206 Electromagnetic Waves & Antennas – S. J.

Orfanidis

7 δi =
ω
c0
ni li cos θi =
2π
λ
ni li cos θi =
2π
λ

 n2 sin2 θa
li ni 1 − a 2
ni
(7.1.2)

 
Multilayer Film Applications where we used Eq. (7.1.1) to write cos θi = 1 − sin2 θi = 1 − n2a sin2 θa /ni2 . The
transverse reflection coefficients at the M + 1 interfaces are defined as in Eq. (5.1.1):

nT,i−1 − nTi
ρTi = , i = 1, 2, . . . , M + 1 (7.1.3)
nT,i−1 + nTi
where we set nT0 = nTa , as in Sec. 5.1. and nT,M+1 = nTb . The transverse refractive
indices are defined in each medium by Eq. (6.2.13):

⎧ ni
⎨ , TM polarization
nTi = cos θi , i = a, 1, 2, . . . , M, b (7.1.4)
⎩
7.1 Multilayer Dielectric Structures at Oblique Incidence ni cos θi , TE polarization

Using the matching and propagation matrices for transverse fields that we discussed To obtain the layer recursions for the electric fields, we apply the propagation matrix
in Sec. 6.3, we derive here the layer recursions for multiple dielectric slabs at oblique (6.3.5) to the fields at the left of interface i + 1 and propagate them to the right of the
incidence. interface i, and then, apply a matching matrix (6.3.11) to pass to the left of that interface:
Fig. 7.1.1 shows such a multilayer structure. The layer recursions relate the various    
ETi+ 1 1 ρTi ejδi 0 ET,i+1,+
field quantities, such as the electric fields and the reflection responses, at the left of =
ETi− τTi ρTi 1 0 e−jδi ET,i+1,−
each interface.
Multiplying the matrix factors, we obtain:

  
ETi+ 1 ejδi ρTi e−jδi ET,i+1,+
= , i = M, M − 1, . . . , 1 (7.1.5)
ETi− τTi ρTi ejδi e−jδi ET,i+1,−

This is identical to Eqs. (5.1.2) with the substitutions ki li → δi and ρi → ρTi . The
recursion is initialized at the left of the (M + 1)st interface by performing an additional
matching to pass to the right of that interface:
  

ET,M+1,+ 1 1 ρT,M+1 ET.M+1,+
= (7.1.6)
ET,M+1,− τT,M+1 ρT,M+1 1 0

Fig. 7.1.1 Oblique incidence on multilayer dielectric structure. It follows now from Eq. (7.1.5) that the reflection responses, ΓTi = ETi− /ETi+ , will
satisfy the identical recursions as Eq. (5.1.5):
We assume that there are no incident fields from the right side of the structure.
The reflection/refraction angles in each medium are related to each other by Snell’s law ρTi + ΓT,i+1 e−2jδi
ΓTi = , i = M, M − 1, . . . , 1 (7.1.7)
applied to each of the M + 1 interfaces: 1 + ρTi ΓT,i+1 e−2jδi

and initialized at ΓT,M+1 = ρT,M+1 . Similarly, we obtain the following recursions for
na sin θa = ni sin θi = nb sin θb , i = 1, 2, . . . , M (7.1.1)
the total transverse electric and magnetic fields at each interface (they are continuous
It is convenient also to define by Eq. (6.3.8) the propagation phases or phase thick- across each interface):
nesses for each of the M layers, that is, the quantities δi = kzi li . Using kzi = k0 ni cos θi ,
where k0 is the free-space wavenumber, k0 = ω/c0 = 2πf /c0 = 2π/λ, we have for   
ETi cos δi jηTi sin δi ET,i+1
i = 1, 2, . . . , M: = −1 , i = M, M − 1, . . . , 1 (7.1.8)
HTi jηTi sin δi cos δi HT,i+1

205
7.2. Single Dielectric Slab 207 208 Electromagnetic Waves & Antennas – S. J. Orfanidis

where ηTi are the transverse characteristic impedances defined by Eq. (6.2.12) and re-
lated to the refractive indices by ηTi = η0 /nTi . The wave impedances, ZTi = ETi /HTi ,
satisfy the following recursions initialized by ZT,M+1 = ηTb :

ZT,i+1 + jηTi tan δi

ZTi = ηTi , i = M, M − 1, . . . , 1 (7.1.9)
ηTi + jZT,i+1 tan δi

The MATLAB function multidiel that was introduced in Sec. 5.1 can also be used
in the oblique case with two extra input arguments: the incidence angle from the left
and the polarization type, TE or TM. Its full usage is as follows:

[Gamma1,Z1] = multidiel(n,L,lambda,theta,pol); % multilayer dielectric structure

Fig. 7.2.1 Oblique incidence on single dielectric slab.
where theta is the angle θ = θa and pol is one of the strings ’te’ or ’tm’. If the angle
and polarization arguments are omitted, the function defaults to normal incidence for
which TE and TM are the same. The other parameters have the same meaning as in
λ0 f f
Sec. 5.1. δ1 = 2π L1 cos θ1 = 2π L1 cos θ1 = π (7.2.2)
λ f0 f1
In using this function, it is convenient to normalize the wavelength λ and the optical
lengths ni li of the layers to some reference wavelength λ0 . The frequency f will be
f0
normalized to the corresponding reference frequency f0 = c0 /λ0 . f1 = (7.2.3)
2L1 cos θ1
Defining the normalized thicknesses Li = ni li /λ0 , so that ni li = Li λ0 , and noting
that λ0 /λ = f /f0 , we may write the phase thicknesses (7.1.2) in the normalized form: At frequencies that are integral multiples of f1 , f = mf1 , the reflection response
vanishes because 2δ1 = 2π(mf1 )/f1 = 2πm and e−2jδ1 = 1. Similarly, at the half-
λ0 f integral multiples, f = (m + 0.5)f1 , the response is maximum because e−2jδ1 = −1.
δi = 2π Li cos θi = 2π Li cos θi , i = 1, 2, . . . , M (7.1.10)
λ f0 Because f1 depends inversely on cos θ1 , then as the angle of incidence θa increases,
cos θ1 will decrease and f1 will shift towards higher frequencies. The maximum shift
Typically, but not necessarily, the Li are chosen to be quarter-wavelength long at
will occur when θ1 reaches its maximum refraction value θ1c = asin(na /n1 ) (assuming
λ0 , that is, Li = 1/4. This way the same multilayer design can be applied equally well
na < n1 .)
at microwave or at optical frequencies. Once the wavelength scale λ0 is chosen, the
Similar shifts occur for the 3-dB width of the reflection response notches. By the
physical lengths of the layers li can be obtained from li = Li λ0 /ni .
same calculation that led to Eq. (4.5.9), we find for the 3-dB width with respect to the
variable δ1 :
7.2 Single Dielectric Slab
∆δ1 1 − ρ2T1
tan =
Many features of oblique incidence on multilayer slabs can be clarified by studying the 2 1 + ρ2T1
single-slab case, shown in Fig. 7.2.1. Assuming that the media to the left and right are
Setting ∆δ1 = π∆f /f1 , we solve for the 3-dB width in frequency:
the same, na = nb , it follows that θb = θa and also that ρT1 = −ρT2 . Moreover, Snell’s
law implies na sin θa = n1 sin θ1 . 2f1 1 − ρ2T1
Because there are no incident fields from the right, the reflection response at the ∆f = atan (7.2.4)
π 1 + ρ2T1
left of interface-2 is: ΓT2 = ρT2 = −ρT1 . It follows from Eq. (7.1.7) that the reflection
response at the left of interface-1 will be: The left/right bandedge frequencies are f1 ± ∆f /2. The dependence of ∆f on the
incidence angle θa is more complicated here because ρT1 also depends on it.
ρT1 + ρT2 e−2jδ1 ρT1 (1 − e−2jδ1 ) In fact, as θa tends to its grazing value θa → 90o , the reflection coefficients for
ΓT1 = = (7.2.1)
1 + ρT1 ρT2 e−2 jδ1 1 − ρ2T1 e−2jδ1 either polarization have the limit |ρT1 | → 1, resulting in zero bandwidth ∆f . On the
other hand, at the Brewster angle, θaB = atan(n1 /na ), the TM reflection coefficient
This is analogous to Eq. (4.5.4). According to Eq. (7.1.10), the phase thickness can be
vanishes, resulting in maximum bandwidth. Indeed, because atan(1)= π/4, we have
written in the following normalized form, where L1 = n1 l1 /λ0 :
∆f max = 2f1 atan(1)/π = f1 /2.
7.2. Single Dielectric Slab 209 210 Electromagnetic Waves & Antennas – S. J. Orfanidis

Fig. 7.2.2 illustrates some of these properties. The refractive indices were na = nb = 7.3 Antireflection Coatings at Oblique Incidence
1 and n1 = 1.5. The optical length of the slab was taken to be half-wavelength at the
reference wavelength λ0 , so that n1 l1 = 0.5λ0 , or, L1 = 0.5. Antireflection coatings are typically designed for normal incidence and then used over
a limited range of oblique incidence, such as up to about 30o . As the angle of incidence
θ θa = 75o θ θa = 85o increases, the antireflection band shifts towards lower wavelengths or higher frequen-
1 1
cies. Any designed reflection zeros at normal incidence are no longer zeros at oblique
incidence.
0.8 0.8
If a particular angle of incidence is preferred, it is possible to design the antireflection
coating to match that angle. However, like the case of normal design, the effectiveness
| ΓT1 ( f )|2

| ΓT1 ( f )|2
0.6 TE 0.6 TE
TM TM of this method will be over an angular width of approximately 30o about the preferred
normal normal
∆ ∆f angle.
0.4 0.4
To appreciate the effects of oblique incidence, we look at the angular behavior of
0.2 0.2 our normal-incidence designs presented in Figs. 5.2.1 and 5.2.3.
The first example was a two-layer design with refractive indices na = 1 (air), n1 =
0 0 1.38 (magnesium fluoride), n2 = 2.45 (bismuth oxide), and nb = 1.5 (glass). The de-
0 1 2 3 0 1 2 3
f/f0 f/f0 signed normalized optical lengths of the layers were L1 = 0.3294 and L2 = 0.0453 at
λ0 = 550 nm.
Fig. 7.2.2 TE and TM reflectances of half-wavelength slab. Fig. 7.3.1 shows the TE and TM reflectances |ΓT1 (λ)|2 as functions of λ, for the
incidence angles θ = 0o , 20o , 30o , 40o .
The graphs show the TE and TM reflectances |ΓT1 (f )|2 as functions of frequency
for the angles of incidence θ1 = 75o and θa = 85o . The normal incidence case is also TE polarization TM polarization
4 4
included for comparison.
 
The corresponding refracted angles were θ1 = asin na asin(θa )/n1 = 40.09o and 0o 0o

| ΓT1 (λ)|2 (percent)

| ΓT1 (λ)|2 (percent)

20o 20o
θ1 = 41.62o . Note that the maximum refracted angle is θ1c = 41.81o , and the Brewster 3
30o
3
30o
angle, θaB = 56.31o . 40o 40o

The notch frequencies were f1 = f0 /(2L1 cos θ1 )= 1.31f0 and f1 = 1.34f0 for the 2 2
angles θa = 75o and 85o . At normal incidence we have f1 = f0 /(2L1 )= f0 , because
L1 = 0.5.
1 1
The graphs also show the 3-dB widths of the notches, calculated from Eq. (7.2.4).
The reflection responses were computed with the help of the function multidiel with
the typical MATLAB code: 0 0
400 450 500 550 600 650 700 400 450 500 550 600 650 700
λ (nm) λ (nm)
na = 1; nb = 1;
n1 = 1.5; L1 = 0.5;
Fig. 7.3.1 Two-layer antireflection coating at oblique incidence.
f = linspace(0,3,401);
theta = 75;
We note the shifting of the responses towards lower wavelengths. The responses
G0 = abs(multidiel([na,n1,nb], L1, 1./f)).^2; are fairly acceptable up to about 20o –30o . The typical MATLAB code used to generate
Ge = abs(multidiel([na,n1,nb], L1, 1./f, theta, ’te’)).^2; these graphs was:
Gm = abs(multidiel([na,n1,nb], L1, 1./f, theta, ’tm’)).^2;
n = [1, 1.38, 2.45, 1.5]; L = [0.3294, 0.0453];
The shifting of the notch frequencies and the narrowing of the notch widths is evi- la0 = 550; la = linspace(400,700,101); pol=’te’;
dent from the graphs. Had we chosen θa = θaB = 56.31o , the TM response would have
G0 = abs(multidiel(n, L, la/la0)).^2 * 100;
been identically zero because of the factor ρT1 in Eq. (7.2.1). G20 = abs(multidiel(n, L, la/la0, 20, pol)).^2 * 100;
The single-slab case is essentially a simplified version of a Fabry-Perot interferometer G30 = abs(multidiel(n, L, la/la0, 30, pol)).^2 * 100;
[187], used as a spectrum analyzer. At multiples of f1 , there are narrow transmittance G40 = abs(multidiel(n, L, la/la0, 40, pol)).^2 * 100;
bands. Because f1 depends on f0 / cos θ1 , the interferometer serves to separate different
plot(la, [G0; G20; G30; G40]);
frequencies f0 in the input by mapping them onto different angles θ1 .
7.3. Antireflection Coatings at Oblique Incidence 211 212 Electromagnetic Waves & Antennas – S. J. Orfanidis

As we mentioned above, the design can be matched at a particular angle of incidence. late the normalized optical lengths from Li = δi /(2π cos θi ), i = 1, 2. The following
As an example, we choose θa = 30o and redesign the two-layer structure. MATLAB code illustrates these steps:
The design equations are still (5.2.2) and (5.2.1), but with the replacement of ni ,
n = [1, 1.38, 2.45, 1.5];
ρi by their transverse values nTi , ρTi , and the replacement of k1 l1 , k2 l2 by the phase
tha = 30; thi = asin(na*sin(pi*tha/180)./n);
thicknesses at λ = λ0 , that is, δ1 = 2πL1 cos θ1 and δ2 = 2πL2 cos θ2 . Moreover, we
must choose to match the design either for TE or TM polarization. nt = n.*cos(thi); % for TM use nt = n./cos(thi)
Fig. 7.3.2 illustrates such a design. The upper left graph shows the TE reflectance r = n2r(nt);

matched at 30o . The designed optical thicknesses are in this case, L1 = 0.3509 and
c = sqrt((r(1)^2*(1-r(2)*r(3))^2 - (r(2)-r(3))^2)/(4*r(2)*r(3)*(1-r(1)^2)));
L2 = 0.0528. The upper right graph shows the corresponding TM reflectance, which de2 = acos(c);
cannot be matched simultaneously with the TE case. G2 = (r(2)+r(3)*exp(-2*j*de2))/(1 + r(2)*r(3)*exp(-2*j*de2));
The lower graphs show the same design, but now the TM reflectance is matched at de1 = (angle(G2) - pi - angle(r(1)))/2;
if de1 <0, de1 = de1 + 2*pi; end
30o . The designed lengths were L1 = 0.3554 and L2 = 0.0386.
L = [de1,de2]/2/pi;
TE matched at 30o TM unmatched at 30o L = L./cos(thi(2:3));
4 4

la0 = 550; la = linspace(400,700,401);

30o 30o
| ΓT1 (λ)|2 (percent)

| ΓT1 (λ)|2 (percent)

3 20o 3 20o
40o 40o
G30 = abs(multidiel(n, L, la/la0, 30, ’te’)).^2 * 100;
0o 0o G20 = abs(multidiel(n, L, la/la0, 20, ’te’)).^2 * 100;
G40 = abs(multidiel(n, L, la/la0, 40, ’te’)).^2 * 100;
2 2
G0 = abs(multidiel(n, L, la/la0)).^2 * 100;

plot(la, [G30; G20; G40; G0]);

1 1

Our second example in Fig. 5.2.3 was a quarter-half-quarter 3-layer design with re-
0 0 fractive indices n1 = 1 (air), n1 = 1.38 (magnesium fluoride), n2 = 2.2 (zirconium oxide),
400 450 500 550 600 650 700 400 450 500 550 600 650 700
λ (nm) λ (nm) n3 = 1.63 (cerium fluoride), and nb = 1.5 (glass). The optical lengths of the layers were
TM matched at 30o TE unmatched at 30o L1 = L3 = 0.25 and L2 = 0.5.
4 4
Fig. 7.3.3 shows the TE and TM reflectances |ΓT1 (λ)|2 as functions of λ, for the
30o 30o
incidence angles θ = 0o , 20o , 30o , 40o .
| ΓT1 (λ)|2 (percent)

| ΓT1 (λ)|2 (percent)

3 20o 3 20o
40o 40o TE polarization TM polarization
0o 0o 4 4

2 2
0o 0o

| ΓT1 (λ)|2 (percent)

| ΓT1 (λ)|2 (percent)

3 20o 3 20o
30o 30o
1 1 40o 40o

2 2

0 0
400 450 500 550 600 650 700 400 450 500 550 600 650 700
λ (nm) λ (nm)
1 1

Fig. 7.3.2 Two-layer antireflection coating matched at 30 degrees.

0 0
400 450 500 550 600 650 700 400 450 500 550 600 650 700
The design steps are as follows. First, we calculate the refraction angles in all media λ (nm) λ (nm)

from Eq. (7.1.1), θi = asin(na sin θa /ni ), for i = a, 1, 2, b. Then, assuming TE polariza-
tion, we calculate the TE refractive indices for all media nTi = ni cos θi , i = a, 1, 2, b. Fig. 7.3.3 Three-layer antireflection coating at oblique incidence.
Then, we calculate the transverse reflection coefficients ρTi from Eq. (7.1.3) and use
them to solve Eq. (5.2.2) and (5.2.1) for the phase thicknesses δ1 , δ2 . Finally, we calcu- The responses are fairly acceptable up to about 20o –30o , but are shifted towards
lower wavelengths. The typical MATLAB code used to generate these graphs was:
7.4. Omnidirectional Dielectric Mirrors 213 214 Electromagnetic Waves & Antennas – S. J. Orfanidis

n = [1, 1.38, 2.2, 1.63, 1.5]; L = [0.25, 0.50, 0.25]; A necessary (but not sufficient) condition for omnidirectional reflectivity for both
polarizations is that the maximum angle of refraction θH,max inside the first layer be
la0 = 550; la = linspace(400,700,401);
less than the Brewster angle θB of the second interface, that is, the high-low interface,
G0 = abs(multidiel(n, L, la/la0)).^2 * 100; so that the Brewster angle can never be accessed by a wave incident on the first interface.
G20 = abs(multidiel(n, L, la/la0, 20, ’te’)).^2 * 100; If this condition is not satisfied, a TM wave would not be reflected at the second and all
G30 = abs(multidiel(n, L, la/la0, 30, ’te’)).^2 * 100; subsequent interfaces and will transmit through the structure.
G40 = abs(multidiel(n, L, la/la0, 40, ’te’)).^2 * 100; 
Because sin θH,max = na /nH and tan θB = nL /nH , or, sin θB = nL / n2H + n2L , the
plot(la, [G0; G20; G30; G40]); condition θH,max< θB , or the equivalent condition sin θH,max < sin θB , can be written
as na /nH < nL / n2H + n2L , or

7.4 Omnidirectional Dielectric Mirrors nH nL

na <  (7.4.3)
Until recently, it was generally thought that it was impossible to have an omnidirectional n2H + n2L
dielectric mirror, that is, a mirror that is perfectly reflecting at all angles of incidence
We note that the exact opposite of this condition is required in the design of multi-
and for both TE and TM polarizations. However, such mirrors are possible and have
layer Brewster polarizing beam splitters, discussed in the next section.
recently been manufactured [322,323] and the conditions for their existence clarified
In addition to condition (7.4.3), in order to achieve omnidirectional reflectivity we
[322–326].
must require that the high-reflectance bands have a common overlapping region for all
We consider the same dielectric mirror structure of Sec. 5.3, consisting of alternating
incidence angles and for both polarizations.
layers of high and low index. Fig. 7.4.1 shows such a structure under oblique incidence.
To determine these bands, we note that the entire discussion of Sec. 5.3 carries
There are N bilayers and a total of M = 2N + 1 single layers, starting and ending with
through unchanged, provided we use the transverse reflection coefficients and trans-
a high-index layer.
verse refractive indices. For example, the transverse version of the bilayer transition
matrix of Eq. (5.3.5) will be:


1 ej(δH +δL ) − ρ2T ej(δH −δL ) −2jρT e−jδH sin δL
FT = −j(δH +δL ) (7.4.4)
1 − ρT2 2jρT ejδH sin δL e − ρ2T e−j(δH −δL )

where ρT = (nHT − nLT )/(nHT + nLT ) and:

⎧ nH ⎧ nL
⎨ ⎨ (TM polarization)
nHT = cos θH nLT = cos θL (7.4.5)
⎩ ⎩ (TE polarization)
nH cos θH nL cos θL

Fig. 7.4.1 Dielectric mirror at oblique incidence. Explicitly, we have for the two polarizations:

nH cos θL − nL cos θH nH cos θH − nL cos θL

The incidence angles on each interface are related by Snell’s law: ρTM = , ρTE = (7.4.6)
nH cos θL + nL cos θH nH cos θH + nL cos θL
na sin θa = nH sin θH = nL sin θL = nb sin θb (7.4.1) The trace of FT is as in Eq. (5.3.13):

The phase thicknesses within the high- and low-index layers are in normalized form:
cos(δH + δL )−ρ2T cos(δH − δL )
a= (7.4.7)
1 − ρ2T
f f
δH = 2π LH cos θH , δL = 2π LL cos θL (7.4.2)
f0 f0 The eigenvalues of the matrix FT are λ± = e±jKl , where K = acos(a)/l and l = lH +lL .
where LH = nH lH /λ0 , LL = nL lL /λ0are the optical thicknesses normalized to some λ0 , The condition a = −1 determines the bandedge frequencies of the high-reflectance
bands. As in Eq. (5.3.16), this condition is equivalent to:
and f0 = c0 /λ0 . Note also, cos θi = 1 − n2a sin2 θa /n2i , i = H, L.
7.4. Omnidirectional Dielectric Mirrors 215 216 Electromagnetic Waves & Antennas – S. J. Orfanidis

At normal incidence, we have λc = 2(lH nH + lL nL ). For quarter-wavelength designs


2 δH + δL
 
2 δH − δL
 at λ0 at normal incidence, we have L+ = 1/4 + 1/4 = 1/2, so that λc = λ0 .
cos = ρ2T cos (7.4.8)
2 2 The accuracy of the approximate solution (7.4.14) depends on the ratio d = L− /L+ .
Even if at normal incidence the layers were quarter-wavelength with LH = LL = 0.25,
Defining the quantities L± = LH cos θH ± LL cos θL and the normalized frequency
the equality of LH and LL will no longer be true at other angles of incidence. In fact, the
F = f /f0 , we may write:
quantity d is an increasing function of θa . For larger values of d, the exact solution of
δH ± δL f (7.4.10) can be obtained by the following iteration:
=π (LH cos θH ± LL cos θL )= πFL± (7.4.9)
2 f0
initialize with F1 = F2 = 0,
Then, taking square roots of Eq. (7.4.8), we have:
for i = 0, 1, . . . , Niter , do:
1  
cos(πFL+ )= ±|ρT | cos(πFL− ) F1 = acos |ρT | cos(πF1 L− ) (7.4.17)
πL+
The plus sign gives the left bandedge, F1 = f1 /f0 , and the minus sign, the right 1  
F2 = acos −|ρT | cos(πF2 L− )
bandedge, F2 = f2 /f0 . Thus, F1 , F2 are the solutions of the equations: πL+

Evidently, the i = 0 iteration gives the zeroth-order solution (7.4.14). The iteration
cos(πF1 L+ ) = |ρT | cos(πF1 L− )
(7.4.10) converges extremely fast, requiring only 3–4 iterations Niter . The MATLAB function
cos(πF2 L+ ) = −|ρT | cos(πF2 L− ) omniband implements this algorithm. It has usage:

The bandwidth and center frequency of the reflecting band are: [F1,F2] = omniband(na,nH,nL,LH,LL,theta,pol,Niter) % bandedge frequencies
[F1,F2] = omniband(na,nH,nL,LH,LL,theta,pol) % equivalent to Niter = 0

∆f fc F1 + F2
= ∆F = F2 − F1 , = Fc = (7.4.11) where theta is the incidence angle in degrees, pol is one of the strings ’te’ or ’tm’ for
f0 f0 2
TE or TM polarization, and Niter is the desired number of iterations. If this argument
The corresponding bandwidth in wavelengths is defined in terms of the left and right is omitted, only the i = 0 iteration is carried out.
bandedge wavelengths: It is straightforward but tedious to verify the following facts about the above solu-
tions. First, f1 , f2 are increasing functions of θa for both TE and TM polarizations. Thus,
λ0 c0 λ0 c0 the center frequency of the band fc = (f1 + f2 )/2 shifts towards higher frequencies with
λ1 = = , λ2 = = , ∆λ = λ2 − λ1 (7.4.12)
F2 f2 F1 f1 increasing angle θa . The corresponding wavelength intervals will shift towards lower
wavelengths.
An approximate solution of Eq. (7.4.10) can be obtained by setting L− = 0 in the
Second, the bandwidth ∆f = f2 − f1 is an increasing function of θa for TE, and a
right-hand sides of Eq. (7.4.10):
decreasing one for TM polarization. Thus, as θa increases, the reflecting band for TE
expands and that of TM shrinks, while their (slightly different) centers fc shift upwards.
cos(πF1 L+ )= |ρT | , cos(πF2 L+ )= −|ρT | (7.4.13)
In order to achieve omnidirectional reflectivity, the TE and TM bands must have a
with solutions: common overlapping intersection for all angles of incidence. Because the TM band is
always narrower than the TE band, it will determine the final common omnidirectional
acos(|ρT |) acos(−|ρT |) band.
F1 = , F2 = (7.4.14)
πL+ πL+ The worst case of overlap is for the TM band at 90o angle of incidence, which must
Using the trigonometric identities acos(±|ρT |)= π/2 ∓ asin(|ρT |), we obtain the overlap with the TM/TE band at 0o . The left bandedge of this TM band, f1,TM (90o ), must
bandwidth and center frequency: be less than the right bandedge of the 0o band, f2 (0o ). This is a sufficient condition for
omnidirectional reflectivity.
2f0 asin(|ρT |) f1 + f2 f0 Thus, the minimum band shared by all angles of incidence and both polarizations
∆f = f2 − f1 = , fc = = (7.4.15)
πL+ 2 2L+ will be [f1,TM (90o ), f2 (0o )], having width:

It follows that the center wavelength will be λc = c0 /fc = 2L+ λ0 or,

∆f min = f2 (0o )−f1,TM (90o ) (minimum omnidirectional bandwidth) (7.4.18)
λc = 2L+ λ0 = 2(lH nH cos θH + lL nL cos θL ) (7.4.16)
7.4. Omnidirectional Dielectric Mirrors 217 218 Electromagnetic Waves & Antennas – S. J. Orfanidis

In a more restricted sense, the common reflecting band for both polarizations and Frequency Response at 450 Frequency Response at 800
for angles up to a given θa will be [f1,TM (θa ), f2,TM (0o )] and the corresponding band- 100 100
width:

| ΓT1 (f)|2 (percent)

| ΓT1 (f)|2 (percent)

80 80
∆f (θa )= f2 (0o )−f1,TM (θa ) (7.4.19)
60 60
In addition to computing the bandwidths of either the TM or the TE bands at any
angle of incidence, the function omniband can also compute the above common band- 40 40
widths. If the parameter pol is equal to ’tem’, then F1 , F2 are those of Eqs. (7.4.18) and
TM TM
(7.4.19). Its extended usage is as follows: 20 TE fc0 fc
20 TE fc
0o 0o

[F1,F2] = omniband(na,nH,nL,LH,LL,theta,’tem’) % Eq. (7.4.19) 0 0

0 1 2 0 1 2
[F1,F2] = omniband(na,nH,nL,LH,LL,90,’tem’) % Eq. (7.4.18) f/f0 f/f0
[F1,F2] = omniband(na,nH,nL,LH,LL) % Eq. (7.4.18)

Fig. 7.4.3 TM and TE frequency responses for nH = 2.32, nL = 1.38.

Next, we discuss some simulation examples that will help clarify the above remarks.

Example 7.4.1: The first example is the angular dependence of Example 5.3.2. In order to flatten The arrows labeled fc0 and fc represent the (TM) band center frequencies at 0o and 45o or
out and sharpen the edges of the reflecting bands, we use N = 30 bilayers. Fig. 7.4.2 shows 80o . The calculated bandedges corresponding to 90o incidence were λ1 = λ0 /F2,TM (0o )=
the TE and TM reflectances |ΓT1 (λ)|2 as functions of the free-space wavelength λ, for the 429.73 nm and λ2 = λ0 /F1,TM (90o )= 432.16 nm, with bandwidth ∆λ = λ2 − λ1 = 2.43
two angles of incidence θa = 45o and 80o . nm. Thus, this structure does exhibit omnidirectional reflectivity, albeit over a very narrow
Fig. 7.4.3 depicts the reflectances as functions of frequency f . The refractive indices were band. The MATLAB code used to generate these graphs was:
na = 1, nH = 2.32, nL = 1.38, nb = 1.52, and the bilayers were quarter-wavelength
LH = LL = 0.25 at the normalization wavelength λ0 = 500 nm. na = 1; nb = 1.52; nH = 2.32; nL = 1.38;
LH = 0.25; LL = 0.25;
The necessary condition (7.4.3) is satisfied and we find for the maximum angle of refraction
and the Brewster angle: θH,max = 25.53o and θB = 30.75o Thus, we have θH,max < θB .
la0 = 500;
la = linspace(300,800,501);
Reflectance at 450 Reflectance at 800
th = 45; N = 30;
100 100
n = [na, nH, repmat([nL,nH], 1, N), nb];
L = [LH, repmat([LL,LH], 1, N)];
| ΓT1 (λ)|2 (percent)

| ΓT1 (λ)|2 (percent)

80 80
Ge = 100*abs(multidiel(n,L,la/la0, th, ’te’)).^2;
Gm = 100*abs(multidiel(n,L,la/la0, th, ’tm’)).^2;
60 60 G0 = 100*abs(multidiel(n,L,la/la0)).^2;

40 40 plot(la,Gm, la,Ge, la,G0);

TM TM
20 TE 20 TE [F10,F20] = omniband(na,nH,nL,LH,LL, 0, ’te’);
λc λc0 λc
0o 0o [F1e,F2e] = omniband(na,nH,nL,LH,LL, th,’te’);
0 0 [F1m,F2m] = omniband(na,nH,nL,LH,LL, th,’tm’);
300 400 500 600 700 800 300 400 500 600 700 800
λ (nm) λ (nm) [F1,F2] = omniband(na,nH,nL,LH,LL, th,’tem’);

Fig. 7.4.2 TM and TE reflectances for nH = 2.32, nL = 1.38. Because the reflectivity bands shrink with decreasing ratio nH /nL , if we were to slightly
decrease nH , then the TM band could be made to shift beyond the end of the 0o band and
On each graph, we have indicated the corresponding bandwidth intervals calculated with there would be no common overlapping reflecting band for all angles. We can observe this
omniband. The indicated intervals are for 0o incidence, for TE and TM, and for the common behavior in Fig. 7.4.4, which has nH = 2, with all the other parameters kept the same.
band Eq. (7.4.19) at θa . We observe the shifting of the bands towards higher frequencies, At 45o there is a common overlap, but at 80o , the TM band has already moved beyond the 0o
or lower wavelengths, and the shrinking of the TM and expanding of the TE bands, and the band, while the TE band still overlaps with the latter. This example has no omnidirectional
shrinking of the common band. reflectivity, although the necessary condition (7.4.3) is still satisfied with θH,max = 30o and
At 45o , there is still sufficient overlap, but at 80o , the TM band has shifted almost to the θB = 34.61o .
end of the 0o band, resulting in an extremely narrow common band.
7.4. Omnidirectional Dielectric Mirrors 219 220 Electromagnetic Waves & Antennas – S. J. Orfanidis

Frequency Response at 450 Frequency Response at 800 TE and TM bandwidths TE and TM bandwidths
2 2
100 100
| ΓT1 (f)|2 (percent)

| ΓT1 (f)|2 (percent)

80 80 1.5 1.5

omnidirectional band

F1, F2

F1, F2
60 60
1 1

40 40

0.5 0.5
TM TM
20 20 TM band TM band
TE fc0 fc TE fc TE band TE band
0o 0o
0 0 0 0
0 1 2 0 1 2 0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
f/f0 f/f0 θa (degrees) θa (degrees)

Fig. 7.4.4 TM and TE reflectances for nH = 2, nL = 1.38. Fig. 7.4.6 TM/TE bandgaps versus angle for nH = 3, nL = 1.38 and nH = 2, nL = 1.38.

On the other hand, if we were to increase nH , all the bands will widen, and so will the by F2 (0o ), and F1,TM (90o ). For the right graph, the bandedge F1,TM (θa ) increases beyond
final common band, resulting in an omnidirectional mirror of wider bandwidth. Fig. 7.4.5 F2 (0o ) for angles θa greater than about 61.8o , and therefore, there is no omnidirectional
shows the case of nH = 3, exhibiting a substantial overlap and omnidirectional behavior. band. The calculations of F1 (θa ), F2 (θa ) were done with omniband with Niter = 3. 


Frequency Response at 450 Frequency Response at 800 Example 7.4.2: In Fig. 7.4.7, we study the effect of changing the optical lengths of the bilayers
from quarter-wavelength to LH = 0.3 and LL = 0.1. The main result is to narrow the
100 100
bands. This example, also illustrates the use of the iteration (7.4.17). The approximate
solution (7.4.15) and exact solutions for the 80o bandedge frequencies are obtained from
| ΓT1 (f)|2 (percent)

| ΓT1 (f)|2 (percent)

80 80
the two MATLAB calls:

60 60
[F1,F2] = omniband(na,nH,nL,LH,LL,80,’tem’,0);
[F1,F2] = omniband(na,nH,nL,LH,LL,80,’tem’,3);
40 40

TM TM with results [F1 , F2 ]= [1.0933, 1.3891] and [F1 , F2 ]= [1.1315, 1.3266], respectively.
20 TE 20 TE
fc0 fc fc
0o 0o Three iterations produce an excellent approximation to the exact solution. 

0 0
0 1 2 0 1 2
f/f0 f/f0
Frequency Response at 450 Frequency Response at 800

Fig. 7.4.5 TM and TE reflectances for nH = 3, nL = 1.38. 100 100

| ΓT1 (f)|2 (percent)

| ΓT1 (f)|2 (percent)

The minimum band (7.4.18) was [F1 , F2 ]= [1.0465, 1.2412] corresponding to the wave- 80 80

length bandedges λ1 = λ0 /F2 = 402.84 nm and λ2 = λ0 /F1 = 477.79 nm with a width of

∆λ = λ2 − λ1 = 74.95 nm, a substantial difference from that of Fig. 7.4.2. The bandedges 60 60

were computed with Nit = 0 in Eq. (7.4.17); with Nit = 3, we obtain the more accurate
40 40
values: [F1 , F2 ]= [1.0505, 1.2412].
To illustrate the dependence of the TE and TM bandwidths on the incident angle θa , we TM TM
20 TE 20 TE
fc0 fc fc
have calculated and plotted the normalized bandedge frequencies F1 (θa ), F2 (θa ) for the 0o 0o
range of angles 0 ≤ θa ≤ 90o for both polarizations. The left graph of Fig. 7.4.6 shows the 0 0
0 1 2 0 1 2
case nH = 3, nL = 1.38, and the right graph, the case nH = 2, nL = 1.38. f/f0 f/f0
We note that the TE band widens with increasing angle, whereas the TM band narrows. At
the same time, the band centers move toward higher frequencies. In the left graph, there Fig. 7.4.7 Unequal length layers LH = 0.30, LL = 0.15.
is a common band shared by both polarizations and all angles, that is, the band defined
7.4. Omnidirectional Dielectric Mirrors 221 222 Electromagnetic Waves & Antennas – S. J. Orfanidis

Example 7.4.3: Here, we revisit Example 5.3.3, whose parameters correspond to the recently la0 = 12.5; la = linspace(5,25,401);
constructed omnidirectional infrared mirror [322]. Fig. 7.4.8 shows the reflectances as na = 1; nb = 1.48; nH = 4.6; nL = 1.6;
functions of wavelength and frequency at θa = 45o and 80o for both TE and TM polar- lH = 0.8; lL = 1.65; LH = nH*lH/la0; LL = nL*lL/la0;
izations. At both angles of incidence there is a wide overlap, essentially over the desired
th = 45;
10–15 µm band. N = 4;
The structure consisted of nine alternating layers of Tellurium (nH = 4.6) and Polystyrene n = [na, nH, repmat([nL,nH], 1, N), nb];
(nL = 1.6) on a NaCl substrate (nb = 1.48.) The physical lengths were lH = 0.8 and lL = 1.6 L = [LH, repmat([LL,LH], 1, N)];
Ge = 100*abs(multidiel(n,L,la/la0, th, ’te’)).^2;
µm. The normalizing wavelength was λ0 = 12.5 µm. The optical thicknesses in units of
Gm = 100*abs(multidiel(n,L,la/la0, th, ’tm’)).^2;
λ0 were LH = 0.2944 and LL = 0.2112. G0 = 100*abs(multidiel(n,L,la/la0)).^2;

Reflectance at 450 Reflectance at 800 plot(la,Gm, la,Ge, la,G0);

100 100 Ni = 5;
[F10,F20] = omniband(na,nH,nL,LH,LL, 0, ’te’, Ni); band at 0o
| ΓT1 (λ)|2 (percent)

| ΓT1 (λ)|2 (percent)

80 80 [F1e,F2e] = omniband(na,nH,nL,LH,LL, th,’te’, Ni); TE band
[F1m,F2m] = omniband(na,nH,nL,LH,LL, th,’tm’, Ni); TM band
60 60 [F1,F2] = omniband(na,nH,nL,LH,LL, th,’tem’,Ni); Eq. (7.4.19)
[F1,F2] = omniband(na,nH,nL,LH,LL, 90,’tem’,Ni); Eq. (7.4.18)

40 40
Finally, Fig. 7.4.9 shows the same example with the number of bilayers doubled to N = 8.
20
TM
20
TM The mirror bands are flatter and sharper, but the widths are the same. 

λc λc0 TE λc λc0 TE
0o 0o
0 0
5 10 15 20 25 5 10 15 20 25 Reflectance at 450 Reflectance at 800
λ (µm) λ (µm)
Frequency Response at 450 Frequency Response at 800 100 100

| ΓT1 (λ)|2 (percent)

| ΓT1 (λ)|2 (percent)

100 100 80 80
| ΓT1 (f)|2 (percent)

| ΓT1 (f)|2 (percent)

80 80 60 60

60 60 40 40

40 40 TM TM
20 20
λc λc0 TE λc λc0 TE
0o 0o
TM TM
20 TE 20 TE 0 0
fc0 fc fc0 fc 5 10 15 20 25 5 10 15 20 25
0o 0o
λ (µm) λ (µm)
0 0
0 1 2 0 1 2
f/f0 f/f0
Fig. 7.4.9 Omnidirectional mirror with N = 8.

Fig. 7.4.8 Nine-layer Te/PS omnidirectional mirror over the infrared.

Example 7.4.4: The last example has parameters corresponding to the recently constructed
The bandedges at 0o were [F1 , F2 ]= [0.6764, 1.2875] with center frequency Fc0 = 0.9819, omnidirectional reflector over the visible range [323]. The refractive indices were na = 1,
corresponding to wavelength λc0 = λ0 /Fc0 = 12.73 µm. Similarly, at 45o , the band centers nH = 2.6 (ZnSe), nL = 1.34 (Na3 AlF6 cryolite), and nb = 1.5 (glass substrate.) The layer
for TE and TM polarizations were Fc,TE = 1.0272 and Fc,TM = 1.0313, resulting in the lengths were lH = lL = 90 nm. There were N = 9 bilayers or 2N + 1 = 19 layers, starting
wavelengths λc,TE = 12.17 and λc,TM = 12.12 µm (shown on the graphs are the TE centers and ending with nH .
only.) With these values, the maximum angle of refraction is θH,max = 22.27o and is less than the
The final bandedges of the common reflecting band computed from Eq. (7.4.18) were Brewster angle θB = 27.27o .
[F1 , F2 ]= [0.8207, 1.2875], resulting in the wavelength bandedges λ1 = λ0 /F2 = 9.71 The normalizing wavelength was taken to be λ0 = 620 nm. Then, the corresponding optical
and λ2 = λ0 /F1 = 14.95 µm, with a width of ∆λ = λ2 − λ1 = 5.24 µm and band center lengths were LL = nL lL /λ0 = 0.1945 and LH = nH lH /λ0 = 0.3774. The overall minimum
(λ1 + λ2 )/2 = 12.33 µm (the approximation (7.4.15) gives 5.67 and 12.4 µm.) The graphs omnidirectional band is [λ1 , λ2 ]= [605.42, 646.88] nm. It was computed by the MATLAB
were generated by the following MATLAB code: call to omniband with Ni = 5 iterations:
 7.5. Polarizing Beam Splitters 223 224 Electromagnetic Waves & Antennas – S. J. Orfanidis

[F1,F2] = omniband(na,nH,nL,LH,LL,90,’tem’,Ni); component will not be reflected at the bilayer interfaces and will transmit through. The
la1 = la0/F2; la2 = la0/F1; design condition is θH = θB , or sin θH = sin θB , which gives:

(The values of λ1 , λ2 do not depend on the choice of λ0 .) Fig. 7.4.10 shows the reflectance nH nL
at 45o and 80o . The upper panel of graphs has N = 9 bilayers as in [323]. The lower panel na sin θa = nH sin θH = nH sin θB =  (7.5.1)
has N = 18 bilayers or 38 layers, and has more well-defined band gaps. The two arrows in n2H + n2L
the figures correspond to the values of λ1 , λ2 of the minimum omnidirectional band.  
This condition can be solved either for the angle θa or for the index na of the incident
medium:
Reflectance at 450 Reflectance at 800
n n nH nL
100 100 sin θa = H L or, na =  (7.5.2)
na n2H + n2L sin θa n2H + n2L
| ΓT1 (λ)|2 (percent)

| ΓT1 (λ)|2 (percent)

80 80

In either case, the feasibility of this approach requires the opposite of the condition
60 60
(7.4.3), that is,
TM TM
TE TE
40 40 nH nL
0o 0o
na >  (7.5.3)
n2H + n2L
20 20

If the angle θa is set equal to the convenient value of 45o , then, condition Eq. (7.5.2)
0 0
400 500 600 700 800 900 400 500 600 700 800 900 fixes the value of the refractive index na to be given by:
λ (nm) λ (nm)
Reflectance at 450 Reflectance at 800 √
2nH nL
na =  (7.5.4)
100 100
n2H + n2L
| ΓT1 (λ)|2 (percent)

| ΓT1 (λ)|2 (percent)

80 80
Fig. 7.5.1 depicts such a multilayer structure sandwiched between two glass prisms
with 45o angles. The thin films are deposited along the hypotenuse of each prism and
60 60
TM TM the prisms are then cemented together. The incident, reflected, and transmitted beams
TE TE are perpendicular to the prism sides.
40 0o 40 0o

20 20

0 0
400 500 600 700 800 900 400 500 600 700 800 900
λ (nm) λ (nm)

Fig. 7.4.10 Omnidirectional mirror over visible band.

7.5 Polarizing Beam Splitters

The objective of an omnidirectional mirror is to achieve high reflectivity for both polar-
izations. However, in polarizers, we are interested in separating the TE and TM polariza-
tions. This can be accomplished with a periodic bilayer structure of the type shown in Fig. 7.5.1 Polarizing beam splitter.

Fig. 7.4.1, which is highly reflecting only for TE and highly transmitting for TM polariza-
tions. This is the principle of the so-called MacNeille polarizers [189,193,196,215,218,233– Not many combinations of available materials satisfy condition (7.5.4). One possible
239]. solution is Banning’s [196] with nH = 2.3 (zinc sulfide), nL = 1.25 (cryolite), and na =
If the angle of incidence θa is chosen such that the angle of refraction in the first 1.5532. Another solution is given in Clapham, et al, [218], with nH = 2.04 (zirconium
high-index layer is equal to the Brewster angle of the high-low interface, then the TM oxide), nL = 1.385 (magnesium fluoride), and na = 1.6205 (a form of dense flint glass.)
7.5. Polarizing Beam Splitters 225 226 Electromagnetic Waves & Antennas – S. J. Orfanidis

Fig. 7.5.2 shows the TE and TM reflectances of the case nH = 2.3 and nL = 1.25. The nH = 2.3; nL = 1.25;
incident and output media had na = nb = 1.5532. The maximum reflectivity for the TE LH = 0.25; LL = 0.25;

component is 99.99%, while that of the TM component is 3% (note the different vertical
na = nH*nL/sqrt(nH^2+nL^2)/sin(pi/4); nb=na;
scales in the two graphs.)
The number of bilayers was N = 5 and the center frequency of the TE band was [f1e,f2e] = omniband(na,nH,nL,LH,LL,th,’te’,5);
chosen to correspond to a wavelength of λc = 500 nm. To achieve this, the normal- lac = 500;
la0 = lac*(f1e+f2e)/2; because λc = λ0 /Fc
izing wavelength was required to be λ0 = 718.38 nm. The layer lengths were quarter-
wavelengths at λ0 . The TE bandwidth calculated with omniband is also shown. la = linspace(300,800,301);
The Brewster angles inside the high- and low-index layers are θH = 28.52o and
θL = 61.48o . As expected, they satisfy θH + θL = 90o . N = 5;
n = [na, nH, repmat([nL,nH], 1, N), nb];
L = [LH, repmat([LL,LH], 1, N)];
TE Reflectance TM Reflectance
4
Ge = 100*abs(multidiel(n,L,la/la0, th, ’te’)).^2;
Gm = 100*abs(multidiel(n,L,la/la0, th, ’tm’)).^2;
100

plot(la,Ge);

| ΓTM (λ)|2 (percent)

| ΓTE (λ)|2 (percent)

80 3

60
2 7.6 Reflection and Refraction in Birefringent Media
40
Uniform plane wave propagation in biaxial media was discussed in Sec. 3.6. We found
1
20 that there is an effective refractive index N such that k = Nk0 = Nω/c0 . The index N,
λc
given by Eq. (3.6.8), depends on the polarization of the fields and the direction of the
0
300 400 500 600 700 800
0
300 400 500 600 700 800
wave vector. The expressions for the TE and TM fields were given in Eqs. (3.6.18) and
λ (nm) λ (nm) (3.6.27).
Here, we discuss how such fields get reflected and refracted at planar interfaces
Fig. 7.5.2 Polarizer with nH = 2.3 and nL = 1.25. between biaxial media. Further discussion can be found in [187] and [246–264].
Fig. 6.1.1 depicts the TM and TE cases, with the understanding that the left and
Fig. 7.5.3 shows the second case having nH = 2.04, nL = 1.385, na = nb = 1.6205. right biaxial media are described by the triplets of principal indices n = [n1 , n2 , n3 ]
The normalizing wavelength was λ0 = 716.27 nm in order to give λc = 500 nm. This and n = [n1 , n2 , n3 ], and that the E-fields are not perpendicular to the corresponding
case achieves a maximum TE reflectivity of 99.89% and TM reflectivity of only 0.53%. wave vectors in the TM case. The principal indices are aligned along the xyz axes, the
The typical MATLAB code generating these examples was: xy-plane is the interface plane, and the xz-plane is the plane of incidence.
The boundary conditions require the matching of the electric field components that
TE Reflectance TM Reflectance
4
are tangential to the interface, that is, the components Ex in the TM case or Ey in TE.
100 It proves convenient, therefore, to re-express Eq. (3.6.27) directly in terms of the Ex
component and Eq. (3.6.18) in terms of Ey .
| ΓTM (λ)|2 (percent)
| ΓTE (λ)|2 (percent)

80 3
For the TM case, we write E = x̂ Ex + ẑ Ez = Ex (x̂ − ẑ tan θ̄), for the electric field of
the left-incident field, where we used Ez = −Ex tan θ̄. Similarly, for the magnetic field
60
2 we have from Eq. (3.6.26):

40 N N Ez
H= ŷ(Ex cos θ − Ez sin θ)= ŷ Ex cos θ 1 − tan θ
1 η0 η0 Ex
20
λc N n2 N n23 cos2 θ + n21 sin2 θ
= ŷ Ex cos θ 1 + 12 tan2 θ = ŷ Ex cos θ
0
300 400 500 600 700 800
0
300 400 500 600 700 800
η0 n3 η0 n23 cos2 θ
λ (nm) λ (nm)
N n23 n21 Ex n21
= ŷ Ex cos θ = ŷ
Fig. 7.5.3 Polarizer with nH = 2.04 and nL = 1.385. η0 N2 n23 cos2 θ η0 N cos θ

where we replaced Ez /Ex = − tan θ̄ = −(n21 /n23 )tan θ and used Eq. (3.7.10). Thus,
7.6. Reflection and Refraction in Birefringent Media 227 228 Electromagnetic Waves & Antennas – S. J. Orfanidis

Assuming that the interface is at z = 0, the equality of the total tangential electric
n2 fields (Ex component for TM and Ey for TE), implies as in Sec. 6.1 that the propagation
E(r) = Ex x̂ − ẑ 12 tan θ e−j k·r
n3 phase factors must match at all values of x:
(TM) (7.6.1)
Ex n21 Ex 
e−jkx+ x = e−jkx− x = e−jkx+ x = e−jkx− x


H(r) = ŷ e−j k·r = ŷ e−j k·r

η0 N cos θ ηTM
which requires that kx+ = kx− = kx+ = kx− , or, because kx = k sin θ = Nk0 sin θ:
Similarly, we may rewrite the TE case of Eq. (3.6.18) in the form:

N sin θ+ = N sin θ− = N sin θ+ = N sin θ−

−j k·r
E(r) = Ey ŷ e
This implies Snell’s law of reflection, that is, θ+ = θ− ≡ θ and θ+ = θ− ≡ θ , and
Ey Ey (TE) (7.6.2) Snell’s law of refraction,
H(r) = n2 cos θ(−x̂ + ẑ tan θ)e−j k·r = (−x̂ + ẑ tan θ)e−j k·r
η0 ηTE
N sin θ = N sin θ (Snell’s law for birefringent media) (7.6.8)
The propagation phase factors are:
Thus, Snell’s law is essentially the same as in the isotropic case, provided one uses
the effective refractive index N. Because N depends on the polarization, there will be
e−j k·r = e−jk0 xN sin θ−jk0 zN cos θ (TM and TE propagation factors) (7.6.3) two different refraction angles† for the same angle of incidence. In particular, Eq. (7.6.8)
can be written explicitly in the two polarization cases:
Unlike the isotropic case, the phase factors are different in the TM and TE cases
because the value of N is different, as given by Eq. (3.6.8), or, n1 n3 sin θ n1 n3 sin θ
 =  (TM) (7.6.9a)
⎧
2 2
n1 sin θ + n3 cos2 θ
2
n1 sin2 θ + n32 cos2 θ
2
⎪
⎪ n1 n3
⎨  , (TM or p-polarization)
N= n21 sin2 θ + n23 cos2 θ (7.6.4) n2 sin θ = n2 sin θ (TE) (7.6.9b)
⎪
⎪
⎩ n , (TE or s-polarization)
2
Both expressions reduce to Eq. (6.1.6) in the isotropic case. The explicit solutions of
In Eqs. (7.6.1) and (7.6.2), the effective transverse impedances are defined by ηTM = Eq. (7.6.9a) for sin θ and sin θ are:
Ex /Hy and ηTE = −Ey /Hx , and are given as follows:

n1 n3 n3 sin θ
sin θ =  
ηTM = η0
N cos θ
, ηTE =
η0
(transverse impedances) (7.6.5) n12 n32 (n21 − n23 )−n21 n23 (n12 − n32 ) sin2 θ + n12 n32 n23
n21 n2 cos θ (7.6.10)
n1 n3 n3 sin θ
sin θ =  
Defining the TM and TE effective transverse refractive indices through ηTM = η0 /nTM n21 n23 (n12 − n32 )−n12 n32 (n21 − n23 ) sin2 θ + n21 n23 n32
and ηTE = η0 /nTE , we have:
The MATLAB function snell, solves Eqs. (7.6.9) for θ given the angle of incidence
θ and the polarization type. It works for any type of medium, isotropic, uniaxial, or
n21 n1 n3 biaxial. It has usage:
nTM = = 
N cos θ 2
n3 − N2 sin2 θ (transverse refractive indices) (7.6.6)
thb = snell(na,nb,tha,pol); % refraction angle from Snell’s law
nTE = n2 cos θ
The refractive index inputs na, nb may be entered as 1-, 2-, or 3-dimensional column
where we used Eq. (3.6.23) for the TM case, that is, or row vectors, for example, na = [na ] (isotropic), na = [nao , nae ] (uniaxial), or na =
[na1 , na2 , na3 ] (biaxial).

n1 Next, we discuss the propagation and matching of the transverse fields. All the
N cos θ = n23 − N2 sin2 θ (7.6.7)
n3 results of Sec. 6.3 translate verbatim to the birefringent case, provided one uses the
proper transverse refractive indices according to Eq. (7.6.6).
In the isotropic case, N = n1 = n2 = n3 = n, Eqs. (7.6.6) reduce to Eq. (6.2.13). Next,
† Hence,
we discuss the TM and TE reflection and refraction problems of Fig. 6.1.1. the name birefringent.
7.6. Reflection and Refraction in Birefringent Media 229 230 Electromagnetic Waves & Antennas – S. J. Orfanidis

In particular, the propagation equations (6.3.5)–(6.3.7) for the transverse fields, for
the transverse reflection coefficients ΓT , and for the transverse wave impedances ZT , n1 − n1
ρTM = (7.6.16)
remain unchanged. n1 + n1
The phase thickness δz for propagating along z by a distance l also remains the same
as Eq. (6.3.8), except that the index N must be used in the optical length, and therefore, Another interesting case is when both media are uniaxial and n3 = n1 and n1 = n3 ,
δz depends on the polarization: that is, the refractive index vectors are n = [n1 , n1 , n3 ] and n = [n3 , n3 , n1 ]. It is
straightforward to show in this case that ρTM = ρTE at all angles of incidence. Multilayer
2π films made from alternating such materials exhibit similar TM and TE optical properties
δz = kz l = kl cos θ = Nk0 l cos θ = lN cos θ (7.6.11) [246].
λ
The MATLAB function fresnel can evaluate Eqs. (7.6.14) at any range of incident
Using Eq. (7.6.7), we have explicitly: angles θ. The function determines internally whether the media are isotropic, uniaxial,
 or biaxial.
2π n1
δz = l n23 − N2 sin2 θ , (TM) (7.6.12a)
λ n3
2π 7.7 Brewster and Critical Angles in Birefringent Media
δz = ln2 cos θ , (TE) (7.6.12b)
λ
The maximum angle of refraction, critical angle of incidence, and Brewster angle, have
The transverse matching matrix (6.3.11) and Fresnel reflection coefficients (6.3.12)
their counterparts in birefringent media.
remain the same. Explicitly, we have in the TM and TE cases:
It is straightforward to verify that θ is an increasing function of θ in Eq. (7.6.9). The
maximum angle of refraction θc is obtained by setting θ = 90o in Eq. (7.6.9).
n21 n2
−  1  For the TE case, we obtain sin θc = n2 /n2 . As in the isotropic case of Eq. (6.5.2), this
nTM − nTM
ρTM = = N cos θ N cos θ
requires that n2 < n2 , that is, the incident medium is less dense than the transmitted
nTM + nTM n21 n2
+  1  (7.6.13) medium, with respect to the index n2 . For the TM case, we obtain from Eq. (7.6.9a):
N cos θ N cos θ
nTE − nTE n2 cos θ − n2 cos θ
ρTE = =
nTE + nTE n2 cos θ + n2 cos θ n3 n3
sin θc =  (maximum TM refraction angle) (7.7.1)
n23 n32 + n12 (n32 − n23 )
Using Eq. (7.6.6) and the TM and TE Snell’s laws, Eqs. (7.6.9), we may rewrite the
reflection coefficients in terms of the angle θ only:
This requires that n3 < n3 . On the other hand, if n3 > n3 , we obtain the critical
  angle of incidence θc that corresponds to θ = 90o in Eq. (7.6.10):
n1 n3 n32 − N2 sin2 θ − n1 n3 n23 − N2 sin2 θ
ρTM =  
n1 n3 n32 − N2 sin2 θ + n1 n3 n23 − N2 sin2 θ n3 n3
sin θc =  (critical TM angle) (7.7.2)
 (7.6.14)
n23 n32 + n21 (n23 − n32 )
n2 cos θ − n22 − n22 sin2 θ
ρTE = 
n2 cos θ + n22 − n22 sin2 θ whereas for the TE case, we have sin θc = n2 /n2 , which requires n2 > n2 .
In the isotropic case, a Brewster angle always exists at which the TM reflection coeffi-
The quantity N2 sin2 θ can be expressed directly in terms of θ and the refractive cient vanishes, ρTM = 0. In the birefringent case, the Brewster angle does not necessarily
indices of the incident medium. Using Eq. (7.6.4), we have: exist, as is the case of Eq. (7.6.16), and it can also have the value zero, or even be imagi-
nary.
n21 n23 sin2 θ The Brewster angle condition ρTM = 0 is equivalent to the equality of the transverse
N2 sin2 θ = (7.6.15) refractive indices nTM = nTM . Using Eq. (7.6.6), we obtain:
n21 sin2 θ + n23 cos2 θ

The TE reflection coefficient behaves like the TE isotropic case. The TM coefficient n1 n3 n1 n3
nTM = nTM ⇒  =  (7.7.3)
exhibits a much more complicated behavior. If n1 = n1 but n3 = n3 , it behaves like the n23 − N2 sin2 θ n32 − N2 sin2 θ
TM isotropic case. If n3 = n3 but n1 = n1 , the square-root factors cancel and it becomes
independent of θ: where N2 sin2 θ is given by Eq. (7.6.15). Solving for θ, we obtain the expression for the
Brewster angle from the left medium:
7.7. Brewster and Critical Angles in Birefringent Media 231 232 Electromagnetic Waves & Antennas – S. J. Orfanidis

In case (c), the Brewster angle does not exist because n3 = n3 , and in fact, the TM reflection


 2 coefficient is independent of the incident angle as in Eq. (7.6.16). The corresponding critical
n3 n3 
 2
 n1 − n1
tan θB = (Brewster angle) (7.7.4) angles of reflection are: θc,TE = 56.4o and θc,TM = 90o .
n12
n3 − n32
2
Finally, in case (d), because n2 > n2 but n3 < n3 , the Brewster angle will be imaginary,
and there will be a TE critical angle of reflection and a TM maximum angle of refraction:
Working instead with N sin θ = N sin θ, we obtain the Brewster angle from the
θc,TE = 60.1o and θc,TM = 74.1o .
right medium, interchanging the roles of the primed and unprimed quantities:
Fig. 7.7.1 shows the TM and TE reflection coefficients |ρTM (θ)| of Eq. (7.6.14) versus θ in

the range 0 ≤ θ ≤ 90o .
 2
 n3 n3  2
 n1 − n1
tan θB = 2 (Brewster angle) (7.7.5)
n1 n3 − n32
2
TM Reflection Coefficients TE Reflection Coefficients

1 1
Eqs. (7.7.4) and (7.7.5) reduce to Eqs. (6.6.2) and (6.6.3) in the isotropic case. It is
(a)
evident from Eq. (7.7.4) that θB is a real angle only if the quantity under the square (b)
(b)
0.8 0.8 (c)
root is non-negative, that is, only if n1 > n1 and n3 > n3 , or if n1 < n1 and n3 < n3 . (c)
(d)
(d)

Otherwise, θB is imaginary. In the special case, n1 = n1 but n3 = n3 , the Brewster

|ρTM(θ)|

|ρTE(θ)|
0.6 0.6
angle vanishes. If n3 = n3 , the Brewster angle does not exist, since then ρTM is given by
Eq. (7.6.16) and cannot vanish. 0.4 0.4

The MATLAB function brewster computes the Brewster angle θB , as well as the
critical angles θc and θc . For birefringent media the critical angles depend on the po- 0.2 0.2

larization. Its usage is as follows:

0 0
0 15 30 45 60 75 90 0 15 30 45 60 75 90
[thB,thc] = brewster(na,nb) % isotropic case θ θ (degrees) θ θ (degrees)
[thB,thcTE,thcTM] = brewster(na,nb) % birefringent case
Fig. 7.7.1 TM and TE birefringent Fresnel reflection coefficients versus incident angle.
In multilayer systems, it is convenient to know if the Brewster angle of an internal
interface is accessible from the incident medium. Using Snell’s law we have in this case The TE coefficient in case (a) is not plotted because it is identically zero. In order to expand
Na sin θa = N sin θ, where θa is the incident angle and Na the effective index of the the vertical scales, Fig. 7.7.2 shows the TM reflectances normalized by their values at θ =
incident medium. It is simpler, then, to solve Eq. (7.7.3) directly for θa : 0o , that is, it plots the quantities |ρTM (θ)/ρTM (0o )|2 . Because in case (a) ρTM (0o )= 0, we
have plotted instead the scaled-up quantity |100ρTM (θ)|2 .

n23 n32 (n21 − n12 )

Na2 sin θ2a = N2 sin2 θB = (7.7.6) Normalized TM Reflectance
n21 n23 − n32 n12 2

(a)
Example 7.7.1: To illustrate the variety of possible Brewster angle values, we consider the fol- 1.5
(b)
(c)
lowing birefringent cases: (d)

Reflectance
(a) n = [1.63, 1.63, 1.5], n = [1.63, 1.63, 1.63] 1

(b) n = [1.54, 1.54, 1.63], n = [ 1 . 5 , 1 .5 , 1 .5 ]

(c) n = [1.8, 1.8, 1.5], n = [ 1 . 5 , 1 .5 , 1 .5 ]
0.5
(d) n = [1.8, 1.8, 1.5], n = [1.56, 1.56, 1.56]

These cases were discussed in [246]. The corresponding materials are: (a) birefringent 0
0 15 30 45 60 75 90
polyester and isotropic polyester, (b) syndiotactic polystyrene and polymethylmethacrylate θ (degrees)
(PMMA), (c) birefringent polyester and PMMA, and (d) birefringent polyester and isotropic
polyester. Fig. 7.7.2 TM reflectances normalized at normal incidence.
Because n1 = n1 in case (a), the Brewster angle will be zero, θB = 0o . In case (b), we
calculate θB = 29.4o . Because n2 > n2 and n3 > n3 , there will be both TE and TM critical The typical MATLAB code used to compute the critical angles and generate these graphs
angles of reflection: θc,TE = 76.9o and θc,TM = 68.1o . was:
7.8. Multilayer Birefringent Structures 233 234 Electromagnetic Waves & Antennas – S. J. Orfanidis

th = linspace(0,90,361); % θ at 1/4o intervals ⎧

⎪ 
⎪
⎪  2 2
na = [1.63,1.63,1.5]; nb = [1.63,1.63,1.63]; % note the variety of ⎪
⎪
2π
l n 1 − Na sin θa , (TM)
⎪
⎪ i i 1 2
[rte1,rtm1] = fresnel(na,nb,th); % equivalent ways of 2π ⎨ λ ni3
[thb1,thcTE1,thcTM1] = brewster(na,nb); % entering na and nb δi = li Ni cos θi =
(7.8.2)
λ ⎪
⎪ 
⎪
⎪  2 2
⎪
⎪ 2π li ni2 1 − Na sin θa ,
na = [1.54,1.63]; ⎪
⎩ λ 2 (TE)
nb = [1.5, 1.5]; % FRESNEL and BREWSTER ni2
[rte2,rtm2] = fresnel(na,nb,th); % internally extend
where we used Eq. (7.6.7) and Snell’s law to write in the TM and TE cases:
[thb2,thcTE2,thcTM2] = brewster(na,nb); % na,nb into 3-d arrays
⎧


⎪   
na = [1.8, 1.5]; % same as na=[1.8,1.8,1.5] ⎪
⎪ n  N 2
sin 2
θ  N2 sin2 θa
⎪ n2i3 − Ni2 sin2 θi = ni1 1 − i 2 = ni1 1 − a 2
i1 i
⎪
⎪
nb = 1.5; % and nb=[1.5,1.5,1.5] ⎪
⎨ ni3 ni3 ni3
[rte3,rtm3] = fresnel(na,nb,th);
Ni cos θi =

[thb3,thcTE3,thcTM3] = brewster(na,nb); % in this case, θB = [] ⎪
⎪  
⎪
⎪  Na2 sin2 θa
⎪
⎪ 
na = [1.8,1.5];
⎪
⎩ ni2 cos θi = ni2 1 − sin θi = ni2 1 −
2
n2i2
nb = 1.56;
[rte4,rtm4] = fresnel(na,nb,th);
To use a unified notation for the TM and TE cases, we define the layer optical lengths
[thb4,thcTE4,thcTM4] = brewster(na,nb);
at normal-incidence, normalized by a fixed free-space wavelength λ0 :
plot(th, abs([rtm1; rtm2; rtm3; rtm4])); ⎧
⎪
⎪ li ni1
⎪
⎨ , (TM)
λ0
We note four striking properties of the birefringent cases that have no counterparts Li = , i = 1, 2, . . . , M (7.8.3)
⎪
⎪ li ni2
for isotropic materials: (i) The Brewster angle can be zero, (ii) the Brewster angle may not ⎪
⎩ , (TE)
λ0
exist, (iii) the Brewster angle may be imaginary with the TE and TM reflection coefficients
both increasing monotonically with the incident angle, and (iv) there may be total internal We define also the cosine coefficients ci , which represent cos θi in the TE birefringent
reflection in one polarization but not in the other. case and in the TM isotropic case:
⎧

⎪ 
⎪
⎪  2 2
⎪
⎪ 1 − Na sin θa , (TM)
7.8 Multilayer Birefringent Structures ⎪
⎪ 2
⎨ ni3
ci =
, i = 1, 2, . . . , M (7.8.4)
⎪
⎪ 
With some redefinitions, all the results of Sec. 7.1 on multilayer dielectric structures ⎪
⎪  2 2
⎪
⎪ 1 − Na sin θa ,
translate essentially unchanged to the birefringent case. ⎪
⎩ 2 (TE)
ni2
We assume the same M-layer configuration shown in Fig. 7.1.1, where now each layer
is a biaxial material. The orthogonal optic axes of all the layers are assumed to be aligned At normal incidence the cosine factors are unity, ci = 1. With these definitions,
with the xyz film axes. The xz-plane is the plane of incidence, the layer interfaces are Eq. (7.8.2) can be written compactly in the form:
parallel to the xy-plane, and the layers are arranged along the z-axis.
The ith layer is described by the triplet of refractive indices ni = [ni1 , ni2 , ni3 ], λ0 f
δi = 2π Li ci = 2π Li ci , i = 1, 2, . . . , M (7.8.5)
i = 1, 2, . . . , M. The incident and exit media a, b may also be birefringent with na = λ f0
[na1 , na2 , na3 ] and nb = [nb1 , nb2 , nb3 ], although in our examples, we will assume that
where λ is the operating free-space wavelength and f = c0 /λ, f0 = c0 /λ0 . This is
they are isotropic.
the birefringent version of Eq. (7.1.10). A typical design might use quarter-wave layers,
The reflection/refraction angles in each layer depend on the assumed polarization
Li = 1/4, at λ0 and at normal incidence.
and are related to each other by the birefringent version of Snell’s law, Eq. (7.6.8):
The reflection coefficients ρTi at the interfaces are given by Eq. (7.1.3), but now the
transverse refractive indices are defined by the birefringent version of Eq. (7.1.4):
Na sin θa = Ni sin θi = Nb sin θb , i = 1, 2 . . . , M (7.8.1)

where Na , Ni , Nb are the effective refractive indices given by Eq. (7.6.4). The phase ⎧
⎪
⎪ n2i1 ni1 ni3
thickness of the ith layer depends on the polarization: ⎪
⎨ =  , (TM)
Ni cos θi ni3 − Na2 sin2 θa
2
nTi = , i = a, 1, 2, . . . , M, b (7.8.6)
⎪
⎪ 
⎪
⎩
ni2 cos θi = n2i2 − Na2 sin2 θa , (TE)
7.9. Giant Birefringent Optics 235 236 Electromagnetic Waves & Antennas – S. J. Orfanidis

With the above redefinitions, the propagation and matching equations (7.1.5)–(7.1.9) Example 7.9.1: We consider a GBO mirror consisting of 50-bilayers of high and low index
remain unchanged. The MATLAB function multidiel can also be used in the birefrin- quarter-wave layers with refractive indices nH = [1.8, 1.8, 1.5], nL = [1.5, 1.5, 1.5] (bire-
gent case to compute the frequency reflection response of a multilayer structure. Its fringent polyester and isotropic PMMA.) The surrounding media are air, na = nb = 1.
usage is still: The layers are quarter wavelength at the normalization wavelength λ0 = 700 nm at normal
incidence, so that for both polarizations we take LH = LL = 1/4.
[Gamma1,Z1] = multidiel(n,L,lambda,theta,pol); % birefringent multilayer structure
Because the high/low index layers are matched along the z-direction, nH3 = nL3 , the TM
where the input n is a 1×(M + 2) vector of refractive indices in the isotropic case, or a reflection coefficient at the high/low interface will be constant, independent of the incident
3×(M + 2) matrix, where each column represents the triplet of birefringent indices of angle θa , as in Eq. (7.6.16). However, some dependence on θa is introduced through the
each medium. For uniaxial materials, n may be entered as a 2×(M + 2) matrix. cosine factors cH , cL of Eq. (7.9.2).
The left graph of Fig. 7.9.1 shows the reflectance |ΓT (λ)|2 as a function of λ for an an-
gle of incidence θa = 60o . The TM and TE bandedge wavelengths were calculated from
7.9 Giant Birefringent Optics omniband2 to be: [λ1 , λ2 ]= [540.24, 606.71] and [λ1 , λ2 ]= [548.55, 644.37] nm.

The results of Sec. 7.4 can be applied almost verbatim to the birefringent case. In Reflectance at 0o and 60o 25% thickness gradient
Fig. 7.4.1, we assume that the high and low alternating layers are birefringent, described
100 100
by the triplet indices nH = [nH1 , nH2 , nH3 ] and nL = [nL1 , nL2 , nL3 ]. The entry and exit TM TM
media may also be assumed to be birefringent. Then, Snell’s laws give: TE TE

| ΓT (λ)|2 (percent)

| ΓT (λ)|2 (percent)
80 0o 80 0o

Na sin θa = NH sin θH = NL sin θL = Nb sin θb (7.9.1) 60 60

The phase thicknesses δH and δL within the high and low index layers are:
40 40

f f
δH = 2π LH cH , δL = 2π LL cL (7.9.2) 20 20
f0 f0
0 0
where LH , cH and LL , cL are defined by Eqs. (7.8.3) and (7.8.4) for i = H, L. The effective 400 500 600 700 800 900 1000 400 500 600 700 800 900 1000
λ (nm) λ (nm)
transverse refractive indices within the high and low index layers are given by Eq. (7.8.6),
again with i = H, L.
Fig. 7.9.1 Reflectance of birefringent mirror.
The alternating reflection coefficient ρT between the high/low interfaces is given by
Eq. (7.6.14), with the quantity N2 sin2 θ replaced by Na2 sin2 θa by Snell’s law:
The typical MATLAB code used to generate the left graph and the bandedge wavelengths
  was as follows:
nH1 nH3 n2L3 − Na2 sin2 θa − nL1 nL3 nH2 2
3 − Na sin θa
2
ρTM =  
LH = 0.25; LL = 0.25;
nH1 nH3 n2L3 − Na2 sin2 θa + nL1 nL3 nH2 2
3 − Na sin θa
2

  (7.9.3) na = [1; 1; 1];

n2H2 − Na2 sin2 θa − n2L2 − Na2 sin2 θa nH = [1.8; 1.8; 1.5];
ρTE =   nL = [1.5; 1.5; 1.5];
n2H2 − Na2 sin2 θa + n2L2 − Na2 sin2 θa nb = [1; 1; 1];

The multilayer structure will exhibit reflection bands whose bandedges can be cal- la0 = 700;
culated from Eqs. (7.4.7)–(7.4.17), with the redefinition L± = LH cH ± LL cL . The MATLAB la = linspace(400,1000,601);
function omniband2 calculates the bandedges. It has usage:
th = 60; % angle of incidence
[F1,F2] = omniband2(na,nH,nL,LH,LL,th,pol,N);
N = 50; % number of bilayers
where pol is one of the strings ’te’ or ’tm’ for TE or TM polarization, and na, nH, nL n = [na, repmat([nH,nL], 1, N), nb]; % 3×(2N + 2) matrix
L = [repmat([LH,LL], 1, N)];
are 1-d, 2-d, or 3-d row or column vectors of birefringent refractive indices.
Next, we discuss some mirror design examples from [246] that illustrate some prop- Ge = 100*abs(multidiel(n, L, la/la0, th, ’te’)).^2;
erties that are specific to birefringent media. The resulting optical effects in such mirror Gm = 100*abs(multidiel(n, L, la/la0, th, ’tm’)).^2;
structures are referred to as giant birefringent optics (GBO) in [246,823]. G0 = 100*abs(multidiel(n, L, la/la0)).^2;
7.9. Giant Birefringent Optics 237 238 Electromagnetic Waves & Antennas – S. J. Orfanidis

plot(la,Gm,’-’, la,Ge,’--’, la,G0,’:’);

1 1
L+,TM = LH,TM cH,TM + LL,TM cL,TM == (cH,TM + cL,TM )= (cL,TE + cH,TE )= L+,TE
[F1,F2]=omniband2(na,nH,nL,LH,LL,th,’tm’,3); 4 4
la1 = la0/F2; la2 = la0/F1; % TM bandedge wavelengths 1 1
L−,TM = LH,TM cH,TM − LL,TM cL,TM == (cH,TM − cL,TM )= (cL,TE − cH,TE )= −L+,TE
4 4
The right graph shows the reflectance with a 25% thickness gradient (the layer thicknesses Because the computational algorithm (7.4.17) for the bandwidth does not depend on the
LH , LL decrease linearly from quarter-wavelength to 25% less than that at the end.) This sign of L− , it follows that Eq. (7.4.17) will have the same solution for the TM and TE cases.
can be implemented in MATLAB by defining the thickness vector L by: The typical MATLAB code for this example was:

L = [repmat([LH,LL], 1, N)]; LH = 0.25; LL = 0.25;

L = L .* (1 - linspace(0, 0.25, 2*N)); % 25% thickness gradient
na = [1.4; 1.4; 1.4];
nb = [1.4; 1.4; 1.4];
The thickness gradient increases the effective bandwidth of the reflecting bands [245].
nH = [1.8; 1.8; 1.5];
However, the bandwidth calculation can no longer be done with omniband2. The band nL = [1.5; 1.5; 1.8];
centers can be shifted to higher wavelengths by choosing λ0 higher. The reflecting bands
can be made flatter by increasing the number of bilayers. 
 la0 = 700;
la = linspace(400,1000,601);
Example 7.9.2: In this example, we design a 30-bilayer GBO mirror with nH = [1.8, 1.8, 1.5]
tha = 45;
and nL = [1.5, 1.5, 1.8], so that nH1 = nH2 = nL3 and nH3 = nL1 = nL2 . As we discussed
in Sec. 7.6, it follows from Eq. (7.6.14) that ρTM = ρTE for all angles of incidence.
N = 30;
As in Ref. [246], the media a, b are taken to be isotropic with na = nb = 1.4. The n = [na, repmat([nH,nL], 1, N), nb];
normalization wavelength at which the high/low index layer are quarter-wavelength is L = [repmat([LH,LL], 1, N)];
λ0 = 700 nm.
Ge = 100*abs(multidiel(n, L, la/la0, tha, ’te’)).^2;
The left graph of Fig. 7.9.2 shows the reflectance for a 45o angle of incidence. Because Gm = 100*abs(multidiel(n, L, la/la0, tha, ’tm’)).^2;
ρTM = ρTE , the reflection bands for the TM and TE cases are essentially the same. G0 = 100*abs(multidiel(n, L, la/la0)).^2;

plot(la,Gm,’-’, la,Ge,’--’, la,G0,’:’);

Reflectance at 0o and 45o TM and TE bandwidths
800
100
700 In Fig. 7.9.3, the low-index material is changed slightly to nL = [1.5, 1.5, 1.9]. The main
TM
TE
behavior of the structure remains the same, except now the TM and TE bands are slightly
| ΓT (λ)|2 (percent)

80 600
0o different.
λ1, λ2 (nm)

500
60 Reflectance at 0o and 45o TM and TE bandwidths
400 800
100
40 300 700
TM
200 | ΓT (λ)|2 (percent) TE 600
80 0o
20

λ1, λ2 (nm)
100 500
60
0 0 400
400 500 600 700 800 900 1000 0 10 20 30 40 50 60 70 80 90
λ (nm) θa (degrees)
40 300

200 TM
Fig. 7.9.2 Birefringent mirror with identical TM and TE reflection bands. 20 TE
100

The right graph depicts the asymptotic (for large number of bilayers) bandedges of the 0 0
400 500 600 700 800 900 1000 0 10 20 30 40 50 60 70 80 90
reflecting band versus incident angle. They were computed with omniband2. Unlike the λ (nm) θa (degrees)
isotropic case, the TM and TE bands are exactly identical. This is a consequence of the
following relationships between the cosine factors in this example: cH,TM = cL,TE and Fig. 7.9.3 Birefringent mirror with slightly different TM and TE reflection bands.
cH,TE = cL,TM . Then, because we assume quarter-wave layers in both the TE and TM cases,
LH = LL = 1/4, we will have:
The MATLAB code used to compute the right graph was:
 7.9. Giant Birefringent Optics 239 240 Electromagnetic Waves & Antennas – S. J. Orfanidis

theta = linspace(0,90,361); % incident angles nL = [1.57; 1.57; 1.57];

F1e = []; F2e = []; la0 = 700;

F1m = []; F2m = []; la = linspace(400,1000,601);

Ni = 3; % refinement iterations N = 80;

n = [na, repmat([nH,nL], 1, N), nb];
for i=1:length(theta), L = [repmat([LH,LL], 1, N)];
[f1e,f2e] = omniband2(na,nH,nL,LH,LL,theta(i),’te’,Ni); L = L .* linspace(1,0.75,2*N); % 25% thickness gradient
[f1m,f2m] = omniband2(na,nH,nL,LH,LL,theta(i),’tm’,Ni);
F1e = [F1e,f1e]; F2e = [F2e,f2e]; Ge = 100*abs(multidiel(n, L, la/la0, 0, ’te’)).^2;
F1m = [F1m,f1m]; F2m = [F2m,f2m]; % frequency bandedges Gm = 100*abs(multidiel(n, L, la/la0, 0, ’tm’)).^2;
end
plot(la,Gm,’-’, la,Ge,’--’);
la1e = la0 ./ F2e; la2e = la0 ./ F1e; % wavelength bandedges
la1m = la0 ./ F2m; la2m = la0 ./ F1m;
A 25% thickness gradient was assumed in both cases. In the first case, the x-direction
plot(theta,la1m,’-’, theta,la2m,’-’, theta,la1e,’--’, theta,la2e,’--’); indices are different and the structure will act as a mirror for the TM polarization. The TE
polarization will be reflected only by the air-high interface.

As the incident angle increases, not only does the TM band widen but it also becomes wider In the second case, the materials are matched in their y-direction indices and therefore,
than the TE band—exactly the opposite behavior from the isotropic case. 
 the structure becomes a mirror for the TE polarization, assuming as always that the plane
of incidence is still the xz plane. 

Example 7.9.3: GBO Reflective Polarizer. By choosing biaxial high/low layers whose refractive
indices are mismatched only in the x or the y direction, one can design a mirror structure Giant birefringent optics is a new paradigm in the design of multilayer mirrors and
that reflects only the TM or only the TE polarization. polarizers [246], offering increased flexibility in the control of reflected light. The re-
Fig. 7.9.4 shows the reflectance of an 80-bilayer mirror with nH = [1.86, 1.57, 1.57] for cently manufactured multilayer optical film by 3M Corp. [823] consists of hundreds to
the left graph, and nH = [1.57, 1.86, 1.57] for the right one. In both graphs, the low index thousands of birefringent polymer layers with individual thicknesses of the order of a
material is the same, with nL = [1.57, 1.57, 1.57]. wavelength and total thickness of a sheet of paper. The optical working range of such
films are between 400–2500 nm.
TM Polarizer TE Polarizer Applications include the design of efficient waveguides for transporting visible light
100 100 over long distances and piping sunlight into interior rooms, reflective polarizers for
TM TM improving liquid crystal displays, and other products, such as various optoelectronic
| ΓT (λ)|2 (percent)

| ΓT (λ)|2 (percent)

80 TE 80 TE components, cosmetics, and ”hot” and ”cold” mirrors for architectural and automotive
windows.
60 60

40 40
7.10 Problems
20 20

0 0
400 500 600 700 800 900 1000 400 500 600 700 800 900 1000
λ (nm) λ (nm)

Fig. 7.9.4 TM and TE mirror polarizers.

The angle of incidence was θa = 0o . The typical MATLAB code was:

LH = 0.25; LL = 0.25;

na = [1; 1; 1];
nb = [1; 1; 1];
nH = [1.86; 1.57; 1.57];