tutorial: merging spintronics with photonics · 2018-11-13 · what happens after fs laser...

Tutorial:

Merging Spintronicswith PhotonicsLaser-induced spin currents & all-optical switching of spintronic devices

Bert KoopmansSPICE: Ultrafast Spintronics WorkshopSML, Mainz 2018

What happens after fs laser excitation?

Quenching magnetic moment Beaurepaire et al., PRL 1996

1

S N

S N Launching spin waves Van Kampen et al., PRL 2002

S N AF F phase transition Ju et al., PRL 2004;

Thiele et al. APL 2004

S N N S

S N

+

N Slinear

Switching by circularly polarized light Stanciu et al., PRL 2007

“Toggle switching” ferrimagnets Radu et al., Nature 2011

Nijmegen group 2007

[Co/Pt]n

Outline: Local dynamics vs. spin transport2

FM

M

Heating Spin transport

1 2

Dissipation of angular momentum

Transport of angular momentum

Outline: Towards Integrated MagnetoPhotonics

Stanciu, Rasing et al., PRL 2007 Stanciu, Rasing et al., PRL 2007 Stanciu, Rasing et al., PRL 2007

Parkin et al.

3D ractetrack3

Essential ingredient: Fs loss of magnetization4

0.5

1.0

MO

con

trast

0 5 10 15

∆t (ps)

Ni thin film

50 fspump/probe

excited electronslaser pulse

electr. lattice

spins

Jean-Yves Bigot1956-2018

Eric Beaurepaire1959-2018

E. Beaurepaire, J.-C. Merle, A. Daunois, and J.-Y. Bigot., Phys. Rev. Lett. 76, 4250 (1996)

0 1 2300

400

500

0,8

0,9

1,0

Tem

pera

ture

(K)

Delay (ps)

E

M magnetization

3 Temperature Model 5

excited electronslaser pulse

electr. lattice

spins

Te Tp

Ts

Cs

Ce

Cp

gep

gspgesTe

Tp

Ts

Different materials, different response6

0.5

1.0

MO

con

trast

0 5 10 15

∆t (ps)

Ni thin film

50 fspump/probe

E. Beaurepaire, J.-C. Merle, A. Daunois, and J.-Y. Bigot., Phys. Rev. Lett. 76, 4250 (1996)

-50 0 50 100 150

0,5

0,6

0,7

0,8

0,9

1,0

M /

M0

Delay (ps)

Nickel3d, Tc = 630 K, µat = 0.6 μB

50 ps0.2 ps

Wietstruk, Bovensiepen et al., Phys. Rev. Lett. (2011)

Weinelt(Tutorial)

Gadolinium4f, Tc = 295 K, µat = 7.5 μB

Claim: it’s just (non-equilibrium) thermodynamics

Photons and hot e- do not play a significant role…7

0 1 2 40,4

0,6

0,8

1,0

M /

M0

delay (ps)BK, Tobias Roth et al., Nature Mat. 2010

Cobalt 10 nm

Electron thermalization within 100 fs!

pump

MO probe

Direct heating

Indirect via hot electronsBergeard, Mangin, BK, Malinowski et al.,

Phys. Rev. Lett. 2016

Mangin(Tutorial),

Bokor

8Proposed microscopic mechanisms / theories

Photon field + s.o. scattering Zhang & Huebner (PRL 2000,etc.) Bigot et al. (Nature Physics 2010)

Superdiffusive spin transport Battatio, Oppeneer, et al. (PRL 2010)

Spin orbit-induced spin-flip scattering Krieger, Sharma, Gross et al. (JCTC 2015) TDDFT Toews & Pastor (PRL 2015) many-body cluster

Atomistic LLG Chantrell, Nowak, Muenzenberger

Landau-Lifzhitz-Bloch approach Kazantseva, Atxitia, Chubykalo-Fesenko

Phonon mediated Elliott-Yafet spin-flip scattering + Weiss This lecture

e-e mediated Elliott-Yafet spin-flip scattering + μ(T) Mueller, Schneider, Rethfeld et al. PRL 2013

discussedlater

Lecture byOppeneer

Lecture bySharma

Angularmomentum

dumped in lattice– very similar

results

9

e-e

Microscopic 3TM - Model Hamiltonian

e

eU

e

e e

e

peσ

e-σ

pasf

Assumption: τth ~ 0 Spin-flip upon momentum scattering (Elliott-Yafet)

Leading to: τE ~ 0.4 ps

Koopmans et al., PRL 2005, Nat. Mater. 2010

e-p e-p + spin flip

kTe

EEF

ex

+½

-½

(½ + n)hp

electrons spins phonons

Spin-less free electrons Debye or EinsteinMean field Weiss

Spintronics:2 2

0 026

Fsf

sf sf

val

10

12

2

ps 150 /

8 sfatatD

Cepsf a

VEkTg

aR

Using Golden Rule & Solve Boltzmann eqs.:

e

C

C

p

peepp

p

peepe

e

TmT

mTT

Rmdtdm

TTgdt

dTc

tzPTTgdtdT

c

coth1

),(

Eep

D

at

C

gE

T

fit to, meV 36

6.0K 630

For nickel:

sMMm

electron and lattice dynamics

magnetization dynamics

~ 60 fs for asf ~ 0.1

Outline: Local dynamics vs. spin transport11

FM

M

Heating Spin transport

1 2

Dissipation of angular momentum

Transport of angular momentum

Fs laser-induced spin transport12

Malinowski et al. Nature Physics 2008

Ru

Pt Co

0,0 0,5 1 2 3

Nor

mal

ized

(arb

.u.)

Delay (ps)

Ru

NiO

Battiato et al. PRL 2010

Superdiffusive Spin Transport

Majority spins travel further

Greg Malinowski

Oppeneer(Tutorial)

Optical generation of fs spin currents13

EF

3d4sp

Ni Super-diffusiveMajority spin

maj. > min.

Spin-dependent Seebeck

T T

dT/dx

P Balistic

FM FM or NM

Due to:- Matrix elements- DOS- Transmission- Screening

P’

P’Spin dep. life time:

tup > tdown

So pure spin current!

Battiato, Oppeneer et al., PRL 2010

Fs spin currents confirmed14

Spinaccumulation

Magnetization

Malinoswki et al., Nat. Phys. 2008Rudolf et al., Nat. Comms. 2011

strong non-equilibrium! huge splitting chemical

potential…

Local dissipation angular momentum (100 fs)

Or spin currents (also fs time scale)

Melnikov et al., PRL 2011Choi, Cahill et al.,

Nat. Comms.2014Barkowski, BK, Aeschlimann,

et al., submitted

Fs spin currents confirmed15

Spinaccumulation Torque

Schellekens et al., Nat. Comms. 2014

Choi et al., Nat Comms. 2014Razdolski, Melnikov et al,

Nat. Comms. 2017

Magnetization

Malinoswki et al., Nat. Phys. 2008Rudolf et al., Nat. Comms. 2011

Melnikov et al., PRL 2011Choi, Cahill et al.,

Nat. Comms.2014Barkowski, BK, Aeschlimann,

et al., submitted

in-plane

PMA

Experimental demonstration Optical STT16

in-plane

PMA

Co

Cu

[Co/Ni]n

HK

HK

Schellekens, BK et al., Nature Comms. 2014See also: Choi, Lee et al., Nature Comms. 2014

Polar MOKEmeasuring Mz

More quantitative studies

Materials engineering: Only bottom to top spin currents17

∆M / M of Bottom layer (%)Delay (ps)

Can

ting

angl

e(m

deg.

)

Pol

arM

O s

igna

l

Co(3)

Cu(5)

[Co(.2)/Ni(.6)]4

DemagnetizationBottom layer

PrecessionTop layer

Mark Lalieu, Paul Helgers et al., Phys. Rev. B (2017)

efficiency η = ∆Mz,top

∆Mz,bottom

B

z

y±0.08o

±0.4o

Canting angle

∆M/Mbottom

How is spin current generated?18

0 1 2 3 4 5 60

5

10

Effic

ienc

y,

(%)

Repeats0 1 2 3 4 5 6

0

5

10

15

20

25

30

Can

ting

angl

e (m

deg.

/ %

)

Ni equiv. thickness (nm)

Transfered spinproportional with thickness!

~ Thickness

Co(3)

Cu(5)

[Co(.2)/Ni(.6)]N/Co(.2)N = 1 N = 2

N = 3N = 4

?N = 2

N = 1

N = 3

N = 4?

N = 1 23 4

Helpful in idenmtifyingmechanism!


How is transverse momentum absorbed?19

0 1 2 3 4 5 6 70

1

2

3

4

(%

)

Thickness (nm)0 2 4 6 8

0

10

20

30

Can

ting

angl

e (m

deg.

/ %

)

Thickness (nm)

Co Variable Thickness

Cu(5)[Co(.2)/Ni(.6)]4

Pt(2)

90% absorbed in 1.5 nm

~ 1 / Thickness

See also Razdolski, Melnikov et al., Nature Comms. 2017


Fs spin transport (and THz magnons)20

Razdolski, Bovensiepen, Melnikov et al., Nat. Comms. 2017

10.23 GHz = 0.010 THz

0.56 THz

5.5 nm Co top layer

0 5 20 40 60 80 100 120 140 160 180 200-0,5

0,0

0,5

1,0

1,5

2,0

2,5

3,0

3,5

4,0

MO

KE

sig

nal (

arb.

uni

t.)

Delay (ps)

uniform precessionstanding

spin wave

Lalieu et al., Phys. Rev. B (2017)

Resolving dispersion & q-dependent damping

? ~ 1/t2 ~ q2

Mark Lalieu, preprint

Outline

• Introduction: fs All-Optical Switching (AOS)

• AOS of spintronic materials

• Integration of “AOS” and spintronic functionality

• Conclusions & take home

22

3

Writing Magnetism with Light – Opto-magnetism23

Stanciu, Rasing et al., Phys. Rev. Lett. 2007

50 fs laser pulses

leftright

Left polarized

up

down

up

down

Writing Magnetism with Light – Opto-magnetism24

Stanciu, Rasing et al., Phys. Rev. Lett. 2007

50 fs laser pulses

leftright

Left polarized

up

down

up

down

Ostler et al., Nature Comms. 2012

20 µm

Toggle mechanism (linearly polarized!)

Khorsand et al., Phys. Rev. Lett. 2012

Helicity dependence just due to circular dichroism

Detailed insight using fs X-ray pulses (XMCD)25

Radu et al., Nature 2011

Ferrimagnetic GdFe

element-specific!

FeGd

heating

FeGd

demagnetization

FeGd

exchange-scattering

FeGd

reversal

AF

FM

Composition near “compensation”

needed


needed

AOS in Microscopic 3-Temparature model26

,ie

asf

,ie

,ie

,ie ,je

,je

1 , 1.6 2i at BS

Co-Co 0J Cobalt

Gadolinium1 , 7.5 2i at BS

( )epep e

e e

p ep p ambe p

p diff

gdT P tT Tdt C CdT g T T

T Tdt C

Electron and lattice

EY spin-flip + exchange scattering

Spin-flip processes in 2 sub-lattice

Similar to Schellekens and BK, PRB 87, 020407(R) (2013)Co-Gd 0J Gd-Gd 0J

e p s ee ep ep sH H H H H H H rate equationsGolden rule

eeHep sH

AOS phase diagram for CoxGd1-x27

M =

0 @

RT

switch

no-switch

Maarten Beens et al., in preparation

+1

0

-1

Co

Gd

0 1 2∆t (ps)

Concentration Co

+1

0

-1

Co

Gd

0 1 2∆t (ps)

P0

(108

Jm-3

)

+1

0

-1

Co

Gd

0 1 2∆t (ps)

+1

0

-1

Co

Gd

0 1 2∆t (ps)

τdiff = 2 ps

Requirements AOS vs. Fast CI-DWM

1. Anti-parallel sub-lattices = reduced M

2. Different (EY) demagnetization times

3. Exchange scattering4. PMA (useful)

28

1. Anti-parallel sub-lattices = reduced Me in white ink

2. Strong SOC SHE3. Strong SOC DMI and

so forth4. PMA

Yang, Parkin et al., Nature Nanotechnol. (2015)

(M1 – M2) / MsVe

loci

ty(m

s-1

)M1

M2

DMIPtCo

SHE

700 m/s

PMA

()

Our dream: Spintronic Photonic Memory29

• If we can engineer the proper magnetic stack

291000 m/s50 nm bits= 20 GHz

1000 m/s50 nm bits= 20 GHz

Outline





30

What about AOS of synthetic ferrimagnets?

• Theoretical predictions

31

M1

M2spacer ?

• Experiments:

Evans, Chantrell et al. APL 2014 (Fe/FePt)

• But no single-pulse switching…Mangin et al., Nature Materials 2014

Gerlach, Nowak et al. PRB 2017 (Fe/Gd)

Pt/Co/Gd … 32

Proximity induced FM @ RT

Co

Gd

Pt

few nm

~ 1 nmStrong spin-orbit(DMI & SHE)

Partly inspired by Pham, Pizzini et al., EPL 2016

VSM SQUID 0.45 nm FM Gd @ RT Msat,Gd = 1.8 MA/m

(bulk: 2.1 MA/m)

MGd: 0.45 nm

Pt/Co/Gd … 33

Mark Lalieu, Peeters, Lavrijsen et al., Phys. Rev. B 96, 220411 (Rapid) 2017

Single-pulse toggle switching

Proximity induced FM @ RT

Co

Gd

Pt

few nm

~ 1 nmStrong spin-orbit(DMI & SHE)

Partly inspired by Pham, Pizzini et al., EPL 2016

Fluence dependence

50 fJ for 50 x 50 nm2

34

Mark Lalieu, Peeters, Lavrijsen et al., Phys. Rev. B 96, 220411 (Rapid) 2017

• No helicty dependence• > 107 successful switches and @ 100 kHz

Fit assuming fixed threshold temperature

Co thickness0.8 nm

1.0 nm

1.2 nm

1.4 nm

Switching far from compensation point

• How come?

35Alloys:


needed

Alloys:Composition near “compensation”

needed

Co

Gd

Pt[Co/Ni]n

Gd

Pt

Co [Co/Ni]n

6

5

4

3

n = 2

Mark Lalieu, Maarten Beens et al., in preparation

Layered multi-sublattice M3TM36

( )epep e

e e

p ep p ambe p

p diff

gdT P tT T

dt C CdT g T T

T Tdt C

,ie

asf

,ie

,ie

,ie ,je

,je

1, 1.6

2i at BS , , 0i i i jJ J

, , 0i i i jJ J

Cobalt

Gadolinium

1 , 7.5 2i at BS

1234567

i

, , 0i i i jJ J

Electron and lattice

EY spin-flip + exchange scattering

slab-wise M3TM

Mark Lalieu, Maarten Beens et al., in preparation

AOS of alloy versus bi-layer37

• How come?

M =

0 @

RY

Alloy Bi-layer

switch

no-switch

switch

no-switch

Mark Lalieu, Maarten Beens et al., in preparationEffective medium

3 Gd layers

M3TM time-dependence38

Lalieu, Beens, Deenen, BK, to be published

5 ML

3 MLFM

Outline





39

Current (SHE) induced motion in Pt/Co/Gd

500 ns pulses0.4 x 1012 A/m2

vDW = 8 m/s

1 μm wide Pt/Co/GdMagnetic racetrack

Lalieu, Lavrijsen and BK, arXiv 1809.02347

Lalieu, Lavrijsen and BK, arXiv 1809.02347

All-optical writing “on the fly”41

laser pulses

DC current

M measured via Anom. Hall Effect

FIB

Conclusions & Take home

• Converging of spintronics and fs magnetism rapidly progressing – two routes discussed

• First step towards integrated magneto-photonics

42

World’s fastest logo…Mark Lalieu, Peeters, Lavrijsen et al., Phys. Rev. B 96, 220411 (Rapid) 2017M.L.M. Lalieu, R. Lavrijsen & BK, arXiv 1809.02347 (2018)Lalieu, Deens, BK et al., in preparation

Acknowledgements43

tutorial: merging spintronics with photonics · 2018-11-13 · what happens after fs laser...

Documents