Theory of individual dopants in semiconductors
Transcript of Theory of individual dopants in semiconductors
Michael E. FlattéOptical Science and Technology CenterDepartment of Physics and Astronomy
Department of Electrical and Computer EngineeringUniversity of Iowa
supported by AFOSR and ARO
Theory of individual dopants in semiconductors
review of “solotronics”: Nature Materials 10, 91 (2011)
V. R. Kortan, C. E. Pryor, C. Şahin, J. M. Tang (U. Iowa)P. M. Koenraad, A. Silov, J. Bocquel, J. K. Garleff, J. van Bree, W. Van Roy,
C. Celebi, A. P. Wijnheimer, A. Yakunin (TU/e)R. Wiesendanger, J. Wiebe, F. Marczinowski, F. Meier, M. Morgenstern (Hamburg)
introduction to dopantsshallow levelsionization and effect of dopants on transportpositioning of dopants, length scalesenvelope function theory and tight-binding theorydeep levels and lattice symmetry (radial, angular)
control of confinement and changes in angular momentummagnetic deep levels (Mn and Fe in GaAs)effect of lattice symmetry, even within effective mass modelsstrain and electric field effects on magnetic deep levels
Outline
local bonding symmetry remains the sameideally, if the dopant is on the same row, it behaves as an extra proton and electron
+-
a simple change to the materialsubstitutional dopants with different valence
GaAs As
As
As
As As
As
As
Si
“hydrogenic" impurities in semiconductors
Group III: Al,Ga,InGroup IV: C,Si,GeGroup V: P,As,Sb
+-+-
theory of hydrogenic dopants
built from general effective mass theory PR 97, 869 (1955) EConduction
EValence
m becomes m*ε becomes εrelative
ERydberg becomes (m*/ε2relative)ERydberg
+-
donor states
theory of hydrogenic dopants
built from general effective mass theory PR 97, 869 (1955) EConduction
EValence
m becomes m*ε becomes εrelative
ERydberg becomes (m*/ε2relative)ERydberg
+-
donor states
ionization
ionization energies ~ meV central cell correction
Length scales of defect statesMetalInsulator
electronic states localizedon the atomic scale
electronic states extendedthrough the entire solid
short length scales = a ~ aB
a
short length scales = kF
-1 ~ aB
E
k kF
E
disorder
Semiconductor
dopant
localized electronic statesare “big”, R>> a
B
E
k
extended electronic statesvary slowly, k
F-1 >> a
B
Large distances characteristic of electronic states in semiconductors imply
Long-distance interactions and perturbations (e.g. Thomas-Fermi length)Interactions very sensitive to external perturbations (E, B fields)
using dopants in transistors
emitter base collectoron
off
Sze
, Se
mic
on
du
cto
r Ph
ysic
s
junction FET+- +-+- +-
+- +-+- +-
+ ++ +
on
off
trapping/detrapping in MOSFETs Tsai et al.,
APL 61, 1691 (1992)
+- +-using dopants in transistors
emitter base collectoron
off+ +-
Asenov et al, IEEE Trans. Elec. Dev. 50, 1837 (2003)
Shinada, Nature 437, 1128 (2005)
charge device reproducibility and dopants
Transport through a few dopants
towards asingle-dopant transistor
Tan et al.,Nano Lett. 10, 11 (2010)
electronic levelsassociated with
individual dopantspermit resonant
transport
conductance
gate voltage
sour
ce-d
rain
vol
tage
single-dopant transistor
electronic levelsassociated with a
single dopantpermit resonant
transport
conductance
gate voltage
sour
ce-d
rain
vol
tage
Fueschsle et al., Nature Nanotechnology 2012
Nature 442, 436 (2006)
Positioning Mn in the surface layer
valence edge andacceptor state
LDOS
positioning a single dopant spin
http://chemwiki.ucdavis.edu/
point-group symmetries and dopants
s
p
d
f
schematic wave functions of different symmetry
breaking degeneracy (radial potential)
Coulomb potential: energy only depends on shell #, not on which spherical harmonicgeneral central potential: energy only depends on shell # and total angular momentum l
Central
l = 0l = 1l = 2l = 3
Coulomb
all l
breaking degeneracy (angular potential)
crystalisotropicpotential
all m
As As
As
As
Si
lattice: split among the m statesalso called “crystal field”
non-magnetic: equal amount of +m and -m
breaking degeneracy (tetrahedral group)s
As As
As
As
Si
p
d
f
A1
T2
T2
E
?
deep levels - how do they form?As As
As
As
SiZunger: they are atomic levels, set relative to the vacuum
s
p
A1
T2
related observation: symmetry of tetrahedral bonds below is A1+T2
Hjalmarson et al: PRL 44, 810 (1980)
deep level is a host-like antibonding state
accessible atomic features in solids“Shallow states” - meV rather than eV
due to small effective mass and large dielectric screening
responsive to external fields
“Core states” ~1 eV - d-states, spin, spin-orbit
Nuclei (hyperfine)very long spin coherence times
Dots have hyperfine, shallow states
- missing the core, lower symmetry
Fe2+ to Fe3+
transition in GaAs
change in d-shell occupancy
S=2 to S=5/2
PRB 87, 075421 (2013)
Si in GaAsGarleff et al., PRB 84, 075459 (2011)
STM-driven electronic transitions of dopant core states
produces HUGE (x100) spin-orbit interactions EPL 98, 17013 (2012)
structural changes
Michael E. FlattéOptical Science and Technology CenterDepartment of Physics and Astronomy
Department of Electrical and Computer EngineeringUniversity of Iowa
supported by AFOSR and ARO
Spin-orbit correlations of dopants in semiconductors
review of “solotronics”: Nature Materials 10, 91 (2011)
V. R. Kortan, C. E. Pryor, C. Şahin, J. M. Tang (U. Iowa)P. M. Koenraad, A. Silov, J. Bocquel, J. K. Garleff, J. van Bree, W. Van Roy,
C. Celebi, A. P. Wijnheimer, A. Yakunin (TU/e)R. Wiesendanger, J. Wiebe, F. Marczinowski, F. Meier, M. Morgenstern (Hamburg)
introduction to dopantsshallow levelsionization and effect of dopants on transportpositioning of dopants, length scalesenvelope function theory and tight-binding theorydeep levels and lattice symmetry (radial, angular)
control of confinement and changes in angular momentummagnetic deep levels (Mn and Fe in GaAs)effect of lattice symmetry, even within effective mass modelsstrain and electric field effects on magnetic deep levels
Outline
breaking degeneracy (tetrahedral group)s
As As
As
As
Si
p
d
f
A1
T2
T2
E
?
deep levels - how do they form?As As
As
As
SiZunger: they are atomic levels, set relative to the vacuum
s
p
A1
T2
related observation: symmetry of tetrahedral bonds below is A1+T2
Hjalmarson et al: PRL 44, 810 (1980)
deep level is a host-like antibonding state
p-d hybridized deep levelsAs As
As
As
deep level is a host-like antibonding state
Sn
p-d hybridized deep levelsAs As
As
As
p
deep level is a host-like antibonding state
d
T2
E
Vogl and Baranowski, Acta Phys. Pol. A 67, 133 (1985)
Mn
Mn dopant in GaAscrystalfield
d
T2
E
d-exchange
T2
E
T2
E106
43
3
2
2
p-d interactionwith As
dangling bonds
AB-T2
E
AB-T2
EB-T2
B-T2
conduction band
valence band
S=5/2
1 heavyhole
j=3/2
sL
groundstateJ = 1
R0
R1
Ga
As
Mn
Mn in GaAs: single-ion p-d hybridization
p-d hybridization well establishede.g. for GaMnAs TcDietl et al., Science 287, 1019 (2000)
single ion: mean field theory cannot be usedwe want the spatial structure
PRL 92, 047201 (2004)
[1-gV] [G] = [g] x
x''
x''
x'
x
x'
[G] = [1-gV] [g] -1
G(x,x';ω) = g(x,x';ω) + g(x,x'';ω)V(x'')G(x'',x';ω)Σx''
sp3 tight-binding Hamiltonian with single-ion p-d exchange
R0
R1
Ga
As
Mn
Mn in GaAs: single-ion p-d hybridization
p-d hybridization well establishede.g. for GaMnAs TcDietl et al., Science 287, 1019 (2000)
single ion: mean field theory cannot be usedwe want the spatial structure
L=1
PRL 92, 047201 (2004)
sp3 tight-binding Hamiltonian with single-ion p-d exchange
experiment/theory comparison for Mn/GaAs
experiment/theory comparison for Mn/GaAs
theoretical prediction
PRL 92, 047201 (2004)
experiment/theory comparison for Mn/GaAs
PRL 92, 216806 (2004)
calculations
STM experiment Nature 442, 436 (2006)
theoretical prediction
PRL 92, 047201 (2004)
experiment/theory comparison for Mn/GaAs
Mn: InAs spectra - spin-orbit-split states
PRL 99, 157202 (2007)
Mn: InAs spectra - spin-orbit-split states
PRL 99, 157202 (2007)
hybridization of Mn states
R0
R1
Ga
As
Mn
core d states split in tetrahedral crystal field into t2 states - do hybridize via pdσ hoppinge states - do not hybridize via pdσ hopping
69
by the Hamiltonian 4.3.3 the total d-like component is spherically symmetric [see
figure 4.4(Total)]. In order to obtain the anisotropy, the problem has to be solved in
the cubic approximation.
In the cubic approximation the symmetry of the wave-function 4.3.5 is reduced.
If the condition 4.3.2 is satisfied, the acceptor wave-function transforms according to
the same irreducible representations as the top of the valence band, Γ8 point
Γ±8 × Γ±
8 = 2Γ+15 + 2Γ+
25 + Γ+12 + Γ+
2 + Γ+1 . (4.3.14)
The wave-function that transforms according to these representations has the form
Ψ3/23/2 =
cΓ1
2R0
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
Y0,0
0
0
0
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
+cΓ25
2√
2R2
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
0
−2Y2,1
Y2,2 − Y2,−2
0
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
+cΓ12
2√
2R2
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
√2Y2,0
0
Y2,2 + Y2,−2
0
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
, (4.3.15)
Ψ3/21/2 =
cΓ1
2R0
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
0
Y0,0
0
0
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
+cΓ25R2
2√
2
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
2Y2,−1
0
0
Y2,2 − Y2,−2
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
+cΓ12
2√
2R2
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
0
−√
2Y2,0
0
Y2,2 + Y2,−2
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
. (4.3.16)
The angular dependence of the spherical functions Y (ϕ, θ) as a function of the az-
imuthal angle θ with respect to the quantization direction is shown in figure 4.4.
In the equations 4.3.15 and 4.3.16 the angular part of the d-component includes
terms that transform according to the Γ12 (dx2−y2−like) and Γ25 (dxy−like) repre-
sentations of the tetrahedral point group. Their corresponding coefficients are the
constants cΓ12 and cΓ25 , respectively, whose ratio is denoted here as
η = cΓ12/cΓ25 , 0 ≤ η ≤ 1. (4.3.17)
The coefficients cΓ12 , cΓ25 and cΓ1 in the equations 4.3.15 and 4.3.16 are usually
evaluated variationally and should allow for the normalization condition given by the
effective mass wave functions (e.g. Bir, Pikus)
69
by the Hamiltonian 4.3.3 the total d-like component is spherically symmetric [see
figure 4.4(Total)]. In order to obtain the anisotropy, the problem has to be solved in
the cubic approximation.
In the cubic approximation the symmetry of the wave-function 4.3.5 is reduced.
If the condition 4.3.2 is satisfied, the acceptor wave-function transforms according to
the same irreducible representations as the top of the valence band, Γ8 point
Γ±8 × Γ±
8 = 2Γ+15 + 2Γ+
25 + Γ+12 + Γ+
2 + Γ+1 . (4.3.14)
The wave-function that transforms according to these representations has the form
Ψ3/23/2 =
cΓ1
2R0
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
Y0,0
0
0
0
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
+cΓ25
2√
2R2
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
0
−2Y2,1
Y2,2 − Y2,−2
0
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
+cΓ12
2√
2R2
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
√2Y2,0
0
Y2,2 + Y2,−2
0
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
, (4.3.15)
Ψ3/21/2 =
cΓ1
2R0
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
0
Y0,0
0
0
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
+cΓ25R2
2√
2
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
2Y2,−1
0
0
Y2,2 − Y2,−2
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
+cΓ12
2√
2R2
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
0
−√
2Y2,0
0
Y2,2 + Y2,−2
∣∣∣∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣∣∣∣
. (4.3.16)
The angular dependence of the spherical functions Y (ϕ, θ) as a function of the az-
imuthal angle θ with respect to the quantization direction is shown in figure 4.4.
In the equations 4.3.15 and 4.3.16 the angular part of the d-component includes
terms that transform according to the Γ12 (dx2−y2−like) and Γ25 (dxy−like) repre-
sentations of the tetrahedral point group. Their corresponding coefficients are the
constants cΓ12 and cΓ25 , respectively, whose ratio is denoted here as
η = cΓ12/cΓ25 , 0 ≤ η ≤ 1. (4.3.17)
The coefficients cΓ12 , cΓ25 and cΓ1 in the equations 4.3.15 and 4.3.16 are usually
evaluated variationally and should allow for the normalization condition given by the
hybridization of Mn states
R0
R1
Ga
As
Mn73
8x8 nm2; ML 6th=1 =0.8
=0.6 =0.4
=0.2 =0
[001]]101[
Figure 4.6: Calculated cross-sections of the Mn acceptor wave-function in GaAs bya {110} plane for various values of the cubic parameter η. Mn is located in the 6th
subsurface layer counting the surface layer as zero. 1ML = 0.2 nm. Images havesizes of 8 × 8 nm2.
core d states split in tetrahedral crystal field into t2 states - do hybridize via pdσ hoppinge states - do not hybridize via pdσ hopping
PRL 92, 216806 (2004)
effect of spin-orbit interaction on shape
Mn:GaAs Cd:GaPBinding energy
102 meVBinding energy
113 meV
crystalfield
d
d-exchange
E
E
T2
E
p-d interactionwith As
dangling bonds
AB-T2
AB-T2
B-T2
B-T2
conduction band
valence band
T2
T2
~ -1.5 eV
~ -1.25 eV
~ -2.0 eV
~ -4.0 eV
~ 0.35 eV
~ -2.0 eV
~ -0.5 eV
E ~ -4.0 eV
E ~ 0.35 eV
~ -3.5 eV
~ 0.75 eV
~ -1.5 eV
~ -0.5 eV
Fe dopant in GaAs
crystalfield
d
d-exchange
E
E
T2
E
p-d interactionwith As
dangling bonds
AB-T2
AB-T2
B-T2
B-T2
conduction band
valence band
T2
T2
~ -1.5 eV
~ -1.25 eV
~ -2.0 eV
~ -4.0 eV
~ 0.35 eV
~ -2.0 eV
~ -0.5 eV
E ~ -4.0 eV
E ~ 0.35 eV
~ -3.5 eV
~ 0.75 eV
~ -1.5 eV
~ -0.5 eV
Fe dopant in GaAs
e.g. Mahadevan and Zunger, PRB 69, 115211 (2004)
Including the d states
Energy (eV)
Loca
l Den
sity
of S
tate
s (1
/eV
)
d(t2)
total spds*
p
(a)
total sp�
(b) (c) (d) (e)
[001
]
[110]
33.9 A
Experiment sp3 sp3d5d5s*Energy (eV)
Loca
l Den
sity
of S
tate
s (1
/eV
)
d(t2)
total spds*
p
(a)
total sp�
(b) (c) (d) (e)
[001
]
[110]
33.9 A
asymmetry in [001] is similar, but more pronounced X feature
hybridization of Fe states
R0
R1
Ga
As
Fe
core d states split in tetrahedral crystal field into t2 states - do hybridize via pdσ hoppinge states - do not hybridize via pdσ hopping
STS
1
24
A comprehensive study of single Fe impurities in GaAs by X-STM
Figure 1.17: Spatially resolved I-V spectroscopy experiment was performed on the sameFe impurity at 4 K. a) dI/dV cross-section taken across the Fe impurity along the [001]direction. Two peaks are resolved in the bandgap. b) dI/dV maps taken at 0.46 V and1.02 V. The spatial extent of these states is consistent with those expected for the deepFe states of e (lower energy) and t2 symmetry (higher energy).
in the bandgap. dI/dV (x,y) maps taken at 0.46 V and 1.02 V, energy positionscorresponding to the two peaks, are presented in Fig. 1.17b. The spatial extent ofthese two states is clearly different. The lower energy state is strongly localizedon the Fe impurity itself. The wave function of this state almost isotropic andextend over ≈ 0.75 nm. The spatial extent of this state is consistent with the oneexpected for the deep Fe states of e symmetry from the group theory and thetight binding calculation presented in Fig. 1.15. The higher energy state is alsomainly localized on the Fe impurity itself but presents extensions in a cross-likeshape. The wave function of this state is anisotropic and extends over ≈ 2.5 nmalong the [001] direction and 2 nm along the [110] direction. These two statesdo not exhibit the even and odd symmetry expected for states resulting from asplitting of the t2 state by the effect of the reconstructed surface. Instead, thespatial extent of this state is consistent with the one expected for the deep Festates of t2 symmetry from the group theory and the tight binding calculationpresented in Fig. 1.15.
To summarize, two levels related to Fe are found the bandgap of GaAs in theI-V spectroscopy data. From the relative energetic position of these Fe states aswell as their spatial extent, they are here attributed to the Fe states e and t2
15
11.25
7.5
3.75
0
Current (pA) arb. units
[0.88-1.18] V
t2 t2
a) b)
0.63 V
e
[001]
e
[110]
2
1.5
1
0.5
0
spds*theorycurrent (pA)
hybridization of Fe states
R0
R1
Ga
As
Fe
t2
e
core d states split in tetrahedral crystal field into t2 states - hybridize via pdσ hoppinge states - hybridize via pdπ hopping
Fe in GaAssp3 + d orbital on impuritynot larger pdS because would cause small amount of T2 ldos on NN E level to have some ldos at pdP=0
pdP x0.0 pdP x0.010 pdP x0.020 pdP x0.030 pdP x0.040 pdP x0.050 pdP x0.060 pdP x0.070 pdP x0.080
E : 0
.512
eV
T2 : 0
.876
eVla
yer o
f Fe
E : 0
.512
eV
T2 : 0
.876
eV
l3 a
yers
abo
ve o
f Fe
pdS x0.080
16 A
[001]
[110]
increasing pdπ hopping strength
introduction to dopantsshallow levelsionization and effect of dopants on transportpositioning of dopants, length scalesenvelope function theory and tight-binding theorydeep levels and lattice symmetry (radial, angular)
control of confinement and changes in angular momentummagnetic deep levels (Mn and Fe in GaAs)effect of lattice symmetry, even within effective mass modelsstrain and electric field effects on magnetic deep levels
Outline
a quantum dot locally strains the material surrounding the Mn acceptor
Manipulating the Mn wave function
Mn1
Mn2
GaAs host
[00-1]
[-110]
3.3 nm
Mn1Mn1(X(X--shape)shape)
Mn3
Mn2Mn2(S(S--shape)shape)
Mn3Mn3(Mirrored S(Mirrored S--shape)shape)
InAsInAsQD1QD1
InAsInAsQD2QD2
Nature Materials 6, 512 (2007)
Mn2
m2
m1
1.3 nmMn1 1.3 nm
m2
m1
[00-1]
[1-10]
TBM
0.0
-1.0
-5.0
-2.0
-3.0
-4.0
low-energy Hamiltonian for manipulation
magneto-elastic constants derived from this effective Hamiltonian agree withMasmanidis et al,
PRL 95, 187206 (2005)
Measured - STM
Tight-binding modelNature Materials 6, 512 (2007)
Mn: InAs spectra - spin-orbit-split states
PRL 99, 157202 (2007)
Mahieu et al., PRL 94, 026407 (2005)hybridization of impurity state with excited states?
Jancu et al., PRL 101, 196801 (2008)hybridization of impurity state with intrinsic surface states?
residual asymmetry at surface - strain?
experiment
theory (no strain)
theory (with strain)
PRL 104, 086404 (2010)
explains residual asymmetry at surfacePRL 104, 086404 (2010)
Mn:GaAs
different hosts/dopants
Cd:GaP Zn:GaP
all in 5th layer
PRL 104, 086404 (2010)
L=1
E-field control of a Mn spin in GaAsPRL 97, 106803 (2006)
1 Tesla effective field for 40 kV/cm electric field
Depth of Mn [atomic layer]PRB 82, 035303 (2010)
splitting by surface E-field
Surface field splits the J=1 state into three separate peaks
Detecting Mn spin orientation in GaAs
0 0.1 0.2 0
1
2
3
4
5
6
7
Loca
l den
sity
of s
tate
s (eV
-1 )
px + i py
0 0.1 0.2 Energy (eV)
pz
0 0.1 0.2
px − i py
Mn
As
Ga
Ga
PRL 92, 047201 (2004); PRB 72, 161315(R) (2005)
eV−1A−310−4 10−3
[ ]001
110[ ]−
1 nm
[001]
[100] [010]
PRB 72, 161315(R) (2005)
HB⊥ ¼ gμBðBxσx þ ByσyÞ
¼ gμBffiffiffi2
p B⊥e−iϕBðjþ 1ih0jþ j0ih−1jÞ þ H.c.; (2)
HE⊥ ¼ −d⊥Exðσ2x − σ2yÞ þ d⊥Eyðσxσy þ σyσxÞ
¼ −d⊥E⊥eiϕE jþ 1ih−1jþ H.c.; (3)
where jii is defined tobe jms ¼ ii,B⊥ ðE⊥Þ andϕB ðEÞ are themagnitude and phase, respectively, of the magnetic (electric)field in the plane transverse to the c axis, and H.c. denotesthe Hermitian conjugate. The main difference between HB
⊥and HE
⊥ is that HB⊥ connects triplet pairs with Δms ¼ %1,
whereasHE⊥ connects triplet pairs withΔms ¼ %2. As such,
in the same way that applying resonant transverse magneticfields can be used to drive magnetic-dipole (Δms ¼ %1)transitions, resonant transverse electric fields can be used todrive magnetic-dipole forbidden (Δms ¼ %2) transitions.Our experiments use both ac electric and magnetic field
control, for which we use separate driving elements. Open-circuit interdigitated metal electrodes on the chip's topsurface are used to drive transverse electric fields betweenadjacent digits [Fig. 1(b)], and a short-circuited striplinebeneath the chip is used to drive transverse magnetic fields
over the electrode region. A flow cryostat cools our deviceto the temperature T ¼ 20 K and a permanent magnetprovides a static B∥. QL1 color centers were produced inour 6H-SiC substrates via a carbon implantation andannealing process designed to generate defects in a400 nm thick layer immediately below the surface (seeSupplemental Material [33]). The QL1 spins betweenadjacent electrode digits are optically addressed by non-resonantly pumping their 1.09 eV near-infrared opticaltransition with 1.27 eV laser light [see Fig. 1(c) and theSupplemental Material [33] for details]. In our experiments,we simultaneously address approximately 104 QL1 spins.Future efforts to extend optical addressability to the single-spin limit in SiC will use strategies such as materialspurification [35], high-efficiency infrared detection, andlocalized ion implantation [10].Much like the NV center in diamond [34], QL1 has a
spin-dependent optical cycle, which allows nonresonantlaser illumination to both polarize and read out its ground-state spin. Because its photoluminescence intensity (IPL)depends on whether its spin state is j0i or j% 1i, we cantrack the QL1 spin dynamics by the measuring differentialphotoluminescence (ΔIPL) between an initial state and onethat has been evolved by magnetic or electric field pulses.
0.8 1 1.2 1.4 1.6 1.8
x 109
x 10-4
frequency (GHz)
(a)
(d)
778 MHz
QL1
QL2 QL6
QL5
∆IP
L / I
PL
x 10
5
10
0
-10
10
0
-10
-20 0.8 1.0 1.2 1.4 1.6 1.8
frequency (GHz)
B|| = 0 G T = 20 K
B|| = 139 G T = 20 K
+1
0
hD
2gµBB||
∆ms = ±1 ∆ms = ±2 ∆ms = ±1
∆ms = ±2
electric ∆ms = ±2
magnetic ∆ms = ±1
QL3
−1
c
dichroic
photodiode laser
PL
10 µm
6H-SiC Au (b) (c)
FIG. 1. (a) The orbital ground-state spin structure of the QL1 defect, with Δms ¼ %1 transitions (orange arrows) and the Δms ¼ %2transition (blue arrow) indicated. (b) Scanning electron microscope image of the electrode pattern. (c) QL1 spins localized within a 400 nmthick layer immediately beneath the 6H-SiC surface are optically pumped with a 1.27 eV laser in a 1.5 μm diameter spot, addressing ∼104QL1 defects at once. Photoluminescence is filtered from the pump laser with a dichroic mirror and is measured with a photodiode. Spins aredriven electrically by the electrodes and magnetically by the stripline. The electrode pattern from part (b) maps to the greendashed parallelogram. (d) The optically detected magnetic resonance (ODMR) signal when the stripline is driven at B∥ ¼ 139 G (upper)and B∥ ¼ 0 G (lower). The two Δms ¼ %1 resonances are shaded orange, and the Δms ¼ %2 resonance (at 778 MHz, shaded blue) ismagnetic-dipole forbidden and not seen in ODMR.
PRL 112, 087601 (2014) P HY S I CA L R EV I EW LE T T ER Sweek ending
28 FEBRUARY 2014
087601-2
Klimov et al, PRL 112, 087601 (2014)
electrical drive of spin-1 single defects
elements of dopant theoriesshallow or deep? hydrogenic? sharp distinction between the twocrystal field splittings and hybridizations dominate deep level defect propertiesenvelope function/effective mass theories and tight-binding theories are closer in results and assumptions than one might expect
control of confinement and changes in angular momentumstrain and electric field control of defect spin distinguishing between different symmetry d states is possible theoretically and experimentally via wave function extent
Highlights