Beruflich Dokumente
Kultur Dokumente
k with
k = e
1
e
2
represent, respectively, the
central and noncentral wave vectors of the incident beam,
where z
i
= x sin + z cos with x and z being unit vectors
directed along the laboratory x and z axes and being the
central angle of incidence; the polarization unit basis vectors
e
1
and e
2
are chosen in such a way that e
1
lies in the plane
containing both k and z axis, i.e., the plane of incidence with
respect to k, and e
2
is orthogonal to such a plane. Usually,
when the electric eld of a plane wave is parallel to either e
1
or e
2
, the wave is called either a TM (or p) or a TE (or s)
wave.
In the angular spectrum representation, the electric eld of
the incident beam on the air side can be written as [33]
E
I
(X
i
,Y
i
,Z
i
) =
E
I
(U,V)e
j(UX
i
+VY
i
+WZ
i
)
dUdV, (1)
where the incident angular spectrum
E
I
(U,V) = ( e
1
, e
2
)
(
2
)A(U,V); U, V, and W are the direction cosines of k in
the coordinate systemO x
i
y
i
z
i
; the dimensionless variables
X
a
= k
0
x
a
, Y
a
= k
0
y
a
, Z
a
= k
0
z
a
, with a = i or r; A(U,V)
and
m
(U,V)(m = 1,2) are the scalar and the vector spectral
amplitudes of the eld, respectively. A(U,V) determines the
spatial characteristics of the beam [34]. For a collimated beam
with a Gaussian spectral amplitude
A(U,V) = e
(
U
2
+V
2
2
0
)
e
jWD
, (2)
where
0
= 2/(k
0
w
0
) is the angular spread of the incident
beam with its waist w
0
at Z
i
= D [33].
m
(U,V) describes
the beams polarization [34]. Suppose the polarizer plate
across which the incident beam has passed is perpendicular
to the central wave vector k
0
, and the complex-valued unit
vector
f = f
P
x
i
+f
S
y
i
with |f
P
|
2
+|f
S
|
2
= 1 denotes the
orientation of the polarizer, then in the paraxial approximation
(U
2
+V
2
1) we have [35]
1 V cot
V cot 1
f
P
f
S
. (3)
In light of the Fourier transformations, the electric eld of
the reected beam obeys
E
R
(X
r
,Y
r
,Z
r
) =
E
R
(U,V)e
j(UX
r
+VY
r
+WZ
r
)
dUdV, (4)
where the reected angular spectrumis related to the boundary
distribution of electric eld through [35]
E
R
= ( e
r1
, e
r2
)r
A(U,V), (5)
and the unit vectors e
r1
, e
r2
for a well-collimated paraxial beam
are determined by [35]
e
r1
e
r2
1 V cot
V cot 1
x
r
y
r
. (6)
As known, in general, anisotropic or chiral media TE and
TM waves are not the eigenmodes of Maxwells equations,
so that the reected wave must be a combination of the
perpendicular and parallel components so as to meet the
boundary conditions. In this way, we describe the reection
of a wave by using the so-called reection matrix r, which is
dened as [36]
r =
r
pp
r
ps
r
sp
r
ss
, (7)
where r
ab
is the ratio of the amplitudes of a reected wave with
a polarization to an incident wave with b polarization; each a
and b denotes either s or p waves. Notably, r
ab
is a function
of U, V and dependent on , and thus it can be expanded in a
Taylor series. In the linear approximation, this yields [19,35]
r
ab
(U,V) = r
ab
(0,0) +U
r
ab
(U,V)
U
U = 0
V = 0
= r
ab
(0,0) +U
r
ab
(0,0)
, (8)
where r
ab
(0,0), which depends on , is obviously the ordinary
reection coefcient with respect to the central wave vector of
the incident beam. It follows fromEqs. (3), (5)(8) that the two
components of
E
R
(U,V) in the coordinate systemO x
r
y
r
z
r
are expressible as
E
x
r
(U,V) = {[r
pp
+Ur
pp
+(r
sp
r
ps
)V cot ]f
P
+[r
ps
+Ur
ps
+(r
ss
+r
pp
)V cot ]f
S
}A(U,V),
(9a)
and
E
y
r
(U,V) = {[r
sp
+Ur
sp
(r
pp
+r
ss
)V cot ]f
P
+[r
ss
+Ur
ss
+(r
sp
r
ps
)V cot ]f
S
}A(U,V),
(9b)
where r
ab
r
ab
(0,0) and r
ab
r
ab
(0,0)
m
r
I(X
r
,Y
r
,Z
r
)dX
r
dY
r
I(X
r
,Y
r
,Z
r
)dX
r
dY
r
, (10)
determines the centroid of the reected beam [34]. As a result,
the TS can be derived from
Y
r
=
Y
r
I(X
r
,Y
r
,Z
r
)dX
r
dY
r
I(X
r
,Y
r
,Z
r
)dX
r
dY
r
=
j
E
R
V
E
R
dUdV +Z
r
V
W
|
E
R
|dUdV
E
R
|
2
dUdV
. (11)
Keeping up to the rst-order term of U and V only and
after a cumbersome calculation, we obtain the TS Y
r
, which
consists of two parts, i.e., Y
r
=
ry
+Z
r
ry
. The rst part
ry
is Z
r
independent and linked to the spatial TS S
ry
, which
is given by
S
ry
=
ry
k
0
=
C
k
0
D
cot , (12)
and the second one is associated with the angular TS
ry
,
obeying
ry
=
1
2
F
D
2
0
, (13)
with
C = Im[r
pp
r
ps
+r
sp
r
ss
+2(r
sp
r
pp
|f
P
|
2
+r
ss
r
ps
|f
S
|
2
)
f
P
f
S
(|r
pp
|
2
+|r
ss
|
2
+2r
pp
r
ss
+|r
sp
|
2
+|r
ps
|
2
2r
sp
r
ps
)], (14a)
D = (|r
pp
|
2
+|r
sp
|
2
)|f
P
|
2
+(|r
ss
|
2
+|r
ps
|
2
)|f
S
|
2
+2Re[f
P
f
S
(r
pp
r
ps
+r
sp
r
ss
)], (14b)
and
F = Re[(|f
S
|
2
|f
P
|
2
)(r
pp
r
ps
+r
sp
r
ss
)
+f
P
f
S
(|r
pp
|
2
|r
ss
|
2
+|r
sp
|
2
|r
ps
|
2
)], (14c)
where Im(. . .) and Re(. . .) imply the evaluation of the
imaginary and the real parts of a complex number, respectively,
and the asterisk (
S
is purely imaginary, and hence
F = 0. It follows from Eqs. (13) and (14c) that circularly
polarized beams impinging on any planar interfaces do not
suffer the angular TSs.
Here, we should stress that Eqs. (12) and (13) can
be applied to various planar interfaces, including isotropic,
anisotropic, chiral, bi-isotropic, and bi-anisotropic interfaces,
etc., provided that the reection matrix can be determined. For
an air-chiral interface, the elements of the reection matrix are
given by [36]
r
pp
=
(1 g
2
)(cos
+
+cos
) cos 2g(cos
2
cos
+
cos
)
(1 +g
2
)(cos
+
+cos
) cos +2g(cos
2
+cos
+
cos
)
, (15a)
r
ps
= r
sp
=
2jg(cos
+
cos
) cos
(1 +g
2
)(cos
+
+cos
) cos +2g(cos
2
+cos
+
cos
)
, (15b)
r
ss
=
(1 g
2
)(cos
+
+cos
) cos +2g(cos
2
cos
+
cos
)
(1 +g
2
)(cos
+
+cos
) cos +2g(cos
2
+cos
+
cos
)
, (15c)
with
g =
/, cos
1
sin
2
n
2
, (16)
where
and n
= 90
c
= arcsin |n
|, (17)
where we have allowed for the existence of the negative
refraction, because recent studies have shown that > n is
possible at least at or near the resonant frequency of the
permittivity of a chiral medium (called chiral nihility [37]).
For > 0, it is distinct that
c+
is always greater than
c
, and
then there exist three possibilities, namely,
(i) When <
c
, both the RCPand LCPwaves propagate.
Their directions of propagation are given by Eq. (16).
(ii) When
c
< <
c+
, only the RCP wave can propa-
gate, and the LCP wave becomes evanescent.
(iii) When >
c+
, neither of the waves propagates and
there is total internal reection of the two waves into air.
053828-3
GUODING XU, TAOCHENG ZANG, HONGMIN MAO, AND TAO PAN PHYSICAL REVIEW A 83, 053828 (2011)
Clearly, the central angle of incidence , characteristic
parameters of the chiral medium , (or n), and affect
the propagation properties of the waves in the chiral medium.
III. NUMERICAL RESULTS AND DISCUSSION
In this section, we apply Eqs. (12)(15a) to the specic
air-chiral interfaces. For convenience, we introduce the polar-
ization operator = 2Im(f
p
f
S
) to describe the polarization
state of a beam. In our calculations, we choose the nonmagnetic
chiral medium, viz., its permeability = 1 and utilize the
right- and left-elliptically polarized beams as the incident
ones with the polarization states = 2/3 corresponding,
respectively, to (f
P
,f
S
) = (1, 1 j)/
3. As is known, the
paraxial approximation requires that the transverse size of
a beam is much greater than its wavelength [38]; as a
consequence, we choose the wavelength = 1.06 m and the
waist of beam w
0
= 25 m. Below, we consider two different
cases.
Case (a) n < 1; air is optically denser than the chiral
medium. In this case, the total internal reection can take
place for a proper chirality parameter . To shed light on
the connections between TS and the propagation of the
waves in the chiral medium, choosing n = 0.6 we depict the
dependences of the reectivity, the spatial and the angular
TSs on the central angle of incidence in Figs. 2(a)2(c)
for = 0.3 and in Figs. 2(d)2(f) for = 1. It is clear from
Fig. 2(a) that all the reectivity curves (the solid, dashed, and
dash-dot lines for |r
ss
|
2
, |r
pp
|
2
, and |r
ps
|
2
, respectively) attain
their two extrema near the two critical angles of incidence
c
= 17.46
and
c+
= 64.16
, respectively. It should be
noted that when >
c+
, the reectivity curves |r
ss
|
2
and
|r
pp
|
2
merge due to |r
pp
|
2
+|r
sp
|
2
= 1 and |r
ss
|
2
+|r
ps
|
2
= 1
on the total internal reection.
c
divides the reectivity
curves into three distinct parts, corresponding to the three
propagation types of waves, respectively, as demonstrated in
Sec. II. Obviously, the propagation behaviors of waves in
the chiral medium impose a great effect on the reection
characteristics of waves, and thereby on the TSs. We plot the
spatial and angular TSs as functions of in Figs. 2(b) and 2(c),
respectively. As expected, both the spatial and the angular TSs
are closely associated with the propagation behaviors of the
waves in the chiral mediumas well as their polarizations. More
specically, when both the RCP and LCP waves propagate in
the chiral medium( <
c
), the spatial TSs that the right- and
left-elliptically polarized beams undergo are almost identical
in magnitude but opposite in direction, which is very similar to
the result in the isotropic interface [1619,24,39]. This is due
to nearly vanishing |r
ps
|
2
(=|r
sp
|
2
) for <
c
[see Fig. 2(a)].
In the above case, both the right- and left-elliptically polarized
beams experience negative angular TSs, and the angular TSs
become larger as increases. When the RCP wave propagates
and the LCP wave becomes evanescent (
c
< <
c+
), the
right-elliptically polarized beam( = 2/3) suffers the positive
spatial TSs and the large angular TSs but varying from
negative to positive values with , whereas the left-elliptically
polarized beam( = 2/3) experiences the very minor spatial
and angular TSs. When both the RCP and LCP waves
become evanescent, i.e., the total internal reection takes place
( >
c+
), neither of the two elliptically polarized beams
suffers the spatial or angular TSs. We conclude from these
results that when the L(R)CP wave becomes evanescent in the
20 40 60 80
0
0.2
0.4
0.6
0.8
1
(deg)
R
e
f
l
e
c
t
i
v
i
t
y
20 40 60 80
4
2
0
2
4
(deg)
S
y
r
/
20 40 60 80
100
50
0
50
(deg)
y
r
(
s
)
20 40 60 80
0
0.2
0.4
0.6
0.8
1
(deg)
R
e
f
l
e
c
t
i
v
i
t
y
20 40 60 80
4
2
0
2
4
(deg)
S
y
r
/
20 40 60 80
60
40
20
0
20
(deg)
y
r
(
s
)
=2/3
=2/3
|r
ss
|
2
|r
pp
|
2
|r
sp
|
2
, |r
ps
|
2
(a)
(b)
(c)
=2/3
=2/3
(d)
(e)
(f)
=2/3
=2/3
=2/3
=2/3
|r
ss
|
2
|r
pp
|
2
|r
sp
|
2
, |r
ps
|
2
FIG. 2. (Color online) Dependences of the
reectivity, the spatial TS and the angular TS
(in seconds) on for n = 0.6. The reectivity
curves for (a) = 0.3 and (d) = 0.6, where
the solid, dashed, and dash-dot lines correspond
to |r
ss
|
2
, |r
pp
|
2
, and |r
sp
|
2
(|r
ps
|
2
), respectively.
(b) and (e) for the spatial TS curves, and (c) and
(f) for the angular TS curves, where the solid and
dashed lines correspond to the right- and left-
elliptically polarized incident beams with =
2/3, reectivity; (b) and (c) for = 0.3; (e)
and (f) for = 0.6.
053828-4
TRANSVERSE SHIFTS OF A REFLECTED LIGHT BEAM . . . PHYSICAL REVIEW A 83, 053828 (2011)
chiral medium, the left (right)-elliptically polarized incident
beamsuffers a smaller TS; when the L(R)CP wave propagates,
the left (right)-elliptically polarized incident beamexperiences
a greater TS; when the total internal reection happens, neither
of the two elliptically polarized incident beams suffer TSs.
Anyway, the propagation properties of the waves in the chiral
medium indeed affect the TSs of the elliptically polarized
incident beams.
For the case = 1, we have n
+
> 1 and n
= 0.4, and
then the RCP wave always propagates in the chiral medium,
while the LCP wave is refracted negatively for <
c
(= 23.58
are greater
than unity; therefore, both the RCP and LCP waves propagate
in the chiral medium. Figure 3(b) illustrates that the spatial TSs
suffered by the right- and left-elliptically polarized beams are
almost the same in magnitude but opposite in direction, which
is just the physical mechanism underlying the SHE of light.
However, it is noteworthy that their angular TSs are the same
in sign [see Fig. 3(c)]. The above results concur with those in
the isotropic interface [1619,24,39]. This is because is by
far smaller than n, and the mediums chirality cannot show up
sufciently, as can be veried from the tiny cross-polarized
reection coefcients r
ps
(r
sp
) [see Fig. 3(a)].
When = 1.1, we have n
+
> 1 and n
= 0.4. Apparently,
this case is similar to that for n = 0.6, = 1 shown in
Figs. 2(d)2(f); that is, the RCP wave always propagates,
while the LCP wave propagates for <
c
(=23.58
) and
becomes evanescent for >
c
. Interestingly, however, the
reectivity curves for |r
ss
|
2
and for |r
pp
|
2
nearly interchange, as
compared with Fig. 2(d). Owing to the interchange, the angular
TSs become positive for the two elliptically polarized beams,
which differs from the case n = 0.6, = 1 [cf. Figs. 2(f)
and 3(f)].
To facilitate the following discussion, we here demonstrate
the two special points concerning the chiral medium: the chiral
nihility [37] mentioned above, where n = 0, while = 0, and
the traditional limitation point (n = ) n
= 0, a singular
point, where is very close to n and the strong spatial
dispersion occurs.
It was reported that the refractive index gradient plays a
crucial role in the TSs [1622,24,26]. In practice, the alteration
of n or of the chiral medium implies the change in refractive
index gradient. The advent of metamaterials offers us the
possibility of tailoring n and articially. Hence, in what
follows we discuss the dependences of the TSs on n and .
The propagating RCP wave can be easily achieved in the
20 40 60 80
0
0.2
0.4
0.6
0.8
(deg)
R
e
f
l
e
c
t
i
v
i
t
y
20 40 60 80
4
2
0
2
4
(deg)
S
r
y
/
20 40 60 80
8
6
4
2
0
(deg)
r
y
(
s
)
20 40 60 80
0
0.2
0.4
0.6
0.8
1
(deg)
R
e
f
l
e
c
t
i
v
i
t
y
20 40 60 80
4
2
0
2
4
(deg)
S
r
y
/
20 40 60 80
20
0
20
40
60
r
y
(
s
)
=2/3
=2/3
|r
ss
|
2
|r
pp
|
2
|r
ps
|
2
, |r
sp
|
2
|r
ss
|
2
|r
pp
|
2
|r
ps
|
2
(a)
(b)
(c)
=2/3
=2/3
=2/3
=2/3
=2/3
=2/3
(d)
(e)
(f)
FIG. 3. (Color online) Dependences of the
reectivity, the spatial TS, and the angular TS
(in seconds) on the central angle of incidence
for n = 1.5. The reectivity curves for (a)
= 0.4 and (d) = 1.1, where the solid,
dashed, and dash- dot lines correspond to |r
ss
|
2
,
|r
pp
|
2
, and |r
sp
|
2
(|r
ps
|
2
), respectively. (b) and
(e) for the spatial TS curves, and (c) and (f)
for the angular TS curves, where the solid
and dashed lines correspond to the right- and
left-elliptically polarized incident beams with
= 2/3, reectivity; (b) and (c) for = 0.4;
(e) and (f) for = 1.1.
053828-5
GUODING XU, TAOCHENG ZANG, HONGMIN MAO, AND TAO PAN PHYSICAL REVIEW A 83, 053828 (2011)
00.158 0.5 0.8421 1.5 2
0.6
0.4
0.2
0
0.2
0.4
0.6
0.8
n
S
r
y
/
r
y
(
s
)
0 0.5 1 1.5 2
0.8
0.6
0.4
0.2
0
0.2
0.4
0.6
n
S
r
y
/
0 0.5 1 1.5 2
60
40
20
0
20
40
60
80
n
r
y
(
s
)
(a)
(b)
(c)
(d)
A
B
C
A
B
C
20
30
40
40
20
30
20
30
40
20
30
40
FIG. 4. (Color online) Dependences of the spatial TS and the angular TS (in seconds) on n for the right-elliptically polarized incident beam
with = 2/3. The spatial TS curves for (a) = 0.5 and (c) = 1, and the angular TS curves for (b) = 0.5 and (d) = 1, where the solid,
dashed, and dash-dot lines correspond to = 20
,30
, and 40
, reectivity. In Figs. 4(a) and 4(b), A and C are the critical points where the
total internal reections happen for the LCP wave; B is the traditional limitation point, where n
= 0.
chiral medium thanks to n
+
= n + , and appears to evoke
a great angular TS for the right-elliptically polarized incident
beam. Aiming at such an incident beam and different , we
plot the curves of the spatial and angular TSs versus n in
Figs. 4(a) and 4(b) for = 0.5 and in Figs. 4(c) and 4(d) for
= 1, respectively. We can rst see from these gures that
at the chiral nihility, the spatial TSs have negative maxima
and the angular TSs are almost absent for these angles of
incidence, and that the greater is, the smaller the spatial and
angular TSs become. Below, we focus only on the solid lines in
Figs. 4(a) and 4(b), i.e., the case = 20
= 0.342. Since n
+
= n +0.5 >
0.342 for n > 0, the RCP wave always propagates in this case.
However, considering the negative index, for the LCP wave we
can obtain from n