Compensation of birefringence in lead-free polarizing beam splitters for LCOS projectors

Transcription

1 Compensation of birefringence in lead-free polarizing beam splitters for LCOS projectors David J. W. Aastuen Charles L. Bruzzone Abstract Most optical designs for delivering light to LCOS imagers and then from the imagers to the projection lens use polarizing-beam-splitter (PBS) technology. Most of the PBSs used in commercial LCOS projectors contain glass with a significant amount of lead (Pb). Such glasses have inherently low stress birefringence, and therefore maintain the polarization state of light passing through them. However, Pb-bearing glass is an expensive, difficult to process, and hazardous material with special disposal requirements and is therefore not desirable in consumer-electronic products. On the other hand, Pb-free wire-grid plate PBSs require a longer back focal length than would be optimal. Data and modeling results show that uniform high-contrast dark states may be obtained from lead-free-glass Cartesian PBS prisms when a quarter-wave compensator is used between the imager and the PBS. Keywords PBS, LCOS, polarization, compensation. 1 ntroduction The function of a PBS is to reflect one polarization of light and transmit the other. Typically, the polarization selective surface is placed on the 45 diagonal of a right rectangular prism. The PBS typically transmits nominally p-polarized light andreflectsnominallys-polarized light (Fig. 1). Recently, Cartesian PBSs, such as those used in the Vikuiti Optical Core, 1 have been used for LCOS projection at lower f/#s and with higher contrast and transmission than a MacNeille PBS can achieve. This is due to the fact that a Cartesian PBS, instead of separating s-pol from p-pol light, transmits or reflects light based on the alignment of the polarization direction with fixed Cartesian axes in the PBS material itself. n what follows we will continue to speak of s- andp-polarization, however, with the understanding that in the case of a Cartesian PBS the polarization state is not actually directly dependent on the azimuth of the reflection plane, as it would be for a MacNeille PBS. 2 t has long been known that it is necessary to compensate for a number of effects to obtain high contrast for PBSbased LCOS imaging systems. For example, if a MacNeille PBS 3 is used, it is necessary to compensate for the geometrical effects arising from rotation of the reflection plane on reflection from the imager. 4,5 t is also widely understood that, in addition to these effects, imager compensation is required to achieve high contrast (over 1:1), especially with the popular vertically aligned nematic (VAN) mode imagers. Schmidt et al. 6 observed as early as 1993 that quarterwave (QW) or slightly larger retardation compensators improved system contrast for JVC LA imaging systems, and JVC shipped the first such super contrast LA projectors with QW compensation in Subsequently, in published work, 8,9 JVC indicated that QW compensation was used in their products primarily for the geometrical effects of the PBS. 4 FGURE 1 Function of the PBS. A polarizing beam splitter reflects s-polarized light and transmits p-polarized light. Another source of contrast degradation can occur if there is retardation in the input glass prism. n this case, the contrast of an imaging system can be reduced when the nominally s-polarized light becomes elliptically polarized, thereby acquiring a phase-delayed p-polarized component. This will cause leakage of light after it is reflected back from the imager, increasing the level of brightness in the dark state. Retardation in the glass may result from prism processing, or be created by stress birefringence induced by fixturing of the PBS, or by thermal expansion due to absorption of light in the PBS glass. 1 Because of this problem, Nikon 11 and other glass manufacturers developed glasses with a low stress-optic coefficient (SOC) to minimize birefringence in response to mechanical stress. PBH56 and SF57 are glasses of this type, made by Ohara and Schott, respectively, and with a very low SOC of ~ nm/cm/pa. However, such glasses always contain Pb in significant quantities; both of these glasses The authors are with 3M Optical Systems Div., 3M Center, Bldg E-54, St. Paul, MN ; clbruzzone@mmm.com. Copyright 26 Society for nformation Display /6/ $1. Journal of the SD 14/3,

2 contain >7% PbO by weight. These glasses are therefore not environmentally desirable materials, in addition to being expensive and difficult to process. The only practical Pb-free alternative has been the wire-grid-plate PBS, but these require a relatively long back focal length for the projection lens and somewhat more complicated assembly than do glass-prism PBSs. We have recently found that it is possible to use more common glasses, such as NSK5, S-BAL35, or ZK3 (equivalent glasses made by Schott, Ohara, and Chendu Guangming, with an SOC of nm/cm/pa) in PBS applications when two conditions are met: (1) The PBS must have very high transmission of the nominally p-polarized light in the incident beam, and (2) There must be a quarter-wave compensator (QWC) between the PBS and the mirror or imager. 2 Theory The somewhat surprising result described above can be understood by means of rather simple calculation. The effect of the QWC can be modeled by Mueller Matrix polarization modeling. The simplest model that shows the observed effects uses crossed linear leaky polarizers for the PBS function. Between the polarizers we place the birefringent glass, the QWC, and an LCOS imager comprising some stray retardance and a mirror. The effect is plainly evident for a ray that travels parallel to the optic axis, i.e., its angle of incidence on the modeled imager is. This confirms that we are not observing a skew-ray effect. n what follows we will model the combined effects of retardation in the imager dark state and birefringence in the glass. This is done to demonstrate that both may be compensated simultaneously. The consequence of non-zero imager retardation will be seen to be a small angle between the birefringence axis of the QWC and the polarization direction of the incident light an angle that approaches zero as the imager retardation approaches zero. The basic mechanism for compensation of the glass birefringence is that outlined here, and described mathematically in the Appendix. The steps are: 1. Pre-polarized light entering the PBS develops polarization mixing due to birefringence, becoming elliptically polarized. 2. The polarization state of the light returns to substantial linear s-polarization upon reflection from the beam-splitting surface. This depends critically on the PBS having very high transmission of p-polarized light, T p, so that only a small fraction of the p- polarized light is reflected. 3. The s-polarized light reflects from the beam-splitting surface, travels on a fixed path through the glass, and exits the PBS toward the mirror (or imager). While passing through the glass, the light may again become elliptically polarized, with both s and p components with respect to the beam-splitting surface. 4. The light passes out of the PBS glass and through the QWC. t then reflects off the mirror and through the QWC again. This is functionally equivalent to passing through a half-wave film. As a result, the handedness of the ellipticity of the light ray is reversed. 5. The light returning from the mirror or imager follows nearly the same path back to the PBS film, and therefore experiences nearly the same retardation as on the incoming path. Due to the inversion of ellipticity in step 4, the effect of the birefringence on the return path is to substantially undo the ellipticity induced on the incident path, returning the polarization state nearly to its original condition (s-polarization). The effect in step 4 is similar in nature to the effects described by Miyatake 4 when dealing with linearly polarized light in a MacNeille PBS. n that case, the QWC acts to reflect the polarization direction of the skew ray through the plane of reflection of the central ray, so that incoming s-polarized light, which must be polarized at an angle θ to the s-polarization direction of the central ray, becomes polarized at an angle θ after reflection. f this were not done, then there wouldbeafractionofp-polarized component in the outgoing ray of magnitude sin(2θ), but this action by the QWC results in the outgoing ray also having purely s-polarization. While we necessarily must deal with the phase factor and ellipticity of the light in the glass-birefringence case, the effects of the QWC are the same on the p-polarized components in both the skew-ray and birefringence cases. This compensation approach works very well for on-axis lightifthet p of the PBS film is nearly 1%. As T p drops below 1%, the effects of the birefringence prior to the FGURE 2 Transmission of p-polarized light through 3M MOF PBS film. This chart includes only losses at the polarization selective surface. Error in the measurement becomes large at short wavelengths due to the use of incandescent illumination with low blue light content. 286 Aastuen and Bruzzone / Lead-free polarizing beam splitters for LCOS

3 FGURE 3 Contrast for VAN mode imager with 5 nm of in-plane retardation at 45 to the polarization direction, as a function of the level of glass retardation also oriented at 45. The result is similar for any residual imager retardation, although the angle of QWF orientation will vary with the level of retardation. Note that in this case, even when there is no birefringence in the glass, it is necessary to have the QWC at 1 to vertical in order to obtain contrast levels over 1:1, and that orientation to 1.8 provides the optimum value of contrast. refection off the PBS film are less perfectly cleaned up at that reflection. Only the birefringence effects after reflection from the PBS film are compensated by the polarization changes caused by the QWC, so to the extent that birefringence effects prior to the PBS film reflection remain, the contrast will be degraded. Thus, it is very important that the T p be very high. Such high T p is available from 3M multi-layer optical-film (MOF) polarizing-beam-splitter technology (Fig. 2). 12 Since non-mof PBSs typically do not have a T p above 9% for extreme rays of beams with a useful F/# (say, 3. or less), this approach does not work for those PBSs. This is why the work of Cline et al. 1 with MacNeille PBSs did not discover this effect. The effectiveness of the glass-retardation compensation is also affected by the NA of the system. As the light rays diverge from on-axis, the birefringence histories for the incoming and outgoing rays become increasingly different and the ellipticity induced on the outgoing leg will not exactly match that introduced on the incoming leg. While this may eventually limit the allowable NA of the system, we will show below that performance at F/2.3 is quite adequate. Calculated contrast contours for glass birefringence oriented at 45 are shown in Fig. 3, for a typical VAN-mode imager and with a variety of levels of glass retardance. The polarizers in the calculation are leaky so that the maximum contrast is limited to 1,:1. This analysis shows that a QWC will exactly correct for any level of glass birefringence oriented at 45 to the polarization direction. Similar calculation shows that, as expected, glass birefringence oriented at or 9 has no effect on the polarization state of the beam, while glass birefringence oriented at 22.5 (modulo 45 ) to the polarization direction will be the most damaging to the contrast. This isindicatedinfig.4.notethatinfig.3 the effect of birefringence in the imager dark state is to require that the QWC be at an angle of about 1.8 to the polarization direction of the illuminating beam, rather than as would be the case if the imager were replaced with a mirror. Figure 5 shows the result for a glass-birefringence orientation of 22.5, the worst-case orientation. t is clear that at some level of retardation, the QWC will not perfectly compensate the birefringence. However, this level is quite high relative to those obtained in a well-constructed PBS, and we have found that it works acceptably in practice. Figure 6 shows the effect of the glass-birefringence angle on the QWC orientation required for best contrast. This chart is generated for a high level of birefringence, 1 nm, and clearly shows that a single orientation of the QWC (as indicated by the horizontal blue line) will provide very high contrast performance at all angles. Figure 7 shows an example of the difference between the uncompensated and compensated cases when the orientation of the glass-retardance optic axis is 25.Withoutcompensation the contrast drops rapidly at small values of retardation. About.2 waves of retardation (~1 nm) reduce FGURE 4 Directions of birefringence causing the most and least effect on the resulting dark state. FGURE 5 Contrast for the same VAN mode imager as for Fig. 3, with glass retardance oriented at This is the worst-case angle for compensation. Journal of the SD 14/3,

4 FGURE 6 Effect of orientation angle of glass retardation on required orientation of QWF compensator for 1-nm glass retardation. t can be seen that excellent contrast is maintained with a single QWF orientation and for any orientation of the glass retardation. contrast to 5:1. However, with the QWC the retardation magnitude must exceed.3 (~15 nm) to cause a similar level of contrast reduction. Useful PBS contrast levels should exceed 25:1 for most RPTV applications. Looking at Fig. 7, we see that without compensation the PBS glass must have no more than.4 waves of retardation at 25 orientation to achieve this level of contrast, while the compensated system can tolerate ten times this level of birefringence. Also shown in Fig. 7 is a comparison of calculated results for obtainable contrast when the imager has a small amount ofretardationinthedarkstate,andwhenitisa perfect mirror. n the calculation, the stray retardance of the imager dark state is assumed to lie along the +45 orientation and has a value of 5.5 nm, or 1/1th of a wave at 55 nm. Such retardance would occur in a VAN imager in which there is a small amount of LC pretilt. By rotating the optic axis of the QWC to 1.8 the QWC compensates both the glass retardance and the retardance of the imager. Very little contrast is lost due to the added retardance from the imager, as long as the QWC is rotated to allow for it. (This effect is independent of the whether a MacNeille or MOF PBS is used.) f the QWC is left at to the polarization direction, we can only achieve a contrast for our model VAN mode imager of 25:1, even when the glass retardation is zero. t is interesting to note that replacing the VAN-mode imager with a ferroelectric LCOS (F-LCOS) device would provide these compensation effects without the need to supply a separate QWC. This is because the F-LCOS itself works by presenting either a quarter wave of birefringence at 45 to the polarization direction of the incoming beam (forabrightstate)orat to that polarization direction (for a dark state). Since no other retardation is present in the imager aside from this quarter-wave layer, the dark state would be naturally compensated for any birefringence effects. FGURE 7 The modeled contrast of the system. The solid red curve represents the contrast of an uncompensated PBS as the glass retardance increases. The dotted blue curve represents the contrast of a compensated PBS as the glass retardance increases when the compensator s optic axis is oriented at. The long dashed green line represents the contrast of a compensated PBS with an as the glass retardance increases and includes an imager with a small amount of stray retardance in the off state. Equivalent performance is obtained by the latter case when the compensator is re-oriented to Measurements Previous attempts to use Pb-free glass in imaging prism PBS applications have not been successful. (Although wire-grid polarizers can be used successfully, the attraction of a glassprism PBS lies in the simpler assembly of the engine and shorter BLF of the projection lens, as well as better transmission of p-polarized light, T p, especially in the blue channel). Cline et al., for example, tested a large number of glasses and showed that most can only work at very low contrast and brightness levels. 1 Even when a stress-free PBS is prepared initially, it is quite difficult to use the PBS with a mirror dark state and maintain good dark-state uniformity as the PBS heats, cools, and is fixtured in place. n order to demonstrate our theoretical results experimentally we used PBSs that meet the high T p requirement discussed above, using 3M MOF PBS technology. 2 n a series of tests in which we intentionally introduced stress-birefringence by swelling the polymer layers in the PBS, we noted that the brightest points in the PBS (which were in the long-path-length corners) became 1/1th to 1/5th as bright with a QWC dark state as for a mirror dark state. (The amount of stress induced in this test is greater than what we expect in normal use environments. The effects produced are similar to what we observe in prototype projectors, but exaggerated.) Figure 8(a) shows the obvious polarization mixing in the dark state of a swelled PBS when a bare mirror is used to reflect the illumination light back to the PBS and then through a projection lens. A schematic of the experimental setup is shown in Fig. 9. The illumination beam is F/ Aastuen and Bruzzone / Lead-free polarizing beam splitters for LCOS

5 FGURE 8 (a) Mirror dark state projected through a Cartesian PBS. (b) QW-compensated dark state projected through a Cartesian PBS. n Fig. 8(a), the left side of the image is brighter than the right side. A ray of light on this side of the screen corresponds to a ray that travels a longer distance through the input half of the PBS, after reflection off of the PBS surface. We will now explain why it is this half of the PBS that generates the most polarization mixing by birefringence in the glass. The multilayer films that form the polarizing beam splitting surface, plane 2 in Fig. 1, are very efficient in transmitting p-polarized light. n fact, the reflectivity of p- polarized light, R p, is typically less than 2% for a 3M MOF PBS. Therefore, if there is birefringence in the glass, any degradation of the linear polarization state in the input beam will be largely cleaned-up upon reflection from plane 2. Once reflected, both rays A and B are highly polarized. However, Ray B has a longer path length in the glass than does Ray A both before and after reflection from the mirror/imager. Therefore, if there is any stress birefringence in the glass, Ray B will acquire a greater amount of FGURE 9 Experimental setup showing the relationship between the PBS, illumination system, mirror, and the projection system. A light ray enters from the bottom, and is reflected by the PBS toward the mirror. The mirror reflects the light ray back toward the PBS where the PBS will allow a dark-state leak to travel to the projection lens if the polarization state has been altered. The details of the illumination system and projection system are not shown. FGURE 1 Depiction of ray paths through PBS. Ray B has its polarization mixed more than Ray A because it travels through more of the PBS after reflecting off the PBS surface. ellipticity (and therefore more p-polarization component) than will Ray A. The greater path-length through the input prism after reflection from the PBS film is why one side is brighter than the other as is shown in Fig. 8(a). The effect of placing a QWC between the PBS and the mirror of Fig. 9 is shown in Fig. 8(b). n this test, the QWC is oriented about the optic axis to provide the darkest state possible. The reader may note that the regions of brightest mirror dark-state measurements vastly improve but that the best mirror dark-state regions actually worsen. At these locations, the birefringence in the glass is in a direction or of a magnitude that does not modify the polarization state of the light. For example, at center left and center right the symmetry of the PBS highly favors the stress lines to be horizontal. As shown in Fig. 4 and discussed above, retardation in this direction does not mix the polarization states. When the QWC is added, small imperfections in the compensator (e.g., variation in the direction of the birefringence) can actually reduce the contrast ratio from its original, very high values. We have found that QWC quality plays a strong role at very high contrast levels. Measurements of the contrast ratio (CR) were made in the four locations shown in the Figs. 8(a) and 8(b). The bright state is taken to be the brightness of the projected image when the QW compensator is rotated 45 from the incoming polarization, and the dark-state data is obtained using the procedure described immediately above. The data is tabulated in Table 1. From the results in Table 1 we can estimate that that the prism has a peak stray retardance of about.85 waves (assuming alignment at 22.5 ), which is more than the uncompensated system can tolerate but well within the levels of retardation our calculations have shown to be acceptable with a compensated system. The data in Table 1 was taken with white light and the QWC used for this test was not spectrally neutral but rather had a birefringence of about 16 nm. t therefore only provided true quarter-wave retardation in the red channel; in Journal of the SD 14/3,

6 TABLE 1 Contrast-ratio measurements of the mirror dark state and a QW dark state as shown in Fig. 1. The four circles in the figure indicate the measurement location. the green channel this compensator is nominally.29 waves, and in the blue channel it is around.35 waves. Even better results could have been obtained by narrowing the bandwidth of the light used or by using a spectrally flat QW such as Nitto Denko NRF that provides close to QW retardation over the entire visible wavelength Conclusion We have shown in theory and in practice that the combination of quarter-wave compensation and very high transmission of p-polarized light allows the use of common, Pb-free glasses in a Cartesian PBS prism, while delivering high, uniform contrast. Many LCOS engines already employ quarterwave films for compensation of skew rays and/or imager dark-state retardance. Thus, the same infrastructure and materials as engine makers currently use to build engines based on Pb-bearing MacNeille PBSs may be used to build projection engines with short back focal length for the projection lens, while avoiding the disadvantages that arise from the use of Pb-bearing glasses. Appendix A: Mueller matrix calculation We are able to demonstrate the effect of compensating the glass with a QWC with a simplified model that ignores offaxis illumination light. We will therefore consider the simplified situation where a beam of light transmits normally through the following list of components: Vertical polarizer Glass with a small amount of birefringence QW compensator VAN imager with a small amount of stray retardance Mirror Same VAN imager with small amount of stray retardance Same QW compensator Same glass with a small amount of birefringence Horizontal polarizer. We will ignore all absorptive, scattering, and fresnel (reflective) losses. Note that the function of the PBS has been assumed to be that of a crossed polarizer and analyzer. We follow the Mueller Matrix method as presented by Azzam and Bashara. 14 We define the polarizers so that they are limited to a contrast ratio of 1,:1, i.e., for the vertical and horizontal polarizers and 1 V = 2 1 H = 2 F HG F HG 1+ L 2 1-L 2 1- L 2 1+ L 2 2 L 2 L where L = 1/1, is the leak term. The contrast ratio, CR, may be calculated by Eq. (2). U is the Stokes vector for unpolarized light of unit intensity. The subscript " in the numerator and denominator indicates the value of the th, or top, component of the resultant Stokes vector. This component represents the intensity of the light after the polarizers have modified it. According to Eqs.(1)and(2)withL as defined above, CR = 1,:1. We now define the Mueller matrix for a generalized retarder of retardance w waves with its optic axis aligned at an angle θ away from vertical. where R(θ) isthemuellerrotationmatrix Cascading Mueller matrix multiplications and taking the th component of the resultant Stokes vector reveals the intensity of light transmitted through the stack of optics referenced above. The retardance for each optic can be written in terms of the generalized retarder matrix M and is given in Table 2. We have simplified the calculation by assuming no reflection losses and that the imager has its optic axis at exactly 45. The mirror is a special case. Because the symmetry that occurs upon reflection, the effect of the mirror on normally incident light can be modeled as a half-wave film with its optic axis aligned vertically. Each of the components that the ray encounters after reflection will have its optic axis direction mirror reflected about the vertical axis. Therefore, KJ KJ 1+ L 2 L 2-1 L L 2, 2 L 2 L [ V V U] CR = [ H V U] F where U = F1 HG KJ 1 1 M( w, q) = R( -q) R(), q cos( 2p w) sin( 2p w) G -sin( p w) cos( p w) J 2 2 H R( q) = F HG 1 cos( 2q) sin( 2q). -sin( 2q) cos( 2q) 1J K. K (1) (2) (3) 29 Aastuen and Bruzzone / Lead-free polarizing beam splitters for LCOS

7 TABLE 2 Mueller matrices of modeled optics. The specific Mueller matrix for each optic modeled in the stack in terms of Eqs. (1) (4) is listed. These definitions are used in Eq. (5) and the following equations. if the beam transmitted through a retarder oriented at an angle β prior to reflection by the mirror, the retarder will look like it is oriented at an angle β after the beam has been reflected. Therefore the contrast ratio given by our model can be written as in Eq. (5), which is similar to Eq. (2). CR( g, q, q, a, s) = F 1 [ ( g, -q) ( q, -a),-, ( q, a) ( g, q) U] H K F 1 H + K r H G C R C G V 4 4 r. [ H G( g, -q) C( q, -a) ( s, -) R ( s, + ) C( q, a) G( g, q) V U] n this case, the bright state is modeled by giving the imager a retardation value of a quarter-wave rather the value s. We have found that as the stray retardance, s, of the off-state imager grows, it can be nearly perfectly compensated by rotating the angle of the QWC to larger negative angles (see Fig. 7, for example). However, the on-state brightness will dim due to incomplete polarization conversion. For example, if s is at a value that the maximum contrast occurs when the compensator angle α = 2.5, then the bright state throughput is reduced 1% when the imager is a perfect QW retarder. However, the retardance of the imager may be increased so that this brightness loss is mostly recovered. For this example, if the on-state retardance of the imager is increased from.25 to.265, then the brightness loss is less than.3%. Because the subject of this paper is dark-state performance, we will ignore the dependence of the contrast ratio on the on-state imager retardance. Therefore, we use Eq. (6) in our analysis and in Figs CR( g, q, q, a, s) = 5.. [ H G(, g -q) C(, q-a) (,) s - R (,) s + C(, q a) G(, g q) V U r ] Equation (6) will overestimate the contrast, but for worst case, for example, when α = 5, the contrast will be overestimatedbynomorethan2%whentheimagerissettoa slightly higher retardance than a quarter-wave. f the imager (5) (6) were to be set to a quarter-wave, the bright state would be depressed 1%. t is not at all unreasonable to expect that driving the imager beyond QW is what happens in today s LCOS engines. Normally, the orientation of the QWC is tuned by careful orientation during manufacture. The example of Fig. 3 showsthisangleisverysensitive to alignment. The contrast falls from over 9:1 to less than 1:1 when the QWC is allowed to rotate by less than 1, the optimum being 1.8. f the QWC is set to, the contrast falls all the way to 25:1. After theqwc is set, theon-statevoltageof theimager will be chosen to deliver the brightest state possible. As long as there is enough voltage and the nd of the liquid crystal is large enough, a voltage will be found that maximizes the brightness beyond which voltage the brightness will begin to decrease. References 1 C L Bruzzone, J J Ma, D J W Aastuen, and S K Eckhardt, High-performance LCOS optical engine using cartesian polarizer technology, SD Symposium Digest Tech Papers 34, (23). 2 S K Eckhardt, C L Bruzzone, D J W Aastuen, and J J MA, 3M PBS for high performance LCOS optical engine, in Projection Displays X, M H Wu, ed., Proc SPE-S&T Electronic maging 52, (23). 3 S M MacNeille, U.S. Patent 2,43,731 (1946). 4 Y Miyatake, U.S. Patent 5,327,27 (1994). 5 A E Rosenbluth, D B Dove, F E Doany, R N Singh, K H Yang, and M Lu, Contrast properties of reflective liquid crystal light valves in projection displays, BM J Research Development 42, (1988). 6 J H Schmidt, N Nestorovic, R D Sterling, J M Haggerty, J A Ruiz, R Edwards, and R Hollister, U.S. Patent 5,576,854 (1996). 7 R Bleha (Personal Communication). 8 R Sterling and W Bleha, LA projectors for advanced applications, Proc DW 97, (1997). 9 A Nakano, A Honma, S Nakagaki, and K Koi, Reflective active matrix LCD: D-LA, SPE Projection Displays V 3296, 1 14 (1998). 1 R Cline, M Duelli, and M Greenberg, Thermal stress birefringence in LCOS projection displays, Display 23, (22). 11 H Simomura, K Numazaki, Y Oikawa, N Simamura, M Ueda, and T Hasegawa, U.S. Patent 5,88,795 (1998). 12 U.S. Patents 6,69,795, 6,486,997, and 6,721, Nitto Denko, Wide-Band Retardation Film, specification sheet. 14 R M A Azzam and N M Bashara, Ellipsometry and Polarized Light, Paperback Edition (North-Holland, Amsterdam, 1997). David J.W. Aastuen received his M.S. degree and, in 1989, his Ph.D. degree in condensed matter physics from the University of Colorado, Boulder, in recognition of his work studying phase transitions of colloidal crystals. Since 1991, he has worked at 3M on product development of display-related components, including polymer LCD substrates, color filter processing and materials, rear-projection screens, and most recently, electronic projection components. Journal of the SD 14/3,

8 Charles L. Bruzzone received his Ph.D. degree for work in high-pressure solid-state physics from the University of Washington, Seattle, W.A., in He has worked in the areas of nuclear physics, solar electrical power, magnetic recording, RF and R control materials, vacuum thin-film processing, polymer LCD substrates, mulitlayer optical films, and electronic projection. His current focus is on components enabling efficient optical systems for LCoS projection systems. 292 Aastuen and Bruzzone / Lead-free polarizing beam splitters for LCOS