A Map for Phase-Change Materials!

Transcription

1 A Map for Phase-Change Materials Invited talk at the IEEE Nano Symposium on "Emerging Non-volatile Memory Technologies" in Santa Clara, CA, USA April, 6th 2012 Martin Salinga

2 Why are people interested in phase change materials? Application in memory devices Samsung press release Google Data Center, Finland Unique physics - (meta-)stable in amorphous and crystalline state - very fast phase transitions - strong property contrast between both states - unusual electronic behavior 6/IV/2012 M. Salinga IEEE Nano Symposium

3 Switching principle of phase change materials Power Resistivity Ω Ω Ω crystallization amorphization Temperature T m regime of fast crystallization t 1 t 2 Time 6/IV/2012 M. Salinga IEEE Nano Symposium

4 Relevant scientific questions Why is there such a strong property contrast between amorphous and crystalline phase? physics of bonding How does the electronic excitation of the material work? electronic properties Power Resistivity Temperature T m Ω crystallization t 1 t 2 How do switching times depend on structure size and pulse parameters? crystallization kinetics Ω amorphization Ω regime of fast crystallization GeTe Time The ultimate question: How does all this change upon a variation of composition? How is the amorphous phase stabilized against crystallization at moderate temperatures? => physics of glass formation 6/IV/2012 M. Salinga IEEE Nano Symposium

5 Overview Introduction Part 1: Crystallization kinetics - Viscosity and glass formation - Crystallization speed - Distinction between nucleation and growth Part 2: Physics of bonding Part 3: Compositional trends Part 4: Electronic behavior 6/IV/2012 M. Salinga IEEE Nano Symposium

6 What drives crystallization? System minimizes Gibbs free enthalpy in equilibrium Crystallization is not favorable above T m Crystalline state is favored below T m G V is driving force However, the system is hindered from following the thermodynamic driving force by a limited atomic mobility. (Locally there must be an activation barrier, otherwise it would just crystallize immediately) 6/IV/2012 M. Salinga IEEE Nano Symposium

7 Connection between crystallization speed and viscosity How is a glass formed? How does this influence crystallization? temperature = viscosity = atomic mobility = crystallization rate 6/IV/2012 M. Salinga IEEE Nano Symposium

8 Crystallization a kinetic process D. Lencer, M. Salinga, M. Wuttig Advanced Materials, 23, (2011) Temperature T / K 1000 Liquid 900 Crystalline % % 10 yrs Amorphous 300 T rt Time t / s η 1 / (Pa s) 1 crystalline amorphous T l T g Atomic mobility ΔG / H f crystalline Driving force 6/IV/2012 M. Salinga IEEE Nano Symposium

9 Connection between crystallization speed and viscosity How is a glass formed? viscosity η drops at glass transition Definition of fragility: steepness of temperature dependence of viscosity at glass transition temperature log 10 ( / Poise) T in K m=30 m=40 m=50 m=60 m=70 m=80 m=90 m=100 m=110 m=120 m=130 m=140 How does this influence crystallization? 2 m log 10 (T ) Tg T T =T g /(k b T) in 1/eV temperature = viscosity = atomic mobility = crystallization rate 6/IV/2012 M. Salinga IEEE Nano Symposium

10 Fragility Using ultra-fast calorimetry Greer et al. found a strong deviation from Arrhenius behavior in Ge 2 Sb 2 Te 5 at high temperatures: crystal growth velocity / m/s Si SiO2 Ge2Sb2Te T m /T M. Wuttig and M. Salinga, Nature Materials 11, (2012); M. Salinga, PhD thesis RWTH Aachen (2008) log 10 ( / Poise) m=30 m=40 m=50 m=60 m=70 m=80 m=90 m=100 m=110 m=120 m=130 m=140 J. Orava, A. L. Greer, et al., Nature Materials 11, (2012) T in K /(k b T) in 1/eV High fragility => pronounced increase of crystallization speed above glass transition temperature 6/IV/2012 M. Salinga IEEE Nano Symposium

11 Measuring fast crystallization by laser heating M. Salinga PhD Thesis, RWTH Aachen (2008) Temperature T / K % 90% Amorphous Liquid Crystalline 10 yrs. T rt Time t / s T l T g Reflectivity change upon recrystallization of a melt-quenched amorphous mark: pulse power (mw) Ge 15 Sb pulse width (ns) Transmission Electron Microscopy on Ge 2 Sb 2 Te 5 : evidence for formation of multiple crystallites in one laser spot 6/IV/2012 M. Salinga IEEE Nano Symposium

12 Formation of critical nuclei Consider interfacial energy G cluster (r) = 4 3 r3 G V +4 r 2 r c 2σ = ΔG V G cluster (r) ΔG c = 16π 3 3 σ ( ΔG ) 2 V 2r G C Interface ~r² Total Enthalpy r c r Driving Force ~r³ 6/IV/2012 M. Salinga IEEE Nano Symposium

13 Nucleation and growth: AgInSbTe J. Kalb, PhD Thesis, RWTH Aachen (2006) nucleation growth 5 min 7 min 9 min AFM scans on AgIn-Sb2Te: Crystals (dark) are visible in amorphous surrounding (bright). Dimensions: 3 µm by 3 µm. Anneal temperature: 160 C (DSC furnace). Film thickness: 30 nm. Direct measurement of growth velocity and nucleation rate at a certain temperature 6/IV/2012 M. Salinga IEEE Nano Symposium

14 4.7GB 2000is Why it 23.3GB 2003 important to be aware of µm 0.32 µm How "#"'&% nm 0.41 µm µm "#"$%& nm long 50GB 2004 nucleation µm 0.32 µm does it take "#"'&% nm and growth? to crystallize a certain volume? How does it crystallize? => It depends on the material mm 0.85 Often materials with high nucleation rates were 0.075mm crystallization. 0.1mm chosen for faster mm µm 1µm 0.19µm growth dominated amorphous 1µm amorphous amorphous Nucleation very rare crystalline crystalline crystalline 6/IV/2012 nucleation dominated M. Salinga IEEE Nano Symposium

15 Why is it important to be aware of nucleation and growth? SiO 2 Top electrode Switchable region Heating contact Bottom electrode Phase-change material How long does it take to crystallize a certain volume? How does it crystallize? The smaller the volume the more growth dominated it will crystallize. => It depends on the size. 20nm x 50nm For electronic memory: materials with low nucleation rates can be good candidates, since crystal growth velocities become crucial for switching speeds. growth dominated nucleation dominated amorphous amorphous amorphous 1µm Nucleation very rare crystalline crystalline crystalline 6/IV/2012 M. Salinga IEEE Nano Symposium

16 Crystallization kinetics in electronic memory 2.3 ± 0.7 M 4.2 ± 0.5 M 5.2 ± 0.7 M 6.2 ± 1.0 M Current in ma Voltage in V 1.2 Current in ma Voltage in V Current in ma Length in ns 2.0 Voltage in V Length in ns Resistance in Ohm Resistance in Ohm Current in ma Voltage in V Current in ma Length in ns 2.0 Resistance in Ohm Voltage in V Length in ns GeTe Length 16 in ns Length 16 in ns Length 2 16 in 4 ns Length in ns Pulse Length in ns Length in ns current 0 Extremely fast switching speeds 1 In the range of DRAM, 2 but non-volatile 3 voltage 6/IV/2012 M. Salinga IEEE Nano Symposium 2012 Current in ma Resistance in Ohm Resistance in Ohm Resistance in Ohm Resistance in Ohm Resistance in Ohm Resistance in Ohm SiO 2 Resistance in Ohm G. Bruns et al., Applied Physics Letters 95, (2009) Reduction of switching time by reduction of amorphous volume as expected for growth dominated crystallization Top electrode Switchable region Heating contact Bottom electrode Phase-change material 1

17 Reduction of energy per switching event M. Salinga and M. Wuttig Science 332, 543 (2011) Electrical pulses shorter and with lower amplitude => lower energy consumption F. Xiong et al., Science 332, 568 (2011) Reset Current / µa CNT electrodes State of the art Current Density / µa / nm 2 H. P. Wong et al., Proc. IEEE 98, 2201 (2010) less material to be heated up = less power necessary scaling down device size => reduction of power consumption Contact Area / nm 2 6/IV/2012 M. Salinga IEEE Nano Symposium

18 Overview Introduction Part 1: Crystallization kinetics - Viscosity and glass formation - Crystallization speed - Distinction between nucleation and growth Part 2: Physics of bonding Part 3: Compositional trends Part 4: Electronic behavior 6/IV/2012 M. Salinga IEEE Nano Symposium

19 Measurement of FTIR reflectance M. Woda, PhD Thesis, RWTH Aachen (2010) 1.0 Optical bandgap (oscillations decrease) Reference Sample R Difference in reflectance oscillation frequency due to change of (n t) on crystallization layer stack: 1- substrate 2- gold layer 3- material under investigation Analysis: model dielectric function 0 Drude contribution Amplitude decrease from complete reflectance level due to absorption Amorphous Crystalline E (ev) 6/IV/2012 M. Salinga IEEE Nano Symposium

20 Modeling dielectric function K. Shportko et al., Nature Materials 7, (Aug. 2008) What is the origin of this polarizability enhancement? 6/IV/2012 M. Salinga IEEE Nano Symposium

21 Resonance bonding in the crystalline phase - with N sp ~ 5 resp. N p ~ 3 - octahedral coordination ~ 6 bonds - covalent bonding, but unsaturated bonds - prototype: GeTe groundstate ψ K. Shportko et al., Nature Materials 7, (Aug. 2008) Resonance bonding (delocalized bonds) = groundstate ψ (unsaturated bonds) is superposition of saturated bond-configurations Φ x Strong coupling between phonons and electronic states Large Born effetive charges Z * T Large value of dielectric constant ε Φ I Φ II resonant bonding relies on long-range order, only feasible for crystalline phase contrast ( 1939 NY, Pauling, Nature of Chemical Bond (Cornell Univ. Press, ( 1973 ) Nr. 2 Lucovsky and White, Phys. Rev. B, Vol. 8, ( 1979 ) Vol. 12 Littlewood and Heine, J. Phys. C.: Solid State Phys. ( 1979 ) Vol. 12 Littlewood, J. Phys. C.: Solid State Phys. Robertson et al, Thin Solid Films 515, (2007). 6/IV/2012 M. Salinga IEEE Nano Symposium

22 Overview Introduction Part 1: Crystallization kinetics - Viscosity and glass formation - Crystallization speed - Distinction between nucleation and growth Part 2: Physics of bonding Part 3: Compositional trends Part 4: Electronic behavior 6/IV/2012 M. Salinga IEEE Nano Symposium

23 Impact of local distortions Static local distortion (Peierls) => reduced resonance bonding effects What materials have such slightly distorted structures? D. Lencer, M. Salinga, et al., ( 2008 (Dec. Nature Materials 7, Model potential: V(x) = a x 2 + b x 4 None or strong distortions: ~ harmonic behaviour Slight distortions: pronounced anharmonicity flat, 'box-like' potential large atomic fluctuations ( factor (~ Debye-Waller possibly enabling fast kinetics ( 2004 ) Matsunaga and Yamada, JJAP 43, 6/IV/2012 M. Salinga IEEE Nano Symposium

24 Classification of structures based on valence radii D. Lencer, M. Salinga, et al., ( 2008 (Dec. Nature Materials 7, ( 1973 ) Nr. 6 Simons and Bloch, Phys. Rev. B Vol. 7 ( 1974 ) Nr. 18 St. John and Bloch, PRL Vol. 33 ( 1980 ) Phys. 13 Littlewood, J. Phys. C.: Solid St. Littlewood, CRC Critical Reviews in Solid State ( 1985 ) Vol. 11, 3 and Material Sciences 6/IV/2012 M. Salinga IEEE Nano Symposium

25 Introduce coordinates for a map for PCMs D. Lencer, M. Salinga, et al., ( 2008 (Dec. Nature Materials 7, Littlewood: Generalization for non-binaries: two coordinates: based on valence radii of s- and p- orbitals derived from pseudopotential calculations ionicity ~ size difference hybridization ~ s-p-splitting treat materials as effective binaries average cations = A average anion = B ( 1974 ) Nr. 18 St. John and Bloch, PRL Vol. 33 ( 1980 ) Phys. 13 Littlewood, J. Phys. C.: Solid St. ( 1977 ) Vol. 22 Phillips, Solid. State. Communications ( 1978 ) Nr. 6 Chelikowsky and Phillips, Phys. Rev. B Vol. 17 6/IV/2012 M. Salinga IEEE Nano Symposium

26 Map for phase change materials 4 Phase Change Materials D. Lencer, M. Salinga, et al., ( 2008 (Dec. Nature Materials 7, A V A IV B VI A IV 2 BV 2 CVI 5 SiO A IV B V 2 CVI 4 Hybridization r π -1 3 P A IV B V 4 CVI 7 SiS GeO SnO 2.5 SiTe As SiSe GeSe GeS SnS PbO 2 GeTe Sb Bi SnTe PbTe SnSe PbSe PbS Ionicity r σ 6/IV/2012 M. Salinga IEEE Nano Symposium

27 Map for phase change materials (zoomed in) D. Lencer, M. Salinga, et al., ( 2008 (Dec. Nature Materials 7, SiSe 2.4 As 2.3 SiTe Si 2 Sb 2 Te 5 SiSb 2 Te 4 Ge 2 Sb 2 Te 4.5 Se 0.5 Ge 2 Sb 2 Se 5 Hybridization r GeTe GeSb 4 Te 7 GeSb 2 Te 4 Ge 2 Sb 2 Te 5 Ge 4 Sb 2 Te 7 Ge 4 SbTe 5 (GeSb 2 Te 4 ) 0.85 (SnBi 2 Te 4 ) 0.15 Ge 2 Sb 2 Te 3.5 Se 1.5 SnSb 2 Te 4 Ge 2 Sn 2 Sb 2 Te 7 GeSe 2 Sb Ge 4 Bi 0.5 Sb 0.5 Te 5 GeBi 2 Te 4 Ge 2 Bi 0.3 Sb 1.7 Te 5 GeBiSbTe 4 SnTe 1.9 Ge 2 Bi 0.5 Sb 1.5 Te 5 Ge 3 Sn 1 Sb 2 Te Ionicity r " 6/IV/2012 M. Salinga IEEE Nano Symposium

28 Overview Introduction Part 1: Crystallization kinetics - Viscosity and glass formation - Crystallization speed - Distinction between nucleation and growth Part 2: Physics of bonding Part 3: Compositional trends Part 4: Electronic behavior 6/IV/2012 M. Salinga IEEE Nano Symposium

29 Voltage-time-dilemma D. Lencer, M. Salinga, M. Wuttig Advanced Materials, 23, (2011) Time / s Crystallization kinetics is already extremely sensitive to temperature But in phase-change materials an additional strong non-linearity is active Temperature T / K Voltage / V Liquid Crystalline 90% 10% 10 yrs. Amorphous T rt Time t / s η 1 / (Pa s) 1 T l T g Atomic mobility Driving force ΔG / H f 6/IV/2012 M. Salinga IEEE Nano Symposium

30 Electrical switching D. Krebs, PhD Thesis, RWTH Aachen (2010) 6/IV/2012 M. Salinga IEEE Nano Symposium

31 Threshold switching: field dependent D. Krebs et al., Applied Physics Letters 95, (2009) Linear dependence => existence of a threshold field Offset => interface effect with electrode material 6/IV/2012 M. Salinga IEEE Nano Symposium

32 Conclusion on voltage-time-dilemma in phase change materials Temperature T / K Liquid T l Crystalline 90% 10% 10 yrs. T g Amorphous T rt 10 9 Time t / s + = Time / s Voltage / V Crystallization kinetics strongly thermally activated Heating strongly Extreme non-linearity + dependent on electrical = of voltage and time threshold switching 6/IV/2012 M. Salinga IEEE Nano Symposium

33 Thank you for your attention funded by Jülich Aachen Research Alliance Fundamentals of Future Information Technologies Questions? Contact: 6/IV/2012 M. Salinga Source: maps.google.de IEEE Nano Symposium