Sei sulla pagina 1di 3

Some notes on Fourier-transform spectroscopy

Elson Liu

General references
A good review of the principles of Fourier-transform spectroscopy is contained in Vanasse
and Sakai (1967).

Resolution
The mathematical basis of Fourier-transform spectroscopy is the Wiener-Khinchine theorem,
which states that the autocorrelation ΓU (τ ) of a signal u(t) and its power spectral density
GU (ν) form a Fourier-transform pair:
F
ΓU (τ ) ⇐⇒ GU (ν)

where
Z ∞
ΓU ( τ ) , u(t + τ )u(t) dt
−∞
|UT (ν)| 2
GU (ν) , lim
T →∞ T   
t
UT (ν) , F {uT (t)} = F u(t) rect
T

Our notation and definitions are based on that of Goodman (1985). The interferogram
recorded by a Fourier-transform spectrometer is the autocorrelation of the electric field.
The delay τ is generated by the unequal lengths of the arms, so we have τ = 2l/c, where
l is the difference in the length of the arms, and c is the speed of light in vacuum. In
Lab 1, we obtained ΓU (n), where n is the index of the stage position. By multiplying n
by the calibration factor C of microns per step, we rescale ΓU (n) to obtain ΓU (2l ). Fi-
nally, by dividing by c, we rescale ΓU (2l ) to obtain ΓU (τ ). In MATLAB, all of this can be
accomplished by rescaling the x-coordinate vector.
To determine the power-spectrum, we can perform a discrete Fourier transform on the
sampled interferogram data. The sampling interval in the frequency domain is

1
∆ν = ,
N∆τ

1
where N is the total number of samples, and ∆τ is the sampling interval in the time
domain. We can transform the frequency axis into a wavelength axis by taking the recip-
rocal and multiplying by c. (Note: uniform sampling in frequency becomes non-uniform
sampling in wavelength.)
The Fourier-transform relationship established by the Wiener-Khinchine theorem as-
sumes that the full interferogram is recorded. If we truncate the interferogram (i.e. if the
spectrometer only scans a finite range) then, by the convolution theorem, we have
τ
F
ΓU (τ ) rect ⇐⇒ GU (ν) ∗ T sinc(Tν)
T
where ∗ denotes the convolution operation. The convolution of the power spectral den-
sity GU (ν) with T sinc(Tν) limits the resolution of the spectrometer.

Importing .adf files into MATLAB


Listing 1: lab1plot.m
function lab1plot ( filename , c a l i b r a t i o n)
% LAB1PLOT
% Plot extracted data and power spectrum from Lab 1
%
5 % Usage: lab1plot(filename,calibration)
%
% Example: lab1plot(’ElsonHeNeExtract.adf’,0.047)
%
% Inputs:
10 % filename = file name
% calibration = calibration factor in microns/step

% import data file


data = dlmread ( filename , ’ \ t ’ );
15 c = 3 * 10^14; % speed of light in microns per second
dx = c a l i b r a t i o n;
x = dx * data (: ,1); % path difference
tau = x / c ; % delay
dtau = tau (2) - tau (1);
20 N = length ( x );
y = data (: ,2); % interferogram data
ybar = mean ( y ); % DC component

% plot interferogram
25 subplot (2 ,1 ,1)
plot (x , y )
xlabel ( ’ Position [\ mum ] ’ );
title ( ’ E x t r a c t e d i n t e r f e r o g r a m ’)

2
30 % compute power spectrum
Y = fft (y - ybar );
f = [0: N -1] ’/( N * dtau ); % frequency coordinate
lambda = c ./ f ; % frequency to wavelength conversion

35 % plot power spectrum


subplot (2 ,1 ,2)
plot ( lambda ( floor ( sqrt ( N )): floor ( N /8)) , ...
abs ( Y ( floor ( sqrt ( N )): floor ( N /8))) , ’o - ’)
xlabel ( ’ W a v e l e n g t h [\ mum ] ’ );
40 title ( ’ Power Spectrum ’)

References
Joseph W. Goodman. Statistical Optics. Wiley, New York, 1985.

G. A. Vanasse and H. Sakai. Fourier spectroscopy. In E. Wolf, editor, Progress in Optics,


volume VI, chapter VII. North Holland, Amsterdam, 1967.

Potrebbero piacerti anche