Transcript enee702spin

Motivation: Embracing Quantum Mechanics
34 Years of History: More, More, Moore
http://www.intel.com/research/silicon
# transistor per IC
High Performance Processor
@ 90nm Technology Node
• 256 million transistors
• 37nm gate length
• PNO gate: 10 nm EOT
• NiSi2/Poly gate
• 8 levels Cu with low-k
interlevel dielectric
Feature Size
Transistor Density
Chip Size
Transistors/Chip
Clock Frequency
Power Dissipation
Fab Cost
WW IC Revenue
WW Electronics
Revenue
1970
Today
Change
6 um
90 nm
~10 mm2
1000
100 kHz
~100 mW
~$10 M
$700 M
$70 B
~400 mm2
200 M
> 1 GHz
~100 W
>$1 B
$170 B
$1.1 T
70x Reduction
5000x Increase
40x Increase
200,000x Increase
>10,000x Increase
~1000x Increase
>100x Increase
240x Increase
16x Increase
• Miniaturization
 power dissipation
 short channel effect
 statistical error in dopant distribution
 quantum mechanical effects
– ballistic
– tunneling
• Alternative schemes to embrace
quantum mechanical effects
– low power
– fast
– new functionality, e.g. spin
• Intel and IBM embracing spintronis
1
Successful Examples of Spintronics
metal-based ferromagnetic devices
spin valve
hard drive read head
GMR
magnetic tunneling
junction memory
• Rapid transition from discovery to commercialization
– 1988 giant magneto resistance (GMR)
– 1998 IBM read head extends storage from 1 to 20Gbits ($100B)
– 2004 Motorola MRAM with 10ns access time
– 2010 10Gbit memory chip projected
• Non-volatile, no wearout, fast write time, fast read time, low energy for
writing, radiation-hard
2
www.sematech.org/meetings/ past/20021028/17_Hummel_Mram.pdf
Objective
• Demonstrate the essential elements required for
ballistic spin devices – semiconductor spin-FET
Essential elements:
electrical (rather than optical)
100nm
1. injection of highly spinpolarized electrons
2. manipulation of electron spin
orientation during coherent
transport
spin
InAs ballistic
spin transistor
length
3. spin-sensitive detection
• All existing semiconductor devices operate in the diffusive transport regime,
where scattering results in heat dissipation and limits frequency response
• Spin transport in the ballistic regime offers opportunities which are
heretofore unexplored and unexploited
3
Semiconductor spintronics: Spin FET
Datta and Das, Applied Physics Letters, 56, 665 (1990); 816 citations
• Ferromagnetic
source injects spin
polarized electrons
• Gate controlled spin
precession through
Rashba spin-orbit
coupling in 2D
channel
• Ferromagnetic drain that
provides:
 low resistance if spins are
parallel
 high resistance if spins are
anti-parallel
• Not experimentally demonstrated yet
• Key processes face challenges
4
Our Technical Approach: Electrical + InAs
Key obstacle
• to study spin manipulation in InAs, ferromagnetic/InAs needed
• to develop FM/InAs hybrid junction, spin detection needed
• optical detection is not practical, electrical detection needed
• FM/InAs hybrid junction needed for electrical detection
Our solution: a novel configuration for electrical spin injection and detection
Manipulation
Injection
Electrical tunability of
Rashba SO coupling
in InAs
Efficiency of spin
transport across
FM/InAs junction
Challenges
• determination of R
5
Detection
Electrical spin
detection
Challenges
• optimal ferromagnetic
• control of R
• Interface integrity
• optimization of R
• etching selectivity
Current Status: Ferromagnetic/GaAs Hybrid Junction


1 mA, 2 V
4.5 K
Magnetic Metal / Tunnel Barriers
spin-LED EL Intensity
(NRL)
32%
Fe
meV
55meV
3T
2
Fe
1.53
1.54
1.55
1.56
Schottky
PQW
Al2O3
1
0
GaAs
40%
GaAs
Metal oxide
Al2O3
Tunneling
PQW to 40% (lower bound, 5 K)
Photon Energy (eV)

6
Fe
AlGaAs
GaAs
APL 80, 1240 (2002) (NRL)
APL 82, 4092 (2003) (NRL)
Benchmark performance at 5K
• well-defined system state
• detector “purely” excitonic
• most well studied temperature
Past Spin Research : Optical + GaAs
• GaAs band gap is ideal for optical experiments
• optically generate spin-polarized electrons
• detect spin populations with circularly polarized PL
Pump-Probe Studies of Spin Precession
D. D. Awschalom, UCSB
Spin-Light Emitted Diode
Fe/barrier/GaAs
NRL patent (1999)
Pcirc


7
PQW
Rashba Spin-Orbit Coupling
Structural Inversion Asymmetry
• In nature, spins are controlled with
magnetic field.
• Thanks to spin-orbit coupling, now
spins can also be controlled by
using conventional gates.
rest frame


relativistic
effect
z
z
2DEG in
quantum well
confinement
potential
moving frame
Bin


Bin
In the absence of an external B
8
H R  αR(  x k y   y k x )  σ g μB B*
Optimal Material for Rashba Spin-Orbit Coupling
  
H R  αR (σ k ) z  σ g * μB Bin  SO  αR k F
αR : Rashba coefficient
2ħSO
(meV)
InAs
2  SO
Si
GaAs
BIN (Tesla)
Advantage of InAs
• Larger g* (15 vs. 2 vs. 0.44)  Larger spin splitting (meV)
• Smaller m* (0.023 vs. 0.067 vs. 0.19)  Larger quantization energy (10meV)
• Higher RT mobility (40k vs. 8k vs. 1.4k)  Longer mean free path (700nm)
9