Barium Ions for Quantum Computing
Matthew Dietrich, Ryan Bowler, Gary Howell, Adam Kleczewski, Nathan Kurz, Viki Mirgon,
Joanna Salacka, Jeff Sherman, Gang Shu, Jana Smith, Li Wang, Boris Blinov
University of Washington Physics Department
Introduction
Trapping Barium 138
Shelving with a femtosecond 455nm laser
We report progress on investigating 137Ba+ as a trapped ion qubit candidate.
The hyperfine structure and visible spectrum cooling transitions of 137Ba+ make
it an excellent qubit candidate. Here we report trapping 138Ba+, an even
isotope of barium, in a linear Paul trap. Cooling is provided by two diode lasers,
one at 650 nm and the other at 493 nm is generated by a doubled infrared
laser. To create the sidebands necessary for trapping 137Ba+, an EOM is
applied to the blue light, while the red is modulated directly using a bias-T on
the diode’s operating current. Shelving to the D5/2 state from the ground S1/2
state has been accomplished with a 1.76 μm fiber laser and during qubit
readout direct adiabatic rapid transfer will shelve the state with high fidelity.
Rabi flops between the ground hyperfine levels will be performed using
microwave pulses whose waveforms can be shaped using a homebuilt pulse
sequencer. A 400 fs pulsed Ti:sapphire laser is doubled in a single pass of
BBO to 455 nm, and can be used for coherent population transfer and single
photon production, using the S1/2 to P3/2 transition.
We have successfully trapped 138Ba+ in a linear Paul trap using two
cooling lasers, one at 493 nm and the other at 649 nm. Trapping
138Ba+ is greatly simplified by its lack of hyperfine structure and high
natural abundance, about 71%. The 649 nm laser light is easily
provided with a 10 mW laser diode, while 30 mW of 493 nm is created
by doubling in a bowtie cavity the light created by a 986 nm laser
diode. The two wavelengths are combined with a dichroic mirror and
coupled into a single mode optical fiber. This allows us to clean the
spatial mode, ensure laser overlap, and bring the laser light into
another room where the trap and ultrafast laser reside.
The light from a 5.5W 532 nm Verdi is used to pump a Ti:Sapphire Mira
ultrafast laser to obtain 500 mW of 910 nm pulsed light. The pulses are 400 fs
long and have a repetition rate of 76 MHz. Because of the high intensity of
each pulse, it is possible to double this light without a cavity in a single pass of
a BBO crystal. This gives us 10-20 mW of pulsed 455 nm laser light, which is
resonant with a Ba+ transition to the P3/2 state. This state decays into the D5/2
with a branching ratio of about 30%. Using a train of 3 individual pulses, we
have observed shelving efficiencies of up to 10.5%, and have made progress
towards observing a π Rabi flop between the S and P states with a single
femtosecond pulse. In order to speed up the deshelving transition, we shine a
high power, broad spectrum orange LED onto the ion, exciting the D5/2 to P3/2
transition and allowing for a decay back into the ground state. This has been
shown to deshelve the ion reliably within 50 ms.
τ ~ 6.3 ns
493 nm
doubled-diode
1762 nm
fiber laser
BBO
Verdi-5
Mira-900B
910 nm, 400 fs, 76 MHz
AOM
pulse-picker
Fig 2. Schematic of experimental setup.
connected with optical fiber.
The whole setup is divided between two rooms, and
20
15
10
Shelving percentage
0
0
1
2
3
4
5
6
7
8
9
10
5
960
830
700
570
0
440
tellurium
saturation
spectroscopy
lock
2
Fig 5. Shelving Probability as a function of
frequency offset during a scan with the 1762
nm laser.
Power broadening made the
transition easier to find.
25
50
EOM
Transitions to the metastable D5/2 state can be accomplished directly
from the ground state. This metastable state has a lifetime of 35 s,
and while in this state, the ion will not fluoresce when illuminated with
the cooling lasers. This allows for reliable detection of a shelving
event with about 10 ms of observing time. Using a narrow band 1762
nm fiber laser, we have successfully shelved 138Ba+ with 50%
efficiency, indicating saturation (see fig 5). We have been making
progress towards performing the shelving operation using rapid
adiabatic passage, which will greatly increase our shelving efficiency.
Because this shelving process is done with a narrow laser, as
opposed to the broadband femtosecond laser, it will be possible to
shelve an ion in the F=2 ground state hyperfine level while leaving the
F=1 state untouched. This will allow for highly efficient qubit readout.
4
30
310
beam stop
6
35
-80
fiber
coupler
Shelving with a 1762 nm narrow band laser
8
40
180
~613 nm
10
45
-210
Orange LED
12
Laser intensity (mW)
-340
ion trap
576 MHz
8.037 GHz
is however an ineffective qubit, since its nuclear spin is I=0 and
so has no hyperfine structure. 137Ba+ has I=3/2, giving its ground
state two hyperfine levels. In order to trap this species of barium,
occurring with an 11% abundance, we need to add sidebands to both
of the cooling lasers. In the case of the blue, this is accomplished
with an EOM resonating at the frequency of the hyperfine splitting. In
the case of the red, two sets of sidebands are created by modulating
the laser diode directly with a bias-T.
-470
414 MHz
dichroic
mirror
138Ba+
Fig 7. Shelving probability as a function of
intensity. This probability is for shelving due
to a short train of 3 pulses. At some intensity,
the Rabi frequency will be such that each
pulse will transfer the ground state to the P3/2
state with perfect efficiency. There, because
the lifetime is shorter than the time between
pulses, we expect to see 66% shelving
probability.
-600
bias-T
PMT
Trapping Barium 137
Fig 6. Picture of Mira femtosecond pulse laser.
-730
Andor
Ixon
EMCCD
Iodine
saturation
spectroscopy
lock
649 nm
diode
Fig 4. Picture of 4 138Ba+ ions
as imaged by our iXon CCD
camera. The ions are about 24
μm apart.
-860
{
Fig 3. Picture of linear Paul trap before
bakeout. Electrodes are held in place
with alumnia holders. The ovens can be
seen in the upper left corner. An aperture
(not shown) prevents the barium from
coating the entire apparatus.
Shelving Probability
P3/2
The two hyperfine levels of the
614 nm
137Ba+ ground state will serve as our
P1/2 τ ~ 7.8 ns
Deshelving
LED
ionic qubit. The energy splitting
649 nm
493 nm
Repump
between them, 8.037 GHz, is such
τ ~ 35 s
D5/2
Cooling
that qubit rotations can be
455 nm
D3/2
performed directly using microwave
Ultrafast
τ ~ 83s
pulses.
These pulses will be
1762 nm
Readout
controlled using a home built pulse
programmer, capable of forming
|1>
nanosecond
scale
pulses
of
F=2
8.037 GHz
radiation. Working with barium is S1/2
|0>
F=1
facilitated by its many visible
wavelength transitions. Also, since
137
Fig 1.
Energy level diagram for
Ba+,
the even isotope has no nuclear
showing qubit levels and relevant transitions.
spin, the path to trapping is
Not to scale.
simplified.
-990
Barium as a Qubit
Frequency Offset (MHz)
Future Work
We intend to confirm the trapping of 137Ba+, complete the pulse picker that will
allow us to excite the ion with single femtosecond pulses of 455 nm laser light,
observe a complete Rabi flop with the those pulses, observe adiabatic rapid
passage with the 1762 nm laser, and observe Rabi flops between the qubit
states using the pulse programmer (see Ryan Bowler’s poster).
Supported by UW Royalty Research Fund