Professional Documents
Culture Documents
Verdeyen Laser - Electronics.3e
Verdeyen Laser - Electronics.3e
} (1.6.1)
In communications, this theorem says that a minimum bandwidth Aw is required to pass
a pulse with a rise time Ar. If we multiply both sides of the equation by A = h/2z, we
obtain formally a relation equivalent to the Heisenberg* uncertainty principle:
h
AEAr> (1.6.2)
Itis not a very interesting exercise in transform theory to prove that any two conjugate
variables (such as and 1), which are related by the Fourier transform, obey (1.6.1). The
genius of Heisenberg was in relating a physical problem to a mathematical abstraction.
Let us now turn to other conjugate variables. For instance, k; is the Fourier transform
variable with its conjugate x, ky with y, and k, with z. Once (1.6.1) is accepted, the same
theory of Fourier transforms yields
Vv
Ak,Ax >
AkyAy >
Ak. Az >
Vv
(1.6.3)
If we again multiply # = h/27 and identify Hk as the momentum, we obtain the conven-
tional form of Heisenberg’s uncertainty relations. These relationships are summarized in
“Whether the factor in (1.6.1) should be 1, +, or some other number close to 1 depends on how Aw and
Ar are defined14 Review of Electromagnetic Theory Chap. 1
TABLE 1.1
Conjugate
Item Physical variable Relation
o Angular frequency + (time) AwAt >
ke Propagationalongx x AkyAx >
ky Propagation along yy Ak, Ay >
k, Propagationalongz z AK Az >
E fiw = energy t AEAr > h/4n
Ps Momentumalongx x Ap.Ax >
Py Momentumalongy —y Ap, Ay >
Momentum along z z Ap.Az = h/4a
Table 1.1 Note that the uncertainty principle says nothing whatsoever about the relation
between nonconjugate variables.
Before we leave this topic, it is worthwhile to have a more precise definition of the
term “uncertainty”: it is the rms value of the deviation of the parameter from its average
value. For instance, if the transverse variation of the electric field of an optical beam were
given by
yy
E(y) = Eo exp | — (2) (1.6.4)
wo
then the average location of the field is at y = 0 and the “uncertainty” Ay is found from
400
f © - OP E*Q)dy
f E°(y)dy
ey
(Ay)? (1.6.5)
Inother words, the mathematical formula for the field can also be interpreted as a probability
function. The Fourier transform (in k, space) is given by
z
E(ky) = 2'?woEo exp [- (2) | (1.6.6)
Thus there is a distribution of ky wave vectors around ky = 0 and thus the “uncertainty” of
kyis
-to0
(ky — 0)? E(ky)dky
(ky = =2
ee
f E°(ky)dky
(1.6.7)Sec. 1.7 Spreading of an Electromagnetic Beam 15
Itis left for a problem to show that this particular field distribution has the minimum value
permitted: Ay - Ak, = 1/2.
1.7 SPREADING OF AN ELECTROMAGNETIC BEAM
Let us use the uncertainty relationships to predict the spread of a beam of light energy. Now
we know that this beam is traveling more or less at the velocity of light, c; hence, the wave
vector k; is very well defined atk. = «w/c (and, sure enough, the beam is almost everywhere
along the z axis). But if this is a “beam,” its extent in the transverse dimension is limited to
the beam diameter, as shown in Fig. 1.2.
If we assume that this “beam” has a smooth “Gaussian-like” spatial extent in the y
direction of the form given by (1.6.4)
2
E(y) = Eoexp [(-Z) |
then we must also allow for a spread in wave vectors centered around ky
kyu \?
Eky) = 2"? wo Eo exp - ( ot) |
This interrelationship is sketched in Fig. 1.3 on page 16.
Although a Gaussian spatial envelope is unique in the sense that itis also a Gaussian
in k space, the conclusions are the same irrespective of what is chosen for E(y).
uncertainty relations to predict the beam.
diameter along the propagation path.
4
'
I
\
i
'
1s
i>
i— sy
| :
| oa.
=
| . w
1 be
| es
—
aN
\—> yy
|= o [-(3)'}
1
tf we FIGURE 1.2. Beam of light diameter 21,
i passing the surface z = 0. We will use the
'
I16 Review of Electromagnetic Theory Chap. 1
EO)
@ )
ship between (a) the spatial extent of a beam and (b) the wave
‘Thus, we can construct a diagram for the propagation vectors ky and k; as shown in
Fig. 1.4, It is obvious that the angle 0/2 is given by
sb
2k mw
or (17.1)
a.
% = ——
zw
Thus, a large beam does not spread. Indeed, a uniform plane wave (one with wp = 00) has
a zero spread, in accordance with every elementary text on electromagnetic theory. (It has
no place to go!)
FIGURE 1.4, Vector addition of k, and
‘Ak, to estimate the beam spread.
Itis instructive to consider some numbers here. Let. = 694.3nm and 2w9 = 0.1 em;
then 0p is 8.8 x 10~ rad, To achieve the same beam spread at 10-cm wavelength would
require an antenna aperture 219 of 144 m, Such a small divergence of an optical beam
justifies the simple ray-tracing approach of Chapter 2.
1.8 WAVE PROPAGATION IN ANISOTROPIC MEDIA
Materials that are anisotropic to electromagnetic waves have many uses in optical electron-
ics: modulation, sensing, and harmonic generation are just a few examples. Indeed, most
crystalline materials are anisotropic and even some of the amorphous ones, such as glass,
become so when subjected to an electric field, a magnetic field, or mechanical stress. This
section introduces the formalism for handling such cases.Sec. 1.8 Wave Propagation in Anisotropic Media 7
‘We limit our attention to uniaxial media whose dielectric “constant” depends on the
direction of the electrical field, and thus the displacement vector D is described by a matrix
multiplication of € with the electric field E.
Dy ed: 0 Ey
Dy |=] 0 « 0 E, (8.1)
D: 00 oe] |E
Our goal is to predict the value of the wave vector k as the wave propagates at an angle 0
with respect to the z axis (the optical axis) as shown in Fig. 1.5.
From the algebraic form of Maxwell's equations, we know that the wave vector k is
perpendicular to D in any and all cases—anisotropy or no anisotropy!
kx h=-oD
k-(k x H) =0 = —ok-D
(1.8.2)
Hence there is one orientation of the electric field where we know’ the answer for the
orientation of the fields with respect to k. This is shown in Fig. 1.5(b), and since the case
is so “ordinary,” it is given that name. Note that if k is constrained to the yz plane, then
D is always in the x direction, and thus E = £,a,. The same argument can be applied to
the case where the displacement vector is perpendicular to the plane containing k and the
z axis, the so-called optic axis. For such cases, the propagation constant is given by
= wppever
or
ki 1 independent of 0) 1.8.3)
“0 = (independent o 8.
7 a pe ¢
If, however, D is not perpendicular to the plane containing k and the optic axis (i.e.,
(a, x k]-D = 0) as shown in Fig. 1.5(c), we have a problem. D is still perpendicular to k,
since (D-k = 0), but E isnot! Hence we can expect a mixture of ¢; and € in the expression
for the propagation constant, and a somewhat “extraordinary” behavior as a function of 8,
a task to which we turn.
For this polarization shown in Fig. 1.5(c), k and D can be expressed as
k = k(cos a, + sind ay) (1.8.4a)
D = D(-cos@ a, + sind a.) (1.8.4)
(Note thatk - D = 0.)
We use (1.8.4b) in conjunction with (1.8.1) to find E:
D
Ey = —[-cos 6] (1.8.5a)
0€1
£, = 2 Isin9] (1.8.5)
&e2Review of Electromagnetic Theory
@
2 (b)
(extraordinary)
y H
©
FIGURE 1.5. Orientation of k, E, and D for a uniaxial crystal. (a) The general problem.
(b) The ordinary wave. (c) The extraordinary wave.
Chap. 1Sec. 1.8 Wave Propagation in Anisotropic Media 19
Now it is a straightforward exercise in vector analysis to show (see Problem 1.3) that
D-D
P= Ome 5 (1.8.6a)
.
mY 1 ED
( : ae DD (1.8.66)
where the effective index is defined by k/ky = neq. Combining (1.8.6b) with (1.8.5) yields
1 cos? 6 sin? 0
a 1.8.7,
mad o
The forms of normalized propagation vector (k/ko) expressed by (1.8.3) and (1.8.7)
are conveniently shown on a graph called the index surface (see Fig. 1.6). Equation (1.8.3)
states that the effective index for the ordinary wave is independent of the angle 0. Hence
it is shown as a circle. The effective index for extraordinary wave does depend on @ in the
form of an ellipse.
Itis apparent from Fig. 1.6 and from (1.8.3) and (1.8.7) that the phase constants for the
ordinary and extraordinary waves are not equal for 9 # 0. This fact plays a critical role in
nonlinear optics where it is crucial that the phase constants of, for example, the fundamental
wave and any harmonic or intermodulation terms, must be synchronized. Fortunately, the
dielectric constants are not constant with frequency (i.¢., 4), and thus it is possible to choose
a phase matching angle 8, such that the effective index for the fundamental frequency «,
when propagated as an ordinary (extraordinary) wave, equals the effective index for the
second (third, etc.) harmonic when it is propagated as an extraordinary (ordinary) wave.
Ordinary wave
FIGURE 1.6, The index ellipsoid for a
uniaxial crystal.20. Review of Electromagnetic Theory Chap. 1
1.9 ELEMENTARY BOUNDARY VALUE PROBLEMS IN OPTICS
The propagation of electromagnetic waves is determined by Maxwell’s equations, but these
are incomplete without a specification of boundary conditions. After all, they are partial
differential equations that presume that all field variables and material properties are con-
tinuous functions of the coordinates. However, we will have many occasions to consider
abrupt junctions between different materials (windows, mirrors, etc.) where the electrical
parameters are different, and, as a consequence, the field variables change discontinuously.
Most elementary texts derive the relationship between the tangential and normal
components of the field at each side of an abrupt interface:
a, x (E, — Ey) = 0 (1.9.1a)
an «(Di — Da) = per (1.9.1b)
an x (Hy — H2) = Jeo (1.9.10)
a, - (B, — B2) = 0 (1.9.14)
where a, is a unit vector from 2 to 1 and perpendicular to the interface. The concept of
a surface charge, ps, and surface current, J,, both existing in zero depth in medium 2,
are useful approximations at low frequencies, v < 10'? Hz, but those approximations are
almost never utilized in the optical domain. Hence, we will let the right-hand side of (1.9.1b)
and (1.9.1¢) be zero.
The formal method of handling the interface problem is to first solve Maxwell’s equa-
tions in the two media and then match the fields at the boundary with (1.9.1a) and (1.9.1c).
Its sufficient to match tangential components only, because the normal components will
then be matched automatically, provided the fields in the respective media obey Maxwell’s
equations.
Many times we can sidestep a lot of dull mathematics implied by what we just did
by applying some elementary physical reasoning. Some very important examples of this
approach are shown below.
1.9.1 Snell's Law
Consider a uniform plane wave (upw) impinging on the interface shown in Fig. 1.7 making an
angle 6, with respect to the normal to the surface. The discontinuity generates a second wave
Reflected
‘Transmitted
FIGURE 1.7. Geometry for Snell’s Law.Sec. 1.9 Elementary Boundary Value Problems in Optics 2
at an angle 6, and a reflected wave. We could grit our teeth and match field components at
the interface and solve the problem completely. This procedure is necessary if the amplitude
and phase of the transmitted and reflected waves are desired. However, if only the direction
is desired, the procedure can be greatly simplified.
The point to be remembered is that the incident wave is the source, and the transmitted
and reflected waves are the responses. Hence the phases of both responses, whatever they
are, must be synchronized with respect to the source along the boundary where the responses
are generated.
The relative phase of the source along the interface is
$ = (w/c)n sin (1.9.2)
and this must be the phase of both responses as measured along the interface. If medium 1
is isotropic, this fact forces the incident and reflected waves to make the same angle with
respect to the normal.
For the transmitted field, we force the phases along the boundary to be the same:
(w/c) ny sin 6; = (w/c) ng sin (1.9.3a)
or
ny sin 8, = nz sin (1.9.36)
For an anisotropic medium for 2, the incident wave can generate two transmitted waves,
but both must remain tied to the phase of incident wave along the interface.
1.9.2 Brewster's Angle
Windows oriented at Brewster's angle are commonly used on gas lasers because, in principle,
they transmit waves without reflection for one polarization of the electric field. The geometry
of the electromagnetic problem is shown in Fig. 1.8 for two possible polarizations of the
incident field. In both cases, Snell’s law is applicable, and thus the wave vector k is bent
toward the normal in the window material.
‘There are some artifacts added to Fig. 1.8 to help visualize the physical situation:
The orientations of the induced dipoles in the dielectric material are shown, for it is their
reradiation that generates the reflected wave.
Now every elementary test in electromagnetic theory shows that electric dipoles ra-
diate perpendicular to the axis and not along it. Thus for the TE orientation there is no
problem in generating a reflected wave. However, for the TM case and a particular angle of
the incident wave, the reflected wave would try to come off the ends of the dipole, which is
impossible. Hence there is 10 reflected wave when the angles 0; + #2 = 2/2. Combining
this fact with Snell’s law yields an expression for an angle of zero reflection:
=
DU
ny sin @; = np sin 62 (Snell’s law) (1.9.4)
Hence ny sin 0, = ng sin(r/2 ~ 0)) = nz cos 6fe)
Review of Electromagnetic Theory Chap. 1
(a) TM or“p" polarized
(b) TE ors” polarized
(©) Dipole radiation
FIGURE 1.8, Brewster's angle windows.Sec. 1.10 Coherent Electromagnetic Radiation 23
Therefore tan, = “2 Brewster’ angle) (1.9.5)
nm
It should be emphasized that mathematics involved in matching fields across an i
terface will lead to the same result, but we should appreciate the physical reasoning just
presented also.
COHERENT ELECTROMAGNETIC RADIATION
Let us reiterate the goals of this book: to understand the physical bases for the generation,
transmission, and detection of electromagnetic radiation in the “optical” portion of the
spectrum. But we should be more precise and focus our attention on a specific characteristic
that distinguishes the laser from a simple lamp.
The distinguishing characteristic is the generation of coherent electromagnetic radia~
tion. Now, the topic of coherence is most involved and complex to describe with precision,
but it is relatively easy to understand the first-order consequences.
Most who have had electronic experience at low frequencies, say less than 30 GHz,
with classical generators never address this subject, because most of our generators had a
long coherence time or length. In other words, they are almost perfectly coherent. But what
does this mean, and how would we measure either coherence time or length?
Ina loose sort of way, coherence time is the net delay that can be inserted in a wave
train and still obtain interference. Since electromagnetic waves travel with a velocity of c,
the longitudinal coherence length is simply c times the coherence time. Note that the key
word is interference. Let us illustrate these ideas with a “thought” experiment taken from
low-frequency electronics and compare it with a similar experiment at optical frequencies
(visible wavelengths).
Reflector
Detector Vou o EF
FIGURE 1.9. Simple interference experiment.24 Review of Electromagnetic Theory Chap. |
Consider a simple transmission-line measurement of the standing-wave ratio on a
short-circuited transmission system as shown in Fig. 1.9. To make the conventional “slotted-
line” measurement of the “voltage” standing-wave ratio (VSWR), We move a short dipole
antenna and a rectifying diode along the z axis. The output of the detector is proportional
to the square of the electric field (usually); hence, the relative output of the detector would
be as shown in Fig. 1.10. The VSWR, Vnuax/Vains is very large, and for all practical purposes
it is infinite. This is precisely what we observe in a normal laboratory." Even elementary
theory would predict this result, as is demonstrated next.
The electric field traveling to the right is given by
E* = Eyexp(—jkz) z