Electrical Generation of Spin Currents - Andrew D. Kent Center for Quantum Phenomena Department of Physics New York University
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Electrical Generation of Spin Currents
Andrew D. Kent
Center for Quantum Phenomena
Department of Physics
New York University
Funding:
Energy Frontiers Research Center
Quantum-Materials for Energy Efficient Neuromorphic-Computing
SPICE-SPIN+X Seminar: June 16th, 2021 1
Electrical Generation of Spin Currents
Chris Safranski, IBM Jonathan Sun, IBM Jun-Wen Xu, NYU Haowen Ren, NYU Laura Rehm, NYU
C. Safranski et al., PRL 124, 197204 (2020) L. Rehm et al., APL 115, 182404 (2019)
L. Rehm et al., PR Appl. 15, 034088 (2021)
SPICE-SPIN+X Seminar: June 16th, 2021 2
Electrical Generation of Spin Currents
Outline
• Introduction: Spin torques and spin-orbit torques
• Charge-to-spin conversion efficiency in switching perpendicular magnetic
tunnel junction nanopillars
• Planar Hall effect spin torques
SPICE-SPIN+X Seminar: June 16th, 2021 3
Electrical Generation of Spin Currents
Outline
• Introduction: Spin torques and spin-orbit torques
• Charge-to-spin conversion efficiency in switching perpendicular magnetic
tunnel junction nanopillars
• Planar Hall effect spin torques
SPICE-SPIN+X Seminar: June 16th, 2021 4
NYU Physics Prize
Prediction
2013 Oliver of Spin-Transfer
E. Buckley CondensedTorques
Matter Physics Priz
2013 Oliver E. Buckley Prize
2013 Oliver E. Buckley Condensed Matter Physics Priz
John Slonczewski
s Citation:
Luc Berger
"For predicting spin-transfer torque and opening the field of current-
ws Carnegie
induced Mellon University
Citation: control over magnetic nanostructures."
ureships “For predicting
Citation: spin-transfer torque and opening the
field of current-induced control over magnetic
Background:
nanostructures.”
"For predicting spin-transfer torque and opening the field of current-induced contro
ureships John Slonczewski received
over magnetic nanostructures."
Foundational papers: the Physics BS at Worcester Polytechnic
Institute in 1950
J. C. Slonczewski, Phys. and
Rev. B.the Physics
39, 6996 (1989) PhD at Rutgers University in 1955. He
J. C. Slonczewski, J. Magn. Magn. Mater. 159, L1 (1996)
pursued
Background:
L. solid-state
Berger, Phys. theory
Rev. B 54, 9353 (1996) as Research Staff Member in IBM from 1955
s,
until retiring in 2002. He was located at Yorktown Heights, NY USA,
hips
except
Luc Berger for sabbaticals
1933-May-02 in 1965-6,
Emeritus 1970-1, andof
professor 1987 at Rueschlikon,
Physics Carnegie Mellon Universit
s, Switzerland.
5000 Forbes Avenue Pittsburgh, PA 15213 B.Sc., Mathematical Sciences, U. of
hips His
SPICE-SPIN+X Seminar
research
Lausanne, in
1955. solids
Ph.D., included
Physics, fundamental
U. of theories
Lausanne, of
1960. graphite bands,
Postdoctoral research 5
NYU
Prediction of Spin-Transfer Torques
The spin-up electrons are those with spin orientati
electrons have anti-parallel alignment with the exte
polarizations of the two ferromagnets, P1 and P2:
If no voltage is applied to the junction, electrons tunn
tunnel preferentially to the positive electrode. With
can be described in a two-current model. The total cur
another for the spin-down electrons. These vary depen
J. C, Applications:
SLONCZE~S~ New types of MRAM
In magnetic tunnel junctions There are two possibilities to obtain a defined anti-pa
1.2
—ik
(by using different materials or different film thickne
&p
0.8
antiferromagnet (exchange
~ bias). In this case the mag
~
0.4
O
:// . d a. / /T 0_ a a c
Q
h N
04
Fg~ ~ - Schelne of ector dy a~&cs due to the trans
1ss j at&v e exchange co
oupljng inducced by
b an 0.2 0.4 0.6 0.8
. bafgQt I 2/k 2
e&ternal voltagge across the baarriver.
FICx.. 7. Dissipativ
p 1v h ge coup ling D e (sn
In magnetic metallic multilayers (F~g 6)
Applications: Magnetic Random
g oPt~cal parlance
wh p~~fers t
mor»
Access Memory, STT-MRAM barrier
is losing
direct~on
polarized
J. C. Sloncewski, JMMM 159, L1-L7 (1996) electrons and S rift away fr sS shown.
er ace greatly in'
interface
tht o dt'1 t 'ldt n includi
at actually th e spin cu«ennt through
L. Berger, PRB 54, 9353 (1996) . Nature
the junct' n (includin g bothNanotechnology, March
longit u d inal and transverse 2015 , sur ace statess a nd the like r
Spin-transfer-torque memory
SPICE-SPIN+X Seminar
erms) is giv
1ven simply
Ils(W
by
= V —VA )(S„+Sa)
a)=D( (5.7)
The polarizat'
za ion coefficien
(3.5 ) for the two-ba , ---" hec6
, g
NYU
Basic Physics of Spin Transfer
Based on conservation of angular momentum
~int ~f
S ~i
S
dS
! ~⌧
θ θ dt ~
S
~int
dS ~
= IP sin ✓
€ dt 2e
1 dM~ ~int
dS
+ =0
dt dt
magnetization itinerant charge
‣Reference layer ‘sets’ spin-polarization of current
‣Enables readout of magnetization state through the tunnel magnetoresistance (TMR),
giant magnetoresistance (GMR), or anisotropic magnetoresistance (AMR) effects
SPICE-SPIN+X Seminar 7
NYU
Charge Current to Spin Current Conversion
Ferromagnetic layers to Spin-orbit torques
polarize the current
Spin-polarization direction set by Spin-polarization direction set by layer geometry
reference layer magnetization and current flow direction
direction
Spin torque foundational theory papers:
Heavy metals/Ferromagnet bilayers
J. C. Slonczewski, Phys. Rev. B. 39, 6996 (1989)
M. Miron et al., Nature Materials 2010
J. C. Slonczewski, J. Magn. Magn. Mater. 159, L1 (1996)
L. Liu et al., Science 2012
L. Berger, Phys. Rev. B 54, 9353 (1996)
Review article: J. Sinova et al., Spin Hall Effects, RMP 87, 1213 (2015)
SPICE-SPIN+X Seminar 8
NYU
Charge Current to Spin Current Conversion
Ferromagnetic layers to Spin-orbit torques
polarize the current z ℓ
y
x
t
FL
RL
spin-current
density
charge current
Js /Jc ≃ P Js /Jc = θSH
density
Is /Ic ≃ P Is /Ic ≃ θSH(ℓ/t)
spin current is −ℏ
ℏJs /(2e) Q ∼ m̂ RL ⊗ z ̂ Q= ξσSHE(z ̂ × E) ⊗ z ̂
2e
Polarization ⊗ Flow direction Polarization ⊗ Flow direction
Spin torque foundational theory papers:
Heavy metals/Ferromagnet bilayers
J. C. Slonczewski, Phys. Rev. B. 39, 6996 (1989)
M. Miron et al., Nature Materials 2010
J. C. Slonczewski, J. Magn. Magn. Mater. 159, L1 (1996)
L. Liu et al., Science 2012
L. Berger, Phys. Rev. B 54, 9353 (1996)
Review article: J. Sinova et al., Spin Hall Effects, RMP 87, 1213 (2015)
SPICE-SPIN+X Seminar 9
Electrical Generation of Spin Currents
Outline
• Introduction: Spin torques and spin-orbit torques
• Charge-to-spin conversion efficiency in switching perpendicular
magnetic tunnel junction nanopillars
• Planar Hall effect spin torques
SPICE-SPIN+X Seminar 10NYU
memories
LETTERS
Andrew D. Kent and Daniel C. Worledge
Solid-state memory devices with all-electrical read and write operation
information storage.
PUBLISHED ONLINE: 11 JULY 2010 | DOI: 10.1038/NMAT2804
S
pin-transfer-torque magnetic random Spin-transfer torque provides a
access memory (STT-MRAM) devices mechanism to write information. On
store information in the orientation the other hand, information can be read
of the magnetization of nanometre-scale by measuring the device resistance. The
A perpendicular-anisotropy CoFeB–MgO Applied Physics Letters ferromagneticARTICLE
elements. As such, they are
like hard disk drives, which use magnetic
states to store information. In contrast to
hard disk drives, however, STT-MRAM is
scitation.org/journal/apl
magnetoresistance refers to the percentage
change in resistance between parallel and
antiparallel magnetization alignment of the
electrodes in a magnetic tunnel junction,
magnetic tunnel junction
written and read electrically, and does not which is made of a ferromagnetic metal/
have moving parts. This is a key difference insulator/ferromagnetic metal stack5. Until
that enables the integration of magnetic 2004, the maximum magnetoresistance
devices with semiconductor chips. Such reported6–8 at room temperature was 70%.
TABLE I. TMR andmight
devices fit parameters
fulfil the speed from the pulsed switching
requirements measurements
Magnetoresistance greater thanfor vari-
100% had,
ous temperatures and theworking
of a computer’s corresponding
memory while optimal write energies.
however, been predicted in crystalline
S. Ikeda *, K. Miura , H. Yamamoto , K. Mizunuma , H. D. Gan , M. Endo , S. Kanai ,
1,2 1,2,3 1,2,3 2 1 2 2 the inherent advantages of using
having
magnetic states — that no energy is needed
Fe/MgO/Fe tunnel junctions9 and was then
observed experimentally 10,11. Subsequent
LETTERS 3
news & views *
J. Hayakawa , F. Matsukura and H. OhnoNATURE MATERIALS
1,2 1,2 DOI: 10.1038/NMAT2804
T TMRadvances
V c (mV)
to retain information.
STT-MRAM is the result of important
in physics and materials science
s 0 (ns)rapid advances in the growth E (fJ) of thin-film
materials have led to junctions with large
magnetoresistance of several hundred per
(K) (%) made APover !P the pastP! APTheAP
20 years. first!key P P cent, AP AP
! through the use!ofPtransition
P ! metalAP
finding was the theoretical prediction of electrodes (typically CoFeB).
a Magnetic tunnel junctions (MTJs) with ferromagnetic replaced by the demagnetization energy E
a a
b , resulting in large E,
of the interface are needed to correlatedemag spin-transfer torque between conduction
V¬ tCoFeB = 2.0 nm Applications: New types of MRAM 4
structure and properties in these materials.
200 399and magnetization:
electrons 421 spin-polarized 0.94 1.03 103 286
electrodes possessing a perpendicular magnetic easy axis which is the reason why perpendicular anisotropy is required for the
1 Another question is related to the75 fact 193 electrical393 currents can 416 transfer spin0.94angular 1.05 Bit line
98 283
are of great interest as they have a potential for realizing reduction of switching current. This equation shows that low ↵ is that CoFeB films show low damping.
momentum to the magnetic moments of a
Cr/Au 150 182 ferromagnet, 381 thus reorienting
403 them0.96 1–3
. In 1.10 94 287
M (T)
next-generation high-density non-volatile memory and logic needed for low switching current for a given E. However, commonly
m Free layer 0 However, in the devices fabricated a ferromagnet, the majority and minority
+m
chips with high thermal stability and low critical current for known perpendicular-anisotropy materials and structures use –m by Ohno and colleagues, which have
In plane 295 117 electron 225spin states are 305 1.48
shifted in energy. 1.38 51 195
5 nm Ru very thin CoFeB layers, the damping is Thus, if the spin polarization of electrons
1–3 18
current-induced magnetization switching . To attain perpen- noble metals with high spin–orbit interaction , which increases ↵
5 nm Ta ¬1 stronglyOutenhanced,
of plane and the origin of this incident on a ferromagnetic layer is not Word line
enhancement is not well understood. In aligned with its magnetization (that is,
dicular anisotropy, a number of material systems have been (refs 3,19–21). For example, the typical ↵ is larger than 0.1 for Co/Pt
tCoFeB CoFeB acquiring a PMA, the damping apparently The devices
the electronare is notfirst characterized
in a definite majority by measuring their field and
explored as electrodes, which include rare-earth/transition- (ref. 19). In addition, there is no established material system that
Al2O3 mp Hard layer ¬0.5 0 0.5 comparable to other
becomes 1.0thin-film
current pulse or minority spin-state), the electron spin
resistance hysteresis loops. Figure 1(b) shows the free
precesses rapidly around a momentum-
tMgO MgO 4,5 V+ 3,6,7 U µ 0H (T) materials with PMA , which typically
metal alloys , L1 -ordered (Co, Fe)–Pt alloys
0 and Co/(Pd, provides high tunnel magnetoresistance (TMR) ratio apart from the 7
layer hysteresis
dependentloop (i.e.,field
internal a ofhysteresis
the ferromagnet. loop in which the applied field is
1,8–10 have large distributions in their magnetic Electron spins dephase because of the
Pt) multilayers
tCoFeB CoFeB . However, none of them so far satisfy well-known body-centred cubic (bcc) (001) CoFe(B)–MgO system.
b properties that may lead to variations always
in lessdistribution
than theofcoercive field associated
electron momenta of the SAF layers) of a 40 nm diame-
Figure
b 1 | A perpendicularly magnetized magnetic tunnel t
junction. =a,1.3 nm
Structure of a perpendicular MTJ,
high thermal stability at reduced dimension, low-current The crystal structures of perpendicular-anisotropy materials are
5 nm Ta consisting of hard (fixed) and free (switchable) 1
magnetic
CoFeB
layers separated by a thin MgO tunnel barrier.
1 characteristics. Do such distributions
device ter pMTJ device measured
with current flow 4. As a in antheapplied perpendicular
result, field at 4 K. We
Bit-line complement
exist in CoFeB–MgO bilayers? To answer component of spin-polarization transverse
K. tCoFeB
(mJ m–2)
current-induced magnetization switching and high tunnel usually different from bcc, and on annealing the initially amorphous
Arrows represent the direction of magnetizations of the hard (mp) and free (m) magnetic layers. b, The
this question, further studies of filmsobserve
and sharp to the switching
magnetization from the P to AP state and vice versa with a
decays, transferring
10 nm Ru
M ( T)
PMA of the free layer leads to a large energy barrier (U) to the magnetization switching between up and 0
magnetoresistance ratio all at the same time. Here, we CoFeB tends to crystallize in structures other than the wanted bcc
down orientations. 0 field
device arrays are necessary. In addition, offset spin angular momentum to the ferromagnet. Figure 1 | STT-MRAM bit cell. A magnetic tunnel
of 56 mT,
In transition
reflecting
metal ferromagnets,
thethisfringe field from the SAF acting on
junction is formed by a fixed reference layer
5 nm Ta the switching speed and energy are critical
use interfacial perpendicular anisotropy between the ferro- because they are deposited in direct contact with the perpendicular- metrics for applications. In
the
perpendicular
free layer. 18
This occurs
dephasing typically sample at theexhibits
interface a (purple),
tunnel magnetoresistance
a tunnel barrier (grey) and a free-layer
¬1 of the ferromagnet, on a length scale of element (red), with both layers magnetized
magnetic electrodes and the tunnel barrier of the MTJ by anisotropy materials . In the following, we show that all three 0 10 junctions,
Si/SiO2 sub. magnetization states
40
employing the material combination of CoFeB–MgO, a system conditions
at
nm Field induced
room temperature
thin-film elements of less than only 10 nmFIG. 1.
in
(a)
¬1 of plane. A steady increase in the PMA
Schematic of a Current-induced
pMTJ device with the pulse and
spin-valve
readout
1
measurement
tCoFeB
for high-performance perpendicular MTJs can be met
is seen with decreasing CoFeB thickness, switching with (nm)
pulses
2 (TMR)
magnetization-
as short as 0.3
1(c)
ratio of
ns showsferromagnetic
203%
voltage-induced
and an
layer responds
average
several atomic layers. However, the entire
switching
coercive
to the torquesof arrows).
the same
field toofthe283
perpendicular
The bit is 40
plane mT. Figure
of the junction
nmby adiameter
selected
(black
word line and
circuit. Nanosecond
clearly duration
indicatingwrite
that pulses
PMA is are applied throughhas the
beencapacitive port of
widely adopted to produce a giant tunnel magnetoresistance free layer with the CoFeB–MgO
in diameter. Moreover, these magnetization
states are expected to be readily altered a biasusing ¬0.5
effect.
tee, while theAnother
DC portimportant
is used for0
an interface
switching
result is that
device the
readout. 0.5
demonstrated,
(b)of Resistance
0.1 pJ in thermally
with energies
standard material system that is widely used
1.0 device
stable elements8.
vs perpen- in
because of the strong exchange coupling of
zero field with 100
moments throughout its thickness. ms duration
transistor, and operated by applying biases to the
voltage
bit lines. pulses. We observe a
Figure 1 | MTJ structure. a, Schematic of an MTJspin-transfer
ratio in MTJs with in-plane anisotropy device for torque
TMR 2,3and . free
This
CIMS
, a new mechanism11–13
layerapproach
hysteresis loop for of a in-plane-anisotropy
µ 0 diameter device at 4MTJs.
for anisotropy of the interface is shown toHbe(T) bistable region
However, the ultimate switching speed of around zero applied voltage and voltage-induced
measurements. b, Top view of an MTJ no
requires magnetization
pillarmaterial
taken by scanning
other switching
electron
than that
those
dicular field
makes
ratiopossible
used
is switching
in
203%. sufficient
(c) to
conventional overcome
Voltage-induced the strong
switching
40 nm
Alltendency
withthelongstack CoFeB–MgO
duration structures
(100 ms)
K. The TMR
requires
in
pulses further
this
of study
study and
are
switching
NATURE NANOTECHNOLOGY | VOL 10 | MARCH 2015 | www.nature.com/naturenanotechnology
prepared
with pulse on ther-
amplitudes of 405 mV for AP ! R. P switching
Beach et al., IEDM and 2008
spin transfer torque MRAM (STT-MRAM). of CoFeB layers to be magnetized in plane as optimization. Devices with non-collinear © 2015 Macmillan Publishers Limited. All rights reserved
microscope. in-plane-anisotropy This MTJs. The perpendicular
has led to intense the Figure
same
worldwide efforts to MTJs 2 |
deviceIn-plane
at 4 K in
consisting
a result and
zero out-of-plane
applied
of magnetic shapemally field. magnetization
The junction
anisotropyoxidized
(that curves
resistance for
for the
Si(001) substrate
magnetizations data in
can switch evenby RF
$358
faster 9
, mVsputtering
for P !atAProom switching. Table I shows the TMR values
realize STT-MRAM show panels
with perpendicularly (b) andresults
CoFeB/MgO. (c) isa,from
measured
tCoFeB with
= 2.0
magnetic a 30b,
nm.
dipole mV DC=bias,
tCoFeB
interactions) a without
1.313nm.bias much
Inset: less dependence
tCoFeB
nanosecond thanincubation
the delays 10
of Ta/CoFeB/MgO/CoFeB/Ta a high tunnel magnetore- temperature . The MTJ structures consist
extracted of, from
from the the pulsed
substrate
voltage loops from 4 to 295 K. We observe
(µ0 : permeability in free space). Theover magnetized
saturation layers. The focus
magnetization of switching
the efforts voltage.
for layer thicknesses of about 1 nm seen in collinearly magnetized structures.
hasAlso, D.C. Worledge et(~3
atal.,
sistance ratio, 120%, highmultilayers
thermal of the product ofApplied Physics
K and tCoFeB , where Letter
the intercept the98,
to(10)/Ta vertical022501
axis andFe(2011)
been on complex of stability dimension
monolayers of CoFeB).side, Ta (5)/Ru Nonetheless, (5)/Co
there is great20 B20
for (tCoFeB
60almost
potential : 1.0–1.3)/MgO
a factor of two increase in the TMR at 4 K compared to its value
is 1.58 T. The perpendicular-anisotropy
as low as 40 nm diameter energy
magnetic density
and
transition lowKswitching
aelements at as Co
such the slope ofThe
current thework
of linear
49 extrapolation
goes
µA. further to(t of the data
: 0.85
demonstrate or
the correspond
0.9)
perpendicularly
/Co to Kmagnetized
Fe i and
B (1.0–1.7)/Ta
CoFeB–MgO
at room (5)/Ru
temperature, (5) (num-
which is consistent with earlier studies. 5,19
MgO 20 60 20
5 3 The Perspective:
this CoFeB thickness, which determines
three
andthe
conditions
magnetic
CoA.
Ni, orthermal
that
elements
D.
and Fe withKent,
stability,
heavier
high-performance
like Pt and Pd.
Perpendicular
non-
time
These
2
b MS incorporation
Kdecreases µ0 . Circles
/2 with
perpendicular
in a device. A
all
of this
and
temperature, the
interface
squares
in
bers
perpendicularly
way,
anisotropy
are obtained
contrast
are
magnetized
Nature
to
nominal
devices.
macrospin ItMaterials
should
from magnetization
thicknesses
increase their
model pre-
in
anisotropy
and9, 699
be straightforward
nanometres)
further to
to (2010)
permit (Fig. 1a),spin-torque
High-speed which areswitching was studied by applying a less
is 2.1 ⇥ 10 J m , a value comparable materials to thatareof theto have
Co–Pd FMR The
dictions. measurements,
largest reductionrespectively.
inprocessed
switching time
MTJs need to satisfy impose a stringent
known set of
perpendicular requirements
CoFeB-MTJ on theis shown
device to have a largeintooccurscircular
stable between devices
magnetization room
stateswith asmaller
in eventhan405 ns or duration
150 nmcurrentdiameter pulse and determining the junction state (P
SPICE-SPIN+X Seminar multilayers
perpendicular 25
and high enough
materials to be used
to secure
magnetic anisotropy
in the MTJ structure.(PMA),goodandtemperature andthe 150thermal
are already magnetoresistance
First of all, K. Further,
(>100%).atWhen
by low temperatures,
these
electron-beam magnetic there is a factor
elements,
lithography for example, by
and Ar-ion oradding
AP)milling
before and(Fig.after
1b).theForpulse. We start the measurement sequence 11ectrodes
NYU (typically CoFeB). moment of the free layer to be reversed by
spin-transfer torque.
Magnetic Tunnel Junction Nanopillars
FOCUS | COMMENTARY
Bit line Device attributes and applications
a
STT-MRAMs b
are potentially suitable for
a variety of uses, including as replacement
of battery-backed static random access Micromagnetic simulations
1.0 1.0
Resistance (kΩ)
Resistance (kΩ)
memory (SRAM) and as a fast-write
0.75 0.75
buffer in a hard disk or solid-state drive.
Word line
0.5
Table 1 lists0.5the key features of existing
and emerging −0.6 memory −0.4 -0.2
technologies.
0 0.2 0.4 0.6
−40 −20 0 20 40
Magnetic field (mT)STT-MRAM is the only Voltagenon-volatile
(V)
c
memory expected 5
to have unlimited
+m -m 3 endurance. This is because there is no
inherent magnetic wear-out mechanism
1/t (GHz)
2.5
Eb 2
for switching magnetic moments back and
forth. No atoms are moved during writing N. Statuto et al., PRB 103, 014409 (2021)
I/Ic0
0
P. Bouquin et al., APL 113, 222408 (2018)
operations, as is the case in phase change
0 1 2 3
I/I
) & 60
c0
Bit-line
Eb /(kTcomplement 1 I. Volvach et al., APL 116, 192408 (2020)
memory (PCM) or resistive random access J. B. Mohammadi et al., APL 118, 132407 (2021)
memory (RRAM); only the magnetization
ℏ ( P bit )cell.
4e 1 + P 2 Eb
Vc = α
gure 1 | STT-MRAM GP A magnetic−10tunnel−9 is rotated. There is,−5 however, an −3 electrical
0
−8 −7 −6 −4
log t (s)
nction
J. C.isSlonczewski,
formed by PRB a 71,
fixed
024411reference
(2005) layer wear-out mechanism — the dielectric
Figure 2 | STT-MRAM electrical characteristics. a, Resistance versus applied magnetic field, showing bistable resistance states near zero-field associated
urple),
J. C. a tunnel barrier
Slonczewski &with
J. Z. (grey)
Sun,
parallel
JMMM and310,a169
(P) and antiparallel free-layer
(AP)
(2007) breakdown
magnetized bits. b, Resistance versus voltage, showingof switching
the MgO between AP tunnel
and P states barrier. Tovice versa
(positive bias) and
ement (red),A.with both
the
D. ofKentpulse layers
duration
and D.magnetized
proportional
C. to pulse amplitude, characteristic
Worledge, “A new avoid
of spin this,
the ballistic
on the
switching
magnetic limitwrite voltage
at shortmemories,”
times, must
while the dashed-dotted
Nature beline kept
(negative bias). c, Pulse switching amplitude versus pulse duration, on a logarithmic scale for fixed switching probability. The dashed line shows the inverse of
isNanotechnology
characteristic 10, 187 (2015)
erpendicular to themagnetization
plane
et al.,
H. Liuduration
ofreversal
theΔjunction
"Dynamics = U/(kT) (ref.
(black
of
20); spin torque
measurements of
sufficiently
these switching
device characteristics
low
incan
(roughly
all-perpendicular
thus be used to
400
estimate Δ.
mV
The spin
inset
across
the long-time behaviour, thermally activated transitions assisted by STT. The slope of the dashed-dotted line is inversely related to the energy barrier to
shows valve
the nanopillars,"
inverse pulse JMMM 358, 233 (2014)
rows). The bit is selected versusby a amplitude
pulse word inline and
the short
the tunnel barrier) 12
. STT-MRAM can be
time limit. The slope of this curve is the STT dynamic parameter A, and the intercept with the x axis occurs at I , the
c0
SPICE-SPIN+X Seminar threshold current for STT switching, permitting determination of key device parameters from short-time pulse switching data. 12NYU
High Speed Magnetization Switching
40 nm diameter nanopillar
Pulse
1.0
Amplitude (mV)
CoFeB/W/CoFeB FL 700 AP→P
Bias Tee MgO
600 295 K
DC
500 0.8
SAF
400
1 T = 295 K
Probability
a) 300
AP to P 0.6
P to AP
Switching Error Rate
0.1 200
9 9
Resistance (kΩ)
Resistance (kΩ)
Amplitude (mV)
8 8 800
0.01 P→AP
7 7 700 0.4
6 6
295 K
5 5
0.001 600
4 b) 4
c) 500 0.2
1E-4
3 3
400
-400 -200 0 200 400 -900 -600 -300 0 300 600 900
Field (mT) Voltage (mV) 1E-5
-600 -400 200 400 600 300
Pulse Amplitude (mV) 0.0
Laura Rehm et al., Appl. Phys. Lett. 115, 182404 (2019)
Laura Rehm et al., Phys. Rev. Appl. 15, 034088 (2021)
✓ ◆
For pulse duration τ0 the 1 1 I Ic
spin-charge conversion Ns (mFL /μB) =
= ≃ 0.23 ⌧ ⌧0 Ic
efficiency in a macrospin Nc (Icτ0 /e)
model: ⌧0 Ic = I⌧ Ic ⌧ Ic ≃ 100 μA
critical number
Ns
=
P
≃ 0.11 P = mr /(2 + mr) of transmitted eNc eN dissipation
Nc (1 + P 2)ln(4 πΔ)
Δ = Eb /(kBT ) charges Nc
SPICE-SPIN+X Seminar Write energy: . 250 fJ 13Electrical Generation of Spin Currents
Outline
• Introduction: Spin torques and spin-orbit torques
• Charge-to-spin conversion efficiency in switching perpendicular magnetic
tunnel junction nanopillars
• Planar Hall effect spin torques
SPICE-SPIN+X Seminar 14NYU
Spin Orbit Torques
• Separating read/write paths can
• Allow separate optimization of read/write
channels
• Increase barrier longevity
• Eliminate reference layer switching
instabilities
• Spin Hall and Rashba effects can
be used to generate spin current.
L. Liu et al., "Spin-Torque Switching with the Giant Spin Hall
Effect of Tantalum." Science 336, 6081 (2012)
I. M. Miron et al. "Current-driven spin torque induced by the
Rashba effect in a ferromagnetic metal layer," Nature
Materials 9, 230(2010)
SPICE-SPIN+X Seminar 15PHE M JPHE
NYU
z z z
Q PHE Q PHE +Idc Q PHE
Ta/Au/FM/Ta
Ta/Pd/FM/Ta
Spin Currents Set By Magnetization
FM
Au
FM
Au Au
Ta/Pt/FM/Ta Ta Ta Ta
0 Symmetry
90 180of antidamping
270 360 spin torques:
f
ϕ (°)
Ta/Au/FM/Ta FM
z
NM y M
Ta/Pd/FM/Ta
PHT
Blue=antidamping
SHT
Red=damping
+I x
dc
Ta/Pt/FM/Ta
0 90 180 270 360
θ (°)
From C. Safranski, E. A. Montoya & I. N. Krivorotov, Nature Nanotechnology 14, 27 (2019)
tryCharge-to-spin
and material dependenceconversion
of SOTs. a,b, dΔH/dKdc measured at room temperature for different NM1 layers characterizes the
ping SHT in the xy plane (a) and that of the antidamping PHT in the xz plane (b). c–e, Schematics of the flow of pure spin current
• Spin-orbit
NM/FM interface couplingplanar
by spin-polarized in non-magnetic
Hall current densitymaterials
JPHE in the(e.g. by for
FM layer Spin-Hall, Rashba
Idc!>!0 (c) and and
Idc!a!0.charge
Error bars show the
current standard
into a spinerror of least
current squares
with fit (Methods).
a controlled spin polarization that can
exert torques on an adjacent magnetic layer and switch perpendicularly magnetized elements
tudinal spin current (p̂ ∣∣m̂ ) by the FM can give net angular momentum transfer to the FM and zero PHT, as illus-
ng torque. An example
SPICE-SPIN+X Seminar of such a mechanism trated in Fig. 3e. In contrast, we found that Au/AlO is a poor spin 16NYU The spin current flow in the z-direction for the anomalous Hall effect (AHE) mechanism is
Planar Hall Driven Spin Current
proportional to my , and, again, with a SP direction m̂. So the maximum AHE spin current
with a perpendicular spin component is when the magnetization is tilted in the y − z plane;
it is zero when m̂ = ẑ because in this case there is no y-component of magnetization.
The spin currents and spin torques associated with planar Hall effect (PHE) have a dif-
ferent symmetry. For example, they are zero when m̂ = ŷ and can be non-zero when the
ℏ magnetization is canted in the x − z plane. These symmetries provide a means to classify
Js = η (m̂ ⋅ Jdc) m̂
the response experimentally, as indicated in Fig. 2b. Recently, a torque with a symmetry
associated with the planar Hall effect has been identified in single Ni|Co multilayers [95]
2e and a torque consistent with the anomalous Hall effect was observed [96–98]. Spin injec-
tion/detection with anomalous Hall effect has also been reported [99]. Recent research from
the PIs of this proposal has identified a spin torque from Ni|Co multilayers with symmetry
associated with the planar Hall effect acting on a second (CoFeB) detector ferromagnet [34],
which we plan to explore further in the proposed project.
• Spin current needs to be injected into nearby
3.1.2FM
Spin polarization associated with magnetic/nonmagnetic interfaces: It is
Non-magnetic
now appreciated that the nature of the interfaces can change the layer NM
picture significantly. (For
an up-to-date review see [100].) Most notably, the spin polarization of interface-generated
• Thus only the z component of current is important spin currents is not necessarily in the plane of the interface or bound by crystal symmetry
constraints [101]. Amin et al. has shown that at an interface between a ferromagnet and non-
magnet in the presence of strong spin-orbit interactions, the resulting spin current entering
ℏ ℏ
the nonmagnet can have its polarization rotated from the plane of the interface [102–104].
J = η (m̂ ⋅ x)(
̂ m̂ ⋅ z)J
̂ = η cos θ sin θ
The spin-current polarization can be written as [102–104]: j = jf ŝ + jp m̂ × ŝ + jm ŝ × (m̂ × ŝ),
sz dc
where j is the spin polarization direction and ŝ = ẑ × E. At nonmagnetic interfaces, jp and
2e 2e jm vanish, as expected for spin-Hall and Rasha-Edelstein effects (i.e. the spin polarization is
in the plane of the interface perpendicular to the electric field). The new theoretical result
is that at FM/NM interfaces the net spin current can acquire a polarization at an angle set
by ŝ and the magnetization direction m̂. For
T. Taniguchi, J. Grollier & M. Stiles, PR Applied 3, 044001 (2015)
example, if the magnetization is parallel to x̂
K. D. Belashchenko et al., PR Materials 3, 011401 (2019) & (parallel
PRB 101, 020407
to the current(2020)
flow direction) the term
V. P. Amin, J. Zemen & M. D. Stiles. Interface-generated spin currents.
with PRL 121,
the prefactor 136805
jp gives (2018) of
a component
V. P. Amin, P. M. Haney & M. D. Stiles, "Interfacial spin-orbitspin
torques," arXiv:2008.01182
polarization in the ẑ direction, perpen-
dicular to the interface. In brief, the micro-
scopic mechanism is spin precession of spin-
SPICE-SPIN+X Seminar polarized electrons in the interface spin-orbit 17NYU
Patterned Samples
• CoNi has large AMR and is grown with
perpendicular anisotropy
• CoFeB has small AMR and is grown to be
weakly in-plane
• A ~0.1 Tesla field can saturate both
layers
1
0.5
M/Ms
0
−0.5
OOP
IP
−1
• 400 nm x 3 µm bridges −4 −2 0 2
Field (kOe)
4
SiO2/Ta(3)/Pt(3)/[Co(0.65)/Ni(0.98)]x2/Co(0.65)/Au(3)/CoFeB(1.5)/Ta(3)
SPICE-SPIN+X Seminar 18NYU
ST-FMR Angular Dependence
VIEW LETTERS 124, 197204 (2020) θ
• Angular dependence of resonance field
with f=14 GHz drive can identify the H in-plane H out-of-plane
layer’s modes.
14 GHz
(γ)
2
ω 2 2
= μ0 (H − Meff cos 2θ)(H − Meff cos θ)
Meff = Ms − Hp
• CoNi has out-of-plane anisotropy (Meff < 0):
Hres large for H in-plane
• CFB has in-plane anisotropy (Meff > 0):
Hres large for H out-of-plane
SPICE-SPIN+X Seminar 19NYU
ST-FMR Bias Dependence
f = 14 GHz, θ = 330 degrees
• We work at high enough frequency so that the
applied fields saturate both FM’s
magnetizations parallel to the applied field.
• Application of DC current changes the
resonances linewidths
ω ℏ η Idc
ΔH = ΔH0 + 2α + cos θ sin θ
γ e MstFM wttot
SPICE-SPIN+X Seminar 20NYU
Linewidth vs Bias
f = 14 GHz, θ = 330 degrees
• Observe linear modulation of
resonance linewidth for both layers.
• The slope is proportional to the
charge to spin conversion efficiency
dΔH ℏ η cos θ sin θ
=
dIdc e Ms tFM wttot
SPICE-SPIN+X Seminar 21NYU
Linewidth vs Bias
H
PHYSICAL REVIEW LETTERS 124, 197204 (2020)
θ = 330 degrees θ = 205 degrees
H
• Changing the z component of the field changes the sign of the spin-torques
-FMR signal at θ ¼ 330 degrees for three different dc bias. Resonance linewidth for each layer as a function of dc bias at
grees and (c) θ ¼ 205 degrees.
SPICE-SPIN+X Seminar 22the CFB layer from a spin current produced in the CoNi layer.
NYU
layer. In Ref. [21], a similar torque was observed in a single sourc
Angular Dependence of Spin Torque ferromagnet paired with a spin sink. When comparing the We
torque on the CoNi layer in this work to Ref. [21], the θ sign Here
is consist with a larger spin current flow from CoNi in the by a
direction of the Au layer. While there is a Pt layer on the angle
•Slope of linewidth vs bias follows other interface, its resistivity is high due to the thin nature versio
expected angular dependence: of the layer [40] and therefore its spin orbit generated spin damp
Here,
dΔH ℏ η cos θ sin θ the A
= conta
dIdc e MstFM wttot Assum
we w
linew
•Overall charge to spin conversion Supp
efficiency:
•CFB η = 0.05
where
•CoNi η = 0.09 partic
thickn
800 e
•Similar conversion efficiency as spin dΔH
Hall effect in materials like Pt. efficie
FIG.
C. Safranski, J. Z. Sun, J-W. Xu and 3. PRL
ADK, Angular
124,dependence of dΔH=dI dc for the CFB (top)
197204 (2020) from
SPICE-SPIN+X Seminar and CoNi (bottom) layers with a fit to the expected angular the23Cto charge-current flow in the z direction, but the thin-film
σ~ E ¼the corrections due to the charge
including current itself
NYU ∇ðδμ=eÞ ¼ ð∂ z δμ=eÞez . The electric field adjusts itself so σ þ σ AMR m2z
geometry
that treated
no electric here prevents
current flows inthat. theExcept for the applied
z direction. being zero. The effective conductivities are
− ðζσ AH my − ησ AMR mz mx Þ;
Spin Currents Set By Magnetization
electric potential eE x, only the z components of ∇μ̄ and ð10Þ
In a particular ferromagnetic layer, we can solve Eqs. (6)
x
m z Þðσ AH my − σ AMR mz mx Þ
2
∇δμ are nonzero, i.e., ∇ð μ̄=eÞ ¼ E e þ ð∂ μ̄=eÞe and ðβσ þ ησ AMR
and (7) together with the diffusion equation x x z [63], z σ~ E ¼
and
∇ðδμ=eÞ ¼ ð∂ z δμ=eÞez . The electric field adjusts itself so σ þ σ AMR m2z
that no electric current flows in the↑ z direction.
∂ 2 μ − μ ↓ − ðζσ AH my − ησ AMR2mz mx Þ; ð10Þ
In a particular ferromagnetic
ðμ ↑
− μ ↓
Þ ¼layer, we can ; solve Eqs. (6)ð8Þ σ
~ δμ ¼ σ þ σ AMR z m
and (7) together∂zwith the diffusion equation l2sf ! "
2
[63],
Spin Hall effect in non-magnetic metals and
− ðβσ þ ησ AMR mz Þ 2Spinβσ þHall mz2
ησ AMREffect
: ð11Þ
(a) σþ σ AMR m2z
z∂ ðμ↑ − μ↓ Þ ¼ μ − μ ;
2 ↑ ↓
∂z2 y l2sf
ð8Þ σ~ δμ ¼ σ þ
−ℏ
σ AMR m2z
! "
x
− ðβσ Qþ ησ= m Þ ξσ
While the effective conductivities
βσ þ ησ
2
SHE (
appear
z ̂ ×
AMR
: E)
ð11Þ ⊗
m complicated, 2
z z ̂
2emain results of this paper. If
dF σ~ simplifies considerably inσ þ certain mlimits and gives
AMR z 2
(a) m
E σ AMR z
z simple illustrations of the
x y dN the
While anisotropic
the effective magnetoresistance
conductivities appear ⊗
Polarization
can be neglected,
complicated, Flow direction
F dF
→ ðβ − ζÞm
σ~ Eσ~ Esimplifies considerably
y σ AH , i.e.,in certain
there limitscurrent
is a spin and giveswhenever
m
simple illustrations Polarization
the magnetization ofhas
theamaincomponentresultsand of this
along flowpaper.
the direction
If
y direction, set by geometry
N dN theQ anisotropic magnetoresistance canmagnetization
be neglected,out of
F
Ex iz ∼ m i m y . Thus, by tilting the
→ ðβ −it ζÞm
σ~ Eplane, y σ AH , i.e.,tothere
is possible get is ana out-of-plane
spin current whenever
component of
thethe magnetization
spins flowing has aintocomponent
the other alonglayer,
the y something
direction, not
N
Spin Orbit Interaction
(b) in magnetic Ex
metals Qiz ∼ mi my . Thus, by tiltingAnomalous the magnetization Hallout Effect
of
achievable with the spin Hall effect in nonmagnetic
d1 plane, it is possible to get an out-of-plane component of
materials. This feature is illustrated in Fig. 1(b). The factor
m the spins flowing into the other layer, something not
(b) dN
of ðβ − ζÞ arises from two contributions; the term propor-
achievable with the spin Hall effect in nonmagnetic
d1 tional toThis
materials. ζ is directly
feature is from theinpolarized
illustrated Fig. 1(b). current
The accom-
factor
F1 m
d of panying
ðβ − ζÞ the anomalous
arises from two Hall current. The
contributions; the termpropor-
term proportional
Flow perpendicular to m and E
dN 2
to β to
tional comes
ζ is from
directly thefrom
polarization
the of thecurrent
polarized counterflow
accom- current
N F1
d2
that cancels
panying the the anomalous
anomalous Hall current.Planar
Hall The termHall
current. Effect
proportional
F2N p
Whenfrom
to β comes the the anomalous
polarizationHall of the effect can becurrent
counterflow neglected,
Ex → ðη −the
σ~ Ecancels βÞm m σ σ
. This expression is more
that anomalous Hall
x z AMR σþσ AMR current.
m2z
F p
When the anomalous
complicated than that Halltheeffect
for anomalouscan beHall neglected,
effect above
FIG. 1. (a) Schematic geometry forExspin-Hall-effect-induced σ~ → ðη − βÞm m σ
2 σ
Ebecause thex anisotropic 2 . This expression is more
z AMR σþσ AMR m magnetoresistance affects the
spin-transfer torques. In this geometry, the dampinglike torque is
complicated than that for the
z
anomalous
conductivity in the z direction, as captured by the last Hall effect Flow
above parallel to m
with
FIG.respect
1. (a)toSchematic
the y axis, i.e., mfor
geometry × ðspin-Hall-effect-induced
ŷ × mÞ (with a smaller
because
factor in thethisanisotropic
expression.magnetoresistance
As with the previous affects
case,thean out-
Figures from
spin-transfer
fieldlike
with T.
torques.
torque).
Taniguchi
respect to the y
In this geometry,
(b) Schematic
et al.,
axis, PR
i.e., m
the dampinglike
geometry
Applied
× ŷ × mÞ
torque is
for anomalous-Hall-
3, 044001
(with
effect-induced spin-transfer torques. In this case, the dampinglike
ð a smaller conductivity
(2015) in Polarization
the z direction, asand
of-plane component of the magnetization gives an out-of-capturedflow by direction
the last set by magnetization!
fieldlike
torque torque).
is with (b) Schematic
respect geometry
to the fixed-layer direction factor
for anomalous-Hall-
magnetization plane in this expression.
component to Asthewithspinthecurrent,
previousQcase, an out-
iz ∼ mi mx mz . As
p,effect-induced
i.e., m × ðp × spin-transfer
mÞ (with atorques.
smallerIn fieldlike
this case, torque).
the dampinglike of-plane
with the component
previousofcase, the magnetization
the factor of ðη gives
− βÞ anappears
out-of- from
torque is with respect to the fixed-layer magnetization direction plane component to the spin current, Qiz ∼ mi mx mz . As
p, i.e., m × ðp × mÞ (with a smaller fieldlike torque). with the previous case, the factor of ðη − βÞ appears from
SPICE-SPIN+X Seminar 24dΔH=dI dc angular symmetry of ðm̂ · ŷÞ. This is incon- CFB layer, t
NYU sistent with our observation. Spin currents produced by be expected
Angular Symmetry AHE have a polarization following m̂ as well, however the
flow direction follows m̂ × x̂. When magnetization lies in
momentum
decreases a
the xz plane as studied here, there is no flow of spin current known to h
in the ẑ direction towards CFB [17]. higher AM
*
The angular dependence of the observed torques is
consistent with the absorption of angular momentum in
Material [29
higher, resu
the CFB layer from a spin current produced in the CoNi layer. As su
layer. In Ref. [21], a similar torque was observed in a single source of p
ferromagnet paired with a spin sink. When comparing the We next
torque on the CoNi layer in this work to Ref. [21], the sign Here we aim
is consist with a larger spin current flow from CoNi in the by a dimen
direction of the Au layer. While there is a Pt layer on the angle. We
other interface, its resistivity is high due to the thin nature version effi
of the layer [40] and therefore its spin orbit generated spin damping-lik
*Symmetry of anti-damping torque under field reversal with fixed current direction Here, interfa
the Au laye
contained i
Assuming a
we would e
linewidth li
Supplement
where M s
particular l
thickness, a
800 emu=cm
C. Safranski, J. Z. Sun, J-W Xu and ADK, PRL 124, 197204 (2020) dΔH=dI dc
efficiency o
SPICE-SPIN+X Seminar 25NYU
Electrical generation of spin currents
Summary
• Spin torque switching in perpendicular MTJ nanopillars
-Charge-to-spin conversion efficiency can be 0.23 for switching!
• Spin orbit torques with planar Hall effect symmetry have been
observed in CoNi multilayers
-Charge-to-spin conversion efficiency (~0.05) is on par with the Spin
Hall effect in Pt.
-The spin polarization can be partially out-of-plane, making the PHE
a candidate for deterministic switching of perpendicularly
magnetized MTJs L. Rehm et al., APL 115, 182404 (2019)
https://www.spintalks.org/talks/safranski PRL 124, 197204 (2020) L. Rehm et al., PR Appl. 15, 034088 (2021)
SPICE-SPIN+X Seminar 26NYU
Kent Group
Washington Square Park-June 2010
June 2012
Washington Square Park, February 2013
June 2018
January 2020
http://www.physics.nyu.edu/kentlab/
SPICE-SPIN+X Seminar 27NYU
POSTDOCTORAL POSITION: ANTIFERROMAGNETIC
SPINTRONICS @ NYU
Description: A postdoctoral position is available in Prof. Andrew Kent’s research group
in the Center for Quantum Phenomena of the Department of Physics. The research
focus is on antiferromagnetic spintronics, specifically spin-transport phenomena in thin
films of antiferromagnetic insulators and at their interfaces with heavy metals,
ferrimagnets and ferromagnets. The successful candidate will work within a multi-
university and national lab team with expertise in thin film materials, magnetic imaging,
measurement, nanofabrication and modeling. Experience with electronic transport and
magnetic measurements, magnetic imaging (e.g. x-ray), thin film deposition,
nanofabrication and magnetic characterization
RES EARCH | R E S E A R C Hmethods
ARTICLE is desirable. Good
communication, writing and interpersonal skills are essential. →
with an effective field H eff comprising the ex-
change field (m0HE = 47.05 T), the anisotropy field
(m0 HA = 0.82 T), the externally applied field (H),
Email: andy.kent@nyu.edu
→
and the microwave field ½H m ¼ ðHo cosð2pftÞ;
Ho sinð2pft þ qÞ; 0Þ&, where the polarization is
determined by changing the phase factor q
Posted: NYU Physics/Interfolio websites soon
from 0 to 2p, and i = 1, 2 labels the two sub-
lattices. We used the following parameter
values for the calculation: g ¼ ge , saturation
magnetization Ms = 47.7 kA/m, and a = 0.001,
in agreement with previously reported values
(24, 25). The theoretical results are displayed
in Fig. 1, together with the measured spectro-
scopic antiferromagnetic resonance (AFMR)
absorptions; the experimental data are rep-
resented by solid symbols corresponding to
three different samples studied at four avail-
able frequencies (horizontal orange arrows).
Figure S1 shows the corresponding spectra.
The upper left inset to Fig. 1 shows the electron
Downloaded from http://scie
paramagnetic resonance (EPR) spectrum ob-
tained at frequency f = 395 GHz (red curve)
for the magnetic field range corresponding to
the HFM resonance (blue triangle at m0 H ¼
4:70 T). The EPR signal is markedly distorted
by saturation of the probe, owing to the large Fig. 1. Antiferromagnetic resonance of MnF2. Positions of the EPR spectroscopy resonances of MnF2.
R. Lebrun et al., Nature 561, 222 (2018) thickness of the MnF2 single crystal used in
P. Vaidya et al., Science 268, 160 (2020)
The solid curves are the computed resonance frequencies associated with the low- and high-frequency AF
these experiments, but still allows us to deter- modes (upper right inset), the SF transition (at m0 HSF e9:4 T), and the QFM at high fields. We use the
SPICE-SPIN+X Seminar mine the location of the resonances spectro- fitting parameters m0 HA ¼ 0:82 T and m0 HE ¼ 47:05 T in Eq. 1. The different colors correspond to different 28NYU
Acknowledgments
NYU Team: Dirk Backes*, Gabriel Chaves*, Eason Chen*, Ege Cogulu, Daniel Gopman*, Christian
Hahn*, Jinting Hang*, Yu-Ming Hung*, Marion Lavanant*, Ferran Macia*, Jamileh Beik Mohammadi*,
Daniele Pinna*, Laura Rehm, Debangsu Roy*, Sohrab Sani*, Nahuel Statuto, Volker Sluka*, Georg
Wolf*, Li Ye* Yassine Quessab, Haowen Ren & Junwen Xu (*=group alumni)
Collaborators -NIST: Hans Nembach and Justin Shaw
-Advanced Light Source, Berkeley: Rajesh V. Chopdekar & Hendrik Ohldag
-IBM T. J. Watson Research Center: Chris Safranski & Jonathan Z. Sun
-NYU: Gabriel Chaves and Dan Stein
-University of Barcelona and ICMAB-CSIC: Nahuel Statuto & Ferran Macia
-Ohio State University: Fengyuan Yang
-UVA: Joseph Poon and Avik Ghosh
-UC Santa Cruz: David Lederman
-University of Central Florida: Enrique del Barco
-BBN Raytheon: Tom Ohki, Colm Ryan & Graham Rolands
-Spin Memory: Georg Wolf, Bartek Kardasz, Steve Watts & Mustafa Pinarbasi
-University of Buffalo: Igor Zutic
-Wayne State: Alex Matos Abiague
-University of Lorraine: Stephane Mangin
-U. Paris Saclay, C2N: Dafine Ravelosona
-UCSD: Eric Fullerton
SPICE-SPIN+X Seminar 29You can also read
