1. Introduction

A wide viewing range is usually desired for a liquid crystal display (LCD). In case of a cellular phone and a notebook PC, however, we sometimes wish to limit its viewing range for privacy protection. Some authors have studied electronic switching of the viewing range of a LCD [1-3]. First, a polymer-dispersed liquid crystal (PDLC) layer is inserted between a liquid crystal (LC) panel and a backlight unit illuminating in a narrow angular range [1]. The PDLC layer either transmits or scatters the backlight output and the viewing range is altered accordingly. Second, two backlight units are stacked and are placed below an LC panel [2]. One unit illuminates in a wide angular range and the other unit emits in a narrow range. Turning on the first unit enables the narrow viewing mode and operation of the second unit facilitates the wide viewing mode. Third, a special LC panel with two LC mode switching function is proposed [3]. Three-electrode structure in a pixel allows one to select one of the two LC modes, i.e., the in-plane switching (IPS) mode for the wide viewing range and the vertical alignment (VA) mode for the narrow viewing range. Obviously, the first two approaches increase the thickness of an LCD and this is not desired for mobiles applications. Although the third approach does not increase the LCD thickness, its switching contrast may not be adequate. Namely, images may be displayed with inverted gray levels at large emission angles even in the narrow viewing range setting. In addition, a careful design of various compensation films may be required for a good color performance.

We have proposed a single backlight idea [4]. Here, a switchable light-deflecting device is placed between a light source and a light-guide of an edge-lit backlight. A device based on the liquid crystal technology is suited for this. A bias applied to such an LC device controls the angular distribution of the light propagating inside the light-guide. Output couplers built in the light-guide translate the propagation angles into the output angles. Hence, the viewing range can be switched. This approach neither increases the LCD thickness nor affects the LC panel design. We have also reported some component-level experimental results [5, 6].

In the next section, we consider the configuration of the whole backlight unit and each component. In the following section, we discuss how to attain uniform illumination and reveal a new output coupler design.

2. Backlight with emission angle control

2-1. Configuration

In our original configuration (Fig. 2 in Ref. 4), an LC device is attached to the edge of a light-guide. However, integration of optical functions previously performed by independent films has significantly decreased the thickness of the edge-lit backlight units [7-12]. Now, the light-guide can be thinner than 0.3mm. Due to an extra space required for containing the LC material inside the LC device, it is not practical to couple it to the edge of such a thin light-guide.

Here, we illustrate a modified configuration in Fig. 1. A plain view of the whole unit is shown in Fig. 1(a). A cross-section along the dotted line A-A’ in Fig. 1(a) is depicted in Fig. 1(b) and a part of it is magnified in Fig. 1(c). A linear light source and an LC device are stacked together and are placed on top of the light-guide. The mirror formed at the edge of the light-guide directs the light emerging from the LC device into the light-guide. A careful optical design and accurate assembly will minimize the insertion loss for the LC device. This level of alignment is routinely carried out over a large area when an LCD panel is assembled. As an output coupler, a transparent film with curved microstructures is in contact with the light-guide [7]. As shown schematically in Fig. 1(c), rays hitting the area called “optical window” enter the curved section. They go through total internal reflection (TIR) at the curved boundary of the structure and are directed toward an observer.

 

Fig. 1. Configuration of our proposed backlight with emission angle control.

Download Full Size | PDF

2-2. Light source

The configuration in Fig. 1 relieves the size requirement for the light source. We can use either a linear source such as a cold cathode fluorescent lamp (CCFL) or a combination of a point source and some optical components that convert the point source into a linear source. A rod lens array coupled to an array of light emitting diodes (LEDs) is routinely used as such a linear light source in a facsimile machine and a printer.

A collimated source is advantageous for compact implementation because it can couple the light into a thin light-guide. For a CCFL, a reflector such as one in Fig. 1(c) can redirect the light toward the mirror. Alternatively, we can use a laser diode (LD) coupled to an optical fiber with some built-in light-extraction structure [13]. Use of three LDs emitting at different wavelengths expands the color gamut considerably [14]. Coherent light has an advantage that will become evident below. But it also has a drawback: a spatially-stable granular pattern known as speckle has to be properly dealt with [13, 14]. Design of such a light source is a subject of our future studies.

Emission properties of a light source must be carefully considered together with the characteristics of an LC device. For example, a polymer-dispersed liquid crystal (PDLC) cell can turn a collimated source into a Lambertian source but not the other way around. In our experiment, we needed to make the angular distribution from a chip-type LED narrower by inserting an optical film [6].

2-3. Liquid crystal device

The LC device alters the emission characteristics of the light source. Either a PDLC cell, an LC lens or an LC phase grating can be employed [4]. An LC phase grating [15] is particularly promising for this application because it does not scatter or absorb the light significantly. Its structure and characteristics are described in detail in [15]. When used as the LC device in Fig. 1(c), it can change the angular distribution only in the y - z plane for example. We have shown that light emitted from an LED is diffracted by an LC phase grating to some degree [4]. Coherent light from an LD is diffracted much more efficiently because it is in phase. We can make the transmitted light disappear almost completely if required [15].

An LC phase grating usually creates a periodic distribution of refractive indices in one dimension so that the diffraction switching takes place in one plane. It would be interesting to see if this switching can be extended to two planes by modifying its device structure. Such an LC phase grating would be able to control the angular profile in the x – z plane in Fig. 1 as well. A PDLC cell and a polymer-network liquid crystal (PNLC) cell can do this but they also scatter light significantly in their off-state [6]. In the simulation in Section 3, we will set up a wide and a narrow profile in every azimuth direction so that the viewing range is switched in both directions.

2-4. Output coupler

Once we choose a light source and an LC device, the angular distributions for the light propagating inside a light-guide are known. Next, we consider the output coupler design. An output coupler translates the change in the propagation angle θ_p to a change in the output angle θ_o. These angles are defined in Fig. 1(c). When a ray hits the curved boundary in Fig. 1(c), it goes through multiple TIRs and is extracted toward an observer. For small propagation angles, the range of the output angles is small. As the propagation angle becomes larger, the number of TIRs decreases and the output angle range becomes larger. Therefore, widening the range for the propagation angle leads to a wider output angle range.

Note that the optical window in Fig. 1(c) is along the z axis. The light propagating inside the light-guide with a large angle has a higher probability for hitting these optical windows and is quickly depleted near the light source. A common practice is to adjust the density of output couplers to compensate this depletion. The positions of the output couplers on the light-guide are carefully tailored for specific emission characteristics of a light source to ensure uniform light extraction. As described in [9], ray tracing simulation and prototyping are often repeated to reach an optimum design.

3. Illumination uniformity

In our configuration for emission angle control, a single output coupler design must achieve uniform light extraction for different angular distributions for the light inside the light-guide. We address this problem in this section.

3-1. Vertical optical window

In case of the conventional optical window in Fig. 2(a), the probability of reaching this window is higher for a ray with a larger propagation angle. This is caused by the fact that the optical window is along the z axis. Instead, we can make the optical window vertical as shown in Fig. 2(b). The top plate has quarter cylinders and the bottom plate has quadratic prisms. The surface binding the quarter cylinder and the quadratic prism constitutes the optical window. For a quantitative discussion, we have carried out ray tracing simulations. As described in [16], the shape of the angular distribution for the light inside the light-guide is more or less the same along the z axis. Therefore, we expect that a single array of these output couplers can provide uniform light extraction for both settings for the viewing range.

 

Fig. 2. A vertical optical window extracts the propagating light with different angles equally.

Download Full Size | PDF

3-2. Uniformity and switching characteristics

We now adjust the intervals between the output couplers to compensate the effect of ray depletion near the light source. The model for this analysis is shown in Fig. 3(a). The cross section of the structure is shown in Fig. 3(b). The material for the light-guide and the top film is Poly(methyl methacrylate) (refractive index 1.49) and its absorption property is taken into account. The length of the quadratic prism (denoted as L) affects the angular profile of the propagating light. We have found that setting L to 190μm gives satisfactory profiles [16] and adopt this value here. A Lambertian light source with a rectangle emission area of 37mm × 0.05mm is placed 10μm away from the edge of the light-guide. A detector array is placed 50μm above the top plate and it monitors all the rays extracted from the top 60mm × 37mm surface. The area covered by each detector is 3mm × 37mm. Later in the analysis, these detectors are grouped into three regions. The incident directions as well as the total number of the rays are calculated for each region. Other geometrical dimensions are indicated in Fig. 3. Tracing is terminated when one of the following events occurs: (1) a ray hits a detector, (2) it is absorbed and (3) the number of reflections exceeds a certain large number (1,090). A commercial software package LightTools (Optical Research Associates) is used.

First, we set all the intervals between the output couplers according to a certain formula. Light intensity recorded by the detector array was plotted as a function of the distance z. Then we modified the formula and repeated the analysis until a more or less uniform distribution was obtained. The result is shown by the solid triangles in Fig. 4(b). The open circles represent the intervals between adjacent output couplers. Then we limited the range of the Lambertian source somewhat arbitrary to within ±15 degrees in an axially symmetric manner. The two distributions are compared in Fig. 4(a). The open triangles in Fig. 4(b) show the result for this second ray tracing. In both cases, the total number of rays generated was 1×10⁷ and the intervals of the output couplers were the same. Thus, illumination uniformity was maintained for both settings of the emission angle range. For further improvement of the uniformity, the oscillation behavior in Fig. 4(b) must be taken out. Further studies are needed to understand its origin.

 

Fig. 3. Model for analyzing illumination uniformity and angular range switching characteristics

Download Full Size | PDF

 

Fig. 4. To obtain uniform illumination, the intervals between the output couplers are adjusted.

Download Full Size | PDF

We are interested in how the angular distribution is switched when the emission characteristics of the light source is switched. We divide the ray data into three groups based on the z coordinates of their detector positions as shown in Fig. 3(a). Angular distributions are calculated for the rays detected in each of these regions. The distributions in the x - y plane are compared in Fig. 5(a)-(c) and those in the y - z plane are shown in Fig. 5(d)-(f). These are normalized at the peak intensity. The open and solid circles correspond to the two settings in Fig. 4(a). As shown in Fig. 5, good switching contrast is obtained in the x - y plane. A similar emission characteristic is obtained in the region near the light source and the region near the other end of the light-guide. Unlike the technique dealing with the LC modes [3], this single-backlight approach gives no leakage of light at a large viewing angle for the narrow mode setting. However, the distributions in the y - z plane are narrow even when the emission angle of the Lambertian source is not limited. Ideally, we may wish to limit the viewing range in both planes but this is not possible with the current design. In practice, however, switching in only one plane may still be useful.

 

Fig. 5. Angular distributions of the light emitted at the three locations along the z axis.

Download Full Size | PDF

Finally, light utilization efficiency is an important characteristic. In the simulation model in Fig. 3, we define this efficiency as the ratio of the number of rays detected by the top detector array to the number of rays generated. This is justified because each ray carries equal weight in the simulation. We placed additional detector arrays close to other surfaces of the light-guide and repeated the two simulations. The number of rays generated was 1×10⁶ for each setting of the light source. The efficiency was 0.72 for the first simulation with the Lambertian source and 0.76 for the second simulation with the limited Lambertian source emitting within ±15 degrees. The ratio of the rays escaping from the edge of the light-guide was 0.13 and 0.16 for the first and the second simulation, respectively. Some rays were also emitted toward downward in Fig. 3 after going through refraction and/or Fresnel reflection at the output coupler structure. However, this contribution was only 0.00039 and 0.00018 for the first and the second simulation, respectively. About 15% and 8% of the rays in the first and the second simulation were not accounted for, i.e., no detector counted these rays. We believe that these are either absorbed or the number of reflections exceeded the maximum number. In future, we will consider placing reflectors at the edges to improve this light utilization efficiency.

4. Conclusions

In order to control viewing range of an LCD illuminated by an edge-lit backlight, one can insert an LC device between the light source and the edge of the light-guide. This approach does not affect the LCD thickness much and the weight increase can be negligible. The LC panel design remains intact. However, its output coupler design needs attentions. We have proposed an output coupler with a vertical optical window through which light is extracted from the light-guide. Rays with large propagation angles and those with small angles have a comparable probability of reaching this window. Hence, a single array of these output couplers can provide uniform illumination for both settings of wide and narrow viewing range. Ray tracing simulations confirm that this is really the case and that good switching contrast is obtained in the plane perpendicular to the direction of light propagation. In future, we will consider the output coupler design further to improve the switching contrast in the other plane as well as the total system design to improve the light utilization efficiency.