rf input

dc input

µw input

2x µw inputs 2x dc inputs

rf chokerf capacitor

UHVfeedthrough

SMA cables

filter capacitor

µwfilters

todoidaltransformer

xy

z

I1 I2 I3

B1

B3

B2

a) b)

ion

D T C Allcock T P Harty L Guidoni D Prado Lopes Aude H A Janacek C J Ballance N M Linke D N Stacey A M Steane D M Lucas

The Oxford planar ion trap projectIon Trap Quantum Computing Group - Department of Physics - University of Oxford

www.physics.ox.ac.uk/users/iontrap/ Department of Physics, Clarendon Laboratory, Parks Road, Oxford, OX1 3PU, United Kingdom d.allcock@physics.ox.ac.uk

Prototype Trap - Allcock et. al. - New J. Phys. 12,053026 (2010)

Sandia Trap Testing - Allcock et. al. - App. Phys. B (2011)

- Similar design to Oxford trap above but with slot for integrated optics.- Monolithic silicon, glass and aluminium construction.- Fabrication by Sandia National Laboratories (group of M. Blain) and funded by iARPA. - See Stick et. al. (arXiv:1008.0990) for fabrication info.- Three traps tested at Oxford.

10

µm

25µ

−200 −150 −100 −50 0 50 100 150 2000

0.02

0.04

0.06

0.08

0.1

0.12

P−P

opul

atio

n

Repump laser detuning (MHz)

Heatingregion

Ion

150µm

dccontrol

rfCCEs1 2 rf dc

control

y (μ

m)

x (μm)−5 0 5

145

150

155

x (μm)−5 0 5

x (μm)−5 0 5

Rf pseudopotential Tilt quadrupole Net pseudopotential

Repumper

Doppler cooling

Ion

Trap

‘6-wire’ design (below) with split centre dc electrode allows arbitrary orientation of trap principle axes

Laser Cleaning - Allcock et. al. - New J. Phys. 13,123023 (2011)

250μ

m260μ

m

0 500 1000 1500

-60

-40

-20

0

20

40

60

80

100

120

Time (s)

x-co

mpe

nsat

ion

(V/ m

)

30s charging

60s charging

300s charging

S i(100) S i(100)

MetalInter-levelDielec tric

E lec tric alV ia

C apac itor T op E lec trode

C apac itorDielec tric

C apac itorB ottom

E lec trodeS pacer

105

10610

-10

10-9

Axial frequency (Hz)

Ele

ctr

icfield

spectr

al

nois

edensi

ty,S

E(V

2m

-2H

z-1) Ce ntre (pre-c l eaning): α = 0.93(5)

A (post-c l eaning): α = 0.57(3)

B (post-c l eaning): α = 0.82(7)

43Ca+ Field-Insensitive Qubit

Alternative 40Ca+ Repumping Schemes Standard Doppler cooling re-pumping scheme uses 866nm laser (scheme I).This allows the use of D5/2 as a shelve for qubit readout.If qubit readout isn’t the goal though better schemes exist.

50 Ω CPW feed lines

λ/4 coupling section

open end of resonator

dc electrodes rf electrodes

ground planecentre microwave electrode

λ/2 resonator

2 mm

to SMAconnectors

x

y

z

2.5 3 3.5 40

0.2

0.4

0.6

0.8

1

Fra

cti

on

of

pow

er

couple

din

totr

ap

Frequency (GHz)

2.5 3 3.5 40

0.2

0.4

0.6

0.8

1

Frequency (GHz)

Port 1

Port 3

Port 2

Towards Microwave-driven Entanglement

- Linke et. al. - App. Phys. B (2012)

0 5 100

0.2

0.4

0.6

0.8

1

Time tR µs)(

Network analyser data shows that >75%of input microwave power is coupled intothe trap.

Microwave field measured with ion is within 15% of simulation.

times of ~3µs with <1mW microwave power typical (see below).

Trap TestingTrap Design

Proposed (2008) and demonstrated (2011) by Ospelkaus and coworkers at NIST.

Gate driven by oscillating microwave, rather than optical, field gradient. Ion is trapped in the near-field <100µm from a microwave conductor to obtain high enough gradients.

393nm+

D5/2 ShelfD3/2

S1/2

M=+

4M

=+5

P3/2

M=+

4M

=+5

854nm89.9%

850nm

850nm+

-400 -300 -200 -100 0 100 200 3000.01

0.015

0.02

0.025

0.03

0.035

0.04

0.045

0.05

866 detuning (MHz)

P 1/2 p

opul

atio

n

ExperimentTheory

Cooling at 146G Readout

Shelving scheme allows >99.9% readout fidelity without the need for a laser to address the quadrupole transition.

- S1/2-P1/2-D1/2 system has 64 states and no closed transitions.- Optical Bloch equations used to simulate.- Straightforward cooling solution found: - Polarizations chosen so only a few states populated - Needs only one sideband on cooling laser (from EOM) - Single frequency 866nm repumping laser - P1/2 level population of up to 0.16 simulated

- Intermediate-field clock qubits preferable to zero-field clock qubit as Zeeman shift lifts state degeneracies.- Until now intermediate-field clock states only demonstated in 9Be+ and 25Mg+ (NIST).- 43Ca+ has the following advantages: - No UV lasers which can charge up the trap - Laser diodes available at all wavelengths - D-states for electron shelving (high readout fidelity)- However no closed cooling transition...

Measured ‘clock’ and ‘stretch’ qubit frequencies using a single 43Ca+ ion at 146G.

Preliminary comparison of experiment and theory for a frequency scan of the 866nm repumping laser.

M=0

M=+4

M=+3

M=0

‘Stretch’ qubit

‘Clock’qubit

~50MHz

S1/2GroundLevel

3.20

0 G

Hz

Ion is initialised in M=+4 using optical pumping then 3 microwave pulses (green) are used to prepare the ‘clock’ qubit.

- ‘Anomolous heating’ in ion traps thought to be caused by adsorbates on surface.- Pulsed-laser cleaning of the trap significantly reduces heating rate (by ~50%).- First reported in situ reduction of heating rate by removal of source.

Diagram of laser cleaning geometry.Trap is the Sandia one described above.

Laser is 355nm tripled Nd:YAG- 2-5ns pulses at 0.2Hz rep rate- up to 350mJ/cm2 in 300µm dia. spot

Results (left) showing ~50% reduction in heating rate in cleaned area (A) compared to uncleaned area (B). Note also the change in frequency dependency of SE (ω-α).

Results backed up by recent Ar+ ion cleaning of trap at NIST (arXiv:1112.5419). - Two orders of magnitude reported by NIST - Laser cleaning is experimentally much simpler though so further experiments to optimise the techique should be performed.

Image of the cleaning spot on the trap (left) and the ablation plume caused when the laser is first applied (right).

−2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2−2

0

2

4

6

843Ca+ Clock qubit transition (M=0 to M=+1)

B − 146.094 (G)

ν cloc

k − 3

199

941

150

Hz

(kH

z)

−2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2−6

−4

−2

0

2

4

643Ca+ Stretch qubit transition (M=+4 to M=+3)

B − 146.094 (G)

ν stre

tch −

2 8

73 6

31 8

42 H

z (M

Hz)

1.208 kHz/G2

−2.362 MHz/G

Advantages: - Microwave electronics more mature and scalable technology than lasers - No photon scattering as in laser-driven Raman gates - No requirement for sub-Doppler cooling

Disadvantages: - Microwave field not as well localised as laser field (crosstalk) - Careful nulling of microwave field at ion required to suppress AC Zeeman shifts - Fast gates (~10µs) will require small traps and high microwave current densities

Optimisation of the trapmicrowave design wascarried out using HFSS,finite element microwavesimulation software. Figures (left and right) show simulations of current density in the trap. Sideband measurements coming soon...

DC ControlRF + MW +DCMW + DCGround

I1I2I3

Trap is gold on sapphire for good thermal conductivity.

The trap region is in the centre of a half-wave resonator to increase currents. Quarter-wave coupling sections provide a good 50Ω impedence match.

Simulations show that combining the microwave and rf electrodes (above) gives higher microwave gradients than alternative layouts for a given current.

Diplexer (right) is used to combine microwaves, rf voltages and dc offsets. The toroidal transformer steps-up the trap rf supply.

Background free detection can also be achieved (right) by using an interference filter in the detection system to block 397nm light ans detect only 393nm fluorescence. We achieve 29000s-1 fluorescence with only 1s-1 scattered laser light, at moderate saturation of the cooling transition.

850nm and 854nm repumpers (scheme II) give higher count and cooling rates and allow easier interpretation of some experiments (eg Doppler recooling heating rate measurements) due to the lack of 2-photon coherent effects, such as dark resonances.

Early versions of trap without on-package capacitors (shown above) had un-compensatable micromotion issues. Future versions will have ‘trench’ capacitors under the electrodes themselves.

Charging of trap (see graph below) investigated by crashing a 10µW beam into the chip (right). Aluminium itself charges (trap has no exposed dielectrics) due to native oxide.

Future traps will be gold coated (above) to reduce charging.

Gold on quartz fabrication at Oxford (left).

Micromotion compensation in all directions is possible by using an out of plane repumper and observing a modulated Raman effect (dip in the scan to the right).

cleaning beam

Microwave currents in trap electrodes

Microwave B-field around ion

F=4

F=3