|
The
Faraday Effect
Michael Faraday (1791-1867) was born in a village near London into the family of a blacksmith. His family was too poor to keep him in school and, at the age of 13, he took a job as an errand boy in a bookshop. A year later he was apprenticed as a bookbinder for a term of seven years. Faraday was not only binding the books but was also reading many of them, which excited in him a burning interest in science. When his position in the bookshop was expiring, he applied for the job of assistant to Sir Humphry Davy, the celebrated chemist, whose lectures Faraday was attending during his apprenticeship. When Davy asked the advice of one of the governors of the Royal Institution of Great Britain about the employment of a young bookbinder, the man said: �Let him wash bottles! If he is any good he will accept the work; if he refuses, he is not good for anything.� Faraday accepted, and remained with the Royal Institution for the next fifty years, first as Davy�s assistant, then as his collaborator, and finally, after Davy�s death, as his successor. It has been said that Faraday was Davy's greatest discovery. In 1823 Faraday
liquefied chlorine and in 1825 he discovered the substance known as benzene.
He also did significant work in electrochemistry, discovering the laws
of electrolysis. However, his greatest work was with electricity. In 1821
Faraday built two devices to produce what he called electromagnetic rotation,
that is, a continuous circular motion from the circular magnetic force
around a wire. Ten years later, in 1831, he began his great series of
experiments in which he discovered electromagnetic induction. These experiments
form the basis of modern electromagnetic technology. Apart from numerous
publications in scientific magazines, the most remarkable document pertaining
to his studies is his Diary, which he kept continuously from the
year 1820 to the year 1862. (This was published in 1932 by the Royal Institution
in seven volumes containing a total of 3236 pages, with a few thousand
marginal drawings.) Queen Victoria rewarded Faraday�s lifetime of achievement
by granting him the use of a house at Hampton Court and a knighthood.
Faraday accepted the cottage but gracefully rejected the knighthood.1 On 13 September, 1845 Faraday discovered the magneto-optical effect that bears his name. This day�s entry in his Diary reads: �Today worked with lines of magnetic force, passing them across different bodies (transparent in different directions) and at the same time passing a polarized ray of light through them and afterwards examining the ray by a Nichol�s Eyepiece or other means.� After describing several negative results in which the ray of light was passed through air and several other substances, Faraday wrote in the same day�s entry: �A piece of heavy glass which was 2 inches by 1.8 inches, and 0.5 of an inch thick, being silico borate of lead, and polished on the two shortest edges, was experimented with. It gave no effects when the same magnetic poles or the contrary poles were on opposite sides (as respects the course of the polarized ray) � nor when the same poles were on the same side, either with a constant or intermitting current� BUT, when contrary magnetic poles were on the same side, there was an effect produced on the polarized ray, and thus magnetic force and light were proved to have relation to each other. This fact will most likely prove exceedingly fertile and of great value in the investigation of both conditions of natural force.� Electromagnetic
basis of the Faraday effect. Magneto-optical (MO) effects are
best described in terms of the dielectric tensor e
of the medium in which the interaction between the light and
the applied magnetic field (or the internal magnetization of the medium)
takes place.2
In an isotropic
material (such as ordinary glass) the three diagonal elements are identical
and, in the presence of a magnetic field along the Z-axis, there is a
non-zero off-diagonal element e�,
which couples the x- and y-components of the optical E-field,
that is,
In general, e and
e� are
wavelength-dependent, but over a narrow range of wavelengths they might
be treated as constants. In a transparent material, where there is no
optical absorption, e
is real and e�
is imaginary. However, in the most general case of an absorbing MO material
both e and e�
may be complex numbers. For dia- and paramagnetic media e�
is proportional to the applied magnetic field H, while for ferro- and
ferrimagnetic materials the spin-orbit coupling is the dominant source
of the MO interaction, making e�
proportional to the magnetization M of the medium.2 Since B = H + 4pM
(in CGS units), we denote the B-field inside the medium as the source
of the MO effects. When a polarized
beam of light propagates in a medium along the direction of the magnetic
field B, The material interacts
with the right- and left-circularly polarized (RCP and LCP) components
of the beam with different refractive indices, n� = (e � ie�)�.
For fused silica glass at the wavelength of l = 550 nm,
for example, e � 2.25
and e� � 10-7i
per kOe of applied magnetic field. (Note that both n+ and n� in this case are real-valued
and, therefore, there is no absorption.) For linearly polarized light
passing through a length L of the material under the influence
of a B-field, the two circular polarization components suffer a relative
phase-shift Df = 2pL (n+
n�)/l.3,4
As shown in Figure 1, a change of relative phase between the RCP and LCP
components is equivalent to a rotation of the plane of polarization by
the Faraday angle qF = �Df.
In the above example, qF ~ 0.22�
at l = 550 nm
for a 1 cm-thick slab immersed in a 1 kOe magnetic field. The
figure of 0.22�/cm/kOe
is known as the Verdet constant of fused silica at the specified wavelength.3 Certain magnetic
materials (e.g., magnetic garnets) are transparent enough to transmit
a good fraction of the light while producing a fairly large Faraday rotation.
These materials can be magnetized in a given direction and sustain their
magnetization when the external field is removed. Therefore, the Faraday
effect in these media may be observed in the absence of an external magnetic
field. At l = 550 nm,
for instance, a typical crystal of bismuth-substituted rare-earth iron
garnet may have e � 5.5 + 0.025i
and e� � 0.002 � 0.01i.
The complex refractive indices for RCP and LCP light are thus (n + ik)+ � 2.347 + 0.006i
and (n + ik)� � 2.343 + 0.005i,
yielding a Faraday rotation angle qF � 1.3�
for a micron-thick slab of this crystal. The absorption coefficient of
the material a = 4pkL/l,
where k is the imaginary part of the complex refractive index.
For the above garnet, therefore, a � 0.12
per micron, which is equivalent to 1dB loss of light for every 2 mm
of crystal thickness. In other words, this garnet delivers 2.6�
of polarization rotation per dB of loss. These crystals can be grown in
a range of thickness from a fraction of a micron to about 100 mm.
Thicker crystals are useful at longer wavelengths, where the losses are
small, but the Faraday rotation generally decreases with the increasing
wavelength as well. Faraday
rotation in a transparent slab. For the sake of simplicity
we ignore the effects of absorption in the Faraday medium and consider
a transparent slab of magnetic material having a real e
and a purely imaginary e�.
Thus consider a 20 mm-thick
slab having e = 5.5,
e� = 0.01i.
The material is magnetized perpendicular to the plane of its surface,
and a linearly polarized beam of light (E-field along the X-axis) is sent
at normal incidence through the slab, as in Figure 2(a).5,6
Real sources of light, of course, are never perfectly monochromatic and,
therefore, we assume a finite spectral bandwidth for the light source,
covering the range l = 545 nm � 555 nm.
Figure 3 shows computed plots of the transmitted amplitudes, |tx|
and |ty|, as well as the polarization rotation and ellipticity
angles, qF
and hF,
versus l.7
Due to multiple reflections at the front and rear facets of the slab these
functions vary periodically with l.
(The same interference phenomena are responsible for the non-zero values
of hF,
which would otherwise be absent in a transparent medium.) The net Faraday
rotation angle is the average value of qF
over the relevant range of wavelengths, but one should also recognize
that the wavelength-dependence of the direction of emergent polarization
produces a certain amount of depolarization in the emergent beam. The
Faraday rotation combined with the spectral bandwidth of the light source
thus causes partial depolarization as a direct consequence of interference
among the multiple reflections. Figure
3. A plane wave, linearly polarized along the
X-axis, is normally incident on a 20mm-thick
slab, as shown in Figure 2(a). The slab (e = 5.5,
e� = 0.01i)
is magnetized along the Z-axis. (a) Plots of |tx| and
|ty|, the transmitted polarization components along
the X- and Y-axes, as functions of l.
(b) Plots of polarization rotation angle qF
and ellipticity hF
versus l. Oblique
incidence. Figure 4 shows the transmitted amplitudes and polarization
angles versus the angle of incidence q
in the case of a 20 mm-thick
slab magnetized along the Z-axis (e = 5.5,
e� = 0.01i)
when, as shown in Figure 2(b), a p-polarized plane wave at the single
wavelength of l = 550 nm
is incident on the slab.7 The oscillations in the transmitted
amplitudes/polarization angles are caused by interference among the multiply-reflected
beams at the facets of the slab.� Aside from the interference effects,
however, note that the Faraday effect does not show any signs of abatement
with the increasing angle of incidence. The reason is that, while the
direction of propagation of the beam increasingly deviates from the direction
of the B-field, the propagation distance simultaneously increases, keeping
the net interaction between the magnetic material and the beam of light
at a constant level. Figure 5 shows
the case of oblique incidence at q = 85�
on the same slab as above in the range of l = 545 nm � 555 nm.7
As in the case of normal incidence depicted in Figure 3, we note a significant
variation of the Faraday angles/amplitudes within this narrow range of
wavelengths. Although the beam inside the slab travels at ~ 25�
relative to the direction of magnetization of the material, the maximum
Faraday effect as exemplified by |tsp| is the same as
in normal incidence, because the propagation distance is correspondingly
adjusted. The wavelength-averaged Faraday rotation may be lower at larger
angles of incidence, but this is just a consequence of interference; it
is not caused by the reduction of the intrinsic optical activity of the
slab. If, for instance, the facets are anti-reflection coated, or if the
beam enters and exits through index-matched spherical surfaces, then multiple
reflections would be eliminated and the Faraday rotation becomes independent
of the incidence angle. The above discussions
were confined to the case of p-polarized incident beam, but the conclusions
remain valid for s-polarized light as well. For example, Figure 6 is the
counterpart of Figure 4, showing the transmitted amplitudes and polarization
angles versus the angle of incidence for a s-polarized incident beam.7
Note that the magneto-optically generated component of polarization tps
in Figure 6 is identical to tsp in Figure 4. This is
an important and completely general result, indicating that the amount
of light converted from one polarization to another is independent of
the state of incident polarization. Faraday
medium in a Fabry-Perot resonator. Because the Faraday effect
is amplified as a result of the beam propagating back and forth within
a magnetized medium, it is interesting to observe the enhancement of the
Faraday effect in a Fabry-Perot resonator. Figure 7 shows the diagram
of a system that may be used to monitor such enhancement in a range of
angles of incidence. The first objective lens (NA = 0.8) focuses
a linearly polarized beam of light onto the Fabry-Perot resonator, and
the second, identical lens collimates the transmitted beam, thus allowing
observation at the exit pupil. For a 20 mm-thick
slab of transparent magnetic material sandwiched between a pair of dielectric
mirrors, Figure 8 shows the computed patterns of intensity and polarization
angles at the exit pupil of the collimator.7 This figure indicates
that the rings of maximum transmission also correspond to locations of
maximum polarization rotation. The maximum and minimum rotation angles
in Fig. 8(c) are +63� and -23�, respectively, well in excess of the
rotations obtained from the bare slab. Also note in Figures 8(c, d)
the asymmetrical nature of the polarization angles in the first and third
quadrants on the one hand and in the second and fourth quadrants on the
other.
Figure
8. Intensity and polarization patterns in the
exit pupil of the collimating lens of Figure 7. (a) Intensity distribution
of the emergent X-polarized component. The bright rings indicate the regions
where the conditions of resonance are met and the light passes through
the resonator. (b) Intensity distribution of the emergent Y-polarized
component. The bright rings coincide with those in (a), indicating that
the conditions of resonance for the incident polarization are the same
as those for the magneto-optically induced polarization. (c) Polarization
rotation angle qF
of the emergent beam encoded in gray-scale. The range of values of qF
is -23� (black) to +63� (white). (d) Polarization ellipticity hF
of the emergent beam encoded in gray-scale. The range of values of hF
is -32� (black) to +42� (white). Longitudinal
and transverse geometries. When the direction of the B-field
is in the plane of the slab as well as in the plane of incidence, as in
Figure 9(a), one observes the longitudinal Faraday effect. In this case
e� occupies
the position of eyz
in the dielectric tensor. The transverse effect occurs when the B-field,
while in the plane of the sample, is perpendicular to the incidence plane,
as in Figure 9(b). In this case e�
occupies the position of exz. In the longitudinal
case at normal incidence no polarization rotation occurs, but the effect
begins to show up with an increasing angle of incidence. For a p-polarized
plane wave (l = 550 nm)
obliquely incident on a 20 mm-thick
slab of magnetic material (e = 5.5,
e� = 0.01i),
Figure 10 shows the computed amplitudes of the transmitted p- and s-polarized
light as well as the angles of rotation and ellipticity versus the incidence
angle q. One
could readily compute similar results for a s-polarized incident beam
as well. In both cases the MO effect is bipolar, meaning that a reversal
of the direction of the B-field reverses the signs of qF
and hF.
Moreover, as in the polar case discussed earlier, the magneto-optically
generated component of polarization turns out to be the same for both
directions of incident polarization, that is, tsp = tps. The transverse
effect is very different from both polar and longitudinal effects. With
s-polarized incident light, where the optical E-field is parallel to the
direction of the B-field in the slab, there is no MO effect whatsoever,
but for p-polarized light the medium exhibits an effective refractive
index n = [e + (e�2/e)]�.
Thus in the transverse case neither s- nor p-polarized beams undergo polarization
rotation, but the magnitude of the transmitted p-light shows a weak dependence
on magnetization, that is, Tp = |tp|2 becomes
a function of the strength of the B-field. The transverse effect is not
bipolar, so changing the direction of the B-field from +Y to �Y does not
alter the magnitude of Tp. For a 20mm-thick
slab of transparent material with a fairly large MO coefficient (e = 5.5,
e� = 0.1i)
Figure 11 shows computed plots of Tp(0) (i.e.,
transmission in the absence of the B-field, when e� = 0)
and DTp = Tp - Tp(0)
versus the angle of incidence q.7
Note, in particular, that DTp � 0
around the Brewster angle qB = 66.9�,
where vanishing surface reflectivity results in minimal interference effects. Figure
11. Transverse Faraday effect, arising when
a p-polarized plane wave (l = 550 nm)
is incident at oblique angle q
on a 20 mm-thick
slab. The slab (e = 5.5)
is magnetized along the Y-axis, as shown in Figure 9(b). In the absence
of the B-field, e� = 0,
and the transmission of the slab for a p-polarized incident beam is denoted
by Tp(0). When a strong B-field is introduced
(corresponding to e� = 0.1i
in this case), the transmission changes to Tp. Shown
here is the transmission differential DTp = Tp - Tp(0)
as function of q.
References
Home
| About MM Research, Inc. | Online
Publications © Copyright 1987-2011, MM Research, Inc. 5748 N. Camino del Conde, Tucson, Arizona 85718 |