Malacara OSA2013

1814
J. Opt. Soc. Am. A / Vol. 30, No. 9 / September 2013
Valencia Estrada et al.
Explicit representations of all refractive optical

interfaces without spherical aberration
Juan Camilo Valencia Estrada,1,,* lvaro Hernn Bedoya Calle,2, and Daniel Malacara Hernndez1,
1
2
Centro de Investigaciones en ptica, A.C., CIO., Col. Lomas del Campestre., Len de los Aldama, GTO 37150, Mexico
Centro de Innovacin, Investigacin y Desarrollo en Ingeniera y Tecnologa.CIIDITUANL., Av. Alianza 101 Sur, Km 10
de la Nueva Carretera Internacional de Monterrey, PIIT Monterrey, Apodaca, NL 66600, Mexico
*Corresponding author: camilo@cio.mx
Received May 7, 2013; revised July 23, 2013; accepted July 24, 2013;
posted July 25, 2013 (Doc. ID 189249); published August 21, 2013
The following explicit model, valid for high aperture refraction with homogenous and isotropic materials, encompasses all explicit solutions of the first-order nonlinear differential equation representing the perfect imageforming process of any axial object point into its axial image point. Solutions include well-known cases, such
as flats, spheres, prolate ellipsoids, prolate hyperboloids, and other sections of nondegenerate Cartesian ovals
of revolution, now classified according to the recurrent explicit solution introduced herein. We also present some
series expansions, given in cylindrical coordinates zr, for more efficient computation. Explicit solutions allow
accurate and expedite thickness calculation as compared to the regular series, parametric, or implicit solutions
commonly used. The results of this study are useful in the design of centered optical systems that are perfectly
aligned. 2013 Optical Society of America
OCIS codes: (080.0080) Geometric optics; (080.1010) Aberrations (global); (080.2720) Mathematical
methods (general); (080.2740) Geometric optical design; (220.1250) Aspherics; (080.4225) Nonspherical lens
design.
http://dx.doi.org/10.1364/JOSAA.30.001814
q
q
1 cos2 i n 1 cos2 r ;
1. INTRODUCTION
The classical centered refractive optical system is built with
an emphasis on light refraction at spherical and /or aspherical
surfaces of revolution. The mathematical and physical models
for refractive ray tracing on spherical surfaces show a blurred
image of an object point on an axis (Fig. 1), because spherical
interfaces are far from being optimum to ensure a perfect image. This phenomenon is well-known as spherical aberration
and presents multiple forms depending on the physical,
chemical, and geometrical characteristics of the interface.
This situation was studied initially by Descartes [1], and reproduced in books by Hawking [2], Luneburg [3], Born and Wolf
[4], D. Malacara-Hernndez and Z. Malacara-Hernndez [5],
Winston et al. [6], Dijkterhuis [7], and Ghatak [8] for some
of the cases included in this study. The conventional and
elegant approach to deal with this problem is by employing
Fermats principle; however, we do it in conjunction with
Snells law.
and squaring both sides,

1 cos2 i n2 1 cos2 r :
sin i n sin r ;
(1)
where i and r are the angles of incidence and refraction,

respectively, which can also be expressed as
1084-7529/13/091814-11$15.00/0
(3)
Using the scalar vector product

cos i
an
an
and
cos r
n b
n b

;
nb nb
(4)
wherein the vector a is a vector in the opposite direction of the

incident light ray with norm a, the vector b is a vector in the
direction of the light ray refracted with norm b, and n is a
normal vector to the refracting surface with norm n, as
shown in Fig. 2.
Substituting Eq. (4) in Eq. (3),
2. PERFECT IMAGE
If we assume a refractive interface with relative refractive index n at the wavelength of yellow light by convention, where
n nD , corresponding to the sodium spectral D line (average
of Fraunhofer doublet D1 and D2 lines, 589.29 nm),
Snells law is
(2)

1

an 2
n b 2
n2 1
an
nb
(5)
and simplifying and organizing terms,

b2 a2 n2 a n2
n2 ;
a2 b2 n2 b n2
where the vectors involved are
2013 Optical Society of America
(6)
Vol. 30, No. 9 / September 2013 / J. Opt. Soc. Am. A
r
10
-20
-10
Air
10
30
20
IMAGE
(0, z1)
50
40
60
70
80
90
110
100
120
40
PMMA
-10
-n
z1
b 1; mr ;
and
n
20

dz
; 1 ;
n
dr
z(r)
20
Fig. 1. Refraction at a spherical interface with R 10, z0 25, and

n 1.4875 for an airPMMA interface.
a 1; mi ;
1815
10
10
20
(7)
-z0
20
OBJECT
40
so,
(0, z0)

a n2 mi
a2 1 m2i ;
dz
dr
2
;

dz 2
2
b n mr
;
dr
b2 1 m2r ; and n2 1
(8)
dz2
dr
Fig. 2. Vector representation of refraction, on the meridional section

of an optical surface of revolution.

q1
dz
n n mr m2r 1 :
dr
(9)
where mi and mr are the slopes of the incident and refracted
rays, respectively, and dzdr is obtained from the meridional
section of the optical interface of revolution.
Substituting Eqs. (7)(9) into Eq. (6), we have
(14)
When the refracted rays are parallel, with mr ,

2
2
dz
dz
2
2
mi 1 n mi 1
;
dr
dr
(15)
solving for dzdr,

2
1 m2r 1 m2i 1 dz
dr mi
2
1 m2i 1 m2r 1 dz
dr mr
2
dz
dr
2
dz
dr
n2 ;

q1
dz
mi n m2i 1 :
dr
(10)
(16)
and simplifying and rearranging terms,

dz 2
dz 2
1 m2r 1 mi
n2 1 m2i 1 mr
;
dr
dr
(11)
solving for
dz

dr
q
mi 1 m2r n2 mr 1 m2i nmr mi 1 m2r 1 m2i
n2 m2r 1 m2i m2i 1 m2r
The above equation has many applications in two special

situations:
When the incident rays are parallel, with mi in
Eq. (12),
m2r

2
2
dz
dz
1 ;
1
n2 mr
dr
dr
solving for dzdr,
It is assumed that from an object point P 0 r 0 ; z0 , multiple

rays emerge that will reach the section of the surface of
revolution z f r at arbitrary points P r r; f r, where they
are refracted in particular directions, with the provision that
all refracted rays converge toward a single-image point
P 1 r 1 ; z1 for perfect imaging; thus,
(13)
mi
z z0
r r0
(12)
and mr
z z1
:
r r1
(17)
Substituting Eq. (17) into Eq. (11) and simplifying, we have

2
2
2
r r 0 z z0 dz
dr r r 1 z z1
n2 ;
dz 2
2
r r 1 z z1 dr r r 0 z z0 2
(18)
and solving for dzdr, we obtain the first-order nonlinear differential equation ensuring a perfect image on the optical axis
1816

dz

dr
2
2
2
2
2
r pr
0 z z0 r r 1 z z1 n r r 1 z z1 r r 0 z z0

n r r 1 2 z z1 2 r r 0 2 z z0 2 r r 0 z z1 r r 1 z z0 2
n2 z z1 2 r r 0 2 z z0 2 z z0 2 r r 1 2 z z1 2
The implicit solution of Eq. (19) when r 0 r 1 0 (axial

points), is well-known
Equation (20) is the Fermats principle interpretation of perfect point-to-point imaging (both on the optical axis), meaning
that perfect conjugate imaging occurs when all rays passing
through these conjugate points have the same optical path
length.
The result (19) has applications in the following special
situations:
When the object is far away (z0 ), rays arrive parallel to the optical axis. Substituting Eq. (17) into Eq. (14),
s

1

dz
z z1
1
z z1 2

1
lim
z0 dr
n
r r1
r r1
nr r 1
p :
nz z1 z z1 2 r r 1 2
(21)
(19)
if f < 0. In this case, r 0 r 1 0, z0 , and z1 f , so

dz
r
p :

dr f z 1n z f 2 r 2
q
q
r r 0 2 z z0 2 n r r 1 2 z z1 2 Constant:
(20)
(25)
This equation has four mathematical real solutions for z (two

valid for real materials), with the boundary condition
z0 0:
p
nn 1f Sign f n 1n2 n 1f 2 n 1r 2
z
:
n2 1
(26)
Performing ray tracing, it can be verified that the two valid
solutions for real materials are
p
nn 1f Signf n 1n 1f 2 n 1r 2
; (27)
z
n2 1
corresponding to ellipsoids of revolution with meridional
section
q
z A A 1 rB2 ;
If the refracted rays will converge to the focal point

P 1 r 1 ; z1 P 1 0; f , we have
(28)
and with semi-axes

dz
r
p :

dr f z 1n z f 2 r 2
(22)
When the object is at the front focal point, refracted rays

emerge from the optical interface parallel to the optic axis
(z1 ). Substituting Eq. (17) into Eq. (16),
s

1
dz
z z0
z z0 2

n 1
lim
z1 dr
r r0
r r0
r r0
p
:
z z0 n z z0 2 r r 0 2
A f nn 1 and B f
p p
n 1 n 1:
(29)
To compare the geometry of these interfaces with the geometry of a spherical optical interface, we have the following
analysis.
Expanding Eq. (26) by Taylor series around the origin,

2

n
r
nn1 r 4
nn12 2r 6

Or 8 ;

n 1 2f
n 12 8f 3
n 13 32f 5

z
(30)
(23)
If the rays start from the object space focal point

P 0 r 0 ; z0 P 0 0; f , we have
where the direction of the surface is defined by the sign of f ,

since all the coefficients are odd functions of f . The first sign
(upper sign) corresponds to the valid solutions. Comparing
the above expansion (30) with spherical surfaces for all cases,
dz
r
p :

dr f z n z f 2 r 2
p
z R SignR R2 r 2 ;
(24)
3. ANALYTICAL EXPLICT SOLUTIONS

A. First Case: When r0 r1 0, z0 , and z1 f in
Eq. (19)
When the object is distant and the incident rays are parallel to
the optic axis, the refracted image rays converge into a single
real focus if f > 0. However, rays diverge from a virtual focus
(31)
where R n 1f , and f is the paraxial focal length of the

spherical interface if the incident or refracted rays are paraxial. So
q
z n 1f Signf n 1f 2 r 2 :
An example of this solution is shown in Fig. 3.
(32)
1817
b
10
50
-30
40
-20
10
-10
30
20
30
-10
f
20
-20
10
-30
-f
-30
10
-20
20
-40
-10
-50
Fig. 3. Refraction by an elliptical interface. n 1.5. Convex case (a) ray tracing for f 50. In this case the expected solution is obtained, where
the rays match the estimated focus, which corresponds to the focus of the ellipse. Concave case (b) ray tracing for f 50. Phantom or discontinue
ray tracing is included to show the virtual focal point.
4
6
2

1
r
1
r
1
2r

Or 8 ;

n 1 2f
n 15 32f 5
n 13 8f 3
(33)

z
which can be compared with the formula (30), showing the

difference between their coefficients.
B. Second Case: When r0 r1 0, z0 f , and z1
in Eq. (19)
In this case, the object is near and the image is formed at
infinity. f is real if f < 0 and virtual if f > 0. In this case
r 0 r 1 0, z0 f , and z1 , so
dz
r
p :

dr f z n z f 2 r 2
4
6

1 r 2
1
r
1
2r

Or 8 :

n 1 2f
n 15 32f 5
n 13 8f 3
(39)
Expanding Eq. (31) in Taylor series, we have
(34)
This result has four real solutions for z, (two valid for
real materials with n > 0), with the boundary condition
z0 0:
z
Figure 4 shows an example of this case, using Eq. (36).

C. Third Case: When r0 r1 0, z0 f , and z1 f
in Eq. (19)
Assuming f is positive, this is the case in which the incident
rays from an axial object point with negative object distance
f are refracted to form an image point with positive axial
focal distance f . In this case, r 0 r 1 0, z0 f , and
z1 f . In this document f is used to indicate the absolute
vertex-image distance. Substituting these values in Eq. (19),
we have
z
b
p
n1f Signf n1n1f 2 n 1r 2
z
: (35)
n2 1
40
40
20
20
f
Performing ray tracing, it was verified that the two valid
solutions for natural materials are given by
p
n 1f Signf n 1n 1f 2 n 1r 2
z
; (36)
n2 1
corresponding to hyperboloids of revolution, with the
meridional section
q
z A A 1 rB2 ;
(37)
and with semi-axes

A f n 1 and B f
p p
n 1 n 1:
Expanding Eq. (36) about the origin, we have
(38)
-20
-10
10
20
-20
-10
10
20
-20
-20
-f
-40
-40
Fig. 4. Refraction by a hyperbolical interface. n 1.5. Convex case

(a) ray tracing for f 50. In this case, the expected solution is obtained, where the rays match the estimated focus, which corresponds
to the focus of the complementary branch of the hyperbola. Concave
case (b) ray tracing for f 50. Phantom tracing is included to show
the virtual focal point.
1818

dz

dr

r 3 z p
f
rz f z f 2 n2 rz f r 2 z f 2
2f nr r 2 z f 2 r 2 z f 2
:
n2 z f 2 r 2 z f 2 z f 2 r 2 z f 2
(40)
The solution obtained, expanded in power series, is [9]

z c2
r2
r4
2r 6
5r 8
2r 10

c

c
c4 3 c6
8
10
2f
32f 5
128f 7
8f
512f 9
c12
X
k1
An example of this power series solution is shown in Fig. 5. In

Table 1 are presented the best simplified approximations
using one, two, and three terms of the expansion (41) using
Eq. (43).
Also, the solution (41) using Eq. (43) corresponds to the
series expansion of a natural Cartesian oval section with
the exact complex function, explicit and recurrent, obtained
just by means of Fermats principle:
14r 12
12r 14
3r 16
c14
c16
Or 18
11
13
2048f
32768f 15
8192f
c2k
I k r 2k
;
2f 2k1
z
(41)
f W n2 1
;
n2 1
(44)
where the recurrent variables are
where
n1
;
n1
n1
;
c4
n1
n 1n2 6n 1
c6
;
n13
n1
c8
;
n1
n 17n4 124n3 122n2 124n 7
c10
;
n15
c2
c12
n 13n4 44n3 46n2 44n 3

;
n15
c14
11n7 545n6 1371n5 2169n4 2169n3 1371n2 545n 11

;
n17
c16
n 1143n6 8794n5 18335n4 27948n3 18335n2 8794n 143

:
n17
(42)
When checking by means of exact ray tracing with both signs, it was found that the positive sign solution is valid for natural
materials, and the negative sign solution is valid for metamaterials with negative refractive index. Therefore,
n1
;
n1
n1
;
c4
n1
n 1n2 6n 1
;
c6
n 13
n1
c8
;
n1
n 17n4 124n3 122n2 124n 7
;
c10
n 15
c2
c12
n 13n4 44n3 46n2 44n 3

;
n 15
c14
11n7 545n6 1371n5 2169n4 2169n3 1371n2 545n 11

;
n 17
c16
n 1143n6 8794n5 18335n4 27948n3 18335n2 8794n 143

:
n 17
(43)
z
z
20
40
40
40
40
40
20
20
20
20
20
10
1819
10
20
20
20
10
10
20
20
20
10
10
20
20
10
10
20
20
20
20
10
10
20
20
f
40
40
40
40
40
Fig. 5. Natural Cartesian ovals approximations. Solution in series with the first sign of the solution of the third case. Ray tracing for n 1.5 and
f 50. Refraction is calculated using 15 terms in the series, (a)(e), respectively. It is clearly seen that using three terms, the residual aberration is
quite small.
Table 1. Natural Cartesian Oval Approximations

Terms
1
2
3
zr Function
n 1r 2 2f n 1
n 14f 2 r 2 r 2 8f 3 n 1
n 18f 4 2f 2 r 2 n 12 r 2 n2 6n 1r 6 16f 5 n 13
A n 1;
B
function, explicit and recurrent, obtained just by means of

Fermats principle:
24nA2 ;
C 5 n2 5nB;
z
D 3n 1;
E 41 n2D n2 n 2 rf n2 12 ;
G 6B2 EF C;

p 13
;
H Re G G2 F 3
D. Fourth Case: When r0 r1 0, z0 f , and

z1 f in Eq. (19)
In this case, the incident rays come from an object point on
the optical axis with negative object distance
I
z0 f ;
(45)
Equation (45) is much more accurate than the series expansion (41) using Eq. (43) itself.
When the incident rays, coming from a virtual object point
on the optical axis with a positive object distance f , are refracted to diverge from a virtual focal point with a negative
axial distance f , i.e., with r 0 r 1 0, z0 f , and z1 f
in Eq. (19), we obtain a valid solution that obeys Eq. (41) with
n1
n1
;
c4
;
n1
n1
n 1n2 6n 1
;
c6
n 13
(47)
with same recurrent variables A, B, C, D, E, F, G, I, and W

used with Eq. (45).
F E 2 2C;
p
E 2H3;
1 p
W 2E H 2BI3 I;
2
p p
p
jrj f n 1A D D A2n2 :
f W n2 1
;
n2 1
c2
(48)
where the factor is a levered ratio. and f are positive definite, and they are refracted to converge to an image point on
the optical axis with a positive axial distance z1 f
[r 0 r 1 0 in Eq. (19)]. The differential Eq. (19) may be
simplified:
dz
dr
r 3 z f rz f z f 2 n2 rz f r 2 z f 2

p
f nr 1 r 2 z f 2 r 2 z f 2

n2 z f 2 r 2 z f 2 z f 2 r 2 z f 2
(46)
(49)
corresponding to the series expansion of the same natural

Cartesian oval section (virtualvirtual), with the complex
with solutions expressed in power series, according to

Eq. (41) as
1820
c2
n 1
;
n1
c4
3 n2 3 22 2 1n 1
;
3 n12
5 n3 25 34 33 2 3 1n2 5 34 3 32 3 2n 1

;
5 n13
1
0
7 n4 37 46 45 24 23 42 4 1n3
C
B
1
C:
B 37 86 84 83 8 3n2
c8 7
A
@
4
n1
7
6
5
4
3
2
4 4 2 2 4 4 3n 1
c6
(50)
When checking these results by means of exact ray tracing with both signs, it was determined that the solution to the first sign is
valid for natural materials.
Thus, the coefficients of the solution of the differential equation that ensures a perfect image are
c2
n 1
;
n 1
c4
3 n2 3 22 2 1n 1
;
3 n 12
5 n3 25 34 33 2 3 1n2 5 34 3 32 3 2n 1

;
5 n 13
1
0
7 n4 37 46 45 24 23 42 4 1n3
C
B
1
C:
B 37 86 84 83 8 3n2
c8 7
A
n 14 @
7
6
5
4
3
2
4 4 2 2 4 4 3n 1
c6
(51)
Several examples of this power series solution are shown in Fig. 6.
A n ;
The solution of Eq. (41) using Eq. (51) corresponds to a

Cartesian oval section with the exact complex function,
explicit and recurrent, obtained just by means of Fermats
principle:
B 121 nA2 ;
C 22 n2 n n n 12 1 B;
D 2 n;
E 4n4 2n3 22 1n2
z
f W n2
;
n2 1
2n 2 rf n2 12 ;

p p
p
jrj f n 1A D D n 11 n2 ;
(52)
following with F, G, H, I, and W according to the same

definitions given in Eq. (45).
where the recurrent variables are
20
40
40
40
40
40
20
20
20
20
20
10
10
20
0
20
(53)
20
10
10
20
20
20
10
10
20
20
20
10
10
20
20
20
10
10
20
20
Fig. 6. Cartesian ovals approximations. Solution in series with the first sign of the solution of the fourth case. Ray tracing for n 1.5 and f 50.
Refraction is observed with object distance z0 f with 50, 10, 2, 0.7, and 0.6 [(a)(e), respectively], using the first four summands of the
series. When < 1, more summands must be applied to reach a more accurate solution.
When the leverage ratio , the oval degenerates into

an ellipse, and when 1, it degenerates into a natural oval.
Similarly, the divergent refraction occurs when the incident
rays are coming from an object point on the optical axis, with
c2
n 1
;
n 1
c4
3 n2 3 22 2 1n 1
;
3 n 12
1821
negative object distance z0 f (both and f are as defined

positive), with r 0 r 1 0; the refracted rays form a virtual
focus image distance z1 f . Substituting these values into
the differential Eq. (19), and solving it, we get
5 n3 25 34 33 2 3 1n2 5 34 3 32 3 2n 1

;
5 n 13
1
0
7 n4 37 46 45 24 23 42 4 1n3
C
B
1
7
6
4
3
2
C.
B
c8 7
A
4 @ 3 8 8 8 8 3n
n 1
7 46 45 24 23 42 4 3n 1
c6
(54)
An example of this power series solution is shown in Fig. 7.

The series expansion solution (41) applying Eq. (54) corresponds to a Cartesian oval section with the exact complex function,
explicit and recurrent, obtained just by means of Fermats principle:
z
f ReW n2
n2 1
n 1;
r; If jImW j < error 0;
(55)
where
A n ;
B 121 nA2 ;
C 22 n2 n n n 12 1 B;
nf
;
n1
E 4n4 2n3 22 1n2
D
2n 2 n2 1rf 2 ;

F E2 2C;
G 6B2 EF C;
p 13
J G G2 H 3 ;
H FJ J2;
p
I E 2H3;
p
1
W I Sign n 2E H 2BI3;
2
p
If n z D SignD D2 r 2 ; jrj jDj:
z
20
20
10
20
20
10
20
20
10
20
20
10
20
20
10
20
(56)
z
20
10
20
20
10
20
20
10
20
20
10
10
20
20
f
40
40
40
40
40
60
60
60
60
60
Fig. 7. Cartesian ovals approximations. Solution in series with the second sign of the solution of the fourth case. Ray tracing for n 1.5 and
f 50. Refraction is observed with an object distance z0 f with 50, 2, 1, 0.7, and 0.5, [(a)(e), respectively], using the first four terms in the
series. Note the virtual ray tracing showing the f point position. When < 1, more summands must be used to obtain a more accurate solution.
1822
E. Fifth Case: When r0 r1 0, z0 f , and z1 f in

Eq. (19)
In this case, the incident rays come from an object point
on the optical axis (r 0 0), with a positive virtual object
distance
z0 f ;
complex function, explicit and recurrent, obtained just by

means of Fermats principle:
z
f W n2
;
n2 1
(61)
(57)
with factor or leverage ratio ( and f are positive definite),

and refract at the surface diverging from a virtual image
point on the optical axis (r 1 0) with negative axial
distance z1 f . Evaluating the differential Eq. (19), we can
obtain

dz

dr
with A, B, C, D, and E, and domain according to Eq. (53)

and then evaluating F, G, H, I, and W according to
Eq. (45).
Similarly, the divergent refraction occurs when the
incident rays from a virtual object point at a positive
r 3 z f p
rz
f z f 2 n2 rz f r 2 z f 2
f nr 1 r 2 z f 2 r 2 z f 2
n2 z f 2 r 2 z f 2 z f 2 r 2 z f 2

;
(58)
with solutions obtained in power series, according with Eq. (40), where now
c2
n1
;
n 1
c4
3 n2 3 22 2 1n 1
;
3 n 12
5 n3 25 34 33 2 3 1n2 5 34 3 32 3 2n1

;
5 n 13
1
0
7 n4 37 46 45 24 23 42 4 1n3
C
B
1
7
6
4
3
2
C:
B
c8 7
A
4 @ 3 8 8 8 8 3n
n 1
7 46 45 24 23 42 4 3n 1
c6
(59)
When checking by means of exact ray tracing with both signs, it was determined that the solution corresponding to the
second sign is valid for natural materials, and is the same obtained before Eq. (51) but with commuted sign:
c2
n 1
;
n 1
c4
3 n2 3 22 2 1n 1
;
3 n 12
5 n3 25 34 33 2 3 1n2 5 34 3 32 3 2n 1

;
5 n 13
1
0
7 n4 37 46 45 24 23 42 4 1n3
C
B
1
C:
B 37 86 84 83 8 3n2
c8 7
A
n 14 @
7 46 45 24 23 42 4 3n 1
c6
An example of this power series solution is shown in Fig. 8. It

can be readily seen that this geometry is symmetrical to that of
the first solution in the fourth case (Fig. 5).
The solution (41) applying Eq. (60) corresponds to the
series expansion of a Cartesian oval section with the exact
(60)
axial distance z0 f , where and f are defined

positive, with r 0 r 1 0, and the refracted rays form
a real image point with image distance z1 f . Substituting
these values into the differential Eq. (19) and solving,
we get
1823
Fig. 8. Cartesian ovals approximations. Solution in series with four terms, using the second sign of the solution of the fifth case with a virtual
image. Exact ray tracing for n 1.5 and f 50. Refraction is observed with a virtual object distance z0 f , 1, 0.9, 0.75, and 0.6, respectively,
(a)(d).
c2
n 1
;
n 1
c4
3 n2 3 22 2 1n 1
;
3 n 12
5 n3 25 34 33 2 3 1n2 5 34 3 32 3 2n 1

;
5 n 13
1
0
7 n4 37 46 45 24 23 42 4 1n3
C
B
1
C:
B 37 86 84 83 8 3n2
c8 7
A
n 14 @
7 46 45 24 23 42 4 3n 1
c6
(62)
When checking by means of exact ray tracing with both signs, it was determined that the solution to the second sign is valid for
natural materials, so
c2
n 1
;
n 1
c4
3 n2 3 22 2 1n 1
;
3 n 12
5 n3 25 34 33 2 3 1n2 5 34 3 32 3 2n 1

;
5 n 13
1
0
7 n4 37 46 45 24 23 42 4 1n3
C
B
1
C:
B 37 86 84 83 8 3n2
c8 7
A
@
4
n 1
7
6
5
4
3
2
4 4 2 2 4 4 3n 1
c6
(63)
Equation (63) is the same as Eq. (54) but with commuted sign. An example of this power series solution is shown in Fig. 9.
Therefore the solution (41) using Eq. (63) correspond to the

series expansion of a Cartesian oval section with the complex
function, explicit and recurrent, obtained just by means of
Fermats principle:
f ReW n2
n 1;
n2 1
r; If jImW j < error 0
z
(64)
with recurrent variables A, B, C, D nf n 1, E, F, G, J,

H, I, and W according to Eq. (56).
F. Other Cases: When r0 r1 0 and z0 z1 f in
Eq. (19)
Other cases correspond to classical degenerate solutions, the
well-known flat surfaces when the object and image are at
infinity and there is no distortion, and spherical surfaces when
the point object matches the image point on its center of
1824
Fig. 9. Cartesian ovals approximations. Solution in series with four terms using the second sign of the solution of the fifth case. Exact ray tracing
for n 1.5 and f 50. Refraction is observed with a virtual object distance z0 f with 5, 1, 0.6, and 0.5, respectively, (a)(d).
curvature on the optical axis. When the incident rays come

from an axial point object with object distance negative f ,
and refract to a focal point with the same focal distance negative f , the focus is virtual with r 0 r 1 0, z0 f , and
z1 f in Eq. (19):
dz
rf z
dr
(65)
with a valid solution that corresponds to a circular section of

radius f , and center (0, f )
z f
q
f 2 r2 ;
(66)
If the incident rays come from a virtual object point on the

optical axis with a positive object distance f , and are refracted
in a real focal point with positive axial distance f , i.e.,
with r 0 r 1 0, z0 f , and z1 f , Eq. (19) yields a valid
solution corresponding to a circular section of radius f ,
centered at (0, f )
zf
q
f 2 r2 ;
(67)
4. CONCLUSIONS
This study allows us to design and manufacture oval lenses,
without the use of approximate series, parametric (Michaelis
et al. [10]), or implicit functions, quite popular in the optical
industry, especially in lighting systems design, minimizing calculations by means of the explicit functions given here.
Explicit solutions zr allow us to calculate thicknesses more
easily than series, parametric, or implicit solutions. These solutions enable the fast design of perfect optical systems, minimizing the number of lenses required to attain high-quality
images and definition, which is of vital importance in the
nanotechnology industry.
All possible explicit surfaces found have the potential to
generate a new family of aspherical oval lenses with several
designs, currently submitted in international patent processes
[11]. All described explicit geometries can be easily machined
by means of CNC turning or grinding in Cartesian coordinates
x; z; C with nanometer resolution using diamond tools to

maximize surface finish quality and minimize the subsequent
polishing process. It is also possible to design and build machines in polar coordinates r; ; ; C to make the refractive
surfaces proposed herein.
ACKNOWLEDGMENTS
The authors acknowledge Centro de Investigaciones en
ptica, AC (CIO), Leon, Guanajuato, Mexico; Universidad
Autnoma de Nuevo Leon (UANL); and National Council of
Science and Technology of Mexico (CONACYT) for their
financial support.
J. C. V. E. and A. H. B. C. developed the physical model and

mathematical programming. D.M.H. reviewed and verified all
results. All the authors contributed equally in writing this
article.
REFERENCES AND NOTE

1. R. Descartes, Discours de la mthode pour bien conduire sa
raison, et chercher la vrit dans les sciences (Ian Maire, 1637).
2. S. Hawking, God Created Integers (Editorial Crtica, 2005).
3. R. K. Luneburg, Final correction of optical instruments by
aspheric surfaces, in Mathematical Theory of Optics, classroom notes (Brown University, 1944), Chap. 24, pp. 139151.
4. M. Born and E. Wolf, Principles of Optics, 6th ed. (Pergamon,
1959), Chap. 4.1, pp. 197200.
5. D. Malacara-Hernndez and Z. Malacara-Hernndez, Handbook
of Optical Design, Centro de Investigaciones en ptica
(Dekker, 1994).
6. R. Winston, J. Miano, and P. Bentez, Nonimaging Optics
(Academic, 2005).
7. F. J. Dijkterhuis, Lenses and Waves, Christiaan Huygens and
the Mathematical Science of Optics in the Seventeenth Century
(Springer, 2005), pp. 1216.
8. P. Ghatak, Optics, 4th ed. (McGraw-Hill, 2009).
9. The word focus is generally used to indicate geometrical optics real or virtual points where light rays converge or diverge as
the image of an object at infinity. In this document it will also be
used to indicate the position of the image, regardless of whether
the object is located at an infinite or finite distance. Therefore, it
is equivalent to declaring the focal length f as the image distance measured from the origin.
10. D. Michaelis, P. Schreiber, and A. Bruer, Cartesian oval
representation of freeform optics in illumination systems,
Opt. Lett. 36, 918920 (2011).
11. J. Valencia and A. Bedoya, Lentes asfricas ovales, Mexican
patent application. Instituto mexicano de la propiedad industrial
(IMPI). MX/a/2012/010025 (August 30, 2012).

Malacara OSA2013

Cargado por

Información del documento

Derechos de autor

Formatos disponibles

Compartir este documento

Compartir o incrustar documentos

Opciones para compartir

¿Le pareció útil este documento?

¿Este contenido es inapropiado?

Copyright:

Formatos disponibles

Malacara OSA2013

Cargado por

Copyright:

Formatos disponibles

1814

J. Opt. Soc. Am. A / Vol. 30, No. 9 / September 2013

Valencia Estrada et al.

Explicit representations of all refractive optical

and squaring both sides,

where i and r are the angles of incidence and refraction,

Using the scalar vector product

wherein the vector a is a vector in the opposite direction of the

and simplifying and organizing terms,

Valencia Estrada et al.

Vol. 30, No. 9 / September 2013 / J. Opt. Soc. Am. A

Fig. 1. Refraction at a spherical interface with R 10, z0 25, and

Fig. 2. Vector representation of refraction, on the meridional section

When the refracted rays are parallel, with mr ,

solving for dzdr,

and simplifying and rearranging terms,

The above equation has many applications in two special

solving for dzdr,

It is assumed that from an object point P 0 r 0 ; z0 , multiple

Substituting Eq. (17) into Eq. (11) and simplifying, we have

J. Opt. Soc. Am. A / Vol. 30, No. 9 / September 2013

Valencia Estrada et al.

The implicit solution of Eq. (19) when r 0 r 1 0 (axial

if f < 0. In this case, r 0 r 1 0, z0 , and z1 f , so

This equation has four mathematical real solutions for z (two

If the refracted rays will converge to the focal point

and with semi-axes

When the object is at the front focal point, refracted rays

If the rays start from the object space focal point

where the direction of the surface is defined by the sign of f ,

3. ANALYTICAL EXPLICT SOLUTIONS

where R n 1f , and f is the paraxial focal length of the

Valencia Estrada et al.

Vol. 30, No. 9 / September 2013 / J. Opt. Soc. Am. A

which can be compared with the formula (30), showing the

Expanding Eq. (31) in Taylor series, we have

Figure 4 shows an example of this case, using Eq. (36).

and with semi-axes

Expanding Eq. (36) about the origin, we have

Fig. 4. Refraction by a hyperbolical interface. n 1.5. Convex case

J. Opt. Soc. Am. A / Vol. 30, No. 9 / September 2013

Valencia Estrada et al.

The solution obtained, expanded in power series, is [9]

An example of this power series solution is shown in Fig. 5. In

where the recurrent variables are

n  13n4 44n3 46n2 44n  3

11n7  545n6  1371n5  2169n4  2169n3  1371n2  545n  11

n  1143n6 8794n5 18335n4 27948n3 18335n2 8794n  143

n  13n4 44n3 46n2 44n  3

11n7  545n6  1371n5  2169n4  2169n3  1371n2  545n  11

n  1143n6 8794n5 18335n4 27948n3 18335n2 8794n  143

Valencia Estrada et al.

Vol. 30, No. 9 / September 2013 / J. Opt. Soc. Am. A

Table 1. Natural Cartesian Oval Approximations

function, explicit and recurrent, obtained just by means of

D. Fourth Case: When r0 r1 0, z0 f , and

with same recurrent variables A, B, C, D, E, F, G, I, and W

r 3 z  f  rz  f z f 2 n2 rz f r 2  z  f 2

corresponding to the series expansion of the same natural

with solutions expressed in power series, according to

J. Opt. Soc. Am. A / Vol. 30, No. 9 / September 2013

Valencia Estrada et al.

5 n3  25  34 33  2  3  1n2  5  34  3 32  3  2n  1

5 n3  25  34 33  2  3  1n2  5  34  3 32  3  2n  1

n 13n4 44n3 46n2 44n 3

11n7 545n6 1371n5 2169n4 2169n3 1371n2 545n 11

n 1143n6 8794n5 18335n4 27948n3 18335n2 8794n 143

n 13n4 44n3 46n2 44n 3

11n7 545n6 1371n5 2169n4 2169n3 1371n2 545n 11

n 1143n6 8794n5 18335n4 27948n3 18335n2 8794n 143

r 3 z f rz f z f 2 n2 rz f r 2 z f 2

5 n3 25 34 33 2 3 1n2 5 34 3 32 3 2n 1

5 n3 25 34 33 2 3 1n2 5 34 3 32 3 2n 1

2n 2 rf n2 12 ;

5 n3 25 34 33 2 3 1n2 5 34 3 32 3 2n 1

r; If jImW j < error 0;

2n 2 n2 1rf 2 ;

5 n3 25 34 33 2 3 1n2 5 34 3 32 3 2n1

5 n3 25 34 33 2 3 1n2 5 34 3 32 3 2n 1

5 n3 25 34 33 2 3 1n2 5 34 3 32 3 2n 1

5 n3 25 34 33 2 3 1n2 5 34 3 32 3 2n 1

with recurrent variables A, B, C, D nf n 1, E, F, G, J,