therefore warrant a detailed investigation. The synthesis of GeTe and Sb2Te3 * To whom correspondence should be addressed. E-mail: byu@ arc.nasa.gov (B.Y.);
[email protected] (X.S.).
nanowires by vapor transport method has been reported previously;15-16 however, the nanowires were mixtures of different morphologies (e.g., straight and helical)15-16 or with other impurities (amorphous GeO2 nanowires)16 and in addition, they were rhombohedral structures which is the stable phase requiring more activation energy for phase-transition between crystalline and amorphous phases.17 In this paper, we report the synthesis, material characterization, and melting point measurement of high-purity, single crystalline (f.c.c structure) phase change germanium telluride (GeTe) nanowire. Experimental Section Our experiment was carried out in a high vacuum, two zone furnace system. High-purity GeTe powder (99.999%, Aldrich) was placed in the middle of the high-temperature zone. Gold nanoparticles (20 nm) were dispersed as catalyst on thermal SiO2 coated Si (100) substrate for GeTe nanowire growth. The substrate was placed downstream in the reactor tube (∼1/4 tube length from the end) in the lower temperature zone. The reactor tube was evacuated to a base pressure of 10-2 Torr prior to the experiment. The carrier gas, argon mixed with hydrogen (20%), was introduced at a flow rate of 25 sccm (standard cubic centimeters per minute) and at a pressure of 200 Torr. The furnace temperature was increased to 680∼720 °C and maintained for 1 h. The temperature in the downstream zone was ∼450 °C as monitored by a thermocouple. This thermocouple measures the outer wall temperature in this zone, and the temperature of the substrate located near the center of the tube may be lower by about 10 degrees as determined by controlled experiments using another thermocouple. The morphology and structure of the synthesized products were characterized by scanning electron microscopy (Hitachi S-4000 FEG) and transmission electron microscopy (TEM Philips CM20, operated at 200 kV) equipped with energy dispersive X-ray spectroscopy (EDS) for chemical composition analysis. X-ray photoelectron spectroscopy (XPS) measurement was performed in an SSI S-Probe Monochromatized Spectrometer using Al KR radiation (1486 eV) as a probe. GeTe nanowires were dispersed onto the
10.1021/jp0658804 CCC: $37.00 © 2007 American Chemical Society Published on Web 01/24/2007
2422 J. Phys. Chem. C, Vol. 111, No. 6, 2007
Sun et al.
Figure 1. SEM image of the as-synthesized GeTe nanowires on SiO2coated Si (100) substrate using 20 nm Au nanoparticles as the growth catalyst. Scale bar is 1 µm. Inset is a close-up view of a catalyst bead at the tip of the single GeTe nanowire. Scale bar (inset) is 200 nm.
SiOx-film-coated TEM grids and were subjected to a real-time melting point measurement experiment under the TEM. Results and Discussion Figure 1 shows a SEM image of the as-synthesized GeTe nanowires on the surface of the SiO2 substrate, showing highyield nanowire growth. The nanowires have a diameter in the range of 40∼80 nm and are up to tens of micrometers in length. Au catalyst beads are visible at the tips of the nanowires as seen in the inset, implying that the nanowires were possibly grown under the vapor-liquid-solid (VLS) mechanism. No growth of GeTe nanowire was observed on SiO2 substrate without Au catalyst in a controlled experiment, further confirming the metal-catalyzed VLS growth. Further structural characterization of the GeTe nanowires was performed with TEM. Figure 2a shows a low-magnification TEM image of a single GeTe nanowire with uniform diameter of about 40 nm along its entire length. A selected area electron diffraction (SAED) pattern taken from the nanowire (shown in the inset) reveals that the nanowire is a single crystal with cubic lattice structure. The EDS analysis of an individual nanowire with locally focused beam spot confirms that the nanowire is composed of only germanium and tellurium with an atomic ratio close to 1:1. The trace O peak is from the oxide outer layer. The high-resolution TEM (HR-TEM) image analysis further indicates that the as-prepared GeTe nanowires are high-quality single crystals. Panels a and b of Figure 3 show that the GeTe nanowire is structurally uniform and contains no noticeable defects such as dislocations and stacking faults. A layer of 1∼3 nm amorphous oxide shell, covering the GeTe nanowire surface which exhibits an atomically sharp interface, was formed probably either during the growth due to oxygen leakage or afterward when the nanowires were exposed to the ambient environment. It is known that the surface of GeTe is oxidized spontaneously in the presence of atmospheric air to form GeO2‚TeO2.18-20 The lattice spacing of ∼3.5 Å in Figure 3a and ∼2.1 Å in Figure 3b corresponds to the d-spacing of the
Figure 2. Morphology and composition of germanium telluride nanowire. (a) Low-magnification TEM image of an individual GeTe nanowire with a diameter of about 40 nm. The nanowire length is over 5 µm. The inset shows the SAED pattern of an fcc cubic lattice structure. (b) EDS spectrum of the same GeTe nanowire as shown in panel a.
(111) and (220) crystal planes, respectively, of GeTe with a cubic lattice structure. HR-TEM imaging results and SAED pattern suggest that the GeTe nanowire is single crystal of the cubic structure with lattice constant a of ∼6.01 Å, which is consistent with the JCPDS PDF 03-065-0415 (cubic structure, Fm3m, a ) 6.02 Å). The GeTe nanowires have a preferential growth direction in the 〈110〉 crystalline orientation. At room temperature, the GeTe crystal has rhombohedral structure (R3m), which is transformed to cubic structure (NaCl structure) at high temperatures (∼446 °C).21 The rhombohedral
Germanium Telluride Nanowire
J. Phys. Chem. C, Vol. 111, No. 6, 2007 2423
Figure 4. XPS spectra of germanium telluride nanowires with peak representing (a) Ge 3d and (b) Te 3d5/2. The binding energy information indicates the existence of GeO2 and TeO2 in each case, respectively.
Figure 3. High-resolution TEM images of the crystal structure of germanium telluride nanowires. (a) From one sample, the individual GeTe nanowire shows (111) planes with an interplane spacing of ∼0.35 nm. (b) From another sample, the GeTe nanowire shows (220) planes with an interplane spacing of ∼0.21 nm. Both nanowire samples show an elongation along the preferential 〈110〉 crystalline orientation. A thin (1∼3 nm) amorphous oxide layer is observed on the surface of GeTe nanowire in both samples.
structure can be viewed as a distorted NaCl-type structure. The degree of the distortion from the cubic NaC1-type structure is reflected in the angle R ) 88.35° between the axes of the facecentered rhombohedral unit cell, which differs slightly from the undistorted value of 90°. With increasing temperature, the angle R increases from the room-temperature value of 88.35° continu-
ously to the undistorted value of 90° at about 446.5 °C. The growth temperature of GeTe nanowires is above the transition temperature of GeTe crystal from rhombohedral to cubic structure. Therefore, the nanowires grown at high-temperature exhibit cubic structures. The electronic and chemical state of GeTe nanowires were studied by XPS. Figure 4, panels a and b, shows the XPS spectra of Ge 3d and Te 3d5/2, respectively. The binding energies at 30.0 eV in Figure 4a and 572.4 eV in Figure 4(b) are attributed to Ge 3d and Te 3d5/2 in GeTe, respectively, which is in agreement with values reported in the literature.20,22-24 The XPS analysis indicates that the nanowires are GeTe compound. The peaks at 32.8 eV in Figure 4a and 576.2 eV in Figure 4b are due to GeO2 and TeO2, respectively, from the thin oxide outer layer. Because XPS is a surface sensitive tool with a sampling depth of several nanometers, even 1∼3 nm GeO2‚TeO2 outer layer on GeTe nanowires can be clearly detected. A quantitative analysis from XPS data further confirms that the atomic ratio of Ge and Te is close to 1:1. For phase change materials, the melting point (Tm) is an important physical property relevant to the performance of data storage devices. Lower Tm represents that the c-to-R phase transition, corresponding to the memory reset activity, can be executed with a lower thermal energy. In our investigation, the Tm of GeTe nanowires was measured via an in situ heating experiment in a TEM chamber under real-time morphology
2424 J. Phys. Chem. C, Vol. 111, No. 6, 2007
Sun et al. Upon reaching the Tm (390 °C), the contrast of GeTe nanowire abruptly turned dark and the single crystalline electron diffraction pattern disappeared as the material changed from solid crystalline phase to liquid phase (ED data shown in the inset of Figure 5). The molten nanowire further vaporized out from the two open ends of the nanowire, and a nanotube structure was left behind. At 460 °C, clear evaporation was noticed with the exception of a few very long nanowires in the sample. A TEM image was taken during the evaporation process, as shown in Figure 5. The very thin wall of the remaining tube, confirmed from EDS analysis, is the GeO2 oxide outerlayer of GeTe nanowire. Because bulk GeO2 (Tm ) 1115 °C) has much higher melting point than that of bulk GeTe and TeO2 (Tm ) 733 °C), it turns out that only the GeO2 shell was left behind when the temperature was increased up to the Tm of GeTe nanowire. The native GeO2 outerlayer provides an excellent natural protection in the device performance in which no extra outerlayer is needed in the device fabrication to prevent oxidation and evaporation. A real-time video of the recorded melting and evaporation process is available in the Supporting Information. The disappearance of electron diffraction pattern, the contrast and meniscus at the interface in Figure 5 and the video confirm that the phase change process is indeed melting rather than sublimation of GeTe. Similar melting process and reduction in melting point were also reported for In2Se3 nanowires25 and germanium nanowires.26 Finally, it is noted that even though the TEM image in Figure 5 and the video capture the process of a single wire, the sample itself consisted of an ensemble of wires. Therefore, the observed reduction in melting point is for nanowires with diameters in the 40-80 nm range. The dependence on diameter as reported for metal and semiconductor particles26-28 cannot be determined using the present setup and approach. When the targeted heating temperature (i.e. melting point) in a nanowire changes from 725 to 390 °C, we can expect a corresponding decrease in the necessary amount of input heat based on the analytical solution to the heat conduction equation for a long, very thin cylinder with a uniform heat source such as electric current.29 This is very critical since a major limitation in thin-film-based PRAM devices is the large amount of heat required for melting which necessitates large programming current for switching the memory state. It is expected that further reduction in the GeTe nanowire diameter may possibly result in even lower melting point and therefore further reduction in programming energy.
Figure 5. Measurement of the melting point of a single germanium telluride nanowire under real-time TEM morphology monitoring. (a) The GeTe nanowire is at room temperature. The inset shows an SAED pattern of the crystalline structure with fcc crystal structure. (b) The nanowire is molten and its mass is gradually lost through evaporation starting from 390 to 460 °C. The remaining oxide nanotube can be clearly seen from the image.
monitoring. In the setup, the sample stage was resistively heated at a rate of 10 °C/min. The melting temperature of the nanowire is identified as the point at which the electron diffraction pattern disappears and the nanowire starts to melt. The melting and evaporation process of the GeTe nanowires was monitored in real time by TEM and recorded by a video camera in the brightfield mode. We observed a significant drop of Tm for GeTe nanowire from that for bulk GeTe. GeTe nanowires started melting at 390 °C, which is 46% less than the bulk melting point of GeTe (725 °C). Figure 5a shows the TEM image of a single GeTe nanowire before melting with its ED pattern shown in the inset.
Summary We have presented a bottom-up approach to synthesize a onedimensional chalcogenide phase-change material: germanium telluride nanowire. Single-crystal GeTe nanowires with nonlithographically defined diameter down to 40 nm have been produced via metal-catalyzed VLS growth. Gold was used as the catalyst metal which appears at the tip of the nanowire. Previous works30 on device fabrication either used a chemical mechanical polishing or acid wash step to remove the catalyst. Alternatively industry benign catalysts or self-catalyzing schemes reported31,32 for Si, Ge, and oxide wires may be attempted. The GeTe nanowires possess excellent morphology with fcc cubic lattice structure. The 40-80 nm diameter nanowires exhibit a melting point as low as 390 °C, significantly reduced from that of bulk GeTe, and therefore exhibit great potential for future data storage application with low thermal energy required for phase switching operation.
Germanium Telluride Nanowire Acknowledgment. B.Y. and X.S. are with the University Affiliated Research Center (UARC) at NASA Ames Research Center and their work was supported by a NASA contract to UARC. G.N. is a graduate student at SJSU, supported by NASA Ames Associate Program. Supporting Information Available: Video clip (MPEG file) of the germanium telluride nanowire melting process as recorded by a video camera. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Ovshinsky, S. R. Phys. ReV. Lett. 1968, 21, 1450. (2) Adler, D.; Shur, M. S.; Silver, M.; Ovshinsky, S. R. J. Appl. Phys. 1980, 51, 3289. (3) Chen, M.; Rubin, K.; Barton, R. Appl. Phys. Lett. 1986, 49, 502. (4) Yamada, N.; Ohno, E.; Nishiochi, K.; Akahira, N.; Takao, M. J. Appl. Phys. 1991, 69, 2849. (5) Coombs, J.; Jongenelis, A.; van Es-Spiekman, W.; Jacobs, B. J. Appl. Phys. 1995, 78, 4906. (6) Volkert, C. A.; Wuttig, M. J. Appl. Phys. 1999, 86, 1808. (7) Yamada, N.; Matsunaga, T. J. Appl. Phys. 2000, 88, 7020. (8) Lai, S.; Lowrey, T. IEDM Tech Dig. 2001, 803. (9) Lai, S. IEDM Tech. Dig. 2003, 255. (10) Pirovano, A.; Lacaita, A. L.; Benvenuti, A.; Pellizzer, S.; Bez, R. IEEE Trans Elec. DeV. 2004, 51, 452. (11) Kotz, J.; Shaw, M. P. J. Appl. Phys. 1984, 55, 427. (12) Hwang, Y. N.; Lee, S. H.; Ahn, S. J. IEDM Tech. Dig. 2003, 893. (13) Pirovano, A.; Lacaita, A. L.; Benvenuti, A.; Pellizzer, F.; Hudgens, S.; Bez, R. IEDM Tech. Dig. 2003, 699. (14) Lankhorst, M. H. R.; Ketelaars, B. W. S. M. M.; Wolters, R. A. M. Nat. Mater. 2005, 4, 347-352.
J. Phys. Chem. C, Vol. 111, No. 6, 2007 2425 (15) Yu, D.; Wu, J. Q.; Gu, Q. A.; Park, H. K. J. Am. Chem. Soc. 2006, 128, 8148-8149. (16) Meister, S.; Peng, H. L.; McIlwrath, K.; Jarausch, K.; Zhang, X. F.; Cui, Y. Nano Lett. 2006, 6, 1514-1517. (17) Yamada, N.; Ohno, E.; Nishiuchi, K.; Akahira, N.; Takao, M. J. Appl. Phys. 1991, 69, 2849-2856. (18) Gogishvili, O. S.; Degaltsev,.A. N.; Kononov, G. G.; Lavrinenko, I. P.; Lalykin, S. P. Inorg. Mater. 1988, 24, 944. (19) Sheveleva, T. F.; Plaksina, Y. B.; Markholiya, T. P. Inorg. Mater. 1976, 12, 791. (20) Yashina, L. V.; Kobeleva, S. P.; Shatalova, T. B.; Zlomanov, V. P.; Shtanov, V. I. Solid State Ionics 2001, 141, 513. (21) Chattopadhyayt, T.; Boucherlet, J. X.; von Schnering, H. G. J. Phys. C: Solid State Phys. 1987, 20, 1431. (22) Lippens, P. E.; Brousse, E.; Jumas, J. C. J. Phys. Chem. Solids 1999, 60, 1663. (23) Shalvoy, R. B.; Fisher, G. B.; Stiles, P. J. Phys. ReV. B 1977, 15, 1680. (24) Shevchik, N. J.; Tejeda, J.; Langer, D. W.; Cardona, M. Phys. ReV. Lett. 1973, 30, 659. (25) Sun, X.; Yu, B.; Ng, G.; Nguyen, T. D.; Meyyappan, M. Appl. Phys. Lett. 2006, 89, 233121. (26) Wu, Y.; Yang, P. AdV. Mater. 2001, 13, 520. (27) Buffat, P.; Borel, J. P. Phys. ReV. 1976, 13, 2287. (28) Goldstein, A. N.; Echer, C. M.; Alivisatos, A. P. Science 1992, 256, 1425. (29) Carslaw, H. S. Jaeger, J. C. Conduction of Heat in Solids; Oxford Clarendon Press: New York, 1947. (30) Ng, H. T.; Han, J.; Yamada, T.; Nguyen, P.; Chen, Y. P.; Meyyappan, M. Nano Lett. 2004, 4, 1247. (31) Sun, X. H.; Calebotta, G.; Yu, B.; Selvaduray, G.; Meyyappan, M. (submitted). (32) Nguyen, P.; Ng, H. T.; Meyyappan, M. M. AdV. Mater. 2005, 17, 1773.
