Vous êtes sur la page 1sur 36

Spin Polarized Transport in Magnetic Tunnel Junctions

I: The Effects due to DOS, Barrier Shape, and Impurity


States

John Q. Xiao
Dept of Physics & Astronomy, University of Delaware,
Newark, DE 19716
Acknowledgement

UD: IOP, Chinese Academic of Science:


Dr. Tao Zhu
Prof. C. Ni
Dr. Xiaohai Xiang Nanjing University
Weigang Wang Prof. Z.Z. Li
Takahiro Moriyama

Supports: NSF DMR Grant No. 0405-136


University of Delaware

16,000 undergraduates
3,000 graduate students
www.udel.edu
Outlines

• Introduction to Magnetism & spintronics


• MTJ preparation and general properties
• Barrier energy profile: electron hologram
• Bias dependence of TMR
– DOS
– Barrier shape
– Impurity states
• Conclusions
Origin of Magnetism

• Magnetism is originated from the motion of electrons around the


nucleus (planetary like motion)

The magnetic moment due to The magnetic moment due to


orbital motion: spin motion:

- -
N
Total magnetic moment:µ T = µ L + µ S Represented as a
tiny bar magnet
S
• Electrons are arranged in electronic shells
• Atoms with incomplete shells have permanent magnetic moment.
• Each atom can be viewed as a tiny bar magnet with its north and south poles
Classification of Magnetism

Types of
Susceptibility Atomic / magnetic behavior Examples
magnetism

Diamagnetism Small & negative Ag, Cu

Paramagnetism Small & positive. Pt, Pd

Large & positive,


Ferromagnetism Fe, Co, Ni
M(H).*

Anti-
ferromagnetism Small & positive. Cr

Large & positive,


Ferrimagnetism ferrites
M(H).*
Spintronics

Electronics: All current devices are based on the charge of electrons

Spintronics: In additional to charge, we also want to control the


spin of electrons

Benefited from recent developments in nanostructures, one can


achieve huge resistance change by controlling spins using
external magnetic field: Giant MagnetoResistance (GMR).

Current Challenges: Generation, manipulation, and detection of


pure spin currents in metals and semiconductors
Magnetoresistance (MR)

MR: Resistance in the presence of external magnetic field

H H
H
J J J
ρ || ρΤ ρ⊥

To assess MR, more than one geometry is needed

Positive MR: ρ increases with H.


Negative MR: ρ decreases with H
Various MR

Drude Metal (free electron gas) No MR ρ


MR description ρ||
• identical charge carriers
• single isotropic relaxation
time (τ)
ρΤ
H

Non-magnetic Metal Ordinary MR


• different meff
• energy dependent (τ) ρΤ
ρΤ > ρ|| ρ||
H
Various MR: Cont.

Ferromagnetic Metal Anisotropic MR (AMR)


ρ
ρ||
ρΤ < ρ||
∆ρ ∆ρ
ρav ≈ a few % ρΤ
H

Layers with AF coupling and granular solids ρ


Giant MR (GMR)
ρΤ ρ||
ρΤ = ρ||
H
Magnetic Tunnel Junction (MTJ)
Jullière, Phys. Rev. Lett., 54A, 225 (1975)

Miyazaki et.al: J. MMM, 139, L231, 1995.


Moodera, et.al. Phys. Rev. Lett. 74, 3273
(1995)

Parkin, et.al. Nature Materials,3, 862 (2004).


Yuasa, et.al. Nature Materials, 3: 868 (2004).

FM
I
FM
AFM
Substrate
TMR Mechanism
E E
1. Asymmetric electronic band
structures in ferromagnetic materials
¾ spin up & spin down bands
¾ DOS at EF: N↑ >> N↓
2. Two current model in FM: I=I↑+I↓
3. I ↑↑ = N ↑L N ↑R + N ↓L N ↓R
E E
I ↑↓ = N ↑L N ↓R + N ↓L N ↑R

I ↑↑ − I ↑↓ 2 PL PR
TMR = =
I ↑↑ 1 − PL PR
N↑ − N↓
P=
N↑ + N↓
Applications

magnetic field sensor for high-density recording


Non-Volatile Magnetic Memory
Datta-Das Spin FET

• Rotate spins instead of


changing number of
electrons
– Less energy and much
faster
Detecting Magnetic Beads with
Integrated (on-chip) Sensors

106 detectors /cm2

To detect 16 nm Fe3O4
(Ms=480 emc/cc)
• spin valve: 20 particles
• MTJ: 2 particles

two binding events:


¾ tag to target
¾ target to detector
MTJ Fabrication Sequency

O2 Plasma
80 Cu

16 Co
Al2O3
7 Co
10 FeMn
7 Cu

20 Py
Silicon Substrate
MTJ preparation

Optimally
oxidized MTJ
Oxide FM2
Al wedge

FM1
Si
Wedged layer preparation
TMR Dependence on Wedged Barrier

Si/Ta(5)/NiFe(5)/FeMn(12)/NiFe(6)/AlOx/NiFe(6)/Ta(5)
20
Py/AlOx(t)/Py
20
oxidized by 15

TMR (%)
30s 10
40s
16 5
150s
0
TMR (%)

-100 0 100
12 H (Oe)

0
0.5 1.0 1.5 2.0 2.5 3.0
tAl (nm)
1.26 nm
I-V curves for ideal samples

0.10

Barrier High (eV) Barrier Thickness(Å)


tAl=1.22 nm
0.05
22
0.00

-0.05
20
-0.10 18
-600 -400 -200 0 200 400 600 Positive
16
0.004
Negative
tAl=1.67nm 14
0.002
I(mA)

0.000 1.6
-0.002
1.5
-0.004

-600 -400 -200 0 200 400 600 1.4


0.0010
tAl=1.93nm 1.3
0.0005

0.0000 1.2
-0.0005 12 13 14 15 16 17 18 19 20
-0.0010
Nominal Al Thickness (Å)
-600 -400 -200 0 200 400 600
V(mV)
Barrier Height and Thickness Along
the Wedge

Si/FeNi/Al2O3/FeNi/FeMn/Cu

+J e
barrier height (eV)

2.0

1.8

1.6

1.4 FM
1.2 Low R+
Low R-
Oxide
High R+
FM
barrier thickness (A)

20
High R-

18

16

14

12
10 12 14 16
Al thickness (A)
Bias along the wedge

1 .0
(a )
Norm alized M R 0 .8

0 .6 V 1 /2
tAl
0 .4 1 5 .3 Å (O )
1 6 .8 Å (C )
0 .2 1 7 .3 Å (U )
1 9 .2 Å (U )
0 .0 2 1 .3 Å (U )

-6 0 0 -4 0 0 -2 0 0 0 200 400 600


B ia s V o lt a g e ( m V )
320 Pos
(b )
280
V 1/2 (m V)

240 N eg

200
O v e r o x id iz e d U n d e r O x id iz e d

160
14 16 18 20 22
N o m in a l A l T h ic k n e s s ( Å )
Barrier Energy Profile

• No experiment directly probes the barrier


energy profile. Barrier height and thickness
extracted from i-v curves are often
erroneous.
• Barrier energy profile is very important to
determine the transport properties
Optimum Oxidized Barrier

∆Φ = CE V0 t
CE: 7.3x10-3 rad/Vnm
V0: internal potential
t: sample thickness

tAl=1.3 nm
TEM: tAlO=1.7 nm
1.7nm Hologram: 1.7 nm
Energy Profiles of Barriers under

Bad interface

Over
Top Bottom

Both interfaces are sharp

Similar
Sharp

Under Negative value


Optimum
HRTEM image

• random
• small grain
• Coherent Growth
• large grains
Magnetic Tunnel Junction (MTJ)
Jullière, Phys. Rev. Lett., 54A, 225 (1975)

Miyazaki et.al: J. MMM, 139, L231, 1995.


Moodera, et.al. Phys. Rev. Lett. 74, 3273
(1995)

Parkin, et.al. Nature Materials,3, 862 (2004).


Yuasa, et.al. Nature Materials, 3: 868 (2004).

FM
I
FM
AFM
Substrate
Bias Dependence

Co/Al2O3/Co

200

RAP
180
R(Ω)

RP
160

30
TMR(%)

25

20
-0.4 -0.2 0.0 0.2 0.4
Bias (V)
Various models on bias dependence

Magnon scattering at Defect state assisted tunneling


the interface
DOS effects on bias dependence


⎛ 2d ⎞
J ∝ ∫ exp⎜ − ∫ {[ϕ ( x, V ) − E x ]2m} dx) ⎟⎟ N ( E x )[ f ( E ) − f ( E − eV )]dE x
⎜ 1/ 2

−∞ ⎝ h0 ⎠
⎧N1↑ ( E x ) N 2↑ ( E x ) + N1↓ ( E x ) N 2↓ ( E x ) for parrallel configuration
N (Ex ) = ⎨
⎩N1↑ ( E x ) N 2↓ ( E x ) + N1↓ ( E x ) N 2↑ ( E x ) for antiparallel configuration
Experimental Results

200 Rap exp 0 .4

P o la r iz a t io n
Rp exp 2.0
0 .3
180 Rap fit
R(Ω)

Normalized DOS
Rp fit 1.5
0 .2

160
- 0 .4 - 0 .2 0 .0 0 .2 0.4
E x - E F (e V )
1.0

30
0.5
Majority
MR(%)

exp Minority
25 0.0
fit -0.4 -0.2 0.0 0.2 0.4

Ex-EF (eV)
20
-0.4 -0.2 0.0 0.2 0.4

V(v)
Is DOS the Only Contribution?

• Why TMR is always maximum at zero


bias?
Bias Dependence

Co/Al2O3/Co

200

RAP
180
R(Ω)

RP
160

30
TMR(%)

25

20
-0.4 -0.2 0.0 0.2 0.4
Bias (V)
Inversed TMR at high bias

Si/.../Co/AlOx/ZrOx/Co 1520 4mV

R(Ω)
8 1480

1440
-150 -100 -50 0 50 100 150
4 H(Oe)
TMR(%)

910
500 mV

R(Ω)
900
0
Over-oxidized 890
Perfectly Oxidized -150 -100 -50 0 50 100 150
Under-Oxidized H(Oe)
-4
860 -800 mV
-0.8 -0.4 0.0 0.4 0.8
R(Ω)

Bias(V)
855
-150 -100 -50 0 50 100 150

H(Oe)
Inversed TMR at High Bias

• What determines the sign of TMR?


– Jullière Model I ↑↑ − I ↑↓ 2 PL PR
TMR = =
I ↑↑ 1 − PL PR
N↑ − N↓
P=
N↑ + N↓
– Negative Polarization around Fermi level in some transition metals, such as
Co, Ni etc.
– No negative TMR found in MTJ with Al2O3 barrier.
• Most inversed TMR are observed at zero bias.
– Bonding at the interfaces: J. M. D. Teresa et al. Science 286, 507 (1999)

Vous aimerez peut-être aussi