1612

J. Opt. Soc. Am. A / Vol. 18, No. 7 / July 2001

Franco Gori

Polarization basis for vortex beams

Franco Gori
` Dipartimento di Fisica, Universita degli Studi Roma Tre, and Istituto Nazionale per la Fisica della Materia, Via della Vasca Navale 84, I-00146 Rome, Italy Received September 6, 2000; revised manuscript received January 31, 2001; accepted February 1, 2001 Attention is called to a polarization basis formed by four Jones vectors that is reducible to two basic structures. In these states the eld is linearly polarized at any point, but the polarization direction changes with the angular coordinate. Very simple equations hold for the eld lines of these basis vectors. The adoption of this basis is of interest for beams possessing a single vortex of any order, say, m, as well as for beams with two vortices of opposite charges, m and m. As an example, the application to vectorial BesselGauss beams is briey discussed. 2001 Optical Society of America OCIS codes: 260.5430, 350.5500.

1. INTRODUCTION
The vectorial properties of nonuniformly polarized beams in the paraxial regime have revealed signicant features since the advent of lasers. In recent times, interest in this subject has been growing, and several vectorial treatments have been presented in both the coherent (see, for example, Refs. 16) and the partially coherent (see, for example, Refs. 712) cases. The simplest approach to studying arbitrarily polarized beams is to decompose the representative eld vector (either electric or magnetic) at any point of a section into orthogonal linearly polarized parts. Free-propagation problems can then be treated as a pair of scalar problems. The same is true if two orthogonal circular-polarization states are used as a basis. On the other hand, signicant results have been obtained by the representation of the eld at a typical point as the sum of a radial and an azimuthal component. As an example, Jordan and Hall2 showed that the propagation process of an azimuthally polarized BesselGauss beam can be treated by means of a single one-dimensional propagation integral with a suitable kernel. In this paper, we wish to put into evidence that there is at least one class of beams for which yet another approach can be fruitful. This is the class of beams that possesses a vortex of charge m, say, a superposed vortex with the opposite charge m, or both. Vortices rst appeared as rather exotic singularities in optical elds, but it was soon realized that they are present in common optical beams as well as in speckle patterns.1316 Familiar examples are LaguerreGauss17 and Bessel18 beams. Experimental production of LaguerreGauss beams on the basis of mode converters,19,20 interferometers,21,22 or intracavity laser devices23,24 is rather well established. Transversely windowed forms of Bessel beams are also realizable.25,26 For beams of this type a basis composed of four nonuniform polarization states can be used. The four basis Jones vectors correspond to elds that are linearly polarized at any point in a typical cross section of the beam. The polarization direction is a function of the angular coordinate. An attractive feature of these states
0740-3232/2001/071612-06$15.00

is that the corresponding polarization patterns are quite simple and elegant. The equations of the eld lines can easily be derived and specify curves that are well known from geometry. As for the intensity distribution, the four basis states are circularly symmetric. The case of m 1 deserves special mention. It includes as particular cases the azimuthal and the radial polarizations with a continuous passage from one to the other through intermediate states of polarization in which the eld lines have the form of spirals. It further is seen that, except for m 1, two of the basis states are simply rotated versions of the other two. For any of the four basis polarization states, and, indeed, for any linear combination of them, the polarization pattern remains angularly invariant at any cross section of the beam, and the propagation process can be studied by means of a single one-dimensional integral. As an example, we discuss BesselGauss beams of an arbitrary order m and show how they can be used to describe other BesselGauss structures that were recently studied in connection with elds that are generated by concentriccircle-grating, surface-emitting (CCGSE) semiconductor lasers.

2. PARAXIALLY PROPAGATING BEAMS AS A SUPERPOSITION OF VORTEX FIELDS

In this section, we briey recall how a paraxially propagating eld with linear polarization can be expressed through a Fourier series with respect to the angular variable and emphasize the interpretation of this expansion in terms of vortex elds. In the paraxial regime the electromagnetic eld is approximately TEM and can be described by a single vector (for example, the electric or the magnetic eld). Suppose that the function V( , , z) species, in cylindrical coordinates, the eld of a monochromatic, linearly polarized beam. The z axis is assumed to be the mean axis of the beam. The propagated eld can be evaluated through the Fresnel formula17
2001 Optical Society of America

Franco Gori

Vol. 18, No. 7 / July 2001 / J. Opt. Soc. Am. A

1613

, ,z
2

i exp ik z z ik exp 2z
2

2z
0

V
0

,0

ik z cos d d , (1)

where k 2 / ( being the wavelength). We expand V( , , 0) into a Fourier series with respect to V p, , 0
m

v 0m

exp im

(2)

the same charge at any plane z constant 0. Wellknown examples of elds of this type are the Laguerre Gauss beams in both their ordinary and elegant versions28 and Bessel beams.18 It should be stressed that, in accord with Eqs. (2) and (6), the eld of any beam can be thought of as a superposition of vortex elds. This concept holds true even if the eld to be expanded does not exhibit any vortex structure. Let us remark that converting m into m does not change the propagation law, as can be seen from Eq. (7) if we take into account the property J m (t) ( 1) m J m (t) of the Bessel functions. Suppose now that two elds with opposite charges and the same radial distributions are superposed at z 0, in accord with a formula of the form V , ,0 v 0m v 0m a exp im a a Cm a exp i a im a Sm . (8) Here, a and a are arbitrary complex coefcients, and we let C m cos(m ) and S m sin(m ). It can be seen that an angular intensity modulation is introduced by the superposition. In view of the above remark about the change of m into m, we conclude that the propagated eld is given by V , ,z v zm a a Cm i a a Sm . (9)

where 1 v 0m 2
2

V
0

, , 0 exp

im

d .

(3)

We refer to the v 0m as the Fourier coefcients, although they are functions of . On inserting Eq. (2) into Eq. (1) and letting /2, we obtain
2

ik exp ik z V , ,z z i
m m

2z

exp im
0

v 0m

k exp i 2z

k Jm z d , (4)

where the integral representation of Bessel functions27 1 Jm t 2

2

In other words, the angular dependence remains unchanged on propagation. Because Eq. (9) accounts for vortices of both m and m values, from now on m is taken to be nonnegative. A beam described by Eq. (9) is said to be of the order m.

exp i m
0

t sin

(5)

3. POLARIZATION BASIS
Up to now, we assumed the eld to be linearly polarized. We now exploit the linearity of Maxwells equations to construct more general polarization states through superposition. More precisely, we combine beams of the order m with linear polarization along the x and the y axes that exhibits identical radial dependences. By use of the Jones formalism29 the representative vector of the wave, say, the electric eld E, will be represented, in the most general case, by Ex v zm c3 c1 1 exp im 0 c4 c2 1 exp 0 im im

was used. Equations (2) and (4) represent the eld at z 0 and the propagated eld, respectively, as superpositions of elds with a single vortex whose charge m runs from to . Let us write Eq. (4) in the form V where v zm i
m 1

, ,z
m

v zm

exp im

(6)

k z exp ik z

/ 2z k

Ey

v 0m
0

exp ik

/ 2z J m

d . (7)

0 exp im 1

0 exp 1

(10)

Equations (6) and (7) put into evidence a signicant feature of the present expansions: Each vortex eld propagates independently from the others. Indeed, the coefcient v zm ( ) at any plane z constant 0 depends on only the corresponding coefcient v 0m ( ) at the plane z 0. In other words, a single-vortex distribution with a charge m at z 0 gives rise to a single-vortex eld with

where the c i are arbitrary complex coefcients. Alternatively, it is easily seen that any resultant state of polarization can be expressed through a linear combination of the following Jones vectors (denoted by a caret): v 1m v 3m Cm , Sm Sm , Cm v 2m v 4m Cm , Sm Sm . Cm (11)

1614

J. Opt. Soc. Am. A / Vol. 18, No. 7 / July 2001

Franco Gori

Therefore vectors (11) constitute a basis for representing a polarized beam of order m. We now discuss the vectors main features. First, it can be seen that they represent polarization states in which the eld is linearly polarized at any point of a beam section. The polarization direction, however, changes with the angular coordinate . We also note that the intensity distribution associated with any one of them is circularly symmetric (being proportional to cos2 m sin2 m ). Let us now investigate the polarization patterns that correspond to vectors (11). Basically, we have to consider the two following structures: u 1m u 2m In fact, we have v 1m v 3m u 1m u 1m 0 , , v 2m v 4m u 2m u 2m 0 , . (13) cos m sin m cos m sin m , . (12)

0. For m 1, instead, the seen in the x, y frame for components of E have the same expressions in any rotated frame. Let us now consider the case in which E x , E y are pro( portional to the components of u 2m ) ( ). Proceeding as before, we nd that Ex Ey E cos m E sin m m m 1 1 , . (20) (21)

In this case the effect of is always equivalent to a rota/(m 1) of the reference frame. tion by There remains to be found the form of the eld lines. Thanks to the above results, we can limit ourselves to ( ( v 1m ) with m 1 and v 2m ) for any m and deal separately (1) with u 1 ( ). The eld lines are found by means of solving the differential equation dy dx Ey Ex . (22)

( If we insert the components of v 1m ) (with m 1) into Eq. (22) and express dx and dy through polar coordinates, we obtain sin d cos d cos d sin d sin m cos m . (23)

To see the effect of on the polarization pattern, we consider a frame that is x , y rotated by an angle with respect to x, y (see Fig. 1). The Cartesian components E x , E y and E x , E y of a typical vector E in the two frames are related by Ex Ey E x cos E x sin E y sin , E y cos . (14) (15)

Equation (23) is easily transformed into d cos m sin m If is limited to the interval 0 is integrated at once to yield
m 1

1 1 d . /(m 1 , (24) 1) Eq. (24) (25)

If we assume that E x and E y are proportional to the com( ponents of u 1m ) ( ) or Ex Ey and use the relation and (15) Ex Ey E cos m E sin m E cos m E sin m , (16) (17) , we obtain from Eqs. (14) m m 1 1 , . (18) (19)

Rm

sin m

0 the polarizaExcept for the case of m 1, when tion pattern seen in a frame that is x , y rotated by /(m 1) is identical to the one that would be

where R is a constant. The curves described by Eq. (25) are known as sinusoidal spirals.30 To complete the pattern, we must nd how the eld lines behave in the interval /(m 1) 2 . By use again of Eqs. (14) and (15) it is not difcult to show that the eld lines for /(m 1) 2 /(m 1) are obtained from those /(m 1) by a rotation of of the interval 0 /(m 1) and by an inversion of the eld vector. Proceeding in this way for the remaining values of , we nd that the polarization pattern resembles a sort of ower with 2(m 1) petals. As an example, curves corresponding to m 4 are shown in Fig. 2. Certain sinusoidal spirals have specic names. As an example, the case of m 3 gives rise to the lemniscate of Bernoulli. ( Passing on to v 2m ) and proceeding as before, we nd the equation
m 1

Rm cos m

(26)

Fig. 1.

Geometrical sketch of two mutually rotated frames.

for / 2(m 1) / 2(m 1) . In this case the rotation and the inversion processes for the eld lines take place in steps of /(m 1) so that the overall pattern is made up of 2(m 1) sections. Curves specied by Eq. (26) also belong to the family of sinusoidal spirals.30 Essentially, they are obtained from those of Eq. (25) by the geometrical process of inversion with re-

Franco Gori

Vol. 18, No. 7 / July 2001 / J. Opt. Soc. Am. A

1615

spect to the origin taken as a pole.30 The case of m 2 is illustrated in Fig. 3. The case of m 1 gives rise to the equilateral hyperbolas. ( Let us now derive the eld-line equation for u 11 ) ( ). ( On inserting the components of u 11 ) ( ) [see Eqs. (12)] into Eq. (22), we obtain sin d cos d cos d sin d sin cos cos cos cos sin sin sin , (27)

and Eq. (27) can be transformed by simple algebra into d cos sin d . (28)

The general integral of Eq. (28) is R exp cot . (29)

Fig. 4. Field lines for the polarization state described by Eq. (29) with /2 0.1.

Equation (29) is also a well-known curve called the logarithmic spiral,30 or the equiangular spiral. The latter name reects the property that any line through the origin cuts the spiral at a constant angle, namely, .31 An example is shown in Fig. 4 in which /2 0.1. /2 the spirals degenerate into circles, For whereas for 0 they tend to lines through the origin. Accordingly, the general structure (29) includes azimuthal and radial polarizations as particular cases. Incidentally, although azimuthal and radial polarizations are well known, the intermediate case of spiral polarization does not seem to have received much attention.

4. BESSELGAUSS BEAMS
Fig. 2. Field lines for the polarization state described by Eq. (25) with m 1.

As an example, we consider in this section simple vectorial forms for BesselGauss beams. The extension from the scalar treatment of the zero-order BesselGauss beam32,33 to vectorial analyses for beams of higher order has been considered by several authors.2,3 In particular, very interesting results have been obtained in relation to beams emitted by CCGSE lasers for which BesselGauss structures furnish elegant models.34,35 An appealing feature of the basis in Eq. (29) is that the propagation of any eld whose polarization state is speci( ed by an arbitrary combination of the vectors v i m ) can be treated by the one-dimensional integral in Eq. (7). This method somehow generalizes the beautiful result by Jordan and Hall2 that relates to the azimuthal BesselGauss beam (corresponding to m 1). Then let us suppose that the radial part of the eld distribution at z 0 is
2

v 0m

AJ m

exp

2 w0

(30)

Fig. 3. Field lines for the polarization state described by Eq. (26) with m 2.

where A is a constant, w 0 species the spot size of the Gaussian envelope, and determines the transverse scale of the Bessel function. The integral appearing in Eq. (7)

1616

J. Opt. Soc. Am. A / Vol. 18, No. 7 / July 2001

Franco Gori

can be evaluated by means of Eq. 6.633.2 of Ref. 27. After some algebra the radial part of the propagated eld can be cast into the form v zm A w0 w z Jm
2

with the orders m 1 and m 1. Similar results hold for the other solutions derived in Ref. 5.

exp i k

2k

5. CONCLUSIONS
We have examined a polarization basis for paraxial vortex elds. It has turned out that the corresponding eld lines are elementary curves of classical geometry. As an application, we have showed that modes of the radiation generated by CCGSE lasers are readily expressed by use of such a basis. In view of the relevance of vortex elds the polarization basis discussed in this paper should prove to be useful in other applications as well.

iz/L 1 ik 2R z
2 2 2

exp

w z

k2

(31)

where L kw 0 2 /2 is the Rayleigh range of the Gaussian component and w(z), R(z), and (z) have the wellknown expressions17 z w z w0 1 L2 R z z z
1 2 1/2

L , z L .

REFERENCES AND NOTES

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. R. Simon, E. C. G. Sudarshan, and N. Mukunda, GaussianMaxwell beams, J. Opt. Soc. Am. A 3, 536540 (1983). R. H. Jordan and G. D. Hall, Free-space azimuthal paraxial wave equation: the azimuthal BesselGauss beam solution, Opt. Lett. 19, 427429 (1994). A. A. Tovar and G. H. Clark, Concentric-circle-grating, surface-emitting laser beam propagation in complex optical systems, J. Opt. Soc. Am. A 14, 33333340 (1997). S. R. Seshadri, Electromagnetic Gaussian beam, J. Opt. Soc. Am. A 15, 27122719 (1998). P. L. Greene and D. G. Hall, Properties and diffraction of vector BesselGauss beams, J. Opt. Soc. Am. A 15, 3020 3027 (1998). C. J. R. Sheppard, Polarization of almost-plane waves, J. Opt. Soc. Am. A 17, 335341 (2000). L. Mandel and E. Wolf, Optical Coherence and Quantum Optics (Cambridge U. Press, Cambridge, 1995). D. F. V. James, Change of polarization of light beams on propagation in free space, J. Opt. Soc. Am. A 11, 1641 1643 (1994). F. Gori, Matrix treatment for partially polarized, partially coherent beams, Opt. Lett. 23, 241243 (1998). F. Gori, M. Santarsiero, S. Vicalvi, R. Borghi, and G. Guattari, Beam coherencepolarization matrix, Pure Appl. Opt. 7, 941951 (1998). G. Piquero, J. M. Movilla, P. M. Mejas, and R. MartnezHerrero, Beam quality of partially polarized beams propagating through lens-like birefringent elements, J. Opt. Soc. Am. A 16, 26662668 (1999). S. R. Seshadri, Partially coherent Gaussian Schell-model electromagnetic beams, J. Opt. Soc. Am. A 16, 13731380 (1999). M. Berry, Singularities in waves and rays, in Physics of Defects, R. Balian, M. Kleman, and J.-P. Poirer, eds. (NorthHolland, Amsterdam, 1981), p. 454. G. Indebetouw, Optical vortices and their propagation, J. Mod. Opt. 40, 7380 (1993). I. Freund, Optical vortex trajectories, Opt. Commun. 181, 1933 (2000). F. Vivanco and F. Melo, Surface spiral waves in a lamentary vortex, Phys. Rev. Lett. 85, 21162119 (2000). A. E. Siegman, Lasers (University Science, Mill Valley, Calif., 1986). J. Durnin, J. J. Miceli, and J. H. Eberly, Diffraction-free beams, Phys. Rev. Lett. 58, 14991501 (1987). E. Abramochkin and V. Volostnikov, Beam transformations and nontransformed beams, Opt. Commun. 83, 123 135 (1991). M. W. Beijersbergen, L. Allen, H. E. L. O. van der Veen, and J. P. Woerdman, Astigmatic laser mode converters and transfer of orbital angular momentum, Opt. Commun. 96, 123132 (1993). S. C. Tidwell, D. H. Ford, and W. D. Kimura, Generating

tan

(32)

Any state of polarization specied by a linear combination ( of the v i m ) preserves its angular dependence on propagation, whereas the radial part is given by Eq. (31). In particular, if the polarization state corresponds to only one of ( the vectors v i m ) the intensity at any cross section of the beam is circularly symmetric. To describe beams generated by CCGSE lasers, it was necessary to introduce more sophisticated eld structures. Because they are also known under the name BesselGauss beams, we refer to the elds whose radial part is specied by Eq. (31) as elementary BesselGauss beams. It is interesting to study how the more sophisticated structures are related to the elementary beams. Without going into a detailed analysis, we refer to a particular solution derived by Greene and Hall5 that reads f f , ,z , ,z a mQ z J m
1

u
1

Jm u

u cos m
1

, (33)

12. 13. 14. 15.

, (34) where f and f are the radial and the azimuthal components, respectively, of the electric eld. Furthermore, a m is a constant. The quantities u and Q are given by i u 1 iz/L , Q z exp 1
2

a mQ z J m

Jm

u sin m

z/ 2k iz/L . (35)

16. 17. 18. 19. 20.

The eld specied by Eqs. (33) and (34) is representative of the general structure of modes for CCGSE lasers. By use of simple trigonometric manipulations it can be shown that the eld associated with Eqs. (33) and (34) can be written in Jones notation as Ex Ey a mQ z J m
1

u v 2m

Jm

u v 1m

(36)

Accordingly, the eld appears as a superposition of two elementary BesselGauss beams of the form of Eq. (31)

21.

Franco Gori radially polarized beams interferometrically, Appl. Opt. 29, 22342239 (1990). S. C. Tidwell, G. H. Kim, and W. D. Kimura, Efcient radially polarized laser beam generation with a double interferometer, Appl. Opt. 32, 52225229 (1993). A. A. Tovar, Production and propagation of cylindrically polarized LaguerreGaussian beams, J. Opt. Soc. Am. A 15, 27052711 (1998). R. Oron, N. Davidson, A. A. Friesem, and E. Hasman, Efcient formation of pure helical laser beams, Opt. Commun. 182, 205208 (2000). J. Turunen, A. Vasara, and A. T. Friberg, Holographic generation of diffraction-free beams, Appl. Opt. 27, 39593962 (1988). R. M. Herman and T. A. Wiggins, Bessel-like beams modulated by arbitrary radial functions, J. Opt. Soc. Am. A 17, 10211032 (2000). I. S. Gradshteyn and I. M. Ryzhik, Tables of Integrals, Series, and Products (Academic, New York, 1980). T. Takenaka, M. Yokota, and O. Fukumitsu, Propagation

Vol. 18, No. 7 / July 2001 / J. Opt. Soc. Am. A

1617

22. 23. 24. 25. 26. 27. 28.

29. 30. 31. 32. 33. 34. 35.

of light beams beyond the paraxial approximation, J. Opt. Soc. Am. A 2, 826829 (1985). C. Brosseau, Fundamentals of Polarized Light (Wiley, New York, 1998). J. D. Lawrence, A Catalog of Special Plane Curves (Dover, New York, 1972). Jacques Bernoulli was so fond of this curve that, in accord with his desire, a spiral was engraved on his tombstone. C. J. R. Sheppard and T. Wilson, Gaussian-beam theory of lenses with annular aperture, Microwaves Opt. Acoust. 2, 105112 (1978). F. Gori, G. Guattari, and C. Padovani, BesselGauss beams, Opt. Commun. 64, 491495 (1987). R. H. Jordan, D. G. Hall, O. King, G. Wicks, and S. Rishton, Lasing behavior of circular grating surface-emitting semiconductor lasers, J. Opt. Soc. Am. B 14, 449453 (1997). C. Olson, P. L. Greene, G. W. Wicks, and D. G. Hall, Highorder azimuthal spatial modes of concentric-circle-grating surface-emitting semiconductor lasers, Appl. Phys. Lett. 72, 12841268 (1998).