The Faraday effect

DNS/Revised AL October 2023 OP26

The equipment can get hot during operation due to the currents flowing in the solenoid. Don’t
touch it, and only run the equipment when making measurements, to avoid overheating.

1 Introduction
The optical properties of a dielectric are altered by the presence of a magnetic field. A particularly im-
portant case is that of plane polarised light travelling parallel to the direction of the field; it is found that
the plane of polarisation rotates. The effect is named after Faraday. It can be exploited as a spectroscopic
tool; the spectrum of Faraday rotation has been used a great deal to investigate properties of matter in
the solid state, and the method has also been applied to atomic vapours. It gives similar information
to that which can be derived from a study of the absorption spectrum, and in some respects is more
powerful than absorption spectroscopy.

The rotation produced by gaseous samples under typical conditions is very small, and in practice it
is measured only in the vicinity of absorption lines. However, a small length of a solid dielectric can
produce large rotations far from the regions of strong absorption, and this gives rise to the great practical
importance of the Faraday effect. In modern optics, it is often very convenient to be able to rotate the
plane of polarisation of a laser beam through a chosen angle reliably and reproducibly without serious
loss of intensity.

This experiment studies the Faraday effect in a sample of glass with a large Faraday rotation. We will
first study the theory and then verify:

(a) that the angle of rotation is proportional to the applied magnetic induction B.

(b) that it varies with wavelength.

(c) that the observations can be understood semi-quantitatively on the basis of very crude assump-
tions about the optical properties of the glass.

As one would expect, the rotation is also proportional to the length, l, of the sample; this leads to the
definition of the wavelength-dependent Verdet constant, V (λ), of a material which specifies the amount
of rotation, θ, as

θ = VBl (1)

Our experiment amounts to a determination of V (λ); the wavelengths to be used are 546 nm (mercury
green) and 589 nm (sodium yellow).

02Oct2023 Copyright c 2023 University of Oxford, except where indicated. OP26- 1

 The Faraday effect

2 Theory
2.1 Polarisation plane rotation
Consider a electromagnetic plane wave propagating along the (horizontal) z-axis with angular fre-
quency ω and wavenumber k = ω/c, where c is the speed of light. We assume that the electromagnetic
wave is linearly polarized along the x axis, so its electric field vector can be written as:

~E(z, t) = Aî cos(kz − ωt), (2)

where A is the amplitude and î is the unit vectors along the x axis. In the complex form:

~E(z, t) = ARe[îeikz−iωt ]. (3)

We rewrite this equation as follows:

  
~E(z, t) = √A Re 1 1
√ î + i ĵ + √ î − i ĵ eikz−iωt



(4)
2 2 2

and notice that the vectors √1 (î ± i ĵ) define the left and right circular polarizations, respectively. In
2
other words, we have written our linearly polarized electromagnetic wave as a superposition of two
circularly polarized waves, which we denote as σ+ and σ− .

As we shall see later, applying a magnetic field along z will render the refractive indices for these
two polarization components to different. Denoting them as n+ and n− , respectively, we have differ-
ent wavenumbers k± = n± ωc for the two circular components, and hence at the end point z = l of a cell
of length l, we have
  
1 + 1 −
~E(z, t) = ARe î + i ĵ ei(k )l + î − i ĵ ei(k )l e−iωt



2 2
  
1 1 + −
î + i ĵ eiθ + î − i ĵ e−iθ e((k )+(k ))l/2−iωt



= ARe
2 2
n  ((k+ )+(k− ))l/2−iωt o
= ARe î cos θ − ĵ sin θ e ,

where
(k+ )l − (k− )l ωl +
θ= = ( n − n − ). (5)
2 2c

We see that, because the σ+ and σ− waves travel at different speeds and develop a phase difference 2θ
as they progress through the cell, they recombine into a linearly polarized wave that is at angle θ to the
original polarization.

2.2 The classical theory of dispersion

In order to understand the effect of the field on the refractive index, let us recall why a material has
a refractive index different from unity at all. The following treatment is based closely on the classi-
cal theory of dispersion [1]. We represent the sample in the simplest possible way, as an assembly of
non-interacting atoms each with a bound electron capable of undergoing simple harmonic motion, the
natural frequency of vibration being ω0 . Each of these responds to the forcing action of the light wave,
causing the atom to have an oscillating induced dipole moment p = αE. The quantity α is called the
polarisability of the atom; we need to calculate it because it determines the relative permittivity er of
the material, and hence the speed of light through it (n2 = er , for a material with relative permeability
µr = 1).

The equation of motion of an electron is

OP26-2
 The Faraday effect

m( ẍ + ω02 x ) = −eE0 exp(iωt) (6)

(Damping is not included, because we are only interested in the response far from resonance where it
is unimportant.) E0 is the electric field amplitude, ω its angular frequency, and −e, m are the electronic
charge and mass respectively. The steady state solution is

eE0 exp(iωt)
x=− (7)
m(ω02 − ω 2 )

Since p = −ex, we have

e2
α=− (8)
m(ω02 − ω 2 )

The induced dipole moment P per unit volume of the sample is P = Np = NαE, where N is the atomic
number density. Recalling that

D = e r e0 E = e0 E + P (9)

we can now find er

Ne2
er = n 2 = 1 + (10)
e0 m(ω02 − ω 2 )

· Re-derive Eq. (9) in presence of a non-zero damping constant γ [so the term in parentheses in
Eq. (6) becomes ( ẍ + γ ẋ + ω02 x )].

What is the interpretation of the imaginary part of the refractive index? Is the sign of the imagi-
nary part consistent with your expectations?

We thus have a simple expression for the refractive index of the sample far from resonance, and to the
approximation we are working n clearly depends only on N, ω and ω0 , apart from fundamental con-
stants. It is real, because we have neglected damping; we have also neglected the (large) local field
effects, and taken the oscillator strength of the resonance to be unity. For our purposes these approxi-
mations are not as bad as they might seem, because we shall determine N and ω0 for the glass by fitting
to experimental values of n.

2.3 Zeeman effect

The theory of the Zeeman effect, whether treated classically or quantum mechanically, shows that a
single spectral line of angular frequency ω0 splits when a magnetic field is applied. In the simplest
case, the splitting is into two components symmetrically disposed about the original frequency by
amounts ±eB/2m. If the optical field propagates along the direction of the field, these two spectral
components correspond to optical fields with opposite circular polarisations. In other words, when
circularly polarised light is interacting with the atoms, the resonant frequencies change according to
ω0 → ω0 ± eB/2m, depending on the handedness of the polarisation – and so do the refractive indices,
which depend on ω0 according to Eq. (10). Because this change in ω0 is much less than both ω0 and ω,
we can write
eB dn
n+ − n− = . (11)
m dω0

OP26-3
 The Faraday effect

Figure 1 summarises the situation. The two refractive index curves, one for left-handed and one for
right-handed circularly polarised light, are shifted in opposite directions when the magnetic field is
applied, giving rise to a difference n+ − n− at any given frequency.

from equation (10) better approximation

n−1
(including damping)

without field

n−1
n+
Zeeman splitting
with field
n−
ω

typical frequency
of observation
ω0

Figure 1: Effect of Zeeman splitting on refractive index.

Reconciling this result with Eqs. (1) and (5), we find

ω e dn
V= . (12)
2c m dω0

· Derive Eqs. (11) and (12). Then find the derivative dn/dω0 from Eq. (10) to obtain the final
expression for the Verdet constant.

3 Estimating the Verdet constant

To invoke the harmonic oscillator model to describe a piece of glass is obviously very crude. However, it
is sufficient for our purposes as are interested in making an order of magnitude estimate of the Faraday
rotation. Our basic input information is the length l of the sample (they are all around 50 mm) and the
refractive index, which has been measured at three wavelengths:

λ(nm) n(λ)
486 1.95869
589 1.92663
656 1.91457

Fit the data from the table with Eq. (10) to estimate ω0 (you should find it in the ultraviolet region) as
well as Ne2 /e0 m.

· Is the value of Ne2 /e0 m you get of the same order of magnitude of what one might expect for a
solid?

OP26-4
 The Faraday effect

You now have an expression which completely specifies n(ω ) over the range of interest. The next step
is to estimate the Verdet constants for the glass at 546 nm and 589 nm using Eqs. (12) and (11).

· What is the estimated ratio of the Faraday rotation angles at 546 nm to that at 589 nm (use the
fact that ω0  ω)?

We will now carry out an experiment to verify your predictions.

4 Measuring the Faraday effect

Eyepiece Analyzer Solenoid Polariser Lamp

Figure 2: The apparatus.

The apparatus is shown in Fig. 2. It consists of the following components.

• A gas lamp. A mercury lamp with a green interference filter produces light at 546 nm while a
sodium lamp at 589 nm.

• A polariser (marked “2-Part Polariser”).

• A solenoid, in which the magnetic field is generated. The glass sample in its brass mount, in which
the Faraday effect is produced, can be inserted into the core of a solenoid.

• An analyzer with an eyepiece (marked “Analyser Polariser”).

The purpose of the analyzer is to measure polarization rotation. A simple way to do so is to use another
polariser after the sample. By rotating it to the angle φext of complete extinction (i.e. such that no light
can get through), one ensures that the polariser’s optic axis is orthogonal to the polarization of the
incoming field. One can measure how that angle depends on the magnetic field applied and therefore
characterise the Faraday effect.

A shortcoming of this simple approach is lack of precision: at the minimum of transmission, there is no
change in the intensity in the first order of δφ as φ is varied. To address this issue, a divided polariser is
used. The idea is to divide the field of view into two parts, each of which has a polariser with its axis
inclined slightly to that of the other (figure 3). One then adjusts the angle φ of the analyser until one
has equal brightness in the two halves of the field of view, rather than extinction in either. The type

OP26-5
The Faraday effect

of divided polariser used in this experiment is a Lippich polariser; it contains a large polarising prism
followed by a small Nicol, covering half the field of view and with its axis inclined at a small angle to
that of the first. This angle cannot be varied.

incident
beam polariser 2

polariser 1

Figure 3: Division of field of view in to two parts with two separate polarisers.

Assemble and align the optical setup. Begin with turning the lamp on and placing it in front of the
solenoid so you could see it shine through the solenoid opening. Then insert the polariser and the
analyzer, adjusting the height to be able to see the lamp’s light all the way through.

Focus the eyepiece by moving it slightly in and out of the analyzer. You should see an unsmeared
bright circle.

Rotate the analyzer and understand the operation of the divided polarizer.

Familiarize yourself with the solenoid power

 supply. It should be operated inconstant current mode
as follows. Switch the POWER
and OUTPUT
(if available) on, turn the CURRENT
knob to
minimum (fully anticlockwise)
 and the VOLTAGE
knobs to maximum (fully clockwise). Then set
the current using the CURRENT
knob only. The supply can provide currents up to 3–4 amps.

Determine the Verdet constants for the glass for light of wavelengths 546 nm and 589 nm, and com-
pare the results with your predictions. The length of the glass cylinder is marked on its holder. Since
some of the specimens are slightly anisotropic, it is advisable not to alter the positions of the glass
or the analyser during a set of measurements. Check that the field divider method is more sensitive
than simply looking for extinction. Compare the spread you get in several measurements of the same
angle (i) by looking for extinction in one half of the field (ii) by looking for equality of brightness in
the two halves.

· Will the rotation reverse direction when the direction of the magnetic field is reversed?

By setting up the apparatus as shown in figure 4, determine if the rotation will be doubled or reduced
to zero if the light were to be reflected back through the glass by a mirror.

Check your predictions experimentally. The polaroids give you a larger aperture to work with than
do the crystal polarisers. Set their axes so that the overall transmission of the system is small, and see
what happens when you turn the field on and off. You will notice back reflections from components
other than the mirror, but these should not present any serious problems.

· Explain carefully in your logbook what you did, what you observed and how you interpreted
your observations, including any sources of experimental error.

OP26-6
 The Faraday effect

Eye
Analyzer
Polarizer Sample
Glass plate
Mirror

Lamp

Figure 4: Investigating Faraday effect observed in double passage of the beam back and forth fthrough
the sample.

An interesting application of the results is the optical isolator, a device which prevents light from getting
back into a laser by reflection from some optical component at which the beam is directed. (This can
cause instability.) The isolator consists of a piece of glass between two polarisers with their axes at 45◦ ,
a magnetic field being applied to produce a Faraday rotation of just this amount.

· Satisfy yourself that an optical isolator would work.

A Calculation of the field due to the solenoid

The accuracy of your values for the Verdet constant is likely to be limited by uncertainty in the magnetic
field you apply to the sample. Given the moderate accuracy with which you can measure the current
through the solenoid, there is no point in going over the top about this. However, for completeness we
give the expression for the field at the centre of a coil of the type you use in this experiment; apply it,
and see how far out you would be if you simply assumed that the coil were infinitely long but had the
same number of turns per unit length.

If m is the total number of turns (this is stamped on the casing of the solenoid) and I the current flowing
(as shown in figure 5), then
2µ0 mI 1
| B| = (13)
L sec φ1 + sec φ2

φ1 φ2

Figure 5: Geometry for the magnetic field of a coil as used in this experiment.

Even this result is not strictly applicable because the samples have finite length. However, the correction
is only at the 1% level, and we ignore it.

OP26-7
The Faraday effect

Bibliography
[1] B. I. Bleaney, B. Bleaney, Electricity and Magnetism, Volume 1, 3rd edition, Oxford Classic Texts in the
Physical Sciences, 2013.

Appendix A Equipment
• Mercury Lamp with Green Filter
• Sodium Lamp
• Condenser Lens

• Polariser
• Telescope
• Mirror
• Glass Sheet

• Solenoid
• Power Supply

OP26-8