Thick film multilayer reflector with tailored layer thickness profile Number:7,385,763 from the United States Patent and Trademark Office (PTO) owispatent

Title: Thick film multilayer reflector with tailored layer thickness profile

Abstract: A multilayer reflector useable to reflect or transmit light over the visible wavelength range includes optically thick constituent layers. The optical thickness of the constituent layers through the thickness of the reflector defines a layer thickness profile. The layers are arranged so that the thickness profile has a tailored non-uniform distribution, such as a graded distribution or a randomized distribution. The layers desirably have an optical thickness in a range from about one to five or one to ten design wavelengths. The reflector can be a polarizer, reflecting only one normally incident polarization state, or a mirror, reflecting two normally incident orthogonal polarization states.

Patent Number: 7,385,763 Issued on 06/10/2008 to Nevitt,   et al.

---

Inventors:	Nevitt; Timothy J. (Red Wing, MN), Ouderkirk; Andrew J. (Woodbury, MN)
Assignee:	3M Innovative Properties Company (Saint Paul, MN)
Appl. No.:	11/109,212
Filed:	April 18, 2005

---

Current U.S. Class:	359/584 ; 359/494; 359/589
Field of Search:	359/494,495,498,584,585,586,587,588,589 349/105,113

---

References Cited [Referenced By]

---

U.S. Patent Documents

4525413	June 1985	Rogers et al.
4937134	June 1990	Schrenk et al.
5103337	April 1992	Schrenk et al.
5122905	June 1992	Wheatley et al.
5122906	June 1992	Wheatley
5126880	June 1992	Wheatley et al.
5360659	November 1994	Arends et al.
5448404	September 1995	Schrenk et al.
5486949	January 1996	Schrenk et al.
5540978	July 1996	Schrenk
5568316	October 1996	Schrenk et al.
5808798	September 1998	Weber et al.
5867316	February 1999	Carlson et al.
5872653	February 1999	Schrenk et al.
5882774	March 1999	Jonza et al.
5968666	October 1999	Carter et al.
6113811	September 2000	Kausch et al.
6157490	December 2000	Wheatley et al.
6207260	March 2001	Wheatley et al.
6335051	January 2002	Kausch et al.
6368699	April 2002	Gilbert et al.
6531230	March 2003	Weber et al.
6583930	June 2003	Schrenk et al.
6590707	July 2003	Weber
6610356	August 2003	Kausch et al.
6827886	December 2004	Neavin et al.
2002/0180107	December 2002	Jackson et al.
2002/0190406	December 2002	Merrill et al.
2004/0099992	May 2004	Merrill et al.
2004/0099993	May 2004	Jackson et al.

Foreign Patent Documents

	WO 95/17303		Jun., 1995		WO

Other References

US. Appl. No. 60/672,964 entitled "Multifunctional Thick Film Reflective Polarizer For Displays", filed Apr. 18, 2005. cited by other.

Primary Examiner: Amari; Alessandro

---

Claims

---

What is claimed is:

1. A multilayer reflector that substantially reflects light over the visible spectrum for at least one polarization state, the reflector comprising: a stack of light-transmissive polymer layers comprising at least a first and second alternating polymer material, the first and second materials having different refractive indices along at least one in-plane axis, the polymer layers having an average optical thickness of at least about 2.lamda..sub.0, where .lamda..sub.0 is a visible wavelength of interest, the polymer layers contributing substantially to the reflectivity of the multilayer reflector; wherein the polymer layers in the stack have a non-linear distribution of optical thicknesses along a thickness axis of the reflector.

2. The reflector of claim 1, wherein the distribution of optical thicknesses is selected to decrease variability in reflection or transmission over visible light wavelengths.

3. The reflector of claim 1, wherein the distribution of optical thicknesses is selected to decrease variability in reflection of normally incident light of the at least one polarization state over visible light wavelengths.

4. The reflector of claim 1, wherein the non-linear distribution includes a randomized distribution.

5. The reflector of claim 1, wherein the polymer layers have an average optical thickness of about 10.lamda..sub.0 or less.

6. The reflector of claim 5, wherein the polymer layers have an average optical thickness of about 5.lamda..sub.0 or less.

7. The reflector of claim 1, wherein the first polymer material is substantially birefringent and the second polymer material is substantially isotropic.

8. The reflector of claim 7, wherein the polymer layers of the first material are thinner than the polymer layers of the second material.

9. The reflector of claim 1, wherein the refractive index difference along an in-plane x-axis is .DELTA.n.sub.x and a refractive index difference along an in-plane y-axis perpendicular to the x-axis is .DELTA.n.sub.y, and where |.DELTA.n.sub.x|.apprxeq.|.DELTA.n.sub.y|.

10. The reflector of claim 1, wherein the refractive index difference along an in-plane x-axis is .DELTA.n.sub.x and a refractive index difference along an in-plane y-axis perpendicular to the x-axis is .DELTA.n.sub.y, and where |.DELTA.n.sub.x|>|.DELTA.n.sub.y|.

11. The reflector of claim 10, wherein |.DELTA.n.sub.y| is 0 or no greater than 0.01.

12. The reflector of claim 11, wherein the refractive index difference along a z-axis perpendicular to both the x- and y-axes is .DELTA.n.sub.z, and |.DELTA.n.sub.z| is 0 or no greater than 0.01.

13. The reflector of claim 11, wherein the refractive index difference along a z-axis perpendicular to both the x- and y-axes is .DELTA.n.sub.z, and |.DELTA.n.sub.z|>|.DELTA.n.sub.y|.

14. The reflector of claim 13, wherein the distribution of optical thicknesses is selected to decrease variability in transmission of p-polarized light obliquely incident in the y-z plane over visible light wavelengths.

15. The reflector of claim 1, wherein the reflector has an overall thickness of less than 1 mm.

16. The reflector of claim 1, wherein the reflector has an overall thickness of at least 1 mm.

17. The reflector of claim 1 in combination with at least one light source illuminating the reflector, the light source emitting light in at least one narrow emission peak.

18. A display comprising the reflector of claim 1.

19. The display of claim 18, wherein the reflector is a reflective polarizer.

20. The reflector of claim 1, wherein the non-linear distribution includes a non-uniform distribution.

21. The reflector of claim 20, wherein the non-linear distribution includes a randomized distribution.

22. The reflector of claim 20, wherein the non-linear distribution includes an irregular distribution.

23. The reflector of claim 20, wherein the non-linear distribution includes a mixed distribution.

24. A multilayer reflector that substantially reflects light over the visible spectrum for at least one polarization state, the reflector comprising: a stack of light-transmissive polymer layers comprising at least a first and second alternating polymer material, the first and second materials having different refractive indices along at least one in-plane axis, the polymer layers having an average optical thickness of at least about ( 5/4).lamda..sub.0, where .lamda..sub.0 is a visible wavelength of interest, the polymer layers contributing substantially to the reflectivity of the multilayer reflector; wherein the polymer layers in the stack have a non-uniform distribution of optical thicknesses along a thickness axis of the reflector; wherein the first polymer material is substantially birefringent and the second polymer material is substantially isotropic; and wherein the polymer layers of the first material are thinner than the polymer layers of the second material.

25. A multilayer reflector that substantially reflects light over the visible spectrum for at least one polarization state, the reflector comprising: a stack of light-transmissive polymer layers comprising at least a first and second alternating polymer material, the first and second materials having different refractive indices along at least one in-plane axis, the polymer layers having an average optical thickness of at least about ( 5/4).lamda..sub.0, where .lamda..sub.0 is a visible wavelength of interest, the polymer layers contributing substantially to the reflectivity of the multilayer reflector; wherein the polymer layers in the stack have a non-uniform distribution of optical thicknesses along a thickness axis of the reflector; wherein the refractive index difference along an in-plane x-axis is .DELTA.n.sub.x and a refractive index difference along an in-plane y-axis perpendicular to the x-axis is .DELTA.n.sub.y, and where |.DELTA.n.sub.x|>|.DELTA.n.sub.y|.

26. The reflector of claim 25, wherein |.DELTA.n.sub.y| is 0 or no greater than 0.01.

27. The reflector of claim 26, wherein the refractive index difference along a z-axis perpendicular to both the x- and y-axes is .DELTA.n.sub.z, and |.DELTA.n.sub.z| is 0 or no greater than 0.01.

28. The reflector of claim 26, wherein the refractive index difference along a z-axis perpendicular to both the x- and y-axes is .DELTA.n.sub.z, and |.DELTA.n.sub.z|>|.DELTA.n.sub.y|.

29. The reflector of claim 28, wherein the distribution of optical thicknesses is selected to decrease variability in transmission of p-polarized light obliquely incident in the y-z plane over visible light wavelengths.

30. A multilayer reflector that substantially reflects light over the visible spectrum for at least one polarization state, the reflector comprising: a stack of light-transmissive polymer layers comprising at least a first and second alternating polymer material, the first and second materials having different refractive indices along at least one in-plane axis, the polymer layers having an average optical thickness that is greater than a maximum wavelength of light in the visible spectrum, the polymer layers contributing substantially to the reflectivity of the multilayer reflector; wherein the polymer layers in the stack have a non-linear distribution of optical thicknesses along a thickness axis of the reflector.

31. The reflector of claim 30, wherein the non-linear distribution includes a non-uniform distribution.

32. A multilayer reflector that substantially reflects light over the visible spectrum for at least one polarization state, the reflector comprising: a stack of light-transmissive polymer layers comprising at least a first and second alternating polymer material, the first and second materials having different refractive indices along at least one in-plane axis, the polymer layers having an average optical thickness of at least about 2.lamda..sub.0, where .lamda..sub.0 is a visible wavelength of interest, the polymer layers contributing substantially to the reflectivity of the multilayer reflector; wherein the polymer layers in the stack have a non-uniform distribution of optical thicknesses along a thickness axis of the reflector; wherein the refractive index difference along an in-plane x-axis is .DELTA.n.sub.x and a refractive index difference along an in-plane y-axis perpendicular to the x-axis is .DELTA.n.sub.y, and where |.DELTA.n.sub.x|.apprxeq.|.DELTA.n.sub.y|.

---

Description

---

FIELD OF THE INVENTION

The present invention relates to reflective optical bodies including but not limited to films, sheets, and plates, particularly those used in visible light applications such as liquid crystal display (LCD) devices and other electronic display devices, as well as to methods of making and using such optical bodies.

BACKGROUND

Thin film multilayer reflectors suitable for reflecting visible or near-visible light are known. Such reflectors have long been made by evaporating thin films of different inorganic dielectric light-transmissive materials in succession on a glass substrate in a vacuum chamber. The different refractive indices of adjacent layers form tens or hundreds of interfaces, each of which reflects light by Fresnel reflection, and coherent constructive or destructive interference of reflected light components provides the reflector with its reflection and transmission properties. It is also know to make thin film multilayer reflectors by an extrusion process in which a multitude of alternating light-transmissive polymer materials are coextruded through a die, optionally passed through one or more layer multipliers, then cast onto a casting wheel or surface, and subsequently uniaxially or biaxially stretched. Such a technique can be used to make all-polymeric thin film reflective optical bodies, such as reflective polarizing films and reflective mirror films. See, for example, U.S. Pat. No. 5,486,949 (Schrenk et al.); U.S. Pat. No. 5,882,774 (Jonza et al.); U.S. Pat. No. 6,531,230 (Weber et al.); and U.S. Pat. No. 6,827,886 (Neavin et al.). In contrast to vacuum-coated inorganic dielectric thin film stacks, the multilayer reflectors made by polymer coextrusion techniques do not require a separate substrate for formation or handling.

Because of their reliance on coherent constructive or destructive interference of light from neighboring layer interfaces, and because such constructive or destructive interference is a strong function of the individual layer thicknesses, as well as other geometric factors, great care is typically needed to ensure that the layers are controlled to within a narrow tolerance of a design goal to ensure proper operation of the thin film interference device. As the physical size of the thin film device increases--such as for polymeric thin film reflective polarizers or broadband mirrors used in LCD devices, where the demand for larger screen sizes continues to grow--the need for such layer control can be even more important. Increasing the physical size of thin film reflective polarizers and mirrors that are manufactured in the form of thin, flexible all polymeric sheets or films also magnifies potential mechanical problems such as wrinkling, warping, and delamination.

Certain "thick film" multilayer reflectors are also known. These reflectors, which are generally associated with incoherent light reflection, are variously defined in the literature, for example, structures whose individual layers have an optical thickness of at least 0.45 micrometers, or structures whose individual layers have an average optical thickness of at least 5/4 times the average wavelength of light to be reflected. In any case, principles of incoherent light reflection inform the skilled artisan that, for most practical situations, thick film multilayer reflectors have peak reflectivities well below those achievable by their thin film counterparts, and thus the former are often considered inferior to the latter. Further, several references suggest that thick film multilayer stacks provide incoherent light reflection regardless of how thick the individual layers are, and that the layer thicknesses of a thick film multilayer stack have substantially no effect on such a stack's reflectivity.

BRIEF SUMMARY

We have found that the layer thickness and the layer thickness profile of thick film multilayer reflectors can have a substantial effect on optical performance in the wavelength range of interest, such as the visible region. In effect, certain multilayer constructions that would be expected to provide smooth incoherent light reflection because of the relatively thick nature of the individual layers, have been found to produce substantial variability in reflection and transmission characteristics, even when averaging those characteristics using a suitable bandwidth smoothing filter. Such variability can have deleterious consequences in some applications.

The present application therefore teaches, inter alia, that in order to provide desirable optical performance, such as more uniform reflection and/or transmission characteristics, the layer thickness profile even of a thick film multilayer reflector can be modified or tailored to achieve such characteristics. In particular, more uniform characteristics can be achieved by making non-uniform the optical thickness distribution of the individual layers that make up the reflector. Exemplary non-uniform distributions include graded distributions and randomized distributions. In this regard, the term "randomized", "random", and the like includes but is not limited to a purely statistical definition, and also includes arrangements of layer optical thickness that are mixed or scrambled to the extent that little or no regular patterns are formed. We have found that tailored non-uniform distributions can be particularly effective for multilayer reflectors whose individual layer optical thicknesses are moderately thick, meaning having a thickness from about 5/4 to about 5 or 10 times the wavelength of interest.

These and other aspects of the present application will be apparent from the detailed description below. In no event, however, should the above summaries be construed as limitations on the claimed subject matter, which subject matter is defined solely by the attached claims, as may be amended during prosecution.

BRIEF DESCRIPTION OF THE DRAWINGS

Throughout the specification, reference is made to the appended figures, wherein like reference numerals designate like elements, and wherein:

FIG. 1 is a schematic cross-sectional view of a conventional quarter-wave thin film multilayer reflector;

FIG. 2 is a graph of a graded layer thickness profile for a conventional thin-film multilayer reflector;

FIG. 3 is a schematic cross-sectional view of a thick film multilayer reflector;

FIG. 4a is a graph of modeled transmission, plotted as optical density, versus wavelength for a thick film multilayer reflector having a uniform layer thickness distribution;

FIG. 4b is a graph of modeled transmission, plotted as optical density, versus wavelength for a thick film multilayer reflector having a non-uniform layer thickness distribution;

FIGS. 5a and 5b are layer thickness profiles of the thick film multilayer reflectors associated with the optical density data of FIGS. 4a and 4b, respectively;

FIG. 6 is a graph of three layer thickness profiles, one uniform, one graded, and one randomized, each such profile consisting of N=100 light-transmissive layers;

FIGS. 7a-c are graphs of modeled reflectivity of normally incident block state light versus wavelength for the thickness profiles of FIG. 6, for a thickness scale factor X=1, where the reflectivity data has been spectrally smoothed with a 20 nm wide running average;

FIGS. 8a-c are graphs of modeled reflectivity of normally incident block state light versus wavelength for the thickness profiles of FIG. 6, for a thickness scale factor X=50, where the reflectivity data has again been spectrally smoothed with a 20 nm wide running average;

FIG. 9 shows the average of the modeled, smoothed reflectivity over the visible spectrum for selected values of the scale factor X from 1 to 500;

FIG. 10 shows the variability of the modeled, smoothed reflectivity over the visible spectrum for selected values of the scale factor X from 1 to 500;

FIG. 11 is a graph of three layer thickness profiles, one uniform, one graded, and one randomized, each such profile consisting of N=300 light-transmissive layers;

FIGS. 12a-c are graphs of modeled reflectivity of normally incident block state light versus wavelength for the thickness profiles of FIG. 11, for a thickness scale factor X=1, where the reflectivity data has been spectrally smoothed with a 20 nm wide running average;

FIGS. 13a-c are graphs of modeled reflectivity of obliquely incident p-polarized pass state light versus wavelength for the thickness profiles of FIG. 11, for a thickness scale factor X=1, where the reflectivity data has again been spectrally smoothed with a 20 nm wide running average;

FIGS. 14a-c correspond to FIGS. 12a-c respectively, except for a thickness scale factor X=10;

FIGS. 15a-c correspond to FIGS. 13a-c respectively, except for a thickness scale factor X=10;

FIG. 16 shows the average of the modeled, smoothed reflectivity of normally incident block state light over the visible spectrum for selected values of the scale factor X from 1 to 500;

FIG. 17 shows the variability of the modeled, smoothed reflectivity of normally incident block state light over the visible spectrum for selected values of the scale factor X from 1 to 500;

FIGS. 18 and 19 correspond to FIGS. 16 and 17 respectively, except for obliquely incident p-polarized pass state light; and

FIG. 20 is a schematic illustration of a uniaxial stretching apparatus and process.

DETAILED DESCRIPTION OF THE ILLUSTRATIVE EMBODIMENTS

FIG. 1 shows a portion of a conventional thin film multilayer reflector 10. Reflector 10 is all polymeric and has a stack or packet of optically thin layers sandwiched between optically thick outer skin layers 12. The layer thicknesses shown in the figure are intended to be representative of optical thickness, as described below, rather than physical thickness. The reflector 10 is shown in relation to a Cartesian x-y-z coordinate system, with the z-axis corresponding to a thickness axis of the reflector and the x- and y-axes extending parallel to the plane of the layers and the interfaces between layers. Arrow 14 represents light incident on the film at an angle .theta. relative to the z-axis, with reflected light represented by arrow 14a and transmitted light by arrow 14b.

In the finished product, the optically thin layers are conventionally arranged in a repeating pattern along the z-axis of the reflector. The smallest unit of the repeating pattern is referred to as a unit cell or optical repeat unit. In a simple quarter-wave stack (see again FIG. 1), "A" layers composed of one material are interspersed with "B" layers composed of a different material, making the pair "AB" the optical repeat unit 16. In other cases the optical repeat unit can be more complex, such as a four-layer repeat unit (ABCB) as discussed in U.S. Pat. No. 5,103,337 (Schrenk et al.), or a six-layer repeat unit (7A1B1A7B1A1B) as discussed in U.S. Pat. No. 5,360,659 (Arends et al.).

The individual layers of these films have optical thicknesses--defined as the physical thickness multiplied by the appropriate refractive index of the individual layer--of less than a design wavelength of light .lamda..sub.0, such that constructive or destructive interference for light components reflected at the interfaces between individual layers can occur coherently to produce the desired overall reflectivity at .lamda..sub.0. More specifically, the optical thickness of each optical repeat unit, which equals the sum of the optical thicknesses of its component layers, corresponds to one-half of the design wavelength. In the reflector 10 of FIG. 1, each optically thin A and B layer is nominally a quarter-wave thick, or about .lamda..sub.0/4.

Thin film reflectors can utilize a variety of layer thickness profiles, for example, a thickness gradient, such that the optical thickness of the optical repeat units changes along a thickness axis of the stack in a prescribed manner, in order to achieve desired reflectivity characteristics such as expanding the spectral width of a reflectance band or sharpening the transition edge of such a band. See, e.g., U.S. Pat. No. 6,583,930 (Schrenk et al.) and U.S. Pat. No. 6,157,490 (Wheatley et al.). FIG. 2 shows an example of such a profile, where the horizontal axis plots the number of the individual optical repeat units counted from one side of the multilayer stack to the other, and the vertical axis plots optical thickness of the corresponding optical repeat unit.

Because of their reliance on coherent constructive or destructive interference of light from neighboring layer interfaces, and because such constructive or destructive interference is intimately related to geometrical factors of the illuminated optical body such as layer thickness profile, angle of incidence, polarization, wavelength, and the like, thin film multilayer reflectors can exhibit substantial variability in reflectivity and transmission as a function of these factors. In some cases this variability is not only desirable but necessary for the intended application. In other cases at least some of the variability may be undesirable. As mentioned previously, the known sensitivity of thin film multilayer stacks to individual layer thickness also requires manufacturers to ensure precise thickness control of such stacks.

Optically thick film multilayer reflectors differ from their thin film counterparts in that the individual layers are so thick that a first and second light component reflecting from adjacent interfaces in the optical body presumably combine substantially incoherently from the standpoint of the wavelength of interest. One example is the "pile-of-plates" polarizer. Where the wavelength of interest is human-visible light, for example, this means that, for a visible light ray or beam that impinges upon two adjacent interfaces to produce two corresponding reflected and transmitted light components, an observer viewing the resulting reflected or transmitted light will notice no change in brightness or color (spectral distribution) of the observed light if the layer thickness is varied a small amount. See also, for example, U.S. Pat. No. 5,122,905 (Wheatley et al.), which states in connection with an optically thick multilayer reflector that "the individual layers should have an optical thickness such that no visibly perceived iridescence is reflected from the body". The '905 Wheatley reference describes thick layers as those whose optical thickness is at least 0.45 micrometers. As another example, U.S. Pat. No. 5,808,798 (Weber et al.) describes a "pile of plates" or "thick film" stack of alternating layers of materials A and B in which the average optical thickness of the layers in the stack is at least 5/4 times the average wavelength of the light to be reflected.

Because of the nature of incoherent light interaction, it has been proposed that layer thickness and the layer thickness profile has substantially no effect on the performance of a thick film multilayer reflector. The '905 Wheatley et al. reference states, for example: "Thus, the reflected wavelength of light from the multilayer polymeric body of the present invention is independent of both individual layer and total structure thickness over a wide processing range so long as a substantial majority of the individual layers has an optical thickness equal to or greater than about 0.45 micrometers. Uniformity of reflection is inherent in the design of the body. Moreover, a gradient of layer thickness through the thickness of the body is neither detrimental nor advantageous to the appearance of the body as long as a substantial majority of the individual layers of the polymers maintains an optical thickness equal to or greater than about 0.45 micrometers." As another example, the '798 Weber et al. reference states in connection with a nonpolarizing beamsplitter comprising a "pile of plates" or "thick film" stack, that such a stack has "no wavelength selectivity", and a ratio of p-polarization reflectivity R.sub.p to s-polarization reflectivity R.sub.s "is determined by material properties only, and cannot be significantly affected by layer thickness."

A portion of a thick film multilayer reflector 20 is shown schematically in FIG. 3, where layer thickness is again depicted as an optical thickness rather than a physical thickness. The figure shows a central group of N light-transmissive layers bounded by outer light-transmissive layers 22, for a total of N+2 layers. In some cases the outer layers 22 may be distinguishable in some way from the central N layers. For example, the central N layers may consist essentially of two alternating coextruded light-transmissive polymer materials A,B, and the outer layers may both be made of a different light-transmissive polymer material selected for its optical, mechanical, or chemical properties. See, e.g., U.S. Pat. No. 6,368,699 (Gilbert et al.). The outer layers 22 may, for instance, provide scratch resistance via a hard coating composition, UV protection via ultraviolet absorbers or inhibitors in a matrix material, antistatic properties, slip properties via slip agents, appearance-modifying properties via diffusing agents, colorants, dyes, pigments, and the like, adhesion via heat activated or pressure sensitive adhesive compositions, and/or warp resistance. The outer layers 22 may also have thicknesses substantially different from that of the central N layers, whether substantially thinner or substantially thicker. Alternatively, the outer layers 22 may be indistinguishable from the central N layers, in which case they are simply the endpoints of a pattern established by the other layers. One or both of outer layers 22 may also be omitted. Although the reflector 20 can be composed of as many light transmissive materials as the number of constituent layers in the reflector (e.g. N,N+1, or N+2), with no two layers having the same composition, it is usually more practical to arrange two, three, or another small number of light-transmissive materials in an alternating fashion such as . . . ABAB . . . or . . . ABCABC . . . or the like.

Each layer has refractive indices for light polarized along the x-, y-, and z-axis of n.sub.x, n.sub.y, and n.sub.z respectively. Differences in refractive index between adjacent layers along these axes, which differences may in general be zero or nonzero, are .DELTA.n.sub.x, .DELTA.n.sub.y, and .DELTA.n.sub.z, respectively. For a reflective polarizer, .DELTA.n.sub.y is zero or has a magnitude small relative to that of .DELTA.n.sub.x, where one can use the arbitrary convention of assigning the x-axis to the direction of maximum in-plane refractive index difference. In that case the x-axis corresponds to the block axis of the polarizer and the y-axis corresponds to the pass axis. In polymeric constructions it is often convenient to make the polarizer with "A" layers that develop stress-induced birefringence during stretching and "B" layers that remain isotropic. This is not, however, a requirement, since it is also possible for both layer types to develop stress-induced birefringence, as long as one in-plane refractive index difference (.DELTA.n.sub.x) between adjacent layers is significantly greater than another in-plane index difference (.DELTA.n.sub.y).

In FIG. 3, the individual layers are shown as having optical thicknesses somewhat greater than the design wavelength .lamda..sub.0, for example, (5/4).lamda..sub.0 or more, and at any rate of a thickness that would be considered optically thick by the person of ordinary skill in the art. At least some of the layer thicknesses can also be on the order of .lamda..sub.0 or even less, and some, most, or substantially all can be greater than (5/4).lamda..sub.0, or greater than 2.lamda..sub.0, 5.lamda..sub.0, or 10.lamda..sub.0 or more, with no strict upper limit. In some cases it may also be desirable to combine one or more thick film stack with one or more thin film stack (such as a quarter-wave interference stack) to make a hybrid multilayer reflective body. Thick film multilayer reflectors can also be described as those having a number N of individual layers whose optical thicknesses, or the average thereof, are at least (5/4).lamda..sub.0, and N is large enough such that these optically thick layers contribute substantially to the reflectivity and/or transmission of the reflector. For example, the number N may be large enough that the optically thick layers are responsible for at least half of the reflectivity at the design wavelength .lamda..sub.0, or even for substantially all of the reflectivity at the design wavelength. As will be seen below, an optical thickness range of particular interest is from about (5/4).lamda..sub.0 to about 5.lamda..sub.0 or 10.lamda..sub.0, where for operation over the entire visible spectrum the design wavelength .lamda..sub.0 can be taken to be approximately in the middle of the visible region (about 550 nanometers) or can be taken as a long wavelength end of the visible region (e.g., approximately 700 nm). Thick film layers whose optical thickness is in this range are referred to herein as being moderately thick.

The disclosed reflectors are capable of reflecting a substantial amount of light of at least one polarization over substantially the entire visible spectrum. For example, the disclosed reflectors desirably have a normal incidence average reflectivity for at least one polarization state from about 400-700 nm of at least about 30%, 40%, or even 45%. This can be accomplished by controlling the refractive index relationships of the individual light-transmissive layers to achieve sufficiently high refractive index differences between layers, in combination with ensuring a sufficient number N of such layers in the stack.

In a process of designing a thick film multilayer reflector, the individual layer thicknesses may be determined or calculated by first specifying a desired overall thickness of the finished multilayer reflector and a desired reflectivity or transmissivity of the reflector, provided the specific light-transmissive polymer materials have predictable refractive index characteristics. Depending on whether the desired reflector is to be made with a uniaxial stretching process (whether constrained or unconstrained, in order to make, for example, a polarizer that substantially reflects one polarization state of normally incident light and substantially transmits the orthogonal polarization state) or with a biaxial stretching process (in order to make, for example, a balanced mirror that reflects orthogonal polarization states of normally incident light substantially equally, or an unbalanced mirror), then details of the stretching process and knowledge of the polymer material properties will inform the designer what layer-to-layer refractive index differences can be expected under those processing conditions. With the refractive index information, the number of layers N needed to provide a desired reflectivity, for example, can then be estimated. Then, if the reflector has a target overall physical thickness D, for example in the range from 1 to 10 mm or even from 1 to 4 mm so as to achieve relative mechanical stiffness but in a relatively low profile plate, then the nominal physical thickness of the individual layers can be estimated at D/N. Alternatively, if it is desired to make as thin a reflector as possible, a nominal optical thickness on the order of about (5/4).lamda..sub.0 can be selected, resulting in an overall physical thickness for the reflector on the order of about (5N.lamda..sub.0)/(4n), where n is an average refractive index of the different light-transmissive materials used for the various layers.

FIGS. 4a and 4b demonstrate the sensitivity of thick film multilayer reflectors to the layer thickness distribution. Data in these figures was generated using TFCalc, an optical design software package available from Software Spectra, Inc. of Portland, Oreg. The figures plot optical density versus wavelength, where optical density or OD=log.sub.10(1/T), and T is transmission on a scale of 0 (perfectly absorbing or otherwise non-transmitting) to 1 (perfectly transmitting). A bandwidth smoothing filter of bandwidth .DELTA..lamda.=20 nm was also used, such that the value of T at each wavelength .lamda. was calculated as an average over a 20 nm wavelength band centered on .lamda.. This smoothing filter is used to ensure that modeled transmission variations are meaningful from the standpoint of a human observer. Smoothing filters with different bandwidths .DELTA..lamda. greater or less than 20 nm may also be used. In some cases, such as where the thick film reflector is combined in a system with a very narrow band source or other component, it may be appropriate to use a bandwidth .DELTA..lamda. of 1 nm or less, or no smoothing filter at all. Beyond this, curve 30a in FIG. 4a assumes the following parameters: a stack consisting of exactly 200 layers (N=200); each layer of the stack has an optical thickness of 1.375 micrometers (hence about 2.5.lamda..sub.0, where .lamda..sub.0=550 nm.). The thickness profile of the stack is uniform, as shown in FIG. 5a; the layers are arranged in an alternating AB pattern, with half the layers having an in-plane refractive index of 1.85 and the other half having an in-plane refractive index of 1.56. These values are representative of a uniaxially stretched alternating PEN/coPEN reflective polarizer construction, where the stretch is unconstrained to permit complete relaxation of the optical body in the y- and z-directions, assuming stretching is along the x-direction, so that .DELTA.n.sub.x=0.19 and .DELTA.n.sub.y=.DELTA.n.sub.z=0. The layers are assumed to have no absorption, hence at any given wavelength, transmission plus reflection equals 100% (T+R=100%); the stack is immersed in air, with index of refraction of 1.0; and light is incident normally on the stack, thus .theta.=0. Where the stack is a reflective polarizer, the curve 30a represents the optical density of a polarization state associated with the stated 1.85/1.56 refractive indices of the layers, e.g. light polarized along the x-axis. Alternatively, if the stack were a balanced mirror for which .DELTA.n.sub.x=.DELTA.n.sub.y=0.19, curve 30a would represent the optical density of x-polarized, y-polarized, and unpolarized normally incident light.

As seen in the graph, curve 30a exhibits a low baseline optical density of less than 0.1, with three peaks in the visible region, each less than 0.4 optical density, that result from three narrow reflection peaks. The variability in optical density over the visible region from 410-700 nm can be seen to have a magnitude between 0.3 and 0.4, but the corresponding smoothed percent reflectivity (not shown) varies greatly over that visible region, ranging from a low of 5% (optical density 0.02) to a high of 59% (optical density 0.38), and averaging 19%, with corresponding variability in smoothed percent transmission.

The initially uniform layer thickness distribution was then modified using an optimization routine (variable metric method) to increase the average optical density over the visible region from 400-700 nm. All other parameters used in connection with curve 30a were the same. The modified layer thickness distribution is shown in FIG. 5b. This modified distribution is characterized by a minimum, maximum, and average optical thickness of 0.95, 1.74, and 1.35 micrometers respectively, corresponding to about 1.7, 3.2, and 2.5.lamda..sub.0, where .lamda..sub.0=550 nm. The average optical thickness differed slightly from that of FIG. 5a because the optimization routine was allowed to adjust optical thickness of each of the 200 layers without requiring the average layer thickness to remain constant.

The optical behavior of the modified layer thickness distribution is shown by curve 30b in FIG. 4b. As intended, the optical density averaged over the visible range is increased for the modified multilayer optical body: average optical density from 410-700 nm is 0.58 for curve 30b, and 0.10 for curve 30a. As a further benefit, the smoothed percent reflectivity and transmission (not shown) associated with curve 30b exhibit substantially less variability over visible wavelengths than those associated with curve 30a. The smoothed percent reflectivity associated with curve 30b ranged, over the 410-700 nm region, from a low of 60% (optical density 0.39) to a high of 84% (optical density 0.80), and averaging 73%, compared with the range from 5 to 59% mentioned above for curve 30a.

The increased average optical density and decreased variability in smoothed percent reflectivity and transmission is achieved by changing the uniform layer thickness distribution of FIG. 5a to the non-uniform, randomized distribution of FIG. 5b. In this regard, the term "randomized" is not intended to be limited to a strictly statistical definition, but also include arrangements of layer optical thickness that are mixed or scrambled to the extent that little or no regular patterns are formed. One can, for example, divide the entire layer thickness profile into segments, such as thirds, quarters, fifths, or tenths, and calculate the average optical or physical thickness of the layers in each of the segments. For many randomized distributions the average thickness of the different segments will be substantially the same. For example, the average thickness of a given segment may differ from the average thicknesses of each of the remaining segments by no more than the amplitude or the standard deviation of thickness variation in the given segment. Further, the minimum or maximum layer thickness of a given segment may be less than or greater than (respectively) the average thickness of all of the remaining segments.

Reducing transmission and reflection variability as a function of wavelength, particularly where the transmission and reflection data has been smoothed by averaging over a visually meaningful spectral width .DELTA..lamda. such as 10, 20, or even 50 nm, reduces the amount of noticeable color that the reflector provides to the display or other system in which it is used.

Reducing transmission and reflection variability can also be important when using the reflector in narrow band applications, for example, applications that utilize narrow band components such as narrow band sources, detectors, or filters. By "narrow band" we mean a band that is narrow relative to the visible wavelength spectrum. In some cases such a band may have a full-width-at-half-maximum (FWHM) spectral width on the order of about 50, 20, or even 10 nm or less, and in some cases may be on the order of about 5, 2, or even 1 nm or less. Furthermore, we mean not only components that operate in only one such narrow band, such as a single mode laser light source, but also components that operate over a plurality of distinct narrow bands such as certain cold cathode fluorescent lamp white light sources that emit over the visible wavelength range but in a plurality of discrete narrow bands, or a high finesse Fabry-Perot filter element. When such narrow band sources or other components are used with a conventional thick multilayer reflector, small spectral shifts or variations can lead to extreme changes in system behavior as, for example, the narrow band emission from a narrow band source aligns itself with a local minimum and then with a local maximum in transmission or reflection. Such spectral shifts or variations may be due to any one or a combination of unit-to-unit variability, angular effects, thermal effects, spatial nonuniformities (e.g. caliper variations) of the multilayer reflector, or the like, and they may not be apparent by a cursory visual inspection of the thick multilayer reflector under normal lighting conditions, particularly if the variability occurs over a wavelength scale too small for human visual detection. By tailoring the layer thickness profile to reduce the spectral variability of the multilayer reflector, such extreme system changes can likewise be reduced.

Turning now to FIG. 6, we see there a graph of optical thickness versus layer number for a uniform layer distribution 40 plotted with square-shaped points, a graded layer distribution 42 plotted with square-shaped points, and a randomized layer distribution 44 plotted with diamond-shaped points. Each distribution 40, 42, 44 has 100 points, hence N=100 total layers for the associated multilayer reflectors. Furthermore, distribution 40 has an average (and uniform) layer optical thickness of 137.5 units (.apprxeq.550/4), while distributions 42 and 44 have slightly higher average layer optical thicknesses of 165.7 units (.apprxeq.650/4). Still further, each point of the random distribution 44 has a one-to-one correspondence with a point of the graded distribution 42. That is, the layers in distribution 44 are the very same layers as those of distribution 42 except their order in the stack has been shuffled or mixed to produce the random distribution. Each of the three distributions, however, maintains an alternating AB repeat pattern throughout the stack. In the modeling, half the layers (the "A" layers) had a high refractive index of 1.84, the other half ("B" layers) had a low refractive index of 1.56. These indices are for light polarized along an in-plane x-axis, and are again typical of an unconstrained uniaxially stretched PEN/coPEN two-polymer system. The refractive indices for the mutually perpendicular y- and z-axes can all equal 1.56, or can have other values, but had no effect on the modeling that was performed. The model further assumed that the layers had no absorption, and that the stack was immersed in air, and that light impinged normally on the reflector.

The unit of length for optical thickness on the vertical axis of FIG. 6 is shown as having a scale factor, labeled "X". The variable X can thus be used as a multiplier to investigate the impact of overall layer thickness or dimension on the effect of randomizing the layer thickness distribution. For example, for X=1 or X=10, the layer distribution 40, has an average optical thickness of 137.5 nm or 1,375 nm, respectively, while distributions 42, 44 each have an average optical thickness of 165.7 nm or 1,657 nm, respectively.

In a first case, X was set to 1 such that the layers were optically thin for each of the three modeled reflectors. The spectral reflectivity was calculated for each layer distribution 40, 42, 44, and a smoothing operation was performed so that the reflectivity reported at any given wavelength .lamda. represents an average over a band .DELTA..lamda.=20 nm wide centered on the wavelength .lamda.. The results are shown in FIGS. 7a-c. Curve 41 is the smoothed reflectivity for the uniform layer distribution 40, curve 43 is the smoothed reflectivity for the graded layer distribution 42, and curve 45 is the smoothed reflectivity for the randomized layer distribution 44. The high maximum reflectivities of well over 50% in the visible for these curves are common for coherent multilayer light reflection, with curves 41 and 45 approaching 100% at selected wavelengths. Note that the uniform and randomized layer distributions produce substantially higher variability in reflectivity over the visible spectrum than the graded layer distribution.

In a second case, the scale factor X was set to 50 to provide optically thick layers, but having the same relative profile as the first case. Smoothed spectral reflectivity was calculated as before, and is shown in FIGS. 8a-c, where curve 51 is for uniform layer distribution 40, curve 53 is for graded layer distribution 42, and curve 55 is for randomized layer distribution 44. Note the maximum reflectivities of the three curves in the visible, generally slightly above 50%, and substantially below that of the curves of FIGS. 7a-c, which is characteristic of incoherent multilayer light reflection. Note also that, even for such optically thick layers, the uniform layer distribution produces significant variability over the visible wavelength range (FIG. 8a). In contrast, the non-uniform layer distributions provide variability of much smaller amplitude (FIGS. 8b, 8c).

The above two cases (X=1 and X=50) were then repeated for the additional cases of X=1.5, 2, 3, 5, 10, 100, and 500, and the smoothed spectral reflectivity was calculated as before. For each value of X and for each of the three profiles of FIG. 6, the average R.sub.ave and the standard deviation R.sub.std of the smoothed reflectivity data was calculated over the visible wavelength range 410-700 nm, i.e., about 400-700 nm. From these numbers a variability parameter, or coefficient of variation ("COV"), was calculated as COV=6*R.sub.std/R.sub.ave. The results are shown in FIGS. 9 and 10, which plot R.sub.ave and COV respectively for each value of scale factor X. These graphs have a linear vertical axis but a non-linear horizontal axis. Curves 61 and 71 are for the uniform layer distribution 40, curves 63 and 73 are for the graded layer distribution 42, and curves 65 and 75 are for the randomized layer distribution 44. As can be seen by inspection of the figures, the non-uniform layer distributions 42 and 44 can provide substantial benefits relative to the uniform distribution, i.e., generally higher average reflectivity and generally lower spectral variability over the visible spectrum. Of particular note are the following regions: X=3 to 50, X=5 to 50, X=3 to 10, and X=5 to 10. Keeping in mind that for the values of X=3, 5, 10, and 50, the average layer optical thickness for the uniform layer distribution 40 is about 413, 688, 1,375, and 6,875 nm, respectively, and that the average layer optical thickness for each of the non-uniform distributions 42, 44 is about 503, 838, 1,675, and 8,375 nm, respectively, these ranges of interest include average layer optical thickness ranges from about ( 5/4).lamda..sub.0 to 10.lamda..sub.0, and from about ( 5/4).lamda..sub.0 to 5.lamda..sub.0, for .lamda..sub.0 in the visible, for example, for .lamda..sub.0 of roughly 550 nm.

From the foregoing, providing a thick film multilayer reflector with a non-uniform optical thickness profile can be seen to provide desirable reflectivity and transmission properties, such as reduced spectral variability for design wavelengths in the visible region, relative to thick film reflectors that have a uniform layer thickness profile. For applications involving human visual response, reduced spectral variability yields a reflector with less perceived color, whether viewed in transmission or reflection.

The foregoing discussion emphasizes the reflectivity and transmission characteristics of a thick film multilayer reflector for normally incident light whose polarization state encounters a sufficiently high refractive index difference .DELTA.n and a sufficient number of layers N that it is substantially reflected, e.g., on the order of about 50% or at least about 30% or 40% average over the visible region. For a thick film multilayer polarizer, this may correspond to the blocked polarization state. It is also instructive to investigate the reflectivity and transmission characteristics of the pass polarization state of a multilayer polarizer that is intended to have a high average transmission, low average reflectivity, and low spectral variability over the visible spectrum, as well as to investigate off-axis behavior.

For this purpose, the following refractive index properties were adopted for the alternating "A" and "B" layers of the thick film stack: "A" layers: n.sub.x=1.85, n.sub.y=1.61, n.sub.z=1.51; "B" layers: n.sub.x=1.61, n.sub.y=1.61, n.sub.z=1.61; hence, .DELTA.n.sub.x=0.24, .DELTA.n.sub.y=0, .DELTA.n.sub.z=-0.10 These values are representative of a PEN/coPEN polymer system made with a constrained uniaxial stretch, which produces the different y- and z-axis refractive indices in the birefringent "A" layers. It is assumed that the stretch is performed so as to force .DELTA.n.sub.y to be substantially zero, but the degree of stretch can alternatively be adjusted to provide a suitable balance between the magnitude of .DELTA.n.sub.y and .DELTA.n.sub.z.

Also, the number of total light-transmissive layers was increased from 100 to 300. FIG. 11, which plots optical thickness of each layer versus layer number from one end of the stack to the other, shows the three new thickness distributions modeled, i.e., a uniform distribution 80, a graded distribution 82, and a randomized distribution 84. The distribution 80 has an average (and uniform) layer optical thickness of 137.5 units (.apprxeq.550/4), while distributions 82 and 84 have slightly higher average optical thicknesses of 165.7 units (.apprxeq.650/4). As before, each point of the random distribution 84 has a one-to-one correspondence with a point of the graded distribution 82, i.e., the layers in distribution 84 are the very same layers as those of distribution 82 except their order in the stack has been shuffled or mixed to produce the random distribution. Each of the three distributions 80, 82, and 84 maintains an alternating AB repeat pattern throughout the stack. The model again assumed that the stack was immersed in air, and that the layers had no absorption, so that percent transmission plus percent reflection (T+R) equals 100%. As in FIG. 6, the unit of length for optical thickness on the vertical axis of FIG. 11 is shown as having a scale factor "X", to investigate the impact of overall layer thickness or dimension on the effect of randomizing the layer thickness distribution.

In a first case relating to this constrained polarizer, X was set to 1 so that the layers were optically thin for each of the three modeled reflectors. The spectral reflectivity was calculated for each layer distribution 80, 82, 84, and a smoothing operation was performed so that the reflectivity reported at any given wavelength .lamda. represents an average over a band .DELTA..lamda.=20 nm wide centered on the wavelength .lamda.. These reflectivity calculations were done for two types of incident light: (1) normal incidence, block polarization state; and (2) oblique incidence (.theta.=60.degree., see FIG. 3) pass polarization state. The normally incident light was thus incident parallel to the z-axis and polarized parallel to the x-axis, while the obliquely incident light was incident in the y-z plane and p-polarized in this plane of incidence. Thus, both the blocking properties and the pass properties of this constrained thin film multilayer polarizer were investigated. The results for the block state normally incident light are shown in FIGS. 12a-c, which show the behavior of uniform, graded, and randomized layer distributions (80, 82, 84) respectively. The results for the obliquely incident p-polarized light are shown in FIGS. 13a-c, which show the behavior of uniform, graded, and randomized layer distributions (80, 82, 84) respectively.

In a second case relating to this constrained polarizer, X was set to 10 so that the layers were optically thick for each of the three modeled reflectors. The spectral reflectivity was calculated for each layer distribution 80, 82, 84, and a smoothing operation was performed as described above with .DELTA..lamda.=20 nm. The reflectivity calculations were again done for: (1) normal incidence, block polarization state light (incident parallel to the z-axis and polarized parallel to the x-axis); and (2) .theta.=60.degree. oblique incidence pass polarization state light (incident in the y-z plane and p-polarized in this plane). The results for the block state normally incident light are shown in FIGS. 14a-c, which show the behavior of uniform, graded, and randomized layer distributions (80, 82, 84) respectively. The results for the obliquely incident p-polarized light are shown in FIGS. 15a-c, which show the behavior of uniform, graded, and randomized layer distributions (80, 82, 84) respectively.

The above two cases (X=1 and X=10) relating to the constrained polarizer were then repeated for the additional cases of X=1.5, 2, 3, 5, 50, 100, and 500, and the smoothed spectral reflectivity was calculated as before for both the normally incident block state light and the 60 degree incident p-polarized light. For each value of X, each type of incident light modeled, and each of the three profiles of FIG. 11, the average R.sub.ave and the standard deviation R.sub.std of the smoothed reflectivity data was calculated over the visible wavelength range 410-700 nm (ab