THIRD YEAR LABORATORY

SPECTROMETRY WITH A MICHELSON INTERFEROMETER

1 Aims
To use Michelson interferometry to determine the spectra of a variety of sources of light

2 Objectives
1. To calibrate the interferometer using a HeNe & diode laser, and the mercury green line
and measure the wavelength of yellow light.
2. To measure the separation of the mercury yellow doublet.
3. To investigate the coherence length and spectra of high intensity LED’s.

3 Introduction
Spectrometry involves the Fourier transform of a signal in order to extract its frequency
content. In the case of an ordinary prism or grating spectrometer, as used in optical
spectrometry, the incoming signal is a fluctuating electromagnetic wave. At the entry slit of
the spectrometer the signal may be described by E (t ) , the electric field strength as a function
of time, φ (v ) , the Fourier transform of E (t ) describes the frequency content, where φ (v )dv is
the amplitude observed when the signal has been filtered to allow only the frequency range
v → v + dv to pass. The power transmitted within this frequency range is φ (v )φ * (v ) the
square modulus of φ (v ) .
The usual grating spectrometer sends signals of different frequencies in different directions so
that there is a lateral dispersion of the signal, and a slit of suitable width will isolate the
frequency range v → v + dv . Then, when the slit is moved, the detector traces out the curve of
I (v ) against v the spectrum of the incoming signal.

4 Theory
The Michelson interferometer behaves quite differently. If monochromatic light enters the
instrument, the beam is split into two, with amplitudes A1 and A2, where
A1 A1* + A2 A2* = AA* = I where I is the intensity of the incoming radiation. The intensity of
any radiation is of course proportional to the square modulus of the amplitude, which in this
case we take to be the electric vector E 0 e 2πivt at the beam-splitter.
Consider the moment when t = 0 . Then at the beam-splitter, just before the light is split,
I = A02 . At an infinitesimal distance after the beam-splitter, the amplitudes of the beam are
A0 iφ1 A
e and 0 e iφ 2 , the phase changes on reflection and transmission.
2 2

-1-
 Mirror

Mirror

Beam
splitter

Screen/
detector
Figure 1:Experimental arrangement of a
Michelson interferometer
2πd1 2πd 2
A0 iφ1 i λ A0 iφ2 i λ
At the mirrors the amplitudes are e e and e e because a path change d is
2 2
2π
accompanied by a phase change d.
λ
When the beams recombine at the beam-splitter the resultant amplitude is

A0 i (φ1 +φ2 )  i λ 1 
2π 2 d1 2π 2 d 2 2π 2 d 2π 2 d 2
A0 iφ1 i iφ2 A i i
e e λ
e + 0 eiφ2 e λ
e iφ1
= e e +e λ 
2 2 2  
which takes account of the extra phase-shifts on reflection and refraction. Where d1 and d2 are
the path lengths of the two beams.
This outgoing amplitude, multiplied by its complex conjugate, is the intensity observed at the
detector.
A02
I (∆ ) = [1 + cos(2πv∆ )] (1)
2
Where ∆ = 2(d1 − d 2 ) is the path-difference travelled by the two beams, and , ν = λ−1 is the
wave-number (often measured in cm-1).

Question 1. Show that equation (1) is valid. Remember that half the light intensity that
recombines on the beam splitter is sent to the detector.

Equation (1) shows that when the path-difference is changed steadily the intensity at the
detector changes sinusoidally. The period depends on the rate at which the path-difference is
changing and on the wavelength (λ) of the radiation. The intensity is a maximum at path-
differences 0, λ 2 , λ , 3λ 2 ... . We usually write:
I (0 )
I (∆ ) = [1 + cos(2πv∆ )]
2

-2-
If, instead of monochromatic radiation, we have a continuous spectrum entering the
interferometer, the intensity in the wave-number interval v → v + δv can be described by
I (v )dv . Then the output at the detector is:
I (v )
∞
I (∆ ) = ∫ [1 + cos(2πv∆ )]dv
−∞ 2

the integral comprising all possible wavelengths (and hence wave-numbers) in the incoming
beam. This simplifies to:
I (0 ) ∞ I (v )
I (∆ ) = +∫ cos(2πv∆ )dv
2 −∞ 2

This is the clue to Fourier spectroscopy: the left hand side, I (∆ ) is equal to a constant I (0) 2
plus the Fourier transform of I (v ) . Consequently, if we take the Fourier transform of the
emerging signal, recorded as a function of a path difference ∆ , we get a constant plus a
function proportional to the spectral content of the beam.

Question 2. Calculate the Fourier transform for a number of different I (ν ) , for example
choose I (ν ) to be a Gaussian function, and then choose I (ν ) to be a ‘top hat’ function. This
will give you an idea what you expect to see for I (∆ ) in these cases.

This seems to be a complicated way of doing spectroscopy compared with the usual process
of recording spectra which involves rotating a prism or grating inside a spectrometer, but it
has one enormous advantage: the amount of radiation that traverses the instrument is
hundreds of times greater than can be transmitted by a grating spectrometer with its narrow
entry and exit slits. Furthermore, because there is only one detector, looking at all the
radiation (or at least, half of it!) all the time, the signal/noise ratio is much improved.
This type of spectrometry, where all frequencies are recorded at the same time and analysed
later, is called “multiplex” spectrometry.

5 Practical details
The Michelson interferometer is a two-beam interferometer and when properly adjusted gives
fringes of constant inclination. These fringes are apparently at infinity and can be focussed on
to a screen with a simple lens, where they may form a set of concentric rings. Why do you
see a series of rings? Each ring corresponds to light transmitted through the instrument at a
different angle to the optic axis (which is the axis of the focusing lens), and the angles for
which transmission is possible are given by the equation:
2nd cos θ = mλ
where n is the refractive index (usually ~1 in air), 2d is the path-difference between the two
beams passing through the instrument, θ is the angle between the transmitted light and the
axis of the interferometer, λ is the wavelength and m is an integer when the transmitted
intensity is a maximum. This is the most important formula in physical optics: you should
learn it by heart.

-3-
For a given wavelength, there are various angles θ corresponding to different values of m. We
are concerned here with the light that comes through parallel to the axis, which is focused to a
point, or at least a small circle. Then θ = 0 and 2d = mλ .
Arrange a mercury lamp at the entry port, with a ground glass screen or a piece of thin tissue
to give a bright but fairly uniform light source. If the interferometer is not in adjustment, a
needle or sharp point silhouetted against the source will show two images, one from each
mirror. Adjust the tiltable mirror until the two images coincide. As they do, you should see
fine fringes crossing the light source. Continue the adjustment delicately, to make the fringes
as large as possible. Eventually they should appear circular or elliptical, and the instrument is
properly adjusted when the fringe pattern is stationary as you move your eye across the
aperture. If now you replace your eye with a lens, you should be able to put an image of the
fringe pattern on a white card. If necessary, remove the ground glass screen to increase the
brightness of the image. You can use a green filter in front of the mercury lamp to isolate the
5460.74Ǻ line in the mercury spectrum. This will increase the fringe contrast.
The central spot of the fringe pattern may not be illuminated – it depends on the path-
difference as you set up the instrument. This merely means that the order at the centre is not
integer. At the centre of the pattern you require a small hole to isolate the radiation coming
through near θ = 0 . This will be received by the detector and constitutes the signal I (∆ )
which you are going to Fourier-transform. The path-difference can be changed steadily with
the synchronous motor, and the signal from the photodiode is recorded at the same time, as a
series of “samples”. The references below will show how the spectrum can be reconstructed
from a knowledge of these “samples”, using the Fast Fourier Transform method, a
computation technique which is of great importance in experimental physics and which you
should understand.
The sampling theorem which you should look up, is important1 and should be well
understood, particularly where it describes “Nyquist frequency” and the phenomenon of
“aliasing” artefacts.
You should take precautions against scattered light entering the detector, using metal tubes or
baffles which you can make from black card. Be aware that the photo-diode has a different
response from your eye. Lasers and LEDs are also available to assist you with the calibration
of this instrument.
Use your Michelson Interferometer to establish fringes from the Helium Neon laser. You
should investigate the coherence length of the laser using your interferometer (you will need
to fully understand the meaning of coherence length). Discuss with your demonstrator what
you expect the coherence length to be, and justify why you think this is so. The Helium Neon
laser can be used to calibrate the encoder counts from the translational stage attached to the
mirror. Its wavelength is well defined (λ=632.816nm) and so too is the distance between the
fringes.
Replace the Helium Neon laser with the laser diode. Redo your experiment, and note any
differences which are observed between the two types of laser. Explain where you think these
differences come from.

1
It is important is all spectroscopy and indeed all experimental physics: it shows you how to make maximum use
of the experimental time available, to collect useful information and avoid useless information.

-4-
You have been given a range of LEDs, these LED’s produce light of different colours, and
indeed one of these is a so-called ‘white light’ LED. From your results with these different
LED’s, determine the bandwidth of the radiation that they emit. Can you infer how the so-
called white LED produces white light and how it operates?

6. References
[1] W.H. Steel. “Interferometry” Cambridge University Press, 1967. Chapter 10.
[2] J.F. James & R.S. Sternberg. “The Design of Optical Spectrometers”, Chapman & Hall,
1969. Chapter 8.
[3] E.Hecht. “Optics” Benjamin Cummings, 2003. Revision of Fourier Optics, coherence and
interferometry.
[4] The fast fourier transform (such as that used in MatLab) is described in detail in
‘Numerical Recipes in C or Fortran’, W.H.Press, CUP, 2007.

Use Matlab to analyse the data.

-5-