612/952 ë MB ë 27/vii-01 ë SVERKA ë 3 ÒÑÎÑÔ
ÍÑÏÒ. å 3
ß2001 Kvantovaya Elektronika and Turpion Ltd
Quantum Electronics 31 (5) 464 ë 466 (2001)
PACS numbers: 42.25.Fx
DOI: 10.1070/QE2001v031n05ABEH001980
Mode separator for a beam with an off-axis optical vortex
M V Vasnetsov, V V Slyusar, M S Soskin
Abstract. The Gaussian beam diffraction by a thin amplitude
grating with a dislocation shifted relative to the beam axis is
considered. The érst-order diffracted beam with an off-axis
optical vortex is represented as a superposition of Laguerre ë
Gauss modes. The optical scheme permitting the spatial separation of modes with even and odd mode indices is proposed
and realised experimentally. The method of separation is based
on the difference in Gouy phase shifts for focused beams. The
possibility of separating photons with zero and nonzero orbital angular momentum is discussed.
Keywords: optical vortices, Laguerre ë Gauss modes, Gouy phase
shift, photon orbital angular momentum
Physical optics has been recently enriched by the concept
of optical vortex (OV) [1] used for describing the structure of
a wave éeld with phase defects (wave-front dislocations) [2].
The interest in OVs is due to their peculiar properties as
well as possible applications in problems of manipulation
with microparticles [3]. A distinguishing feature of an OV is
that the éeld amplitude vanishes at its axis, while the phase
becomes indeénite, or singular due to a jump by p or mp in
the case of an m-fold vortex (a review of the OV properties is
given in [4, 5]). If the OV axis coincides with the beam axis
(as, for example, for Laguerre ë Gauss modes with a nonzero azimuthal index), the integer value of m is called the
topological charge. The structure of beams with OVs can be
determined from the expression for the Laguerre ë Gauss
modes LGp l (in the cylindrical coordinates r, j, z):
p j l j
2
2r
w
r2
2r
exp ÿ 2 Lpj l j
E LGp l ELG 0
w
w
w
w2
kr 2
z
lj ÿ Q arctan
exp i kz
,
2R z
zR
(1)
where ELG is the amplitude parameter; w0 is transverse size
of the beam waist; w w0 (1 z 2 =zR2 )1=2 is the transverse
size of the beam at a distance z from the waist; R(z)
z(1 zR2 =z 2 ) is the radius of curvature of the wave front at
M V Vasnetsov, V V Slyusar, M S Soskin Institute of Physics, National
Academy of Science of Ukraine, prosp. Nauki 46, 03039 Kyiv, Ukraine;
e-mail: mvas@iop.kiev.ua
Received 30 August 2000; revision received 31 January 2001
Kvantovaya Elektronika 31 (5) 464 ë 466 (2001)
Translated by Ram Wadhwa
the beam axis; zR kw02 =2 is the Rayleigh length; k is the
wave number; Ljpl j is the associated Laguerre polynomial; l
is the azimuthal mode index; p is the radial mode index [6].
For beams with a nonzero index l, the amplitude vanishes
at the beam axis, while the phase is an exponential function
of azimuth, exp (ilj), which corresponds to an OV with the
topological charge m l. The number Q 2p j l j 1 is
known as the mode index determining the aféliation to a
family of modes with the same Q. The mode index determines the small correction to the phase velocity of the
mode, associated with the Gouy phase shift arctan (z=zR ).
For a Gaussian beam (p; l 0) Q 1 and the additional
phase incursion (relative to a plane wave) over the distance
between the waist and the far-éeld zone due to the Gouy
1
phase shift is ÿ p=2. The `doughnut' mode LG0 ( p 0,
l 1) has the index Q 2, and the corresponding additional phase incursion is equal to ÿp.
The expression for the 3D equiphase surface (wave
l
front) of the LG0 mode at the waist can be obtained
from Eqn (1) putting R(z) ! 1 and z 5 zR :
kz lj const.
(2)
The equation for the wave front (2) describes a helicoidal
surface with a step equal to ll (l is the wavelength). As the
wave front leaves the waist, its shape remains helicoidal
with a singularity (phase discontinuity of lp) at the axis.
Such a wave-front structure leads to the interference fringe
splitting (emergence of l new fringes) during the interference
of the beam containing OVs with a plane wave, which is the
main indication of the presence of OVs in interference
detection [7].
Another important consequence is the presence of the
orbital angular momentum of the beam [8] due to the
circulation of the optical êux around the OV axis. The
orbital angular momentum Lz for an axisymmetric OV
beam with energy W, frequency o, and topological charge m
is deéned as [8]
Lz
mW
,
o
(3)
which gives the value m
h when recalculated per photon.
The quantised value of the orbital angular momentum of
a photon either can be random or may reêect the physical
reality of the existence of the corresponding orbital quantum
number. At the present time, there are arguments in favour
of the quantum origin of the orbital angular momentum as
well as of a purely classical description [9]. For instance, the
coaxial interference of Laguerre ë Gauss modes leads to a
Mode separator for a beam with an off-axis optical vortex
465
combined beam [10] for which the speciéc orbital angular
momentum `per photon' may be arbitrary. This can be
interpreted as the result of superposition of photons with a
nonzero orbital angular momentum and photons belonging
to the LGp 0 modes, which have no orbital angular momentum. The position of OV in a combined beam is displaced
relative to the centre of the beam due to the presence of a
mode with an axial peak [11].
In this work, an attempt is made to separate the vortex
and the vortex-free components in a beam with an off-axis
OV. The idea of the method is to make use of the difference
in the Gouy phase shifts for modes with different values of
Q.
It is often difécult to generate laser radiation at a required Laguerre ë Gauss mode in laboratory conditions. For
this reason, a few extracavity procedures have been worked
out for obtaining OV beams, the simplest technique being
diffraction by a synthesised hologram [12]. The érst-order
diffracted beam has a phase-singularity point in the cross
section, i.e., contains an OV of an appropriate sign and
charge.
Consider a hologram located at the waist of the readout
Gaussian beam:
r2
(4)
EG / exp ÿ 2 .
w0
We will assume that the beam diameter is much smaller
than the transverse size of the hologram, but much larger
than its characteristic spatial period. For a hologram synthesised in the form of a éne amplitude grating with a dislocation, its transmission T has the form
T r; j T0 T1 r cos Kr cos j Mj,
(5)
exp ikz iKx iMj exp ikz ÿ iKx ÿ iMj
,
2
(6)
y
x
a
For a binary grating with a dislocation described by
formula (7), the derivation of the distribution function of
the érst-order diffracted éeld is quite cumbersome [13]. For
this reason, while determining the basic properties of the
diffracted beam, we will assume that T1 ( r) increases linearly with the distance from the grating centre and becomes
equal to unity for r R0 . (Note that the branching point of
the band is shifted relative to the grating centre by ML=2p
[11].) In this case, the diffraction of the Gaussian beam
(w0 5 R0 ) by a grating with M 1 leads to the following
érst-order diffraction éeld distribution in the plane behind
the grating (in Cartesian coordinates):
x iy
x2 y2
exp ÿ
,
(8)
E1 /
R0
w02
1
which corresponds to the Laguerre ë Gauss mode LG0 . Let
us now suppose that the grating centre is slightly displaced
relative to the centre of the readout beam, say, by a
distance x0 < R0 along the x axis. Then the corresponding
éeld distribution in the érst diffracted order has the form
x x0 iy
x2 y2
exp ÿ
.
(9)
E1 /
R0
w02
1
where T0 is the mean transmission; T1 ( r) is the band
contrast; K 2p=L; L being the grating period; M is the
order of the dislocation in the grating [11]. The grating with
M 1 has the splitting of the central band into two bands
(see Fig. 1a). Considering that r cos j x, the diffraction
éeld Ed in the case of grating readout by a plane wave
E0 exp (ikz) can be represented as the sum of the éelds of
the order zero and 1:
Ed E0 T0 exp ikz T1 r
in the 1 order, the topological charge of the OV is m M,
while in the ÿ1 order, m ÿM. The Gaussian beam readout of the grating leads to the same result.
In actual practice, synthesised holograms may have a
binary transmission distribution in the form
0; cos Kx Mj 4 0;
(7)
T x; j
1; cos Kx Mj > 0:
b
Figure 1. (a) Orientation of a Gaussian beam (grey circle) incident on a
grating with phase singularity M 1 (the centre of the grating is shifted
relative to the beam axis) and (b) the cross section of the diffracted beam.
This expression leads to a superposition of the LG0 mode
0
carrying OV and the `admixture' in the form of the LG0
mode (Gaussian beam) with a contribution proportional to
x0 . As a result, a combined beam is formed, in which the
OV shift relative to the beam axis is proportional to displacement x0 . Such a simpliéed approach leads to the following important conclusion: the éeld structure in the érst
diffraction order can be represented as a superposition of
the axial OV and the vortex-free component. For a binary
grating with the transmission function (7), the vortex component is a superposition of the LGp M modes, while the
vortex-free component is formed by the LGp0 modes. In
this respect, we may speak of various types of photons in a
beam with an off-axis OV, i.e., of photons with zero and
nonzero orbital angular momentum.
The schematic of the experiment is shown in Fig. 2. The
initial beam was obtained with the help of a syntesised binary hologram (M 1) giving a beam with a single OV in
the érst diffraction order [11]. The grating centre is displaced relative to the centre of the beam incident on the
grating (see Fig. 1a). The OV position in the diffracted beam
did not coincide with the centre of the beam (see Fig. 1b)
directed to the interferometer serving as a mode separator.
One of the arms of the interferometer formed by beamsplitters 3 and totally reêecting mirrors 4 contained two
confocal lenses 5. The propagation of the beam through the
gap between the lenses leads to an addition phase incursion
equal to the doubled Gouy shift due to beam focusing
followed by expansion. Thus, the resultant phase incursion
for the mode with index Q amounts to ÿ Qp. If, for ex-
466
M V Vasnetsov, V V Slyusar, M S Soskin
B
A
3
2
3
1
5
5
4
2f
4
Figure 2. Experimental scheme: ( 1 ) He ë Ne laser, ( 2 ) binary grating,
( 3 ) beamsplitters, ( 4 ) mirrors, ( 5 ) lenses.
ample, the path difference between the beam propagating to
channel A without multiple reêections and the beam
reêected to channel A from the second arm is equal to
an even number of half-waves for a mode with even Q, the
entire beam will be directed to channel A, whereas radiation
in channel B will be absent due to interference quenching.
For such a tuning of the interferometer, the path difference
for a mode with an odd Q is equal to an odd number of halfwaves, and the entire beam propagates through channel B.
The interferometer was tuned in accordance with beam
quenching at the exit of channel A for a Gaussian beam
(Q 1) at the separator entrance.
Because the beam with a single off-axis OV is a
superposition of modes with even Q (vortex component)
and odd Q (vortex-free component), the separator divides
the vortex and vortex-free components into different
channels. Note that the OV sign is reversed after reêection,
but the beams interfering at the exit A have OV of the same
sign as those in the initial beam due to an even number of
reêections.
With an appropriate tuning, different patterns are
observed at the exits A and B: an axial OV at the exit
A and a beam with a smooth wave front at exit B (Fig. 3a,
b), which was veriéed using the interference with an
additional plane wave (Fig. 3c,d).
Thus, we experimentally divided the modes with an odd
index Q (vortex-free component, l 0) and even Q (vortex
component, l 1). This division may be interpreted as the
possibility of spatial separation of photons, belonging to the
1
0
`vortex' LG0 mode and to a Gaussian beam (LG0 mode)
and forming a combined beam as a result of coherent
addition, using purely optical methods of spatial separation
along different channels.
Note that this method is based on the difference in the
Gouy phase shifts for modes with even and odd mode indices rather than on the existence of orbital angular momen2
0
tum. For instance, the modes LG0 (Q 3) and LG1
(Q 3), one of which has an orbital angular momentum
and the other has zero momentum, cannot be separated
using our scheme.
In our opinion, the existence of an orbital angular
momentum for a solitary photon can be established
only in experiments with light absorption by a medium
capable of detecting the angular momentum transfer. For
this and other problems, the obtaining of a `puriéed' OV is
very important, and the technique proposed by us here may
be used in such cases.
Acknowledgements. This work was carried out under project
No. P051 of the Research Laboratory of USA Air Force
(EOARD).
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
a
b
c
d
Figure 3. Patterns of the beam intensity distribution in the cross section
at exits A (a) and B (b) and the corresponding interferograms obtained as
a result of interference of these beams with an additional plane wave (c)
and (d).
Coullet P, Gil L, Rocca F Opt. Commun. 73 403 (1989)
Nye J F, Berry M V Proc Roy. Soc. Lond. A 336 165 (1974)
He H, Friese M E J, Heckenberg N R et al. Phys. Rev. Letts. 75
826 (1995)
Soskin M, Vasnetsov M Photonics Science News 4 (4) 21 (1999)
Vasnetsov M, Staliunas K (Eds.) Optical Vortices (N.Y.: Nova
Science, 1999)
Anan'ev Yu A Opticheskie Rezonatory i Problema Raskhodimosti
Lazernogo Izlucheniya (Optical Cavities and Laser Beam Divergence Problem) (Moscow: Nauka, 1979)
Baranova N B, Zel'dovich B Ya, Mamaev A V et al. Zh. Eksp.
Teor. Fiz. 83 1702 (1982)
Allen L, Beijersbergen M W, Spreeuw R J C, Woerdman J P,
Phys. Rev. A 45 8185 (1992)
Tiwari S C J. Mod. Opt. 46 1721 (1999)
Soskin M S, Gorshkov V N, Vasnetsov M V, et al. Phys. Rev. A
56 4064 (1997)
Basistiy I V, Bazhenov V Yu, Soskin M S, Vasnetsov M V Opt.
Commun. 103 422 (1993)
Bazhenov V Yu, Vasnetsov M V, Soskin M S Pis'ma Zh. Eksp.
Teor. Fiz. 52 1037 (1990)
Sacks Z S, Rozas D, Swartzlander G A, Jr. J. Opt. Soc. Am. B:
Opt. Phys. 15 2226 (1998)