Phase-Change Ge−Sb Nanowires: Synthesis, Memory Switching, and

Department of Materials Science and Engineering, University of Pennsylvania, Philadelphia, Pennsylvania 19104. Nano Lett. , 2009, 9 (5), pp 2103–210...
16 downloads 16 Views 3MB Size
NANO LETTERS

Phase-Change Ge-Sb Nanowires: Synthesis, Memory Switching, and Phase-Instability

2009 Vol. 9, No. 5 2103-2108

Yeonwoong Jung, Chung-Ying Yang, Se-Ho Lee, and Ritesh Agarwal* Department of Materials Science and Engineering, UniVersity of PennsylVania, Philadelphia, PennsylVania 19104 Received February 25, 2009; Revised Manuscript Received March 25, 2009

ABSTRACT We report the synthesis and characterization of phase-change Ge-Sb nanowires with two different eutectic compositions and their memory switching characteristics. Under application of electric-fields with controlled pulse amplitude and duration times, Sb-rich (Sb g 86 at. %) eutectic Ge-Sb nanowires show phase-change based memory switching, while another eutectic GeSb (Ge:Sb ) 1:1) nanowires do not show electronic memory switching at all. However, under repeated measurements, Sb-rich Ge-Sb nanowires display an increase of resistance of the low resistive state. The observed electrical irreversibility for Sb-rich Ge-Sb nanowires is attributed to the structural and compositional instability due to the phase-separation of Ge out of homogeneous Ge-Sb as observed from rapid thermal annealing and transmission electron microscopy experiments. Implications for design of Te-free nanoscale materials for phase change memory applications are also discussed.

Among a variety of materials developed to date for electronic phase-change memories (PCMs), Te-based ternary systems, such as Ge-Sb-Te alloys, have dominated the field. In PCMs, reversible structural change between crystalline and amorphous phases represents different electronic and optical states corresponding to changes in electrical resistivity and reflectivity, used for storing information.1 Recently, Te-free Ge-Sb binary alloys emerged as a new class of phase-change materials.2-8 In Ge-Sb systems, there are two eutectic compositions of Ge:Sb ) 1:1 and 0.14:0.86.2 In particular, Sb-rich Ge-Sb alloys at/or above its eutectic limit (Sb g 86 at %) display many attributes of good phase-change materials for memory applications and offer several advantages over Ge-Sb-Te alloys. (1) The material is free of Te, which is an environmentally hazardous element and often chemically reacts with electrode metals in devices.3 (2) It has a relatively high crystallization temperature of ∼250 °C leading to extremely long data retention times.4 (3) Recrystallization mechanism is growth-dominated phase-change5 asopposedtothenucleation-dominatedbehaviorinGe-Sb-Te, and the material possesses very fast crystallization speeds,6-8 making it suitable for ultrafast data storages.4-8 Despite such positive attributes, Ge-Sb alloys have been relatively unexplored in comparison to Ge-Sb-Te and the underlying phase-change mechanism remains poorly understood. In particular, the electronic properties of Sb-rich eutectic Ge-Sb and their feasibility for electronic memories have not been * To whom correspondence should be addressed. E-mail: riteshag@ seas.upenn.edu. 10.1021/nl900620n CCC: $40.75 Published on Web 04/24/2009

 2009 American Chemical Society

extensively investigated. Moreover, very few attempts have been made to synthesize Ge-Sb nanostructures,9,10 which are expected to display superior properties due to their confined geometry at nanoscale.11 In fact, recent studies of GeTe and Ge2Sb2Te5 nanowires from our group have demonstrated advantages of using nanowires over thin-film devices, such as lower writing currents and power consumption due to current and heat localization,12-14 size-dependent materials properties,15 and enhanced functionalities by structural/chemical modulation.16 In this paper, we present a comprehensive study of Ge-Sb nanowire systems at different eutectic compositions to understand their phase change behavior at the nanoscale and to investigate their feasibility for phase-change based electronic memory devices. We synthesized Ge-Sb nanowires with two different eutectic compositions, that is, Sb-rich Ge-Sb (Sb g 86 at. %) and GeSb (1:1), and characterized the structure and the composition of the nanowires by using X-ray diffraction (XRD), scanning/transmission electron microscopies (SEM/TEM), and energy-dispersive spectroscopy (EDS). We investigated the memory switching characteristics of two-terminal nanowires devices under sequential application of AC voltage pulses and DC current-voltage (I-V) sweep and correlated their electrical properties with structural change using electron microscopy techniques. Ge-Sb nanowires were synthesized based on the vaporliquid-solid mechanism17 by thermal evaporation of Ge and Sb powders. Au colloid (∼100 nm)-deposited SiO2 substrates were placed at the downstream side of a tube furnace. Ge

Figure 1. Sb-rich Ge-Sb (Sb g 86 at %) nanowires. (a) SEM image of as-grown nanowires on SiO2 substrate. (b) XRD of Sbrich Ge-Sb nanowires corresponding to hexagonal Sb structure. (c) STEM elemental mapping image to show uniform distribution of Ge and Sb with no phase-separation and Au at the nanowire tip. (d) EDS spectrum from the body of the same nanowire in panel c. (e) HRTEM of a Sb-rich Ge-Sb nanowire showing hexagonal (012) lattice fringes. Inset: FFT of the HRTEM confirms [012] growth direction. (f) (left) TEM of a Sb-rich Ge-Sb nanowire to show stacking faults along the nanowire length; (right) HRTEM of the boxed region in the left to show that stacking faults proceeds in [012] direction.

and Sb powders were located inside the furnace with Ge in the higher temperature region than Sb. The furnace temperature was ramped to 880 °C to simultaneously vaporize Ge and Sb with a carrier Ar gas (200 SCCM) at a pressure of 100 Torr. The temperature of the growth substrate was ∼400 °C and the reaction was maintained for 1 h. The two different eutectic compositions of Sb-rich Ge-Sb (Sb g 86 at %) and GeSb (1:1) were determined from the phase-diagram of Ge-Sb2 and were reliably achieved by controlling the location and the amount of Ge and Sb powders. SEM image (Figure 1a) shows that as-grown nanowires are typically 30-300 nm thick and >30 µm long when Ge and Sb powders prepared in 0.14:0.86 molar ratio were thermally evaporated at 850 and 590 °C, respectively. The XRD pattern of the as-grown nanowires (Figure 1b) matches with that of pure Sb in hexagonal structure [JCPDS 85-1324, space group: R3m j (no. 166)], consistent with previous studies to report the solid-solution of Ge-Sb.18-20 With small concentration of Ge, it is known that Ge atoms occupy interstitial sites of Sb lattice and do not distort the Sb structure, therefore the crystalline phase is a solid solution of Ge in Sb with the alloy structure remaining identical to pure Sb.20 The chemical composition and the structure of 2104

the nanowires were characterized with EDS in scanning TEM (STEM) and HRTEM. STEM elemental mapping images (Figure 1c) show the uniform distribution of Ge and Sb throughout the nanowire with Au catalyst at the nanowire tip. EDS point scanning (Figure 1d) at arbitrary positions on the nanowire reveals that Ge and Sb are present in an atomic ratio of ∼0.12:0.88. EDS performed on a large number of nanowires confirmed that Sb is typically in a range of at % 0.86-0.90. HRTEM image (Figure 1e) of a Sb-rich Ge-Sb nanowire shows lattice fringes with spacing of 0.311 nm corresponding to (012) crystal plane of hexagonal Sb (a ) 4.3 Å, c ) 11.22 Å) consistent with the XRD analysis. The (012) orientation is also consistent with the previously reported growth direction of Sb nanobelts,21 as seen in the fast Fourier transform (FFT) of the HRTEM image (Figure 1e, inset). Interestingly, in some nanowires stacking faults are found along the nanowire growth direction. Figure 1f (left) is a TEM image of the side of a Sb-rich Ge-Sb nanowire showing that stacking faults proceed along the nanowire growth direction, while the HRTEM (right) clearly reveals that alternating twin domains exist along [012] direction. These stacking faults were typically observed in thick nanowires (over ∼150 nm in diameter) that are found in the hotter region of the furnace where the radial diameter of the nanowires increases due to the agglomeration of molten Au. Such structural variations are known to be due to fast nanowire growth kinetics at high temperature, representing the competition between the diffusion rate inside the catalyst droplet and the growth rate on the liquid-solid interface.22,23 EDS study still confirms that Ge and Sb are uniformly distributed in as-grown nanowires in spite of the presence of stacking faults (data not shown). Ge-rich GeSb (1:1) nanowires were synthesized by evaporating Ge and Sb powders prepared in 1:1 molar ratio at 880 and 590 °C, respectively (Figure 2a, SEM image). High-resolution SEM reveals that the nanowires display wellfaceted, rectangular cross sections, suggesting the singlecrystallinity of the nanowires. XRD of the nanowires (Figure 2b) matches well with the tetragonal GeSb pattern with Ge: Sb in 1:1 atomic ratio [ICDS 51652, space group: I4/mmm (no. 139)],2 and clearly differs from the XRD pattern of Sbrich Ge-Sb nanowires (Figure 1b). The tetragonal structure of GeSb (1:1) nanowires is also consistent with the SEM observation of rectangular cross sections. The chemical composition of individual nanowires was found to be Ge: Sb ) ∼1:1 by EDS (Figure 2c), and STEM images show uniform spatial distribution of Ge and Sb in a large number of nanowires (Figure 2d). The tetragonal GeSb structure was further studied by HRTEM. HRTEM images reveal two lattice fringes (0.310 and 0.294 nm) intersecting at 90°, corresponding to (002) and (110) tetragonal planes respectively (Figure 2e, left inset). The indexed FFT (Figure 2e, right inset) obtained from the HRTEM image shows that the single-crystalline GeSb (1:1) nanowires grow along [001] direction. This TEM analysis combined with the XRD analysis (Figure 2b) indicates that the frequently observed rectangular cross sections of GeSb (1:1) nanowires are attributed to the (001) tetragonal basal plane, and the Nano Lett., Vol. 9, No. 5, 2009

Figure 2. Ge-rich GeSb (1:1) nanowires. (a) SEM image of asgrown nanowires; (right) GeSb (1:1) nanowires with rectangular cross sections. (b) XRD of as-grown GeSb (1:1) nanowires. (c) EDS of GeSb (1:1) nanowires. (d) STEM mapping images of GeSb (1:1) nanowires. Scale bar: 300 nm. (e) TEM of a GeSb (1:1) nanowire. HRTEM (left inset) shows tetragonal (002) and (110) lattice fringes. FFT (right inset) confirms [001] growth direction.

nanowire side facet corresponds to the {110} crystal family. The electrical memory switching properties of Ge-Sb nanowires were investigated by fabricating two-terminal devices with Pt contacts (2 µm separation) deposited by focused-ion-beam (FIB). DC I-V sweep of an as-synthesized Sb-rich Ge-Sb nanowire (∼80 nm thick) displays linear I-V behavior with an initial resistance of ∼36 kΩ (Figure 3a, red curve). Upon application of 200 ns voltage pulses with increasing amplitude, the resistance is observed to suddenly increase at ∼6.2 V and saturates to ∼3.7 MΩ (Figure 3b). Nano Lett., Vol. 9, No. 5, 2009

The ratio of the resistance change reaches ∼100, which suggests crystalline-amorphous phase-change based memory switching.1,5,12-14 DC I-V sweep (Figure 3a, blue curve) of the amorphized nanowire device shows the initial high resistance is maintained until ∼4.0 V (threshold switching, Vth), after which the current increases and linear I-V behavior reappears. Subsequent DC sweep (Figure 3a, black curve) displays linear behavior without threshold switching, indicating the completion of amorphous-to-crystalline transition. However, the resistance is found to increase to ∼83 kΩ and does not fully recover its initial low value (∼36 kΩ) of the as-synthesized state prior to amorphization. This observation suggests that the structure of the nanowires after voltage pulses/DC sweeps did not fully switch back to the initial structure, which is in contrast to the completely reversible switching observed in GeTe and Ge2Sb2Te5 nanowires.12-14 Such lack of complete electrical reversibility was also observed for other nanowires when phase-change was introduced by voltage pulses. Figure 3c shows the resistance (R) variation of another Sb-rich Ge-Sb nanowire (∼100 nm thick) under voltage pulses to realize crystalline-to-amorphous (RESET) and amorphous-to-crystalline (SET) transitions. The initially crystalline nanowire (∼31 kΩ) reaches a RESET state (∼3.5 MΩ) with 200 ns voltage pulses above ∼7.3 V (Figure 3c, red triangles) with a resistance ratio of over ∼100. Subsequently, SET voltage pulses of longer duration (1 µs) were applied to the initially amorphized nanowire, and a low resistive crystalline state (∼92 kΩ) was achieved above ∼5.1 V (Figure 3c, blue circles). It should be noted that the resistance of the SET state (∼92 kΩ) is about three times higher than the initial resistance (∼31 kΩ), resulting in the reduced resistance ratio of ∼40 between RESET/SET states. The irreversible change of the resistance is further studied with another Sb-rich Ge-Sb nanowire (∼120 nm thick) under cyclic application of RESET/SET voltage pulses. Under alternating RESET (8 V, 200 ns) and SET (6 V, 1 µs) voltages, the nanowire displays a cyclic resistance change between high/low resistive states (Figure 3d). However, the SET resistance (Figure 3d, red squares) significantly increases (from ∼58 to ∼349 kΩ) as the device is repeatedly switched over 200 cycles, while the RESET resistance (Figure 3d, blue circles) remains relatively constant at ∼3.4 MΩ. We investigated a large number of nanowires and consistently observed similar behavior; SET resistance increases and often saturates under repeated voltage pulses/ DC-sweep leading to reduced RESET/SET resistance ratios, and the initial low resistance is never recovered. These data suggest that phase-change in Sb-rich Ge-Sb nanowires is not completely reversible in contrast to GeTe and Ge2Sb2Te5 nanowires.12-14 We also measured electrical switching properties of Ge-rich GeSb (1:1) nanowires by applying voltage pulses and failed to observe any memory switching behavior in all our devices; a majority of the as-synthesized nanowires do not show a noticeable change in the resistance under application of voltage pulses in our measurement range (up to 10 V), while some of thin nanowires (typically below ∼80 nm) show a drastic resistance increase to over ∼1 GΩ and saturate to this value, indicating electrical/structural 2105

Figure 3. Memory switching characteristics of Ge-Sb nanowire devices. (a) DC I-V sweep of a Sb-rich Ge-Sb nanowire (∼80 nm) device obtained from (1) as-synthesized state showing initial resistance of ∼36 kΩ, (2) after applying the amorphization voltage pulse (6.9 V, 200 ns) showing clear threshold switching, and (3) after completing the DC sweep in (2) showing increased resistance to ∼83 kΩ. (b) Crystalline-to-amorphous phase-change of the same Sb-rich Ge-Sb nanowire in panel a. (c) RESET/SET transition of a Sb-rich Ge-Sb nanowire (∼100 nm) as a function of voltage pulses. (d) RESET/SET resistance variation of a Sb-rich Ge-Sb nanowire (∼120 nm) under cyclic voltage pulses (RESET, 8 V, 200 ns; SET, 6 V, 1 µs). (e) GeSb (1:1) nanowire (∼70 nm) shows electrical breakdown at ∼8 V under application of voltage pulses with increasing amplitude. (f) Constant high resistance of the same GeSb (1:1) nanowires under cyclic application of voltage pulses (red square, 9 V, 200 ns; blue triangle, 6 V, 1 µs) confirms permanent damage to the device.

breakdown of the device (Figure 3e). This high resistance remains unchanged even under a cyclic application of voltage pulses with different amplitudes/durations as well as under DC I-V sweep, which confirms the failure of the device (Figure 3f). The absence of memory switching in GeSb (1: 1) nanowires is possibly due to the higher melting temperature (∼670 °C) of 1:1 GeSb in comparison to Sb-rich Ge-Sb (∼590 °C),2 which will require very high voltage to amorphize the nanowires. Another possible reason to explain the absence of the memory switching could be the unique electronic properties of GeSb (1:1). Recently, Raoux et al. have reported that the resistivity of crystalline and amorphous phases of Ge-Sb thin-films with Ge 59.3% are very similar, therefore very little resistance change (1 month). Figure 4d is a representative lowmagnification TEM of GeSb (1:1) nanowires 12 months after the synthesis (stored at room temperature), displaying a similar roughened surface as the annealed ones, which is in Nano Lett., Vol. 9, No. 5, 2009

Figure 4. TEM characterization showing the phase-separation in Ge-Sb nanowires. (a) TEM image of a Sb-rich Ge-Sb nanowire after annealing at 450 °C for 20 min with the corresponding EDS line-scan profiles superimposed. Bimodal distribution of Ge is observed indicating phase-separation of Ge from Ge-Sb toward the surface. (b) HRTEM of the same nanowire showing the phase separated region. (c) STEM elemental mapping image of a Sb-rich Ge-Sb nanowire after annealing at 550 °C revealing more pronounced Ge phase-separation. (d) TEM image of a GeSb (1:1) nanowire after being stored for 12 months in air at room temperature. (e) HRTEM image of the same nanowire shows polycrystalline Ge grains segregated toward the surface of the GeSb (1:1) nanowire.

clear contrast to the clean, single-crystalline surface of the freshly synthesized nanowires (Figure 2a,e). Representative HRTEM image (Figure 4e) clearly shows polycrystalline Ge grains segregated out of the Sb-rich host, suggesting the spontaneous phase-separation of Ge and Sb. These data indicate the high tendency of phase-separation in Ge-Sb systems, which become more pronounced with higher concentration of Ge and also at higher temperatures. In fact, recent studies of Ge-Sb thin-films (Ge:Sb ) 0.15: 0.85) also report the precipitation of Ge upon high temperature (∼420 °C) annealing, which typically occurs at the interface of thin film/metal contacts.25 We believe that our Ge-Sb nanowires also undergo phase-separation during electrical measurements, particularly in RESET transition which requires Joule-heating above their melting temperatures (∼590 °C).2 The effect of phase-separation on the resistance can be qualitatively understood based on the structural and compositional fluctuation. The resistance of the nanowires is determined by the effective cross-sectional area for carrier transport and the intrinsic resistivity which varies with the ratio of Ge/Sb. During the phase-separation, the nanowire core becomes Sb-rich, which is more electrically conductive25 than the surface which becomes Ge-rich. Meanwhile, the effective cross-sectional area of the Sb-rich Nano Lett., Vol. 9, No. 5, 2009

core becomes smaller with Ge localizing toward the surface, which contributes to increase the resistance of the crystalline phase. In addition, electron scattering from the polycrystalline grain boundaries and defects can also contribute to the resistance increase. On the basis of the above analysis, the reason that GeSb (1:1) nanowires failed to show that the memory switching can also be attributed to their higher degree of phase-separation along with higher density of defects, which will lead to carrier transport through twoseparated Sb-rich and Ge-rich regimes, none of which are ideal for phase-change memory applications. Also, as relative Sb concentration increases, the Sb-rich core will have lower crystallization/melting temperatures than the homogeneous, eutectic Ge-Sb, since the transition temperatures of Sb is lower than Ge (Tm ) ∼630 °C for Sb while Tm ) ∼920 °C for Ge).2 It is also known that Ge in contact with metals generally crystallizes at much lower temperature than its intrinsic crystallization temperature (60-70% of metal-Ge eutectic temperatures), and the crystallization starts from the metal-Ge interface.25,26 This indicates that Ge-Sb phase separation in our nanowire devices may not be isotropic and may initiate from the nanowire/Pt contacts. These factors make the precise interpretation of the resistance increase in Ge-Sb nanowire devices difficult. Nevertheless, we believe the phase-separation at nanoscale causes the irreversibility of memory switching, as observed in our electrical and annealing experiments. We also compare the device performance of Sb-rich Ge-Sb nanowires with our previously reported GeTe and Ge2Sb2Te5 nanowires. Sb-rich Ge-Sb nanowires display similar RESET currents and power consumption in comparison to Ge2Sb2Te5 nanowires of comparable thickness but are lower than those of GeTe nanowire. For example, the writing current measured for ∼100 nm Sb-rich Ge-Sb nanowire is found to be 0.24 mA, corresponding to a power of 1.6 mW, which is comparable to ∼1.5 mW reported for Ge2Sb2Te5 nanowire, but lower than GeTe nanowire (∼2.0 mW).27 The main reason for the observed lower writing currents could be the lower melting temperature of Sb-rich Ge-Sb (∼590 °C in bulk) in comparison to GeTe and Ge2Sb2Te5 (Tm ) 725 °C for GeTe and 616 °C for Ge2Sb2Te5).27 Other materials’ properties that affect the writing currents/power consumption, such as thermal conductivity of Sb-rich Ge-Sb, are not well known in the literature. The initial resistance ratio (∼102) of Sb-rich Ge-Sb nanowire is found to be similar to that of Ge2Sb2Te5 nanowire, while GeTe nanowires typically show larger resistance ratio (∼5 × 102). Other device parameters such as minimum device switching speed and maximum lifetime under cyclic test were not investigated in this study, mainly because of the complication in the measurements owing to the electrical irreversibility. This study presents a relatively unexplored but an important limitation of utilizing nanoscale Ge-Sb systems for PCMs despite their advantages in improved material properties. Even though the precipitation of Ge appears as an apparent problem, pure Sb without introduction of foreign elements (such as Ge) is not a good choice for phase-change 2107

memory applications. The underlying role of Ge in the Ge-Sb system is to enhance the structural stability, which thereby realizes reliable memory switching. Pure amorphous Sb has extremely fast crystallization speed; however, it easily relaxes back to the crystalline phase due to its low crystallization temperature (close to room temperature)28 causing data volatility. An introduction of a small amount of a semiconductor with high melting temperature such as Ge improves the stability of amorphous phase as it increase the phase transition temperatures while preserving their fast crystallization speeds.20 However, at this stage, the physical origin of Ge-Sb phase separation remains unknown, and possible means to circumvent this problem needs further investigation. The present study also establishes the need for investigation of thermal stability of other recently developed Sb-based phase-change materials such as doped Ge-Sb5 or Si-Sb29 at the nanoscale since the presence of a large surface can lead to rapid phase separation. In summary, phase-change Ge-Sb nanowires with two different eutectic compositions of Ge-rich GeSb (1:1) and Sbrich Ge-Sb (Sb g 86 at %) were synthesized. No electricfield induced phase-change was observed in Ge-rich GeSb nanowires. Phase-change induced memory switching was realized in Sb-rich Ge-Sb nanowires however with partial irreversibility. TEM studies of Sb-rich Ge-Sb nanowires show phase-separation of Ge under thermal annealing, which is responsible for the observed electrical irreversibility. Acknowledgment. This work was supported by Grants from NSF (DMR-0706381), Penn-MRSEC seed award (DMR05-20020), and in part by ONR (Grant N000140910116). References (1) Hudgens, S.; Johnson, B. MRS Bull. 2004, 29, 829. (2) Giessen, B. C.; Borromee, C. J. Solid State Chem. 1972, 4, 447. (3) Krusin-Elbaum, L.; Cabral, C., Jr.; K. N. Chen, K. N.; Copel, M.; Abraham, D. W.; Reuter, K. B.; Rossnagel, S. M.; Bruley, J.; Deline, V. R. Appl. Phys. Lett. 2007, 90, 1141902.

2108

(4) Raoux, S.; Salinga, M.; Jordan-Sweet, J. L.; Kellock, A. J. Appl. Phys. 2007, 101, 044909. (5) Chen, Y. C.; et al. Tech. Dig. - Int. Electron DeVices Meet. 2006, 30, 3. (6) Van Pieterson, L.; van Schijndel, M.; Rijpers, J. C. N.; Kaiser, M. Appl. Phys. Let. 2003, 83, 1373. (7) Afonso, C. N.; Solis, J.; Catalina, F.; Kalpouzos, C. Appl. Phys. Lett. 1992, 60, 3123. (8) Solis, J.; Afonso, C. N.; Trull, J. F.; Morilla, M. C. J. Appl. Phys. 1994, 75, 7788. (9) Sun, X.; Yu, B.; Ng, G.; Meyyappan, M.; Ju, S.; Janes, D. B. IEEE Trans. Electron DeVices 2008, 55, 3131. (10) Zhang, Y.; et al. Appl. Phys. Lett. 2007, 91, 013104. (11) Lieber, C. M.; Wang, Z. L. MRS Bulletin 2007, 32, 99. (12) Lee, S. H.; Ko, D. K.; Jung, Y.; Agarwal, R. Appl. Phys. Lett. 2006, 89, 223116. (13) Jung, Y.; Lee, S. H.; Ko, D. K.; Agarwal, R. J. Am. Chem. Soc. 2006, 128, 14026. (14) Lee, S. H.; Jung, Y.; Agarwal, R. Nat. Nanotechnol. 2007, 2, 626. (15) Lee, S. H.; Jung, Y.; Agarwal, R. Nano Lett. 2008, 8, 3303. (16) Jung, Y.; Lee, S. H.; Jennings, A. T.; Agarwal, R. Nano Lett. 2008, 8, 2056. (17) Wagner, R. S.; Ellis, W. C. Appl. Phys. Lett. 1964, 84, 89. (18) Yoo, Y.-G.; Yang, D.-S.; Ryu, H.-J.; Cheong, W.-S.; Baek, M.-C. Mater. Sci. Eng., A 2007, 449, 627. (19) Afonso, C. N.; Morilla, M. C.; Solis, J.; Rizvi, N. H.; Ollacarizqueta, M. A.; Catalina, F. Mater. Sci. Eng., A 1993, 173, 343. (20) Callan, J. P. Ultrafast Dynamics and Phase Changes in Solids Excited by Femtosecond Laser Pulses. Ph.D Thesis, Harvard University, Cambridge, MA, 2002. (21) Zhang, M.; Wang, Z.; Xi, G.; Ma, D.; Zhang, R.; Qian, Y. J. Cryst. Growth 2004, 268, 215. (22) Lee, J. S.; Brittman, S.; Yu, D.; Park, H. J. Am. Chem. Soc. 2008, 130, 6252. (23) Hao, Y.; Meng, G.; Wang, Z. L.; Ye, C.; Zhang, L. Nano Lett. 2006, 6, 1650. (24) Raoux, S. et al. J. Appl. Phys. 2009, 105, 064918. (25) Cabral, C., Jr.; Krusin-Elbaum, L.; Bruley, J.; Raoux, S.; Deline, V.; Madan, A.; Pinto, T. Appl. Phys. Lett. 2008, 93, 071906. (26) Tan, Z.; Heald, S. M.; Rapposch, M.; Bouldin, C. E.; Woicik, J. C. Phys. ReV. B. 1992, 46, 9505. (27) Lee, S. H.; Jung, Y.; Chung, H.-S.; Jennings, A. T.; Agarwal, R. Physica E 2008, 40, 2474. (28) Hashimoto, M.; Niizeki, T.; Kambe, K. J. J. Appl. Phys. 1980, 19, 21. (29) Zhang, T.; Song, Z.; Wang, F.; Liu, B.; Feng, S.; Chen, B. Appl. Phys. Lett. 2007, 91, 222102.

NL900620N

Nano Lett., Vol. 9, No. 5, 2009