Liquid Crystal on Silicon Reflective Microdisplays

Abstract

This section defines the term “LCOS,” then overviews the origins and history of liquid crystal on silicon microdisplay technology. It briefly describes and compares various microdisplay technologies that are, and have been, prevalent. Subsequent sections go into greater detail on selected technologies.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

In the early 1970s, around the time that Alt and Pleshko [1] were quantifying the limitations of passive-matrix or multiplexed addressing and early versions of active matrix addressing were being investigated in order to facilitate the development of TV-quality LCDs, the realization was soon made that crystalline silicon could provide a substrate for reflective AMLCDs and that crystalline silicon–based thin-film device technologies such as metal oxide semiconductor (MOS) were capable of providing active matrix electronics for high pixel count (albeit physically small) AMLCDs built on a crystalline silicon (x-Si) wafer substrate. Thin-film transistor (TFT) technology using amorphous silicon (α-Si) and polycrystalline silicon (p-Si) on a glass substrate was developed separately to allow AMLCDs to be transparent or transmissive and to allow large area. The first report of a “liquid crystal pictorial display” using a silicon active matrix was by Ernstoff et al. in [2] 1973. Other reports followed through the 1970s to the mid-1980s. Armitage et al. [3] offer a list that may not be comprehensive. Many of these efforts looked at wafer-scale displays. Figure  1 shows an example. For wafer-scale displays, the related problems of very low yield and very high cost killed any chance of wide commercialization. In the 1980s, LCOS was developed by several groups primarily as a technology for electronically addressed spatial light modulators (EASLMs) to compliment efforts in optically addressed spatial light modulators (OASLMs). The EASLMs were intended for use in optical systems – including coherent optical systems – for optical computing, optical routing, and optical correlation. The EASLMs tended to be medium to large chip sized or die sized (rather than wafer sized) with many dice per wafer thereby increasing yield and reducing cost to manageable proportions. Figure  2 shows a wafer containing many microdisplay backplanes. When the developed performance of EASLMs was sufficiently high in terms of key parameters such as pixel count and frame rate, attention again turned to their use as ultraminiature displays – initially monochrome and later color. These were no longer wafer-scale direct-view displays but small die-sized displays or microdisplays that would be viewed under optical magnification.

Fig. 1

Wafer-scale LCOS backplane (circa year 1981) (Reproduced with permission from © Bill Crossland 1981)

Fig. 2

Part of multi-chip microdisplay wafer (circa year 2004) (Reproduced with permission from © Peter Tuffy 2006)

The hybrid technology that combines liquid crystal with a CMOS active-matrix substrate has become known as liquid crystal on silicon or LCOS. One of the key factors that has assisted LCOS technology development to progress apace and that has allowed it to become established is that it been able to effectively piggyback progress in its two main component technologies, namely, liquid crystal technology and CMOS silicon technology. In other words, the mainstream development of the backplane technology and, separately, the electro-optical technology are done (and paid for) elsewhere and the LCOS industry has to carry out only the incremental development of bringing the two together and optimizing the combination.

The initial applications recognized for LCOS microdisplays in the early to mid-1990s were military head-mounted displays and electronic projectors. A primary motivation for the development of LCOS microdisplay technology in the late 1990s and early 2000s was the promise of very large, low-cost, rear-projection television screens [4] for domestic use at a time when large low-cost flat panels in direct-view technologies such as LCD and plasma (with hindsight mistakenly) seemed to be very far in the future. The very rapid development, commercialization, and consequent price erosion of PDP and LCD TV technology meant that they came to dominate the domestic large screen TV market before RPTV fulfilled its potential. In the late 2000s, LCOS has established itself in a range of consumer markets from EVF to HMD to pico-projector to HD home-cinema projector, professional and specialist markets from EVF to HMD to data projector and probably the most demanding market in terms of both specification and performance – the digital cinema projector.

A primary enabler for the commercial development of LCOS technology before and after the turn of the century was the ready availability of suitably adapted CMOS wafer substrates. In the case of large corporations such as IBM, Philips, and Intel, CMOS fabrication could be done in-house; for the many start-ups in LCOS such as Microdisplay Corporation, Colorado Microdisplay, Displaytech (now part of Micron), and Micropix Technologies (now part of Kopin), the prevalence of commercial CMOS foundries such as United Microelectronic Corporation (UMC), Taiwan Semiconductor Manufacturing Corporation (TSMC), and others has allowed semi-fabless LCOS microdisplay companies (i.e., those that purchase customized CMOS wafers and add the LC in-house) to compete.

2 LCOS Principle of Operation

The principle of operation of LCOS microdisplays is similar to that of many LCDs as described in Sect. 7, Liquid Crystal Displays of this handbook. The two primary differentiators for LCOS are the reflective nature of the technology as explained below and the very small pixel pitch – of the order of 10 μm.

In a transmissive LCD, the pixel electrode is indium tin oxide (ITO), which is both conductive and transparent. The active matrix circuitry is opaque and the pixel area is shared between opaque circuitry and transparent aperture. This leads to a compromise in fill factor, as shown in Fig. 3 , which can be aided by the use of microlenses aligned to each pixel to concentrate the light through the aperture. In an LCOS device, the pixel electrode is mirror-quality reflective aluminum or similar. This is stacked on top of the active matrix circuit allowing a smaller pixel pitch coupled with a higher pixel fill factor.

Fig. 3

Transmissive (left), conventional reflective (centre) and LCOS reflective (right) LCD pixel configurations

The simplest – in principle – reflective LCD is just a transmissive LCD configuration with a reflector placed behind it as shown in Fig. 3 . In the particular case of LCOS microdisplays, there is no way to position a polarizer under the LC layer (i.e., between the LC layer and the reflective top metal layer of the CMOS backplane), so the configuration shown in Fig. 3 is adopted.

Figure  4 gives a more detailed schematic representation of the cross-sectional structure of an LCOS device.

Fig. 4

Simplified schematic cross section of part of reflective LCOS microdisplay

Specific LC configurations, optimized for reflection with no rear polarizer, must be used. As with transmissive LCDs, there are many options with different strengths and weaknesses. Wu and Yang [5] discuss reflective LCDs in detail while Armitage, Underwood, and Wu [6] analyze LC configurations for LCOS.

Reflective operation offers some advantages. The primary one, noted above, is the increased fill factor. This is emphasized in the plan view of Fig. 5 while the advantage of high pixel fill factor to the appearance of an image is shown indicatively in Fig. 6 . Consider a pixel with optical aperture = $a$ and fixed circuit area = $C$. For a transmissive device, we have the conflicting requirements of minimizing the overall pixel area, A, while wishing to maximize pixel aperture ratio or fill factor, FF,

$$ A = a + C $$

(1)

and

$$ FF = \frac{a}{{a + C}} $$

(2)

Fig. 5

Plan view of pixel showing aperture ratio for transmission and reflection modes

Fig. 6

Illustration of the effect of low and high pixel aperture ratio on appearance of image

For a reflective device with $a$ stacked on top of $C$, we can simultaneously achieve

$$ a < \approx A = C $$

(3)

and

$$ FF < \approx 100\% $$

(4)

In addition, the LC thickness can be halved due to the optical double-pass. This leads to some combination of lower LC drive voltage (i.e., lower power consumption) and/or faster LC switching. The latter opens the opportunity for field sequential color (FSC) operation as described in Chap. 10.3.1.

The primary disadvantage of reflective operation is the relative complication of separating the outgoing modulated light from the incoming illumination in a frontlighting (rather than backlighting) configuration. This is illustrated in Fig. 7 . Solutions to this issue typically add to the depth or profile of the module. In many microdisplay applications, such as EVF and projection, the complication of frontlighting – the additional components and the volume they occupy – can be accommodated. In others, such as head-worn display, the additional depth can be less welcome.

Fig. 7

Conceptual front lighting configuration

The very small pixel pitch of microdisplays has consequences. Maintaining an adequate aperture ratio (or fill factor) becomes more of a challenge at small pixel pitch as the inter-pixel electrode gap typically does not shrink as rapidly as the pixel pitch between product generations. Fringing fields and the associated pixel edge effects reduce the effective fill factor below that of the mirror alone and can be more significant in microdisplays than in larger conventional LCDs. These issues are explored in detail by Armitage et al. [6].

Today there are LCOS displays that use spatial sub-pixelation by RGB color filter [7] and many that operate in FSC mode [8, 9]. In single chip systems for near-to-eye (NTE) applications, RGB LEDs are the illumination source of choice [10]. In single chip projection systems, the historical choice has been high-intensity white light source and rotating color wheel [11]. More recently there has been a move toward laser sources for large area projectors [12] and solid-state laser or high-brightness LED for compact and pico-projectors [13, 14]. The optical and system requirements of NTE systems and those of large-scale projection systems are very different (the issues of pico-projection systems contain elements of both), the requirements of both are complex and the details are dealt with elsewhere in this volume.

3 Ferroelectric Liquid Crystal on Silicon

In contrast to the various forms and configurations of nematic liquid crystal (NLC) that dominate in both conventional liquid crystal displays and liquid crystal microdisplays, surface-stabilized ferroelectric liquid crystal (SSFLC) is used in a small but significant minority of microdisplays. FLC and SSFLC are described in more detail elsewhere in Chapter 7.3.5. SSFLC showed promise in the early 1990s as a potential medium- to large-area e-paper display technology due to its low switching voltage and inherent bistability. But defect-free large area displays proved elusive and poor shock resistance prevented its widespread adoption. It found a niche in digital microdisplays [15, 16] where its use continues to the present day [17, 18].

4 CMOS Technology for LCOS Microdisplays

Silicon microchip technology improves in capability with time according to an empirical observation originally made by Gordon Moore and now known as Moore’s Law [19] which states that the number of transistors that can be placed inexpensively on an integrated circuit doubles approximately every 2 years. This has a beneficial knock-on effect on performance criteria such as processing power, memory capacity, and sensor capability. It has allowed the continued development of higher definition and higher resolution microdisplay backplanes.

Early CMOS technology had a number of shortcomings for use in LCOS. For example, CMOS with only a single metal level meant that the pixel electrode sat alongside bus lines (reducing the optical analysis of the reflective configuration to that of the transmissive as described in 1 and 2), the optical/reflective quality of the metal was very poor, and light was allowed into the substrate where it interfered with circuit operation. Even CMOS with two metal layers suffered from significant surface undulations. Those shortcomings have been overcome with time through mainstream CMOS developments such as multilevel metal (5 or more is quite common in 2010) for high fill factor and chemical mechanical polishing of inter-metal dielectric for flatness. Others have been overcome through custom development of some CMOS process steps purely for LCOS such as optical- or mirror-quality top-metal and reflectivity-enhancing surface layers. Additional enhancements have included microfabricated on-wafer spacer pillars to replace the more conventional spacer-balls or spacer-rods. These can be tailored in size, shape, and position for optimal performance.

The flip side of Moore’s Law for LCOS is that, as CMOS gets smaller in critical dimension, the available voltage scales approximately with critical dimension thus keeping the electric field approximately constant. A 0.18 μm CMOS process, that is, a process with a minimum feature size (usually MOS FET gate width) of 0.18 μm has a standard power supply voltage of 1.8 V. So LCOS microdisplays that use a modern CMOS process must select a process that incorporates the option to include circuit blocks that operate at higher voltage in order to drive the LC effectively.

A mixed blessing of Moore’s law relates to cost. While, on the one hand the manufacture of smaller circuits on larger wafers results in a reduction in unit cost, the one-off cost of design and mask-making in preparation for volume manufacturing rises considerably.

5 Manufacturability of LCOS Microdisplays

There are three main stages of manufacture of LCOS microdisplays as illustrated in Fig. 8 . Wafer fabrication is as per a typical CMOS wafer, as described in [20], with some customization of the last steps that define the surface planarity and reflectivity. LC cell assembly is, in effect, a CMOS wafer-scale (typically 200 mm or 300 mm) version of LCD panel assembly. Finally, dicing and packaging has aspects in common with both LC panel manufacturing and integrated circuit packaging, depending partially upon the packaging requirements for the final application. Figure  9 shows an example of the difference in a microdisplay packaged for NTE and projection. (In this case a DLP^® – rather than LCOS – microdisplay.)

Fig. 8

LCOS top level manufacturing process flow

Fig. 9

Microdisplay package examples for large-scale (left) and pico (right) projection

Large and related historical issues for LCOS have been manufacturability and defectivity. Stated directly – if the microdisplay is magnified by a large factor in use then any visual defects that are created during manufacture will also be magnified and therefore be more visible to the end user. Some years ago, it would have been acceptable to ship a display with a small number of minor visual defects (e.g., stuck-at pixels or pixel-level cosmetic defects). Early TFT LCD TVs, computer monitors, and laptop screens listed the allowed number and distribution of pixel defects in the user manual. This is no longer the case – today the end user demands pixel perfection.

LCOS defects arise from two main sources. Electronic defects produced during manufacture of the CMOS backplane and LC defects that arise from LC cell construction and filling.

In purely electronic CMOS components, an array of techniques including redundancy is used to minimize the effect of CMOS defects. So, for example, a memory chip can be designed to contain some extra memory cells. In the event of a defect, the memory bank can be reconfigured to use the spare cells instead of the defective cells [21]. The physical location of the spare cell is unimportant. Unfortunately in LCOS microdisplays, the vast majority of the area and circuitry is usually in the pixel array where physical location is crucial. A broken pixel or sub-pixel (stuck at black or stuck at white) cannot be replaced by a spare pixel somewhere else. In mitigation, the pixel pitch is often sufficient that the circuit layout can be designed with yield maximization as a high priority.

6 Generic Electronic Architecture

The generic electronic architecture of an LCOS (LCOS implies active matrix; there are no passive matrix LCOS displays) microdisplay is analogous to that of an AMLCD of equivalent definition. However, in the same way that p-Si backplane technology allows greater functional integration than α-Si backplane technology, CMOS provides the ultimate in functional integration as suggested by Fig. 10 . Functional blocks that, in a conventional AMLCD, would be on separate integrated circuits can be integrated on an LCOS backplane thus reducing the chip count. A generic example of LCOS backplane architecture in more detail is offered in Fig. 11 .

Fig. 10

LCD Module integration for a-Si, p-Si, and CMOS

Fig. 11

Block schematic of part of LCOS Backplane

7 Pixel Circuits for LCOS

The small feature size of MOS transistors in a modern CMOS process coupled with the availability of the full pixel area for circuit layout allows the design of pixel circuits with functionality substantially beyond that of the conventional single-switch, single-capacitor pixel shown in the bottom right corner of Fig. 11 . This is of particular benefit to FLCOS devices in which digital pixel functionality can be quite sophisticated. Examples of such circuit functionality include multiple additional storage nodes [22], polarization reversal for ease of DC balancing [23] without the need for full array addressing, triggering for simultaneous pixel switching [24], and analog to PWM conversion [25], as illustrated in Fig. 12 .

Fig. 12

Example of complex pixel circuit used for analog to PWM conversion

8 Application Specific Design of LCOS

LCOS components are used in a wide variety of applications from, at two extremes, electronic viewfinders for digital cameras to projectors for digital cinema. There are many other volume, specialist, and niche applications, each with specific requirements and priorities.

The specification of a LCOS SLM may include the following examples as a subset of all of the parameters to be specified

Definition	Number of lines of pixels
Aspect ratio	4:3, 16:9, …
Method of grayscale	Analog, PWM, …
Method of color	RGB sub-pixels, FSC, 3-chip, …
Color range	CIE color triangle
Color fidelity	Color matching
Maximum luminance	How bright is white
Contrast ratio	How black is black
Optical efficiency	Amount of light transmitted
Maximum frame rate
Switching time
Operating temperature range
Power consumption
Cost

It should be clear that different applications will prioritize different parameters. For example, an LCOS microdisplay for a digital cinema projector may require the highest definition, the highest optical quality (color range, color fidelity, contrast ratio etc.), even if this combination implies high power consumption and cost. An LCOS microdisplay for a battery-powered consumer electronic viewfinder may be able to compromise significantly on all optical parameters in order to achieve low-cost and/or low-power consumption.

9 Non-display Applications of Microdisplays

Microdisplay panels are used in a number of non-display applications. These include structured illumination [26]. LCOS microdisplays in particular can often be adapted to modulate the phase rather than the amplitude of an incoming coherent wavefront thus allowing them to be used as programmable holographic elements. SSFLC LCOS devices offer fast frame rate binary holograms while N-LCOS devices deliver multilevel or analog-phase holograms at more modest frame rates. Note that, in contrast to the situation for amplitude modulation in which fast PWM of a binary source can be used to build up a perceived gray-scale image, when considering phase modulation, PWM cannot be used to create multiple levels from an inherently binary source.

One recent application worthy of particular mention is the use of LCOS devices to display holograms that are projected via a Fourier transforming optical system to recreate the desired real-image sequence [27] resulting in, among other things, a virtual touch screen [28].

10 Forward Look

LCOS technology has survived the early 2000s setback of the RPTV market that disappeared before it had really taken off. It has carved a niche in EVFs from consumer to professional although competition has begun and is increasing from OLED microdisplays. But OLED is not competitive in any form of projection. Several proprietary varieties of LCOS have become established in digital cinema and are capable of 3D digital cinema and 4k definition. The rise of linearly polarized laser illumination sources helps level the playing field against DLP. LCOS has some market share in data projectors and looks promising for the nascent pico-projector market. Furthermore, the ability of LCOS variants to modulate the phase (rather than amplitude) of coherent illumination offers potential in a growing range of niche applications.