Calculations for Nondestructive BEC Imaging

Daniel A. Steck
Theoretical Division (T-8), MS B285
Los Alamos National Laboratory
Los Alamos, NM 87545

4 June 2002

1 Diffraction by a Phase Object

We describe the phase object (BEC) by the deviation δn(x, y, z) of the local index n(x, y, z) from unity:

δn(x, y, z) := n(x, y, z) − 1. (1)

In the thin-lens approximation, we can integrate over the propagation direction (z) of the probe, and consider
only the 2-dimensional index profile:

δnz (x, y) := δn(x, y, z) dz. (2)

This phase shift imparts momentum to the light as follows: in the scalar-wave approximation, the mean-square
transverse wave vector kT
2
 due to the relative phase shift of exp[ikδnz (x, y)] (k is the total wave vector) is


1
kT  =
2
E ∗ (x, y)(kx2 + ky2 )E(x, y) dx dy
|E|


2

= − e−ikδnz (∂x2 + ∂y2 )eikδnz dx dy (3)

 
= k2 (∂x δnz )2 + (∂y δnz )2 dx dy.

Here we have dropped terms of the form ∂x2 δnz , which do not contribute if we assume that δnz is purely real (i.e.,
we ignore absorption effects).
To make this expression more concrete, we can consider a Gaussian approximation to the true BEC atomic
distribution,  
δφa x2 y2
δnz = exp − 2 − 2 , (4)
2πσx σy 2σx 2σy
where δφa is the integrated phase shift:


δφa := δn(x, y, z) dx dy dz. (5)

Then we can write

(δφa )2 (σx2 + σy2 )k 2
kT
2
= (6)
8πσx3 σy3
for the optical transverse momentum in the Gaussian approximation.
2 ATOMIC PARAMETERS 2

This momentum is similarly imparted to the atomic cloud, and the atomic momentum diffusion is given by
h̄2 kT
2
 multiplied by the incident photon flux. Thus we can write

Ih̄ 2
D= k  (7)
Nω T
for the momentum diffusion rate per atom, where I is the probe intensity, ω is the optical frequency, and N is
the number of atoms in the condensate.

2 Atomic Parameters
We can write the local index as (see, for example, Cohen-Tannoudji, Dupont-Roc, and Grynberg, Atom-Photon
Interactions, p. 604)
nd |d|2 ∆
δn = − , (8)
20 h̄ ∆2 + Γ2 /4 + Ω2 /2
where nd is the local number density, d is the appropriate dipole moment (here, approximately the effective dipole
moment for far-detuned, linearly polarized light, see “Rubidium 87 D Line Data” for details). In the far-detuned
limit, the integrated phase shift is then
N |d|2
δφa = − . (9)
20 h̄∆
Thus the diffusion rate in the Gaussian approximation is

N |d|4 I σx2 + σy2

D = (h̄2 k 2 ) , (10)
420 h̄3 ω ∆2 8πσx3 σy3

which has an overall scaling similar to spontaneous scattering, I/∆2 .

3 Signal/Noise
From the Kadlecek et al. paper, the signal/noise ratio is given by

S/N = φ 2ηNp , (11)

where Np is the number of photons striking a particular CCD pixel, η is the CCD quantum efficiency, we consider √
only the limit of large reference-beam intensity, and we have reduced the value quoted here by a factor of 2
from the value quoted in the paper, as is appropriate for spatial-heterodyne mode (since we must average over
the heterodyne phase in the mean-square sense). This expression is also valid only for small local phase shifts φ;
for large φ, we must average also over φ, and in this case the signal/noise is given by the same expression, but
with φ replaced by 1. To evaluate the signal/noise ratio, we simply note that the typical phase shift is

N |d|2 k
φ≈− , (12)
20 h̄∆σxσy

and the photon number is

Itexp Apix
Np = , (13)
h̄ω
where texp is the exposure time of the image, and Apix is the effective CCD pixel area, taking into account any
magnification in the imaging √ system. Thus in the small φ regime, where the experiment is likely to operate, the
signal/noise ratio scales as I/∆2 .
 1
Hz MHz 106 . Hz kHz 103 . Hz GHz 109 . Hz THz 101 2. Hz
sec
mW 10 3 . watt ms 10 3 . sec cm 10 2 . m
farad
hbar 1.054571596. 10 3 4. joule. sec ε0 8.854187817. 10 1 2. µm 10 6 . m
m

Parameters for Rb 87
m c 2. π
c 2.99792458. 108 . ω 2. π . 384.2279818773. THz λ 2. π . k
sec ω λ
Γ 2. π . 6.065. MHz d 2.069. 10 2 9. coul. m
N 2. 105 σx 100. µm σy σx c. ε0. Γ2 . hbar2 mW
Isat Isat = 2.504
∆ 2. π . 2. GHz mW 4. d2 cm2
I 2.
cm2
∆ 2
N. d2 5
4. = 4.35 10
δφa δφa = 3.64826 µm3 Γ
2. ε0. hbar. ∆
I
= 0.799
Isat
Diffusion due to coherent lensing
N. d4 2
. I . σx σy2 .
D hbar2 . k2
. 2. 3. 2 . . 3. 3
4 ε0 hbar ω ∆ 8 π σx σy
2 2
9 hbar . k
D = 4.16 10
ms
Diffusion due to spontaneous emission
1
Γ . I . ∆ 2 I 1
Rsc 1 4. Rsc = 0.035
2 Isat Γ Isat ms

hbar2 . k2
Dsp Rsc. hbar2 . k2 Dsp = 0.035
ms

Signal/noise ratio

N. d2 . k
φ φ = 0.003 Apix ( 10. µm ) 2
2. ε0. hbar. ∆ . σx. σy

I. Apix
η 0.3 Np 6 1
hbar. ω Np = 7.856 10
ms

1
SN1 φ . 2. η. Np SN1 = 6.378 (valid for small phase shift)
ms