Nanometallic RRAM I-Wei Chen Department of Materials Science and Engineering University of Pennsylvania Philadelphia, PA 19104 Nature Nano, 6, 237 (2011) Adv Mater,, 23, 3847 (2011) Adv Func Mater,, 22, 546 (2012)
Size-dependent Metal-insulator Transition Random Materials Crystalline & Amorphous Purely Electronic Switching
A Metal? An Insulator?
Periodic System With a band gap, or without Insulator: : infinite resistivity at 0K Metal: : finite resistivity at 0K
Random System Aperiodic: No sharp band edge Mott : Mobility edge (k space) Anderson: = diffusion i distance @ 0K (elastic tunneling) = extent of wave function Metallic if if < Metal: Insulator: < < Insulating if > Sample size (thickness) =
An insulator, small enough, is a metal! Conducting Insulating
Long- Random Insulator Atomically mixing metal into insulator
Nanometallic Thin Film: Crystalline Top electrode Nanometallic Film Bottom electrode Substrate [001] 10 [100] I (m ma) 5 0-5 30 nm 20 nm -10-3 -2-1 0 1 2 3 Voltage (V) 75%CaZrO 3-25%LaNiO 3 Insulator Metal
After switching (0) 1M ( ) Resis stance ( 100k 10k HR 1k 100 10 0 10 20 30 Voltage-controlled MIT ~ ~ 25 nm 10 6 LR Thickness (nm) As fabricated (1) 0.02 0.01 ( ) Res sistance 20 nm 10 5 Thickness 10 4 30 nm 10 3 20 nm 10 nm 10 2 2.5 nm 10 1-3 -2-1 0 1 2 3 Voltage (V) (1) Low resistance Ohmic I (A) 0.00-0.01 001-0.02-3 -2-1 0 1 2 3 Voltage (V) (0) High resistance non-ohmic
Mechanism for Switching (on/off) LRS TE HRS TE T2 T2 T3 T3 T1 electron flow T1 electron flow BE BE Free electron path ( > ) Isolated traps Filled traps < Repulsion Change Diffusion Length: on/off
Switching: A Metal-Insulator Transition rrent ( A A) Cu High resistance HR state Insulating 8 4 0-4 -8 300K 100 K 7K -0.2-0.1 0.0 0.1 0.2 Temp Cu urrent (m ma) Low resistance LR state Metallic Voltage (V) Voltage (V) Resistan nce ( ) 10 6 10 5 10 4 10 3 200K 300K 10 5 0-5 -10 10 2-3 -2-1 0 1 2 3 Voltage (V) 7K 100 K -0.2-0.1 0.0 0.1 0.2 Temp 300 K Temp
Nanometallic Glasses SiO 2 :Pt SrRuO 3 15 Pt Si Random SiO 2 :f Pt alloy Insulator Metal 10 I [ma A] 5 0-5 -10 5.5 nm 21 nm -15-6 -3 0 3 6 V [V] 2 293K I (ma) 1 0-1 10K -2-4 -2 0 2 4 V (V)
-triggered (SiO 2 :0.2 Pt) Nanoscale MIT 100M f-triggered ( ~20 nm) ) R ( 1M 10k 7 nm 100 5.5 nm -6-3 0 3 6 V (V) 21 nm 100M 12 nm 9 nm 1M R ( ) 10k 7% 13% 20% 25% 30% -6-3 0 3 6 V (V)
Random Nanometallic Materials Insulator : Metal [001] LaAlO 3 :LaNiO 3 [100] LaAlO 3 :SrRuO 3 CaZrO 3 :LaNiO 3 CaZrO 3 :SrRuO 3 Si 3 N 4 :metal A LaNiO 3 -CaZrO 3 Si 3 N 4 :metal B Si 3 N 4 :metal C Si 3 N 4 :Pt SiO 2 :Pt Oxide A:Pt etc. SiO 2 :Pt
Nanometallicity Percolation (kohm-c cm) Res sistivity 10 1 0.1 001 0.01 40%Pt 45%Pt SiO 2 :Pt f c (mol %) f Percolation (mol%) 30%Pt SiO 2 :Pt 20 38 100 150 200 250 300 Temperature (K) Bulk percolation: f ~ 0.4 SiN 3/4 :Pt <<38 38 SiN 3/4 :metal A <<40 40 LaAlO 3 :LaNiO 3 12 - LaAlO 3 :SrRuO 3 11 86 CaZrO 3 :LaNiO 3 25 - CaZrO 3 :SrRuO 3 5.5 75 Not percolation-induced induced d metallicity it
f- Map for MIT Percolation Threshold (nm) 50 25 Insulating 0 000 0.00 025 0.25 050 0.50 f Conducting Nanometallicity I (ma) V-triggered MIT 20 1M R-V 10 0-10 I-V 100k 10k 1k -20 100-6 -3 0 3 6 V (V) R ( )
Nanostructure?
SiN:metal A System: 10 nm TEM Films Insulating Switching Switching Switching 5 nm SiN SiN+ x1% metal SiN+x2% metal SiN+x3%metal SiN+x4% metal SiN+x5% metal SiN+x6% metal metal Switching Switching Conducting Metal-rich Nanoparticles (Percolating) Conducting
Oxide A:Pt System 5 nm Insulating Oxide Oxide+ x1% Pt Oxide+x2% Pt Oxide +x3% Pt Switching Switching Conducting (percolating) Metal-rich Nanoparticles Irrelevant for Nanometallic Switching
Metallic Nanoparticle? (SiO 2 :f Pt) Calibr rated Reflectio on (%) f Pt 16 14 12 10 8 6 4 f ~ 0.2 Fused Silica f ~ 0.5 f ~ 0.37 f ~ 0.3 f ~ 0.33 400 600 800 1000 Wavelength (nm) Pt < 0.3: No Metallic Nanoparticle Plasmon Resonance: Signature of Metallic Nanoparticles
Capacitance (HR): SiN:Metal A f critical C = C diel + C met NP I C met NP ~ n met NP r 2 met NP r Switching % ~ n met NP r 3 met NP % Metal C met NP ~ %/r met NP f metal < f critical : No Metallic Nanoparticle
Reality Check Scaling: x nm times x nm Endurance: >10 7 cycles
Electronic Transition?
Fundamentally Different Features T Initial State Electrode s Work Functions, B UV R C Pulse Width,
Initial State No forming Initial state: metal High breakdown voltage >15 V
Top & Bottom Electrodes Must Be Different ø TE (ev) ø BE (ev) V s Polarity SiO 2 :Pt Pt 5.65 SRO 5.0 + 20 Pt 5.65 Mo 4.6 + 10 10 5 Mo 4.6 SRO 5.0 - I (ma) 0-10 10 4 10 3 R ( ) Ag 4.26 Pt 5.65 - Pt 5.65 Ta 4.25 + -20 10 2-4 -2 0 2 4 6 V (V) + : Counter clockwise SiN 3/4 :Pt Pt 5.65 SRO 5.0 + LaAlO 3 : Pt 5.65 SRO 5.0 + LaNiO 3 Au 5.10 SRO 5.0 + Ag 4.26 SRO 5.0 - LaAlO 3 : Pt 5.65 SRO 5.0 + SrRuO 3 CaZrO 3 : Pt 5.65 SRO 5.0 + LaNiO 3 CaZrO 3 : Pt 5.65 SRO 5.0 + SrRuO 3 I (ma) 10 5 0-5 -10-6 -3 0 3 6 V (V) -: Clockwise 10 5 10 4 10 3 R ( )
UV Resets Memory: Electronic Pt, 5.6eV CaZrO 3 :LaNiO 3 SrTiO 3 SrRuO 3, 5.0eV Inten nsity 100 80 60 40 20 0 UV light 4.2 ev 3.4 ev 3.0 ev 200 300 400 500 600 Wavelength (nm) Res sistance ( ) 1000 500 0 UV off UV on LR UV off 0 100 200 300 Time (sec) UV off UV on HR UV off 0 100 200 300 Time (sec) Not an ionic effect
Electronic Transition: UV-triggered Pt SiO 2 :Pt Fused SiO 2 SrRuO 3 or Mo UV light 100k 100k R 10k UV on R ( 10k 1k UV on 1k 0 20 40 60 100 0 50 100 Time (sec) Time (sec) Not an ionic effect
R HR Dependence Resis stance ( ) 10 9 10 7 10 5 10 9 (100 m) 2 (60 m) 2 (40 m) 2 (20 m) 2 (10 m) 2 10 8 10 3-6 -4-2 0 2 4 6 Voltage (V) Resist tivity ( cm m) 10 7 10 6 10 5 27nm 20nm 17nm 10-6 10-5 10-4 Area (cm 2 ) Obeys Ohm s law (R HRS 1/A) Extreme thickness dependence!
Non-Ohmic Thickness Dependence mm 2 ) A ( m 10 7 10 5 10 3 R ~ exp(δ/ HR ) R x 10 1 5 10 15 (nm) > (HR State) t Wave Function ~ exp(- / HR ) HR Violates Ohm s Law
ζ : Decreased by trapped charge mm 2 ) A ( m 10 7 10 5 10 3 R ~ exp( / ) (nm) )20 10 R x 10 1 5 10 15 (nm) Diffusion length ( ) : HR and LR 0 0.00 0.25 0.50 f f 0.2 0.23 0.26 0.28 0.33 ζ HR (nm) 1.1 1.7 2.1 2.5 2.8 ζ LR (nm) 10.8 13.5 15.9 17.5 20.5
Symmetric Capacitance: Higher for HRS Equivalent Circuit 6000 Line & BE & Film / connecti interface Device on -Z im ( ) 3000 HRS@0V HRS@0.4V HRS@0.8V HRS@1.2V LRS Only at very high 0 0 3000 6000 Z real ( ) No Interface Barrier: Symmetric C(V)
No Pulse Width Effect (>RC time) No Rate Dependence: Not AMemristor! > RC: Same switching voltage regardless of < < RC: Increasing switching voltage at smaller Same observations for HR-LR and LR-HR switching RC Time = R Line Line C Cell Cell ~ Cell Area Prediction: Same switching voltage down to ps in sub- m devices
Device Performance Retention Variance
Memory Retention Resist tance ( ) 10 7 10 6 10 5 10 4 10 3 Retention @85 C HRS: >10ms 100ns LRS: >10ms 100ns Resist tance ( ) t retention /t write 10 7 10 6 10 5 10 4 10 3 HRS: >10ms 100ns LRS: >10ms 100ns 10yr 10 2 10 0 10 1 10 2 10 3 10 4 10 5 10 6 time (sec) 10 2 10 0 10 3 10 6 10 9 10 12 10 15 t retention /t write Good retention regardless of programming speed
Retention per Constant Voltage Stressing Test (sec) LR to HR 10000 V + *= 20 1.1V 1.6V 16 100 2.3V 27V 2.7V 12 4.4V 5.5V 8 1 4 0 0.01 0.6 0.9 1.2 1.5 V m+ (V) 2 ) ln( V m+ -4 10 years 1.1V 1.6V V + *= 2.3V 2.7V 45V 4.5V 5.5V 0.5 1.0 1.5 2.0 2.5 1/V m+ (V -1 ) Read @ < 0.4 V >10 year data retention t Fowler-Nordheim Tunneling over Barrier d: 15nm~ 1.5 26nmthick 2.6
Retention per Constant Voltage Stressing Test HR to LR (sec c) 10000 V *= 1.25V 1000 100 10 1 1.7V 2.6V 3.6V 0.1 0.50 0.75 1.00 1.25 V m (V) Read @ < 0.4 V >10 year data retention
Switching Uniformity R distribution rrent (A) Cu 10-2 100k 1.0 10-3 10-4 10-5 10-6 50 cycles -4-2 0 2 4 Applied voltage (V).1V ( ) R @0 10k 1k ±4V, 100ns 0 250 500 750 1000 # of switching F(x) Cumulative probability, 0.8 0.6 0.4 0.2 0.0 LRS HRS 10 2 10 3 10 4 10 5 Resistance, x ( ) Highly reproducible on-offoff Narrowly distributed resistance
Switching Uniformity: Weibull plot Weibull plot (R) F( x) 1 exp( ( x / x0) k ) Ln(-Ln(1-F(x) )) 2 0-2 -4-6 LRS HRS V off (LRS HRS) -V on (HRS LRS) 6 7 8 9 10 11 12 Ln(x) () k /μ R HRS 50 0.0233 R LRS 145 0.0086 19.2 0.0602 9.8 0.1135 Higher Weibull exponent k: more uniformity (lower /μ)
Switching Uniformity: R HRS & V switching Weibull distribution (analytic relation) / [ (1 2 2 / k) (1 1/ (1 1/ k) k)] 1/ 2 k 100 10 1 0.1 Our Device R HRS 0.01 0.1 1 / TiO 2 (DC) [1] TiO 2 :Pt (DC) [1] HfO 2 (AC) [2] HfO 2 :Al (AC) [2] NiO (DC) [3] NiO (AC) [3] NiO:Ti (DC) [4] NiO:Ti (AC) [4] Cu x O (AC) [5] ZnO:Mn (DC) [6] TaO (AC) [7] x PCMO (DC) [8] our device Weibull plot k 100 TiO 10 2 [1] TiO 2 :Pt [1] HfO 2 [2] HfO 1 2 :Al [2] NiO [3] NiO:Ti [4] ZnO:Mn [6] our device (V off ) 0.1 our device (-V on ) Weibull plot V switching 0.01 0.1 1 / 1.K. Tsunoda et al., Tech. Dig. International Electron Devices Meeting 2007, 767. 2.B. Gao et al., Tech. Dig. Symposium on VLSI Technology 2009, 30. 3.M.-J. Lee et al.,, Tech. Dig. International Electron Devices Meeting 2007, 771. 4.A. Chen et al., Tech. Dig. International Electron Devices Meeting 2005, 746. 5.Y. C. Yang et al., Nano Lett. 2009, 9, 1636. 6.W.-Y. Chang et al., Appl. Phys. Lett. 2009, 95, 042104. 7.Z. Wei et al., Tech. Dig. International Electron Devices Meeting 2008, 293. 8.D. -J. Seong et al., Tech. Dig. International Electron Devices Meeting 2009, 5.4.1.
Intermediate States: Metals or Insulators?
Intermediate States Insulator or Metal? Cur rrent (ma) 15 L1 0-15 1 2 0 3 4-2 0 2 Voltage (V) Resis stance ( ) 10 9 10 7 10 5 L2 0 1 2 3 10 3 4 L1 10 1 0 100 200 300 Temperature (K) Resis stance ( ) 10 9 10 7 L2 10 5 0 1 2 10 3 3 4 10 1 1 10 100 Temperature (K) FIT: Fluctuation-induced Tunneling Elastic tunneling 0 0 l 2 =area R/2 R/2 G ( T) G exp GV ( ) G 0, T C n T w G (1/ ) 10-2 (1/ ) 10-2 10-2 Experimental FIT fitting 10-3 10-4 Experimental Fitting 1 10-5 T T0 10-3 300K 250K 10 1 10 2 10 3 exp( V / V0 ) 1 h exp( V / V ) 1 V 0 G 0.01 0.1 1 Voltage (V) T(K) FIT
Intermediate States Insulator or Metal? Cur rrent (ma) Resis stance ( ) 10 9 L2 15 L1 10 7 10 7 0 1 0 10 5 0 2 1 3 10 5 0 2 1-15 4 10 3 2 3 10 3 3-2 0 2 4 L1 Voltage (V) 10 1 4 0 100 200 300 10 1 1 10 100 Temperature (K) Resist tance ( ) 10 9 L2 Temperature (K) Reduced Activation Energy 10 1 L2 0 Insulator W T dln ( T) dt W 10 0 Metal 1 10-1 2 3 10 100 Temperature (K) Multi-step on-switching: insulator-bad metal-metal transition Multi-state Multi-filament
Nanometallic RRAM Size-dependent d Nanometallicity it Metal Insulator Transition Random Materials Purely Electronic Switching
Conclusions Nanometallicty t : a size effect unique to random materials MIT: : triggered via δ, f, V, & photon; switching due to changing localization length, direction controlled by electrode work function Thank You Device properties: highly reproducible switching parameters, short read/write time independent of voltage, long retention time, adjustable switching parameters Nanostructure: metallic nanoparticles irrelevant