Tse-Ming Chen Department of Physics, NCKU
All-electric Spin Transistors
Figure courtesy of Intel
What is transistor (or MOSFET)?
If is said that if a cell is the building block of life,
then a transistor is the building block of the digital
era.
Semiconductor Scaling
Figure courtesy of wiki
7 nm 5 nm Pentium 4
22nm transistor
R. W. Keyes, IBM J. R&D 32, 26 (1988).
Doom of Moore’s Law (and the semiconductor industry)???
William Shockley, John Bardeen and Walter Brattain
1956 Nobel Prize
Beyond MOSFET
MOSFET
Quantum Computing
(superposition & entanglement)
Spintronic Computing
(spin)
1. Spin switch faster than THz 2. New functionality
3. Less energy consumption
Serge Haroche and David J. Wineland 2012 Nobel Prize
Photo courtesy of NASA Datta-Das spin FET
Giant Magnetoresistance (GMR) Albert Fert and Peter Grünberg
2007 Nobel Prize
Spin-transfer torque RAM STT-RAM
http://www.nature.com/milestones/milespin/index.html
Outline
What is spin FET and Why realization of it still seems to be impossible?
Our contribution to spin FET, and probably many
other spintronic devices
MOSFET vs SpinFET
Figure courtesy of Intel
Figure adapted from Hall and Flatte, APL 2006.
ON
OFF
Figure adapted from Nikonov, Proceedings of the IEEE 101, 2498 (2013).
Datta-Das Spin-FET
Key components of spin FET (proposed by Datta and Das in 1990)
Spin injection/detection – using ferromagnetism
Spin control – using electric field to control the spin rotation
Figure courtesy of NIMS (Japan)
Rashba spin-orbit coupling & spin precession
J. Nitta et al., PRL (1997).
Rashba spin-orbit coupling & spin precession
J. Nitta et al., PRL (1997).
Datta and Das showed in APL 56, 665 (1990) that for electron spins travelling in a 1D channel
the precession frequency is
and the phase shift for spins ballistically travelling in a distance L is
Spin Transistor Research Roadmap
199 0
199 0
199 7
Nitta et al. PRL 78, 1335 (1997)
In order to achieve Dq = 2p, L must be at least 2 mm for Da = 3E-12 eVm
199 0
199 7
Nitta et al. PRL 78, 1335 (1997)
In order to achieve Dq = 2p, L must be at least 2 mm for Da = 3E-12 eVm
2003 2002
Jedema et al. Science 416, 713 (2002)
199 0
199
7 2002 2007
Lou et al., Nature Physics 3, 197 (2007).
Appelbaum et al., Nature 447, 295 (2007)
2003
199 0
199
7 2002 2007
2009
Lou et al., Nature Physics 3, 197 (2007).
Appelbaum et al., Nature 447, 295 (2007)
Koo et al., Science 325, 1515 (2009)
However, the oscillatory resistance (on/off) variation is extremely small,
V/I is only 0.01 W Datta-Das spin
transistor reach a dead end?
Mean free path is ~1.6 mm when T < 10 K
Distance between two contacts is 1.6 and 1.2 mm.
2003
Obstacles to the realization of spin FET…
low spin-injection efficiency
limited spin lifetime
phase spread of accumulated spins
due to resistance mismatch [Schmidt et al., PRB 62, R4790 (2000)]
due to various precession angles resulted from different paths from source to drain picoseconds, due to D’yakonov-Perel’ scatterings
Obstacles again…
low spin-injection efficiency
limited spin lifetime
phase spread of accumulated spins
1. Increase spin injection efficiency of FM 2. Optical spin injection
3. ?
1D transport
1. Improve material quality to reduce scatterings, or
2. ?
due to resistance mismatch
due to various precession angles resulted from different paths from source to drain picoseconds, due to D’yakonov-Perel’ scatterings
Can we find a solution
1) to overcome all the obstacles and
2) for very-large-scale integration (VLSI)?
Our proposed solution is to employ
Figure courtesy of M. Switkes
Quantum Point Contacts
1D transport
conductance quantization at
integer multiples of 2e
2/h
Quantum point contacts & Spin injection
QPC spin injection through e-e interaction
0.7 structure – implying spin polarization.
Possible fully spin polarization (manifested as the 0.25 structure) was first proposed by T.-M. Chen et al., APL 93, 032102 (2008), and then was verified by T.-M. Chen et al.,
PRL 109, 177202 (2012). K. J. Thomas et al., PRL (1996). A. Lassl et al., PRB (2007).
Quantum point contacts & Spin injection
QPC spin injection through spin-orbit coupling
SO coupling results in two spin- polarized 1D subbands shifted
horizontally along the kx wavevector.
Only one spin-species is present in the either right- (+kx) or the left- (-kx) moving direction if the Fermi energy lies below the crossing point, thereby allowing for spin injection/detection.
Unable to tell from QPC conductance.
Quantum point contacts & Spin injection
QPC spin injection through
spin-orbit coupling & e-e interaction
Debray et al. in Nature Nanotech. 4, 759 (2009) proposed that the e-e interaction coexist with SOI when QPCs are operated near threshold.
Debray et al., Nat Nanotech (2009).
Ferromagnetism
Gate
Spin FET
Datta-Das
Quantum Point Contact
Gate Gate Gate
All-electric all-semiconductor Spin FET
Asymmetrically biasing the QPCs to generate a lateral SO coupling
QPC spin injector/detector with 100%
efficiency is created
Spin FET
Oscillatory on/off modulation up to 500%
Resistance modulation is 100,000 times greater than that observed by Koo et al. (Science 2009)
Obstacles resolved…
low spin-injection efficiency
limited spin lifetime
phase spread of accumulated spins
QPC as a spin injection/detection with 100% efficiency; all-electric
Two aligned QPCs set up a quasi-1D path
Two aligned QPCs collect ballistic transport electrons only
due to resistance mismatch
due to various precession angles resulted from different paths from source to drain picoseconds, due to D’yakonov-Perel’ scatterings
Spatial spin separation
The magnetic field, B
f, required to focus electrons at a distance L is
In the presence of SO, up- and down-spin electrons have
different momenta, leading to
different cyclotron orbits and
focusing peaks.
Influence of QPC conductance & temperature
Oscillation amplitude decreases with increases QPC conductance, consistent with the model of 1D + SO coupling
Our prototype spin FET survives up to ~ 17 K only
Working temperature could be improved by introducing a larger spin splitting using wet-etched QPC or InAs nanowires
Simultaneous magnetic and electrical control of spin precession
Our spin FET allows us to combine magnetic and electrical controls of spin precession.
Theoretical simulations performed
Good quantitative agreement
obtained between experiment
and theory
Challenges remain…
on/off ratio still not large enough
working temperature still low
Spin Hall Effect Transistor
Wunderlich et al., Science 330, 1801 (2010).
Non-ballistic spinFET
Schliemann et al., Phys. Rev. Lett. 90, 146801 (2003).
QPC spin filter used in other spin transistors
Acknowledgement
Material growth & Device fabrication
L. W. Smith, F. Sfigakis, M. Pepper, J. P. Griffiths, G. Jones, I. Farrer, H. E. Beere, D. A. Ritchie. T.-M. Chen
