Excitation of A Three-Level Si:P Donor by A Gaussian Pulse
Excitation of A Three-Level Si:P Donor by A Gaussian Pulse
Excitation of A Three-Level Si:P Donor by A Gaussian Pulse
()
Notations used in this report: |0i is the 1sA state, 2p , , = x, y, z, is the 2p- wavefunction for the valley
(including the cos(k .r) term). The explicit forms of these wavefunctions are
" s ! #
1 1 x2 + y 2 z2
|0i = p 2 exp + 2 cos(kz) + cyclic permutations , (1)
3 a0 b0 a20 b0
with a2 = 5.5 nm and b2 = 3.4 nm. The energy of the 2p levels is E2 = 6.1 meV.
where
1 h i
|e2 i = x (|2p(y)
x i + |2p(z)
x i) + y (|2p(x)
y i + |2p(z)
y i) + z (|2p (x)
z i + |2p(y)
z i) (23)
2
(y)
and l2 = 2 h0|x|px i. The magnetic field-split 2p states for the x, y, z valleys are
(x) 1
|2p i = |2py(x) i i |2pz(x) i , (24)
2
(y) 1 (y)
|2p i = |2pz i i |2p(y)x i ,
2
(z) 1 (z)
|2p i = |2px i i |2p(z)y i .
2
One can write |e2 i as the superposition
1
|e2 i = (|2p+ i + |2p i) , (25)
2
where
1 h (z) (y)
(x) (z)
(y) (x)
i
|2p i = x |2p i i |2p i + y |2p i i |2p i + z |2p i i |2p i . (26)
2
For the rest of the report we assume that the magnetic field is directed along the [1,1,1] direction so that the energy
shift for the 2p levels are the same for all valleys. We denote this energy shift .
Delta pulse approximation: Assuming that the split 2 between the |2p+ i and |2p i states satisfy 2(tf ti )
0 tp , we can neglect this energy split during the pulse. Then we can proceed with the calculation in the basis of |0i
and |e2 i in the same way as in the previous section with |e1 i |e2 i. The final wavefunction is
where Sf is the pulse area given in Eq. (20). The excited state population at the end of the pulse is pe = 1
2
[cos (0 tp )] .
In the general case when tp is not negligible we need to work in the basis of |0i , |+i , |i. We expand the
wavefunction as
In the rotating wave approximation, the coefficients vector c(t) = [c0 (t), c+ (t), c (t)] obeys the differential equation
where
0 (t)
2
(t)
2
(t)
2
Hrot (t) = 0
, (30)
(t)
2
0 +
Square pulse: We can approximate the Gaussian pulse by a square pulse with the same pulse area
(
0 if tp /2 < t < tp /2.
(t) = (31)
0 otherwise.
4
If initially the donor electron is in the ground state, the final wavefunction at resonance ( = 0) is
!
c0 (tp ) Z tp /2 1
c+ (tp ) = exp i Hrot (t0 )dt0 0 (32)
c (tp ) tp /2 0
2 2
+ cos(1 tp )
0
1 2
0 [(cos(1 tp )1)+i1 sin(1 tp )]
2
=
21 ,
0 [(1cos(1 tp )))+i1 sin(1 tp )]
2
21
p
where 1 = 2 + 20 . The excited state population at the end of the pulse is
2
2 + 20 cos (1 tp )
2
pe = 1 [c0 (tf )] = 1 . (33)
21
Numerical solution for the Gaussian pulse: For the Gaussian pulse in Eq. (12) we solve for the final wavefunction
(at resonance, =0) using the numerical method of Eq. (21). We fix 0 = 0.01 THz, so that the period of the Rabi
oscillation is TRabi = 2/0 628 ps, and consider two different cases: (1) When the energy shift is = 0.001 THz
0 ; the period of the beat is Tbeat = 2/(2) = 3140 ps, and (2) when = 0 ; the period of the beat is Tbeat = 314
ps. Case 2 is closer to the experimental values of the quantum beat paper. The excited state population as a function
of the pulse duration tp is plotted in Fig. 1 and Fig. 2. The time evolution of the excitation probability during the
interaction window with a Gaussian pulse is also plotted in Fig. 3. The same graph for square pulse is given in Fig. 4
for comparison.
Remarks on the figures: When is comparable to 0 , as in Fig. 2, it is surprising that the result for the square
pulse and the Gaussian pulse are very different. In this regime there is no sustained Rabi oscillation with the Gaussian
pulse no matter how long the pulse duration is. We can also see that sustained Rabi oscillation exists for 0
with the Gaussian pulse (see Fig. 3), and for all values of with the square pulse (see Fig. 4). Im not sure why a
large beat frequency suppresses Rabi oscillation in the case of the Gaussian pulse.
FIG. 1. Excitation probability at the end of the pulse as a function of the pulse duration tp for 0 = 0.01 THz and = 0.001
THz.
5
FIG. 2. Excitation probability at the end of the pulse as a function of the pulse duration tp for 0 = = 0.01 THz.
FIG. 3. Time evolution of the excitation probability during a Gaussian pulse with tp = 1500 ps for 0 = 0.01 THz and various
values of .
FIG. 4. Time evolution of the excitation probability during a square pulse with tp = 1500 ps for 0 = 0.01 THz and various
values of .