|
|
??
|
Barium isotope 137 has
nuclear spin 3/2, which gives rise to hyperfine interaction and
ground state splitting of about 8 GHz. Hyperfine states with there
effectively infinite lifetime and long coherence are excellent
candidates for quantum information storage and processing. We are
working on developing a robust scheme for
137Ba+ qubit detection. |
|
The first step in a
quantum computation is the initialization of the qubit register. In
trapped Ba ions we achieve this by "optical pumping". Laser of
appropriately tuned frequency and polariation causes the electron in
the ion to go to a particular state (to be "pimped") and remain
there. The polarization of the laser is chosen such that when in
that staate the electron does not "see" the laser light. We
optically pump to the F= 2, mF= 0 Zeeman sublevel of the
ground state using a p-polarized 493 nm laser light. The p-polarization means that the
electric field of the laser is aligned with the quantuzation axis of
the system, defined by the direction of the applied magnetic field.
Instead of tuning the polarization itself, we tune the magnetic
field direction. Figure on the right shows the mapping of the
optical pumping in 137Ba+. Pumping is achieved
when the ion "diappears", which corresponds to the "deep-sea" area
on the map. |
 |
|
To measure the state of
the hyperfine qubit, we begin by "shelving" one of the hyperfine
states to the metastable 5D5/2 state. That state lifetime
is about 30 seconds, much longer than any subsequent operations. The
shelving light is coming from a 1762 nm fiber laser stabilized to a
high-finesse Zerodur optical cavity. To achieve high efficiency and
robustness of the shelfing process, adiabatic passage technique
is used. Following the shelving pulse w e apply the 495 nm and 650 nm laser light and detect ion
fluorescence. Ion in the "shelved" state does not scatter any
light and remains dark, while the non-shelved ion scatters many
photons. We can easily sicriminate between the "dark" and the
"bright" ion and thus detect the qubit state. |
| The 6D512 to
5D5/2 transition is a dipole-forbidden, slow transition.
However, with sufficient laser power it can be driven fast. We map
the transition by changing the frequency of the 1762 nm laser in
fine steps and measuring the probablilty of "shelving". Figure on
the right shows such a frequency scan for different conditions: no
optical pumping, with optcal pumping, both with 2 ms laser exposure
time, and with optical pumping and a 5 ms laser exposure time. Note
how the additional lines which correspond to different Zeeman
sublevels of the ground state diappear when the optical pumping is
present. At the longer exposure time, the shelving probablility
reaches 50%, which corresponds to a saturated transition. The large
width of the resonance line is due in part to power-broadening, but
is also a combination of the laser linewidth and the fact that we
transition to multiple Zeeman level of the 5D5/2
state. |
 |
|

|
Finally, putting the
initialization and the detection steps together with the 8 GHz
transition in between we can observe coherent qubit evolution - the
Rabi flops. Graph on the left shows these flops driven by about 1
watt of microwaves. The Rabi frequency is about 10 kHz. The
oscillations only reach about 70% at the maxima due to imperfect
shelving. |
|
Recently, by better
stabilizing the 1762 nm laser, we dramatically reduced the line
width of the shelving transition, as seen on the right. We observe
more than 2 orders of magnitude line width reduction, and now the
spectrum of motion sideband is clearly resolved. We can now apply
the adiabatic passage technique to increase the shelving efficiency
and thus to improve the qubit detection
fidelity. |
 | |