The Spin Torque Lego from spin torque nano-devices to advanced computing architectures J. Grollier et al., CNRS/Thales, France
NanoBrain
Spintronics : roadmap Giant Magneto-Resistance - 1988
Magnetic Nanostructures
reading the magnetization configuration
sensors
HDD read heads
Spin Transfer - 1996 J. Slonczewski JMMM 1996 L. Berger PRB 1996 julie.grollier.free.fr
ISAMMA 2013
writing the magnetization configuration 1
Spintronics : roadmap Giant Magneto-Resistance - 1988
Magnetic Nanostructures
reading the magnetization configuration
New devices - digital memories - nano-oscillators
- memristors J. Slonczewski JMMM 1996 L. Berger PRB 1996 julie.grollier.free.fr
ISAMMA 2013
sensors
HDD read heads
Spin Transfer - 1996 writing the magnetization configuration 1
Spintronics : roadmap Giant Magneto-Resistance - 1988
Magnetic Nanostructures
reading the magnetization configuration
New Computing Architectures ?
New devices - digital memories - nano-oscillators
- memristors
J. Slonczewski JMMM 1996 L. Berger PRB 1996 julie.grollier.free.fr
ISAMMA 2013
sensors
HDD read heads
Spin Transfer - 1996 writing the magnetization configuration 1
Principle of spin-torque devices I(t)
spin
m
torque
R magnetoresistance
magnetization dynamics
julie.grollier.free.fr
ISAMMA 2013
resistance variations
t, ns
2
Principle of spin-torque devices I(t)
m
spin torque
R magnetoresistance
magnetization dynamics
TIP
spin torque =
+
in-plane torque
resistance variations
t, ns
TOOP out-of-plane torque
2 torques 2 knobs to engineer the dynamic response julie.grollier.free.fr
ISAMMA 2013
2
In-plane versus out-of-plane torques
H
Mfixed Tdamping eq. position
Tfield
Mfree
julie.grollier.free.fr
ISAMMA 2013
3
In-plane versus out-of-plane torques in-plane torque anti-damping H
Mfixed Tdamping eq. position
TIP P
Tfield TIP
E
AP
destabilizes magnetization
Mfree
julie.grollier.free.fr
ISAMMA 2013
3
In-plane versus out-of-plane torques TIP
in-plane torque anti-damping H
Mfixed Tdamping eq. position
TOOP
E
P
AP
destabilizes magnetization
Tfield TIP
E
Mfree
HOOP
out-of-plane torque field-like torque P
AP
modifies energy barrier julie.grollier.free.fr
ISAMMA 2013
3
In-plane versus out-of-plane torques TIP
in-plane torque anti-damping
TIP
P
E
AP
destabilizes magnetization
Magnetization dynamics with the in-plane torque 3 scenarios depending on H
julie.grollier.free.fr
ISAMMA 2013
3
Binary Memory H < Hc
E STT P
AP
Hysteretic Switching 400
Resistance ()
350
AP
300 250
P
200 150
-2
-1
0
1
2
d.c. current (mA)
julie.grollier.free.fr
ISAMMA 2013
4
Binary Memory H < Hc
E
First observations :
STT
Katine et al. PRL 2000 Grollier et al. APL 2001
P
AP
Hysteretic Switching
Application : STT-MRAM
400
www.everspin.com
Resistance ()
350
AP FREE LAYER TUNNEL BARRIER FIXED LAYER
300 250
P
200 150
Isolation transistor OFF
-2
-1
0
www.nec.co.jp
1
2
target : D-RAM replacement
d.c. current (mA)
julie.grollier.free.fr
ISAMMA 2013
4
Stochastic device H Hc
E STT
P
H
AP
Telegraphic Switching
Resistance ()
340
- 2.7 mA
320 300
340
- 2.9 mA
320 300 0
10
20
30
40
50
60
Time (µs)
julie.grollier.free.fr
ISAMMA 2013
5
Stochastic device H Hc
E STT
First observations :
H
Fabian et al. PRL 2003 Urazhdin et al. PRL 2003
P
AP
Telegraphic Switching
Dwell times controlled by current spin torque =
Resistance ()
340
- 2.7 mA
320
handle to control probabilities
300
340
- 2.9 mA
320
Fukushima et al. SSDM 2010
300 0
10
20
30
40
Time (µs)
julie.grollier.free.fr
50
60
: spin dice
nanoscale random number generators
ISAMMA 2013
5
Spin Transfer Nano-Oscillators H > Hc
E
H
STT P
AP
Precessionnal state 2
Power density (nW/GHz/mA )
7 6 5
1.2 mA 1.0 mA 0.8 mA
4 3 2 1 0 2
4
6
frequency (GHz)
julie.grollier.free.fr
ISAMMA 2013
6
Spin Transfer Nano-Oscillators H > Hc
E
H
STT P
AP
Precessionnal state 2
Power density (nW/GHz/mA )
7 6 5
1.2 mA 1.0 mA 0.8 mA
4 3 2 1 0 2
4
6
frequency (GHz)
julie.grollier.free.fr
ISAMMA 2013
6
Spin Transfer Nano-Oscillators H > Hc
E
H
First observations : Kiselev et al. Nature 2003 Rippard et al. PRL 2004
STT P
AP
Precessionnal state
small - work directly at the GHz tunable with I and H – radiations proof
7 2
Power density (nW/GHz/mA )
ST microwave devices
6 5
1.2 mA 1.0 mA 0.8 mA
4 3
Applications telecommunication, radars, read heads…
2 1 0 2
4
6
frequency (GHz)
julie.grollier.free.fr
ISAMMA 2013
6
Challenges for ST nano-oscillators
initial performances: power 100 pW, linewidth 10 MHz Requirements for applications: - Power > 1 µW - Linewidth < 1 KHz julie.grollier.free.fr
ISAMMA 2013
7
Challenges for ST nano-oscillators
initial performances: power 100 pW, linewidth 10 MHz Requirements for applications: - Power > 1 µW : P DR2 high TMR MgO based MTJs - Linewidth < 1 KHz julie.grollier.free.fr
ISAMMA 2013
7
Challenges for ST nano-oscillators
initial performances: power 100 pW, linewidth 10 MHz Requirements for applications: - Power > 1 µW : P DR2 high TMR MgO based MTJs - Linewidth < 1 KHz julie.grollier.free.fr
ISAMMA 2013
7
Strategies to decrease LW • 1st source of LW : mode hopping (freq. spread)
T0
• 2d source of LW : phase/amplitude noise
Tiberkevich et al, PRB 2008
julie.grollier.free.fr
ISAMMA 2013
8
Strategies to decrease LW • 1st source of LW : mode hopping (freq. spread)
work with a dynamic mode well separated in energy from other modes
• 2d source of LW : phase/amplitude noise
Tiberkevich et al, PRB 2008
Vortex gyrotropic mode
P = 0.6 µW LW = 590 kHz
A. Dussaux , JG et al., Nature Com. 2010 julie.grollier.free.fr
ISAMMA 2013
8
Strategies to decrease LW • 1st source of LW : mode hopping (freq. spread)
• 2d source of LW : phase/amplitude noise
work with a dynamic mode well separated in energy from other modes
rigidify the phase
Vortex gyrotropic mode
Synchronization
P = 0.6 µW LW = 590 kHz
A. Dussaux , JG et al., Nature Com. 2010 julie.grollier.free.fr
ISAMMA 2013
B. Georges , JG et al., PRL 2008 A. Dussaux, JG et al, APL 2011 8
Microwave oscillator I
Idc
stt
ST
t
m
MR
V=RI
V
t
stt
dc current
R
sustained precession
resistance osc.
t ac voltage
Voltage (µV)
40 20 0
-20 -40 0
20
40
Time (ns)
strong advances towards applications julie.grollier.free.fr
ISAMMA 2013
9
Spin wave emitter I
Idc t
dc current
stt
ST
m
exch. inter.
stt local sustained precession
spin wave emission
Tsoi et al. PRL 1998 Demidov et al. Nat. Mat. 2010, Madami et al., Nat. Nano. 2011
Applications: Magnonics (computing with spin waves)
julie.grollier.free.fr
ISAMMA 2013
10
Microwave detector sttI>0
I
Idc
ST
m sttI 600 m/s 800
∆T = 0.8 ns v = 621 m/s
0.8 0.6
DW velocity (m/s)
normalized resistance
1.0
0.4 0.2
J=-7.8
0.0 0
1
2
MA/cm2 3
4
5
time (ns)
julie.grollier.free.fr
ISAMMA 2013
600
400
200
0
0
2
4
6
8 2
Jpulse (MA/cm )
10
12
J. Sampaio, JG et al. in preparation 19
Spin torque Lego
Magnetic Field
stochastic device
microwave detector
Time
memristor Resistance
d.c. voltage
Voltage Time
d.c. current
spin wave emitter
microwave oscillator
Resistance
binary memory Resistance
Resistance
detector (GMR,TMR)
Frequency
d.c. current
Assembling the bricks to compute julie.grollier.free.fr
ISAMMA 2013
20
Spintronic logic • MTJs logic
Ohno et al. IEDM 2010
• DW logic
Allwood et al. Science2005
• Nano-magnet logic
• All-Spin logic
Niemier et al. J. Phys. C. Matter 2011 Behin-Aein et al. Nature Nano. 2010
Boolean logic: compete with CMOS + exploit only two bricks: detector
READ julie.grollier.free.fr
binary memory
WRITE / STORE ISAMMA 2013
21
Spin torque Lego Architectures • innovative, non-boolean, hybrid CMOs/spintronic architectures • take full advantage of spin-torque functionalities
ST-Magnonics
julie.grollier.free.fr
ST-Neuromorphic architectures
ISAMMA 2013
22
Spin Torque Magnonics spin wave creation, manipulation and detection Kruglyak et al, Khitun et al., Serga et al. J.Phys.D: Appl. Phys. 2010
Spin wave emitter
ST-Magnonics gates
ST nanocontact
Spin wave manipulator
Spin wave detector
ST damping/anti-damping
dc detector microwave detector GMR/TMR spin diode
Slavin and Krivorotov, US 7,678,475 B2 julie.grollier.free.fr
ST soliton bursting
ISAMMA 2013
Bonetti and Akerman, Magnonics, 2013 23
Spin Torque Neuromorphic Architectures Synapse
ST memristor
ST stochastic synapse
Neuron
ST nanooscillators
julie.grollier.free.fr
ISAMMA 2013
ST stochastic neuron
24
Neuromorphic architectures : motivation Semiconductor industry hurdles :
- Excessive dissipation - Multicore scaling - Defects
- Massively parallel - Analog - Relatively uniform
- Fast - Low energy demand - Defect tolerant
Artificial Neural Networks algorithms:
- very performant (deep networks) - key applications : « Recognition, Mining and Synthesis » Temam, ISCA 2010 julie.grollier.free.fr
Chen, Temam et al. IISWC 2012 ISAMMA 2013
P. Dubey, Tech. Intel Magazine 2005 25
Neuromorphic architectures : basics Neuron : - processing unit - integrates information sent from other neurons through synapses - Spikes when threshold reached - « integrate and fire »
threshold
1
Synapse : - define how well the information is
w1 2 w2 w3 3
transmitted : synaptic weight - the weigths are adjustable (synaptic plasticity) - all synapes : network memory
xi neuron
julie.grollier.free.fr
ISAMMA 2013
xj
outputs
- interconnectivity (human brain 104 synapses / neuron) - scale of the network
inputs
Network performances :
w1 and w3 reinforced
synapse
wij
26
Spin torque Synapse CMOS implementation
memristor implementation
1) Store synaptic weights
SRAM banks
10 µm
plasticity
R
2) Synaptic plasticity:
OFF
1 memristor = 1 nano-synapse
ON
V
1) Store synaptic weights : non-volatile 2) Synaptic plasticity: tunable Resistance
STDP
STDP Jo et al., Nanoletters 2010 d.c. current
Schemmel et al., IJCNN 2006 julie.grollier.free.fr
Spin torque memristor = ST synapse ISAMMA 2013
27
Spin torque Neurons Biological neuron: « integrate and fire » neuron threshold
CMOS implementation
relaxation oscillators
neuristor
ST neuron
Voltage
~ 100 µm
Time
Zamarreño-Ramos et al., Frontiers Neuroscience 2011
julie.grollier.free.fr
Pickett et al. Nature Mat. 2013
ISAMMA 2013
28
Spin torque Neurons Biological neuron: « integrate and fire » neuron threshold
CMOS implementation
relaxation oscillators
neuristor
relaxation oscillator
~ 100 µm
Zamarreño-Ramos et al., Frontiers Neuroscience 2011
julie.grollier.free.fr
ST neuron
Pickett et al. Nature Mat. 2013
ISAMMA 2013
Petit, Kim, JG et al. Nature Phys. 2012
29
ST oscillators can synchronize Neural synchronization between different parts of the brain is a key operation for information processing, in particular memory Buzsaki, « Rhythms of the brain » 2006
coupling : spin waves
Fell and Axmacher, Nature Reviews Neuroscience 2011
coupling : microwaves
exp. demonstrated : up to 4
RL Mancoff et al. Nature 2005 Kaka et al. Nature 2005 julie.grollier.free.fr
Grollier at al., PRB 2006 ISAMMA 2013
Ruotolo, Cros, JG et al., Nat. Nano 2009 30
ST Synchronization: associative memories Code information in the phase of each oscillator
brain-inspired associative memories
? pattern recognition - classification
Applications: pattern recognition / classification Csaba et al., CNNA 2012 julie.grollier.free.fr
Roska et al., CNNA 2012 ISAMMA 2013
Macia et al., Nanotechnology 2011 31
Spin Torque Neural Networks Several recent proposals of hybrid spintronic/CMOS neural networks Sharad et al., IEEE Trans Nano 2012, IEDM 2012, Arxiv 2012
inspired from all-spin logic inspired from ST-induced DW motion
Krysteczko et al., Adv. Mater. 2012 Synapse = resistive switching
julie.grollier.free.fr
Neuron = stochastic firing due to backhopping
ISAMMA 2013
32
Spin Torque Neural Networks Several recent proposals of hybrid spintronic/CMOS neural networks Sharad et al., IEEE Trans Nano 2012, IEDM 2012, Arxiv 2012
inspired from all-spin logic inspired from ST-induced DW motion
Synapse = resistive switching
Neuron = stochastic firing due to backhopping
stochastic device Resistance
Krysteczko et al., Adv. Mater. 2012
Time
julie.grollier.free.fr
ISAMMA 2013
33
Advantages of stochasticity Noise : key element of neural computation near-threshold signaling/decision making
Compute with stochastic devices = Saving energy 1) Working below threshold • Switching becomes probabilistic • Ex : binary probabilistic synapses
Modha and Parkin, US2010/0220523 A1
2) Decrease non-volatility degree • Long term memory not required for all synapses • Reduce the energy barrier drastically reduce critical currents
Ultra-low power hybric CMOS/ Spintronic stochastic architectures julie.grollier.free.fr
ISAMMA 2013
34
Spin torque Lego - Spin torque versatility: engineering complex functions at the nanoscale f(x)
- Assembling ST bricks: promising for novel computing architectures Let’s build something different !
julie.grollier.free.fr
ISAMMA 2013
35
Spin torque Lego - Spin torque versatility: engineering complex functions at the nanoscale f(x)
- Assembling ST bricks: promising for novel computing architectures
julie.grollier.free.fr
ISAMMA 2013
35
Acknowledgements Nicolas Locatelli, Vincent Cros, Albert Fert, André Chanthbouala, Steven Lequeux, Joao Sampaio, Peter Metaxas, Sören Boyn, Eva Grilmadi, Paolo Bortolotti, Antoine Dussaux, Alexei Khvalkovskiy, Benoit Georges, Olivier Boulle, Sana Laribi, Cyrile Deranlot, Stéphanie Girod, Rie Matsumoto, Akio Fukushima, Hitoshi Kubota, Kay Yakushiji, Shinji Yuasa, Olivier Temam, Damien Querlioz, Pierre Bessière, Jacques Droulez
CNRS/Thales
College de France AIST
julie.grollier.free.fr
INRIA
ISAMMA 2013
IEF
36
Thank you
julie.grollier.free.fr
ISAMMA 2013