All of the special properties described above, especially the property of circular birefringence, can be exploited for the purposes of creating flat-panel displays. Usually what is done is a thin film of liquid crystal is placed between two pieces of glass or transparent plastic. These plates are usually manufactured with transparent electrodes, typically made of indium tin oxide, that make it possible to apply an electric field across small areas of the film of liquid crystal. Polarizing filters are usually placed on one or both sides of the glass to polarize the light entering and leaving the crystal. Usually these polarizers are crossed, which means that, normally, no light would be able to pass through the display. The liquid crystal, however, will modify the polarization of the light in some way that is dependent on the electric field being applied to it. Therefore, it is possible to dynamically create spots where light will get through and spots where it will not.
There are many issues that must be addressed when designing and analyzing the performance of a flat-panel display. The first is the amount of voltage necessary to turn on a pixel in the display. There are two common measures: First is simply a threshold voltage, , which is the amount of voltage across the pixel that is necessary to produce any response whatsoever. The other is a measure of the ``sharpness'' of the response and is calculated by finding the difference in voltage necessary to go from a 10% to a 90% brightness (usually written as .
Another characteristic of displays that must be dealt with is the switching times of the pixels. These are commonly written as and , and they correspond to the amount of time between application/removal of the voltage and a 90% brightness/darkness response. Usually is slightly larger because after voltage is removed, the liquid crystal is merely relaxing back into its off state---no force is being applied, unlike when it is being turned on. Switching times can be changed by controlling the amount of orientational viscosity in the crystal, which is the amount of resistance when being forced to change direction.
The contrast of a liquid crystal display is an important issue as well. One way to measure it is to find the difference in brightness between an on and off pixel, divided by the larger of the two values. A more useful value is the contrast ratio which is simply the larger brightness divided by the smaller brightness. LCD designers want this ratio to be as large as possible in order to obtain ``blacker blacks'' and ``whiter whites.'' Typical LCDs have contrast ratios between 10 and 40. Unfortunately, the contrast will depend highly on the angle the display is viewed from since the effects of the liquid crystal are calibrated to work best on light passing through the display perpendicularly. When viewed from an angle, we are not seeing the light coming out perpendicular from the liquid crystal, so it is not uncommon to see a break down in the contrast. In some cases it is even possible to see a negative image of the display.
In addition to these issues of display design, we must also pay attention to the method we address the display pixels. In direct addressing, there is a direct connection to every element in the display, which is good since we have direct control over the pixels, but it is also bad because in large displays there could be thousands or even millions of pixels that would require separate connections. The method used in the vast majority of large modern displays is multiplexing. In this method, all the pixels across each row are connected together on the plate on one side of the liquid crystal film, and all the pixels in each column are connected on the opposite side. The rows are then ``addressed'' serially by setting all of the column voltages separately for each row and then turning on the row voltages in sequence. There are significantly less physical connections that must be made when multiplexing is used, but there are also several new problems to address. If there are N rows, as we cycle through them, the pixels in one row will only be receiving the necessary voltage 1/N of the time. When other rows are being addressed, these pixels will be receiving smaller voltages originating only from their column electrodes. Therefore, the pixels never really receive full on or off voltages---they are always somewhere in between, and depending on how close together they are, the contrast of the display could be very low. A specific example of this effect is presented later.
With these issues in mind, let's examine the state-of-the-art of LCD technology. Perhaps the most common kind on the market today is the twisted nematic display. In this display, the liquid crystal molecules lie parallel to the glass plates, and the glass is specially treated so that the director of the crystal is forced to point a particular direction near one of the plates and perpendicular to that direction near the other plate. This forces the director to twist by from the back to the front of the display, forming a helical structure similar to chiral nematic liquid crystals. In fact, some chiral nematic crystal is added to make sure all of the twists go the same direction.
Now, this thin film of twisted nematic liquid crystal is circularly birefringent. When linearly polarized light passes through, the optical activity of the material causes the polarization of the light to rotate by a certain angle. The thickness of the film, typically around 6 or 8 micrometers, can be controlled to produce a rotation of the polarization of exactly for visible light. Therefore, when the film is placed between crossed polarizers, this arrangement allows light to pass through. However, when an electric field is applied across the film, the director will want to align with the field. The crystal will lose its twisted structure and, consequently, its circular birefringence. Therefore, linearly-polarized light entering the crystal will not have its polarization rotated (in fact, it is only rotated very slightly), so light will not be able to penetrate through the other polarizer. When the field is turned off, the crystal will relax back into its twisted structure and light will again be able to pass through. In some displays, the polarizers are parallel to each other, thus reversing the on and off states. If red, green, and blue colored filters are used on groups of 3 pixels, color displays can be created.
The twisted nematic system coupled with multiplexed addressing is used in many of today's so-called passive matrix LCDs. Even though it is the most popular kind, it does have a number of disadvantages as well. First of all, the use of polarizers reduces the potential brightness since they allow less than half of the light incident on the display to pass through. The effective viewing angle of the display can be very small because the optical activity and the polarizers are tuned to work best only on light that is propagating perpendicular to the display. The voltage-brightness response curve is often not very sharp, leading to reduced contrast. The display is also affected by crosstalk where voltage meant for a certain pixel can leak through ``sneak paths'' to nearby pixels, causing a ghosting effect. And finally, switching speed of the liquid crystal is often not as high as might be desired---typically around 150 milliseconds. Lower switching speeds are necessary when doing multiplexing since we want the crystal to respond to voltages over the whole scanning cycle to reduce flicker, but such low speeds make passive matrix displays unusable for many applications (such as full motion video).
The degree of multiplexing in twisted nematic displays has a huge influence on the contrast of the display. To understand how this is so, let's examine a typical situation. The liquid crystal will respond to the average voltage applied to it over a certain period of time, depending on its viscosity. Assuming the liquid crystal responds to voltages over one frame period, we can calculate the average voltage felt by a pixel that is on and a pixel that is off. Suppose that there are N rows that are addressed in sequence. A row gets a voltage of 1 if it is being addressed and 0 if it is not. During each row pulse, the column voltages are set according to which pixels in that row are on or off. If a certain pixel is on, it receives a column voltage of -1, otherwise it is 0.
Now, let's calculate the average voltage felt by a specific pixel assuming that other pixels in the same column have equal probability of being on or off. When the row containing our pixel is addressed, it will feel either 2 volts if it is on (row voltage = 1, column voltage = -1), or just 1 volt if it is off (row voltage = 1, column voltage = 0). When the other (N - 1) rows are being addressed, our row voltage is 0 and our column voltage changes between 0 or 1 as the other pixels in the column are addressed. This means we feel an average voltage of 1/2 during that time. Therefore, the total weighted average is
if the pixel is on, and
if the pixel is off. Since the voltage-brightness response curve of the crystal is typically not very steep (see figure), these voltages will not lead to as high of a difference in brightness as would be possible if each pixel received either a full 2 volts or 0 volts continuously over the entire frame period.
Figure 3: Response curve of a twisted nematic LCD. The range of voltage (and therefore brightness) felt by on and off pixels is much less than the full range possible.
Ideally, we would want our response curve to have near infinite slope between the average on and off voltages, but this is very difficult to achieve. The selection ratio of the display is the average on-voltage divided by the average off-voltage---for our example, the ratio is
Typically, a certain minimum ratio is needed to get good contrast. Since the ratio is a function of the number of rows, the more rows we add, the less contrast our display will have. For example, if we knew from our brightness response curve that we needed a selection ratio of at least 1.01 to have a good contrast ratio, then our display cannot have more than about 200 multiplexed rows. For a selection ratio of 1.05, we could only have about 60 rows.
There is another common kind of LCD called an active matrix display that solves this problem by essentially putting a full switch at every pixel in the entire display. This special element ``actively'' drives the voltage to the pixel continuously over the frame period. Essentially, a thin-film transistor at each pixel allows the effects of the column voltages to be felt only by the row that is being addressed. Therefore, outstanding contrast is possible, and a faster liquid crystal substance can be used since it is not necessary for the pixel to respond to the average voltage over a whole frame period. Crosstalk is also cut to a minimum. With increased contrast and switching speed, active matrix displays are more than able to handle many kinds of information, including full-motion video which requires a minimum switching speed of 50 milliseconds.