Chromatic analysis and design of a first-order radial GRIN lens

Joseph N. Mait; Guy Beadie; Predrag Milojkovic; Richard A. Flynn

doi:10.1364/OE.23.022069

1. Introduction

For over 100 years, researchers have recognized the potential offered by graded-index (GRIN) optical elements to improve the performance of optical systems [1]. The technology received considerable attention in the 1970s [2, 3], when it found its primary application in fiber-based imaging systems. Fabrication of GRIN elements for such systems relied on diffusing salts into glass [4]. However, more recently, researchers have fabricated GRIN elements by layering nanometer-thick polymer films of differing indices to produce a material with a desired effective index [5]. Other techniques newly introduced for creating arbitrary three-dimensional distributions of refractive index rely upon rapid prototyping technology [6] and upon semiconductor fabrication [7, 8]. Spurred by these developments, interest in applying GRIN optics to conventional, i.e., non-fiber-based, imaging has been rekindled.

Although the ability to create an arbitrary index profile in space is the most alluring prospect of GRIN optics, a practical, near-term consideration is their utility as achromatic elements. Achromatic design using GRIN has been addressed in the literature [9–18]. But, except for a few cases [19], the dispersive and achromatic properties of a GRIN element have not been presented in a manner useful for optical designers. There exist few rules of thumb or intuition with regard to such design. Such rules would have benefited a recent survey of materials suitable for achromatic design [20].

The first attempt, known to us, to derive formulae for first-order chromatic aberrations of GRIN lenses was performed by Sands [11,13]. His starting point was the analysis by Buchdahl in his seminal work [9]. Starting from the notion of the optical quasi-invariant and its change with wavelength, Sands derived formulae for the first-order longitudinal and lateral chromatic aberrations of gradient-index lenses [11] and applied them to selected examples of axial and radial index distributions [13]. His formulae account for both surface and transfer chromatic aberration components of longitudinal and lateral color. Surface terms are not unlike those for a homogeneous lens, but transfer terms are unique. They do not exist in homogeneous lenses and represent a contribution from gradient dispersion. Others [15, 17] subsequently found that two formulae for transfer aberrations contained errors. These are correct in [15].

Unfortunately, the Buchdahl-Sands approach is not easily understood. It does not provide much intuition about aberrations and their physical meaning. Furthermore, the approach is not based on the materials that create the gradient index. Instead, the index function is described as a purely mathematical object.

The goal of our work is to address some of these deficiencies. To do so, we assume the paraxial approximation is valid. This allows us to use a ray optical model and its Gaussian coefficients (the ABCD-matrix elements) to analyze aberrations. For a GRIN lens, the Gaussian coefficients relate the optical properties of a GRIN singlet, in particular, its effective focal length, to its physical parameters, such as thickness, diameter, and refractive index.

The coefficients derived by Sands [12] and others [18] for a GRIN lens with a quadratic radial index distribution are the starting formulae for our analysis. From this, we derive the first-order chromatic aberration formulae, which allows us to analyze the change in optical power as a function of wavelength and provides us with a measure of longitudinal chromatic aberration. We then apply simplifying assumptions to elucidate the links between an element’s geometrical and material parameters and its optical performance. We leave an analysis of lateral color to a subsequent paper.

To the best of our knowledge, our approach to chromatic analysis of GRIN lenses is new. Our primary goal is to give optical designers a tool that provides them the intuition to generate a good initial design. We also wish to indicate how material properties affect design. Hopefully, this insight might lead to better GRIN designs, development of new GRIN materials, and may help novice designers.

In this paper, we consider the design of cylindrically symmetric GRIN lenses shown in Fig. 1 that have an index distribution n(r, z, λ) [15],

n (r, z, λ) = \sum_{ℓ = 0}^{L} \sum_{m = 0}^{M} Γ_{ℓ m} (λ) z^{ℓ} r^{2 m},

where

Γ_{ℓ m} (λ) = \sum_{k = 0}^{K} ν_{k ℓ m} {(λ - λ_{0})}^{k},

r is the radial distance from the optic axis, z is the distance from the origin along the optic axis, and λ₀ is the wavelength of operation. The coefficients ν_kℓm characterize the chromatic properties of a material. We refer to Γ_ℓm(λ) as the (ℓ, m)-order gradient generator since it is not a gradient per se. To make this point explicit, consider the radial and axial index gradients,

\frac{d n (r, z, λ)}{d r} = 2 \sum_{ℓ = 0}^{L} \sum_{m = 0}^{M} m Γ_{ℓ m} (λ) z^{ℓ} r^{2 m - 1},

\frac{d n (r, z, λ)}{d z} = \sum_{ℓ = 0}^{L} \sum_{m = 0}^{M} ℓ Γ_{ℓ m} (λ) z^{ℓ - 1} r^{2 m},

and the dispersion,

\frac{d n (r, z, λ)}{d λ} = \sum_{ℓ = 0}^{L} \sum_{m = 0}^{M} \frac{d Γ_{ℓ m} (λ)}{d λ} z^{ℓ} r^{2 m} .

Note that the gradient generators are fundamental to describing the refractive index in geometric terms and, therefore, provide a link to the GRIN’s optical properties. In the next section, we discuss in more detail Γ_ℓm(λ) for a first-order radial GRIN.

Fig. 1 GRIN lens.

Download Full Size | PDF

We assume the GRIN is designed to operate nominally at the wavelength λ₀. Unlike others who have examined chromatic behavior of GRIN elements [15], we choose to expand Eqs. (1) and (2) in terms of wavelength λ rather than the Buchdahl variable ω, a dimensionless quantity related to wavelength [9]. Our analysis is valid regardless of the variable we use and we felt a presentation in λ would be less confusing to optical designers.

In Sec. 2 we introduce the radial gradient lens and, in Sec. 3, we derive expressions for its power and chromatic aberration. In Sec. 4 we examine these optical properties for the special case of a Wood lens. The full expressions derived in Sec. 3 are cumbersome in their length, if not their complexity. To uncover the underlying relationships between material and optical properties, we make a series of simplifying approximations in Sec. 5. A key contribution of our analysis is the discussion in Sec. 6 on the material relationship that allows achromatic designs to work over broad bandwidths. We provide concluding remarks in Sec. 7.

2. First-order radial gradients

Equation (1) is sufficiently general that it can represent axial (Γ_ℓm = 0 for all m) and radial (Γ_ℓm = 0 for all ℓ) gradients. In this work we are interested in focusing elements and, since axial gradients do not add optical power, we consider only radial gradients. More specifically, we consider only first-order radial gradients. Thus, we rewrite Eq. (1),

n (r, λ) = n_{0} (λ) + Γ (λ) r^{2} .

Because we use only the (0, 1)-gradient generator in the expansion of the index, we drop the subscript on Γ(λ) to simplify notation.

With reference to Fig. 1, we assume the front and back surfaces of the lens are spherical with radii R_f and R_b. The maximum element thickness is t and its diameter is D. The lens focal length is f. By definition, the indices at the center and at the edge of the GRIN element, n_ctr(λ) and n_edg(λ), respectively, are

n (0, λ) = n_{ctr} (λ),

\begin{array}{l} n (D / 2, λ) & = & n_{edg} (λ), \\ = & n_{ctr} (λ) + Γ (λ) {(D / 2)}^{2}, \end{array}

from which we derive

Γ (λ) = {(\frac{2}{D})}^{2} [n_{edg} (λ) - n_{ctr} (λ)] .

Equation (9) introduces explicitly the notion that the material at the edge of a GRIN lens is different than the material at its center. The material at intermediate locations is a blended combination of these two boundary materials. Thus, if we know the dispersions for the edge and center materials, we can infer not just the index at all radial locations but also the dispersion. We note that for radial gradients higher than first-order, the representation of the gradient generator is not as compact.

If Γ(λ₀) < 0, the lens center has greater refractive index than the edge. Since this corresponds to a lens with positive optical power, we refer to this as a positive GRIN. Similarly, Γ(λ₀) > 0 indicates the lens center has less refractive index than the edge, which corresponds to a lens with negative power. We refer to this condition as a negative GRIN. However, rather than refering to Γ(λ₀) to define the GRIN, we introduce the GRIN sign s_G to increase clarity,

s_{G} = sgn [n_{ctr} (λ_{0}) - n_{edg} (λ_{0})],

where sgn is the signum function. Note that s_G is positive for a positive GRIN and negative for negative GRIN, which we use in the following analysis.

3. Optical properties of a GRIN lens

In this section, we use the expression for optical power to derive chromatic aberration. The form of our equations differs slightly from what is currently in the literature. However, we feel it simplifies analysis and lends itself to a more physical and intuitive presentation than previous work.

3.1. Optical power

We represent the chromatic behavior of focal length by f(λ), which is inversely related to optical power ϕ(λ),

ϕ (λ) = \frac{1}{f (λ)} .

From Sands’ expression for C in Eq. (16) of [12], the optical power for a GRIN lens is

ϕ (λ) = ϕ_{R} (λ) C [Φ (λ)] + [ϕ_{G} (λ) + ϕ_{R 2} (λ)] [\frac{S [Φ (λ)]}{Φ (λ)}],

where, for a negative gradient,

C [ϕ (λ)] = cosh ϕ (λ),

S [ϕ (λ)] = sinh ϕ (λ) .

and for a positive gradient,

C [ϕ (λ)] = cos ϕ (λ),

S [ϕ (λ)] = sin ϕ (λ) .

The optical powers are

ϕ_{R} (λ) = (\frac{1}{R_{f}} - \frac{1}{R_{b}}) [n_{ctr} (λ) - 1],

ϕ_{R 2} (λ) = \frac{{[n_{ctr} (λ) - 1]}^{2} t}{R_{f} R_{b} n_{ctr} (λ)},

ϕ_{G} (λ) = s_{G} \frac{Φ^{2} (λ) n_{ctr} (λ)}{t},

and the gradient is characterized by Φ(λ),

Φ (λ) = \sqrt{(\frac{8 t^{2}}{D^{2}}) [\frac{| n_{edg} (λ) - n_{ctr} (λ) |}{n_{ctr} (λ)}]} .

The term ϕ_G(λ) is the contribution to optical power from the strength of the gradient and ϕ_R2(λ), the contribution from the strength of the surface curvature. Note that, if no gradient is present, Φ(λ) = 0 and Eq. (12) reduces to the Lens Maker’s equation for conventional singlets [21].

We note that some treatments of GRIN elements define the gradient strength α(λ) such that Φ(λ) = α(λ)t. Although representing a gradient as a spatial frequency is useful when one wishes to understand the structure of the gradient in a material, we found that representing the gradient as a phase was more effective for optical design. This is evident in Sec. 5.

3.2. Chromatic aberration

Differentiation of Eq. (12) with respect to λ yields a measure of the element’s chromatic aberration. We use the relationships presented in the Appendix to write this succinctly as

\frac{d ϕ (λ)}{d λ} = \frac{ϕ_{R} (λ)}{V_{R} (λ)} + \frac{ϕ_{R 2} (λ)}{V_{R 2} (λ)} + \frac{ϕ_{G} (λ)}{V_{G} (λ)} .

The expressions for the V -terms are given in the Appendix. These terms are analogous to Abbe numbers in that they relate to dispersion, i.e., large values indicate small dispersion and small values, large dispersion. One can interpret the V -terms as the bandwidth Δλ over which a relative change in power Δϕ/ϕ takes place.

In the following, we use Eqs. (12) and (21) to consider optical and material achromatic lens design. To do so, we consider simplifying assumptions that reduce the complexity of the equations and provide us with physical insight into their meaning.

For our purposes, achromatization means dϕ(λ₀)/dλ = 0, i.e., the lens power near the design wavelength λ₀ does not change with λ. This is, of course, a property local to λ₀. As we discuss below, the range over which one can expect this condition to hold within some tolerance is dictated by the choice of materials.

We define achromatic optical design when, given a gradient material, one achieves achromatization by specifying the lens’ geometric parameters. Conversely, we define achromatic material design when, given a range of geometric parameters, one determines material properties that are conducive to achromatization.

We begin our discussion of achromatic design by considering a Wood lens, a gradient lens that has no refractive power, due to its flat front and back surfaces. The chromatic properties of a Wood lens have been studied previously [14–16] and provide insight into more general achromatic design and GRIN lens behavior.

4. Wood lens

A Wood lens generates no power at its surfaces due to curvature. Thus, the variation in power with wavelength is determined solely by the dispersive properties of the gradient index. If ϕ_R(λ) = 0 and ϕ_R2(λ) = 0, the power and chromatic aberration (Eqs. (12) and (21)) reduce to,

ϕ (λ) = ϕ_{G} (λ) [\frac{S [Φ (λ)]}{Φ (λ)}],

\frac{d ϕ (λ)}{d λ} = \frac{ϕ_{G} (λ)}{V_{G} (λ)} .

If we set dϕ(λ₀)/dλ = 0, use Eqs. (19) and (23), and use the expression for V_G(λ) from the Appendix, we can express the edge material dispersion that insures achromatization in terms of the center material dispersion,

\begin{array}{l} \frac{d n_{edg} (λ_{0})}{d λ} & = & [\frac{d n_{ctr} (λ_{0})}{d λ}] {\frac{2 [\frac{S [Φ (λ_{0})]}{Φ (λ_{0})}] - [\frac{n_{edg} (λ_{0})}{n_{ctr} (λ_{0})}] [\frac{S [Φ (λ_{0})]}{Φ (λ_{0})} - C [Φ (λ_{0})]]}{\frac{S [Φ (λ_{0})]}{Φ (λ_{0})} + C [Φ (λ_{0})]}}, \\ = & [\frac{d n_{ctr} (λ_{0})}{d λ}] [\frac{n_{edg} (λ_{0})}{n_{ctr} (λ_{0})}] {1 + 2 [\frac{n_{ctr} (λ_{0})}{n_{edg} (λ_{0})} - 1] [\frac{\frac{S [Φ (λ_{0})]}{Φ (λ_{0})}}{\frac{S [Φ (λ_{0})]}{Φ (λ_{0})} + C [Φ (λ)]}]} . \end{array}

Note that for achromatization to hold, the dispersion and index of the edge material must maintain a specific relationship with the dispersion and index of the center material.

We compare Eq. (24) to results derived by Krishna and Sharma, who also considered achromatization of a Wood lens [15, 16]. We borrowed their definition of Ψ₁(λ) (Eq. (14) in [15]), which we use in the Appendix and repeat here,

Ψ_{1} (λ) = \frac{1}{n_{ctr} (λ)} \frac{d n_{ctr} (λ)}{d λ} - \frac{1}{n_{edg} (λ)} \frac{d n_{edg} (λ)}{d λ} .

A comparison to our work reveals that, after making suitable substitutions for different variable definitions, Eqs. (6) and (8) in [16] are equivalent to our Eqs. (22) and (23), respectively. However, despite this agreement, [15] indicates that material achromatization occurs when Ψ₁(λ₀) = 0, which is in contrast to our Eq. (24). The two are equivalent only when n_ctr(λ) ≈ n_edg(λ), which is implicit in Krishna and Sharma’s polynomial expansion of the gradient generator (Eq. (17) in [15]). Other than the paraxial approximation, we made no assumptions to derive Eq. (24).

For a thin lens such that Φ(λ₀) << 1, Eq. (23) reduces to,

\frac{d ϕ (λ_{0})}{d λ} = (\frac{8 t}{D^{2}}) [\frac{d n_{ctr} (λ_{0})}{d λ} - \frac{d n_{edg} (λ_{0})}{d λ}] .

This follows from Eqs. (19) and (20), if we assume C[ϕ(λ)] ≈ S[ϕ(λ)]/ϕ(λ) ≈ 1. For a Wood lens to have the same focal length for all wavelengths, the first derivatives of the dispersion curves for the center and edge materials must be equal to one another. In other words, their dispersion curves differ only by a constant index contrast Δn. (This also follows from Eq. (24) when C[ϕ(λ)] ≈ S[ϕ(λ)]/ϕ(λ) ≈ 1.)

Since real materials rarely meet this condition, we note the tradeoffs implied by Eq. (25) for a fixed diameter. For a fixed thickness, the smaller the difference in dispersions, the smaller the amount of chromatic aberration. If one wishes to hold chromatic aberration below some tolerance level, there exists a maximum thickness one can achieve and still meet that criterion. For a small difference in dispersion, the maximum thickness may hinder the design of a lens with sufficient power to meet the application.

If there is a target value for chromatic aberration, to balance the aberration of another element, for example, then the material relationships require a specific thickness to meet that goal; the greater the difference in the center and edge material dispersions, the thinner the lens can be.

This behavior is true also for variations in focal length due to other parameters, such as temperature. So long as the difference in index variation at the edge and at the center is small, the variation in focus is small. For a desired focal deviation, an element with a small variation from center to edge will have a greater thickness than an element with a large variation.

5. GRIN lens assumptions

In this section we consider approximations that simplify subsequent analysis and design. Our objective in doing so is to strip away extraneous details to elucidate core physical relationships that describe the behavior of an achromatic GRIN lens. These relationships enable the optical designer to think about the materials and geometries that work best for achromatic GRIN design. The utility of our approach is borne out in the next section on design, Sec. 6, wherein ray tracing validates the robustness of achromatic GRIN designs generated using paraxial and thin lens approximations.

5.1. Thin lens

In conventional optics, the thin lens approximation for the Lens Makers Equation holds when ϕ_R2(λ) << ϕ_R(λ). This is typically a valid approximation because, for all but weak meniscus lenses, ϕ_R2(λ) and dϕ_R2(λ)/dλ are a small fraction of ϕ_R(λ) and dϕ_R(λ)/dλ, given that the ratio of ϕ_R2(λ) to ϕ_R(λ) is on the order of t/(R_b − R_f).

We now consider the validity of this approximation when we add a GRIN element. For GRIN achromats to provide an advantage over conventional achromats, the achromats should be comparable in weight and volume. This restricts the class of lenses to conventional “pancake”-shaped lenses, as opposed to rod-like lenses, and maintains the above-mentioned condition t/(R_b − R_f) << 1.

Further, we recognize that rays bend more quickly at the abrupt interface between air and a curved high index material than they do through even a strong index gradient; the strong gradient index requires a propagation length on the order of the lens diameter to achieve ray deflections equal to those of homogeneous refraction [22]. This suggests that the most efficient manner to design a GRIN achromat is to use surface refraction to generate optical power and GRIN dispersion to balance the color.

Depending on the materials used in the GRIN, the GRIN power ϕ_G(λ) may or may not be small compared to the refractive power ϕ_R(λ). We, therefore, need to keep the GRIN power term. But, because we want surface curvature to generate considerable optical power, i.e., the surface cannot be a weak meniscus, we require ϕ_R2(λ) << ϕ_R(λ). Finally, for the dispersion from the GRIN power to cancel the dispersion generated by the surface power, GRIN dispersion must dominate dϕ_R2(λ)/dλ just as much as dϕ_R(λ)/dλ does.

For these reasons, the conventional thin lens approximation, which ignores contributions from ϕ_R2(λ), remains a good approximation for designing the GRIN achromats. Under this approximation, the equations for power and chromatic aberration reduce to

ϕ (λ) = ϕ_{R} (λ) C [Φ (λ) + ϕ_{G} (λ)] [\frac{S [Φ (λ)]}{Φ (λ)}],

and

\frac{d ϕ (λ)}{d λ} = \frac{ϕ_{R} (λ)}{V_{R} (λ)} + \frac{ϕ_{G} (λ)}{V_{G} (λ)},

where 1/V_R(λ) and 1/V_G(λ) are given in Table 1.

Table 1. Design Coefficients for Dispersion.

View Table | View all tables in this article

5.2. Thin lens-thin GRIN

Examination of Eq. (12) shows that the lens powers evolve with trigonometric functions of Φ(λ). From Eq. (20), because the term for the relative change in index is order one, the lens thickness t that corresponds to Φ(λ) = 2π is always much greater than the diameter D. For a GRIN lens to be an attractive replacement for conventional achromatic doublets, its thickness should be on the order of a standard doublet, which is generally less than the diameter. For GRIN lenses constrained to thicknesses less than their diameter, we must have Φ(λ) < 1. We refer to this as a thin GRIN, in which case, using a Taylor series expansion, the power reduces to

ϕ (λ) = ϕ_{R} (λ) [1 - s_{G} \frac{Φ^{2} (λ)}{2}] + ϕ_{G} (λ) [1 - s_{G} \frac{Φ^{2} (λ)}{6}],

and, although the chromatic aberration equation remains unchanged,

\frac{d ϕ (λ)}{d λ} = \frac{ϕ_{R} (λ)}{V_{R} (λ)} + \frac{ϕ_{R} (λ)}{V_{G} (λ)},

the terms 1/V_R(λ) and 1/V_G(λ) adopt simpler forms, which are given in Table 1.

5.3. Thin lens-very thin GRIN

For the simplest expressions, we consider Φ(λ) << 1. From Eq. (20), this can occur when the relative difference in refractive index between the center and edge is small and when the lens thickness is smaller than its diameter. Under this condition, the power and chromatic aberration are

ϕ (λ) = ϕ_{R} (λ) + ϕ_{R} (λ),

\frac{d ϕ (λ)}{d λ} = \frac{ϕ_{R} (λ)}{V_{R} (λ)} + \frac{ϕ_{G} (λ)}{V_{G} (λ)} .

The simplified chromatic aberration terms 1/V_R(λ) and 1/V_G(λ) are given in Table 1.

6. Achromatic design

Our analysis reveals that, assuming the paraxial and thin lens approximations are valid, we can model a GRIN lens as a doublet comprised of a homogeneous refractive lens and a Wood lens. See Fig. 2. All other approximations affect only the values that multiply the refractive and GRIN powers in the expressions for optical power and chromatic behavior. The equations that allow a designer to correct first-order chromatic aberration are

ϕ (λ) = A (λ) ϕ_{R} (λ) + B (λ) ϕ_{G} (λ),

\frac{d ϕ (λ)}{d λ} = \frac{ϕ_{R} (λ)}{V_{R} (λ)} + \frac{ϕ_{G} (λ)}{V_{G} (λ)} .

To design a lens with power ϕ₀ at a design wavelength λ₀, we define it locally corrected for color if dϕ(λ₀)/dλ = 0. This condition sets the following relationships:

ϕ_{R} (λ_{0}) = \frac{ϕ_{0} V_{R} (λ_{0})}{A (λ_{0}) V_{R} (λ_{0}) - B (λ_{0}) V_{G} (λ_{0})},

ϕ_{G} (λ_{0}) = \frac{- ϕ_{0} V_{G} (λ_{0})}{A (λ_{0}) V_{R} (λ_{0}) - B (λ_{0}) V_{G} (λ_{0})}

where the coefficients for the approximations we considered above are summarized in Tables 1 and 2.

Table 2. Design Coefficients for Optical Power.

View Table | View all tables in this article

Fig. 2 GRIN lens represented as doublet comprised of a homogeneous refractive lens and a Wood lens.

Download Full Size | PDF

6.1. Materials considerations for achromatic design

A central result of our analysis is that, once chromatic aberration is corrected at λ₀, the lens is also nearly corrected at other wavelengths. The extent over which this is true depends upon the ratio ϕ_G(λ)/ϕ_R(λ) at nearby wavelengths. Since this ratio depends only upon the material dispersion curves,

\frac{ϕ_{G} (λ)}{ϕ_{R} (λ)} = - \frac{V_{G} (λ)}{V_{R} (λ)},

one can calculate the ratio on the right hand side for arbitrary material pairs to determine bandwidths over which the ratio stays constant or near-constant. Other factors will affect how effective the materials are for a design, but the behavior of this ratio will determine how constant the performance is over a wavelength range.

In typical examples of a GRIN achromat, which allow for direct comparisons to conventional doublets, a negative GRIN power produces the dispersion required to balance that of a positive refractive power. To maximize the overall lens power while still achieving chromatic balance, the designer would prefer to build in as little negative GRIN power as possible. This suggests that attractive material pair candidates have a ratio V_G(λ₀)/V_R(λ₀) with a small magnitude. In the thin lens-very thin GRIN approximation, this reduces to:

| \frac{V_{G} (λ_{0})}{V_{R} (λ_{0})} | = | [\frac{d n_{ctr} (λ_{0}) / d λ}{n_{ctr} (λ_{0}) - 1}] [\frac{n_{ctr} (λ_{0}) - n_{edg} (λ_{0})}{d n_{edg} (λ_{0}) / d λ - d n_{ctr} (λ_{0}) / d λ}] | .

We compare this to the results of a similar analysis of a conventional doublet that consists of two homogeneous lenses with indices n₁(λ) and n₂(λ),

| \frac{V_{1} (λ_{0})}{V_{2} (λ_{0})} | = [\frac{n_{1} (λ_{0}) - 1}{n_{2} (λ_{0}) - 1}] | \frac{d n_{2} (λ_{0}) / d λ}{d n_{1} (λ_{0}) / d λ} | .

Note that different material pairs optimize each expression. Materials suitable for a homogeneous doublet include one with high index and low dispersion, and a second material with low index and high dispersion. Materials well suited to an achromatic GRIN have indices that are large and similar but whose dispersions are quite different, one low and one high.

6.2. Thin lens-very thin GRIN design

We now provide examples of both analysis and design using the simplest forms for chromatic aberration. If we normalize the equations by the diameter D and assume we have selected materials for the center and edge, the geometric parameters are

(\frac{D}{R_{f}} - \frac{D}{R_{b}}) = (D ϕ_{0}) \frac{d n_{ctr} (λ_{0}) / d λ - d n_{edg} (λ_{0}) / d λ}{[d n_{ctr} (λ_{0}) / d λ] [n_{edg} (λ_{0}) - 1] - [d n_{edg} (λ_{0}) / d λ] [n_{ctr} (λ_{0}) - 1]}

(\frac{8 t}{D}) = - (D ϕ_{0}) \frac{d n_{ctr} (λ_{0}) / d λ}{[d n_{ctr} (λ_{0}) / d λ] [n_{edg} (λ_{0}) - 1] - [d n_{edg} (λ_{0}) / d λ] [n_{ctr} (λ_{0}) - 1]}

Alternatively, if we fix the geometric parameters, the material parameters that insure achromatic performance are,

\frac{d n_{edg} (λ_{0})}{d λ} = [1 + (\frac{D^{2}}{8 t}) (\frac{1}{R_{f}} - \frac{1}{R_{b}})] [\frac{d n_{ctr} (λ_{0})}{d λ}] .

6.3. Examples

6.3.1. Material selection

We used Eqs. (37) and (38) to search the Schott catalog of optical glasses [23] for material pairs suitable to design homogeneous doublet and GRIN achromats. Because the equations are defined locally around λ₀, to determine their maximum value within the visible spectrum, we evaluated them at several wavelengths between 0.4 and 0.7 μm. Given the difference between indices in the denominator of Eq. (37), the condition for a GRIN element, the sample rate must be sufficiently high to resolve the wavelength at which small differences occur.

The pair that yielded the largest value for the homogeneous condition Eq. (38) was N-FK51A and SF66. Their indices and Abbe numbers are listed in Table 3.

Table 3. Glass pairs selected using homogeneous doublet (pair 1) and GRIN (pair 2) achromatic criteria.

View Table | View all tables in this article

We performed a similar search for material pairs that minimized Eq. (37). Note that this occurs when the edge and center materials have the same index at the operating wavelength but different dispersions. For the wavelength at which the materials share the same index, the GRIN element appears homogeneous despite the material inhomogeneity and Eq. (37) is zero. This implies that the GRIN element provides no optical power. Instead, it only corrects the chromatic dispersion of the refractive lens.

The Schott catalog contains multiple examples of glass pairs that have near identical indices of refraction in the visible spectrum yet distinct dispersion curves. We selected N-SK4 and NKzFS4 as a representative example from the candidate pairs available. Their indices and Abbe numbers are listed in Table 3.

6.3.2. Optical design

For both pairs of materials, we designed an achromatic doublet and radial GRIN lens. Our designs are summarized in Tables 4–6. Given the number of simplifying assumptions in our analysis, many of which are stressed by the applications we believe GRIN can positively impact, we chose to test our approach on a non-paraxial lens design with a 24-mm entrance pupil aperture, 25-mm clear aperture, and 72-mm focal length. The lens should be achromatic over the visible band defined conventionally by the wavelengths λ_F = 0.4861 μm, λ_d = 0.5876 μm, and λ_C = 0.6563 μm.

Table 4. Surface descriptions for homogeneous achromat designs.

View Table | View all tables in this article

Table 5. Surface descriptions for GRIN achromat designs.

View Table | View all tables in this article

Table 6. Additional parameters for GRIN achromat designs.

View Table | View all tables in this article

Except for thickness, the values in Tables 5 and 6 describe the front surface of the material indicated. The thickness value for air is the surface’s back focal length. We assumed the homogeneous doublets are cemented and designed with 1-mm minimum edge and center thicknesses. We constrained the thickness of the GRIN lens to be less than or equal to that of the doublet.

As documented previously [19], addressing chromatic aberration with GRIN exacerbates geometric aberrations, which one can compensate using surface asphere terms. Consequently, to provide a fair comparison to the aspheric GRIN lenses, we allowed the front surface of the doublet to contain aspheres as well. A conic constant sufficed to provide the maximum benefit. Finally, to accommodate the full f /3 aperture, we allowed the GRIN profile to deviate from a simple parabola and included high-order polynomial terms in the optimization process. With reference to Eq. (1) and Table 6, we included conic and even-order asphere surface terms up to order r¹⁸, and both even and odd index polynomial terms from r⁰ to r⁹. The values for the index terms are given for λ_d.

6.3.3. Optical analysis

The optical performance of the resulting four lenses is represented in Figs. 3 by their rayfans and in Table 7 by their calculated spot sizes. A rayfan represents the height of a ray in the image plane y_i as a function of the ray’s height in the entrance aperture y_l, which in our case is the lens front surface. We computed the spot sizes from the ray fans.

Fig. 3 Optical performance of achromats. Rayfan for homogeneous doublet achromat design using (a) glass pair 1 and (b) glass pair 2. Rayfan for GRIN achromat design using (c) glass pair 1 and (d) glass pair 2. The color of each graph is representative of its wavelength: red (λ_C), green (λ_d), and blue (λ_F).

Download Full Size | PDF

Table 7. Calculated spot size in μm for designed achromats.

View Table | View all tables in this article

We produced the rayfans using a commercial raytrace package, ZEMAX^®. However, because native ZEMAX cannot easily model high-order radial GRINs that contain aspheric terms in arbitrary material blends, we generated our own model to communicate surface shape and dispersive GRIN information to ZEMAX’s ray trace engine. To do so, we followed ZEMAX’s prescription for developing dynamically linked libraries.

We validated our model by comparing ZEMAX-computed wavefronts transmitted through GRIN test structures to those calculated from independently developed GRIN propagation code that is based on published algorithms [24, 25]. Wavefronts calculated using the two methods agreed to within picometers of optical path length.

Further, our model confirmed the utility of paraxial power as a valid model for analysis and design. Results predicted by Eq. (12) agreed with those generated by ray tracing to within a few percent.

The homogeneous doublet designs are tabulated in Table 4 and their optical performance is represented in Figs. 3(a) and 3(b) and Table 7. The performance represented in Figs. 3(a) and 3(b) reflects that of a typical homogeneous doublet. The limiting aberration for both designs is spherochromatism, the wavelength-dependent variation in spherical aberration. Note that spherochromatism of blue and red are opposite in sign. As reflected in the spot sizes presented in Table 7, glass pair 1 outperforms glass pair 2. The geometric and RMS spot sizes for glass pair 2 is seven times larger than that of glass pair 1, which we selected using the homogeneous glass criterion.

The GRIN designs are tabulated in Tables 5 and 6 and their optical performance is represented in Figs. 3(c) and 3(d) and Table 7. It is clear from these that the GRIN achromat we designed using a glass pair selected especially for this purpose yields the best optical performance. The nearly straight lines in Fig. 3(d) indicate that all high order aberrations are corrected for glass pair 2. The limiting aberration is now secondary color, which is measured by the distance between the focal plane for λ_d and the focal plane for the extremal wavelengths λ_F and λ_C.

It is interesting to note that glass pair 1 produces a GRIN achromat whose performance is comparable to, but still slightly better than, its homogeneous achromat. As we discuss below, the GRIN achromat from glass pair 1 was a difficult design to produce, especially when compared to the ease with which we found an achromatic GRIN using glass pair 2. Nonetheless, glass pair 2 produces a GRIN achromat whose spot size is approximately 70 percent that of glass pair 1.

Some comments are pertinent to the development of the GRIN lens designs. Although we use several polynomial terms to describe the index variation, as represented in Fig. 4, the indices are smoothly monotonic. Figure 4 represents the volume fraction γ(r) of the material blend, as opposed to its refractive index, where

γ (r) = \frac{n^{2} (r, λ_{0}) - n_{\min}^{2} (λ_{0})}{n_{\max}^{2} (λ_{0}) - n_{\min}^{2} (λ_{0})} .

This makes a material comparison between the different designs simpler. The form of Eq. (42) is a consequence of effective medium theory [5].

Fig. 4 Volume fraction as a function of lens radius for the GRIN achromats. (a) Glass Pair 1. (b) Glass Pair 2.

Download Full Size | PDF

From Fig. 4, it is evident that the optimal design has the most dispersive material at the lens edge. However, neither solution has the pure glass counterpart at the center. Instead, the center material is a blend, 30 percent for glass pair 1 and 15 percent for glass pair 2. Given that we selected the glass pairs from a list of existing materials, as opposed to designing them from scratch, they are not the optimal materials for either lens design. Consequently, the numerical optimizer identified a blend that performed better than a pure glass.

Note that each curve is qualitatively parabolic. It has a flat minimum in the center and rises smoothly at the edges. Its departure from a simple parabola indicates the degree to which computed light rays at relatively low f -numbers achieve good focus in contrast to those of paraxial ray theories.

As mentioned above, although we found a solution to the GRIN achromat design using glass pair 1, it was significantly more difficult to find than the solution using glass pair 2. The design for glass pair 2 converged quickly with minor effort. The design for glass pair 1 required long runs of non-local optimization algorithms before it settled into a practicable solution. From Table 6, it is evident that the solution requires many more index and asphere terms than glass pair 2 to achieve the performance depicted in Fig. 3(c), which is comparable to Fig. 3(d).

During the design process, we observed that we could achieve optical performance similar to that presented in Fig. 3(d) by thinning the glass pair 2 GRIN, but this required us to include more index and surface polynomial terms. However, there existed a minimum thickness beyond which we could no longer achieve the same optical performance. Therefore, we believe that the oscillations visible in the ray fan of Fig. 3(c) can be reduced if we used more index and surface terms, if we allowed the GRIN lens to be thicker, or both. This only highlights our point that design complexity increases when one uses materials for a conventional achromat in a GRIN achromat design, as opposed to materials selected for an achromatic GRIN.

7. Conclusion

From the expression for optical power of a radial first-order GRIN lens with curved surfaces, we derived an expression for chromatic aberration. By applying a set of simplifying assumptions, we combine chromatic behavior with optical power to derive a set of equations with which one can design a paraxial starting point for an achromatic GRIN lens.

We also used our analysis to identify relationships among suitable material pairs for a GRIN achromat. This enabled us to search within a standard glass catalog for attractive GRIN material pairs, along with a side-by-side comparison to a similar search for conventional doublet materials. We demonstrate that our analysis highlights material pairs which perform well for achromatic GRIN lenses which would not generally be considered in conventional achromatic design. We anticipate future progress in the selection of advantageous material pairs by refining the concepts introduced here.

With the advent of new technologies for producing gradient-index optics, these results should be instructive to those taking advantage of the opportunities afforded by GRIN optical design, as well as by those invested in the development of new materials which could open up new application areas for advanced GRIN optics.

Appendix

To derive the chromatic behavior of a GRIN element, we differentiate the GRIN optical power. For a positive GRIN, the optical power is

ϕ (λ) = ϕ_{R} (λ) cos Φ (λ) + [ϕ_{G} (λ) + ϕ_{R 2} (λ)] [\frac{sin Φ (λ)}{Φ (λ)}],

which, using Eq. (19), yields

\begin{array}{l} \frac{d ϕ (λ)}{d λ} & = & {\frac{d ϕ_{R} (λ)}{d λ} cos Φ (λ) - ϕ_{R} (λ) [\frac{sin Φ (λ)}{Φ (λ)}] [Φ (λ) \frac{d Φ (λ)}{d λ}]} \\ + & {\frac{d [Φ^{2} (λ) n_{ctr} (λ) / t]}{d λ} + \frac{d ϕ_{R 2} (λ)}{d λ}} [\frac{sin Φ (λ)}{Φ (λ)}] \\ + & [Φ^{2} (λ) \frac{n_{ctr} (λ)}{t} + ϕ_{R 2} (λ)] \frac{d [sin Φ (λ) / Φ (λ)]}{d λ} . \end{array}

For a negative GRIN, the optical power and chromatic behavior are

ϕ (λ) = ϕ_{R} (λ) cosh Φ (λ) + [ϕ_{G} (λ) + ϕ_{R 2} (λ)] [\frac{sinh Φ (λ)}{Φ (λ)}],

\begin{array}{l} \frac{d ϕ (λ)}{d λ} & = & {\frac{d ϕ_{R} (λ)}{d λ} cosh Φ (λ) + ϕ_{R} (λ) [\frac{sinh Φ (λ)}{Φ (λ)}] [Φ (λ) \frac{d Φ (λ)}{d λ}]} \\ + & {- \frac{d [Φ^{2} (λ) n_{ctr} (λ) / t]}{d λ} + \frac{d ϕ_{R 2} (λ)}{d λ}} [\frac{sinh Φ (λ)}{Φ (λ)}] \\ + & [- Φ^{2} (λ) \frac{n_{ctr} (λ)}{t} + ϕ_{R 2} (λ)] \frac{d [sinh Φ (λ) / Φ (λ)]}{d λ} . \end{array}

From Eqs. (17)–(20), we determine

\frac{d ϕ_{R} (λ)}{d λ} = [\frac{1}{n_{ctr} (λ) - 1}] [\frac{d n_{ctr} (λ)}{d λ}] ϕ_{R} (λ),

\frac{d ϕ_{R 2} (λ)}{d λ} = [\frac{1}{n_{ctr} (λ)}] [\frac{n_{ctr} (λ) + 1}{n_{ctr} (λ) - 1}] [\frac{d n_{ctr} (λ)}{d λ}] ϕ_{R 2} (λ),

Φ (λ) \frac{d Φ (λ)}{d λ} = [\frac{Φ^{2} (λ)}{2}] [\frac{n_{edg} (λ)}{n_{ctr} (λ) - n_{edg} (λ)}] Ψ_{1} (λ) .

The form of Eq. (47) removes the need to include radicals in our equations. We note also that

\frac{d [sin Φ (λ) / Φ (λ)]}{d λ} = [cos Φ (λ) - \frac{sin Φ (λ)}{Φ (λ)}] [\frac{1}{Φ (λ)} \frac{d Φ (λ)}{d λ}],

\frac{d [sinh Φ (λ) / Φ (λ)]}{d λ} = [- cosh Φ (λ) - \frac{sinh Φ (λ)}{Φ (λ)}] [\frac{1}{Φ (λ)} \frac{d Φ (λ)}{d λ}] .

We combine Eqs. (43) and (44) with Eqs. (45)–(49) and use algebra to yield

\begin{array}{l} \frac{1}{V_{R} (λ)} & = & [\frac{C [Φ (λ)]}{n_{ctr} (λ) - 1}] [\frac{d n_{ctr} (λ)}{d λ}] \\ + & s_{G} [\frac{S [Φ (λ)]}{Φ (λ)}] [\frac{Φ^{2} (λ)}{2}] [\frac{n_{edg} (λ)}{n_{edg} (λ) - n_{ctr} (λ)}] Ψ_{1} (λ), \end{array}

\begin{array}{l} \frac{1}{V_{R 2} (λ)} & = & {\frac{n_{ctr} (λ) + 1}{n_{ctr} (λ) [n_{ctr} (λ) - 1]}} [\frac{S [Φ (λ)]}{Φ (λ)}] [\frac{d n_{ctr} (λ)}{d λ}] \\ + & (\frac{1}{2}) [\frac{S [Φ (λ)]}{Φ (λ)} - C [Φ (λ)]] [\frac{n_{edg} (λ)}{n_{edg} (λ) - n_{ctr} (λ)}] Ψ_{1} (λ), \end{array}

\begin{array}{l} \frac{1}{V_{G} (λ)} & = & {\frac{1}{2 [n_{edg} (λ) - n_{ctr} (λ)]}} \\ \times & ([\frac{d n_{ctr} (λ)}{d λ}] {2 [\frac{S [Φ (λ)]}{Φ (λ)}] - [\frac{n_{edg} (λ)}{n_{c t r} (λ)}] [\frac{S [Φ (λ)]}{Φ (λ)} - C [Φ (λ)]]} \\ - & [\frac{d n_{edg} (λ)}{d λ}] [\frac{S [Φ (λ)]}{Φ (λ)} + C [Φ (λ)]]), \end{array}

We borrow the definition of Ψ₁(λ) from Krishna and Sharma [15],

Ψ_{1} (λ) = \frac{1}{n_{ctr} (λ)} \frac{d n_{ctr} (λ)}{d λ} - \frac{1}{n_{edg} (λ)} \frac{d n_{edg} (λ)}{d λ} .

References and links

1. R. W. Wood, Physical Optics (Macmillan, 1905).

2. D. T. Moore, “Gradient-index optics: a review,” Appl. Opt. 19, 1035–1038 (1980). [CrossRef] [PubMed]

3. E. W. Marchand, Gradient Index Optics (Academic Press, 1978).

4. S. N. Houde-Walter and D. T. Moore, “Gradient-index profile control by field-assisted ion exchange in glass,” Appl. Opt. 24, 4326–4333 (1985). [CrossRef] [PubMed]

5. M. Ponting, A. Hiltner, and E. Baer, “Polymer nanostructures by forced assembly: process, structure, properties,” Macromol. Symp. 294, 19–32 (2010). [CrossRef]

6. H. Liu, T. Maekawa, N. M. Patrikalakis, E. M. Sachs, and W. Cho, “Methods for feature-based design of heterogeneous solids,” Computer-Aided Design 36, 1141–1159 (2004). [CrossRef]

7. I. I. Smolyaninov, “Graded index metamaterial lens,” International Patent Application, WO2013/032758 (7 March 2013).

8. D. H. Werner, T. S. Mayer, C. Rivero-Baleine, N. Podraza, K. Richardson, J. Turpin, A. Pogrebnyakov, J. D. Musgraves, J. A. Bossard, H. J. Shin, R. Muise, S. Rogers, and J. D. Johnson, “Adaptive phase change metamaterials for infrared aperture control,” in Unconventional Imaging, Wavefront Sensing, and Adaptive Coded Aperture Imaging and Non-Imaging Sensor Systems, J. J. Dolne, T. J. Karr, V. L. Gamiz, S. Rogers, and D. P. Casasent, eds., Proc. SPIE8165, 81651H (2011). [CrossRef]

9. H. A. Buchdahl, Optical Aberration Coefficients (Dover Publications, 1968).

10. P. J. Sands, “Third-order aberrations of inhomogeneous lenses,” J. Opt. Soc. Am. 60, 1436–1443 (1970). [CrossRef]

11. P. J. Sands, “Inhomogeneous lenses, II. Chromatic paraxial aberrations,” J. Opt. Soc. Am. 61, 777–783 (1971). [CrossRef] [PubMed]

12. P. J. Sands, “Inhomogeneous lenses, III. Paraxial optics,” J. Opt. Soc. Am. 61, 879–885 (1971). [CrossRef]

13. P. J. Sands, “Inhomogeneous lenses, V. Chromatic paraxial aberrations of lenses with axial or cylindrical index distributions,” J. Opt. Soc. Am. 61, 1495–1500 (1971). [CrossRef]

14. K. Nishizawa, “Chromatic aberration of the Selfoc lens as an imaging system,” Appl. Opt. 19, 1052–1055 (1980). [CrossRef] [PubMed]

15. K. S. R. Krishna and A. Sharma, “Chromatic aberrations of radial gradient-index lenses. I. Theory,” Appl. Opt. 35, 1032–1036 (1996). [CrossRef] [PubMed]

16. K. S. R. Krishna and A. Sharma, “Chromatic aberrations of radial gradient-index lenses. II. Selfoc lenses,” Appl. Opt. 35, 1037–1040 (1996). [CrossRef] [PubMed]

17. F. Bociort, “Chromatic paraxial aberration coefficients for radial gradient-index lenses,” J. Opt. Soc. Am. A 13, 1277–1284 (1996). [CrossRef]

18. G. Greisukh, S. Bobrov, and S. Stepanov, Optics of Diffractive and Gradient-Index Elements and Systems (SPIE Optical Engineering Press, 1997).

19. R. A. Flynn, E. F. Fleet, G. Beadie, and J. S. Shirk, “Achromatic GRIN singlet lens design,” Opt. Express 21, 4970–4978 (2013). [CrossRef] [PubMed]

20. J. A. Corsetti, P. McCarthy, and D. T. Moore, “Color correction in the infrared using gradient-index materials,” Opt. Eng. 52, 112109 (2013). [CrossRef]

21. E. Hecht, Optics, 2nd ed. (Addison-Wesley, 1987).

22. K. M. Jones, S. Lundgren, and A. Chakravorty, “A calculus of variations demonstration: The gradient index lens,” Am. J. Phys. 56, 1099 (1988). [CrossRef]

23. http://www.schott.com/advanced_optics/english/download/schott-optical-glass-pocket-catalog-january-2014-row.pdf. (Accessed March 5, 2015).

24. B. D. Stone and G. W. Forbes, “Differential ray tracing in inhomogeneous media,” J. Opt. Soc. Am. A 14, 2824–2836 (1997). [CrossRef]

25. A. Sharma, D. V. Kumar, and A. K. Ghatak, “Tracing rays through graded-index media: a new method,” Appl. Opt. 21, 984–987 (1982). [CrossRef] [PubMed]

approximation	$\frac{1}{V_{R} (λ)}$	$\frac{1}{V_{G} (λ)}$
thin lens	$\begin{matrix} [\frac{C [Φ (λ)]}{n_{ctr} (λ) - 1}] [\frac{d n_{ctr} (λ)}{d λ}] \\ + s_{G} [\frac{S [Φ (λ)]}{Φ (λ)}] [\frac{Φ^{2} (λ)}{2}] [\frac{n_{edg} (λ)}{n_{edg} (λ) - n_{ctr} (λ)}] Ψ_{1} (λ) \end{matrix}$	$\begin{matrix} {\frac{1}{2 [n_{edg} (λ) - n_{ctr} (λ)]}} \\ \times ([\frac{d n_{ctr} (λ)}{d λ}] {2 [\frac{S [Φ (λ)]}{Φ (λ)}] - [\frac{n_{edg} (λ)}{n_{ctr} (λ)}] [\frac{S [Φ (λ)]}{Φ (λ)} - C [Φ (λ)]]} \\ - [\frac{d n_{edg} (λ)}{d λ}] [\frac{S [Φ (λ)]}{Φ (λ)} + C [Φ (λ)]]) \end{matrix}$
thin lens-thin GRIN	$\begin{matrix} [\frac{1}{n_{ctr} (λ) - 1}] [\frac{d n_{ctr} (λ)}{d λ}] [1 - s_{G} \frac{Φ^{2} (λ)}{2}] \\ + s_{G} [\frac{Φ^{2} (λ)}{2}] [\frac{n_{edg} (λ)}{n_{edg} (λ) - n_{ctr} (λ)}] Ψ_{1} (λ) \end{matrix}$	$\begin{matrix} {\frac{1}{[n_{edg} (λ) - n_{ctr} (λ)]}} \\ \times ([\frac{d n_{ctr} (λ)}{d λ}] {1 - s_{G} [\frac{Φ^{2} (λ)]}{6}] [\frac{n_{edg} (λ)}{n_{ctr} (λ)} + 1]} \\ - [\frac{d n_{edg} (λ)}{d λ}] [1 - s_{G} \frac{Φ^{2} (λ)}{3}]) \end{matrix}$
thin lens-very thin GRIN	$[\frac{1}{n_{ctr} (λ) - 1}] [\frac{d n_{ctr} (λ)}{d λ}]$	$[\frac{1}{n_{edg} (λ) - n_{ctr} (λ)}] [\frac{d n_{ctr} (λ)}{d λ} - \frac{d n_{edg} (λ)}{d λ}]$

approximation	A(λ)	B(λ)
thin lens	C[Φ(λ)]	$\frac{S [Φ (λ)]}{Φ (λ)}$
thin lens-thin GRIN	$1 - s_{G} \frac{Φ^{2} (λ)}{2}$	$1 - s_{G} \frac{Φ^{2} (λ)}{6}$
thin lens-very thin GRIN	1	1

material parameter	glass pair 1		glass pair 2

	N-FK51A	SF66	N-SK4	N-KZFS4
n_d	1.48656	1.92286	1.61272	1.61336
Abbe number	84.5	20.9	58.6	44.5

design parameter	glass pair 1			glass pair 2

	N-FK51A	SF66	Air	N-SK4	N-KzFS4	Air
Radius (mm)	32.495	−142.895	−476.704	29.851	−19.427	81.611
Conic	−0.481	0.000	0.000	−0.366	0.000	0.000
Thickness (mm)	4.000	1.000	68.655	8.500	1.000	63.235
Aperture (mm)	25.000	25.000	25.000	25.000	25.000	25.000

design parameter	glass pair 1		glass pair 2

	N-FK51A+SF66	Air	N-SK4+N-KzFS4	Air
Radius (mm)	38.012	−139.813	44.963	−1841.900
Conic	0.000	0.000	−0.682	0.000
Thickness (mm)	5.000	69.136	9.000	66.548
Aperture (mm)	25.000	25.000	25.000	25.000

Chromatic analysis and design of a first-order radial GRIN lens

Abstract

1. Introduction

2. First-order radial gradients

3. Optical properties of a GRIN lens

3.1. Optical power

3.2. Chromatic aberration

4. Wood lens

5. GRIN lens assumptions

5.1. Thin lens

5.2. Thin lens-thin GRIN

5.3. Thin lens-very thin GRIN

6. Achromatic design

6.1. Materials considerations for achromatic design

6.2. Thin lens-very thin GRIN design

6.3. Examples

6.3.1. Material selection

6.3.2. Optical design

6.3.3. Optical analysis

7. Conclusion

Appendix

References and links

Cited By

Figures (4)

Tables (7)

Equations (56)

Optics Express

Index	Pair 1	Pair 2	Asphere	Pair 1	Pair 2

n₀	1.63473	1.61281	r⁴(×10⁻⁶)	8.261	6.005e−2
r²(×10⁻⁴)	7.327	2.768e−2	r⁶(×10⁻⁷)	−3.418	6.869e−4
r³(×10⁻⁶)	2.288	9.818e−3	r⁸(×10⁻⁹)	7.321	−2.820e−4
r⁴(×10⁻⁶)	1.351	1.322e−3	r¹⁰(×10⁻¹¹)	−8.738	−2.487e−5
r⁵(×10⁻⁶)	1.241	1.907e−4	r¹²(×10⁻¹³)	6.119
r⁶(×10⁻⁷)	−3.581	8.579e−6	r¹⁴(×10⁻¹⁵)	−2.452
r⁷(×10⁻⁸)	4.598		r¹⁶(×10⁻¹⁸)	5.337
r⁸(×10⁻⁹)	−1.953		r¹⁸(×10⁻²¹)	−5.237
r⁹(×10⁻¹¹)	−5.282
r¹⁰(×10⁻¹²)	5.246

calculated value	doublet		GRIN

	pair 1	pair 2	pair 1	pair 2
geometric spot size	3.06	20.70	2.48	1.73
RMS spot size	1.59	7.11	1.18	0.92