The crystallization temperature of the phase change data storage material Ge2Sb2Te5 (GST) is known to vary depending on the thickness of the GST film and the presence of an interfacial film “cap”. This effect is most obvious for GST films that are thinner than 10 nm. With the ever increasing demand for high density electrical data storage devices and the concomitant reduction in the volume of phase change material, it is important to fully understand the affect of interfacial films on the phase change switching process.
The exact mechanism that causes the increased crystallization temperature is still a matter of debate. One explanation is based on chemical changes in the GST through oxidization or inward diffusion of chemical species from the interfacial caps, however it has also been argued that the crystallization temperature is influenced by the mechanical stress that is applied from the interfacial cap to the GST film. In this work, the affect of stress on the crystallization process in GST was investigated using density functional theory molecular dynamics (DFT-MD). In particular the affect of stress on the formation of GST crystal nucleates (formation of Sb-Te-Sb-Te and Ge-Te-Ge-Te rings), crystal growth and the resultant dielectric properties was studied.
There is a clear trend that shows the rate of crystal growth decreasing with increasing compressive stress. This has important implications for the scaling of phase change memory cells. On one hand the crystal growth rate decreases (due to compressive stress) when the GST is scaled to a smaller volume but on the other hand the reduced volume means the crystal need only grow over a smaller region of space.

Phase change memory (PCM) is attracting a great interest as future technology for reliable, scalable and non-volatile memory [1]. To extend the PCM functionality to universal and/or storage-class memories [2], energy reduction to the sub-pJ range and acceleration of the switching time in the few-ns regime should be achieved [3-5]. To this aim, the crystallization kinetic at high temperature (>300°C) in the pulsed-regime (<1 mu;s) must be thoroughly assessed. Previous results obtained by fast calorimetry in bulk Ge2Sb2Te5 (GST) suggested a non-Arrhenius crystallization kinetic above 300°C [6], which was attributed to the breakdown of the Stokes-Einstein equation in the supercooled liquid [7]. Here, we directly demonstrate non-Arrhenius crystallization in the pulsed-regime of PCM devices, indicating two different crystallization regimes in the temperature range below/above about 300°C.
We first studied the crystallization kinetic in the thermal regime, where the amorphous GST was annealed at constant temperature below 300°C. Data display Arrhenius behaviour with an activation energy Ex = 2.5 eV, in agreement with previous results for GST [8-10]. Then we concentrated to the pulsed regime, where crystallization is induced by Joule heating during the application of voltage pulses in the 50-1000 ns range. Time-dependent crystallization was studied at different current and ambient temperature (200 and 300 K). The local temperature in the amorphous GST was extracted through an electrothermal model for Joule heating, which was calibrated for a PCM in the amorphous phase, in the crystalline phase and in the ON state, namely the transient regime soon after threshold switching. Results display a very low Ex of 0.35 eV, similar to previous results for high-T viscosity in GeTe [6]. The results at 200 and 300 K overlap with each other, thus supporting the accuracy of our experimental/modelling procedure.
The different Ex below/above 300°C for the same device provides strong evidence for non-Arrhenius crystallization in GST. The high Ex in the thermal regime allows good reliability, while the lower Ex in the pulsed regime allows good glass formation and controllable crystallization, which might be useful for, e.g., multilevel cell applications [11] or Boolean/neuromorphic computation [12].
[1] H.-S. P. Wong, et al., Proc. IEEE 98 2201 (2010).
[2] G. W. Burr, et al., IBM J. Res. Dev., 52(4/5), 449 (2008).
[3] F. Xiong, et al., Science 332, 568 (2011).
[4] G. Bruns, et al., Appl. Phys. Lett. 95, 043108 (2009).
[5] D. Loke, et al., Science 336, 1566 (2012).
[6] J. Orava, et al., Nature Mater. 11, 279 (2012).
[7] G. C. Sosso, et al., Phys. Status Solidi B 249, 1880 (2012).
[8] G. Ruitenberg, et al., J. Appl. Phys. 92, 3116 (2002).
[9] J. Kalb, et al., Appl. Phys. Lett. 84, 5240 (2004).
[10] U. Russo, et al., IEEE T. Electron Dev. 54, 2769 (2007).
[11] A. Sebastian, et al., J. Appl. Phys. 110, 084505 (2011).
[12] D. C. Wright, et al., Adv. Mater. 23, 3408 (2011).

Phase change memory (PCM) is becoming one of the premier candidates to replace flash memory. It involves switching chalcogenide glass between two states with very different electrical properties—amorphous (logic 1, highly resistive) and crystalline (logic 0, less resistive). As cell size is scaled down to meet size and power requirements, variability between cells becomes increasingly important. As cells are operated, crystalline and amorphous regions can build up within the cell leading to resistance drift, void formations, and other instabilities within the phase change material GeSbTe (GST).
PCM cells with a heater width of 10 nm and GST thickness of 100 nm are studied using 2D finite element simulations with COMSOL Multiphysics. The cells consist of crystalline grains of random size and location, generated using a random function of normal distribution with a user-specified mean and standard deviation, in an otherwise more resistive amorphous region; each cell has a different grain map. Electric field breakdown is included by adding an exponential term to the conductivity of amorphous GST. The material parameters are modeled with full temperature dependency from 300 to 1000 K, including electrical resistivity, thermal conductivity, and Seebeck coefficient.
Results show that nanocrystalline GST mushroom cells present a degree of variability. The greatest amount of current finds the path of least resistance along the crystalline grains, which causes the path to vary based on the grain map. This can be detrimental to device operation unless crystal growth is directly controlled by other means.

Phase change memory is a non-volatile memory technology that a small volume of phase change material can be reversibly changed between amorphous and crystalline phases by suitable electrical pulses to define electrically resistive (amorphous) and conductive (crystalline) states. Phase change from crystalline to amorphous occurs when the material is melted and quenched, and from amorphous to crystalline is achieved by heating the material to just above the crystallization temperature or melting and slowly cooling the material. Crystallization dynamics at device level are key in understanding phase-change memory technology. Crystallization times of nanoscale Ge₂Sb₂Te₅ wires experiencing electrical current are studied through amorphization of the wires using single voltage pulses followed by crystallization at various substrate temperatures (400 K to 675 K). Current through the wire is measured for long periods after the voltage pulse using an offset signal to determine crystallization time. The current levels used for monitoring resistivity of the devices is kept low so that there is no substantial self-heating of the elements. Crystallization times are significantly faster if there is substantial self-heating during the measurement. The crystallization time of Ge₂Sb₂Te₅ decreases with increasing temperature.

The difference in bonding between the bulk amorphous and crystalline phases of GeSbTe phase change materials is accounted for in terms of resonant bonding in the crystalline phase. This accounts for the optical contrast. The electrical contrast is explained in terms of the different behavior of gap states. In the crystalline phase Ef is pinned at the valence band edge by Ge vacancies. Ge vacancies are a low energy defect because of reconstruction along resonant bonding directions. In the amorphous phase, Te excess is accommodated by incorporation into the network, and overlapping band tails tend to pin Ef in midgap. Valence alternation pairs are not a low energy defect. This difference extends to interfacial phase change materials.

Phase-change materials show remarkable electrical transport properties in their amorphous state such as the threshold switching or the resistance drift effect [1]. Threshold switching denotes the sudden decrease of the amorphous state resistivity at a material dependent critical threshold field, which is of the order of 10 MV/m [2]. The amorphous state resistivity below the threshold shows thermally activated behaviour and is observed to increase with time. A better understanding of the physical mechanisms driving threshold switching and resistance drift phenomena is mandatory to improve non-volatile phase-change memories. Even though both phenomena are often attributed to localized defect states in the band gap [3,4], little is known about the defect density in amorphous phase-change materials.
This work presents a model for the density of states of amorphous Germanium telluride (a-GeTe), a pioneering phase change alloy combining D-RAM like switching speeds with a high crystallization temperature [5, 6]. Recently, it was shown that the density of states of a-GeTe can be experimentally probed by Modulated Photocurrent Experiments (MPC) and Photothernal Deflection Spectroscopy [7-9]. In this communication we propose a density of states derived by means of simulation consisting of donor, acceptor and band tail states, which reproduces our experimental MPC and PDS results as well as the measured steady state photoconductivity.
[1] M. Wuttig and N. Yamada , Nat. Mat. 6, 824-332 (2007)
[2] D. Krebs et al. , Appl. Phys. Lett., 95, 082101 (2009)
[3] A. Pirovano et al. , IEEE., 51, 714 (2004)
[4] D. Ielmini, Phys. Rev. B, 78, 035308 (2008)
[5] G. Bruns et al., Appl. Phys. Lett., 95, 043108, (2009)
[6] M. Chen et al., Appl. Phys. Lett., 49, 502 (1986)
[7] J. Luckas et al. , PSS C, 7, 852 (2010)
[8] J. Luckas et al., J. Appl. Phys., 110, 013719 (2011)
[9] D. Krebs et al., J. Non-Cryst. Solids 358 (17), 2412 (2012)

Phase change memory devices are based on the electrical resistivity contrast between the amorphous and the crystalline phases of chalcogenide materials [1]. Since melting is required to amorphize the material, knowledge of the liquid state properties is critical for device design. Two previously reported values for the electrical resistivity of liquid Ge₂Sb₂Te₅ (GST) differ significantly: 4 m#8486;.cm obtained from a measurement on a thin film [2] and 0.41 m#8486;.cm obtained from bulk measurements [3]. We have obtained the electrical resistivity of liquid GST from thin films using DC current-voltage measurements and from small-scale encapsulated devices using microsecond pulse voltage and current measurements. The devices have metal contacts and have lengths from 315 nm to 675 nm, widths from 60 nm to 420 nm and thickness of 20 and 50 nm [4]. Thin film measurements yield 1.26 ± 0.15 m#8486;.cm (film thickness: 50, 100 and 200 nm), however, there is significant uncertainty regarding the integrity of the film in liquid state. The results we obtained from device-level measurements, 0.31 ± 0.04 m#8486;.cm and 0.21 ± 0.03 m#8486;.cm from 20 nm and 50 nm thick wires arrays, are close to those obtained from measurements on bulk GST [3]. Details of the measurements and analysis will be presented.
References
[1] H. -. P. Wong, S. Raoux, S. B. Kim, J. Liang, J. P. Reifenberg, B. Rajendran, M. Asheghi and K. E. Goodson, "Phase Change Memory," Proc. IEEE, vol. 98, pp. 2201-2227, 2010.
[2] T. Kato and K. Tanaka, "Electronic Properties of Amorphous and Crystalline Ge₂Sb₂Te₅ Films," Japanese Journal of Applied Physics Part 1 Regular Papers Short Notes and Review Papers, vol. 44, pp. 7340, 2005.
[3] R. Endo, S. Maeda, Y. Jinnai, R. Lan, M. Kuwahara, Y. Kobayashi and M. Susa, "Electric Resistivity Measurements of Sb₂Te₃ and Ge₂Sb₂Te₅ Melts Using Four-Terminal Method," Japanese Journal of Applied Physics, vol. 49, pp. 5802, 2010.
[4] K. Cil, F. Dirisaglik, L. Adnane, M. Wennberg, A. King, A. Azer Faraclas, M. Akbulut, Y. Zhu, C. Lam, A. Gokirmak and H. Silva, "Electrical Resistivity of Liquid Ge₂Sb₂Te₅ Based on Thin Film and Nanoscale Device Measurements " Electron Devices, IEEE Transactions on (2012) Accepted.

Phase-change memory materials are being actively researched for non-volatile resistive random-access memory (RRAM) applications. Multilevel programming, wherein several different resistance states can be stored in the memory material, allows more than one bit to be stored per memory cell. However, this route to increasing data density without size down-scaling is threatened by the phenomenon of ‘resistance drift&’, wherein the electrical resistance of the amorphous state increases inexorably with time after being written with a voltage pulse. Here, we provide a new theoretical model for this electrically-induced effect, akin to that for the decay of the persistent photoconductivity widely observed in disordered semiconductors. This is based on a model involving long-time charge-carrier detrapping and recombination being operative in a-GST, and this is responsible for the resistance drift observed in this material. In principle, this insight will then allow for the resistance drift to be mitigated by suitable materials ‘engineering&’ via control of the band-tail electronic states.

Phase change memory (PCM) devices have today reached the level of a mature technology with enabling functionalities ranging from non-volatile data storage [1], random access memory [2] and analog/digital computing [3,4]. Although PCM offers outstanding properties in terms of fast switching, area scaling and extended endurance, reliability is still affected by temperature-sensitive crystallization and resistance drift. The latter phenomenon consists of a time-dependent increase of resistance and threshold voltage Vt, as a result of the structural relaxation (SR) in the amorphous phase. Despite several attempts to attenuate/compensate drift effects [5,6], SR appears an intrinsic process in all amorphous semiconductors [7-10].
This work discusses the impact of electrical pulses on the drift of Vt, which is the voltage for the onset of threshold switching in PCM devices. First, we show that Vt increases for increasing times after the reset pulse. This is attributed to the increasing activation energy for conduction, resulting in a lower Poole-Frenkel current and a consequently larger voltage to induce threshold switching [10]. Then, Vt is measured as a function of the duration tp of the measurement pulse following the reset pulse. Two regimes are evidenced: at short tp, Vt increases for decreasing tp. This is because the switching probability Ps increases with voltage, therefore a short duration of the pulse must be compensated by a larger voltage to achieve a critical Ps [11]. At long tp, instead, we evidence an anomalous increase of Vt for increasing tp. Such electrical-acceleration is most probably due to pulse-induced Joule heating within the active volume, causing a faster SR and Vt drift.
Our interpretation was validated by applying a sequence of pulses to measure Vt. In the pulse sequence, the amplitude is increased by 10 or 20 mV after each pulse, until Vt is reached. Results show that Vt is higher when measured by multiple pulses instead of by a single pulse, since each pulse contributes to the acceleration of Vt drift. Results are interpreted based on the logarithmic time increase of Vt drift, thus providing support for our picture of electrically-induced drift in PCM. Our findings may pave the way for a purely-electrical suppression of drift, thus enabling highly-reliable multilevel PCM.
[1] H.-S. P. Wong, et al., Proc. IEEE 98 2201 (2010).
[2] G. Bruns, et al., Appl. Phys. Lett. 95, 043108 (2009).
[3] D. Kuzum, et al., Nano Lett. 12, 2179 (2011).
[4] D. C. Wright, et al., Adv. Mater. 23, 3408 (2011).
[5] M. Mitra, et al., Appl. Phys. Lett. 96, 222111 (2010).
[6] A. Sebastian, et al., J. Appl. Phys. 110, 084505 (2011).
[7] K. P. Chik, Solid State Commun. 33, 1019 (1980).
[8] L. L. Snead, et al., Nucl. Instrum. Meth. B 191, 497 (2002).
[9] N. M. J. Conway, et al., Appl. Phys. Lett. 73, 2456 (1998).
[10] N. Ciocchini, et al., Mater. Res. Soc. Symp. Proc. Vol. 1431 (2012).
[11] S. Lavizzari, et al., IEEE T. Electron Dev. 57, 1047 (2010).

While phase change memory technology has become more mature in recent years, fundamental problems linked to the electrical transport properties in the amorphous phase of phase change materials remain to be solved. The increase of resistance over time, called resistance drift, for example prompts a major challenge for the implementation of multilevel storage which eventually is necessary to compete in terms of high storage densities.
A lot of studies have been performed to gain better understanding of resistance drift and the underlying transport mechanism. Thus, the literature on the phenomenological description of resistance drift and its time and temperature dependence[1] on one side and theoretical studies of the structure of amorphous phase change materials on the other side is broad. However, to link structural changes with electrical transport a broader knowledge of (i) changes in the density of states (DoS) upon structural relaxation and (ii) the influence of defects on electrical transport is required.
In this work, we present temperature dependent conductivity and photo-conductivity measurements on GeTe and other phase change materials[2]. It is shown that trap limited band transport at high temperatures (above 165 K for GeTe) and variable range hopping at low temperatures is the dominating transport mechanism.
Based on measurements of the temperature dependence of the optical band gap[3], modulated photo-conductivity[4] and photo-thermal deflection spectroscopy[1] a DoS for GeTe has been proposed previously[5]. Using this DoS, the temperature dependence of conductivity and photo-conducticity has been simulated. Our work shows how changes in the DoS (band gap and defect distributions) will affect the electrical transport before and after temperature accelerated drift[2]. The decrease in conductivity upon annealing can entirely be explained by an increase of the band gap by about 12%. However, low temperature photo-conductivity measurements also suggest a change in the defect density.
[1] Krebs et al., Journal of Non-Crystalline Solids 358, 2412 (2012).
[2] Oosthoek et al., Journal of Applied Physics 112, 084506 (2012).
[3] Luckas et al., Journal of Applied Physics 110, 013719 (2011).
[4] Luckas et al., Physica Status Solidi (c) 7, 852 (2010).
[5] Longeaud et al., On the Density of States of Germanium Telluride, (submitted).

In recent years electronic memories based on phase change materials have matured into a technology that is realized in commercial devices. The pronounced contrast in resistivity between the amorphous and the crystalline phase is especially attractive since it facilitates the realization of multilevel storage multiplying data densities on phase change chips [1]. However, this concept faces an obstacle as the resistivity of the amorphous phase drifts over time, which eventually leads to a loss of data [2].
Various studies on phase change thin films as well as on integrated memory devices have provided empirical evidence that the temporal evolution of the resistance in amorphous phase change materials follows a power law.
The physical mechanism underlying this process, however, is discussed controversially. Some researchers see the fundamental reason for resistance drift in mechanical stresses inside the amorphous material caused by the pronounced density contrast between the different phases [3]. Others discuss it as a generic structural relaxation common to amorphous semiconductors in general [4]. Either way, when speculating about the physics behind resistance drift it is typically assumed that it must be accompanied by a rise of the activation energy for conduction [5].
To confirm or refute this assumption we have studied experimentally how the activation energy for conduction in amorphous thin films of Ge₂Sb₂Te₅ and other phase change materials actually changes while the resistance drifts. While for Ge₂Sb₂Te₅ the results are in agreement with a direct connection between the drifting resistance and the activation energy, for other materials the contribution from the pre-factor in the Arrhenius law is dominating over the change of activation energy over time. This new experimental evidence must stir up the allegedly save assumptions about what is behind the drift of resistance in amorphous phase change materials.
[1] Nirschl, T., et al., Write strategies for 2 and 4-bit multi-level phase-change memory. 2007 IEEE International Electron Devices Meeting, Vols 1 and 2, 2007: p. 461-464.
[2] Pirovano, A., et al., Low-field amorphous state resistance and threshold voltage drift in chalcogenide materials. IEEE Transactions on Electron Devices, 2004. 51(5): p. 714-719.
[3] Braga, S., A. Cabrini, and G. Torelli, Dependence of resistance drift on the amorphous cap size in phase change memory arrays. Appl. Phys. Lett., 2009. 94(9).
[4] Boniardi, M. and D. Ielmini, Physical origin of the resistance drift exponent in amorphous phase change materials. Appl. Phys. Lett., 2011. 98(24).
[5] Boniardi, M., et al., A physics-based model of electrical conduction decrease with time in amorphous Ge(2)Sb(2)Te(5). J. Appl. Phys., 2009. 105(8).

Recently, phase change random access memory (PC-RAM) has been studied extensively as a promising alternative to existing flash memories. PC-RAM stores information as a chalcogenide alloy goes through phase transition between the low resistive crystalline phase (SET state) and the high resistive amorphous phase (RESET state). Most of previous microstructural characterization studies have reported only final snapshot images of the crystalline and the amorphous phases after switching, which do not allow a direct correlation between structural phase transition and the electrical properties during switching processes.
In this presentation, using in-situ transmission electron microscopy (TEM), we show the real-time switching operation of GeTe-Sb₂Te₃ (GST) alloy based PC-RAM vertical cells by directly applying DC and AC biases. The common ‘mushroom-like&’ cell structure was fabricated by conventional method. The cross-sectional TEM specimens were prepared by conventional focused ion beam (FIB). The in-situ TEM experiments were performed by using a field emission TEM (JEOL JEM-2100F) operated at 200kV and a single-tilt STM-TEM holder (Nanofactorytrade;). A function generator was installed additionally to the system in order to apply the pulses.
In DC biasing experiments, we used the RESET state specimen and then applied the DC bias from 0 to 1.5 V in total 150 seconds and simultaneously measured DC current while observing the microstructure of GST in real-time (25 fps). In an initial bright-field TEM image of the RESET state of GST, we confirmed that the contrast of programming volume was very uniform without any trace of diffraction contrast. In the range of 0 to 1.2 V, the contrast was changed little and the corresponding resistance value remained to around 25 kOmega;. Around 1.2 V, the local contrast change and concurrent electrical threshold behavior was observed. In the range of 1.2 to 1.5 V, the contrast variation and abrupt increase of current level (corresponding resistance was 0.64 kOmega;) was clearly observed, indicating that the RESET state was transformed to the SET state by crystallization process due to local joule heating.
In AC biasing experiments, the waveforms for switching were programmed such as 20/500/20ns and 100/500/10000ns (leading edge/width/trailing edge), respectively. When the SET pulse is applied to the RESET state specimen with progressively increasing the flat voltage from 0.1 to 1.5 V, the local contrast variation was observed at 1.5 V, which is similar to what observed in the DC experiments. During back switching to the RESET state, the diffraction contrast became relatively uniform but small voids and pores were formed. As the switching was repeated, the concentration of voids and pores increased in size even further. The phase transition behaviors observed during DC and AC switching will be discussed in greater detail in conjunction with the elemental mapping and the simulation of heat distribution around the programming volume.

Since S. R. Ovshinsky discover the reversible phase transition characteristics of chalcogenide glasses, it have roused interest of many researchers about this phenomenon [1]. In many studies about new chalcogenide glasses, Ge-Sb-Te ternary material, especially Ge2Sb2Te5, have attracted public attention owing to its good reversibility in phase transition and fast phase transition speed. From this merit, Ge2Sb2Te5 has been used widely in optical memory devices (CD, DVD, Blue-ray). However, Ge2Sb2Te5 has relatively low crystallization temperature of around 150°C. This low Tm is unsuitable for phase change memory, because data retention can be a serious problem [2]. Our research team tried to find new materials having sufficiently high crystallization temperature. In this work we chose GeTe as the base materials for new PCMs owing to its high crystallization temperature, high resistance ratio, and good crystallization rate [3]. From many previous studies, we could know that metallic doping in chacogenide glasses have an effect on the electrical characteristics [3, 4]. Also it is well-known to control the short-range order through compositional variation for target characteristics [5]. In this work, we investigate overall phase change behaviors of Ge-Te-Ag ternary system in the broad Ag composition by co-sputtering GeTe and Ag. We could observe that the crystallization temperature decrease until threshold Ag composition, which is 7.74mol%, and then increase continuously. This reverse tendency of characteristics is also found in other results [4, 5, 6]. We focused on the reason of this phenomenon from the crystallographic differences between low and high Ag concentration systems. In our work the significant changes of crystalline was found from various analysis. We expect this research can give a way to control the intrinsic characteristics of PCMs for target performance.

Phase change memory (PCM) is a non-volatile memory technology that is based on the electrical resistivity contrast between amorphous (high resistivity) and crystalline (low resistivity) phases of phase change materials. Amorphization and crystallization can be achieved reversibly by suitable electrical pulses. Phase change memory devices operate in nanosecond time scales, and these phase change materials do not have enough time to change phase until melting due to crystallization times that are much longer than the typical reset pulse durations; hence phase transitions are not seen at metastable state. Typical thin film resistivity vs. temperature characterization is based on slow measurements (~1 to 100 K/min) which cannot capture metastable material characteristics as the phase change materials continuously change phase during the measurement.
In this work, the resistivity of metastable amorphous Ge₂Sb₂Te₅ is measured at elevated temperatures (300 K to 675 K) using tailored voltage waveforms including a melting pulse on a large number of PCM line-cells with varying dimensions.

Composition-phase-tuned GeSbTe (GST) nanowires (NWs) were synthesized by the thermal evaporation method of GeTe/Sb2Te3 powders. As the Sb content increased up to x=[Sb]/([Ge]+[Sb])=0.8, rhombohedral-cubic-hexagonal (rhombohedral) phase GST NWs were produced consecutively, showing a distinctive phase evolution. Remarkable superlattice structures were identified for the Ge8Sb2Te11, Ge3Sb2Te6, Ge3Sb8Te6, and Ge2Sb7Te4 NWs. The coexisting cubic-hexagonal phase Ge3Sb2Te6 NWs exhibited an exclusively uniform [111]C superlattice structure consisting of 2.2 nm-period slabs with straight and helical morphologies. The hexagonal phase Ge3Sb8Te6 and Ge2Sb7Te4 NWs adopted an innovative superlattice structure consisting of alternating GeTe-rich and Sb-rich domains stacked along the [0001] growth direction. We suggest that three Sb2 layers intercalated the Ge3Sb2Te6 and Ge2Sb1Te4 domains, respectively, producing 3.4 nm and 2.7 nm period slabs of Ge3Sb8Te6 and Ge2Sb7Te4.

During the repeated set/reset operations of GeSbTe (GST) PRAM cell, the GST film should be fast quenched and crystallized. However, it is known that if Ge-Te bonds are remained in the fast quenched GST film, activation energy for crystallization increases and subsequently, crystallization speed decreases at the set pulse. Therefore, in this work, we have investigated microstructure of the GST PRAM cell during the set/reset operations with high resolution transmission electron microscope (HR-TEM) and understand the failure mechanism of the GST film. After the repeated set/reset operations with 20ns pulses, when the GST is heated up to the melting temperature and then, the cooling rate is not fast enough to quench the GST film below than the crystallization temperature, the HR-TEM, the fast Fourier transform images and selected area electron diffraction (SAED) patterns show that the GST is dominantly changed to the amorphous, but hexagonal structured GST and GeTe phases are also locally observed due to incomplete melt-quenching during the reset operation. The GeTe-Sb2Te3 pseudo-binary phase diagram also indicates that the hexagonal Ge2Sb2Te5 is appeared below than 600°C and the GeTe phase is existed with melted GST at 630°C. The HR-TEM and SAED patterns show that at the set operation since the GST film is already crystallized into the hexagonal as well as the face centered cubic GeTe. As a result, the PRAM cells are operated with relatively lower current, but the switching speed is not decreased. This means that the crystallization speed of the GST is not sensitive to incomplete quenching and the existence of GeTe phase. However, during the repeated set/reset operations, besides of these incomplete hexagonal GST and GeTe, a void is formed and the GST atoms seems to be depleted at the interface between the electrode and the GST film. Eventually, when these voids and GST depleted regions become to be enlarged the PRAM cell does not return to the set state again. In this work, the microstructural behavior of the GST PRAM cell will be discussed with the direct observation of the cross sectional devices to provide clear evidence for the failure mechanism of the GST PRAM cells.

Ge₂Sb₂Te₅ (GST) is a key material used in rewritable phase-change optical memories and phase-change random access memories. The understanding on the local order in amorphous structures would be not only fundamentally stimulating but also technologically important to resolve various technological obstacles. There have been several studies in this direction, but unfortunately there is serious discrepancy between first-principles results and the extant experimental observations. In particular, averaged Ge-Te distance is longer than experimental values by 0.1~0.2 Å in both liquid and amorphous structures. This is closely related to description of electron-electron interaction by semilocal functionals which tend to favor octahedral Ge atoms. Being similar to the crystalline bonding geometry, the octahedral Ge atoms lead to more delocalized electrons. However, in disordered systems such as liquid and amorphous phase, the electrons at band edges are inherently localized and chemistry-based functionals such as BLYP or Hatree-Fock can be more valid.

In this presentation, we discuss on the liquid and amorphous structures of GST studied with molecular dynamics simulations using various exchange-correlation functionals such as PBE, BLYP, and HSE06 with 0.25 and 0.5 mxing parameter. We confirm the previous conclusion that the PBE functional results vary from the experimental data as mentioned above. On the other hand, the hybrid functionals significantly increase the population of t-Ge atoms and the liquid structures are in much better agreement with experiment compared to results with semilocal functionals. This implies that the chemistry-based functionals such as hybrid functionals might better describe the disordered systems than PBE. The amorphous structures are obtained through the melt-quench process and hybrid functionals produce amorphous structures that satisfy 8-N rule more closely than semilocal functionals. Lastly, we examine relative preference between octahedral and tetrahedral Ge atoms by using random phase approximation (RPA) method based on adiabatic conection fluctuation dissipation theory (ACFDT) for which it is believed that the accuracy is higher than semilocal or hybrid functionals.
Contact: [email protected]

Recently the use of germanium-based precursors metal-organic chemical vapor deposition (MOCVD) and atomic layer deposition (ALD) of phase-change chalcogenide alloy films with the prototype of germanium antimony telluride (GST), which undergo a thermally induced crystalline-amorphous phase transition for data storage, has been demonstrated in the next-generation nonvolatile phase change random access memory (PCRAM) to replace conventional dynamic and flash random access memory technology. In this presentation, we will disclose the synthesis and characterization of novel germanium precursors on the basis of molecular design. We have synthesized and characterized novel germanium chalcogenide precursors, Ge complex (1) and Ge-Te complex (2), for Ge2Sb2Te5 (GST) materials. Gemanium-based chalcogenide materials have been synthesized by thermal decomposition of single precursor with surfactant, and characterized by powder X-ray diffraction (XRD), scanning electron microscope (SEM), and energy dispersive spectroscopy (EDS). XRD and EDS data showed the thermally decomposed product of the complex 2 is germanium telluride with small amounts of Te.

Phase Change Memory (PCM) attracts increasing interest as low-power storage device due to its fast switching speed and good scalability.[1] Its characteristics are modeled by thermal simulation, which is an essential tool for understanding the behavior of PCM cells and for investigating new materials. This study reports multiphysics modeling of PCM with simple analytical expressions which take into account the temperature and electrical field dependence of material properties. This approach can describe phase change physics with the limited number of parameters. Electrical conductivity was fitted with Poole-Frenkel equation. Thermal conductivity was approximated by linear function with respect to temperature. Phase change dynamics were described by the following equation, dc/dt=(1-c)*fc-c*fa, where c is the “degree of crystallization”, namely, the fraction of crystal in a mesh. Two material parameters, fa and fc, are characteristic of amorphization and crystallization frequency, respectively. The model assumed fa to be non-zero constant above melting temperature, and fc was extracted by experimental data. Electric conductivity of mixed phase (0

Among the rich vanadate familly performing metal to insulator transitions (MIT), vanadium dioxide is the most intensively studied for the past several decades. Indeed, vanadium dioxide undergoes an abrupt reversible first order phase transition within temperature from insulating monoclinic phase to metallic rutile phase. The remarkable character of this transition is what takes place at a temperature close to room temperature, at about 68°C. The insulator to metal transition can be achieved with other external stimuli than temperature such as light, electric field, mechanical strain. MIT mechanism in vanadium dioxide has been a topic of debate between two fundamental descriptions: a Mott transition driven by electron-electron correlation and a Peierels transition driven by electron-lattice interactions. Recently, numerous studies and experiments shed new light on the nature of the phase transition mecanism. To this purpose, micro or nanoscale imaging have been successfully used to study the MIT on polycristalline thin films and single cristalline VO2 nanobeams.
In our study, vanadium dioxide films about 160nm thick were grown on c-saphirre (Al2O3) substrates by pulsed laser deposition. X ray diffraction patern and AFM topological studies reveals that PLD films exhibit polycrystalline structures corresponding to the exclusive formation of monoclinic vanadium dioxide. We only observe one orientation (0 l 0).
We performed also electrical resistivity measurements using a four point probe method, in the temperature range from 100K to 380K. The change in the VO2 resistivity is found to be around four orders of magnitude
A Jobin Yvon-UVISEL spectroscopic ellipsometer (SE) with a temperature cell has been used to characterize the optical properties of these thin films. The SE measurements were performed at two different incidence angles 60 and 70° from room temperature to 95°C. The optical constants of VO2 thin films as a function of the frequency were determined by a classical dispersion model based on the sum of Lorentz and Drude oscillators. Optical constants of low temperature (20°C) insulating VO2 and high temperature (95°C) metallic VO2 were then used to model the dispersion via effective medium approximation.

We investigated the coherent acoustic and optical phonon dynamics of Ge2Sb2Re5 phase change material using pump-probe spectroscopy. Ge2Sb2Te5 thin films were grown on 100-nm thick SiO2/Si substrates using radio frequency (RF) sputtering. Oscillating reflectivity changes with oscillation period on the order of sub-picosecond and tens of picoseconds were observed. The longer oscillation with oscillation period and dephasing time on a time-scale of tens of picoseconds hasn&’t been reported previously and can be consistently explained by coherent acoustic phonon generation in the form of a stain pulse propagating into substrate. The shorter period oscillations corresponds to coherent A1 optical phonon by diplacive excitation of coherent phonon mechanism. Its temperature dependent frequency shift was due to a three-phonon anharmonic decay, while its de-phasing was dominated by temperature independent phonon-vacancy scattering. Laser fluence-dependent coherent A1 phonon mode softening and a dephasing rate increase were observed and attributed to the high density electrons and accompanying lattice distortion as well as to the lattice heating.

Phase-change materials are nowadays used for writing and storing information on common data carriers due to their ability to switch between a crystalline and an amorphous state [1-4]. Besides the change in long-range order of these two phases, the switching behavior is also accompanied by a change in the local order and therefore influences the bonding properties of the atoms. This can be observed from major differences of the lattice dynamics of the two structures which assumably also affect the functionality. It was experimentally found that GeSb₂Te₄ reveals both a hardening of the low-energy acoustic phonons and a softening of the high-energy optical phonons upon crystallization [5].
Here, we present the results of various density-functional calculations on the vibrational properties of GSTs as well as the cornerstone compounds GeTe and Sb₂Te₃ using the ab initio force-constant method [6]. The influence of dispersion-corrected DFT as well as different parametrizations of exchange and correlation is investigated with a special focus on GeTe and Sb₂Te₃. For the crystalline GST materials we have carried out calculations based on a rock salt-related 64 atom-containing model, especially looking at the low-energy acoustic phonons and the high-energy optical phonons.
Moreover, it has been shown that within the amorphous phase the coordination number lowers substantially [7,8]. Therefore a massive atomic rearrangement will occur upon amorphization/crystallization, which of course implies atomic movement on a local scale. We investigate the atomic motion in GeTe by looking at the actual pathways of the atoms by calculating the activation barriers using the nudged-elastic band method. Moreover, the direct influence of vacancies and doping atoms on the energy barrier is examined.

References
[1] D. Lencer, M. Salinga, M. Wuttig, Adv. Mater. 23, 2030 (2011).
[2] M. Wuttig, W. Bensch, Chem. unserer Zeit 44, 92 (2010).
[3] S. Raoux, M. Wuttig (Eds.): Phase Change Materials: Science and Applications, Springer, New York (2009).
[4] D. Lencer, M. Salinga, B. Grabowski, T. Hickel, J. Neugebauer, M. Wuttig, Nature Mater. 7, 972 (2007).
[5] T. Mastunaga, N. Yamada, R. Kojima, S. Shamoto, M. Sato, H. Tanide, T. Uruga, S. Kohara, M. Takata, P. Zalden, G. Bruns, I. Sergueev, H. C. Wille, R. P. Hermann, M. Wuttig, Adv. Funct. Mater. 21, 2232 (2011).
[6] A. Togo, F. Oba, I. Tanaka, Phys. Rev. B 78, 134106-1-9 (2008).
[7] J. Akola, R. O. Jones, J. Phys: Condens. Matter 20, 465103 (2008).
[8] R. Mazzarello, S. Caravati, S. Angioletti-Uberti, M. Bernasconi, M. Parrinello, Phys. Rev. Lett. 104, 088503 (2010).

In view of the application as active element in ultra-scaled phase change memory devices, self-assembled chalcogenide nanowires are becoming a valid option. In particular, in order to reduce the writing current and power consumption necessary for low power devices, both the contact size and the volume to be programmed could be reduced by the introduction of self-assembled chalcogenide nanostructures. Reaching the nanometer scale, the electrical properties of single crystalline nanowires can be influenced by the reduced dimensionality, due to peculiar structural characteristics, such as the high surface to volume ratio and the absence of intra-grain defectivity.
In this work we present the structural and electrical properties of Sb2Te3 and Ge-doped Sb2Te3 (Gele;13%) nanowires self-assembled by metalorganic chemical vapour deposition (MOCVD) using the Au-catalysed vapour-liquid-solid mode. The structural and chemical properties are discussed by using both larger area (total reflection X-ray fluorescence and X-ray diffraction) and local area (scanning electron microscopy, SEM, transmission electron microscopy, TEM, coupled with high angle annular dark field detectors, X-ray microanalysis and Fourier transform analysis of selected areas) techniques. For 5le;Gele;13%, the nanowires show a disordered structure, with random stacking of slabs with different stoichiometry. For 0le;Ge<5%, the nanowires are single-crystal and defect free. Crystallization occurs within the R-3m phase of Sb2Te3, with different growth directions. In some particular cases, kinking and bending of the nanowires have been observed by both SEM and TEM. These variations of the growth direction are not related to the presence of any interfacial defect. They can be understood in the light of the stabilization of metastable phases by doping of Sb2Te3 with Ge or by reducing the dimensionality of the nanostructure. For pure Sb2Te3 nanowires, different growth parameters have been explored, including the variation of the total precursor concentration of the Sb to Te precursor ratio and different size of the Au seed nanoparticle. For the same growth parameters, pure Sb2Te3 nanowires have a higher diameter than the Ge-doped ones (230nm vs 50-80 nm). The reduction of the dimensionality of Sb2Te3 pure nanowires was achieved by increasing the total precursor&’s concentration. In this case, thinner and longer single crystalline Sb2Te3 nanowires have been achieved, with diameter ranging from 70 nm to 100 nm and length up to 4 mu;m.
Pulsed I/V measurements have been carried out on single nanowires though simple devices, obtained by depositing Pt electrodes on the nanowire tips, using focused ion beam. The electrical characteristics, including cyclability, nanowire resistance and threshold voltage, are discussed in order to elucidate the effect of Ge doping and of the dimensionality of the Sb2Te3 nanowires.

It was reported that phase change can result in a large stress in chalcogenide owing to 6-10% volume change [1]. Such a huge volume change can have a great influence on the performance of phase-change memory (PCM) if we consider the cycling of the memory. Voids tend to form at the interface between chalcogenide and electrodes, resulting in a serious reset-stuck failure [2]. However, there is no report aimed for the reduction of volume change in chalcogenides upon crystallization or amorphizaion. GeTe is adopted here as a base chalcogenide because GeTe-based PCM was reported to have fast speed of ns order [3].
In this study, we systematically investigated GeTe-based chalcogenides to know the influence of N-doping into GeTe on the volume change and crystallization temperature for improving the performance of PCM. Experimental results show that N-doping with a proper quantity led to an ultralow volume change upon crystallization by annealing up to 400oC, nearly 0, which can reduce internal stress during cycling to greatly improve endurance of memory. N-doping can also increase the crystallization temperature above 300oC from 223oC based our differential scanning calorimetry measurements to greatly improve retention of memory. Resistivity change of 7-8 orders of magnitude upon crystallization was observed for the film with a low volume change. Therefore, improved performance of PCM can be expected by adopting the low-volume-change high -crystallization-temperature N-doped GeTe.
This work was financially supported by Grant-in-Aid for Young Scientists and Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan (Grant Nos. 24686042, and 24360003).
[1] T. P. L. Pedersen, J. Kalb, W. K. Njoroge, D. Wamwangi, M. Wuttig, and F. Spaepen, Appl. Phys. Lett. 79 (2001) 3597.
[2] H. Lee, D. H. Kang, Jpn. J. Appl. Phys. 44 (2005) 4759.
[3] G. Bruns, P. Merkelbach, C. Schlockermann, M. Salinga and M. Wuttig, T. D. Happ, J. B. Philipp, and M. Kund, Appl. Phys. Lett. 95 (2009) 043108.

Ge2Sb2Te5 (GST) is widely used in phase-change memory (PCM), regarded as the best candidate for next generation nonvolatile memory for it has been demonstrated to have almost perfect performances (e.g., fast speed, excellent endurance, low power, low cost) required by the growing demand. Ultrafast speed, the order of 1ns for an operation, was demonstrated by Bruns et al using a GeTe vertical PCM cell with a very small size of contact area of 60 nm [1]. Growth dominated recrystallization was thought to be the reason of such a fast speed. However, there is a lack of study on the application of GeTe to the lateral device and further multilevel storage [2].
In this study, we demonstrated that the operations of both set (crystallization) and reset (amorphization) with a resistance change of around two orders of magnitude can be conducted as fast as 10 ns in our lateral TiSi/GeTe PCM device. Comparatively, 100 ns operation time was necessary for the TiSi/GST device. Furthermore, we investigated freely achievable multilevel storage in our lateral TiSi/GeTe PCM device by staircase-like pulse programming. Staircase-like pulse here consisted of first sub-pulse and second sub-pulse. The former sub-pulse was used to melt the whole programmable region and the latter was used to control the crystallization degree or region only by changing the pulse width or the pulse amplitude. Experimental results exhibited that freely achievable multilevel storage with a short second sub-pulse of 50 ns in our lateral TiSi/GeTe PCM device could be realized while a second sub-pulse of 500 ns was necessary for the TiSi/GST device.
This work was financially supported by Grant-in-Aid for Young Scientists and Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan (Grant Nos. 24686042, and 24360003).
[1] G. Bruns, P. Merkelbach, C. Schlockermann, M. Salinga and M. Wuttig, T. D. Happ, J. B. Philipp, and M. Kund, Appl. Phys. Lett. 95 (2009) 043108.
[2] Y. Yin, T. Noguchi, and S. Hosaka, Jpn. J. Appl. Phys., 50 (2011) 105201.

The extraction of temperature-dependent electrical resistivity ρ(T) and thermal conductivity k(T) of self-heated Ge₂Sb₂Te₅ (GST) micro/nano-bridges with Hexagonal Close Packed (HCP) crystalline phase by matching simulated and experimental current-voltage (I-V) characteristics will be discussed. We have recently used this technique to extract ρ(T) and k(T) of nanocrystalline silicon microwires self-heated to melting temperature [1]. GST wires on silicon dioxide are self-heated using electrical stress up to the melting point and electrical resistivity of GST is extracted assuming a constant thermal conductivity (since the thermal transport in this case is mostly downward to the silicon dioxide). The extracted electrical resistivity is then used to simulate a suspended wire, where the lateral conduction through the GST wire dominates the thermal transport, to extract the thermal conductivity of GST. Several other wires of different dimensions are simulated using the extracted ρ(T) and k(T) at different ambient temperatures and the results are compared to the corresponding experimental I-V characteristics to validate the extracted temperature-depended parameters.
[1] Gokhan Bakan, Lhacene Adnane, Ali Gokirmak, and Helena Silva, J. Appl. Phys. 112, 063527 (2012).

In the last few years atomistic simulations based on density functional theory have provided useful insights on the properties of phase change materials. However, several key issues such as the crystallization dynamics, the properties of the crystalline/amorphous interface and the thermal conductivity at the nanoscale, just to name a few, require large simulation cells and long simulation times beyond the reach of fully DFT simulations.
A route to overcome the limitations in system size and time scale of DFT molecular dynamics is the development of classical interatomic potentials. Traditional approaches based on the fitting of simple functional forms are very challenging due to the complexity and variability of the chemical bonding in the crystal and amorphous phases revealed by DFT simulations. A possible solution has been demonstrated recently by Behler and Parrinello [1] who developed empirical interatomic potentials with close to ab-initio accuracy for elemental carbon, silicon and sodium by fitting large DFT databases within a neural network (NN) scheme.
By means of this technique, we have recently developed an interatomic potential for GeTe [2] which is one of the compounds under scrutiny for applications in phase change memories. The NN potential has allowed simulating several thousands of atoms for several tens of ns which provided insights on the thermal transport properties of amorphous GeTe [3] and on the fragility of the supercooled liquid phase [4].
We will present the results of NN simulations of the crystallization dynamics of the supercooled liquid and overheated amorphous states of GeTe in the homogeneous phase and at the interface with the crystal that allow extracting the crystal growth velocity as a function of temperature and density.
[1] J. Behler and M.Parrinello, Phys. Rev. Lett. 14, 146401 (2007).
[2] G. C. Sosso, G. Miceli, S. Caravati, J. Behler, and M. Bernasconi, Phys. Rev. B 85, 174103 (2012).
[3] G. C. Sosso, D. Donadio, S. Caravati, J. Behler, and M. Bernasconi, Phys. Rev. B 86, 104301 (2012).
[4] G. C. Sosso, J. Behler, and M. Bernasconi, Phy. Status Solidi B 249, 1880 (2012).

The speed of crystallization is a critical parameter in phase-change materials to be applied in non-volatile memory and cognitive computing. Its temperature dependence must be such as to permit ultra-fast (<100 ns) phase change during device operation, while also providing sufficient resistance to crystallization for data retention under ambient conditions. Recent analyses of the crystallization kinetics in relevant systems such as GST have focused on: (i) the temperature dependence of the viscosity of the supercooled liquid, noting that this is not expected to have an Arrhenius form, and may indeed deviate strongly from that form by showing an exceptionally high fragility, and (ii) the substantial decoupling of crystal growth from viscous flow expected for very fragile liquids. Recent advances in such kinetic analyses will be reviewed. The application of ultra-fast DSC to characterize crystallization kinetics in thin films will be considered, focusing on its use to explore (i) the differences between nucleation-driven (GST) and growth-driven (AIST) phase-change materials and (ii) the effects of layers constraining the chalcogenide thin films. A clear picture emerges of where chalcogenide phase-change materials sit in the general spectrum of glass-forming systems.

Phase change alloys have specific properties that allow their use in data storage devices. The storage concept is based on the huge change of the optical and electrical properties associated with the reversible transitions between amorphous and crystalline phases. The write-erase cycles are induced by proper laser or electrical pulses. The data retention capability of memories based on phase change materials is determined by the stability of a melt-quenched amorphous volume against crystallization at low temperatures, whereas the maximum writing speed can be determined by studying the fast regime. In order to achieve full understanding of crystallization, quantitative experimental data are needed describing the processes, ideally over a large range of temperatures and switching speeds, respectively. Until now, there has been a clear lack of such experimental data due to a series of fundamental challenges [1-2]. We have developed a new approach towards a meaningful quantification of the temperature dependence of crystallization kinetics. The method, that we will describe here, allows the investigation of the technologically relevant melt-quenched amorphous phase under the most favorable isothermal conditions [3]. At the same time, the new method is capable to provide experimental data in both the slow and the fast crystallization regimes providing information that are essential from a technological point of view. This has been achieved using an optical tester that allows time resolved reflectivity measurements from a second to a nanosecond timescale. The crystallization process occurs under isothermal conditions since a stable temperature can be obtained by the combination of laser irradiation and substrate annealing. Employing this new setup we have collected data on the crystallization of several phase change materials. To further validate this approach, transmission electron microscopy (TEM) measurements have been performed and the thermal response of the sample during the phase transitions has been checked by finite element calculations.
REFERENCES
[1] J. Kalb, F. Spaepen & M Wuttig, Appl. Phys. Lett. 84, 5240-5242 (2004)
[2] J. Orava, A. L. Greer, B. Gholipour, D. W. Hewak & C. E. Smith. Nat. Mater 11, 279-283 (2012).
[3] K.F. Kelton, Mat. Sci. Eng. a-Struct. 226, 142-150 (1997).

Phase change materials are the key ingredient for phase change random access memory (PCRAM). Their properties determine to a large degree the functionality of PCRAM devices. The ultimate test of a novel phase change material for PCRAM lies in the testing of PCRAM devices containing such a material, but PCRAM fabrication can take several months for integrated wafers, and fast turn-around testing of new phase change materials is desirable. We have investigated here the usefulness of fast static laser testing for PCRAM material optimization.
Crystallization time is one of the major optimization criteria for PCRAM materials and is determined by nucleation and growth. We observed a large difference between the crystallization behavior for as-deposited amorphous materials compared to re-crystallization of amorphous areas which were produced in crystalline film by laser melt-quenching. For all phase change materials tested the re-crystallization time of melt-quenched amorphous material was faster than the crystallization of as-deposited amorphous films, for some materials such as Ge-Sb and Ge-Te alloys orders of magnitude faster. The re-crystallization of melt-quenched material surrounded by crystalline material reflects closer the situation in a PCRAM cell where e.g. the mushroom-shaped amorphous region in a mushroom cell is surrounded by crystalline material. In this case nucleation is not required and crystallization can occur just by growth from the amorphous-crystalline interface. We found that therefore re-crystallization times are the relevant values when characterizing a material for PCRAM, and crystallization speed of as-deposited amorphous material often grossly underestimates crystallization speed in PCRAM.
On the other hand, we observed that data retention was grossly overestimated in as-deposited films and predicted a much better behavior of a new material compared to it&’s actual performance in PCRAM. Only when data retention was evaluated using again melt-quenched amorphous material reliable data for prediction of device performance were obtained. These are crucial results because they allow material optimization with excellent predictability for device performance in much shorter time than actual device fabrication and testing.
Finally, a class of phase change materials (Sb-rich GaSb alloys) was tested that shows three distinct levels of reflectance and resistance. It is possible to switch repeatedly between these three levels. Using time-resolved x-ray diffraction the three states were identified as fully amorphous (highest resistance, intermediate reflectance), GaSb crystalline and excess Sb amorphous (intermediate resistance, lowest reflectance), and both GaSb and excess Sb crystalline (lowest resistance, highest reflectance). Such a material has the potential for multi-level optical and PCRAM storage. All these experiments demonstrate the great usefulness of static laser testing for material development and optimization for PCRAM.

Chalcogenide alloys have specific properties that allow their use in data storage devices. The data storage concept is based on the huge change in the optical and electrical properties associated to the reversible amorphous to crystal transition induced by laser or electrical current pulses. Crystallization of the amorphous regions will determine the maximum speed and the data retention capability of the devices.
It has been recently demonstrated by EXAFS and Raman spectroscopy that a local rearrangement of the amorphous short range order occurs during ion or laser irradiation promoting the system to a state closer to the crystalline phase by the breaking of homopolar “wrong bonds” [1]. As a consequence the crystallization kinetics of melt-quenched chalcogenide samples noticeable differs from the as-deposited behavior [2][3].
In this study, the crystallization of as sputtered and melt-quenched Ge₂Sb₂Te₅ (GST) and GeTe thin films has been investigated by Time Resolved Reflectivity (TRR), optical microscopy and in situ and ex situ Transmission Electron Microscopy (TEM) in the low temperature regime (130-160°C).
Amorphous GST, 50 nm thick, and GeTe, 35 nm thick, films were deposited at room temperature on SiO₂ layers. Some films were then irradiated by pulsed (asymp;30 ns) XeCl excimer laser (lambda; = 308 nm) and by pulsed (10 ns) Nd:YAG laser (lambda; = 532 nm second harmonic) at an energy density as high as 180 mJ/cm², to obtain melt-quenched amorphous samples.
For GeTe films, simultaneous record of the optical microscope images and TRR provides a clear relationship of the change in the reflectivity with the crystallized area as measured by optical images. As deposited GeTe films require an annealing time of about 40 min at 160°C to crystallize. A temperature of about 142°C is instead enough to accomplish the phase transition for the same time interval in the melt-quenched amorphous films.
In comparison with the behavior of GST, it seems that the enhancement in the nucleation rate of melt-quenched GeTe prevails over the corresponding increase of the grain growth velocity. For this reason at the end of the crystallization process at 142°C the grain size amounts to about 1 mu;m against tens of microns in the as deposited samples. In GST, instead, the ratio between the grain sizes amounts to about five.
Preliminary measurements of crystallization in melt-quenched amorphous GeTe layers, pre-heated at 110°C, do not show appreciable changes with respect to the untreated samples. A more detailed investigation will allow to ascertaining if during a heat treatment below the glass temperature, relaxation induces a change in the crystallization kinetics [4].
[1] Carria et al., Electrochem. Solid State Lett., 14 (12), H480-H482 (2011).
[2] Khulbe et al. J. Appl. Phys., 88, 3926 (2000).
[3] De Bastiani et al., Appl. Phys. Lett. 92, 241925 (2008).
[4] Kalb et al., J. Appl. Phys. 94, 4908 (2003).

Recently, there has been renewed interest in the use of GeTe as a phase change material (PCM) for memory applications because of its high crystallization temperature [1,2] and rapid switching speed [2]. For PCMs used in optical or resistance-based memory, crystallization of the amorphous phase must be achieved within nanoseconds. While knowledge of crystallization kinetics is critical to the understanding phase transformations in PCMs, basic quantities, such as crystal growth rates, are difficult to measure experimentally during highly-driven laser- or current-induced crystallization. Under such conditions, crystal growth may achieve its maximum rate, which cannot be reliably extrapolated from low temperature data (near the crystallization temperature) where growth rates are many orders of magnitude lower.
We use the dynamic transmission electron microscope (DTEM), an instrument capable of nanosecond-scale time-resolved electron imaging and diffraction, to study laser-driven crystallization in GeTe. The DTEM has been used to study laser-induced crystallization in PCMs, such as Ge₂Sb₂Te₅ [3] and GeTe [4]. In the previous work on GeTe, only a single 15-ns electron image could be formed during each laser-induced event. This left open questions about the true growth rates since the incubation time for the imaged crystalline grains was unknown. Here we capture nine-frame movies of the laser-crystallization of amorphous GeTe, in which each frame consists of an electron image with 17.5-ns time resolution. Individual growing grains are tracked during different points in their growth and unambiguous measurements of growth rate may be made. Crystal growth rates exceeding 3 m/s are observed and changes in growth rate are observed during a single crystallization event.
Finite element analysis simulations are used to model the rapidly changing spatial and temporal temperature profiles during laser heating. Observed changes in growth rate during crystallization along with calculated temperature profiles and models of crystal growth rate indicate these experiments probe the crystallization of GeTe at temperatures where the maximum crystal growth rate may be attained.
This work performed under the auspices of the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Materials Sciences and Engineering by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344.
References
[1] S. Raoux et al., Applied Physics Letters95 (2009) 143118.
[2] G. Bruns et al., Applied Physics Letters95 (2009) 043108.
[3] M. K. Santala et al., Journal of Applied Physics111 (2012) 024309.
[4] M. K. Santala et al., Physica Status Solidi B249 (2012) 1907-1913.

Phase change materials (PCM) function as rewritable optical and computer memory because the amorphous-crystalline transition is rapid, reversible, and accompanied by changes in the optical and electrical properties. The structures of crystalline phases can be studied by x-ray diffraction, and it is obvious that focused laser heating of a nanosized crystalline spot can result in rapid melting. However, the amorphous structures are difficult to determine, and the nature of the crystallization mechanism remains the subject of much study and speculation. We have performed two large scale density functional simulations (several hundred atoms over hundreds of picoseconds) of crystallization in the prototype "nucleation dominated" phase change material Ge2Sb2Te5 (GST-225) and can provide details of the changes in order as crystallization proceeds.
The amorphous structure of GST-225 and other alloys of Ge, Sb, and Te can be characterized by "ABAB alternation" (A: Ge,Sb, B: Te) with four-membered ABAB rings as a dominant motif. This pattern is also prevalent in the metastable (rock salt) crystalline structure, and it is plausible that the rapid amorphous-to-crystalline transition be viewed as a re-orientation (nucleation) of disordered ABAB squares supported by the space provided by cavities. Our first DF/MD simulations [1] were on a 460-atom sample at 500, 600, and 700 K for up to 600 picoseconds, where crystallization was promoted by fixing the structure of a crystalline "seed" (58 atoms, 10% vacancies). The second simulations were also on a 460-atom sample at 600 K, but with no constraints on the geometry. The density was adjusted during the simulation to allow for the difference between the amorphous and crystalline forms. Crystallization occurs in stages and was complete after 1.1 ns. The structural changes were monitored by calculating the (partial) pair distribution functions, appropriate order parameters, the number of "wrong bonds" (bond pairs that do not occur in the ordered form), the cavities, and the vibrational and electronic densities of states. We shall present a consistent picture of all these findings.
1. J. Kalikka, J. Akola, J. Larrucea, R. O. Jones, Phys. Rev. 86, 144113 (2012)

The crystallization properties of fast-growing Sb-rich GeSb films were studied using optical microscopy and transmission electron microscopy (TEM). Films of 50 nm to 200 nm were deposited on poorly heat-conducting glass or polycarbonate substrates, and capped with a 5 nm SiO-ZnS layer.
Isothermal measurements using optical microscopy showed that the dominant crystal growth morphology for these films changes from a triangular shaped dendritic-like one to a circular shaped isotropic one when increasing the percentage of germanium. In the range of Ge8Sb92 to Ge10Sb90 a transition regime is found where both morphologies are present, although both morphologies have different incubation times and growth rates. The density and optical reflectivity of both crystal morphologies changes with a minor change in temperature. Additional isothermal measurements were done in-situ using TEM providing further information about the crystal growth rates over four orders of magnitude.
Using TEM more insight is gained into the crystal structure of the dendritic growth in Ge7Sb93 films. The centers of the crystals are orientated with their [0001] axis in an upright position (normal to the film surface). However, during subsequent lateral growth of the crystals the crystal orientation at the three tips of the triangular shaped crystals gradually becomes tilted till an angle of about 55°. This tilted orientation appears then to be a stable during further lateral growth.
Currently we are addressing the crystallization time and reflectivity changes in these GeSb compositions using a static laser tester, as well as comparing the crystal structure after crystallization from the melt and melt-quenched amorphous phase with the isothermally grown crystals in the as-deposited amorphous films.

The relationship between the polycrystalline nature of phase change materials (such as Ge2Sb2Te5) and the intermediate resistance states of phase change memory (PCM) devices has not been widely studied. A full understanding of such states will require knowledge of how polycrystalline grains form, how they interact with each other at various temperatures, and how the differing electrical (and thermal) characteristics within the grains and at their boundaries combine through percolation to produce the externally observed electrical (and thermal) characteristics of a PCM device.
We address the first of these tasks (and introduce a vehicle for the second) by studying the formation of fcc polycrystalline grains from the as-deposited amorphous state in undoped Ge2Sb2Te5. We perform ex situ transmission electron microscopy membrane experiments and then match these observations against numerical simulation. Ramped-anneal experiments show that the temperature ramp-rate strongly influences the median grain size. By truncating such ramped-anneal experiments at various peak temperatures, we convincingly demonstrate that the temperature range over which these grains are established is quite narrow. Subsequent annealing at elevated temperature appears to change these established distributions of grain sizes only slightly.
Our numerical simulator—which models nuclei formation through classical nucleation theory and then tracks the subsequent time- and temperature-dependent growth of these grains—can match these experimental observations of initial grain distributions and crystallization temperature both qualitatively and quantitatively. These simulations show that the particular narrow temperature range over which crystallization occurs shifts as a function of temperature ramp-rate, which allows us to quantify the lower portions of the time-temperature-transformation map for Ge2Sb2Te5.
Future experiments and extensions of the simulator to investigate temperature-dependent interactions between neighboring grains, and to study nucleation from within the melt-quenched amorphous state, are discussed.

Synaptic electronics is an emerging field, which aims to reach brain-like performance in real time processing of sensory data. Parallelism, configurability, integration of memory and computation, and energy efficiency are crucial for brain&’s computational performance. A compact nanoelectronic device emulating the functions and plasticity of biological synapses will be the key building block for brain-inspired computational systems. Phase change materials offer several advantages such as scalability, reliability, endurance, multiple resistance levels and low device-to-device variation [1]-[4], which enables fabrication of industrial-scale arrays for nonvolatile memory applications. These properties make phase change materials a suitable candidate for implementing large-scale synaptic systems for neuromorphic computation.
Here, we describe programming techniques to implement gradual set and reset of phase change devices and discuss partial crystallization and amorphization resulting in gradual resistance change. We demonstrate how gradual programming capability can be utilized for synaptic plasticity using phase change materials [5]. Then we will review different synaptic plasticity schemes from optimization of energy consumption perspective [6]. Finally, we will present our network-level simulation results based on measured device characteristics of phase change synaptic devices, demonstrating hippocampus-like learning in synaptic arrays. Our analyses show that immunization against variation and noise can be achieved using recurrent networks with plastic synapses [7].
References
[1] G. Servalli, “A 45nm generation phase change memory technology”, IEDM Tech. Dig., pp. 5.7.1-5.7.4, 2009.
[2] R. Bez, “Chalgonide PCM: a memory technology for next decade”, IEDM Tech. Dig., pp. 5.1.1-5.1.4, 2009.
[3] J.H.Oh,J.H.Park,Y.S.Lim,H.S.Lim,Y.T.Oh,J.S.Kim,J.M. Shin, Y. J. Song, K. C. Ryoo, D. W. Lim, S. S. Park, J. I. Kim, J. H. Kim, J. Yu, F. Yeung, C. W. Jeong, J. H. Kong, D. H. Kang, G. H. Koh, G. T. Jeong, H. S. Jeong, K. Kim, “Full integration of highly manufacturable 512Mb PRAM based on 90nm technology”, IEDM Tech. Dig., pp 49-53, 2006.
[4] S. Lai, “Current status of the phase change memory and its future”, IEDM Tech. Dig., pp. 255-258, 2003.
[5] D. Kuzum, R. J. D. Jeyasingh, B. Lee, H.-S. P. Wong, “Nanoelectronic Programmable Synapses Based on Phase Change Materials for Brain-Inspired Computing”, Nano Letters, pp. 2179-2652, 2012.
[6] D. Kuzum, R. J. D. Jeyasingh, H.-S. P. Wong, “Energy efficient programming of nanoelectronic synaptic devices for large-scale implementation of associative and temporal sequence learning”, IEDM Tech. Dig., pp. 30.3.1.-30.3.4, 2011.
[7] D. Kuzum, R. J. D. Jeyasingh, S. Yu, H.-S. P. Wong, “Low Energy, Robust Neuromorphic Computation Using Synaptic Devices”, IEEE Trans. Elec. Dev., in press

To date the main applications of phase-change materials and devices have been limited to the provision of non-volatile memories. Recently however the potential has been demonstrated for using phase-change devices as the basis for new forms of computing, by exploiting their multi-level resistance capability to provide electronic mimics of biological synapses [1,2]. Here we exploit a different and previously under-explored property also intrinsic to phase-change materials and devices, namely accumulation, to demonstrate that phase-change devices can also be used to implement a simple form of integrate-and-fire neuron. Thus, we might imagine 'all phase-change' based neural networks or neuromorphic processors [3]. Furthermore, the accumulation regime can also be used to carry out arithmetic computing and perform Boolean logic, leading potentially to a new type of phase-change based arithmetic logic unit (ALU) for general purpose computing. A major difference between such a phase-change ALU and conventional silicon processors is that the phase-change system would be non-volatile; if the power were to be removed from the phase-change system it would remain in its pre-existing state, and processing could recommence from where it left off when power is re-supplied. Moreover, since the phase-change ALU both computes (processes) and stores the result of such computations simultaneously and at the same physical location, the detrimental von Neumann 'bottleneck' inherent to conventional computing systems is avoided [4]. Finally, we note that both the neuromorphic and ALU computing capabilities of phase-change materials and devices are accessible in both the optical (photonic) and the electrical (electronic) domains, or indeed via a 'mixed-mode' approach. This opens up the route towards various attractive non-von-Neumann phase-change computing possibilities. For example, we might implement 'all-photonic computing' that avoids the well-known high-frequency limitations of electrical buses [5]. Alternatively we could used a mixed optical-electrical approach, in which the ultra-fast optical switching capabilities of phase-change materials [6] is used to provide high-speed computing operation, but with data output (sensing) being in the electrical domain. In this paper we discuss such novel approaches to phase-change based neuromorphic and general purpose computing and present experimental proof-of-principal of some of the underlying concepts.
[1] D. Kuzum, R.G.D. Jeyasingh, B. Lee, H-S. P. Wong, Nano Lett., 12, 2179, 2012
[2] M. Suri, V. Sousa, L. Perniola, D. Vuillaume, B. DeSalvo, Int. Conf. on Neural
Networks (IJCNN), 619 (doi: 10.1109/IJCNN.2011.6033278), 2011
[3] C.D. Wright, Y. Liu, K.I. Kohary, M.M. Aziz, R.J. Hicken, Adv. Mater. 23, 3408, 2011
[4] C.D. Wright, P. Hosseini, J. A. Vazquez Diosdado, Adv. Func. Mater., doi:10.1002/adfm.201202383, 2012
[5] W.H.P. Pernice and H. Bhaskaran, Appl. Phys. Lett. 101, 171101, 2012
[6] J. Siegel , A. Schropp , J. Solis , C. N. Afonso , M. Wuttig , Appl. Phys. Lett. 84, 2250, 2004

Phase change materials, already largely investigated for the realization of rewritable optical disks and nonvolatile memories, are also good candidates for reconfigurable electronic components, for example in analog circuits. In particular, for high precision analog devices, such as operational amplifiers or voltage references, very precise high resistance values, stable over a large temperature range, are required. Thin film resistors, usually Si:Cr based alloys, with thickness of only a few nanometers, are commonly employed for this purpose, since they are characterized by low temperature coefficient of resistance (<100 ppm/K). To obtain high precise resistance values a trimming operation at the end of manufacturing can be required. This is usually done through the use of an appropriate laser (laser trimming) or by other electrical methods, such as electrical fusing, self-heating or heater resistors, offering the advantage to adjust also shifts due to packaging.
Ge2Sb2Te5 exhibits a polymorphism with a metastable fcc phase, which converts into the stable hcp structure at temperature around 400°C. The first formed phase (fcc) is characterized by a decrease of resistance versus temperature, typical of semiconductor materials, while the resistance of hcp phase increases with increasing temperature, thus behaving like a metal. Thanks to this property, mixed phase Ge2Sb2Te5 thin films, containing both the metastable polycrystalline fcc phase and the stable hcp phase could be employed to obtain resistors with temperature coefficient of resistance close to zero.
In order to investigate this aspect, the electrical properties of Ge2Sb2Te5 films deposited in the amorphous phase have been studied through in situ four point probe resistance measurements during annealing. Mixed phase films containing both the fcc and the hcp phase have been obtained under several annealing conditions (ramp rate, maximum annealing temperature and annealing time). After the annealing, the temperature dependence of the resistance in the mixed phase samples has been measured in the temperature range 300-400K. The structure of the material has been investigated through X ray diffraction.
The reconfigurability properties of Ge2Sb2Te5 films have also been investigated by applying electric pulses to patterned resistors with optimized layout.
The results are promising and suggest Ge2Sb2Te5 as a valuable material for manufacturing of high precision reconfigurable resistors with high sheet resistance values, far more viable from a point of view of the device fabrication, compared to thin film resistors.

A multi-contact device based on the phase-change material Ge2Sb2Te5 (GST) is simulated to study its potential as a reconfigurable switch. COMSOL multi-physics is used to simulate the proposed device. The structure is a rectangular shaped GST patch of 100 nm x 50 nm x 20 nm with a TiN contact/heater on each side. The whole structure is surrounded by several nanometers of silicon dioxide for electrical and thermal insulation. Each of the four contacts is controlled with a series MOSFET. Using appropriate pulsing schemes one or more contacts can be isolated by amorphizing small volumes of GST around the contacts, enabling or disabling connectivity between different contacts. The details and results of the simulations and possible device implementations will be presented.

The density advantage, power-efficiency, and non-volatility of PCRAM provide increasing attractions to be used for today&’s electronic systems and the studies at architectural level. However, at the device level, there are still some technical hurdles need to be solved to meet the different application requirements, especially the high RESET current. To reduce current, besides scaling down the cell size, the reported methods are mainly focused on reducing the heat loss, and increasing the heating efficiency. In this talk, the high current issue of PCRAM will be tackled from both material and carrier transportation aspects. Artificial structures were implemented to manipulate the materials&’ properties in order to minimize the heat loss from the surrounding dielectric and contact electrodes. An asymmetric electrode structure was proposed to have reduced thermal conductivity while still maintaining the high electrical conductivity. Success RESET current reduction has been demonstrated by both simulation and experiments. Moreover, a different carrier injection mechanism was proposed and implemented to further reduce the current. More details will be presented at the conference.

Chalcogenide based phase-change memory (PCM) utilizing Ge2Sb2Te5 (GST) has the potential to replace DRAM and Flash memory in future electronics [1], but current understanding of power dissipation and thermal transport in PCM is incomplete. The influence of thermal and electrical boundary resistance and thermoelectric effects at PCM contacts are expected to decrease PCM energy consumption [2]; however, these effects are also poorly understood.
We present the first study of nanometer-scale temperature mapping near GST-TiW contacts. Thermometry is achieved by scanning Joule expansion microscopy (SJEM), which measures temperature with sub-50 nm spatial resolution and sub-100 mK temperature resolution [3,4]. Lateral PCM devices were fabricated on SiO2/Si substrates by sputtering of 25 nm of GST between 50 nm thick TiW contacts. Devices were encapsulated by 10 nm SiO2 for stability during testing. The GST was crystallized by heating on a hot plate at 200 degrees Celsius for five minutes. The lateral GST devices had 1.5-10 mu;m lengths and a fixed 300 mu;m width. Fabrication was completed by covering the devices with 70 nm of poly(methyl methacrylate) (PMMA), whose thermal expansion acts to amplify the thermomechanical expansion signal in the SJEM measurements [3].
Temperature profiles of the GST were measured with SJEM at various forward and reverse biases and were understood by comparison to an electrical-thermal-thermoelectric finite element analysis (FEA) model. The observed temperature fields show heating due to Joule effects in the GST and due to interfacial resistance at GST-TiW contacts. Comparing our SJEM experiments with FEA simulations, we estimate an electrical interface resistance of ~25 kOmega;-mu;m2 and a current transfer length of ~1 mu;m at GST-TiW contacts. We observe a clear asymmetry of the temperature profile at the contacts with respect to the direction of current flow, due to Peltier heating and cooling at the GST-TiW contacts. Fitting our simulations to measurements we extract a Seebeck coefficient of ~200 mu;V/K for 25 nm thick crystalline GST. Our combined measurement-modeling study represents the first direct observation of nanometer-scale Joule heating and thermoelectric effects at GST-metal contacts.
[1] G.W. Burr et al, Journal of Vacuum Science & Technology B, 28, 223 (2010).
[2] J. Lee et al, Nanotechnology, 23, 205201 (2012).
[3] J. Varesi et al, Applied Physics Letters, 72, 37 (1997).
[4] K. L. Grosse et al, Nature Nanotechnology, 6, 287 (2011).

Phase change materials (PCM) are a class of compounds that can be switched reversibly and on a nanosecond timescale between the amorphous and crystalline state. Since these two states differ significantly in their bonding mechanism the contrast in physical properties such as optical reflectivity or electrical conductivity is very pronounced. This generic feature makes PCM one of the most promising materials for emerging non-volatile data storage applications.
The insight in the bonding scheme of the crystalline phase led to a first map of PCMs that successfully predicts the appearance of resonance bonding [1]. While this map enables the identification of alloys with a pronounced optical contrast between the crystalline and amorphous state, the question arises how to predict transport properties which are also crucial for applications of phase change materials.
Therefore, this study focuses on the thermal transport properties of alloys along the pseudo-binary line between GeTe and Sb2Te3. The thermal conductivity κ of several PCM thin films was measured as a function of stoichiometry and annealing temperature. A clear trend for the dependence of thermal conductivity upon stoichiometry is observed which will be discussed and explained.
[1] Lencer, D. et al. A map of phase-change materials. Nature Materials 7, 972-977 (2008).

Chalcogenide phase change material Ge2Sb2Te5 (GST) usually has two crystalline phases that exhibit different electrical resistance and fast phase transition kinetics. The one is amorphous and the other is crystalline. Phase Change Memory (PCM) is a unique nonvolatile memory that has write and erase times of tens of nanoseconds. Currently, GST is widely used phase change material for PCM applications.
In the development of high density PCM below 50-nm technical nodes, GST Chemical Mechanical Planarization (CMP) process is highly recommended because it has a smooth surface and high scalability. In recent years, many researchers have studied GST slurry [1-5], but they show only laboratorial results not real polishing properties in device manufacturing.
In this study, we tried to investigate the polished surface with 300mm ReflexionLK polisher and IC1010-K GRV. The surface defect was measured and verified in various slurries. Also the surfaces damaged layer was observed with a high resolution transmission electron microscopy (HR-TEM), an auger electron spectroscopy (AES) and Atomic Force Microscopy (AFM).
We found that the alumina particle slurry has more defects than the colloidal silica one and increasing hydro peroxide concentration decrease the roughness. The surface damaged layer also changed with increasing hydro peroxide percentage. The surface damaged layer is broadening resistance distribution, so we need to control the damage layer minimized.

In July 2012 a milestone has been put in the history of semiconductor based non-volatile memories. For the first time a 1 Gigabit product, based on 45 nm technology, has been put on the market in volume production by Micron. The stand-alone memory chip, released for wireless applications, will be included in mobile phone and will offer better performance and lower cost with respect to the state of the art conventional architectures. Aim of the talk is to describe the main technology features focusing on chip architecture, process challenges and specifically developed innovative devices. The developed chip is compatible with the NOR specification enabling the random access to the memory with a reduced latency time of 85 ns, increased read and program throughput with respect to the conventional NOR based solutions. To achieve the goal an innovative bipolar selector has been fabricated, coupled with the so called “wall” cell architecture for the chalcogenide based storage element. The developed cell architecture allows to properly tune the programming current of the cell thanks to the self-aligned heater that enables a reduced active volume involved in the phase transition. An outlook to the future will be also given: a possible path for scaling of the PCM will be illustrated. Chalcogenide material engineering will be presented as a viable way to improve chip performance and enlarge the application spectrum of PCM enabling new products. Finally innovative concept such as the chalcogenide superlattice, will be also discussed as a possible solution to overcome specific PCM limitations that would enable PCM as a “universal memory”.

Phase change memory (PCM) strongly entered the market of high density, fast and non-volatile memory and is expected to replace the current memory technologies. Ge₂Sb₂Te₅ (GST) is the most commonly studied phase change material due to the high resistance contrast between the amorphous and crystalline phases. PCM devices are resistive memory elements where the crystalline (set) and amorphous (reset) states typically have up to four orders of magnitude difference in resistance values. The transitions between these states are achieved through localized self heating with large current densities. Scaling device dimensions improves packing density, speed as well as peak current, power, and total energy requirements for switching. Recent experimental studies have shown that the current confinement can further be enhanced by integrating a thin oxide atop the heater, where the oxide layer is ruptured with application of an initialization pulse, forming a narrow filament in the oxide. This nano-scale filament can be smaller than what is lithographically achievable; resulting in reduced reset current due to smaller active region size and better heat confinement.
In this work, 2D rotational symmetric simulations using COMSOL Multiphysics are utilized to assess the rupture oxide cell with an nFET access device. The joule heating and electric current modules were coupled such that the current continuity and heat transport equations are solved self-consistently. The rupture oxide is modeled as a 3 nm thick SiO2 layer separating the TiN bottom contact and the GST film with a filament in the center with electrical and thermal properties of TiN. The filaments&’ diameter, the width of the transistor, and the supply and gate voltages were varied to study the cell performance.
Rupture oxide cells require less operating power and smaller transistor width compared to the conventional mushroom cells. The behavior of the rupture oxide cell significantly depends on the diameter and resistivity of the filaments formed by the rupturing process.

Phase change material is the heart of phase-change memory (PCM) technology. The development of phase-change materials for PCM has relied on materials originally developed for phase-change rewritable optical storage in the past two decades. Today almost all phase change memory IC&’s still use Ge2Sb2Te5 (GST-225) inherited from optical disk technology, even though high reset current and poor data retention at elevated temperature make it difficult for new applications such as automotive. Many material modifications were made trying to solve these issues however, conflicting material properties between switching speed and thermal stability still persisted. In addition, large programming current is also a key issue that limits the adoption of PCM for advanced applications. In this study, we will discuss how materials engineering and electrode architecture can help make high performance PCM possible.
Firstly, we introduce a special group of Ge1SbxTe1 (GST-1x1) materials located along an isoelectronic tie line with equal amounts of Ge and Te. We selected the best compromise material (GST-212) on the isoelectronic tie line and further enhanced its high temperature stability by moving along a second tie line between Ge and Sb2Te3, taking advantage of the assumption that higher Ge concentration should increase the thermal stability. The “golden” composition was successfully found by maneuvering between these two tie-lines and the promising properties of this new material are demonstrated both at wafer level and in packaged chips [1].
Next, the Ge/N concentration of the “golden” composition is further engineered to meet requirements for automotive application. A laser melt-quenching method is adopted that provides fast turn around retention data on blanket films which are highly predictive for device results. The optimized material demonstrated excellent retention in a 256 Mb test chip with projected 10-year retention at 120 oC, suitable for industrial and some automotive (in-cabin) application [2].
Finally, bottom electrode (BE) engineering for better power efficiency is adopted. The new thermally confined electrode with TaN/TiN/TaN BE has demonstrated 30 uA reset current, representing a 90% reduction compared with solid TiN electrode. The low reset current also improves the reliability and cycling endurance. This dramatic reduction in switching power enables the scaling of the selection device, which in turn allows the scaling of the cell size.
References
[1] H. Y. Cheng, T. H. Hsu, S. Raoux, J.Y. Wu, P. Y. Du, M. BrightSky, Y. Zhu, E. K. Lai, E. Joseph, S. Mittal, R. Cheek, A. Schrott, S. C. Lai, H. L. Lung and C. H. Lam, Tech. Dig.- Int. Electron Devices Meet., 3.4 (2011).
[2] H. Y. Cheng, J. Y. Wu, R. Cheek, S. Raoux, M. BrightSky, D. Garbin, S. Kim, T. H. Hsu, Y. Zhu, E. K. Lai, E. Joseph, A. Schrott, S. C. Lai, A. Ray, H. L. Lung and C.H. Lam, Tech. Dig.- Int. Electron Devices Meet., 31.1 (2012).

Phase change materials (PCMs) are being studied for non-volatile memory applications due to their scalability, fast switching and low power at small dimensions [1]. Recently, performance of nanoscale PCM devices has been examined by confining their bit either as nanowires, or by contacting the PCM with carbon nanotube electrodes [2]. While such structures are useful for scalability and performance analysis, their large-scale integration remains challenging.
In this work, we present the first study of low-power PCM devices with patterned graphene electrodes for wafer-scale integration. The thin structure of these devices (with thin PCM and graphene layers) makes them ideal for flexible and transparent electronics which have strict low-power requirements. In addition, graphene interconnects can also be integrated with conventional CMOS substrates. Our devices switch at threshold voltages as low as ~3 V with low programming currents (<1 mu;A set, <10 mu;A reset) and excellent on/off ratios (>100), enabled by the sharp contact area with their atomically thin graphene electrodes.
Graphene was grown by chemical vapor deposition (CVD) on copper and then transferred to SiO₂/Si substrates [3]. We demonstrate how interconnects ranging from monolayer to several graphene layers can be integrated in this way, depending on circuit requirements. Graphene ribbon electrodes are shaped by lithography and O₂ plasma etching, contacted by Au/Ti metal pads for testing. Memory bits are then defined by e-beam lithography, Ge₂Sb₂Te₅ (GST) sputtering and lift-off. Devices were encapsulated by thin SiO₂ (~10 nm) for stability during testing. GST bit dimensions range from 20-100 nm long and wide by ~10 nm thick.
Electrical measurements, atomic force microscopy and Raman spectroscopy confirm good quality and uniformity of the graphene electrodes. Estimated minimum contact resistivities between graphene and GST in the amorphous and crystalline states are 50 and 0.2 kOmega;mu;m² respectively, comparable to values reported for other GST contacts (e.g. with TiW [4]). Our comprehensive analysis of the first GST-based memories with graphene electrodes is important for future large-scale integration of very low power memory devices. In addition, the techniques demonstrated can also be used for fabrication of other types of memories (e.g. resistive random access memory) with graphene electrodes.
[1] H.-S. P. Wong, et al., Proc. IEEE 98, 2201 (2010); D. Loke, et al., Science 336, 1566 (2012).
[2] F. Xiong, et al., Science 98, 206805 (2011); D. Yu, et al., Nano Lett. 8, 3429 (2008).
[3] A. Behnam, et al., Nano Lett. 12, 4424 (2012).
[4] D. Roy, et al., IEEE Electron Dev. Lett. 31, 1293 (2010); E.K. Chua, et al., Appl. Phys. Lett. 101, 012107 (2012).

The doping of crystalline semiconductors, in particular Si, has proven to be the key technological step that underpins the majority of today&’s electronic technologies. Of all the effects observed, the ability to control the electronic properties of these materials, providing n-type, p-type conducting and insulating regions via ion-implantation, has revolutionised manufacturing and enabled Moore&’s law to continue to be held. Ion-implantation continues to provide new opportunities for technological advances in microelectronics, for example, such methods can also be used to stabilize or activate specific interactions within the materials within localized regions.
Historically, the use of ion-implantation into chalcogenides has been focused on the formation of optical waveguides, rare-earth doping, and the formation of metal (nano) colloids as a means of increasing the non-linear optical response. It is also well established that high-dose ion implantation can be used to form nanoclusters in amorphous materials. The introduction of such clusters has been utilised in ReRAM devices, most notably the metal-chalcogenide memristors take advantage of metal ions through the chalcogenide glass lattice.
In this work, we report first on ion implantation of a broad range of elements into chalcogenide thin films spanning s sulphides, selenides and tellurides. The properties of these films are investigated pre and post implantation. Second, targeting the most promising dopants and chalcogenide compounds, we describe the design, fabrication and characterisation of a series of ion implanted amorphous devices in the Ge:Se family of glasses. The diodes produced show good rectification while at higher electric field exhibiting memory switching behaviour. This suggests the possibility of unique devices exhibit, rectification along with a controlled asymmetric polarity dependant behaviour, which shows great promise in realising next generation synaptic devices.
We believe that through the ion implantation process, in selected chalcogenide materials, a low cost production line method of producing integrated diode/memory cells with next generation cognitive information processing capability can be realised for use in future cross bar array architectures.

The phase change material (PCM) based device is promising to become next-generation main-stream non-volatile memory technology. But its ultimate scaling limit is still unknown. Recent experiments have proven that sub-2 nm PCM nanostructures still keep phase change properties, making the write operations possible in ultra-scaled dimension. However, until now it is still an open question that how small the PCM nanostructures can be scaled, yet still retain adequate ON/OFF ratio for read operations.
State-of-art scaling experiments have shown that 2 orders of magnitude ON/OFF ratio can be maintained if the PCM ultrathin films is scaled down to 6 nm. While sub-6 nm experimental data is not available, here we compute it using purely ab initio simulations. In the simulations, ab initio molecular dynamics, density functional theory, and Green&’s function algorithms are applied. The crystalline GeTe (c-GeTe) and amorphous GeTe (a-GeTe) ultrathin films are sandwiched by the TiN electrodes. Based on the simulation results we point out, for the first time according to our knowledge, that the aggressive scaling might be limited by the loss of adequate ON/OFF ratio, which makes it difficult to reliably perform read operation.
We observed that that the a-GeTe ultrathin films scaled down to about 38 Å (12 atomic layers) still show band gaps and the electrical conductance is mainly due to the electron transport via intra-gap states. If the ultrathin films are further scaled, the a-GeTe band gap disappears due to overlap of the two metal induced gap states (MIGS) regions near the TiN electrodes, leading to sharp increase of a-GeTe conductance and significant decrease of c-GeTe/a-GeTe conductance contrast (i.e. ON/OFF ratio). As a result, the ON/OFF ratio drops below ten if the ultrathin films are scaled below about 33 Å, making it difficult to reliably perform read operations. This sets up an ultimate scaling limit of phase change memory technology. Our results suggest that this ultimate scaling limit can be pushed to even smaller size, by using PCM with larger amorphous phase band gap than a-GeTe.
Reference:
[1]. Jie Liu and M. P. Anantram, “Low-bias electron transport properties of germanium telluride ultrathin films”, (submitted to Phys. Rev. B) http://www.ee.washington.edu/faculty/anant/publications/JieLiuPaper.pdf
[2]. G. E. Ghezzi, “Effect of carbon doping on the structure of amorphous GeTe phase change material”, Appl. Phys. Lett. 99, 151906 (2011).
[3]. L. L. Chang, P. J. Stiles, and L. Esaki, “Electron Barriers in Al-Al2O3-SnTe and Al-Al2O3-GeTe Tunnel Junctions”, IBM J. of Res. and Dev. 10, 484 (1966).

A range of material systems exist in which nanoscale ionic transport and redox reactions provide the essential for switching as platform for reconfigurable electronic devices and biological like computing. One class relies on mobile cations, which are easily created by electrochemical oxidation of the corresponding electrode metal, transported in the insulating layer, and reduced at the inert counter electrode. These devices are termed electrochemical metallization memories (EMC) or conductive bridge random access memories [1]. The material candidates for electrolytes in such devides have been recently studied. They are amorphous chalcogenides [2, 3] and also oxides (SiO2, WO3, TiO2 and others [1]) containing metal elements (Ag, Cu) or their compounds (Ag2S, CuS) and gaining some portion of ionic conductivity and becoming mixed ionic-electronic conductors [3-7].
The aim of this work is to present our current results on syntesis and resistive switching of chalcogenide based nanowire array cells.
The authors thanks to project CZ.1.07/2.3.00/20.00254 “Research Team for Advanced Non-crystalline Materials" realized by European Social Fund and Ministry of Education, Youth and Sports of The Czech Republic within The Education for Competitiveness Operational Programme for financial support.
[1] W. Lu, D. S. Jeong, M. Kozicki, R. Waser, MRS Bulletin, 37 (2012) 124.
[2] M. Frumar and T. Wagner, Curr. Opinion Solid St. Mat. Sci. 7 (2003) 117.
[3] M. Frumar, B. Frumarova, T. Wágner, Amorphous and Glassy Semiconducting Chalcogenides. In: Bhattacharya P, Fornari R, and Kamimura H, (eds.), Comprehensive Semiconductor Science and Technology, volume 4, pp. 206-261 (2011)Amsterdam: Elsevier.
[4] S. Stehlík, J. Kolár, M. Bartoscaron;, Mil.Vl#269;ek, M. Frumar, V. Zima, T. Wágner, Sol.
State Ionics 181, (2010) 1625.
[5] J. Kolár, T. Wágner, V. Zima, S. Stehlík, B. Frumarová, L. Benescaron;, Mil. Vl#269;ek, M.
Frumar, J. Non-Cryst. Solids, 357 (2011) 2223.
[6] I. Kaban, P. Joacute;vári, T. Wágner, M. Bartoscaron;, M. Frumar, B. Beuneu, W. Hoyer, N.
Mattern, J., Eckert, J. Non-Cryst. Solids, 357 (2011) 3430.
[7] Scaron;. Stehlik, K. Shimakawa, T. Wágner, M. Frumar, J. Phys. D: Appl. Phys. 45
(2012) 205304.

Understanding charge transport in phase change materials is crucial to extend the application range of these exciting materials. The reduction of drift, i.e. the increase of the resistivity of the amorphous phase with time, for example, would facilitate the development of multilevel memories. Tailoring the resistance of the crystalline state could help in reducing the power consumption upon amorphization. With this goal in mind we have studied the resistivity of crystalline phase change materials. A pronounced dependence of the room temperature resistivity upon annealing temperature is observed for crystalline phase change materials such as Ge1Sb2Te4 or Ge2Sb2Te5. This finding is corroborated by low temperature measurements as well as FTIR data, which confirm that a metal - insulator transition is observed without a change in crystallographic state. This is indicative for an electronically driven MIT [1].
Such an MIT can be achieved if the electron correlation exceeds a critical value (Mott MIT). A second route to insulating behavior has been identified by Anderson, who showed that increasing disorder turns a metal with delocalized electronic states at the Fermi energy into an insulator with localized states. In this talk, arguments for a disorder induced localization of charge carriers will be presented. The observations are compared with doped semiconductors such as Si:P, where both disorder and correlations are crucial to describe the charge transport. Experimental and theoretical attempts to unravel the origin of disorder induced localization will be presented. These calculations reveal that it is the ordering of vacancies into vacancy layers which drives the transition to the metallic state [2]. The potential of this remarkable impact of disorder for applications as well as our fundamental understanding of solids is discussed.
[1] T. Siegrist et al., Nature Materials 10, 202, (2011)
[2] W. Zhang et al., Nature Materials, 11, 952, (2012)

It was recently found that phase change memory (PCM) composed of crystalline superlattice using a GeTe/Sb2Te3 stack, which is called interfacial phase-change memory (iPCM) [1], has a huge magnetoresistance at room temperature without a magnetic dopant [2]. In addition, IPCM has other interesting properties under a magnetic field: optical reflection responding to a circular polarized light is discriminated by the magnetic polarity change; the spin polarized state can be stored as spin memory like magnetic random access memory, MRAM. Although it has been speculated that all these unusual magnetisms are attributed to a topological insulating property emerged from the Sb2Te3 sub-layer [3, 4], more theoretical and experimental works are necessary to make clear the mechanism hidden in the structure. IPCM is different from a usual TI on the pint of view of the helical spin state generated at the interface between the heterogeneous solid-solid layers, which makes hard to understand the mechanism. In this paper, we are going to discuss about the iPCM properties from the point of view of the unique band structures and device measurement.
[1] R. Simpson, P. Fons, A. V. Kolobov, M. Krbal, and J. Tominaga, Interfacial phase-change memory. Nature Nano. 6, 501-505 (2011). [2] J. Tominaga, R. Simpson, P. Fons, A. V. Kolobov, Electrical-field induced giant magnetoresistivity in (non-magnetic) phase change films. Appl. Phys. Lett. 99, 152105 (2011). [3] J. Kim, J. Kim, K. Kim, and S. Jhi, Topological Phase Transition in the Interaction of Surface Dirac Fermions in Heterostructures, Phy. Rev. Lett. 109, 146601 (2012). [4] B. Sa, J. Zhou, Z. Sun, J. Tominaga, and R. Ahuja, Topological Insulating in GeTe/Sb2Te3 Phase-Change Superlattice, Phys. Rev. Lett. 109, 096802 (2012). A part of the work was supported by FIRST program initiated by the Council for Science and Technology Policy (CSTP).

In topological insulators, the bulk is insulating while their surfaces support gapless surface states. They are characterized by the topological numbers calculated from the band-structure calculations, and if the topological numbers turn out to be nontrivial, the existence of gapless surface states is guaranteed. The surface states typically form Dirac cones, which have linear dispersions. There are various topological insulator materials found so far, among which are Bi2Te3, Bi2Se3, and Sb2Te3.
On the other hand, the superlattices between Sb2Te3 and GeTe, i.e. the multilayer materials [(GeTe)_l(Sb2Te3)m], are demonstrated to behave as interfacial phase-change memory. Therefore it is of potential interest to show possible relationships between physics of topological insulators and interfacial phase-change memory. In my presentation I give brief introduction on topological insulators, followed by our recent theoretical results on surface states of topological insulators and their transport properties. We then demonstrate how these properties can be related to the interfacial phase-change memory and discuss theoretical interpretations on the experiments by Tominaga et al.

Innovative phase-change memory based on sputtered Sb2Te3/GeTe superlattices (SLs) exhibit reduced switching energies, longer write-erase cycle lifetimes and faster switching speeds than conventional data storage devices employing Ge2Sb2Te5 (GST) alloys. The superior electrical characteristics are attributed to the realization of fiber-like rather than randomly oriented phase-change materials (PCMs), i.e., the larger structural order of the multi-layered chalcogenides limits atomic motion, and hence entropic losses, associated with the switching process [1]. Compared to physical vapor deposition (PVD), molecular beam epitaxy (MBE) is anticipated to improve the PCMs&’ structural perfection, allowing for tight control of layer thickness and interface quality. In this perspective, chalcogenide thin films with composition along the GeTe-Sb2Te3 pseudo-binary line were grown in a MBE chamber mounted on a six-circle diffractometer at the BESSY II storage ring in Berlin, and their structural properties studied by in vivo synchrotron radiation grazing incident x-ray diffraction (SR-GIXRD). GST epitaxial layers (ELs) were successfully grown in the metastable cubic phase on both slightly (GaSb) and highly (Si) lattice-mismatched templates. High structural film quality could be achieved on (111) surfaces, independently of the lattice misfit, owing to the tendency of GST to form layered structures in the stable bulk phase and the presence of vacancies in the metastable cubic phase [2]. In particular, it was demonstrated that GST ELs on Si(111) can be reversibly switched between crystalline and amorphous states over a large area using femtosecond laser pulses, and that the re-crystallized material recovers its as-deposited single crystalline nature [3]. The epitaxy on Si(111) substrates of GeTe and Sb2Te3 was also studied. Thin layers of the former material grow (111)-oriented in a rhombohedrally distorted rocksalt structure [4]. The latter binary compound grows in a rhombohedral layered structure (space group R-3m), exclusively c-plane-oriented. Sb2Te3 is a three-dimensional topological insulator. The weak antilocalization effect in thin Sb2Te3 films indicates that the two topological surface states are robustly maintained [5], attesting to the excellent quality of the MBE-grown layers. Finally, our know-how on the growth of binary compounds was exploited to fabricate epitaxial Sb2Te3/GeTe SLs on Si(111). SR-GIXRD showed that the superlattice heterostructures grown by MBE possess a preferential orientation also in-plane, anticipating improved electrical characteristics with respect to the weakly-textured PCM fibers deposited by PVD.
[1] R.E. Simpson et al., Nat. Nanotechnol. 6, 501 (2011).
[2] P. Rodenbach et al., Phys. Status Solidi RRL (2012), doi: 10.1002/pssr.201206387.
[3] P. Rodenbach, et al., Appl. Phys. Lett. 101, 061903 (2012).
[4] A. Giussani et al., Phys. Status Solidi B 249, 1939 (2012).
[5] Y. Takagaki et al., Phys. Rev. B 86, 125137 (2012).

Chalcogenide materials, in particular those based on Te and Se compounds, having exceptional structural, optical and electrical properties are intensively studied as phase change material and thermoelectric devices. More recently, the topological insulator behavior of these materials opened new application possibilities in spintronics and quantum computing. In order to control the behaviors of these materials, different preparation methods, alloying and doping have been investigated. The high bulk carrier concentration, usually observed in these materials can dramatically influence the properties of the films. It has recently been shown, that the bulk carrier concentration in a single crystal obtained by slow cooling of a melt can be tuned by adjusting the composition of Bi2(1-x)Sb2xTe3, i.e the conductivity type changes from n-Bi2Te3 to p-Sb2Te3 films. In this paper we present studies on Bi2(1-x)Sb2xTe3 epitaxialy grown thin films on Si(111) substrates. The layers were obtained by MBE using similar optimized conditions to those described in Ref. [2], but exploring the complete alloying x range. The morphology, composition and crystalline structure have been studied using AFM, SEM, XRD, Raman spectroscopy and XPS measurements. The robust topological surface behavior of the films has been already demonstrated [3]. For all investigated samples, the deposition temperature and Te flux were constant. The composition was changed by changing the Sb/Bi flux ratio. As found by AFM, the layers show atomic QL steps of about 1 nm thickness, associated with plateaus of 200 - 500 nm width. The roughness of the plateaus was found at the limit of our AFM sensitivity of ~0.1nm from most samples, but some samples showed values of 0.2 - 0,4 nm. The crystalline quality and c-axis orientation perpendicular to the substrate was proved by XRD measurements. In Raman spectra, two main peaks corresponding to Eg2 and A1g2 phonon modes have been detected. The peaks show a blue shift by increasing the Sb concentration (maximum range of ~10 cm-1 and ~32 cm-1 for Eg2 and A1g2, respectively). No splitting of the modes was observed, demonstrating that a solid solution is obtained. The composition of the films was evaluated using linear interpolation of the A1g2 peak energy. The values are in good agreement with XPS results. Comparative studies on Sb2Te3 epitaxialy grown on Si(111) using MOCVD method are also presented. In this case, by adjusting the growth conditions, compact layers were obtained by coalescence of flat hexagonal nano-platelets. Specific Sb2Te3 Raman modes were clearly detected.
1. D. Kong, et al., Nature Nanotechnology, 6, 705 (2011)
2. J. Krumrain, et al., Journal of Crystal Growth 324, 115 (2011)
3. L. Plucinski, et al., Appl. Phys. Lett. 98, 222503 (2011)

Miniaturization, nowadays down to the nanoscale, plays a crucial part in the design of novel materials. Among the latter, the phase-change alloys GeTe and Ge_xSb_yTe_z (“GST”) are of undisputed importance, with compounds like Ge₂Sb₂Te₅ or Ge₈Sb₂Te₁₁ being used in DVD-RAM and re-writeable Blu-Ray disks, respectively [1]. Very recently, various groups reported preparation of precisely defined nanocrystalline phases of GeTe, as well as of GeTe and Ge₂Sb₂Te₅ nanowires with promising prospects for future data-storage applications. Despite such exciting developments, a number of downright fundamental physical and chemical properties within the GST family are not yet completely understood. On the side of quantum-theoretical simulations, much effort has been devoted to modeling the bulk (infinitely expanded) phases, and the last years have seen many questions answered. In the future, however, it is clear that new computational approaches (beyond the “bulk”) will be required to properly model phase-change material surfaces and nanostructures.
Here, we present results of first-principles electronic-structure computations (DFT, PW/PAW technique) for cornerstone phase-change material surfaces from the above-mentioned family. We start with the simplest binary compound, namely, α-GeTe(111) [2]. Interestingly, Ge-terminated surfaces are predicted to stabilize themselves through various reconstruction mechanisms, whereas Te terminated surfaces do not “need” the latter due to lower surface energy. The formation energy of Ge vacancies below the surface is evaluated and compared to the bulk.
Based on these results, we move on to model surfaces of cornerstone ternary “GST” phase-change materials, namely, the stable hexagonal phases GeSb₂Te₄, Ge₂Sb₂Te₅ and GeSb₄Te₇. General trends in surface energies are discussed beyond these explicit compositions. We also assess the influence of dispersion-corrected DFT on predicting surface stabilities of these “van der Waals” type systems—a technical but important aspect also beyond GST.
Finally, we present results [3] for the metastable crystalline phase Ge₂Sb₂Te₅, whose “rocksalt-like” structure has been under vivid debate recently. Structural peculiarities are discussed—in particular, differences between the bulk and the “nanocrystalline” phase, the latter of which is modeled by our computations.
References
[1] M. Wuttig, N. Yamada, Nature Mater. 6, 824 (2007).
[2] V. L. Deringer, M. Lumeij, R. Dronskowski, J. Phys. Chem. C 116, 15801 (2012).
[3] V. L. Deringer, R. Dronskowski, Phys. Rev. Lett., submitted.

The defect-free scaling down of phase change memory (PCM) devices is great importance to realize high performance and low power devices. A way to sensibly lower both the active volume to be programmed and the contact size is offered by the self-assembled nanowire (NW) option, obtainable by chemical deposition methods.
Among the deposition methods, the Metal Organic Chemical Vapor Deposition (MOCVD) is a very attractive technology for growing nanostructures, because it allows process control (including rate, composition and uniformity) and easy industrial transfer, notwithstanding the difficulty due to the very narrow parameter windows for NW self-assembly.
After an overview of different approaches employed to chemically synthesize other phase change, chalcogenide NWs, from the use of catalyst- or catalyst-free methods, to the growth of core-shell NWs for multi-level PCM devices, in this presentation recent and unique results in the MOCVD self-assembly of Ge-Sb-Te NWs by the Au-catalyzed vapor-liquid-solid (VLS) mode will be illustrated [1]. In particular, the growth study and functional analysis will cover single NWs of GeTe, Ge1Sb2Te4 , Ge2Sb2Te5, Ge-doped Sb-Te (Ge content of 1divide;13% at) and Sb2Te3. The effect of the growth conditions in terms of compositional and structural properties, as well as of the employed catalyst nanoseeds, will be discussed on the basis of both large area (XRD, TXRF) and local area (SEM, TEM) analysis. In particular, it will be shown that the single-crystal MOCVD-grown NWs (with diameter down to ~ 50 nm) exhibit a cubic structure when pure GeTe NWs are grown and a hexagonal structure for both pure Sb2Te3 and Ge-Sb-Te NWs. The NW functionality analysis will be discussed on the basis of I/V and voltage pulse measurements; in particular, the phase change memory switching of single NWs will be illustrated, reversibly induced by nanosecond current pulses through metal-contacted NWs, with threshold voltage of about 1.35 V in the case of Ge1Sb2Te4 NWs. Finally, a comprehensive analysis of the MOCVD-grown NW performance will be discussed and compared with other types of chalcogenide NWs, in view of future applications for scaled PCM devices.
REFERENCES
[1] M. Longo, R. Fallica, C. Wiemer, O. Salicio, M. Fanciulli, E. Rotunno, L. Lazzarini, Nano Lett. 2012, 12, 1509minus;1515.

The ever increasing demand for cheap and robust data storage space has been met with the introduction of a number of nonvolatile data storage concepts. Different materials are utilized which change at least one of their properties upon application of an external stimulus [1]. One of these novel concepts is the phase change random access memory where information is stored via amorphous and crystalline bits exhibiting low and high electrical conductivity. A low conducting amorphous phase is formed when the material is heated locally above its melting point and solidified as a glass due to rapid cooling. The highly conducting crystalline phase is formed by heating the material above its glass temperature, resulting in its recrystallization. The required stimuli for the phase transformation may be brought on by electrical or optical pulses.
We introduce the solution based synthesis of the narrow band gap semiconductor Sb2Te3 as a model material for phase change switching. It possesses a highly anisotropic crystal structure, consisting of covalently bonded Te-Sb-Te-Sb-Te layers in ab direction. These quintuple layer stacks are interconnected via v. d. Waals forces resulting in a layered structure in c direction. Up to now switching experiments are typically performed on polycrystalline, sputter deposited thin films or on nanowires grown via vapor liquid solid growth. However, the switching geometries of such structures are either limited by the random orientation of the polycrystalline thin film or restricted to the ab orientation resulting from the growth direction of the Sb2Te3 nanowires.
We apply a solvothermal route for producing Sb2Te3 as a model phase change material following a modified protocol of Zhang et al. [2]. Sb2Te3 is produced as defect free single crystalline hexagonal platelets (HP)s. In the first step of this reaction elementary Sb and Te is formed by the reduction of Sb2O3 and TeO3 in diethylenglycole. Subsequently the metals react to the desired alloy which recrystallizes into HPs via an oriented attachment mechanism [2]. The size of the particles varies from 1 to 10 µm in lateral dimension and from a few tens to 250 nm in thickness depending on the reaction conditions. Single Sb2Te3 HPs were electrically addressed in situ in a SEM in ab and c direction respectively and switching experiments were performed in both directions.
[1] K. Szot, W. Speier, G. Bihlmayer, R. Waser, Nat. Mater. 2006, 5, 312
[2] G. Zhang, W. Wang, X. Lu, X. Li, Cryst. Growth Des. 2009, 9, 145

Different from S- and Se- based chalcogenide glasses that satisfy the 8-N rule, GeTe-based phase-change alloys do not. The unusual Ge(3)Te(3) local bonding geometry in the latter, that is no consistent with the valency of the constituent elements, is due to the formation of dative covalent bonds that utilize Te lone-pair electrons. As a result, the ideal Ge₅₀Te₅₀ composition is not a lone-pair semiconductor. However, the formation of vacancies, e.g. due to Sb substitution, generates two-fold coordinated Te atoms with lone-pair electrons. The latter serve to form three-center four-electron (3c-4e) bonds that determine the properties of the crystalline phase. The role of 3c-4e bonds in the phase-change process is also discussed. Spacial separation of GeTe and Sb₂Te₃ effects the vacancy location resulting in superior properties of the layered structures. This work was supported by JSPS through FIRST Program initiated by CSTP

Phase-change random-access memory (PCRAM) is one of the most promising candidate devices for so-called ‘universal memory&’ that has non-volatile characteristics with DRAM-like fast read/write speeds. The phase transitions involved in device operations are crystallisation/ amorphisation, of which the former speed determines the overall data-transfer rate.
Compared to crystalline phase-change materials, the amorphous counterparts show a larger electrical resistance and the electronic charge distribution is more spatially localised in nature. The amorphous-to-crystalline phase transition thus involves the local redistribution of electronic charges in the amorphous phase during the nucleation and growth processes.
Most of the valence electrons of the atoms in GST are involved in either forming chemical bonds or are spatially localised in the form of lone pairs, forming an asymmetric charge distribution around the atoms. We have studied the chemical bonds and lone pairs in crystalline/amorphous Ge-Sb-Te PC materials using the electron-localisation function (ELF) formalism. By investigating the distribution and strength of the EFL, we have been able to identify the distribution of covalent bonds and lone pairs. This approach, in principle, describes the occurrence of chemical bonds more accurately than the conventionally adopted method solely based on setting maximum (cut-off) atomic distances for defining chemical bonds. By inspecting the dynamical behaviour of the lone pairs, as well as of chemical bonds, insight can be gained into the electronic transitions that may occur during the crystallisation process of GST PC materials.

Using X-ray absorption near edge spectroscopy (XANES) in combination with a Pillatus detector which enables us to probe depth profiles with 2 nm accuracy, we investigated oxidation proces in 50 nm thin films of GeTe-based phase change alloys. We found that as-deposited amorphous films annealled at 100°C and their crystalline counterparts annealled at 100, 250 and 330°C for 4h at ambient conditions do not oxidize readily. We observed that the oxidation proces predominantly undergoes on the surface of thin films. Oxygen atoms diffuse and hook onto Ge atoms very slowly and the oxidation process saturates after a few hours.
While it appears that oxidation does not have strong influences on the structure (and hence properties) of amorphous and crystalline films, we observed that oxygen atoms present in the atmophere harmfully affects the surface and diffuse deeper into thin films during crystallization and re-amorphization cycle, respectively. Although the crystallization increases a concentration of the Ge-O bonds in a lesser extent, the laser re-amorphized films are heavily oxidized and thus a protective layer must be used during SET/RESET cycling.
We believe that the higher predisposition of GeTe-based films to the oxidation during the phase transition is driven by a change in bonding. Applied a sufficient heat flow, a laser or electrical pulse with energy exceeding a threshold energy triggers the crystallization (amorphization) process. As a consequence of this, several Ge-X(Te,Ge,Sb) bonds break and form active centers (non-saturated bonds) which can easily react with the oxygen in the air. The re-amorphization is a special case when the material is transfered to the liquid phase consisting of high amount of disordered Ge-X fragmental units which may as a matter of priority bind with the highly reactive oxygen to form Ge-O bonds. The generated heat accelerates both the diffusion of oxygen into the thin film and the reaction of oxygen with other atoms.

Contrary to almost all other elements, liquid and amorphous phases of pure tellurium have proven difficult to simulate using ab initio molecular dynamics. Standard density functional theory calculations yield structures in relatively poor agreement with available diffraction experiments at low temperature, especially regarding first neighbor distance and coordination number, which are strongly overestimated in the simulations. Tellurium being a key component of many phase change materials, this poor structural description of its disordered phases raises important issues about the ability of ab initio molecular dynamics to generate accurate structural models of amorphous phases.
In this work, we use ab initio molecular dynamics performed under constant volume (experimental values) conditions to simulate liquid Tellurium structure and dynamics along its density anomaly. We test different exchange correlation functionals and approximations, and show their influence on liquid and amorphous structures. In particular, we show that the treatment of dispersion forces is yielding a clear improvement over recent hybrid functional calculations [1], with significant local order modifications in both phases. Especially, the structure evolution along the density anomaly is shown to be related to the creation of many interconnections between Te chains, these chains having increasing lengths upon temperature reduction. In the amorphous phase, Te chains are almost perfectly isolated with specific dihedral angle distributions. These structural changes are reflected on dynamical properties, such as atomic diffusion coefficient and vibrational density of states. We then apply the same method to revisit the structure of some Te based alloys.
[1] J. Akola, R. O. Jones, S. Kohara, T. Usuki, and E. Bychkov, Phys. Rev. B 81, 094202 (2010).

Ge2Sb2Te5 (GST) is one of the most promising materials for electrical and optical data storage due to the fast phase transition that may occur between the amorphous and crystalline state. Coherent optical phonons (COP) provide a fingerprint of the crystallographic structure and can be viewed as a precursor to the phase transition. Several studies of the vibrational modes of polycrystalline (p) and amorphous (a) GST have been performed. However the structure of the GST and the change in bonding that occurs during the phase transition have proved controversial. Here we compare the experimental phonon spectra of epitaxial e-GST(15 nm)/GaSb(001) and as-deposited p-GST (37 nm) and a-GST (57 nm) films grown on a Si(001) substrate. We identify the mode character from COP spectroscopy using a Raman tensor approach due to Merlin1.
Oscillations due to coherent phonons were observed in time resolved measurements of either reflectance (R) or anisotropic reflectance (AR), or both. The dominant 3.4 THz mode in e-GST was observed only in the AR channel and exhibited a strong dependence on the probe polarization. Comparison with Merlin&’s theory suggests that this mode has the 3 dimensional (D) T2 character expected for the Oh point group, confirming that the underlying crystallographic structure is cubic. On the other hand, the 4.5 THz mode in p-GST and 4.2 THz mode in a-GST were observed in both R and AR channels. p-GST is often considered similar to rhombohedral Sb2Te3 of D(5,3d) symmetry for which only 1D and 2D modes are allowed. The 4.5 THz mode of p-GST can be associated with the A1g mode of c-Sb2Te3 crystallites of (110) and (111) orientation, for which the outer Sb and Te layers vibrate along the c-axis of the rhombohedral structure2. This mode is also weakly observed in the R channel for e-GST. The 4.2 THz mode of a-GST has instead been associated with the A1 mode of distorted Te-Te chains3. Additional Raman microscope measurements confirmed the presence of the modes observed in the pump-probe measurements.
After exposure of the films to pump fluences in excess of 2.12 mJ/cm2, new modes were observed at frequencies of 4.2, 3.45 and 3.6 THz in both R and AR channels for e-GST, p-GST and a-GST respectively. A sharp peak at 3.45 (3.6) THz in p (a) GST is suggestive of the degenerate Eg mode of either c-Sb2Te3 or Sb, which may form locally where high pump fluence has been applied. The 4.2 THz mode in e-GST can be assigned to the A1g mode of Sb.
While the modes observed in our pump-probe measurements on p/a- GST films are otherwise similar to those reported previously2, a 3.7 THz mode was not observed within either the p or a as-deposited films. This suggests that substantial differences in coordination and bonding may exist between nominally similar films depending upon their thermal history.
References:
1. R. Merlin, Solid State Commun. 102, 207 (1997)
2. M. Forst et al., Appl. Phys. Lett. 77, 13 (2000)
3. M. Hase et al., Phys. Rev. B 79, 174112 (2009)

High-density photo-excitation of opaque materials with amplified femtosecond laser pulses has led to unique physical phenomena, such as femtosecond laser ablation and laser-induced phase transition. These phenomena in narrow band-gap materials are generally induced by the photo-excitation of electrons from bonding into antibonding states (electronic softening), which will lead to a large lattice distortion along the lattice potential. Under high-density excitation with more than several mJ/cm², the amplitude of the coherent lattice displacement can reach up to several percent of the lattice constant. Under these conditions, large amplitude coherent phonons can play an important role in structural modification. The purpose of this work is to study dynamics of phase change in Ge₂Sb₂Te₅ (GST) materials after the excitation of large amplitude coherent phonons. This idea is based on the coherent control of local lattice vibrations, whose atomic motions play a main role in the rapid phase change in GST materials.
We have continued to use thin films (20 nm thick) of Ge₂Te₂/Sb₂Te₃ superlattice (SL). Significantly lower SET and RESET programming current for the SL cells has been discovered and thus Ge₂Te₂/Sb₂Te₃ SL will become a potential candidate for the future PRAM (Phase-change random access memory) devices. To measure the time-resolved reflectivity change of the sample as a function of the time delay, 120 fs-width optical pulses (with a wavelength of 800 nm and the rate of 100 kHz) from a Ti:sapphire regenerative amplifier was utilized. An optical pump-probe measurement was performed at room temperature.
As the results of the pump-probe measurements on the amorphous Ge₂Te₂/Sb₂Te₃ SL, strongly damped coherent optical (A₁ symmetry) modes were observed at various pump fluences. Most importantly, the Fourier transformed (FT) spectra of the time domain oscillation exhibit red-shift of the A₁ mode frequency (3.8 THz), depending on the pump fluence. Furthermore, new phonon modes at lower frequencies than 2.0 THz were uncovered when the pump fluence was larger than 1 mJ/cm². These phonon dynamics will be discussed in terms of the large lattice distortion and of the phase change phenomena.

Recent density functional theory studies of non-thermal pathways in Ge₂Sb₂Te₅ (GST) have suggested that the strong bonding hierarchy of GST can be utilized to induce ultrafast and more energy efficient transitions [1]. To explore the possibilities of coherent phonon generation and the concomitant changes in lattice symmetry on short time scales, we have carried out ultrafast diffraction-based studies of changes in lattice symmetry under the influence of coherent phonon generation in epitaxial layers of GST [2,3]. The experiments were carried out at ID 9 of the ESRF using a femtosecond optical pump and an x-ray structural probe (lambda; = 1.20 Å) in a pump probe configuration. The pump fluence was restricted to values well below the amorphization threshold of 20 mJ/cm² and were confirmed to be transient and reversible. Time resolved measurements were carried out for delays as short as 20 ps; the ultimate time resolution of the experiment was limited by the bunch length of the fill pattern of the electrons in the storage ring. To ensure that the optical pump volume and the volume seen by the x-ray probe were comparable, the thickness of the epitaxial layer was limited to less than 40 nm. A Si₃N₄ cap of a few nanometers thickness was employed to prevent photothermal-induced oxidation of the film. Large transient changes in the symmetry of the lattice were observed with some reflections loosing over 95% of their initial intensity. The local lattice symmetry changes induced by coherent longitudinal phonon generation will be discussed as well as the relationship of these photo-induced changes to the excitation of non-thermal pathways to induce phase-changes.

The structures of liquid or amorphous phase-change (PC) materials - e.g. Ge-Sb-Te or GeTe - have been intensively investigated by various experimental techniques including X-ray or neutron-scattering methods. This is because their structures might be able to provide useful insights into the microscopic processes involved in the amorphous-to-crystalline phase transition.
Much experimental evidence has suggested that the structural order beyond the scale of interatomic distances, i.e. medium-range structural order (MRSO), may exist in amorphous PC materials. Recently, it was found, from ab initio molecular-dynamics (AIMD) simulations, that MRSO beyond the scale of fourfold rings, i.e. at scales exceeding that for the second-nearest neighbours, might play an important role in the ultrafast crystallisation of PC materials. In this respect, knowledge about the atomic dynamical behaviour of MRSO in amorphous, and also in liquid, phases is important for understanding the physics of the microscopic crystallisation process in PC materials.
We have investigated MRSO in liquid and amorphous PC materials using AIMD by studying the temporal evolution of atomic bonds, as well as structural units constituting their crystalline counterparts. The stability of atomic bonds between different species of atomic pairs shows a characteristic dependence on temperature. Moreover, we have studied for the first time the relationship between the stability of atomic bonds and structural units, and the temporal transition rates between structural units having different degrees of structural order. The observations gained from this study shed light on unravelling the mechanism of ultrafast crystallisation of PC materials from both kinetic and energetic points of view.

We present an x-ray absorption spectroscopy based study of in-situ electrically switched phase change memory material Ge2Sb2Te5 (GST225) and the time resolution of the switching process. Up to this point static measurements on devices previously switched into the desired state at our home laboratory have been presented. This study will focus on in situ switching of devices at the synchrotron while the device is under investigation by the x-ray beam. Synchronization of x-ray beam pulses with the switching event will allow for dynamic measurements of the switching event while adjusting the timing of the x-ray coincidence with the electrical pulse will yield information on the temporal development of the material structure during the switching process. Investigation of individual devices was possible employing synchrotron radiation based x-ray absorption (fluorescence) spectroscopy experiments utilizing an x-ray beam of about 200 nm dimension spot size at beamline BL39XU of SPring-8. The devices under investigation were a special type of Lance cell, where in an otherwise isolating layer a conductive pathway was created by intermixing of the layers via a laser pulse. The ultra small spot size enabled us to contain the x-ray beam entirely within the dimensions of a single device allowing data acquisition exclusively from the devices without interference from the surrounding material.
Employing the change in fluorescence response between the amorphous and crystalline phase of GST225 when excited with a photon energy of 11.105 keV we have recorded spatial 2D arrays of the fluorescence response of the elements contained in the film - most notably germanium - investigating the state of the material and the device structure in and around the device. These maps clearly show the position of the device and the contrast in the state of the phase change material with regard to position. EXAFS and XANES data were collected on the crystalline and amorphous regions for structural information of the two phases and to allow characterization of the differences - if any - between the states of laser and electrically modified phase change material phases. In order to gain a more complete understanding of the spatial properties of the changed phase material in the device spatially overlapping XANES scans have been recorded along a line crossing the device. These scans track the transition from the amorphous to the crystalline phase.

Phase-change films are widely employed as the active layer in optical data storage discs, such as optical DVDs, and due to the scalability, phase transition speed and energy consumption are likely to form part of a future technology that replaces electrical FLASH memories . Significant progresses in the understanding of the phase-change mechanism has been achieved both from a theoretical and an experimental point of view [1]. Recently the gap between theoretical predictions and experimental determination has been filled by testing the structures obtained by density functional theory molecular dynamics (DFT-MD) against experimental XANES data [2]. Herein, first-principles Ge and Te K-edge XANES spectra of Ge2Sb2Te5 alloys have been computed from (DFT-MD) simulations. The spectra are computed within the framework of multiple-scattering theory: individual XANES spectra are computed for every photo-absorbing element in the modeled Ge2Sb2Te5 and averaged together to obtain a single spectrum. A series of such spectra are then computed for structural snapshots extracted from the DFT-MD model every 1 ps [3]. We found that the computed theoretical XANES spectra were extremely sensitive to the dynamical evolution of the simulated sample. In particular, the XANES spectra computed before and after the phase transition closely reproduce the main features of the experimentally determined Ge and Te K-edge XANES spectra of Ge2Sb2Te5 alloys. Furthermore, it was also possible to computed the XANES during the nucleation and growth process of crystallization. We ascribe the changes in the XANES spectra to the formation of four-member square-planar rings, which can be tracked directly in the MD trajectory. We are now using this method to understand the affect of scaling related stresses on the local structure of Ge2Sb2Te5 during the crystallization process.
References
[1] J. Hegedüs and S. R. Elliott, Nature Materials 7, 399 - 405 (2008).
[2] M. Krbal et. al, Phys. Rev. B, 83, 054203 (2011).
[3] O.M. Roscioni et al. Phys. Rev. B, 83, 115409 (2011).

GeSbTe alloys can be switched between the crystalline and amorphous phase using ultrafast laser pulses. The mechanism and time scales of this phase change are not well understood. We have used time resolved pump-probe reflectivity measurements to investigate the response of crystalline and amorphous Ge1Sb2Te4 thin films to optical excitation at a high repetition rate (5 MHz) with pump fluence below and approaching the switching threshold. Both the crystalline and amorphous films show a fast carrier-induced decrease in reflectivity upon pumping, but the longer time scale response differs for the two phases. A long-lived decrease in reflectivity is observed in the crystalline sample, while an increase in reflectivity is observed in the amorphous sample, with both effects lasting for hundreds of picoseconds to nanoseconds. We interpret this response as a precursor to the mechanism of optically-induced switching of the two phases, which relaxes on a nanosecond time scale. This observation is consistent with a theoretical model [1] of a phase change mechanism in which small atomic distortions alter the optical properties of the material below the switching threshold. [1] Kolobov et al., Nat. Chem. 3, 311 (2011).

Phase change materials (PCMs) exhibit rapid and reversible phase transitions between an amorphous and a crystalline state, which can be triggered by short light or electrical pulses. Since the structural changes are associated with large differences in the electronic and optical properties of the two phases, PCMs are widely used in rewritable optical data storage technology (e.g. RW-DVDs) and also considered for future non-volatile electronic memory applications.
We have used time-resolved X-ray scattering at the XPP end station of the Linear Coherent Light Source (LCLS) to directly probe the structural dynamics in PCMs after laser irradiation over an extended time range from fs to µs. Thin films of the PCMs Ge2Sb2Te5, Ag4In3Sb66Te26, and Ge15Sb85 deposited on free-standing Si3N4-membranes were irradiated by fs optical laser pulses. The structural changes during the laser-induced transitions were monitored by scattering of a time-delayed 50 fs X-ray probe pulse at 9.5 keV from the LCLS in normal-incidence transmission geometry. Taking full advantage of the high flux and the short pulse duration of the LCLS the measurements allowed to measure transient scattering pattern with high signal-to-noise and thus to obtain information on the structural properties of PCMs during switching. The behavior of the PCMs has been compared to the response of “simple” materials like Ge and Bi.
From a very preliminary analysis of the data (beamtime finished Oct. 30) we have to conclude that all transitions (amorphous-to-crystalline and crystalline-to-amorphous) seem to involve melting of the material. Depending on the excitation strength melting can occur very fast on a sub-ps time-scale as a non-thermal process driven by the strong laser-induced electronic excitation. However, it takes ns up to tens of µs for the material to resolidify and to reach the final amorphous or crystalline state. While these time-scales imply purely thermal mechanisms, most probably determined by the nucleation and growth kinetics for the given material and sample geometry, we observe in some cases distinct fluence-dependent differences in the structure of the final state. Additionally, we will compare the structural response of PCMs as obtained from the X-ray scattering experiments with time-resolved measurements of the optical properties to be performed soon.
Financial support by the German Research Council (DFG, grant SO 408/9-1 and SFB 616 “Energy dissipation at surfaces”) is gratefully acknowledged. Portions of this research were carried out at the Linac Coherent Light Source (LCLS) at the SLAC National Accelerator Laboratory. LCLS is an Office of Science User Facility operated for the U.S. Department of Energy Office of Science by Stanford University.

New results of the study of non-stoichiometric binary and multinary phase change materials from the systems Ge-(Sb,In)-Te and Sb-rich Sb-Te, (Sb,In)-Te and Sb-Se are presented. The influence of the non-stoichiometry and doping on optical and electrical properties of bulk materials and their thin films are discussed. The stoichiometry strongly influences not only properties but also the mechanism of fast phase changes in these materials.

Novel phase change materials based on Ga-Sb alloys have been proposed for
future memory applications, yet the underlying structural characteristics
governing the kinetics of phase transition remain unresolved. The structure
of the amorphous and crystalline Ga50Sb50 and Ga15Sb85 alloys are
investigated using 71Ga and 121Sb NMR spectroscopy, x-ray and neutron
diffraction. Ga and Sb atoms in amorphous Ga50Sb50 are tetrahedrally
coordinated with the existence of Ga-Ga homopolar bonds indicating a
violation of chemical order and the creation of Sb-Sb bonds, which do not
occur in the corresponding crystalline state. In the amorphous and
crystalline Ga15Sb85, Ga remains tetrahedrally coordinated exclusively to
Sb atoms and no homopolar Ga-Ga bonding is observed. On the other hand, Sb
atoms are in threefold coordination in this material, presumably forming
Sb-Sb3 pyramids as in pure Sb metal. This structural scenario is
reproduced in the Reverse Monte Carlo modeling of the neutron and x-ray
structure factors of these amorphous phases. The ultrafast kinetics of the
phase change observed in Ga15Sb85 could be ascribed to the strong
structural commonalities between the amorphous and crystalline phases that
may allow phase change via local atomic displacements.

Prototypes of phase change memory (PCM) devices, based on two different compositions within the In-Sb-Te system, were analyzed from the electrical point of view. The In-based chalcogenides were deposited in the form of thin films by Metalorganic Chemical Vapor Deposition; all deposited films showed high conformality in the 25 to 70 nm thickness range^[1]. Previous studies on the solubility of In₂Te₃ in InSb have found that a wide range of compounds can be formed, including both binary and ternary alloys featuring different crystallographic ordering and high crystallization temperatures^[2].
To realize the PCM devices, we focused on two compounds of composition (In₂Te₃)_x(InSb)_y, x=9.1 mol.% and 14.2 mol.%, both comprised within their mutual solubility limit (which has been found to be < 15 mol.%)^[3]. The functionality of the two types of devices was proved by nanosecond-pulsed electrical measurements. Devices obtained from 50 nm-thick films of both compositions, on prototype cells featuring 50² and 93² nm² heater area, exhibited reversible memory switching effect. It was found that these alloys realize cells with higher threshold voltage (ge; 2.5 V) than those based on Ge-Sb-Te (asymp; 1.5 V), which could make them viable for high-temperature operations. This result is roughly related to the high temperature (450 °C) required for crystallization of most In-Sb-Te alloys, as proved by X-ray diffraction analysis. Moreover, it was found that increasing the Te content led to a decrease of the threshold voltage, regardless of the cells size. Notably, several hundred programming cycles were successfully achieved before cell failure. Finally, the programming window was nearly one order of magnitude wide (which could be improved by changing device architecture).
References:
[1] T. Stoycheva et al., Proceedings of the E-MRS 2012 Conference, May 13-18, 2012, Strasbourg (FR).
[2] J. C. Woolley, J. Electrochem. Soc. 113, 465 (1966).
[3] J. C. Woolley et al., J. Phys. Chem. Solids 16, 138 (1960).

Phase change memory materials have been the subject of numerous publications in recent years as the materials are good candidate for non-volatile PCRAM applications. Due to its wide usage in optical data storage and its interesting electrical characteristics GeSbTe alloys remains among the most studied materials for these applications. However other chalcogenides based alloys with glass forming regions are worth investigating. The use of high throughput synthesis and screening enables rapid assessment of key parameters for investigating potential systems of interest for phase change memory applications. In the first instance we have employed high throughput physical vapour deposition method to synthesise a large portion of the GeSbTe ternary space on a single sample and studied it by means of automated high throughput characterisation techniques. This study enables rapid identification of compositions of interest for phase change memory applications and enables to prove the usefulness of the high throughput approach as the results obtained on GeSbTe can easily be compared with the extensive data found in the literature. It offers the most complete study, in terms of composition, of the GeSbTe system. The parameters investigated and reported as a function of composition are the crystallisation temperature, the change in reflectivity, the crystallographic structure and the electrical resistivity of the material. Good agreement with published studies of individual compositions has been found. Similar approach has been taken to study an emerging family of glassy materials with promising characteristics GaLaX (X= S or Te). High throughput studies of GaLaTe were carried out on single samples spanning a large portion of the compositional space highlighting the domain of existence of candidate phase change memory materials. Electrical performance of GaLaS used in a device configuration outperformed GeSbTe under the same configuration; it offers a lower power, a higher crystallisation temperature and better data retention.

We show that the amorphization process in phase-change In2Se3 nanowires grown by chemical vapor deposition can be driven by electronic effects and does not require the conventional thermal melt-quench process. In particular, using transmission electron microscopy, in-situ single-nanowire Raman spectroscopy, scanning Kelvin probe microscopy, and finite-element simulations, we demonstrate that the electronic amorphization can be achieved under optical excitations at temperatures far below the thermal melting point. The mechanism of this electronic amorphization is likely related to the presence of atomic bonds with different strengths in the crystalline phase In2Se3 and the weakening of the weaker bonds by non-equilibrium electrons. Our findings suggest that In2Se3 are promising candidates for phase-change memory applications, with the potential advantages including energy-efficient memory switching due to the electronic amorphization process, and highly stable data storage as a result of a high melting point. On a more general level, these results indicate the need to take into account the electronic effects in modeling and understanding the phase transition processes in phase-change memories.

Phase-change random-access memory (PCRAM) is among several leading contenders to replace current Si-based flash memory technology in the near future. Although PCRAM devices based on the prototypical Ge₂Sb₂Te₅ (GST) phase-change material have been commercialised, its properties are not ideal, imposing limitations on switching speed, data-retention times, power consumption and multilevel operation. Thus, enhancing the properties of GST by doping is an area of active research.
Although the most intensively-studied dopants have been p-block elements such as C, N, O and Si, transition-metal (TM) dopants, most notably Fe, have also shown promise. In addition to adjusting the properties of GST, TM dopants also have the potential to introduce magnetism - if magnetic ordering can be controlled by the state of the host, magnetism may serve as an additional property contrast between phases, making PCRAM technology suitable as a component of spin-electronic (“spintronic”) devices. However, while a selection of first-row TM dopants have been investigated both experimentally and theoretically, there has as yet been no systematic study of the entire period.
We have employed ab initio molecular-dynamics (AIMD) simulations of complete melt-quench-anneal PC cycles in small model systems of GST, doped with each of the first-row TMs. We have investigated how the dopant atoms integrate into the host amorphous and crystalline phases, and the influence they have on their properties. Sc-Co exhibit a preference for forming octahedral complexes, while Ni-Zn prefer smaller coordination numbers. While Sc is able to integrate completely into the metastable cubic crystal, Ti-Co have too short bond lengths, and thus adopt distorted vacancy and interstitial sites near the edges of voids. In a similar vein, Ni-Zn reside within voids, and can display significant atomic motion even after crystallisation of the host has completed.
The 3d states from the TMs invariably lie within the host valence and conduction bands. For Sc, the majority of the states lie within the conduction band, while the Zn and Cu states lie completely within the valence band; for Ti-Ni, the states straddle the Fermi level and contribute to both bands. There is a systematic decrease in TM charge with atomic number for the octahedral TMs, which reverses for Ni-Zn. Finally, we find that V-Fe possess magnetic electronic configurations, and thus might introduce magnetism into the host.
By surveying the complete period, trends and patterns can be identified and understood. An improved understanding of the interaction between TM dopants and a GST host could enable the predictive selection of dopants, or combinations of dopants, to tune material properties for specific applications.

Phase change memory (PCM) technology is considered to be among the most promising alternatives to conventional technologies in embedded memories [1]. To allow operation at relatively high temperatures in embedded applications, it is crucial to improve the stability of the amorphous phase. Carbon and nitrogen doping have been shown to significantly increase the crystallization temperature [1-3]. Moreover, the high RESET current requirement [2], which is a limit to the scalability of GeTe and GST, can be reduced by the incorporation of a dopant element [4].
In this presentation we focus on correlating experimental results and ab initio simulations to understand the effect of C and N incorporation in GeTe and GST PCM devices. Understanding the effect of dopants on the change of electronic properties and the mechanisms of the phase transformation requires analysis of the local order and structure of the amorphous to crystalline phases.
In this context, we demonstrate that carbon and nitrogen deeply affects the structure and the dynamical properties of the amorphous phase of GeTe. In particular, the inclusion of N and C dopant elements in GeTe has a drastic effect on the vibrational modes of GeTe therefore improving the stability of the glass. This effect goes with an increased mechanical rigidity explaining why these doped GeTe compounds have a higher crystallization temperature than the undoped ones.
Finally we will explore, mainly by FTIR and XRD measurements, the effect of C and N dopants during the annealing of amorphous PCMaterials towards their crystalline phases. These results will be discussed in order to understand the origin of the differences of the doped PCMaterials amorphous phase stability (data retention) observed between full sheet materials and the materials integrated in PCM devices.
[1] A. Fantini et al., 2010 IEEE International Electron Devices Meeting (IEDM), 2010, pp. 29.21.21-29.21.24.
[2] G. Betti Beneventi et al., Solid-State Electronics, 65-66 (2011) 197-204.
[3] V. Sousa et al., EPCOS 2011.
[4] Q. Hubert et al., IMW 2012.

Ovonic threshold switching (OTS) devices have recently attracted much interest due to the simple device structure, the ability of driving high current and etc. Due to these advantages, it is expected to replace CMOS transistor, BJT, or p-n diode for the cell select switch in memory devices requiring high driving current and to raise the possibility of 3D stackable memory. For the successful implementation as a switching device, it is essential to achieve a method to control the characteristic threshold voltage (Vth) of an OTS device. As a solution, light element addition has been studied extensively with aiming at modulating the electrical properties of chalcogenide materials in its amorphous state. Starting from the modifying, we show the dependence of OTS device characteristics on the nitrogen addition and a role of nitrogen in amorphous GeSe. As the results of this approach, we found 1) reduction of Vth and delay time, 2) increasing of on-state current density, 3) maintain of amorphous stability. As a cause of these results, we suggest that electronic structure of localized stats be changed by nitrogen addition. According to the compensation of defects by nitrogen, localized defect states density within band gap will be decreased. Therefore, addition of nitrogen provides the narrowing of band edges consisted of localized sates related with the threshold switching phenomena. And our predictions were in good agreement with the optical and structural characteristics of N-doped GeSe films examined with infrared spectroscopic ellipsometry and Raman spectroscopy.

Recently, multi-level phase changing properties are strongly demanded for fusion memory application. We have investigated InSbTe alloys and shown its feasibility of multi-level cell (MLC). Among InSbTe alloys, In3SbTe2 (IST) seems to be relatively more adequate for MLC because of its high crystallization temperature (Tx) and activation energy (Ea). Higher Tx and Ea mean a better thermal stability, but there is a tradeoff between the crystallization temperature and operation power. In this work, we have tried to control the Tx by doping Bi atom in the IST. Differential scanning calorimeter (DSC) results show that about 3.2 at.% of Bi doping decreases the Tx from about 300 to 230°C, and phase transformation temperatures are also dropped. In addition, the Ea is reduced from 5.0 to 2.5 eV, which is derived from Kissinger&’s formula. Bi doping decreases Tx and Ea, and the gap between Tx and the melting temperature is widen, increasing the possibility of multi-level phase transformations. In order to confirm the multi-level electrical properties of the Bi-doped IST, we have fabricated PRAM cell devices. The resistance of the Bi-doped IST sharply drops when the crystal structure changes from amorphous to crystalline state at the temperature close to Tx. The switching characteristics shows that the threshold voltage (Vth) of the Bi-doped IST becomes lower than the Vth of the undoped IST, and the Bi doped IST has multi-level differences in the resistance due to the multi-level phase transformations. It also shows that the differences in resistance between each level are distinguishable while the total resistance ratio of amorphous and crystalline state maintains three orders of magnitude. In this work, we will discuss set/reset switching and endurance characteristics of the Bi-doped IST for the application of MLC-PRAM.