electron spin resonance study in a wide parabolic quantum well

1
electron spin resonance study in a wideparabolic
quantum well Russ Bowers Associate Professor of
Chemistry University of Florida, Gainesville and
National High Magnetic Field Lab
2

• Acknowledgements
• Dr. Josh Caldwell presently at Naval Research
  Lab please visit poster!!
• Dr. Alexey Kovalev presently at Pennsylvania
  State University Dept. of Electrical
  EngineeringProfessor Guennadii Gusev,
  Departamento Física dos Materiais e Mecânica,
  University of Sao Paulo, Brazil
• funding
• National Science Foundation grant DMR-0106058.
• CNPq (Brazil) / NSF International Cooperation
  Program.
• University of Florida Division of Sponsored
  Research.

3
The ability to create nanometer-sized
nuclear spin distributions combined with long
solid-state nuclear spin lifetimes has important
implications for the future of dense information
storage, both classical and quantum. In addition,
control over highly localized interactions
between conduction electrons and lattice nuclei
may provide a means to manipulate such
information. - D. Awschalom PRL (2003) 91
article no. 207602
4
g-factor variation in a PQWtime-resolved optical
Faraday rotation detection
realization of electricallytuned g-factor in a
PQW
Poggio et al, PRL (2003) 91 article no. 207602
5
g-factor variation inbulk AlxGa1-xAs
In a PQW, the local g-factor varies along zdue
to variationof Al fraction.
conduction electronspin resonance data
Weisbuch and Hermann, PRB B15 (1977) 816.
6
Can g-factor control be integrated with
conventional solid state electronics? The
Awschalom demonstration employed optical
techniques
but in the context of EDESR?
The answer is YES!!
7
first demonstration of electrically detected
g-factor tuning
H.W. Jiang and E. Yablonivitchdemonstrated
g-factor tuning in a gated Al0.3Ga0.7As/GaAshete
rostructure.
However, the tuning range wasrelatively
smallabout .8 or -0.5.
H.W. Jiang and Eli Yablonovitch PRB, 64 (2001)
article no. 041307 (R).
8
Remotely doped wide parabolic quantum well (WPQW)

• Potential Vaz2 mimics a uniform 3D charge n
  proportional to the curvature.

• Effective potential in loaded well has no
  central barrier
• ? wide, uniform layer of high mobility carriers

• modest gate voltage can induce large
  displacements of wavefunction
• ? permits tuning of the electronic g-factor over
  wide range.

• unique properties for spin-based semiconductor
  devices
• ? i.e. spintronics, spin quantum computing

• model system to study 3D many body effects in
  high magnetic fields.
• ? high mobility, uniform electron density

• g-factor tuning in the context of EDESR has yet
  to be demonstrated in a WPQW.

9
Parabolic potential mimics a uniform change n,
electrons attempt to screen
A.J. Rimberg and R.M. Westervelt PRB 40 (1989)
3970.
10

• Calculation of the bound statesand electron
  density distribution for a 4000Å WPQW
  (AG662).
• (b) Calculated g-factor in each subband i.

11
4000Å WPQW structure(sample AG662)
25Å GaAs/AlxGa1-xAs superlattice
after illumination
before illumination
We 3000Å ? 4000Å
12
The tilt angle q might be expected to have a big
effect in a WPQW
high perpendicular field (q0o)
Usual expression for a 2DES at high magnetic
field (neglecting spin)
high parallel field (q90o)
Magnetic confinement dominatesover the quantum
well confinementbecause the cyclotron diameter
ltlt well width Identical to the energy levels
of a 3DES ? quasi-3D.
13
cyclotron spacing gtgt subband spacing
at high field, bulk Landau levels.
Gusev et al. PRB, 65 (2003) article no. 205316
14
at high field, bulk Landau levels.
Shayegan et al. Appl. Phys. Lett. 53 (1998) 791.
15
transport properties4000Å WPQW (AG662)
3DES
depopulaton of lowest LL
2DES
16
In high B?, changing the density increases the
occupancy of thehigher subbands.
low density, g-factor determinedby ground state
high density, g factor decreases due to
contribution ofexcited states
averaging assumes electron exchange rate is fast
compared to electron Larmor frequency.
17
experimental configuration
coax
m-wave field
B
Vxx
18
PC
YIG Oscillator
10-18 GHz
Oxford Heliox 3He probe
Doubling amp
Lockin 1
m-wave antenna
20-36 GHz
Output
coax
Hall bar patterned sample mounted on rotation
stage
modulator
12 Hz ref out
In
20-36 GHz,12 Hz.
B0
Doubling amp
Lockin 2
Brf
40-72 GHz,12Hz
rf coil
Output
10MW
5V 530 Hz out
coax
In
RF Synth.
50 MHz
19
g-factor temperature dependence
300Å GaAs/Al0.1Ga0.9Assquare well EA124
(60o) 5.5 T, n1
WPQW (AG662) 6.5T, q90o (parallel field)
g
g
-0.6 change
20
-0.6 change in g-factor over 2 ?10 K range
Lower g-factor at higher T is consistent with
increased igt1 sublevel occupancy.
21
temperature dependence of ESR linewidth
300Å QW (EA124,) 5.5 T, n1, q60o
WPQW (AG662), 6.5T
Trend in line broadening is opposite for B vs.
perpendicular orientation.
22
partially filled well
EF
regions of lower g-factor become occupied?
decreasing g-factor? g-factor broadening
increaseddensity
EF
completely full well
n(z)
z
23
DRxx (amplitude) temperature dependence
WPQW (AG662)
300Å QW (EA124,) 5.5 T, n1, q60o
up down

• Much weaker T-dependence in parallel
  orientation.? Signal could be observed all the
  way to 10 K.
• in 2DES, similar to narrow QW sample.

24
g-factor anisotropy in a 150Å GaAs/Al0.35Ga0.65As
QW M. Dobers, K. v. Klitzing, G. Weimann, Solid
State Comm. 70 (1989) 41-44.
N0
25
G-factor anisotropy in a 2DES is due
tonon-parabolicity of conduction band
Energy spectrum
diamagnetic term
Spatial extent of wave function.
Linear relationship between g-factor andLandau
level energy was assumed
26
n36.65 GHz
g-factor anisotropy in the WPQW
gg0-cB
g-factor anisotropy 5. monotonic decrease
with q
27
qualitative interpretation
g-factor decreases monotonically with increasing
tilt angle. ? diamagnetic term
dominates at all angles in lowest landau
level. especially true for WPQW ?
much larger in 400nm well compare to 15nm
well
Anisotropy of g-factor in B smaller than
expected

• calculated g0 should be -0.275. actual -0.428
• selectivity for detection in central layers of
  high mobility

28
dynamic nuclear polarization in PQW
DNP in the q3DES closely resembles DNP in the
2DES WPQW and 30nm QW Direction of DNP
establishes glt0.
29
T1n in parallel and perpendicular fields
30
Overhauser shift decay in high parallel field
Temperature independent nuclear spin relaxation
in q3DES
31
3DES, T1n temperature independent.
2DES, Korringa law
A. Berg, M. Dobers, R.R. Gerhardts, K. v.
KlitzingPRL, 24 (1990) 2563.
for kT ltlt G
where
32
filling factor dependence of T1n around n1
WPQW (AG662)
150Å Al.35Ga.65As/GaAs QW
A. Berg, M. Dobers, R.R. Gerhardts, K. v.
KlitzingPRL, 24 (1990) 2563.
33
Fig. 11.
Frequency swept MDENDOR - see poster for details
5.0
4.8
75As
69Ga
71Ga
4.6
4.4
4.2
DRxx (W)
4.0
3.8
3.6
3.4
3.2
42.6
42.7
59.8
59.9
75.9
76.0
MHz
But no 27Al signal!!!
34
Fig. 12.
(a) AG662
(b) AG662
DRxx (W)
(c) EA124
(d) EA124
35
The solid blue curve is the angle dependence
observed in a sample subjected to planar
stress. Green symbols represent the splitting
in the 30nm QW. Splitting due to selective
detection in the band bending region?
MDENDOR
OPNMR
36
The lowest Landau level, illustrating the change
in EFG with filling factor.
Representative spectra illustrating the filling
factor dependence of the 75As spectrum in EA124.
37
conclusions

• Increased linewidth in parallel field vs.
  perpendicular field.
• g-factor anisotropy, 5
• g-factor decreases monotonically over 0?90o.
• in parallel field, detection is selective to
  higher mobility layers near center.
• Mechanism of EDESR may be different in q3DES
  than in the 2DES
• No Al-27 NMR despite good SN on other isotopes.
• Korringa Relaxation in 2DES, T-independent
  Relaxation in q3DES
• broadening in MDENDOR spectrum could result from
  heterogeneousband bending across z.

38
g-factor tuning range in a WPQW is large
application of electric field
39
Effect of illumination on the g-factor Increasin
g the densityis expected to decreasethe
g-factor However, results inconsistent
5s 7s 10s 60s