Atom Interferometers

Yanbei Chen California Institute of Technology

GWADW, Ft Lauderdale, FL, May 15, 2009 G0900611-v1 2 Why Interferometers? • Not because have short de Broglie wavelength.

• But because atom clouds could be used as test masses for low frequency detection.

• Some recent proposals • R. Chiao et al. • S. Dimopoulos, P.W. Graham, J.M. Hogan, M.A. Kasevich and S. Rajendran, Phys. Rev. D 78, 122002, (2008).

• Paper on how atom interferometers respond to gravitational • A. Roura, D.R. Brill, B.L. Hu, C.W. Misner, and W.D. Phillips, Phys. Rev. D 73, 084018 (2006)

G0900611-v1 3 How does an atom interferometer work?

mirror

k beamsplitter 1 k k2 1

k k2 k1 1 k2 atoms mirror

• Similar to optical interferometer • Phaseshift depends on momentum transfer at x mirror k k • De Broglie wavelength λ=h/(mv) could be 1 2 small, de Broglie k=mv/ħ could be high, but it is δk that matters Φ = (k2 − k1 )⋅x • How do atom mirrors transfer momentum?

G0900611-v1 4 How does an atom mirror work? • Atoms are deflected by photons, for example Stimulated Raman Transition

γ1

absorption p=ħk1

γ2 stimulated γ2 × 2 emission

p~ħ(k1+k2) Stimulated Raman Transition In reality, excited state has negligible Picture from Dimopoulos et al. occupation number

G0900611-v1 5 Mirror vs Beamsplitter: Phase of Rabi Oscillation

γ1 γ2 Rabi oscillation

γ1 γ2

γ1 γ2

γ1 γ2

continuous pumping with two opposite beams: Rabi Oscillation atom sloshes between Picture from Dimopoulos et al. state 1 and state 2

• Depending on phase of the Rabi oscillation, one can make a mirror (π-pulse) or beamsplitter (π/2-pulse)

G0900611-v1 6 An Atom Interferometer

supplies γ1 supplies γ2 pulsed to always on create mirrors & π/2-pulse beamsplitters

T2

π-pulse

T1

π/2-pulse

Space-time diagram of a “linear” atom interferometer

Picture from Dimopoulos et al. G0900611-v1 7 An Atom Interferometer

supplies γ1 supplies γ2 pulsed to always on create mirrors & π/2-pulse beamsplitters

T2

π-pulse

T1 But δk is only 2kγ, and there π/2-pulse won’t be any improvement in length resolution!

Space-time diagram of a “linear” atom interferometer

Picture from Dimopoulos et al. G0900611-v1 8 Large Momentum Transfer & Atom Squeezing • It is possible to transfer more momenta, if the transition is Bragg, not Raman

Pulses can be iterated to give N times the momentum N ~ 100 - 1000 might be realizable

Picture from Dimopoulos et al.

• Roughly equivalent to the use of a cavity with Finesse equal to N, which is not very high.

G0900611-v1 9 Why are we still interested in Atom ? • Atoms might be usable as test masses for low-frequency detectors • they do not need to be suspended • they do not couple to vibration during flight

• Recent proposal • S. Dimopoulos, P.W. Graham, J.M. Hogan, M.A. Kasevich and S. Rajendran, Phys. Rev. D 78, 122002, (2008). • Two versions: terrestrial & space-based

G0900611-v1 10 Dimopoulos et al.’s Detector

T2 T4

T1 T3

• Two co-linear atom interferometers sharing the same control and passive

• Measures delays of pulses due to gravitational : (T4-T3)-(T2-T1) G0900611-v1 11 Ground-based version

IL~10 m because atoms drop during the 2T interrogation time

Advanced version L = 4 km

vacuum 10-10 Torr, average collision time 200s

1 W laser waist size 10 cm

9 8 -17 • T ~ 1 s, keff ~ 1.6×10 (with N~1000), n ~ 10 , 5 clouds simultaneously, h~10 /rtHz at 1 Hz (advanced version is 10x better)

G0900611-v1 12 Sensitivity of Ground-Based Version

ALIGO grav. grad.

G0900611-v1 13 Noises: vibration & laser phase noise

• Phase noise in control pulse unimportant (except for quantum part which seems small anyway) • Vibration & Phase noise: phase noise of passive laser are subtracted between times of L/c (short). Requires δx~10-15m (for 4 km version) or 10-14m (for 1 km version) G0900611-v1 14 Noises: Gradient

mass anomaly

Mass anomaly may increase/ decrease gravity gradient

Different paths Double loop configuration, cancels g gradient sense different g

• Gravity Gradient: mostly due to gradient of g --- fluctuations in atoms’ path cause difference in g. This sets requirement on the path of atoms, and therefore on the atom trap. • Tricks can be used to avoid measuring them • Add compensation mass • Using fancy pulses • Temporal fluctuations in gravity gradient seems unimportant. G0900611-v1 15 Rotation -8 • Coriolis force cause acceleration larger than GW δa=2ωeδv with δv~10 m/s • Can be cancelled if laser axis is not rotating together with earth. Requires accuracy of δω<10-9(rad/s/rtHz)×(1 km/L)1/2 • Laser jittering will cause the atoms to sense different g, and this requires laser beam to have δω<10-9(rad/s/rtHz) (for 1 km version)

G0900611-v1 16 Sensitivity to Stochastic Background

Ground-based versions G0900611-v1 17 Space-Based Version

3 T ~ 100s, d ~30 m to IL ~100 m is the maximum For 10 km version, aiming at waves minimize gravity that spacecraft can provide Laser 1 W below 0.01 Hz gradient caused bias B field. For T ~ 100s, waistsize 0.5m 9 by spacecraft this limits keff to ~10 . [v = ħk/m]

Must be shielded from sunlight, otherwise will only work when it’s behind the earth, when photon-atom scattering timescale is longer than T

Effect from solar wind considered unimportant

Interplanetary magnetic field can be dealt with

G0900611-v1 18 Sensitivity of Space-Based Detector

G0900611-v1 19 Sensitivity to Stochastic Background

Space Versions G0900611-v1 20 Comparison with LISA requirements

Photon shot noise does Limited by photon shot not seem to have been noise taken into account

G0900611-v1 21 Discussions?

G0900611-v1