Letter pubs.acs.org/NanoLett
Polymorphism of GeSbTe Superlattice Nanowires Chan Su Jung,† Han Sung Kim,† Hyung Soon Im,† Young Seok Seo,† Kidong Park,† Seung Hyuk Back,† Yong Jae Cho,† Chang Hyun Kim,† Jeunghee Park,*,† and Jae-Pyoung Ahn*,‡ †
Department of Chemistry, Korea University, Jochiwon 339-700, Korea Advanced Analysis Center, Korea Institute of Science and Technology, Seoul 136-791, Korea
‡
S Supporting Information *
ABSTRACT: Scaling-down of phase change materials to a nanowire (NW) geometry is critical to a fast switching speed of nonvolatile memory devices. Herein, we report novel composition-phase-tuned GeSbTe NWs, synthesized by a chemical vapor transport method, which guarantees promising applications in the field of nanoscale electric devices. As the Sb content increased, they showed a distinctive rhombohedral−cubic−rhombohedral phase evolution. Remarkable superlattice structures were identified for the Ge8Sb2Te11, Ge3Sb2Te6, Ge3Sb8Te6, and Ge2Sb7Te4 NWs. The coexisting cubic−rhombohedral phase Ge3Sb2Te6 NWs exhibited an exclusively uniform superlattice structure consisting of 2.2 nm period slabs. The rhombohedral phase Ge3Sb8Te6 and Ge2Sb7Te4 NWs adopted an innovative structure; 3Sb2 layers intercalated the Ge3Sb2Te6 and Ge2Sb1Te4 domains, respectively, producing 3.4 and 2.7 nm period slabs. The current− voltage measurement of the individual NW revealed that the vacancy layers of Ge8Sb2Te11 and Ge3Sb2Te6 decreased the electrical conductivity. KEYWORDS: GeSbTe, nanowires, polymorphism, superlattices, phase change, cubic−rhombohedral transition
O
switching at a very low power dissipation, suggesting that NWs hold promise for electrical data storage devices.9−17 However, there remain considerable difficulties in synthesizing the PCM NWs with a controlled composition and crystal phase. Furthermore, to fully understand the basis of NW device operation, it is essential to scrutinize the crystal structures, which may be different from those of the bulk or thin films. Herein, we synthesized crystalline GST NWs with six different compositions by a chemical vapor transport (CVT) method. The tuning of the composition over a wide range, x = [Sb]/([Ge] + [Sb]) = 0−0.8, was successfully achieved to produce a series of homologous GST NWs, which have never been identified before. As x increases, the phase evolution from the R to C and R phase takes place. The R structure could be indexed by either the primitive or rhombohedrally centered unit cell. We denoted all R structure indices by rhombohedrally centered unit cell (Supporting Information, SI, Figure S5). To avoid the confusion, we will not the expression of the H phase hereafter unless cited the results of other research groups. This exclusive composition and phase control enables unique superlattice structures of Ge8Sb2Te11 (x = 0.2), Ge3Sb2Te6 (x = 0.4), Ge3Sb8Te6 (x = 0.7), and Ge2Sb7Te4 (x = 0.8) NWs to emerge. We referred to five x ≥ 0.1 samples as GST-1, GST-2, GST-4, GST-7, and GST-8, respectively (Scheme 1). In
ne of most important chalcogenides, a GeTe-Sb2Te3 (GST) pseudobinary compound, exhibits crystalline− amorphous (order−disorder) phase transitions with distinct resistance states that can be reversibly switched by varying the temperature or electric field.1−8 These “phase-change” materials (PCM) have been widely used in the field concerning nonvolatile optical storage applications, such as CDs and DVDs, and are now actively investigated as a medium for phase change random access memory (PRAM) devices that could replace dynamic and flash random access memories. The GST compounds usually have two kinds of crystalline phases: one is a metastable face centered cubic (C) phase, and the other is a stable rhombohedral (R) or hexagonal (H) phase. The reversible transformation between amorphous and C phases is used for memory storage. However, there exist changes in resistivity of the crystalline state over a wide temperature window due to comprising two different crystal phases. Therefore, there are still relentless studies to characterize the amorphous and crystal structures underlying the phase change. It is recognized that the scaling-down of PCM is critical to reducing the current required to make “reset” rapidly, with less power consumption, thereby enabling a fast memory switching speed with high reliability. The use of one-dimensional (1D) nanostructures, such as nanowires (NWs), can be a powerful bottom-up approach to improving the device performance, owing to their unique 1D geometry in which the current and heat localize intrinsically within their structure. A number of pioneering works have already shown nanosecond-level phase © 2013 American Chemical Society
Received: November 3, 2012 Revised: January 3, 2013 Published: January 16, 2013 543
dx.doi.org/10.1021/nl304056k | Nano Lett. 2013, 13, 543−549
Nano Letters
Letter
explicit [211]̅ C/[21̅ 1]̅ C zigzagged directions (with a zigzag angle of 110°). When this NW is rotated around the NW axis by 45° (zone axis = [001]C), the [110]C/[1̅10]C zigzagged directions (with a zigzag angle of 90°) are observed. Upon each 45° rotation, these two zigzagged directions appear alternately. Therefore, this NW has a helical shape consisting of four equivalent growth directions of 110 C with an axial direction of [001]C, which is the identical structure as that of the C phase ZnGa2O4 NW.27 At the [011]C and [1̅12]C (35° rotation from the [001]C) zone axes, the uniform 2.2 nm period superlattice domain appears along the [111]C direction perpendicular to the [110]C growth direction (Figure 2d and e). The SAED pattern at the [011]C zone axis shows clear N = 6 equal divisions of two adjacent basic spots along the 111 C direction. At the [001]C zone axis, there are no superlattice spots, confirming the exclusive superlattice structures along the [111]C direction (Figure 2f). All data show an identical superlattice structure as that of the straight NW. Kooi and De Hosson reported the TEM images and SAED patterns of Ge3Sb2Te6.5 The present superlattice structure of the Ge3Sb2Te6 NWs is coherent with their 11-layer slab structure model. Da Silva et al. calculated the structure of the same 11-layer slab for the C and H phases of Ge3Sb2Te6, which is consistent with our superlattice structures.26 The increase in the average d-spacing (d111 = 3.6 Å vs 3.5 Å of GST-2) is probably due to the formation of the R phase. The coexistence of the C and R phases indicates that the C-R transition temperature of the Ge3Sb2Te6 NW is the growth temperature (350 °C). The fact that the Ge3Sb2Te6 NW has the same helical structure as that of the C structure ZnGa2O4 NW suggests that their C phase plays an important role in determining the morphology. The inherent anisotropy of the surface energies associated with the surface reactivity and/or dipolar interactions was suggested to be the driving force for the growth of the nanocrystals. In the C phase, the following general sequence is observed in the surface energy of the crystallographic planes; {111} < {100} < {110}.28 The growth along the higher surfaceenergy direction may have a lower activation energy barrier. Therefore, the C phase Ge9Sb1Te10 (GST-1), Ge8Sb2Te11 (GST-2), and Ge3Sb2Te6 (GST-4) NWs grew favorably along the highest surface energy direction [110]C. The growth of the helical Ge3Sb2Te6 NW is driven by the four 110 C growth directions. The coexisting R phase would grow preferentially along the active [0001]R direction. The increased disorder of the substituted Sb allows for growth along the lower surface energy direction, [111]C, which needs a higher activation energy barrier. Since the (111)C planes of C phase and (0001)R planes of R phase are atomically identical, the growth of the (111)C planes is made favorable by matching them with the active (0001)R planes. We suggest that the helical morphology is produced as an intermediate form upon the evolution of the growth direction from [110]C to [111]C. The kinetically controlled [110]C growth and thermodynamically controlled [111]C growth compete to produce the periodic turning of the growth direction along their vector summed [100]C axial direction. In the GST-1, where the C and R phases of GeTe coexist, no helical structures formed. The preferred growth direction of the R phase is likely to be either [01̅11]R (GeTe) or [211̅ 0̅ ]R (GST-1 and GST-2), as seen in the SI, Figure S4, which unsatisfied the condition we suggested ([0001]R). Nevertheless, further studies are required to provide the growth mechanism of the morphology of the helical structures.
The HRTEM image of a typical GST-7 (x = 0.7) shows that the bright and dark domains stack periodically along the wire axis (Figure 3a). The intensity line profile integrated along the region (i) reveal that the period is uniform in the length of 3.4 nm. Its corresponding EDX line-scan data along the wire axis and mapping (with high angle annular dark field (HAADF) STEM image) show that the Ge (Te) and Sb elements exist alternately along the wire axis. The average composition is
Figure 3. (a) HRTEM image of GST-7 (x = 0.7; Ge3Sb8Te6) NW exhibiting a superlattice structure along the growth direction. The intensity line profile (i) show 3.4 nm-period slabs. The HAADF STEM images and EDX line profile/mapping of the Ge, Sb, and Te elements along the wire axis show the composition of Ge/Te and Sb alternative modulation with the same period of 3.4 nm. (b) Latticeresolved image and SAED pattern (zone axis = [21̅1̅0]R), confirms d0001 = 3.8 Å. The SAED pattern shows 17 spots (00017̅) subdividing the brighter ED spots corresponding to adjacent atomic layers (1.9 Å). (c) Magnified TEM image (viewed at the zone axis of [011̅0]R) and its corresponding intensity line profile show 11 Ge/Te layers (marked by a red line) including 6 Te layers (marked by a triangle symbol) in the brighter domain. In the darker parts, as marked by blue lines, there exist 6 Sb layers (in avg.). 546
dx.doi.org/10.1021/nl304056k | Nano Lett. 2013, 13, 543−549
Nano Letters
Letter
Ge:Sb:Te = 1:3:2. The concentration of the Ge and Te elements increases and decreases with an average period of 3.4 nm, and that of the Sb element alternatively changes with the same period. The Ge-rich part is brighter with an average composition of Ge:Sb:Te = 1:2:2 and the Sb-rich part is darker with 1:4:2. The average distance between Te−Te layers (= d0001) is estimated to be 3.8 Å (Figure 3b). The 17 periodic spots (00017̅) dividing the distance between two adjacent atomic layers (1.9 Å) are identified along the [0001]R growth direction, which suggests a periodic 17-layer superlattice domain. The intensity line profile for a magnified TEM image viewed at the [011̅0]R zone axis shows clearly 11 atomic layers in the brighter domain and 6 atomic layers in the darker domain (Figure 3c and Figure S6). There are no vacancy layers between domains, which is different from the superlattice structures of GST-2 and GST-4. The HRTEM image of a typical GST-8 shows that the bright-dark striped domains appear uniformly along the wire axis (Figure 4a). The intensity line profile and EDX line-scan data indicates that the Ge/Te and Sb elements are modulated alternatively with an average period of 2.7 nm. Two SAED spots from two atomic layers (1.9 Å) along [0001]R are divided into 13 parts (00013̅) (Figure 4b). A magnified TEM image viewed at the zone axis of [011̅0]R and its corresponding intensity line profiles reveal robustly that each superlattice domain consisted of 13 atomic layers with no vacancy layer (Figure 4c and Figure S6). There are a number of works showing the alternate slab structures with Sb2 layers and no vacancy layers. The Oeckler group reported on the layer structures of Ge1.57Sb10.43Te5 film, in which the slabs of Ge2Sb2Te5 (2GeTe·Sb2Te3) alternate with elementary 4Sb2 slabs.29 In the Sb−Te binary system, a series of Sb2nTe3 (where n is an integer) compounds, in which the nSb2 layers are intercalated between the Te−Te layers of Sb2Te3, was found.30−32 Recently, Tomforde et al. reported the structure of Ge 2 Sb 2 Te 4 that consists of two metastable Ge-rich (Ge2.1Sb1.2Te4) and Sb-rich (Ge1.1Sb2.6Te4) phases.33 The HRTEM images reveal their lamella nanostructures having dark and bright stripes without vacancy layers, which have nearly the same feature as that of the present GST-7 and GST8. We suggest the following possible atomic arrangement for the Sb2 layer intercalated structures. For the GST-7, 3Sb2 layers are intercalated between the Te layers of Ge 3 Sb 2 Te 6 (3GeTe·Sb2Te3). The 11 layers of Ge3Sb2Te6 and 6 layers of Sb element constitute one superlattice slab of the 17 layered Ge3Sb8Te6 (Ge3Sb2Te6·3Sb2), whose composition is fully coherent with the EDX data. For the GST-8, we propose unique 13 layered superlattice structure of Ge2Sb7Te4 in which 3Sb2 layers are intercalated between the Te layers of Ge2Sb1Te4. The 7 layers of Ge2Sb1Te4 and 6 layers of Sb element constitute one slab of Ge2Sb1Te4·3Sb2. The resulting composition accurately matches the EDX data. The phase diagram of the Au−Ge−Sb ternary system shows that the solubility of Sb in Au is higher than that of Ge, and there is an equilibrium between the Sb metal and the alloy liquid at temperatures above the eutectic point (430 °C).34 The EDX data for the Au nanoparticles at the tip of the NW reveals consistently that they have a Sb richer composition than that of the NW (SI, Figure S2). Therefore, when the Sb saturates in the Au nanoparticles, a precipitate of Sb layers is formed. The saturation of Sb would increase the diameter of Au nano-
Figure 4. (a) HRTEM image of GST-8 (x = 0.8; Ge2Sb7Te4) exhibiting a superlattice structure along the [0001]R growth direction. The intensity line profile (i) shows 2.7 nm period slabs. The EDX line profile of the Ge/Te and Sb elements along the wire axis shows an alternative modulation with a period of 2.7 nm. (b) Lattice-resolved image and SAED pattern at the zone axis of [21̅1̅0]R. The average distance between Te−Te layers (= d0001) is 3.8 Å, and two ED spots from adjacent atomic layers (1.9 Å) along [0001]R are divided into 13 parts (00013̅). (c) Magnified TEM image (at the zone axis of [011̅0]R) and its corresponding intensity line profile, showing that the brighter parts consisted of 7 Ge/Te layers (marked by a red line) including 4 Te layers (marked by a green-colored triangle symbol). In the darker parts, 6 Sb layers are marked by a blue-colored triangle symbol.
particles, which is responsible for the increased diameter with increasing the Sb content. The GST-7 and GST-8 are grown with only the R phase, indicating that its C-R phase transition temperature is below the growth temperature of 350 °C. This result is coherent with the prediction that increasing the Sb content decreases the C-R transition temperature. In situ temperature-dependent TEM measurements showed the stability of these Sb-rich NWs (SI, Figure S7). The four-point probe electrical transport measurements of GST NW were performed using a nanomanipulator (Figure 5a). The resistance of individual NW was obtained from the linear current−voltage (I−V) curves (Figure 5b and Table S1). The resistance of 3−5 NW electrodes with the well-defined cross section and length between Pt electrodes provides the reliable averaged value of resistivity (Table 1). The lower resistivity of the GST-1, GST-7, and GST-8 than that of the 547
dx.doi.org/10.1021/nl304056k | Nano Lett. 2013, 13, 543−549
Nano Letters
Letter
Figure 5. (a) SEM images showing the NW between Pt electrode pads with four-probe W tips. (b) I−V characteristics of the (1) GeTe, (2) GST-1, (3) GST-2, (4) GST-4, (5) GST-7, and (6) GST-8 with a typical cross section image of NW-Pt electrodes (inset).
■
GeTe NW can be rationalized by the idea that the incorporation of Sb would increase the concentration of charge carriers, thereby increasing the conductivity. Interestingly, the resistivity of GST-4 is higher than that of GST-1 by 2.6-fold. We suggest that the uniform vacancy layers running parallel to the NW direction would produce energy barriers and reduce electron mobility. Despite the higher Sb content, the same resistivity of GST-2 as that of GST-1 is also probably due to the vacancy layers. The Wang group reported the “Schottky type” I−V property of superlattice-structured ZnO nanohelix and suggested that the nanostripe boundaries lying almost in parallel to the nanobelt direction produce energy barriers.35 The lack of the vacancy layers and the metallic Sb2 layers would contribute to the increase of the conductivity of the GST-7 and GST-8. These electrical properties can provide the fundamental data to predict the promising PCM NW; the higher conductivity of crystalline phase (GST-7 and GST-8) would induce a larger resistance change of set-reset states. In summary, we synthesized a homologous series of GST NWs at 350 °C by CVT of GeTe/Sb2Te3. As the Sb content increased, the R, C, and R phases were produced consecutively, showing a unique phase evolution in which increasing the Sb content decreased the phase transition temperatures. Four distinctive superlattice structures were produced; (1) 3.8 nm period Ge8Sb2Te11, (2) 2.2 nm period Ge3Sb2Te6, (3) 3.4 nm period Ge3Sb8Te6, and (4) 2.7 nm period Ge2Sb7Te4. The C phase Ge8Sb2Te11 NWs consisted of a 3.8 nm period superlattice along the [111]C direction, perpendicular to the [110]C growth direction. The Ge3Sb2Te6 NWs exhibited two types of morphologies with coexisting C and R phases, namely, straight and helical NWs. They consisted of a 2.2 nm period superlattice along the [111]C direction. The R phase Ge3Sb8Te6 and Ge2Sb7Te4 NWs exhibited a unique superlattice structure that consists of 3.4 and 2.7 nm period GeTe-rich and Sb-rich slabs along the [0001]R growth direction, respectively. We suggest an atomic arrangement of Ge3Sb8Te6 in which 3Sb2 layers are intercalated between the Ge3Sb2Te6 slabs. The Ge2Sb7Te4 consisted of alternating Ge2Sb1Te4 and 3Sb2 slabs. We measured the I−V curve of the individual NW, using the FIB nanomanipulator. The results suggested that the vacancy layers of the superlattice slabs running parallel to the NW axis reduced electrical conductivity (Ge8Sb2Te11 and Ge3Sb2Te6).
ASSOCIATED CONTENT
S Supporting Information *
Materials and methods; SEM images, EDX spectra, XRD and SAED patterns, HRTEM images, and structure relations of the studied material. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected];
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This study was supported by NRF (2011-001-5235; 2011-0020090), WCU (R31-2012-000-10035-0), and KETEP (20104010100640). The HVEM (Daejeon) and XPS (Pusan) measurements were performed at the KBSI. The experiments at the PLS were partially supported by MOST and POSTECH.
■
REFERENCES
(1) Yamada, N.; Ohno, E.; Nishiuchi, K.; Akahira, N.; Takao, M. J. Appl. Phys. 1991, 69, 2849−2856. (2) Nonaka, T.; Ohbayashi, G.; Toriumi, Y.; Mori, Y.; Hashimoto, H. Thin Solid Films 2000, 370, 258−261. (3) Friedrich, I.; Weidenhof, V.; Njoroge, W.; Franz, P.; Wuttig, M. J. Appl. Phys. 2000, 87, 4130−4134. (4) Yamada, N.; Matsunaga, T. J. Appl. Phys. 2000, 88, 7020−7028. (5) Kooi, B. J.; De Hosson, J. T. M. J. Appl. Phys. 2002, 92, 3584− 3590. (6) Kolobov, A. V.; Fons, P.; Frenkel, A. I.; Ankudinov, A. L.; Tominaga, J.; Uruga, T. Nat. Mater. 2004, 3, 703−708. (7) Wuttig, M.; Lusebrink, D.; Wamwangi, D.; Welnic, W.; Gillessen, M.; Dronskowski, R. Nat. Mater. 2007, 6, 122−128. (8) Siegrist, T.; Jost, P.; Volker, H.; Woda, M.; Merkelbach, P.; Schlockermann, C.; Wuttig, M. Nat. Mater. 2011, 10, 202−208. (9) Yu, D.; Wu, J. Q.; Gu, Q.; Park, H. J. Am. Chem. Soc. 2006, 128, 8148−8149. (10) Lee, S. H.; Ko, D. K.; Jung, Y.; Agarwal, R. Appl. Phys. Lett. 2006, 89, 223116. (11) Yu, D.; Brittman, S.; Lee, J. S.; Falk, A. L.; Park, H. Nano Lett. 2008, 8, 3429−3433. (12) Meister, S.; Schoen, D. T.; Topinka, M. A.; Minor, A. M.; Cui, Y. Nano Lett. 2008, 8, 4562−4567.
548
dx.doi.org/10.1021/nl304056k | Nano Lett. 2013, 13, 543−549
Nano Letters
Letter
(13) Lee, S. H.; Jung, Y.; Agarwal, R. Nat. Nanotechnol. 2007, 2, 626− 630. (14) Ahn, J. K.; Park, K. W.; Jung, H. J.; Yoon, S. G. Nano Lett. 2010, 10, 472−477. (15) Han, N.; Kim, S. I.; Yang, J. D.; Lee, K.; Sohn, H.; So, H. M.; Ahn, C. W.; Yoo, K. H. Adv. Mater. 2011, 23, 1871−1875. (16) Karthik, C.; Mehta, R. J.; Jiang, W.; Castillo, E.; Borca-Tasciuc, T.; Ramanath, G. Appl. Phys. Lett. 2011, 99, 103101. (17) Longo, M.; Fallica, R.; Wiemer, C.; Salicio, O.; Fanciulli, M.; Rotunno, E.; Lazzarini, L. Nano Lett. 2012, 12, 1509−1515. (18) Goldak, J.; Barrett, C. S.; Innes, D.; Youdelis, W. J. Chem. Phys. 1966, 44, 3323−3325. (19) Rabe, K. M.; Joannopoulos, J. D. Phys. Rev. Lett. 1987, 59, 570− 573. (20) Meister, S.; Peng, H. L.; McIlwrath, K.; Jarausch, K.; Zhang, X. F.; Cui, Y. Nano Lett. 2006, 6, 1514−1517. (21) Sun, X.; Yu, B.; Ng, G.; Meyyappan, M. J. Phys. Chem. C 2007, 111, 2421−2425. (22) Chung, H. S.; Jung, Y.; Kim, S. C.; Kim, D. H.; Oh, K. H.; Agarwal, R. Nano Lett. 2009, 9, 2395−2401. (23) Matsunaga, T.; Kojima, R.; Yamada, N.; Kifune, K.; Kubota, Y.; Tabata, Y.; Takata, M. Inorg. Chem. 2006, 45, 2235−2241. (24) Karpinsky, O. G.; Shelimova, L. E.; Kretova, M. A.; Fleurial, J. P. J. Alloys Compd. 1998, 268, 112−117. (25) Kuypers, S.; Vantendeloo, G.; Vanlanduyt, J.; Amelinckx, S. J. Solid State Chem. 1988, 76, 102−108. (26) Da Silva, J. L. F.; Walsh, A.; Lee, H. Phys. Rev. B 2008, 78, 224111. (27) Kim, H. S.; Hwang, S. O.; Myung, Y.; Park, J.; Bae, S. Y.; Ahn, J. P. Nano Lett. 2008, 8, 551−557. (28) Wang, Z. L. J. Phys. Chem. B 2000, 104, 1153−1175. (29) Schneider, M. N.; Seibald, M.; Oeckler, O. Dalton Trans. 2009, 2004−2011. (30) Nakata, Y.; Suenaga, T.; Ejiri, M.; Shida, S.; Tani, K.; Yiwata, N.; Tashiro, H. Mater. Trans. 2004, 45, 2673−2677. (31) Kifune, K.; Kubota, Y.; Matsunaga, T.; Yamada, N. Acta Cryst. B 2005, 61, 492−497. (32) Sun, C. W.; Lee, J. Y.; Youm, M.; Kim, Y. T. Phys. Status Solidi (RRL) 2007, 1, R25−R27. (33) Tomforde, J.; Bensch, W.; Kienle, L.; Duppel, V.; Merkelbach, P.; Wuttig, M. Chem. Mater. 2011, 23, 3871−3878. (34) Wang, J.; Leinenbach, C.; Roth, M. J. Alloys Compd. 2009, 485, 577−582. (35) Gao, P. X.; Ding, Y.; Wang, Z. L. Nano Lett. 2009, 9, 137−143.
549
dx.doi.org/10.1021/nl304056k | Nano Lett. 2013, 13, 543−549
