Physics Letters A 372 (2008) 25272533

www.elsevier.com/locate/pla

Reflection of the light beam carrying orbital angular momentum

from a lossy medium
V.G. Fedoseyev
Institute of Physics, University of Tartu, Riia 142, 51014 Tartu, Estonia
Received 24 October 2007; received in revised form 27 November 2007; accepted 28 November 2007
Available online 4 December 2007
Communicated by V.M. Agranovich

Abstract
It is shown that after reflection from a lossy medium the s- or p-polarized paraxial light beam carrying the orbital angular momentum suffers
the 2D shift of the beams centre of gravity relative the geometric optic axis. The direction as well as the length of the 2D vector, which describes
the shift, change smoothly with the change of the angle of incidence.
2007 Elsevier B.V. All rights reserved.

PACS: 42.25.-p; 42.25.Gy; 42.60.Jf

Keywords: Orbital angular momentum; Reflection; Transverse shift; GoosHnchen shift

1. Introduction
In the communication [1] a specific nonspecular effect,
which accompanies the process of partial reflection of a parax-
ial light beam at a plane interface of two isotropic transparent
media, has been predicted. It was shown that, if the incident
beam carries the intrinsic orbital angular momentum (OAM),
the centre of gravity of the reflected beam (CGRB) suffers
a transverse shift, i.e. the shift perpendicular to the plane of
incidence. Unlike the previously known, spin-dependent trans-
verse shift (see, for instance, [27] and references therein), the
OAM-dependent transverse shift can take place, when the inci-
dent beam is s- or p-polarized. This effect appears because the
electromagnetic energy inside the beam is redistributed after re-
flection in such a way that the intensity increases at one side of
the incidence plane and decreases at the other [8,9].
The magnitude of the OAM-dependent transverse shift is, if
an angle of incidence is not close to the critical angle for total
reflection or, in the case of p-polarization, to the Brewster an-
* Fax: +372 7383033.
E-mail address: fedo@fi.tartu.ee.
0375-9601/$ see front matter 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.physleta.2007.11.059
gle, rather small: it is of the order of the wavelength or less. But,
at the present, the optical method has been developed, which
permit to detect the positions of the objects with nanometer-
scale precision (see, for instance, [10] and references therein).
This method has been used by Dasgupta and Gupta [11] in order
to define the OAM-dependent transverse shifts of the CGRBs in
the wide region of the angles of incidence, the authors of [11]
have reported a good agreement between their experimental re-
sults and the predicted values of the transverse shifts.
On the other hand, Okuda and Sasada [9] have observed the
above-mentioned redistribution of the electromagnetic energy
inside the reflected beam. Their experiments were carried out
in a situation, where the angle of incidence was close to the
critical angle for total reflection. In this region, the magnitude
of the effect increases significantly, so the energy redistribution
becomes visible (see Figs. 3 and 4 in [9]). The OAM-dependent
transverse shift of the CGRB was not calculated in [9], but this
calculation was possible [12].
The experimental results presented in papers [9] and [11]
show that the shift of the CGRB is a detectable effect.
In this Letter, the 2D shift of the CGRB, which accompanies
the reflection of the s- or p-polarized paraxial light beam, is
considered. It is shown that, if the reflecting medium is lossy
2528 V.G. Fedoseyev / Physics Letters A 372 (2008) 25272533

and the incident beam carries the intrinsic OAM, the direction
as well as the length of the 2D vector, which describe this shift,
change smoothly with the change of the angle of incidence. The
conditions under which the effect is pronounced, are pointed
out.
for both systems. The direction of the x (a) axis of the a-system,
which lies in the plane of incidence, is defined by the unit vector
x (a) = y z (a) .
In the a-system, the 3D radius-vector is represented as fol-
lows:

2. Geometry of reflection = z(a) z (a) + u(a) , (1)

where u(i) and u(r) are the 2D planar radius vectors lying in the
We shall consider the reflection of a monochromatic light planes, which are perpendicular to the axis of the incident beam
beam at a plane interface of two semi-infinite isotropic, nondis- and to the GO axis of the reflected beam, respectively,
persive, and nonmagnetic media. The scheme of the process
is shown in Fig. 1. The position of the interface is defined by u(a) = x (a) x (a) + y y . (2)
equation N = 0, where is the 3D radius vector, and N is
the unit normal to the interface directed from the first (incident) 3. Incident and reflected fields
medium to the second (reflecting) one. The dielectric constants
of the media in the upper half-space (N < 0) and the lower Let us assume that the incident beam is paraxial, that it car-
half-space (N > 0) will be denoted by 1 and 2 , respectively. ries the well-defined OAM (see, for instance, [13,14]), and that
The relative dielectric constant  = 2 /1 . The first medium is it is mainly s- or p-polarized. Due to first assumption, the rela-
assumed to be transparent, while the second medium may be tion
lossy, i.e. we assume that 1 is positive, while 2 and, as a con- /b  1, (3)
sequence,  may be complex, so, in general case,  =   + i  .
Here and further on, the prime and double prime stand for the must be fulfilled, where is the wavelength of light in the first
real and imaginary parts of a quantity. medium, and b is the beam radius.
Throughout this Letter, we shall use the superscripts i and r The electrical field vector of the incident beam under consid-
for indicating the quantities characteristic of the incident and of eration can be written, in its own coordinate system, as follows
the reflected beams, respectively; the superscript a will be used [15]:
in order to designate an arbitrary (incident or reflected) beam,   
(i) (i) d
so, a = i or r. We shall employ two coordinate systems con- E ()  e i z
(i) (i)
e (i) F (i) ()
2 du
nected with the incident and the reflected beams, these systems  
2 (i)
will be termed the i- and r-systems. exp i z + c.c., (4)
In the i-system, the z(i) axis is assumed to coincide with the
incident beams axis. This axis and the unit vector N define the where the subscript denotes the polarization of the incident
position of the beams plane of incidence in 3D space; the an- (i) (i)
beam ( = s or p), while e is the polarization vector: e s y ,
gle between the unit vectors N and z (i) is the beams angle of (i)
and e p x . If the incident beam carries the well-defined
(i)
incidence: = arccos(N z ). The coordinate origin O is taken
(i)
OAM, the function F (i) () can be represented as a product of
to be the point of intersection of the incident beam axis with the
two terms:
interface. In the r-system, the z(r) axis is assumed to coincide
   
with the geometric optic (GO) axis of the reflected beam. This F (i) () = f u(i) , z(i) exp il (i) , (5)
axis is defined as a ray, which is intersected by the interface at
the coordinate origin, and whose direction is characterized by where l is the azimuthal index (l = 0, 1, 2, . . .), and (i) is
the unit vector z (r) , which is related to z (i) through the Snel- the azimuth of the radius vector in the i-system obtained from
ls law and is written as follows: z (r) = z (i) 2(N z (i) )N.
The the relation: tan (i) = y/x (i) . The term f (u(i) , z(i) ) does not
beams angle of reflection is . depend on (i) , for instance, it may be given by the Laguerre
The y axis in every system is defined to be perpendicular Gaussian function [13,14] or, like in [9], by a superposition of
(i)
to the plane of incidence, it is characterized by the unit vector such functions. The time dependence of E () is suppressed.
y = N z (i) /|N z (i) |; this axis (the transverse one) is common The distribution of fields inside the reflected beam depends
upon . The electrical field vector of this beam E(r) (; ) is cal-
culated in the following way. First, the 2D Fourier transform of
the function F (i) () with respect to the coordinates x (i) and y is
to be performed. Then the vector E(i) () will be represented as
a superposition of the plane waves. Next, the Snells and Fres-
nels laws should be applied to the particular plane waves. After
that, the reverse 2D Fourier transform should be performed.
Let us denote the angle of incidence of the particular plane
wave inside the incident beam by (par) . For the actual waves
constituting a paraxial beam, | (par) | /b  1. So, one
Fig. 1. Geometry of reflection. can expand the particular Fresnel field reflection coefficients in
 V.G. Fedoseyev / Physics Letters A 372 (2008) 25272533 2529

the truncated power series with respect to (par) . When the [8] (see also [17]):
second medium is transparent, and   < 1, such an expansion sin
can be substantiated, if is not too close to the critical angle Qs ( ) = , (15)
for total reflection given by ( sin2 )1/2
and
(c) = arcsin   , (6)
Qs ( )
i.e. if the following condition is fulfilled: Qp ( ) = . (16)
 1 sin2 cos2
 
 (c)  /b, if   < 1, and   0. (7) Generally, Q ( ) is a complex quantity, its real and imaginary
parts are caused by the dependence on of the amplitude and
Again, in the case of p-polarization, the expansion is substanti-
the phase, respectively, of the Fresnels field reflection coeffi-
ated, if is not too close to the Brewster angle given by
cient r ().
(r)
(B) = arctan   , (8) The magnetic field vector of the reflected beam H (; ) is
obtained from Eq. (12) by means of the Faradays law.
i.e. if the following condition is fulfilled:
 
 (B)  /b, if p, and   0. (9) 4. 2D shift of the CGRB

But, when 4.1. Definition and results of calculations


| | /b, (10)
Let us select the plane, which is perpendicular to the GO axis
the series expansion is justified at any . of the reflected beam (the output plane in Fig. 1). The position
Let us restrict ourselves with the near-field region (in this of the CGRB on this plane is given by the 2D vector
case the possible angular shifts of the CGRB can be excluded
1
from consideration), i.e. let us assume that U ( ) = (r) u(r) w(r) (; ) du(r) , (17)
W
z(r)  b2 /. (11)
where w(r) (; ) is the electromagnetic energy density inside
If the restriction (11) together with the restrictions (10) or (7), the reflected beam
(9) are fulfilled, it is possible to retain only the zero-order and
1
 (r) 2  2 
first-order terms in the power series expansion. Then, perform- w(r) (; ) = 1 E (; ) + H(r) (; ) , (18)
ing the operations, which are mentioned in two by last para- 8
graph, one obtains the following expression for the electrical and W(r) is the electromagnetic energy per unit length of the
field vector of the reflected beam written in its own coordinate reflected beam:
system:
   W = w(r) (; ) du(r) .
(r)
(19)
d
E (; )  r ( ) e 1 + iQ ( )
(r) (r)
2 dx (r) The vector U ( ) is shown in Fig. 2 in [18]. As the GO axis
 
(r) (r) d   intersects the output plane at the point u(r) = 0 this vector de-
i z e (r) f u(r) , z(r)
2 du scribes the 2D shift of the CGRB relative to this axis.
  Let impose the following restriction on the axial coordinate
2 (r)
exp i z + il (r)
+ c.c. (12) of the output plane:

In the right-hand side of Eq. (12), the following definitions z(r) b/ sin(2 ). (20)
(r)
are used. e is the polarization vector of the reflected beam: If the restriction (20) is fulfilled, the reflected field on the
(r) (r)
e s y , and e p x (r) . (r) is the azimuth of the radius output plane is not overlapped with the incident field. Again,
vector in the r-system obtained from the relation: tan (r) = (r)
w (u(r) ; ) is negligible at the line of intersection of the output
y/x (r) . r ( ) is the Fresnels field reflection coefficient of the plane and the interface; in this case, the lower limit of inte-
-polarized plane wave, which is incident at an angle [16]: gration over x (r) in Eqs. (17) and (19) can be substituted by
sin( ) tan( ) . Once this substitution is made and the expression (12) for
rs ( ) = , rp ( ) = , (13) (r) (r)
sin( + ) tan( + ) E (; ) and respective expression for H (; ) obtained by
means of the Faradays law are used, the calculation of U ( )
where
is the angle of refraction given by the relation
is straightforward. In the first-order approximation with respect
 sin = sin , in a general case is complex. The factor
to /b, one obtains:
Q ( ) is defined as follows:
1 d   U ( ) = X ( )x(r) + Y ( )y, (21)
Q ( ) = ln r () = . (14)
2 d where
Basing on Eqs. (13) and (14), the calculations of the factors
Qs ( ) and Qp ( ) are straightforward, the results are as follows X ( ) = Q ( ) , (22)

2530 V.G. Fedoseyev / Physics Letters A 372 (2008) 25272533

and

Y ( ) = lQ ( ) . (23)

4.2. Analysis

As follows from Eq. (21), the shift of the CGRB relative to

the GO axis is, in a general case, described by the 2D vector
U ( ) on the output plane. Its x (r) -component (X ( )) repre-
sents the GoosHnchen shift (see, for instance, [4,17,1921]
and references therein), if this shift regards the position of the
CGRB. X ( ) does not depend on l. On the other hand, the
y-component (Y ( )) is proportional to l. Apart from the fac-
tor l, the functions X ( ) and Y ( ) are represented by the
imaginary and real parts of the same complex quantity, it is in
spite of the fact that the mechanisms of the appearances of the
orthogonal components of the 2D shift are different [8,22]. The
dependencies of the functions Q and Q on for different
values of   and   are illustrated in Figs. 3(a) and 3(b) in [18].
When the second medium is transparent, the vector U ( ) is
parallel either to the x (r) axis (in the total-reflection regime) or
to the y axis (in the partial-reflection regime). When the second
medium is lossy, the vector U ( ) is inclined to the coordi-
nate axes. By this, the direction as well as the length of the
vector U ( ) smoothly change with the change of the angle of
incidence. In others words, the vector U ( ) rotates around the
coordinate origin, i.e. around the CO axis, when changes.
The direction of U ( ) is defined by its azimuth ( ),
which is given by the relation: tan ( ) = Y ( )/X ( ). Sub-
stituting in its right-hand side Eqs. (22) and (23) one obtains the
following relation between the azimuth ( ) and the argument
of the complex quantity Q ( ) given by ( ) = (ln Q ( )) :
tan ( ) = l/ tan ( ). (24)
In particular, if l = 1,
 
( ) = ( ) . (25)
2 s on for different values of  and l are illustrated in Figs. 4(a)
As for the arguments of the quantities Qs ( ) and Qp ( ), they
are obtained from the expressions (15) and (16) and look as
follows:
   of the angle (c) and, for p-polarization, in the vicinity of the
1  sin2
s ( ) = arccotan , (26)
2 |  |
and
   is similar (see further Eq. (30)). It is seen from Fig. 2 that, if
 ||2 cot2
p ( ) = s ( ) arccotan . (27)
= 0, the effect of the change of the length and direction of the
|  |
Basing on Eqs. (24)(27) one can make the following conclu-
sions about the orientation of the vectors Us ( ) and Up ( ).
As follows from Eq. (26), /2 > s ( ) > at any , i.e.
Qs ( ) lies in the third quadrant of the complex plane; as a
consequence, the vector Us ( ) (see Eq. (24)) lies in the third
quadrant of the output plane (Xs ( ) < 0, Ys ( ) < 0), if l > 0,
and in the second quadrant (Xs ( ) < 0, Ys ( ) > 0), if l < 0.
The positions of Qp ( ) on the complex plane and of the vector
Up ( ) on the output plane, unlike of Qs ( ) and Us ( ), are not
Fig. 2. Dependence of the components Xs and Ys on in the vicinity of (c) .
  = 0.5 ( (c) = 45 ), and   = 0.001. The thin, middle, and thick curves
correspond to l = 0, 1, and 2, respectively.
contained within one quadrant, when changes from 0 to /2.
At small the direction of the vector Up ( ) is approximately
opposite with respect to the direction of the vector Us ( ). As
increases, the sign of Yp ( ) changes once independent of the
value of . The sign of Xp ( ) changes, if the condition
2  > ||2 (28)
is fulfilled. The angles, at which these changes take place, are
obtained from the following equation:
 
||2 cot4 + 2 1 +   ||2 sin2 3 = 0. (29)
If   0 and p, the second term in the right-hand side
of Eq. (27) have the jump discontinuity at the Brewster angle
(B) , where p changes abruptly by . If   0 and   < 1,
the (/2)-jump of s and p takes place at the critical angle for
total reflection (c) . The dependencies of the functions p and
and 4(b) in [18].
The scale of the 2D shift of the CGRB is / , but, if
|  |  1, its magnitude increases significantly in the vicinity
angle (B) , these angles are given by Eqs. (6) and (8), respec-
tively. Fig. 2 illustrates the behavior of the vector Us ( ) in the
former domain, the behavior of the vector Up ( ) in this domain
l
vector U ( ) in the process of the change of is pronounced.
Again, if |  |  1, and | (c) |  1, the position of the vector
U ( ) is sensitive to small changes of the values of   and  
(see Figs. 3 and 4). If the above-mentioned conditions are ful-
filled, the approximate expressions of the functions Q ( ) and
Q ( ) look as follows:
 1/2  1/2
 (2 + 1)1/2
Q,  ,
s ( ) =  Qp ( ) = ,
2|  | 2 + 1
(30)
 V.G. Fedoseyev / Physics Letters A 372 (2008) 25272533 2531

Fig. 3. Dependence of the components Xs and Ys on   in the vicinity of   =

sin2 . l = 1, = 45 . The thick and thin curves correspond to   = 0.001
and   = 0.01, respectively. Fig. 5. Dependence of the components Xp and Yp on in the vicinity of (B) .
  = 2.0 ( (B) = 54.7 ), and   = 0.001. The thin and thick curves corre-
spond to l = 1, and 2, respectively.

while |Qs ( )|is small in comparison with |Qp ( )|, namely,

Qs ( )  1/   , and |Qs ( )|  |Qs ( )|. A numerical analy-
sis of the dependence of the GoosHnchen shift, i.e. of the
quantity X , on has been recently performed by Lai et al.
The authors of [17] have paid special attention to the behavior
of this shift in the vicinity of the Brewster dip. Eq. (31) shows
that, if |  |  1, the function Xp ( ) is approximated in this re-
gion by the Lorentz curve.

4.3. Relative ps shift

Detection of the shift of the CGRB of the s- or p-polarized

Fig. 4. Dependence of the components Xs and Ys on   in the situation, when
is close to (c) . l = 1,   = 0.5. The thin and thick curves correspond to
= 44.9 = (c) 0.1 and = 45.1 = (c) + 0.1 , respectively.
where = ( (c) ) sin(2 (c) )/|  |, the signs minus and plus
in (30) and further on in this paragraph stand for Q ( ) and
Q ( ), respectively. Notice that, relative to the point = (c) ,
i.e. to the point = 0, the functions Q ( ) and Q ( ) given by
Eq. (30) are mutually symmetric. The magnitudes of these func-
()
tions achieve the maximums at the symmetrical points max
() () ()
and max , respectively, given by max = max = 1/ 3. The
values of Q ( ) and Q ( ) at these points are as follows:
Qs (max ) = Qs (max ) = (3/4)3/4 (  /|  |)1/2 .
() ()
The behavior of the vector Up ( ) in the vicinity of (B) is
shown in Fig. 5. Like in the previous case, the effect of the
change of the length and direction of the vector U ( ) with the
change of is pronounced in this domain; again, the position of
Up ( ) is sensitive to small changes of the values of   and   .
When |  |  1, and | (B) |  1,
 1
1 |  |
Qp ( )  + i
(B)
, (31)
2 2(1 +   )  
beam requires a careful determination of the position of the GO
axis. The latter procedure can meet the difficulties. One can es-
cape these difficulties by operating with the difference of the
shifts of two beams having the orthogonal polarizations. Such
a scheme has been realized in the experiments by Dasgupta
and Gupta [11] devoted to the definition of the one-dimensional
OAM-dependent transverse shift of the CGRB. In [11], the ini-
tial beam was randomly polarized, and the s- or p-polarization
were selected by means of rotation of a linear polarizer. The
other elements of the set-up were kept motionless. In view of
these experiments, it is interesting to present the analytic ex-
pression for the difference between the centres of gravity of
the respective reflected beams, which will further be denoted
by Ups ( ). Let us assume that both orthogonal incident beams
have the common axes, as a consequence, both reflected beams
have the common GO axes too. Such a situation takes place, for
instance, if, within the scheme of the experiment [11], the rota-
tion of the polarizer do not affect the position of the axis of the
incident beam. In this case, the expression for the vector Ups ( )
is directly obtained from Eqs. (17) and (19), it looks as follows:
Ups ( ) Up ( ) Us ( )
1  
= (r) (r) u(r) G(r) u(r) ; du(r) , (32)
Ws Wp
2532 V.G. Fedoseyev / Physics Letters A 372 (2008) 25272533
where G(r) (u(r) ; ) is the cross-correlation function,
 
G(r) u(r) ;
   
= wp(r) u(r) + v(r) ; ws(r) v(r) ; dv(r) , (33)
and where v(r) is, like u(r) , the 2D radius vector lying in the
(r)
plane perpendicular to e z . The vector Ups ( ), unlike the vec-
tors Us ( ) and Up ( ), does not depend on the position of the
GO axis. The cross-correlation function G(r) (u(r) ; ) can be de-
fined on the basis of the experimental data, hence, Eq. (32) can
be used as a basis for the experimental definition of Ups ( ). It
is to be compared with the predicted value of this vector. In the
considered case, the calculations presented previously yield the
following expression for the vector Ups ( ):
   sitive to small changes of the values of   and   , what is
U(r) x(r) + lQps ( )y ,
ps ( ) = Qps ( ) (34)
where Qps ( ) = Qp ( ) Qs ( ). Generalization of the above
result for the case, when the axes of the s- and p-polarized in-
cident beams are parallel, is straightforward.
5. Conclusions
When the s- or p-polarized paraxial light beam, which car-
ries the OAM, is reflected from the lossy medium, two mutually
orthogonal shifts of the CGRB to scale appear in the near-
field region, the total shift is described by the 2D vector Us ( )
or Up ( ) on the output plane. The vector U ( ), with = s
or p, is given by Eq. (21) and its orthogonal components are
given by Eqs. (22) and (23). The second medium is transpar-
ent, these expressions are correct, if conditions (7) and (9) are
fulfilled. The second medium is lossy, and condition (10) for
|  | is fulfilled, the above-mentioned expressions are correct at
any .
The vectors Us ( ) and Up ( ) are defined relative the GO
axis of the reflected beam; the difference between them (the
vector Ups ( ) Up ( ) Us ( )) does not depend on the posi-
tion of the GO axis, it is expressed through the cross-correlation
function of two energy distributions across the output plane
(Eqs. (32) and (33)).
The component of the vector U ( ), which is parallel to the
plane of incidence (X ( )) corresponds to the GoosHnchen
shift, it does not depend on the azimuthal index l. The trans-
verse component Y ( ), which represents the OAM-dependent
transverse shift of the CGRB, is proportional to l. Apart of the
factor l, the components X ( ) and Y ( ) are given by the
imaginary and real parts of the common quantity.
In the case of s-polarization the signs of both components
of 2D shift do not depend on : the vector Us ( ) lies in the
third quadrant of the output plane, if l > 0, and in the second
quadrant, if l < 0. On the contrary, the positions of the vector
Up ( ) on the output plane is not contained within one quadrant
when changes from 0 up to /2. The azimuths of the vectors
Us ( ) and Up ( ) are given by Eqs. (24)(27). At small the
directions of these vectors are approximately opposite. As in-
creases, the sign of Yp ( ) changes one time independent of the
value of , while the sign of Xp ( ) changes, if the condition
(28) is fulfilled. If the reflecting medium is lossy, the direc-
tion as well as the length of the vector U ( ) change smoothly
with the change of the angle of incidence. These effects are
most pronounced, if |  |  1 and is close to (c) or, for
p, to (B) . In these domains, the quantity Q ( ) is ap-
proximated by expression (30) or (31). In the vicinity of (c) ,
the dependence of the vector Us ( ) on is shown in Fig. 2,
the behaviors of the vectors Up ( ) and Ups ( ) are, due to the
left relation in (30), similar to behavior of Us ( ). In the vicin-
ity of (B) , the dependence of the vector Up ( ) on is shown
in Fig. 5.
The considered effect can be used in order to investigate
some features of the matter and the light. Thus, in the above-
mentioned domains, the position of the vector U ( ) is sen-
illustrated in Figs. 3 and 4. Hence, the effect may be used as
a tool for the investigation of such changes. Again, in view of
this fact, one can await that, if the reflecting medium is taken
to be nonlinear, the dependence of the position of the vec-
tor U ( ) on the light intensity should be pronounced in the
above-mentioned domains. On the other hand, redistribution of
the electromagnetic energy inside the reflected beam, which is
responsible for the appearance of the component Y ( ), is con-
nected with the rotation energy motion inside the incident beam
[2325]. Hence, if a deflection of the vector U ( ) or Ups ( )
from the x (r) axis is detected, one can conclude that such a mo-
tion takes place.
Acknowledgements
The author thanks V. Hizhnyakov for helpful discussions and
H. Sasada for valuable correspondence. This work was sup-
ported by Estonian Science Foundation (grant No. 6534).
References
[1] V.G. Fedoseyev, Opt. Commun. 193 (2001) 9.
[2] F.I. Fedorov, Zh. Prikl. Spektrosk. 27 (1977) 580.
[3] V.G. Fedoseyev, Opt. Spektrosk. 71 (1991) 992, Opt. Spectrosc.
(USSR) 71 (1991) 570.
[4] W. Nasalski, J. Opt. Soc. Am. A 13 (1996) 172.
[5] F. Pillon, et al., J. Opt. Soc. Am. B 22 (2005) 1290.
[6] M. Onoda, S. Murakami, N. Nagaosa, Phys. Rev. E 74 (2006) 066610.
[7] K.Yu. Bliokh, Yu.P. Bliokh, Phys. Rev. E 75 (2007) 066609.
[8] V.G. Fedoseyev, Proc. SPIE 4403 (2001) 90.
[9] H. Okuda, H. Sasada, Opt. Express 14 (2006) 8393.
[10] J. Gelles, B. Schnapp, M.P. Sheets, Nature 331 (1988) 450.
[11] R. Dasgupta, P.K. Gupta, Opt. Commun. 257 (2006) 9196.
[12] H. Sasada, private communication.
[13] L. Allen, et al., Phys. Rev. A 45 (1992) 8185.
[14] L. Allen, S.M. Barnet, M.J. Padgett (Eds.), Optical Angular Momentum,
Institute of Physics Publishing, Bristol and Philadelphia, 2003.
[15] H.A. Haus, Waves and Fields in Optoelectronics, PrenticeHall, Engle-
wood Cliffs, New Jersey, 1984.
[16] M. Born, E. Wolf, Principles of Optics, third ed., Pergamon, Oxford, UK,
1965.
[17] H.M. Lai, S.W. Chan, W.H. Wong, J. Opt. Soc. Am. A 23 (2006) 3208.
[18] V.G. Fedoseyev, arXiv: 0708.2335.
 V.G. Fedoseyev / Physics Letters A 372 (2008) 25272533 2533

[19] C.W. Hsue, T. Tamir, Opt. Commun. 49 (1984) 383.
[20] M.A. Porras, Opt. Commun. 135 (1997) 369.
[21] I.V. Shadrivof, A.A. Zharov, Yu.S. Kivshar, Appl. Phys. Lett. 83 (2003)
2713.
[22] V.G. Fedoseyev, in preparation.
[23] I.G. Marienko, et al., Opt. Commun. 213 (2002) 1.
[24] V.G. Fedoseyev, Abstracts of the International Seminar Day on Diffrac-
tion-2002, St. Petersburg, Russia, 2002, p. 28.
[25] V.G. Fedoseyev, Proc. SPIE 5481 (2004) 154.