**Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays**

Jian-Chiun Liou

*Industrial Technology Research Institute Taiwan* 

## **1. Introduction**

130 Features of Liquid Crystal Display Materials and Processes

Doom, Z., Dynamic behavior of twisted nematic liquid-crystal layers in switched field, Journal of Applied Physics, Vol.46, No.9, September 1975, pp.3783-3745 Ishihara, K.(2008), Implementation of optical response of thin film transistor, *Proceedings of* 

Jacunski, M. D. (1999). A Short-Channel DC SPICE Model for Polysilicon Thin-Film

Kimura, M. (1994). Simulation Techniques for Horizontal Crosstalk in TFT-LCDs,

Pochi, Y. (1999). *Optics of Liquid Crystal Displays*, pp.248-252, John Wiley & Sons, ISBN 0-471-

Smet, H.D., Electrical model of a liquid crystal pixel with dynamic, voltage historydependent capacitance value, *Liquid Crystal*, Vol.31, No.5, May 2004, pp.705-711 Sze, S. M. (1981). *Physics of Semiconductor Devices Second Edition*, pp.440, John Wiley & Sons,

Watanabe, M.(1996), Development of a 20.1 Diagonal Super Fine TFT LCD, *Proceedings of* 

Watanabe, M.(1997), Novel pixel structre for IPS TFT-LCD with color-shift free,*Proceedings of* 

Watanabe, M.(2007), Macro modeling of liquid crystal cell with VerilogA, *Proceedings of IEEE* 

Wittek,M. (2008), Advanced LC Materials for Ultra-Fast Switching for Active-Matrix-Device (AMD) Applications, Proceedings of IDRC '08 Conference, pp.253-255, 2008

*DEVICES*, Vol.46, No.6, June 1999, pp.1146-1158, ISSN 0018-9383

Transistors Including Temperature Effects, *IEEE TRANSACTIONS ON ELECTRON* 

*Proceedings of AM-LCD'94*, pp.228-231, ISBN 4-930813-62-10, Shinjuku, Tokyo,

*IEEE BMAS 2008 Conference*, pp.39-44, 2008.

Japan, November 30 – December 1, 1994

ISBN 0-471-05661-8, the United States of America

*International Display Research Conference'97*, pp.L-9, 1997.

18201-X, the United States of America

*Euro Display'96*, pp.587-590, 1996.

*BMAS 2007 Conference*, pp.132-137, 2007.

**6. References** 

As modem society has been reshaped rapidly into the multimedia information society, large size of display devices is required. Thin-film-transistor liquid- crystal-displays (TFT-LCDs) are one of the most popular display devices from small to large size. TFT-LCDs need backlight source, because they are not a self luminance device. Backlight for LCDs is becoming more important with growth of the LCD size. Up to now, multiple fluorescent lamps such as cold cathode fluorescent lamp (CCFL), external electrode fluorescent lamp (EEFL) and flat fluorescent lamp (FFL) are generally used as LCD backlight source. Since, the conventional backlight of LCDs illuminates at the full luminance regardless of the images to be displayed, it wasted power and contrast ratio is row due to the light leakage in dark state.

Recently, as the luminance efficiency of light emitting diode (LED) has been improved and the cost of LED is going down, the LED is the substitutive solution for the backlight source. Moreover, since LED has many advantages such as long lifetime, wide color gamut, fast response, and so on, LEDs are expected to replace the conventional fluorescent lamps for backlight source of LCD in near future. Although, LED backlight driving systems have been developed and introduced to the market, further reduction of power consumption and cost reduction are still demanded to be widely used as backlight source. In segmented dimming and local dimming methods which have usually adapted to CCFL and LED backlight source, the whole backlight is divided into several segmented and block, backlight scaling is adapted to each segmented and block, respectively. The more division of backlight, the more power saving can be achieved. Therefore, local dimming method can save power consumption more effectively than segmented dimming one. Especially, segmented dimming method has limitation of power saving when the image is bright vertically. Therefore, local dimming is the best way to have local dimming effects. However, local dimming method results in huge increase in the number of drivers needed for large number of division. To compromise between segmented dimming and local dimming methods, a new LED backlight system for LCD TVs which involves the time division X-Y segmented driving method that utilizes row and column switches to control the individual division screen is proposed.

Even though the refresh time of an LCD panel is as fast as a CRT display, there are still some problems to be solved before this LCD panel can be implemented to be a time-multiplexed

Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays 133

Fig. 1. Schematic diagram of scanning backlight method

Fig. 2. Synchronization Signal LED backlight architecture

stereo display, vision and panel seem to be redundant. First of all, LCD is a hold-type display, which is different from the impulse operation type of CRT. This has been a very good property of an LCD while it works at 2D display mode because it can avoid blinking or flickering at the refresh rate of 50 or 60 Hz. For a CRT display, viewers will still experience slight image flickering at the refresh rate of about 60Hz. Nevertheless, while working in 3D page-flipping mode and watched through shutter glasses, the hold-type operation causes a problem.

When the left-image begins to fill out the pixels, the left-eye shutter opens immediately. However, at this moment and the following 1/120 sec., the left-image at the upper part of the screen becomes more and more image area of screen, the right-image at the lower part of the screen becomes less and less image area of screen. That means, the viewer will see both images at the same time through his right eye. This situation will induce strong double image (or crosstalk) and destroy the 3D perception seriously.

#### **1.1 Intelligent and green energy LED backlighting techniques**

1. Scanning backlight method. The setup is as following:

The backlight unit is separated into several regions. Let's take 4 segmented regions as the example as in Fig.1. the pixel response time is less than three fourths of the frame time when the illumination period is one quarter of the frame time. Arranging for the required illumination period to end just before the new image is written into the panel provides the most relaxed requirement for panel response time; i.e., the response time can be longer (Fig. 1). A novel controlled circuit architecture of scanning regions for 120Hz high frequency and high resolution stereoscopic display is shown in Fig.2. Setup all the parameters of scanning backlight method by counting the amount to decide turning time between 4-regions、and 2-regions LED backlight type. If counted times equal to 100 then jump to next backlight segmented region.

For 4-region scanning backlight method, when the panel is filled in segmented regions 1, 2, 3 and 4 by the new image, the backlight lights up are in the corresponding regions 4, 1, 2 and 3. In anticipation of an image for a left eye and right eye is shown in the region 1 of the panel, we turned on region 3 of the backlight unit. Analogize the image shown in region 2 and turned on region 4 of the backlight unit. For 2-region scanning backlight method, when the panel is filled in regions 1 and 2 by the new image, the backlight lights up in the corresponding regions 2 and 1. For avoiding seeing both Limage and R-image at the same time, the backlight regions R1 have to be off until R1 filled up the image. Analogize the backlight regions R2 have to be off until R2 filled up the image.

2. In backlight strobe method as shown in Fig.3, Setup the parameters of backlight strobe method by counting the amount to decide turning on time of full screen (full screen of one frame 1/120sec counted amount equal to 400). Setup the parameters of backlight strobe duty time by counting the amount to decide turning time on full screen backlight regions. If counted times equal to (400 - 400\*9/10) then jump to next full screen backlight region. The backlight is turned off when the image data refreshes. The backlight only turns on at the system time, or at most a little bit longer than the system time. But the system time is short compared with the time between two adjacent vertical synchronization signals (less than 10%), the display brightness operated under this method is probably quite small.

stereo display, vision and panel seem to be redundant. First of all, LCD is a hold-type display, which is different from the impulse operation type of CRT. This has been a very good property of an LCD while it works at 2D display mode because it can avoid blinking or flickering at the refresh rate of 50 or 60 Hz. For a CRT display, viewers will still experience slight image flickering at the refresh rate of about 60Hz. Nevertheless, while working in 3D page-flipping mode and watched through shutter glasses, the hold-type

When the left-image begins to fill out the pixels, the left-eye shutter opens immediately. However, at this moment and the following 1/120 sec., the left-image at the upper part of the screen becomes more and more image area of screen, the right-image at the lower part of the screen becomes less and less image area of screen. That means, the viewer will see both images at the same time through his right eye. This situation will induce strong double

The backlight unit is separated into several regions. Let's take 4 segmented regions as the example as in Fig.1. the pixel response time is less than three fourths of the frame time when the illumination period is one quarter of the frame time. Arranging for the required illumination period to end just before the new image is written into the panel provides the most relaxed requirement for panel response time; i.e., the response time can be longer (Fig. 1). A novel controlled circuit architecture of scanning regions for 120Hz high frequency and high resolution stereoscopic display is shown in Fig.2. Setup all the parameters of scanning backlight method by counting the amount to decide turning time between 4-regions、and 2-regions LED backlight type. If counted times

For 4-region scanning backlight method, when the panel is filled in segmented regions 1, 2, 3 and 4 by the new image, the backlight lights up are in the corresponding regions 4, 1, 2 and 3. In anticipation of an image for a left eye and right eye is shown in the region 1 of the panel, we turned on region 3 of the backlight unit. Analogize the image shown in region 2 and turned on region 4 of the backlight unit. For 2-region scanning backlight method, when the panel is filled in regions 1 and 2 by the new image, the backlight lights up in the corresponding regions 2 and 1. For avoiding seeing both Limage and R-image at the same time, the backlight regions R1 have to be off until R1 filled up the image. Analogize the backlight regions R2 have to be off until R2 filled up

2. In backlight strobe method as shown in Fig.3, Setup the parameters of backlight strobe method by counting the amount to decide turning on time of full screen (full screen of one frame 1/120sec counted amount equal to 400). Setup the parameters of backlight strobe duty time by counting the amount to decide turning time on full screen backlight regions. If counted times equal to (400 - 400\*9/10) then jump to next full screen backlight region. The backlight is turned off when the image data refreshes. The backlight only turns on at the system time, or at most a little bit longer than the system time. But the system time is short compared with the time between two adjacent vertical synchronization signals (less than 10%), the display brightness operated under

image (or crosstalk) and destroy the 3D perception seriously.

**1.1 Intelligent and green energy LED backlighting techniques** 1. Scanning backlight method. The setup is as following:

equal to 100 then jump to next backlight segmented region.

operation causes a problem.

the image.

this method is probably quite small.

Fig. 1. Schematic diagram of scanning backlight method

Fig. 2. Synchronization Signal LED backlight architecture

Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays 135

Fig. 4. Block diagram of control algorithm for LED backlight array module

fabrication cost from the greater reduction of circuit component numbers.

processing technique for the proposed stereo LCD backlight system.

browsing for remote 3D models; scientific visualisation and medical imaging.

**2. 3D display backlight and application** 

follows.

driving current and shares the same ground with the other resistors. The resistors are individually addressable to provide unconstraint signal permutations by a serial data stream fed from the controller. The shift register is employed to shift a token bit from one group to another through AND gates to power the switch of a LED group. The selection of a LED set is thus a combined selection of the shift register for the group and the data for the specific LED. Such an arrangement allows encoding one data line from the controller to provide data to all of the LEDs, permitting high-speed scanning by shortening the LED selection path and low IC

Throughout the proposed system, lower power consumption are successfully obtained as well as high contrast ratio even with less number of drivers than that of conventional local dimming method. This chapter also contains a new adaptive dimming algorithm and image

Recent progress in stereo display research has led to an increasing awareness of market requirements for commercial systems. In particular areas of display cost and software input to the displays are now of great importance to the programme. Possible areas of application include games displays for PC and arcade units; education and edutainment; Internet

Intelligent and green power LED backlighting techniques of two-dimensional (2D) to threedimensional (3D) convertible type, shutter glasses type, mult-view time multiplexed naked eye type, and multi-viewer tracking type for stereo liquid crystal displays are shown as

Fig. 3. Time scheme of backlight strobe method

In this research, we have successfully designed and demonstrated a decent performance with 120Hz optimized synchronization signal between LED brightness/darkness flash and adjusted shutter glasses signal. It has been demonstrated that the 120Hz scanning characteristic from upper row to lower row of the horizontally arranged of stereoscopic image. A quadrate image for a left eye is projected by the light from the left eye image file and a circle image for a right eye is projected by the light from the right eye file through a liquid crystal panel.

LED scanning backlight stereoscopic display with shutter glasses is provided to realize stereoscopic image viewing even in a liquid crystal display. In a frame time, some kinds of brightness/darkness characteristic from upper row to lower row of the horizontally arranged rows of LED in the backlight module, cooperating with the scanning of the LCD, to thereby realize an effect similar to scanning. The general strategy that we employ is to integrate all relatively small-signal electronic functions into one ASIC to minimize the total number of the components. This strategy demonstrates that both the cost is lowered and the amount of the printed circuit board area is reduced. Based on this concept, a smart three dimensional multiplexed driver for LED switching chip with more than 640 LEDs are proposed and the circuit architecture is shown in Fig. 4. It is difference from the traditional two dimensional arrays driven by scanning scheme. Three lines are employed to control one LED, including voltage, shift register, and data line. Each LED requires a voltage line for the

In this research, we have successfully designed and demonstrated a decent performance with 120Hz optimized synchronization signal between LED brightness/darkness flash and adjusted shutter glasses signal. It has been demonstrated that the 120Hz scanning characteristic from upper row to lower row of the horizontally arranged of stereoscopic image. A quadrate image for a left eye is projected by the light from the left eye image file and a circle image for a right

LED scanning backlight stereoscopic display with shutter glasses is provided to realize stereoscopic image viewing even in a liquid crystal display. In a frame time, some kinds of brightness/darkness characteristic from upper row to lower row of the horizontally arranged rows of LED in the backlight module, cooperating with the scanning of the LCD, to thereby realize an effect similar to scanning. The general strategy that we employ is to integrate all relatively small-signal electronic functions into one ASIC to minimize the total number of the components. This strategy demonstrates that both the cost is lowered and the amount of the printed circuit board area is reduced. Based on this concept, a smart three dimensional multiplexed driver for LED switching chip with more than 640 LEDs are proposed and the circuit architecture is shown in Fig. 4. It is difference from the traditional two dimensional arrays driven by scanning scheme. Three lines are employed to control one LED, including voltage, shift register, and data line. Each LED requires a voltage line for the

eye is projected by the light from the right eye file through a liquid crystal panel.

Fig. 3. Time scheme of backlight strobe method

Fig. 4. Block diagram of control algorithm for LED backlight array module

driving current and shares the same ground with the other resistors. The resistors are individually addressable to provide unconstraint signal permutations by a serial data stream fed from the controller. The shift register is employed to shift a token bit from one group to another through AND gates to power the switch of a LED group. The selection of a LED set is thus a combined selection of the shift register for the group and the data for the specific LED. Such an arrangement allows encoding one data line from the controller to provide data to all of the LEDs, permitting high-speed scanning by shortening the LED selection path and low IC fabrication cost from the greater reduction of circuit component numbers.

## **2. 3D display backlight and application**

Throughout the proposed system, lower power consumption are successfully obtained as well as high contrast ratio even with less number of drivers than that of conventional local dimming method. This chapter also contains a new adaptive dimming algorithm and image processing technique for the proposed stereo LCD backlight system.

Recent progress in stereo display research has led to an increasing awareness of market requirements for commercial systems. In particular areas of display cost and software input to the displays are now of great importance to the programme. Possible areas of application include games displays for PC and arcade units; education and edutainment; Internet browsing for remote 3D models; scientific visualisation and medical imaging.

Intelligent and green power LED backlighting techniques of two-dimensional (2D) to threedimensional (3D) convertible type, shutter glasses type, mult-view time multiplexed naked eye type, and multi-viewer tracking type for stereo liquid crystal displays are shown as follows.

Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays 137

work by simultaneously showing a set of images ("views"), each one seen from a particular viewing angle along the horizontal direction. Such effect is achieved by adding an optical filter, which alters the propagation direction for the information displayed on

Currently, several 2D/3D switched displays had been proposed such as switched barrier and LC-lens. However, both of the parallax barrier and the cylindrical lens arrays still has the issues of narrow viewing angle and low resolution when displaying the 3D images. Besides, in order to balance the horizontal versus vertical resolution of an autostereoscopic a display, a slanted lens array is used. This causes the high crosstalk of stereo display and the subpixels of a view to appear on nonrectangular grid. This type describes the work aimed at developing optical system; active barrier dynamic backlight slit multi-view full resolution and lower crosstalk 3D panel as shown in Fig.7. The panel of 240Hz displays the corresponding images of the four viewing zones by the same time sequence according to

All modern multiview displays use TFT screens for image formation. The light generated by the TFT is separated into multiple directions by the means of special layer additionally mounted on the screen surface. Such layer is called "optical layer", "lens plate" and "optical filter". A characteristic of all 3D displays is the tradeoff between pixel resolution (or brightness or temporal frequency) and depth. In a scene viewed in 3D, pixels that in 2D would have contributed to high resolution are used instead to show depth. If the slanted lenticular sheet were placed vertically atop the LCD, then vertical and horizontal resolution

This type addresses the specific technological challenges of autostereoscopic 3D displays and presents a novel optical system that integrates a real-time active barrier dynamic backlight slit system with a naked eyes multi-view stereo display. With 240Hz display and tunable frequency LED backlight slits, only a pair of page-flipped left and right eye images was necessary to produce a multi-view effect. Furthermore, full resolution was maintained for the images of each eye. The loading of the transmission bandwidth was controllable, and the binocular parallax and motion parallax is as good as the usually full resolution multi-

Fig. 6. Schematic diagram of dynamic backlight method

the screen.

temporal multiplexed mechanism.

view autostereo display.

would drop by a factor equal to the number of views.

## **2.1 Two-dimensional (2D) to three-dimensional (3D) convertible display**

Convertible two-dimensional-three-dimensional display using an LED array based on modified integral imaging as shown in Fig.5. This type propose a two-dimensional (2D) to three-dimensional (3D) convertible display technique using a light-emitting diode (LED) array based on the principle of modified integral imaging. This system can be electrically converted between 3D and 2D modes by using different combinations of LEDs in the LED array without any mechanical movement. The LED array, which is controlled electrically, is used for backlight, and a lens array is used for making a point light source array with higher density. We explain the principle of operation and present experimental results.

Fig. 5. Display configuration

#### **2.2 Shutter glasses type stereoscopic displays**

This type proposes to employ multi-dimensional controller for driving LED backlight scanning in a 120Hz LCD for overcoming the hold-type characteristic of an LCD in timemultiplexed stereoscopic displays. A synchronization signal circuit is developed to connect the time scheme of the vertical synchronization for reducing scanning time. The general strategy is to integrate three dimensional controller and all relatively small-signal electronic functions into one ASIC to minimize the total number of the components. The display panel, LED backlight scanning, and shutter glass signals could be adjusted by vertical synchronization and modulation to obtain stereoscopic images. Each row of LED in a backlight module is controlled by multi-dimensional data registration and synchronization control circuits for LED backlight scanning to flash in bright or dark. LED backlight scanning stereoscopic display incorporated with shutter glasses is provided to realize stereoscopic images even viewed in a liquid crystal display as shown in Fig.6. The eye shutter signal is alternately switched from the left eye to the right eye with 120Hz of LCD Vertical synchronization (V-sync). This kind of low cross-talk shutter glasses stereoscopic display with an intelligent multiplexing control of LED backlight scanning has low crosstalk below 1% through a liquid crystal shutter glasses.

#### **2.3 Multi-view time multiplexed autostereoscopic displays**

Three-dimensional displays which create 3D effect without requiring the observer to wear special glasses are called autostereoscopic displays. A number of techniques exist – parallax barriers, spherical and lenticular lenses, the latter being the most common one. Depending on the design parameters, various tradeoffs between screen resolution, number of views and optimal observation distance exist. The most popular ones, so called multiview 3D displays,

Fig. 6. Schematic diagram of dynamic backlight method

Convertible two-dimensional-three-dimensional display using an LED array based on modified integral imaging as shown in Fig.5. This type propose a two-dimensional (2D) to three-dimensional (3D) convertible display technique using a light-emitting diode (LED) array based on the principle of modified integral imaging. This system can be electrically converted between 3D and 2D modes by using different combinations of LEDs in the LED array without any mechanical movement. The LED array, which is controlled electrically, is used for backlight, and a lens array is used for making a point light source array with higher

This type proposes to employ multi-dimensional controller for driving LED backlight scanning in a 120Hz LCD for overcoming the hold-type characteristic of an LCD in timemultiplexed stereoscopic displays. A synchronization signal circuit is developed to connect the time scheme of the vertical synchronization for reducing scanning time. The general strategy is to integrate three dimensional controller and all relatively small-signal electronic functions into one ASIC to minimize the total number of the components. The display panel, LED backlight scanning, and shutter glass signals could be adjusted by vertical synchronization and modulation to obtain stereoscopic images. Each row of LED in a backlight module is controlled by multi-dimensional data registration and synchronization control circuits for LED backlight scanning to flash in bright or dark. LED backlight scanning stereoscopic display incorporated with shutter glasses is provided to realize stereoscopic images even viewed in a liquid crystal display as shown in Fig.6. The eye shutter signal is alternately switched from the left eye to the right eye with 120Hz of LCD Vertical synchronization (V-sync). This kind of low cross-talk shutter glasses stereoscopic display with an intelligent multiplexing control of LED backlight scanning has low cross-

Three-dimensional displays which create 3D effect without requiring the observer to wear special glasses are called autostereoscopic displays. A number of techniques exist – parallax barriers, spherical and lenticular lenses, the latter being the most common one. Depending on the design parameters, various tradeoffs between screen resolution, number of views and optimal observation distance exist. The most popular ones, so called multiview 3D displays,

**2.1 Two-dimensional (2D) to three-dimensional (3D) convertible display** 

density. We explain the principle of operation and present experimental results.

Fig. 5. Display configuration

**2.2 Shutter glasses type stereoscopic displays** 

talk below 1% through a liquid crystal shutter glasses.

**2.3 Multi-view time multiplexed autostereoscopic displays** 

work by simultaneously showing a set of images ("views"), each one seen from a particular viewing angle along the horizontal direction. Such effect is achieved by adding an optical filter, which alters the propagation direction for the information displayed on the screen.

Currently, several 2D/3D switched displays had been proposed such as switched barrier and LC-lens. However, both of the parallax barrier and the cylindrical lens arrays still has the issues of narrow viewing angle and low resolution when displaying the 3D images. Besides, in order to balance the horizontal versus vertical resolution of an autostereoscopic a display, a slanted lens array is used. This causes the high crosstalk of stereo display and the subpixels of a view to appear on nonrectangular grid. This type describes the work aimed at developing optical system; active barrier dynamic backlight slit multi-view full resolution and lower crosstalk 3D panel as shown in Fig.7. The panel of 240Hz displays the corresponding images of the four viewing zones by the same time sequence according to temporal multiplexed mechanism.

All modern multiview displays use TFT screens for image formation. The light generated by the TFT is separated into multiple directions by the means of special layer additionally mounted on the screen surface. Such layer is called "optical layer", "lens plate" and "optical filter". A characteristic of all 3D displays is the tradeoff between pixel resolution (or brightness or temporal frequency) and depth. In a scene viewed in 3D, pixels that in 2D would have contributed to high resolution are used instead to show depth. If the slanted lenticular sheet were placed vertically atop the LCD, then vertical and horizontal resolution would drop by a factor equal to the number of views.

This type addresses the specific technological challenges of autostereoscopic 3D displays and presents a novel optical system that integrates a real-time active barrier dynamic backlight slit system with a naked eyes multi-view stereo display. With 240Hz display and tunable frequency LED backlight slits, only a pair of page-flipped left and right eye images was necessary to produce a multi-view effect. Furthermore, full resolution was maintained for the images of each eye. The loading of the transmission bandwidth was controllable, and the binocular parallax and motion parallax is as good as the usually full resolution multiview autostereo display.

Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays 139

Due to the LCD with physical delay characteristic (low response speed), both images are alternately switched from one to the other by switching the light emitting full panel. Thus, stereoscopic images are shown to a viewer. As a result, all kinds of low cross-talk stereoscopic display with an intelligent multiplexing control LED scanning backlight have

Intelligent and green energy LED backlighting techniques of stereo liquid crystal displays have been successfully designed and demonstrated a decent performance of all kinds of stereo types display system with optimized synchronization signal between LED brightness/darkness flash and adjusted driving signal enabled by a 3-Dimensional controlling IC. At high scanning rate from upper row to lower row, the system demonstrated horizontally arranged clear stereoscopic images. This method of backlighting also allows dimming to occur in locally specific areas of darkness on the screen. This can show truer blacks and whites at much higher dynamic contrast ratios, at the cost of less

As LED backlight of flat panel display become large in format, the data and gate lines turn into longer, parasitic capacitance and resistance increase, and the display signal is delayed. Three dimensional architecture of multiplexing data registration integrated circuit method is used that divides the data line into several blocks and provides the advantages of high

The design concept can be easily scaled up for large LED backlight array format TFT-LCD elements system without much change in the terminal numbers thanks to the three

Fig. 8. Viewer-tracking 3D display system

**3. Experimental and results** 

low crosstalk below 1% through a liquid crystal display.

detail in small bright objects on a dark background, such as star fields.

accuracy, rapid selection, and reasonable switching speed.

**3.1 Two-dimensional (2D) to three-dimensional (3D) convertible display** 

Fig. 7. The structure of the proposed multi-view 3D display

A lenticular-based 3D display directs the light of neighboring sub-pixels into different directions by means of small lenses placed immediately in front of the sub-pixels. In this manner different pictures can be transmitted into different directions. Usually a multitude of directions is chosen, e.g. 4 different views. Two of these views can be seen by the left and right eye respectively, and as such create a stereoscopic (3D) image. Fig.7. shows the structure of the proposed multi-view 3D display. Only one eye individually receives the image at one corresponding viewing zone at its displaying time period, such as 1/240 second. As a result, the 3D image can be created for the viewer by naked eyes. Each view is 60Hz.

#### **2.4 Multi-viewer tracking stereoscopic display**

Many people believe that in the future, autostereoscopic 3D displays will become a mainstream display type. Achievement of higher quality 3D images requires both higher panel resolution and more viewing zones. Consequently, the transmission bandwidth of the 3D display systems involves enormous amounts of data transfer. This type integrated a viewer-tracking system and a synchro-signal LED scanning backlight module with an autostereoscopic 3D display to reduce the crosstalk of right/left eye images and data transfer bandwidth, while maintaining 3D image resolution. Light-emitting diodes (LED) are a dot light source of the dynamic backlight module as shown in Fig.8. When modulating the dynamic backlight module to control the display mode of the stereoscopic display, the updating speed of the dynamic light-emitting regions and the updating speed of pixels were synchronal. For each frame period, the viewer can accurately view three-dimensional images, and the three-dimensional images displayed by the stereoscopic display have full resolution. The stereoscopic display tracks the viewer's position or can be watched by multiple viewers. This type demonstrated that the three-dimensional image displayed by the stereoscopic display is of high quality, and analyzed this phenomenon. The multi-viewer tracking stereoscopic display with intelligent multiplexing control of LED backlight scanning had low crosstalk, below 1%, when phase shift was 1/160s.

Fig. 8. Viewer-tracking 3D display system

A lenticular-based 3D display directs the light of neighboring sub-pixels into different directions by means of small lenses placed immediately in front of the sub-pixels. In this manner different pictures can be transmitted into different directions. Usually a multitude of directions is chosen, e.g. 4 different views. Two of these views can be seen by the left and right eye respectively, and as such create a stereoscopic (3D) image. Fig.7. shows the structure of the proposed multi-view 3D display. Only one eye individually receives the image at one corresponding viewing zone at its displaying time period, such as 1/240 second. As a result, the 3D image can be created for the viewer by naked eyes. Each view is

Many people believe that in the future, autostereoscopic 3D displays will become a mainstream display type. Achievement of higher quality 3D images requires both higher panel resolution and more viewing zones. Consequently, the transmission bandwidth of the 3D display systems involves enormous amounts of data transfer. This type integrated a viewer-tracking system and a synchro-signal LED scanning backlight module with an autostereoscopic 3D display to reduce the crosstalk of right/left eye images and data transfer bandwidth, while maintaining 3D image resolution. Light-emitting diodes (LED) are a dot light source of the dynamic backlight module as shown in Fig.8. When modulating the dynamic backlight module to control the display mode of the stereoscopic display, the updating speed of the dynamic light-emitting regions and the updating speed of pixels were synchronal. For each frame period, the viewer can accurately view three-dimensional images, and the three-dimensional images displayed by the stereoscopic display have full resolution. The stereoscopic display tracks the viewer's position or can be watched by multiple viewers. This type demonstrated that the three-dimensional image displayed by the stereoscopic display is of high quality, and analyzed this phenomenon. The multi-viewer tracking stereoscopic display with intelligent multiplexing control of LED backlight

Fig. 7. The structure of the proposed multi-view 3D display

**2.4 Multi-viewer tracking stereoscopic display** 

scanning had low crosstalk, below 1%, when phase shift was 1/160s.

60Hz.

Due to the LCD with physical delay characteristic (low response speed), both images are alternately switched from one to the other by switching the light emitting full panel. Thus, stereoscopic images are shown to a viewer. As a result, all kinds of low cross-talk stereoscopic display with an intelligent multiplexing control LED scanning backlight have low crosstalk below 1% through a liquid crystal display.

Intelligent and green energy LED backlighting techniques of stereo liquid crystal displays have been successfully designed and demonstrated a decent performance of all kinds of stereo types display system with optimized synchronization signal between LED brightness/darkness flash and adjusted driving signal enabled by a 3-Dimensional controlling IC. At high scanning rate from upper row to lower row, the system demonstrated horizontally arranged clear stereoscopic images. This method of backlighting also allows dimming to occur in locally specific areas of darkness on the screen. This can show truer blacks and whites at much higher dynamic contrast ratios, at the cost of less detail in small bright objects on a dark background, such as star fields.

## **3. Experimental and results**

#### **3.1 Two-dimensional (2D) to three-dimensional (3D) convertible display**

As LED backlight of flat panel display become large in format, the data and gate lines turn into longer, parasitic capacitance and resistance increase, and the display signal is delayed. Three dimensional architecture of multiplexing data registration integrated circuit method is used that divides the data line into several blocks and provides the advantages of high accuracy, rapid selection, and reasonable switching speed.

The design concept can be easily scaled up for large LED backlight array format TFT-LCD elements system without much change in the terminal numbers thanks to the three

Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays 141

increased by 3 times) thanks to the great reduction of lines for 3D scanning, instead of 2D scanning. Fig. 11 is shown pad connections from 1D, 2D, and 3D control circuits. Fig.12. is

Fig. 9. Photograph of LEDs backlight sets

Fig. 10. 3D scanning display block array

localized dimming LED backlight.

dimensional hierarchy of control circuit design, which effectively reduces the terminal numbers into the cubic root of the total control unit numbers and prevent a block defect of the flat panel. The TFT-LCD unit lights, line(s) in the vertical or horizontal axis appear dim, but not completely on or off. These defects are generally the result of a failure in the row (horizontal) or column (vertical) drivers or their connections. The TFT-LCD includes an extension part defect such as an extension piece overlapping with a pixel electrode of boundary pixels at a boundary data line applying a data signal to the boundary pixels.

#### **3.1.1 Experimental results and discussion**

A LED backlight of flat panel display with three dimensional architecture of multiplexing data registration integrated circuit having a plurality of scanning electrodes, a plurality of data electrodes extending perpendicularly to the scanning electrodes, and liquid crystal filling a space between the scanning electrodes and data electrodes, pixels being formed at each intersection of the scanning and data electrodes together with the liquid crystal, the display panel being divided into an even row part and a odd row part; a scanning control circuit for scanning the scanning electrodes by sequentially supplying scanning voltages to each scanning electrode and by maintaining the same for a predetermined period, the scanning electrodes located in the even row part of the panel and the scanning electrodes located in the odd row part being scanned separately but simultaneously in the same directions from upper to lower of the panel; an image data control circuit for sequentially supplying image data voltages to the data electrodes in synchronism with scanning of the scanning electrodes, the scanning electrodes are scanned in such a manner that the image data is written on the pixels in a selecting period, the written image data is held on the pixels in a holding period and the image data is eliminated in an eliminating period; In traditional control circuit design for TFT-LCD elements array system, each TFT-LCD element requires one driver switch. As a result, when the TFT-LCD backlight of LEDs' pixels scale up into a large array, the numbers of input/output ports will increase enormously. To handle large array of driving circuits for such large pixels array, 2D circuit architecture was employed for the traditional driving circuit to reduce the IO number from n\*n into 2n+1. However, firstly, this reduction still can not meet the requirement for high speed signal scanning with low data accessing points when switch numbers greater than 640×480 pixels. It would be necessary to increase the display frequency to 240 Hz or higher to eliminate flicker. If the display frequency is 240 Hz, a period of time for writing one frame is 4.17 ms. Assuming the number of scanning electrodes is 480, a period of time available for writing one line is only 8.7 microseconds. The number of scanning electrodes has to be larger than 480 to display a high resolution image, making the writing period further shorter. Secondly, no technology is ever completely perfect, of course, and the LCD can still suffer from some defects in the displayed image. In this technology, though, most defects in the basic electronics, such as failure of the backlight or the row or column drivers, result in a completely unusable display, and so when such occur in production they are easily detected and corrected. It is extremely rare for a product to ship with any such problems.

To achieve this, In this study, a three dimensional data registration flat panel display scheme (Fig. 9) to reduce the number of data accessing points as well as scanning lines for large array TFT-LCD element with switch number more than 640×480 is proposed (Fig.10). The total numbers of data accessing points will be N=3 <sup>3</sup> *Y* +1, which is 68 for 640×480 switches by the 3D novel design, the scanning time is reduced up to 30% (The scanning speed is also

Fig. 9. Photograph of LEDs backlight sets

dimensional hierarchy of control circuit design, which effectively reduces the terminal numbers into the cubic root of the total control unit numbers and prevent a block defect of the flat panel. The TFT-LCD unit lights, line(s) in the vertical or horizontal axis appear dim, but not completely on or off. These defects are generally the result of a failure in the row (horizontal) or column (vertical) drivers or their connections. The TFT-LCD includes an extension part defect such as an extension piece overlapping with a pixel electrode of boundary pixels at a boundary data line applying a data signal to the boundary pixels.

A LED backlight of flat panel display with three dimensional architecture of multiplexing data registration integrated circuit having a plurality of scanning electrodes, a plurality of data electrodes extending perpendicularly to the scanning electrodes, and liquid crystal filling a space between the scanning electrodes and data electrodes, pixels being formed at each intersection of the scanning and data electrodes together with the liquid crystal, the display panel being divided into an even row part and a odd row part; a scanning control circuit for scanning the scanning electrodes by sequentially supplying scanning voltages to each scanning electrode and by maintaining the same for a predetermined period, the scanning electrodes located in the even row part of the panel and the scanning electrodes located in the odd row part being scanned separately but simultaneously in the same directions from upper to lower of the panel; an image data control circuit for sequentially supplying image data voltages to the data electrodes in synchronism with scanning of the scanning electrodes, the scanning electrodes are scanned in such a manner that the image data is written on the pixels in a selecting period, the written image data is held on the pixels in a holding period and the image data is eliminated in an eliminating period; In traditional control circuit design for TFT-LCD elements array system, each TFT-LCD element requires one driver switch. As a result, when the TFT-LCD backlight of LEDs' pixels scale up into a large array, the numbers of input/output ports will increase enormously. To handle large array of driving circuits for such large pixels array, 2D circuit architecture was employed for the traditional driving circuit to reduce the IO number from n\*n into 2n+1. However, firstly, this reduction still can not meet the requirement for high speed signal scanning with low data accessing points when switch numbers greater than 640×480 pixels. It would be necessary to increase the display frequency to 240 Hz or higher to eliminate flicker. If the display frequency is 240 Hz, a period of time for writing one frame is 4.17 ms. Assuming the number of scanning electrodes is 480, a period of time available for writing one line is only 8.7 microseconds. The number of scanning electrodes has to be larger than 480 to display a high resolution image, making the writing period further shorter. Secondly, no technology is ever completely perfect, of course, and the LCD can still suffer from some defects in the displayed image. In this technology, though, most defects in the basic electronics, such as failure of the backlight or the row or column drivers, result in a completely unusable display, and so when such occur in production they are easily detected

and corrected. It is extremely rare for a product to ship with any such problems.

To achieve this, In this study, a three dimensional data registration flat panel display scheme (Fig. 9) to reduce the number of data accessing points as well as scanning lines for large array TFT-LCD element with switch number more than 640×480 is proposed (Fig.10). The total numbers of data accessing points will be N=3 <sup>3</sup> *Y* +1, which is 68 for 640×480 switches by the 3D novel design, the scanning time is reduced up to 30% (The scanning speed is also

**3.1.1 Experimental results and discussion** 

Fig. 10. 3D scanning display block array

increased by 3 times) thanks to the great reduction of lines for 3D scanning, instead of 2D scanning. Fig. 11 is shown pad connections from 1D, 2D, and 3D control circuits. Fig.12. is localized dimming LED backlight.

Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays 143

Fig. 13. The structure of 2D/3D switching design for integrated 2D/3D display

Property Specification Resolution 1680 x 1050, SXGA+

Speed 60Hz contrast 800:1

Diagonal 22 inch diagonal

The third at right is a dynamic back-light unit (DBLU), one that is similar, for example, to an LED matrix-based BLU now seen in many laptop PC display panel but with brightness of each (or at least groups) of the LED elements in the matrix separately controllable. There is a second component layer, the 2D/3D switching component, inserted between the first and third that is responsible for the automatic switching of individual pixels in the first between its 2D and 3D display mode. This technology uses a microretarder-based switching device to partially switch various parts of the display screen between 2D mode and 3D mode. Structurally, a microretarder plate has an interleaved pattern of half-wavelength-retardation and zero-retardation stripes. Working together with the microretarder is a liquid crystal (LC) switching panel inserted between the microretarder and the polarized backlight of the device. Each "pixel" of the LC switching panel functions as the switching unit cells between 2D and 3D mode regions. As a general rule of thumb, the smaller these "pixels" are, the smaller the 2D/3D switching cells can be. Though, for any practical application, the number of units required for the 2D/3D switching LC panel lands in where barely enough but with

In such a shutter-glasses type stereoscopic display, the display signal (including vertical synchronization, horizontal synchronization and data) is sent from the display card to the LCD panel. The switching of the shutter glasses is driven by the vertical synchronization signal from the display card. Due to the possible phase lag between the shutter glasses and

Panel

Table 1. Image Liquid Crystal panel specification

the lowest cost case, for example, 16 by 10.

**3.2 Shutter glasses type stereoscopic displays** 

Fig. 11. Pad connections from 1D, 2D, and 3D control circuits

Fig. 12. Localized dimming LED backlight

#### **3.1.2 Localized 2D/3D switchable maked-eye 3D display**

The patented 2D/3D control module is made of a low-resolution panel with micro-retarder film, the LCD paneland backlight dynamically controlled form a patented architecture that can switch 2D/3D mode display area, provides three-dimensional image web application. Area by demand, 2D/3D switchable, 2D coexists with 3D on the one screen, 2D area shows small character clearly and 3D area shows multi-view naked-eye stereo image.

For simultaneous display of 2D and 3D information, we have develop an integrates nakedeye 2D/3D display window technology, called integrated 2D/3D windows (i2/3DW), that can display 2D and 3D images with flexibility and best quality on the same screen. As for stereo 3D gaming, there are dialog windows or pop-up windows in the stereo 3D game frames. Without an integral 2D/3D display, 2D texts in the 3D mode windows appear in broken and blurred characters. This situation is very much annoying especially for small fonts. But, with i2/3DW's localized 2D/3D switchable display technology, 3D gaming becomes true joy because 2D texts will be as clear as they are on a 2D screen while the 3D game scenes would still be the same fascinating as on a 3D device. Fig. 13 shows the construction of an i2/3DW display made according to technology from ITRI. It comprises three "primary "component layers." The first at left is a conventional liquid crystal display panel (LCD panel). Table 1 shows the specifications of the image liquid crystal panel.

Fig. 13. The structure of 2D/3D switching design for integrated 2D/3D display


Table 1. Image Liquid Crystal panel specification

142 Features of Liquid Crystal Display Materials and Processes

**1D>>2D>>3D**

 N=2 dimension N=3 dimension

100000 N=1 dimension

**1D>>2D~3D**

**3D>>2D>>1D**

1 10 100 1000 10000 100000

The patented 2D/3D control module is made of a low-resolution panel with micro-retarder film, the LCD paneland backlight dynamically controlled form a patented architecture that can switch 2D/3D mode display area, provides three-dimensional image web application. Area by demand, 2D/3D switchable, 2D coexists with 3D on the one screen, 2D area shows

For simultaneous display of 2D and 3D information, we have develop an integrates nakedeye 2D/3D display window technology, called integrated 2D/3D windows (i2/3DW), that can display 2D and 3D images with flexibility and best quality on the same screen. As for stereo 3D gaming, there are dialog windows or pop-up windows in the stereo 3D game frames. Without an integral 2D/3D display, 2D texts in the 3D mode windows appear in broken and blurred characters. This situation is very much annoying especially for small fonts. But, with i2/3DW's localized 2D/3D switchable display technology, 3D gaming becomes true joy because 2D texts will be as clear as they are on a 2D screen while the 3D game scenes would still be the same fascinating as on a 3D device. Fig. 13 shows the construction of an i2/3DW display made according to technology from ITRI. It comprises three "primary "component layers." The first at left is a conventional liquid crystal display

panel (LCD panel). Table 1 shows the specifications of the image liquid crystal panel.

small character clearly and 3D area shows multi-view naked-eye stereo image.

Y~Switches(Log Y)

1

Fig. 12. Localized dimming LED backlight

Fig. 11. Pad connections from 1D, 2D, and 3D control circuits

**3.1.2 Localized 2D/3D switchable maked-eye 3D display** 

10

100

1000

X Connections(Log X)

10000

The third at right is a dynamic back-light unit (DBLU), one that is similar, for example, to an LED matrix-based BLU now seen in many laptop PC display panel but with brightness of each (or at least groups) of the LED elements in the matrix separately controllable. There is a second component layer, the 2D/3D switching component, inserted between the first and third that is responsible for the automatic switching of individual pixels in the first between its 2D and 3D display mode. This technology uses a microretarder-based switching device to partially switch various parts of the display screen between 2D mode and 3D mode. Structurally, a microretarder plate has an interleaved pattern of half-wavelength-retardation and zero-retardation stripes. Working together with the microretarder is a liquid crystal (LC) switching panel inserted between the microretarder and the polarized backlight of the device. Each "pixel" of the LC switching panel functions as the switching unit cells between 2D and 3D mode regions. As a general rule of thumb, the smaller these "pixels" are, the smaller the 2D/3D switching cells can be. Though, for any practical application, the number of units required for the 2D/3D switching LC panel lands in where barely enough but with the lowest cost case, for example, 16 by 10.

#### **3.2 Shutter glasses type stereoscopic displays**

In such a shutter-glasses type stereoscopic display, the display signal (including vertical synchronization, horizontal synchronization and data) is sent from the display card to the LCD panel. The switching of the shutter glasses is driven by the vertical synchronization signal from the display card. Due to the possible phase lag between the shutter glasses and

Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays 145

The CS-100 Spot Chroma Meter is used in this research to measure all the luminance values. The images are displayed in page-flipping mode using the resolution and color-depth set in

Stereo/Page-flip Setup as shown in Fig.14 and Fig.15.

Fig. 14. LED scanning backlight duty cycle

Fig. 15. The CS-100 measurement stereoscopic display system

the image liquid crystal panel.

The scanning backlight method turns on several (e.g., 2 or 4) horizontal regions of the backlight in turn, corresponding with the fill-out of the LCD panel. The backlight strobe method is to synchronously apply a control signal to the whole backlight to provide flashing effect rather than scanning. Both methods can control the brightness of the backlight module by adjusting the duty cycle of the control signal (Fig.14). Table 2 shows the specifications of

the LCD panel, a phase lag circuit is set between these two devices. The scanning or the strobe of the dynamic backlight is driven by the same vertical synchronization signal. Two experimental setups are used to implement the two dynamic backlight methods.


In order to tell which method is better for a shutter-glasses stereoscopic display, three experiments are done. They are 4-region scanning backlight, 2-region scanning backlight and backlight strobe methods. One of the most important properties of the 3D display, the crosstalk, is used in these experiments to tell which method is better. The test pattern is in a video stream form, which is like the left diagram.

The crosstalk is given by equations (1) and (2):

$$\mathbf{C}\_{\rm L} = \frac{B\mathcal{W} - BB}{\mathcal{W}\mathcal{B} - BB} \tag{1}$$

and

$$\mathbf{C}\_{R} = \frac{\text{V}\mathbf{B} - \text{BB}}{\text{B}\mathbf{V} - \text{BB}} \tag{2}$$

Where

WB represents a video stream with all-white as left-eye images, all-black as right-eye images),

BW represents a video stream with all-black as left-eye images, all-white as right-eye images),

BB represents a video stream with all-black for both left and right eyes.

CL and CR represent the crosstalk experienced by the left eye and right eye

The CS-100 Spot Chroma Meter is used in this research to measure all the luminance values. The images are displayed in page-flipping mode using the resolution and color-depth set in Stereo/Page-flip Setup as shown in Fig.14 and Fig.15.

Fig. 14. LED scanning backlight duty cycle

144 Features of Liquid Crystal Display Materials and Processes

the LCD panel, a phase lag circuit is set between these two devices. The scanning or the strobe of the dynamic backlight is driven by the same vertical synchronization signal. Two experimental setups are used to implement the two dynamic backlight methods. 1. In scanning backlight method, a novel controlled circuit architecture of scanning regions for 120Hz high frequency. Setup all the parameters of scanning backlight method by counting the amount to decide turning time between 4、and 2 LED backlight regions. If counted times equal to 100 then jump to next backlight region. For 4-region scanning backlight method, when the panel is filled in regions 1, 2, 3 and 4 by the new image, the backlight lights up in the corresponding regions 3, 4, 1 and 2. In anticipation of an image for a left eye and right eye is shown in the region 1 of the panel, we turned on region 3 of the backlight unit. Analogize the image shown in region 2 and turned on region 4 of the backlight unit. For 2-region scanning backlight method, when the panel is filled in regions 1, and 2 by the new image, the backlight lights up in the corresponding regions 2, and 1. For avoiding seeing both L-image and R-image at the same time, the backlight regions R1 have to be off until R1 filled up the image.

Analogize the backlight regions R2 have to be off until R2 filled up the image. 2. In backlight strobe method, Setup the parameters of backlight strobe method by counting the amount to decide turning on time of full screen (full screen of one frame 1/120sec counted amount equal to 400). Setup the parameters of backlight strobe duty time by counting the amount to decide turning time on full screen backlight regions. If counted times equal to (400 - 400\*9/10 ) then jump to next full screen backlight region. The backlight is turned off when the image data refreshes. The backlight only turns on at the system time, or at most a little bit longer than the system time. But the system time is short compared with the time between two adjacent vertical synchronization signals (less than 10%), the display brightness operated under this method is probably

In order to tell which method is better for a shutter-glasses stereoscopic display, three experiments are done. They are 4-region scanning backlight, 2-region scanning backlight and backlight strobe methods. One of the most important properties of the 3D display, the crosstalk, is used in these experiments to tell which method is better. The test pattern is in a

> *WB BB*

*BW BB* 

WB represents a video stream with all-white as left-eye images, all-black as right-eye

BW represents a video stream with all-black as left-eye images, all-white as right-eye

BB represents a video stream with all-black for both left and right eyes. CL and CR represent the crosstalk experienced by the left eye and right eye

(1)

(2)

quite small.

and

Where

images),

images),

video stream form, which is like the left diagram. The crosstalk is given by equations (1) and (2):

CL = *BW BB*

CR = *WB BB*

Fig. 15. The CS-100 measurement stereoscopic display system

The scanning backlight method turns on several (e.g., 2 or 4) horizontal regions of the backlight in turn, corresponding with the fill-out of the LCD panel. The backlight strobe method is to synchronously apply a control signal to the whole backlight to provide flashing effect rather than scanning. Both methods can control the brightness of the backlight module by adjusting the duty cycle of the control signal (Fig.14). Table 2 shows the specifications of the image liquid crystal panel.

Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays 147

Fig. 17. Special pattern with eye shutter.

0.00% 1.00% 2.00% 3.00% 4.00% 5.00% 6.00%

(4 Region)

**3.3 Multi-view time multiplexed autostereoscopic displays** 

(2 Region)

According to the results of and 18, the brightness of the 4R scanning backlight and the widened backlight strobe is about half of the 2R scanning backlight. But the crosstalk performance of the widened backlight strobe crosstalk is the best of three methods. Only about 1.68%. Although the 4R scanning backlight display crosstalk is higher than widened

Multi-view displays use TFT screens for image formation . The light generated by the TFT is split into multiple directions by the means of special optical layer (called also lens plate or optical filter) mounted in front of the TFT. The intensity of the light rays passing through the filter changes as a function of the angle, as if the light is directionally projected. There are two important points to note when considering multi-view screens: (1)They are not directly compatible with standard stereo-scopic footage or software. This is because, rather than display two distinct views to the viewer (as with most other stereoscopic displays), they provide up to 5, allowing the viewer to walk around the screen whilst maintaining the 3D effect. (2)Due to the display showing 4 or 5 views simultaneously, each view contains only 1/4 or 1/5 of the standard resolution of the display panel. An optimal lens, designed to handle a single situation, would be shaped to contain only as many distinct views as local

Fig. 18. Results of crosstalk calculation of different dynamic backlight methods

backlight strobe, it still performs better than the 2R scanning backlight one.

(Strobe)

Left Eye Right Eye


Table 2. Image Liquid Crystal panel specification

## **3.2.1 Results and discussion**

There is a phase difference between the vertical synchronization signal and the shutter glasses switching time. A phase lag circuit is applied to correct the difference. The waveform shown in Fig. 16 is an adjusted result.

Fig. 16. Optimized synchronization signal

The synchronized shutter glasses can be used to separate the left-eye image and the righteye image. However, even the block and transparent function of the shutter glasses is perfect, if the synchronization is not exact, or the response time of the liquid crystal is not fast enough to operate with 100 to 120 Hz, the viewer will still experience serious crosstalk. Therefore, reduction of the crosstalk is very important while making a stereoscopic display. The observation of cross-talk reduction effect special pattern is clearly shown in Fig.17. The display content is a video stream with small squares as the left-eye image and small circles for the right eye.

The luminance of the 3D LCD is measured and recorded as WB, BW and BB charts. The measurement distance is 1 meter and the recording unit is "nit". The crosstalk is calculated by the equations 1 and 2. As a result, the luminance of a 4R scanning backlight display for one viewing zone is 28.825 nits, that of a 2R scanning backlight display for one viewing zone is 57.25 nits, and that of backlight strobe display is 28.45 nits.

The crosstalk under different scanning conditions is calculated from data, including the cases of 4-Region and 2-Region scanning backlight method, and backlight strobe method (1 Region). The crosstalk of 4R and 2R scanning backlight displays are 3.2% and 5.08% respectively, and that of backlight strobe display is 1.68%.

Fig. 17. Special pattern with eye shutter.

contrast 1,000:1 (20,000:1 'MEGA'

There is a phase difference between the vertical synchronization signal and the shutter glasses switching time. A phase lag circuit is applied to correct the difference. The waveform

The synchronized shutter glasses can be used to separate the left-eye image and the righteye image. However, even the block and transparent function of the shutter glasses is perfect, if the synchronization is not exact, or the response time of the liquid crystal is not fast enough to operate with 100 to 120 Hz, the viewer will still experience serious crosstalk. Therefore, reduction of the crosstalk is very important while making a stereoscopic display. The observation of cross-talk reduction effect special pattern is clearly shown in Fig.17. The display content is a video stream with small squares as the left-eye image and small circles

The luminance of the 3D LCD is measured and recorded as WB, BW and BB charts. The measurement distance is 1 meter and the recording unit is "nit". The crosstalk is calculated by the equations 1 and 2. As a result, the luminance of a 4R scanning backlight display for one viewing zone is 28.825 nits, that of a 2R scanning backlight display for one viewing zone

The crosstalk under different scanning conditions is calculated from data, including the cases of 4-Region and 2-Region scanning backlight method, and backlight strobe method (1 Region). The crosstalk of 4R and 2R scanning backlight displays are 3.2% and 5.08%

Response time (G2G) 5ms (2D), 3ms (3D)

Dynamic Contrast)

Property Specification Resolution 1680 x 1050 Panel Diagonal 22 inch diagonal

Speed 120Hz

Table 2. Image Liquid Crystal panel specification

**3.2.1 Results and discussion** 

shown in Fig. 16 is an adjusted result.

Fig. 16. Optimized synchronization signal

is 57.25 nits, and that of backlight strobe display is 28.45 nits.

respectively, and that of backlight strobe display is 1.68%.

for the right eye.

Fig. 18. Results of crosstalk calculation of different dynamic backlight methods

According to the results of and 18, the brightness of the 4R scanning backlight and the widened backlight strobe is about half of the 2R scanning backlight. But the crosstalk performance of the widened backlight strobe crosstalk is the best of three methods. Only about 1.68%. Although the 4R scanning backlight display crosstalk is higher than widened backlight strobe, it still performs better than the 2R scanning backlight one.

## **3.3 Multi-view time multiplexed autostereoscopic displays**

Multi-view displays use TFT screens for image formation . The light generated by the TFT is split into multiple directions by the means of special optical layer (called also lens plate or optical filter) mounted in front of the TFT. The intensity of the light rays passing through the filter changes as a function of the angle, as if the light is directionally projected. There are two important points to note when considering multi-view screens: (1)They are not directly compatible with standard stereo-scopic footage or software. This is because, rather than display two distinct views to the viewer (as with most other stereoscopic displays), they provide up to 5, allowing the viewer to walk around the screen whilst maintaining the 3D effect. (2)Due to the display showing 4 or 5 views simultaneously, each view contains only 1/4 or 1/5 of the standard resolution of the display panel. An optimal lens, designed to handle a single situation, would be shaped to contain only as many distinct views as local

Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays 149

The photographs of displayed images used the luminance meter (Konica Minolta CS-200) as shown in Fig. 20. The distance from the optical sensor to the center of LED backlight panel is in 60 cm, which is the normal range of distance for watching a 3D computer monitor. Measurements of angle are done from observation point -50 degree to observation point +50 degree; view 2 is the central view. The illuminance meter has an analog output to the oscilloscope and the illuminance signal can be recorded and processed by a computer.

In this research, we observed backlight light stain structure for 3D image display based on lenticular lens array. In Fig.21, the photo is illustrated for four viewing zones 1-4 located at the viewing location. Each viewing zone uses 1/240 second to display one image. The light source at specific location is grouped corresponding to each lenticular lens of the lenticular lens array. For the four viewing zones, each lenticular lens has four groups 1-4 of light sources corresponding to four viewing zones 1-4. The four groups of light are sequentially turned on for 1/240 second. The group 1 of light source is turned on, and then the group 2 of light source is turned on next for 1/240 second. Likewise, the groups 3 and 4 of light source are sequentially turned on for 1/240 second. Generally, the multiple viewing zones equally shares 1/60 second for one image frame. The viewable zone area from first viewing zone to be contiguous to second viewing zone is 90 mm, and for 4 viewing zone of viewable area is 360 mm(The separation of viewing zones is about 90 mm, and overall width of

Property Specification Resolution 1024 x 768

Speed 240Hz contrast 1000:1

Diagonal 6.0 inch diagonal

Panel

Table 3. Image Liquid Crystal panel specification

**3.3.1 Experimental results** 

Fig. 20. Four views backlight

viewing group is 360 mm) as shown in Fig. 20.

participants; maximizing each views resolution would require a lens width sufficient to cover the same number of subpixels as views.

Our current prototype system uses a lenticular lens multiview technique with a timemultiplex autostereoscopic display based on active directional backlight (active dynamic backlight). Throughout the proposed system, lower power consumption are successfully obtained as well as high contrast ratio even with less number of drivers than that of conventional local dimming method. This architecture also contains a new adaptive dimming algorithm and image processing technique for the proposed stereo LCD backlight system. Recent progress in stereo display research has led to an increasing awareness of market requirements for commercial systems. In particular areas of display cost and software input to the displays are now of great importance to the program. Possible areas of application include games displays for PC and arcade units; education and edutainment; Internet browsing for remote 3D models; scientific visualisation and medical imaging. Intelligent and green power LED backlighting techniques of two-dimensional (2D) to threedimensional (3D) convertible type, shutter glasses type, multi-view time multiplexed naked eye type, and multi-viewer tracking type for stereo liquid crystal displays are applied for 3D display system.

According to the time sequence for turning the groups of the light source, multiple viewing zones at multiple directions are created. To meet the requirements of different one-eye images, we propose that the real-time active barrier dynamic backlight slit system on stereodisplay. To confirm our design workable, we did the optical simulation using Advanced Systems Analysis Program (ASAP) software. The detector is set at the convergent point, and the intensity profiles of the four views are shown in Fig. 19. The intensity profiles are evolving every 1/240 sec, and the separation of peaks is about 60 mm, quite close to the design value, 65 mm. The small inaccuracy resulted from the absorption of black matrix, and it makes the dead zone explicit. Setting the pixel size larger, or the black matrix smaller would improve the results. On the other hand, the center-viewing group has the least crosstalk, and side lobe groups have larger crosstalk, especially when viewing groups departs from center very much. The increase of crosstalk arises from the abbreviation when light is not close to the optical axis. Thus, the profiles of the four views confirm our design workable. Table 3 shows the specifications of the image liquid crystal panel.

Fig. 19. The optical simulation using ASAP software


Table 3. Image Liquid Crystal panel specification

## **3.3.1 Experimental results**

148 Features of Liquid Crystal Display Materials and Processes

participants; maximizing each views resolution would require a lens width sufficient to

Our current prototype system uses a lenticular lens multiview technique with a timemultiplex autostereoscopic display based on active directional backlight (active dynamic backlight). Throughout the proposed system, lower power consumption are successfully obtained as well as high contrast ratio even with less number of drivers than that of conventional local dimming method. This architecture also contains a new adaptive dimming algorithm and image processing technique for the proposed stereo LCD backlight system. Recent progress in stereo display research has led to an increasing awareness of market requirements for commercial systems. In particular areas of display cost and software input to the displays are now of great importance to the program. Possible areas of application include games displays for PC and arcade units; education and edutainment; Internet browsing for remote 3D models; scientific visualisation and medical imaging. Intelligent and green power LED backlighting techniques of two-dimensional (2D) to threedimensional (3D) convertible type, shutter glasses type, multi-view time multiplexed naked eye type, and multi-viewer tracking type for stereo liquid crystal displays are applied for 3D

According to the time sequence for turning the groups of the light source, multiple viewing zones at multiple directions are created. To meet the requirements of different one-eye images, we propose that the real-time active barrier dynamic backlight slit system on stereodisplay. To confirm our design workable, we did the optical simulation using Advanced Systems Analysis Program (ASAP) software. The detector is set at the convergent point, and the intensity profiles of the four views are shown in Fig. 19. The intensity profiles are evolving every 1/240 sec, and the separation of peaks is about 60 mm, quite close to the design value, 65 mm. The small inaccuracy resulted from the absorption of black matrix, and it makes the dead zone explicit. Setting the pixel size larger, or the black matrix smaller would improve the results. On the other hand, the center-viewing group has the least crosstalk, and side lobe groups have larger crosstalk, especially when viewing groups departs from center very much. The increase of crosstalk arises from the abbreviation when light is not close to the optical axis. Thus, the profiles of the four views confirm our design

workable. Table 3 shows the specifications of the image liquid crystal panel.

Fig. 19. The optical simulation using ASAP software

cover the same number of subpixels as views.

display system.

The photographs of displayed images used the luminance meter (Konica Minolta CS-200) as shown in Fig. 20. The distance from the optical sensor to the center of LED backlight panel is in 60 cm, which is the normal range of distance for watching a 3D computer monitor. Measurements of angle are done from observation point -50 degree to observation point +50 degree; view 2 is the central view. The illuminance meter has an analog output to the oscilloscope and the illuminance signal can be recorded and processed by a computer.

Fig. 20. Four views backlight

In this research, we observed backlight light stain structure for 3D image display based on lenticular lens array. In Fig.21, the photo is illustrated for four viewing zones 1-4 located at the viewing location. Each viewing zone uses 1/240 second to display one image. The light source at specific location is grouped corresponding to each lenticular lens of the lenticular lens array. For the four viewing zones, each lenticular lens has four groups 1-4 of light sources corresponding to four viewing zones 1-4. The four groups of light are sequentially turned on for 1/240 second. The group 1 of light source is turned on, and then the group 2 of light source is turned on next for 1/240 second. Likewise, the groups 3 and 4 of light source are sequentially turned on for 1/240 second. Generally, the multiple viewing zones equally shares 1/60 second for one image frame. The viewable zone area from first viewing zone to be contiguous to second viewing zone is 90 mm, and for 4 viewing zone of viewable area is 360 mm(The separation of viewing zones is about 90 mm, and overall width of viewing group is 360 mm) as shown in Fig. 20.

Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays 151

This study integrated an autostereoscopic display with a viewer-tracking system. Fig. 22. illustrates the basic structure of the display and table 4 shows the specifications of the image liquid crystal panel. In the proposed structure, a retarder inserted between the image panels rotated the light beam at 90°; simultaneously, a lenticular plate adjusted the light direction to show the light slit from the tracking display. Retarder film is a clear birefringent material that alters the phase of a polarized beam of light. A quarter wave plate can convert linearly polarized light (oriented at 45° from the direction of the fast/slow axis) into circularly polarized light. Conversely, the wave plate can convert a circularly polarized beam into

**3.4 Multi-viewer tracking stereoscopic display** 

Fig. 22. The structure of the proposed viewer-tracking display panel

Panel

Table 4. Image Liquid Crystal panel specification

lenticular plate.

Property Specification Resolution 1920 x 1080

Speed 120Hz contrast 1000:1 Response time (G2G) 2ms (3D)

Diagonal 23.6 inch diagonal

In this study, when the polarization direction of the incident light formed an included 45° angle with the optical axis of the retarder, the polarization of the light passing through the /2 retardation regions rotated by 90° and became orthogonal to the polarization of the light passing though the 0° retardation regions. The molding method fabricated the lenticular plate with polymeric film as the substrate material. One of the light slit pattern pairs adjusted the direction of light from the tracking panel to the viewer's eyes through the

In this display, the PDLC panel played an important role in the function of the 2D/3D switch. When the PDLC panel was turned to clear state, the microretarder interacted with the polarizers to form a parallax barrier pattern as shown in Fig.23, making the display autostereoscopic. In a case where the PDLC panel is in a diffusive state, the light passing

linearly polarized light.

Yellow stripe pattern is created by phosphor of yellow color. White LEDs are blue LED chips covered with a phosphor that absorbs some of the blue light and fluoresces with a broad spectral output ranging from mid-green to mid-red. So, the backlight modular was taken on yellow stripe.

The configuration of uni-direction diffusion lens plate is shown in Fig. 21(b). The panel of 240Hz displays the corresponding images of the four viewing zones by the same time sequence according to temporal multiplexed mechanism. The uni-direction diffusion lens plate can condense the light individually belonging to each the lenticular lens at transverse direction. The lenticular lenses of the lens array receive the light and deflect the light into each viewing zone in a time sequence, respectively.

(a) Lenticular/LED

(b) Lenticular/Optical film/LED

Fig. 21. The crosstalk under different observation scanning angles

According to the time sequence for turning the groups of the light source, multiple viewing zones at multiple directions are created. To meet the requirements of different one-eye images, we propose that the real-time active barrier dynamic backlight slit system on stereodisplay. The center viewing group has the least crosstalk, and side lobe groups have larger crosstalk, especially when viewing groups departs from center very much. The crosstalk under different observation scanning angles is showed from data in Fig. 21, including the cases of 4-views field scanning. The crosstalk of view 1 is about 5% respectively, the results are better than slanted lenticuler lens type.

## **3.4 Multi-viewer tracking stereoscopic display**

150 Features of Liquid Crystal Display Materials and Processes

Yellow stripe pattern is created by phosphor of yellow color. White LEDs are blue LED chips covered with a phosphor that absorbs some of the blue light and fluoresces with a broad spectral output ranging from mid-green to mid-red. So, the backlight modular was

The configuration of uni-direction diffusion lens plate is shown in Fig. 21(b). The panel of 240Hz displays the corresponding images of the four viewing zones by the same time sequence according to temporal multiplexed mechanism. The uni-direction diffusion lens plate can condense the light individually belonging to each the lenticular lens at transverse direction. The lenticular lenses of the lens array receive the light and deflect the light into

(a) Lenticular/LED

nits

(b) Lenticular/Optical film/LED

According to the time sequence for turning the groups of the light source, multiple viewing zones at multiple directions are created. To meet the requirements of different one-eye images, we propose that the real-time active barrier dynamic backlight slit system on stereodisplay. The center viewing group has the least crosstalk, and side lobe groups have larger crosstalk, especially when viewing groups departs from center very much. The crosstalk under different observation scanning angles is showed from data in Fig. 21, including the cases of 4-views field scanning. The crosstalk of view 1 is about 5% respectively, the results

Fig. 21. The crosstalk under different observation scanning angles

are better than slanted lenticuler lens type.

nits

Light intensity

A V1 V2 V3 V4

V1 V2 V3 V4 FV(2/25)


**Light intensity**(Polarizer+uR+Polarizer+Len)


taken on yellow stripe.

each viewing zone in a time sequence, respectively.

This study integrated an autostereoscopic display with a viewer-tracking system. Fig. 22. illustrates the basic structure of the display and table 4 shows the specifications of the image liquid crystal panel. In the proposed structure, a retarder inserted between the image panels rotated the light beam at 90°; simultaneously, a lenticular plate adjusted the light direction to show the light slit from the tracking display. Retarder film is a clear birefringent material that alters the phase of a polarized beam of light. A quarter wave plate can convert linearly polarized light (oriented at 45° from the direction of the fast/slow axis) into circularly polarized light. Conversely, the wave plate can convert a circularly polarized beam into linearly polarized light.

Fig. 22. The structure of the proposed viewer-tracking display panel


Table 4. Image Liquid Crystal panel specification

In this study, when the polarization direction of the incident light formed an included 45° angle with the optical axis of the retarder, the polarization of the light passing through the /2 retardation regions rotated by 90° and became orthogonal to the polarization of the light passing though the 0° retardation regions. The molding method fabricated the lenticular plate with polymeric film as the substrate material. One of the light slit pattern pairs adjusted the direction of light from the tracking panel to the viewer's eyes through the lenticular plate.

In this display, the PDLC panel played an important role in the function of the 2D/3D switch. When the PDLC panel was turned to clear state, the microretarder interacted with the polarizers to form a parallax barrier pattern as shown in Fig.23, making the display autostereoscopic. In a case where the PDLC panel is in a diffusive state, the light passing

Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays 153

The autostereoscopic display integrated a webcam as the real-time detection device for tracking of the viewer's head/eye positions, so that the display showed left and right eye images correctly. The computer vision-based tracking method detects viewer's eyes over a specific range and under conditions of low and fluctuating illumination. By capturing the image of the viewer in front of the display, the viewer's position is calculated and the related position data is transferred to the field programmable gate array (FPGA) controller through RS232. When the viewer recognizes that he/she is standing at the borders of the viewing zones, analyzing the captured viewer images determines the border positions of the viewing zones. The resulting eye reference pattern allows the tracker to locate the viewer's eyes in live video images. If an observer moves away from his original position, the tracking slit will vary its pattern according to the viewer's new position. The viewer still perceives two eye images separately before exceeding the webcam detection range. Fig. 25 shows the

This research addresses the specific technological challenges of autostereoscopic 3D displays and presents a novel system that integrates a real-time viewer-tracking system with an autostereoscopic display. Our successfully designed prototype utilized a FPGA system to synchronize between a display panel and tracking slit panel. With 120Hz display and tracking panels, only a pair of page-flipped left and right eye images was necessary to produce a multi-view effect. Furthermore, full resolution was maintained for the images of each eye. The loading of the transmission bandwidth was controllable, and the binocular parallax and motion parallax is as good as the usually lower resolution multi-view

viewer-tracking system.

Fig. 25. Viewer-tracking 3D display system

autostereo display.

Fig. 23. The pattern of a microretarder

through the PDLC destroys the polarization. The microretarder then loses its function as a parallax barrier, and the display becomes a general 2D display.

This study developed autostereoscopic display apparatus and a display method. The autostereoscopic display apparatus included a display panel, a backlight module, a tracking slit panel and an optical lens array. In a frame time, the display panel and tracking panel share the same synchronization signal for the display panel. The tracking panel controls the light of the backlight module. The tracking panel features tracking slit patterns and switches the slit patterns according to the synchronization signal. Until all screen data is updated, the backlight module is inactive during the frame time. A light provided by the part of the backlight regions passes through the tracking slit set, optical lens array, and the display panel in such a way that each eye separately perceives images. As shown in Fig. 24, when the viewer moves to the left, the tracking slit set changes its pattern to display the correct image.

Fig. 24. Relations between viewer and tracking panel

through the PDLC destroys the polarization. The microretarder then loses its function as a

This study developed autostereoscopic display apparatus and a display method. The autostereoscopic display apparatus included a display panel, a backlight module, a tracking slit panel and an optical lens array. In a frame time, the display panel and tracking panel share the same synchronization signal for the display panel. The tracking panel controls the light of the backlight module. The tracking panel features tracking slit patterns and switches the slit patterns according to the synchronization signal. Until all screen data is updated, the backlight module is inactive during the frame time. A light provided by the part of the backlight regions passes through the tracking slit set, optical lens array, and the display panel in such a way that each eye separately perceives images. As shown in Fig. 24, when the viewer moves to the left,

Fig. 23. The pattern of a microretarder

parallax barrier, and the display becomes a general 2D display.

the tracking slit set changes its pattern to display the correct image.

Fig. 24. Relations between viewer and tracking panel

The autostereoscopic display integrated a webcam as the real-time detection device for tracking of the viewer's head/eye positions, so that the display showed left and right eye images correctly. The computer vision-based tracking method detects viewer's eyes over a specific range and under conditions of low and fluctuating illumination. By capturing the image of the viewer in front of the display, the viewer's position is calculated and the related position data is transferred to the field programmable gate array (FPGA) controller through RS232. When the viewer recognizes that he/she is standing at the borders of the viewing zones, analyzing the captured viewer images determines the border positions of the viewing zones. The resulting eye reference pattern allows the tracker to locate the viewer's eyes in live video images. If an observer moves away from his original position, the tracking slit will vary its pattern according to the viewer's new position. The viewer still perceives two eye images separately before exceeding the webcam detection range. Fig. 25 shows the viewer-tracking system.

Fig. 25. Viewer-tracking 3D display system

This research addresses the specific technological challenges of autostereoscopic 3D displays and presents a novel system that integrates a real-time viewer-tracking system with an autostereoscopic display. Our successfully designed prototype utilized a FPGA system to synchronize between a display panel and tracking slit panel. With 120Hz display and tracking panels, only a pair of page-flipped left and right eye images was necessary to produce a multi-view effect. Furthermore, full resolution was maintained for the images of each eye. The loading of the transmission bandwidth was controllable, and the binocular parallax and motion parallax is as good as the usually lower resolution multi-view autostereo display.

Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays 155

The backlight module of the stereoscopic display is a dynamic backlight module featuring many light-emitting regions R(1)~R(4). Fig. 26A excludes the control unit and optical lens array. In the stereoscopic display, the graphic card outputs and transmits the vertical synchro-signal to the control unit. After receiving the synchro-signal, the control unit outputs the synchro-signal to control (turn on or off) the light-emitting regions R(1)~R(4). To meet the requirements of different one-eye images, we propose that the dynamic LED backlight tracking panel has have many backlight slit sets. According to the position information of viewer O and the vertical synchro-signal, one of the slit sets of the tracking panels is selected and turned-on. Each slit set includes either left or right eye slits. Light emitted from the dynamic backlight module passes through the either left or right eye slit and the display panel, and projects onto one eye of viewer O. Similarly, light emitted from the dynamic backlight module passes through the either left or right eye slit and the display panel, and projects onto the other eye of viewer O. In this way, the pair images are projected to the two eyes of viewer O, who can see accurate three-dimensional images. For example, light emitted from the dynamic backlight module passes through the left eye slit of the slit set and the display panel, and projects onto the left eye OL of viewer O. Similarly, light emitted from the dynamic backlight module passes through the right eye slit of the slit set and the display panel, and projects onto the right eye OR of viewer O. The one-eye slits are stripe-shaped and the lengths of the one-eye slits are approximately equal to the

When the display panel displays an image based on the vertical synchro-signal, the slit set of the tracking panel is enabled. Meanwhile, pixels in the updated region of the display panel display a left-eye image, but pixels in the non-updated region of the display panel still display the previous right-eye image. Light passing through the slit set of the tracking panel and the non-updated region of the display panel can be projected onto left eye OL of viewer O (i.e. a crosstalk phenomenon) if no alternative methodology is applied. This research proposes using a dynamic backlight module to suppress the crosstalk. The light-emitting regions R1~R4 of the dynamic backlight module are separately controlled according to the

During a frame period, the light-emitting regions R(1) and R(2) corresponding to the updated region are turned on and the light-emitting regions R(3) and (4) corresponding to the non-updated region are turned off. In this way, only the light-emitting regions R(1) and R(2) provide light, so that no light passes through the slit set of the tracking panel and the non-updated region of the display panel. This reduces the crosstalk phenomenon of the

As shown in Fig. 26A and Fig. 26B, the display method of the stereoscopic display comprises

First, slit data banks (Ds) corresponding to the many viewing angles of the stereoscopic display apparatus is established. Next, the control unit receives information (Dv) on the position of the viewer. The control unit compares the position information and the slit data banks stored in advance. Meanwhile, the control unit outputs the vertical synchro-signal from the graphic card to control the output mode of the dynamic backlight module and operation mode of the tracking panel. The display panel is driven to display images (i.e. image updating) according to the vertical synchro-signal output from the graphics card. Many of the light-emitting regions (R(1)~R(4)) of the dynamic backlight module are stripeshaped and the light-emitting regions R(1)~R(4) extend across the slits of the tracking panel. The extending direction of the light-emitting regions R(1)~R(4) is perpendicular to the

longitudinal length of the display panel.

vertical synchro-signal.

stereoscopic display system.

the following steps:

#### (B) LED Backlight architecture

Many types of LED backlights are applied to 2D or 3D displays. To date, research on 3D display systems has generally focused on providing uniform, collimated illumination of the LCD, rather than addressing low crosstalk issues. This study investigated the method of using an autostereoscopic multi-viewer tracking 3D display with a synchro-signal LED scanning backlight module to reduce the crosstalk of right eye and left eye images, enhancing data transfer bandwidth while maintaining image resolution. Fig. 26A is a schematic view of a stereoscopic display. Fig. 26B is a block diagram illustrating the stereoscopic display; the stereoscopic display can track the viewer's position and be watched by multiple viewers.

Fig. 26A. The schematic view illustrating a stereoscopic display.

Fig. 26B. The block diagram illustrating the stereoscopic display

Many types of LED backlights are applied to 2D or 3D displays. To date, research on 3D display systems has generally focused on providing uniform, collimated illumination of the LCD, rather than addressing low crosstalk issues. This study investigated the method of using an autostereoscopic multi-viewer tracking 3D display with a synchro-signal LED scanning backlight module to reduce the crosstalk of right eye and left eye images, enhancing data transfer bandwidth while maintaining image resolution. Fig. 26A is a schematic view of a stereoscopic display. Fig. 26B is a block diagram illustrating the stereoscopic display; the stereoscopic display can track the viewer's position and be

> **Stereo image display card**

**Image Sync**

Page-flipping

**Dynamic** 

**R(3) R(2) R(1)**

**R(4)**

**display card**

**120Hz BLU Tracking panel**

**Control Unit Stereo image** 

**Dv**

**Display panel**

**Web cam** viewer's head

**viewer's head position (Information)**

**Tracking panel**

Tracking pattern

**120Hz Display panel**

(B) LED Backlight architecture

watched by multiple viewers.

OL OR

**Tracking pattern generation (slit data banks)**

**Ds**

Fig. 26A. The schematic view illustrating a stereoscopic display.

**Dynamic Backlight Unit**

Fig. 26B. The block diagram illustrating the stereoscopic display

Choose pattern

Sync

The backlight module of the stereoscopic display is a dynamic backlight module featuring many light-emitting regions R(1)~R(4). Fig. 26A excludes the control unit and optical lens array. In the stereoscopic display, the graphic card outputs and transmits the vertical synchro-signal to the control unit. After receiving the synchro-signal, the control unit outputs the synchro-signal to control (turn on or off) the light-emitting regions R(1)~R(4).

To meet the requirements of different one-eye images, we propose that the dynamic LED backlight tracking panel has have many backlight slit sets. According to the position information of viewer O and the vertical synchro-signal, one of the slit sets of the tracking panels is selected and turned-on. Each slit set includes either left or right eye slits. Light emitted from the dynamic backlight module passes through the either left or right eye slit and the display panel, and projects onto one eye of viewer O. Similarly, light emitted from the dynamic backlight module passes through the either left or right eye slit and the display panel, and projects onto the other eye of viewer O. In this way, the pair images are projected to the two eyes of viewer O, who can see accurate three-dimensional images. For example, light emitted from the dynamic backlight module passes through the left eye slit of the slit set and the display panel, and projects onto the left eye OL of viewer O. Similarly, light emitted from the dynamic backlight module passes through the right eye slit of the slit set and the display panel, and projects onto the right eye OR of viewer O. The one-eye slits are stripe-shaped and the lengths of the one-eye slits are approximately equal to the longitudinal length of the display panel.

When the display panel displays an image based on the vertical synchro-signal, the slit set of the tracking panel is enabled. Meanwhile, pixels in the updated region of the display panel display a left-eye image, but pixels in the non-updated region of the display panel still display the previous right-eye image. Light passing through the slit set of the tracking panel and the non-updated region of the display panel can be projected onto left eye OL of viewer O (i.e. a crosstalk phenomenon) if no alternative methodology is applied. This research proposes using a dynamic backlight module to suppress the crosstalk. The light-emitting regions R1~R4 of the dynamic backlight module are separately controlled according to the vertical synchro-signal.

During a frame period, the light-emitting regions R(1) and R(2) corresponding to the updated region are turned on and the light-emitting regions R(3) and (4) corresponding to the non-updated region are turned off. In this way, only the light-emitting regions R(1) and R(2) provide light, so that no light passes through the slit set of the tracking panel and the non-updated region of the display panel. This reduces the crosstalk phenomenon of the stereoscopic display system.

As shown in Fig. 26A and Fig. 26B, the display method of the stereoscopic display comprises the following steps:

First, slit data banks (Ds) corresponding to the many viewing angles of the stereoscopic display apparatus is established. Next, the control unit receives information (Dv) on the position of the viewer. The control unit compares the position information and the slit data banks stored in advance. Meanwhile, the control unit outputs the vertical synchro-signal from the graphic card to control the output mode of the dynamic backlight module and operation mode of the tracking panel. The display panel is driven to display images (i.e. image updating) according to the vertical synchro-signal output from the graphics card. Many of the light-emitting regions (R(1)~R(4)) of the dynamic backlight module are stripeshaped and the light-emitting regions R(1)~R(4) extend across the slits of the tracking panel. The extending direction of the light-emitting regions R(1)~R(4) is perpendicular to the

Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays 157

Fig. 27. Optical sensor and CS-100 Spot Chroma Meter of luminance crosstalk measurement

Fig. 28. The synchronization relationship between image display and backlight (from top to

Fig. 29. The crosstalk under different brightness conditions

bottom)

of the 4-regions , 2-regions scanning and strobe backlight method without lenticular

extending direction of the slits of the tracking panel. Many of the light-emitting regions (R(1)~R(4)) of the dynamic backlight module are array in an arrayed manner.

#### **3.4.1 Crosstalk analysis**

To avoid ghost images, the backlight modular provides backlight control signals which are dependent on the position of an associated part of the panel. The system is provided for controlling synchronization timing between backlighting and pixel refresh, in dependence of a location of a section within the display panel. The backlight unit is separated into several regions. Let's take 4 regions as the example, the pixel response time is less than three fourths of the frame time when the illumination period is one quarter of the frame time. Optical sensor and CS-100 Spot Chroma Meter of luminance crosstalk measurement of the 4 regions 、2-regions scanning and strobe backlight method without lenticular. Frame sequential(page flip, temporal multiplexed) process, the process is referred to as alternate frame sequencing.

Crosstalk is a critical factor determining the image quality of stereoscopic displays. Also known as ghosting or leakage, high levels of crosstalk can make stereoscopic images hard to fuse and lack fidelity. Crosstalk is measured by displaying full-black and full-white in lightemitting regions R(1)~R(4) of the display system without lenticular and using an optical sensor to measure the amount of leakage between channels.

For example, the optical sensor is placed at the left eye position (either behind the left eye of 3D glasses, or in the left eye viewing zone for an autostereoscopic display) and measurements are taken for the four cross-combinations of full-white and full-black in the left and right eye-channels. An additional reading is also taken with the display in the off state. These readings can then be used in the crosstalk equations described above. This metric can be called black-and-white crosstalk and this metric is often used because maximum crosstalk occurs when the pixels in one eye-channel are full-black and the same pixels in the opposite eye-channel are full-white. According to the results, the brightness of the 4R scanning backlight and the widened backlight strobe is about half of the 2R scanning backlight. But the crosstalk performance of the widened backlight strobe crosstalk is the best of three methods. Only about 1.68% left. Although the 4R scanning backlight display crosstalk is higher than widened backlight strobe, it still performs better than the 2R scanning backlight one.

In the study, the CS-100 Spot Chroma Meter was used to measure the brightness of the backlights, which can be controlled using the duty cycle of backlight signal, as shown in Fig. 27. Moreover, photodiode s3072 was used to measure the optic characteristics of the display device.

To view the correct image from the tracking display, the synchronization relationship between image display and backlight requires calibration. Fig. 28 shows that the V-sync signal exceeds the backlight signal in 1/160s. If the V-sync signal triggers the backlight signal directly, the observer sees three white regions and one black region (not fully white or fully black). As the human eyes determine light source, a vertical signal must trigger the first region backlight (the dotted line of Fig. 28). Fig. 29 is the crosstalk of right eye and left eye under three different brightness conditions. The phase of the V-sync signal exceeds the phase of the backlight signal in 1/480s. It is too dark if the duty cycle is lower than 50%, so the three chosen duty cycles all exceeded 50%. According to Fig.29, the differences in brightness do not significantly affect the crosstalk. The performances of both eyes were approximately in agreement. The experiment selected maximum brightness.

extending direction of the slits of the tracking panel. Many of the light-emitting regions

To avoid ghost images, the backlight modular provides backlight control signals which are dependent on the position of an associated part of the panel. The system is provided for controlling synchronization timing between backlighting and pixel refresh, in dependence of a location of a section within the display panel. The backlight unit is separated into several regions. Let's take 4 regions as the example, the pixel response time is less than three fourths of the frame time when the illumination period is one quarter of the frame time. Optical sensor and CS-100 Spot Chroma Meter of luminance crosstalk measurement of the 4 regions 、2-regions scanning and strobe backlight method without lenticular. Frame sequential(page flip, temporal multiplexed) process, the process is referred to as alternate

Crosstalk is a critical factor determining the image quality of stereoscopic displays. Also known as ghosting or leakage, high levels of crosstalk can make stereoscopic images hard to fuse and lack fidelity. Crosstalk is measured by displaying full-black and full-white in lightemitting regions R(1)~R(4) of the display system without lenticular and using an optical

For example, the optical sensor is placed at the left eye position (either behind the left eye of 3D glasses, or in the left eye viewing zone for an autostereoscopic display) and measurements are taken for the four cross-combinations of full-white and full-black in the left and right eye-channels. An additional reading is also taken with the display in the off state. These readings can then be used in the crosstalk equations described above. This metric can be called black-and-white crosstalk and this metric is often used because maximum crosstalk occurs when the pixels in one eye-channel are full-black and the same pixels in the opposite eye-channel are full-white. According to the results, the brightness of the 4R scanning backlight and the widened backlight strobe is about half of the 2R scanning backlight. But the crosstalk performance of the widened backlight strobe crosstalk is the best of three methods. Only about 1.68% left. Although the 4R scanning backlight display crosstalk is higher than widened backlight strobe, it still performs better than the 2R

In the study, the CS-100 Spot Chroma Meter was used to measure the brightness of the backlights, which can be controlled using the duty cycle of backlight signal, as shown in Fig. 27. Moreover, photodiode s3072 was used to measure the optic characteristics of the display

To view the correct image from the tracking display, the synchronization relationship between image display and backlight requires calibration. Fig. 28 shows that the V-sync signal exceeds the backlight signal in 1/160s. If the V-sync signal triggers the backlight signal directly, the observer sees three white regions and one black region (not fully white or fully black). As the human eyes determine light source, a vertical signal must trigger the first region backlight (the dotted line of Fig. 28). Fig. 29 is the crosstalk of right eye and left eye under three different brightness conditions. The phase of the V-sync signal exceeds the phase of the backlight signal in 1/480s. It is too dark if the duty cycle is lower than 50%, so the three chosen duty cycles all exceeded 50%. According to Fig.29, the differences in brightness do not significantly affect the crosstalk. The performances of both eyes were

approximately in agreement. The experiment selected maximum brightness.

(R(1)~R(4)) of the dynamic backlight module are array in an arrayed manner.

sensor to measure the amount of leakage between channels.

**3.4.1 Crosstalk analysis** 

frame sequencing.

scanning backlight one.

device.

Fig. 27. Optical sensor and CS-100 Spot Chroma Meter of luminance crosstalk measurement of the 4-regions , 2-regions scanning and strobe backlight method without lenticular

Fig. 28. The synchronization relationship between image display and backlight (from top to bottom)

Fig. 29. The crosstalk under different brightness conditions

Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays 159

luminance measurement experiments. In this backlight structure, no additional brightness

Fig. 32 shows the results of the luminance intensity experiment. The entire measuring process was completed in a darkroom, which provided measurements of fairly high quality. In the luminance intensity experiment, only 40 degrees on both sides of the center of the backlight module was measured; the intensity value was captured every 0.5 degree. When measuring, only one viewing zone of the tracking panel was switched on. The luminance meter was used to scan the viewing zones horizontally. The maximum peak luminance value of the viewing zone was approximately 514 cd/m^2 when CS-200 detected the luminance intensity value near the center of the backlight module. The minimum peak luminance intensity value of the backlight was approximately 364 cd/m^2 at the edge of the viewing group. The luminance intensity range of the viewing group in front of the backlight module ranged between 264 and 514 cd/m^2. The intersection point between two adjacent

After luminance intensity measurement of viewing zones 1 to 8 was completed, data could be combined to yield a luminance intensity distribution figure to verify the optical design parameters. The peak luminance intensity value of viewing zone 4 is in front of the center of the backlight module. The tracking panel was slightly misaligned with the lenticular plate. Due to the light intensity distribution of the tracking panel and the entrance angle difference of the light path between the backlight and the lenticular lens, a stronger luminance intensity curve was measured in the central part of the viewing group. The viewing group of the 3D display system was approximately 53 cm wide at a viewing distance of 63 cm,

In building a viewer-tracking-based autostereoscopic display, the viewer's position and border positions of the viewing zones are the key parameters. To accurately define the viewer's position, the black and white pictures for both eyes were displayed as calibration images. The positions of the borders in the viewing zones could be determined by analyzing the images of the viewer captured while the viewer reported to be at the border

For a 3D display to simulate the natural vision of human beings, both binocular parallax and motion parallax are required. For a multi-view autostereoscopic display system, viewers can

luminance intensity curves may determine the borders of the viewing zones.

Fig. 32. Luminance intensity distribution of lenticular-type BLU

indicating that each viewing zone is 6.625 cm wide on average.

positions.

(B) Motion parallax function result

was lost in the optical path.

Fig. 30 shows the crosstalk of the right eye and left eye with different phase shifts between the V-sync signal and backlight signal, where the duty cycle of backlight signal is 100%. The lowest crosstalk only occurs when phase shifts are 1/160s, not both 1/480s and 1/160s. Here, light leaking to other regions and the response time of the liquid crystal affect the crosstalk (Fig.31). Fig. 31 is the response reaction of the liquid crystal from full black to full white. The horizontal axis is time (5 ms per grid) and the vertical axis is voltage (20 mV per grid). One display frame is 1/120 second, approximately equal to 8 milliseconds. The response waveform can be divided into four sections (2ms per section). The waveform of the liquid crystal still rises (section II) when the phase of V-sync signal exceeds the backlight signal in 1/480s (≒2ms); here, the phase does not reach a bright state. But, the phase shifts of 1/240s (≒4ms) and 1/160s (≒6ms), located at region III and region IV, respectively, gradually near the bright state; this explains the difference in crosstalk.

Fig. 30. The crosstalk under different phase shift conditions

Fig. 31. The response time of liquid crystal from dark state to bright state

#### **3.4.2 Measured results**

(A) The optical properties measurement

Detailed and quantitative measurements were used in the autostereoscopic display. To measure the borders and the performance of the viewing zones, a luminance meter, Minolta CS-200, was located at the designed viewing distance (630mm from the display), which reduced the display panel's optical interference during measurement. Only the backlight module, including a backlight, tracking panel, and lenticular plate, was used in optical

Fig. 30 shows the crosstalk of the right eye and left eye with different phase shifts between the V-sync signal and backlight signal, where the duty cycle of backlight signal is 100%. The lowest crosstalk only occurs when phase shifts are 1/160s, not both 1/480s and 1/160s. Here, light leaking to other regions and the response time of the liquid crystal affect the crosstalk (Fig.31). Fig. 31 is the response reaction of the liquid crystal from full black to full white. The horizontal axis is time (5 ms per grid) and the vertical axis is voltage (20 mV per grid). One display frame is 1/120 second, approximately equal to 8 milliseconds. The response waveform can be divided into four sections (2ms per section). The waveform of the liquid crystal still rises (section II) when the phase of V-sync signal exceeds the backlight signal in 1/480s (≒2ms); here, the phase does not reach a bright state. But, the phase shifts of 1/240s (≒4ms) and 1/160s (≒6ms), located at region III and region IV, respectively,

gradually near the bright state; this explains the difference in crosstalk.

Fig. 30. The crosstalk under different phase shift conditions

Fig. 31. The response time of liquid crystal from dark state to bright state

Detailed and quantitative measurements were used in the autostereoscopic display. To measure the borders and the performance of the viewing zones, a luminance meter, Minolta CS-200, was located at the designed viewing distance (630mm from the display), which reduced the display panel's optical interference during measurement. Only the backlight module, including a backlight, tracking panel, and lenticular plate, was used in optical

**3.4.2 Measured results** 

(A) The optical properties measurement

luminance measurement experiments. In this backlight structure, no additional brightness was lost in the optical path.

Fig. 32 shows the results of the luminance intensity experiment. The entire measuring process was completed in a darkroom, which provided measurements of fairly high quality. In the luminance intensity experiment, only 40 degrees on both sides of the center of the backlight module was measured; the intensity value was captured every 0.5 degree. When measuring, only one viewing zone of the tracking panel was switched on. The luminance meter was used to scan the viewing zones horizontally. The maximum peak luminance value of the viewing zone was approximately 514 cd/m^2 when CS-200 detected the luminance intensity value near the center of the backlight module. The minimum peak luminance intensity value of the backlight was approximately 364 cd/m^2 at the edge of the viewing group. The luminance intensity range of the viewing group in front of the backlight module ranged between 264 and 514 cd/m^2. The intersection point between two adjacent luminance intensity curves may determine the borders of the viewing zones.

Fig. 32. Luminance intensity distribution of lenticular-type BLU

After luminance intensity measurement of viewing zones 1 to 8 was completed, data could be combined to yield a luminance intensity distribution figure to verify the optical design parameters. The peak luminance intensity value of viewing zone 4 is in front of the center of the backlight module. The tracking panel was slightly misaligned with the lenticular plate. Due to the light intensity distribution of the tracking panel and the entrance angle difference of the light path between the backlight and the lenticular lens, a stronger luminance intensity curve was measured in the central part of the viewing group. The viewing group of the 3D display system was approximately 53 cm wide at a viewing distance of 63 cm, indicating that each viewing zone is 6.625 cm wide on average.

In building a viewer-tracking-based autostereoscopic display, the viewer's position and border positions of the viewing zones are the key parameters. To accurately define the viewer's position, the black and white pictures for both eyes were displayed as calibration images. The positions of the borders in the viewing zones could be determined by analyzing the images of the viewer captured while the viewer reported to be at the border positions.

## (B) Motion parallax function result

For a 3D display to simulate the natural vision of human beings, both binocular parallax and motion parallax are required. For a multi-view autostereoscopic display system, viewers can

Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays 161

provides about 160 pairs of stereo images to the viewer moving in the viewing angle of the

The design of the three dimensional hierarchy with control circuit for large LED backlight array, which effectively reduces the terminal numbers into the cubic root of the total control unit numbers and prevent a block defect of the flat panel. The display panel is divided into many scanning block parts, each part is separately and simultaneously scanned in the same directions to write images on the pixels on the respective scanning electrodes. These defects are generally the result of a failure in the row (horizontal) or column (vertical) drivers or their connections. We have reached the advantages of high accuracy, rapid selection, and

Several shutter-glasses type stereoscopic displays have been measured to analysis difference of their 3D performance. The less the backlight regions are, the brighter the display with scanning backlight method is. Therefore, a 2R scanning backlight is brighter than a 4R one. Nevertheless, due to better separation of a 4R scanning backlight, the crosstalk of it is less than that of a 2R scanning backlight. However, from the other aspect, the uniformity of a scanning backlight method is usually not as good as than backlight strobe method. For a higher luminance and lower crosstalk, it is suggested to combine the 4R scanning backlight method and backlight strobe method. In this way, a 120Hz LCD can be made a very good

In full resolution multi-view autostereoscopic display research, we have successfully designed and fabricated the optical system, high density active barrier dynamic LED backlight, the slit pitch is 700um, and the LED chip size is 10×23mil for full resolution multiview autostereoscopic display. From the measurement results, the dynamic LED backlight optical system can yield ideal parabolic curvature and the crosstalk is lower than 5%. Besides, the lenticular lenses of the lens array optical system was successfully received the light and deflected the light into each viewing zone in a time sequence, which could be one

A viewer-tracking-based auto- stereoscopic display of a synchro-signal LED scanning backlight system that can correspondingly send different pairs of stereo images based on the viewer's position. Additionally, an 8-view autostereoscopic display was implemented with full resolution in the display panel, achieving high 3D image quality in the preliminary configuration. Further modifications, e.g. higher precision for viewer-tracking positioning and design of more viewing zones in the display system, may improve system

P.de Greef and H. Groot Hulze, "Adaptive Dimming and Boosting Backlight for LCD-TV

E.H.A. Langendijk, R. Muijs, and W. van Beek."Quantifying Contrast Improvements and Power Savings in Displays with a 2D-Dimming Backlight",IDW(2007). P. J. Bos, K. Roehler/Beran "The pi-cell: a fast liquid-crystal optical-switching device" Mol.

System,"SID Symposium Digest Tech Paper 38, 1332-1335(2007).

of the candidates for future full resolution time-multiplexed 3D applications.

system. The viewer is consistently able to experience the reality of motion parallax.

**4. Conclusion** 

performance.

**5. References** 

reasonable switching speed flat panel.

performance stereoscopic display with shutter glasses.

Cryst. Liq. Cryst., 113, 329-339,(1984).

see stereoscopic images with binocular parallax and motion parallax within a group of viewing zones. However, a high number of viewing zones is necessary to achieve smooth motion parallax for a sufficiently large view. This normally causes significant reduction of image resolution. While maintaining good image resolution, we implemented smooth motion parallax by adopting viewer tracking function and real-time image updating in a two-view autostereoscopic display system.

(a) 35∘image content, (b) 40∘image content, (c) 45∘image content, (d) 50∘image content, (e) 55∘image content, (f) 60∘image content, (g) 65∘image content, (h) 70∘image content.

Fig. 33. The viewer-tracking-based 2D/3D switchable autostereoscopic display

This study used an ad-boost algorithm capable of evaluating important features to quickly track viewers. If the viewer's eyes were detected in specific viewing zones in front of the display, the viewer's position would determine the images of the corresponding viewing angles shown. When the viewer's eyes move inside the same viewing zones, full stop needs moving, the images for the new viewing angles are fed into the same viewing zones. The viewer experiences the motion parallax because he/she sees different images from different angle.

When the viewer's eyes continue to move and finally cross the border of the viewing zones, the images of the new angle are reversed left-and-right and presented in real-time on the display.

In this study, tracking stability was good with viewing angles ranging from -15 degrees and 15 degrees. The refresh rate achieved 30 frames per second when the resolution of the capturing image was set to 160×120. To match the resolution of the display, the resolution of image content for each eye was 840×1050. The image content was rendered from the 3D model built from 3D Max or directly captured using cameras. For the webcam coordinates, the accuracy of one pixel was about 6.25 mm according to the viewing angle of the webcam indicated and the designed viewing distance in the autostereoscopic display. Therefore, the rendering or capturing angle was set to 0.5 degree according to the accuracy of one pixel. As shown in Fig. 33, the 2D/3D switchable auto- stereoscopic display correspondingly provides about 160 pairs of stereo images to the viewer moving in the viewing angle of the system. The viewer is consistently able to experience the reality of motion parallax.

## **4. Conclusion**

160 Features of Liquid Crystal Display Materials and Processes

see stereoscopic images with binocular parallax and motion parallax within a group of viewing zones. However, a high number of viewing zones is necessary to achieve smooth motion parallax for a sufficiently large view. This normally causes significant reduction of image resolution. While maintaining good image resolution, we implemented smooth motion parallax by adopting viewer tracking function and real-time image updating in a

(a) 35∘image content, (b) 40∘image content, (c) 45∘image content, (d) 50∘image content, (e) 55∘image content, (f) 60∘image content, (g) 65∘image content, (h) 70∘image content. Fig. 33. The viewer-tracking-based 2D/3D switchable autostereoscopic display

This study used an ad-boost algorithm capable of evaluating important features to quickly track viewers. If the viewer's eyes were detected in specific viewing zones in front of the display, the viewer's position would determine the images of the corresponding viewing angles shown. When the viewer's eyes move inside the same viewing zones, full stop needs moving, the images for the new viewing angles are fed into the same viewing zones. The viewer experiences the motion parallax because he/she sees different images from different

When the viewer's eyes continue to move and finally cross the border of the viewing zones, the images of the new angle are reversed left-and-right and presented in real-time on the

In this study, tracking stability was good with viewing angles ranging from -15 degrees and 15 degrees. The refresh rate achieved 30 frames per second when the resolution of the capturing image was set to 160×120. To match the resolution of the display, the resolution of image content for each eye was 840×1050. The image content was rendered from the 3D model built from 3D Max or directly captured using cameras. For the webcam coordinates, the accuracy of one pixel was about 6.25 mm according to the viewing angle of the webcam indicated and the designed viewing distance in the autostereoscopic display. Therefore, the rendering or capturing angle was set to 0.5 degree according to the accuracy of one pixel. As shown in Fig. 33, the 2D/3D switchable auto- stereoscopic display correspondingly

two-view autostereoscopic display system.

angle.

display.

The design of the three dimensional hierarchy with control circuit for large LED backlight array, which effectively reduces the terminal numbers into the cubic root of the total control unit numbers and prevent a block defect of the flat panel. The display panel is divided into many scanning block parts, each part is separately and simultaneously scanned in the same directions to write images on the pixels on the respective scanning electrodes. These defects are generally the result of a failure in the row (horizontal) or column (vertical) drivers or their connections. We have reached the advantages of high accuracy, rapid selection, and reasonable switching speed flat panel.

Several shutter-glasses type stereoscopic displays have been measured to analysis difference of their 3D performance. The less the backlight regions are, the brighter the display with scanning backlight method is. Therefore, a 2R scanning backlight is brighter than a 4R one. Nevertheless, due to better separation of a 4R scanning backlight, the crosstalk of it is less than that of a 2R scanning backlight. However, from the other aspect, the uniformity of a scanning backlight method is usually not as good as than backlight strobe method. For a higher luminance and lower crosstalk, it is suggested to combine the 4R scanning backlight method and backlight strobe method. In this way, a 120Hz LCD can be made a very good performance stereoscopic display with shutter glasses.

In full resolution multi-view autostereoscopic display research, we have successfully designed and fabricated the optical system, high density active barrier dynamic LED backlight, the slit pitch is 700um, and the LED chip size is 10×23mil for full resolution multiview autostereoscopic display. From the measurement results, the dynamic LED backlight optical system can yield ideal parabolic curvature and the crosstalk is lower than 5%. Besides, the lenticular lenses of the lens array optical system was successfully received the light and deflected the light into each viewing zone in a time sequence, which could be one of the candidates for future full resolution time-multiplexed 3D applications.

A viewer-tracking-based auto- stereoscopic display of a synchro-signal LED scanning backlight system that can correspondingly send different pairs of stereo images based on the viewer's position. Additionally, an 8-view autostereoscopic display was implemented with full resolution in the display panel, achieving high 3D image quality in the preliminary configuration. Further modifications, e.g. higher precision for viewer-tracking positioning and design of more viewing zones in the display system, may improve system performance.

## **5. References**


Intelligent and Green Energy LED Backlighting Techniques of Stereo Liquid Crystal Displays 163

J.-Y. Son, V. V. Saveljev, Y.-J. Choi, J.-E. Bahn, and H.-H. Choi, "Parameters for designing

M.Sakata, G.Hamagashi, A.Yamashita, K.Mashitani, E.Nakayama "3D displays without

E. Kurutepe, M. R. Civanlar, and A. M. Tekalp, "Interactive transport of multi-view videos

Y. Huang and I. Essa. "Tracking multiple objects through occlusions." In IEEE Conf. on Computer Vision and Pattern Recognition, pages II: 1051-1058 (2005). F. Jurie and M. Dhome. "Real time tracking of 3D objects with occultations." In Int'l Conf. on

Thomas B. Moeslund, Adrian Hilton, and Volker Kruger. "A survey of advances in vision-

J. Pan and B. Hu. "Robust occlusion handling in object tracking." In IEEE Workshop on Object Tracking and Classification Beyond the Visible Spectrum, pages 1-8 (2007). Rajwinder Singh Brar, Phil Surman, Ian Sexton, Richard Bates, Wing Kai Lee, Klaus Hopf,

Jian-Chiun Liou and Fo-Hau Chen, "Design and fabrication of optical system for time-

Won-Sik Oh; Daeyoun Cho; Kyu-Min Cho; Gun-Woo Moon; Byungchoon Yang; Taeseok

U. Vogel, L. Kroker, K. Seidl, J. Knobbe, C. Grillberger, J. Amelung, and M. Scholles,

N. Raman and G. J. Hekstra, "Content Based Contrast Enhancement for Liquid Crystal

M. Doshi and R. Zane, "Digital architecture for driving large LED arrays with dynamic bus

S.-Y. Yseng et al., "LED Backlight Power System with Auto-tuning Regulation Voltage for

Y. Hu and M. M. Javanovic, "LED Driver With Self-Adaptive Drive Voltage,"IEEE Transactions on Power Electronics, Vol. 23, No 6, pp. 3116-3125,Nov.(2008).

plates," Opt. Eng., vol. 42, no. 11, pp.3326–3333 (2003).

Video Workshop 2006, vol. 7, no. 5, pp. 830–836(2006).

1995, Kogakuin University, 6/7 July (1995).

Image Processing, pages I: 413-416 (2001).

Volume: 5 , Issue: 1 , Page(s): 20 – 26(2009).

Understanding, 104(2):90-126 (2006).

1 – 14(2010).

11017 (2011).

(2009).

(2008).

1, pp. 18-21, Feb. (2005).

Conf. (APEC), pp. 287-293 (2007).

autostereoscopic imaging systems based on lenticular, parallax barrier and IP

special glasses by image-splitter method", pp.48-53, Proc. 3D Image Conference

for 3DTV applications,"Journal of Zhejiang University SCIENCE A: Proc. Packet

based human motion capture and analysis." Computer Vision and Image

Frank Neumann, Sally E. Day, and Eero Willman," Laser-Based Head-Tracked 3D Display Research", Journal of Display Technology, Volume: PP , Issue: 99 , Page(s):

multiplex autostereoscopic display", Optics Express, Vol. 19, Issue 12, pp. 11007-

Jang;"A Novel Two-Dimensional Adaptive Dimming Technique of X-Y Channel Drivers for LED Backlight System in LCD TVs" ,Journal of Display Technology,

"OLED backlight for autostereoscopic displays," SPIE, vol. 7237, pp. 72370U-1–9,

Displays with Backlight Modulation," IEEE Trans. Consumer Electron., vol. 51, no.

voltage regulation and phase shifted PWM," in Proc. IEEE Appl. Power Electron.

LCD Panel," in Proc. IEEE Appl. Power Electron. Conf.(APEC), pp. 551-557


P. Bos "Stereoscopic imaging system with passive viewing apparatus", US Patent

L.Lipton, J.Halnon, J.Wuopio, B. Dorworth "Eliminating pi-cells artifacts", Sterescopic Displays and Virtual Reality Systems VII, Vol. 3295, 264-270, (2000). C. vanBerkel and J.A. Clarke "Characterisation and Optimisation of 3D-LCD Module

S. Pastoor and M. Wopking 3-D displays: a review of current technologies. Displays 17,

L. Lipton Synthagram: autostereoscopic display technology. Proceedings of the SPIE, Vol.

Hamagishi, M. Sakata, A. Yamashita, K. Mashitani, E. Nakayama, S. Kishimoto, and K.

Roese, "Stereoscopic electro-optic shutter CRT displays—a basic approach," in Processing

Bohm, J. Frank, and L. Joulain, "An LCD helmet mounted display with stereoscopic view for

P. Surman, K. Hopf, I. Sexton, W.K. Lee, R. Bates, "Solving the 3D problem - The history and

C. van Berkel, D. Parker and A. Franklin, "Multiview 3D LCD," in Proc. SPIE Vol. 3012, pp.

A. Schmidt and A. Grasnick, "Multi-viewpoint autostereoscopic displays from 4D-vision",

C. van Berkel, "Image preparation for 3D-LCD," in Proceedings of SPIE, 1999, vol. 3639,

J. Konrad and P. Angiel, "Subsampling models and anti-alias filters for 3-D automultiscopic displays", IEEE Trans. Image Processing, vol.15, no.1, pp. 128-140, Jan. 2006. Xiaofang Li, Qiong-hua Wang, Yuhong Tao, Dahai Li, and Aihong Wang,"Cross-talk

Jian-Chiun Liou, Kuen Lee, Chun-Jung Chen, "Low Cross-Talk Multi-Viewer Tracking 3-D

Display Technology, VOL. 7, NO. 8, AUGUST 2011,pp.411-419 (2011).

Kanatani, "New stereoscopic LC displays without special glasses," Asia Display

and Display of Three-Dimensional Data II, J. J. Pearson, ed., Proc. SPIE 507, 102–107

helicopter flight guidance," in Helmet- and Head-Mounted Displays V, R. Lewandowski, L. Haworth, and H. Girolamo, Eds., Proc. SPIE 4021, 66–78 (2000). Jian-Chiun Liou and Fan-Gang Tseng,"120Hz Display with Intelligent LED Backlight

Enabled by Multi-Dimensional Controlling IC , Displays, Vol. 30, No. 3, pp. 147-

development of viable domestic 3-dimensional video displays", In (Haldun M. Ozaktas, Levent Onural, Eds.), Three-Dimensional Television: Capture,

in Proc. SPIE Photonics West 2002: Electronic Imaging, vol. 4660, pp. 212-221,

reduction by correcting the subpixel position in a multiview autostereoscopic threedimensional display based on a lenticular sheet", Chinese Optics Letters, Vol. 9,

Display of Synchro-Signal LED Scanning Backlight System", IEEE/OSA Journal of

S.S. Kim, K.H. Cha, J.H. Sung," 3-D Display" SID'02 Digest, pp.1422,(2002).

Transmission, and Display (ch. 13), Springer Verlag, 2007.

4,719,507,(1988).

(1997).

(1984).

153, 2009.

32-39, 1996.

pp.84-91,(1999).

Issue 2, pp. 021001- (2011).

2002.

4660, (2002).

'95, pp. 791–794 (1995).

Design" Proc SPIE Vol. 3012, (1997).


*National Kaohsiung First University of Science & Technology, Kaohsiung* 

The fabrication of TFT-LCD panel includes the growing and etching of a-silicon and silicon nitride films. These fabrication processes utilize significant amounts of silane, phosphine, ammonia, chlorine, boron trichloride, nitrogen trifluoride, fluorine, and hydrogen which are highly flammable, reactive, corrosive, and/or toxic. The amount used and scale of the supply system for these gases are far larger than for other tech industries such as the semiconductor and photovoltaic industries. Accidental leaks and fires of these gases are the

This paper first reviews the hazardous properties of the gases used in the TFT-LCD manufacturing processes. The best practices for handling these hazardous gases are then described. Finally, the past incidents and emergency response actions are also reviewed.

The TFT-LCD manufacturing processes utilize significant amount of gases for thin-film deposition and etching. These include silane, phosphine, ammonia, and hydrogen for polycrystalline silicon and silicon nitride thin film deposition in a plasma-enhanced chemical vapor deposition (PECVD) reaction chamber. In addition, nitrogen trifluoride or fluorine is used in the PECVD chamber cleaning. Chlorine and sulfur hexafluoride are used in the dry etching of thin film. The hazards of these gases can be classified as pyrophoric, flammable and oxidizing gases. There are other bulk gases such as nitrogen, argon and helium which are used

Silane or silicon tetrahydride (SiH4) is the most common silicon source used in TFT-LCD manufacturing. It is a highly flammable gas and has a very wide flammable range, from 1.37% to 96%. Silane is colorless and odorless although it has been reported to have a prudent odor and a reported time weighted threshold limit value (TLV-TWA) of 5 ppm by ACGIH. However, this data is based on analogy of another hydride gas Germane rather than the actual toxicological data. In fact, silane has no odor and its toxicity is low with a high median lethal concentration (LC50) of 9,600 ppm for rat at 4 hours exposure. The prudent odor comes from the impurity of trichlorsilane during early silane production.

for inerting or purging and will not be discussed here owing to their low risks.

**1. Introduction** 

**2.1 Silane** 

major safety concern in the TFT-LCD fabs.

**2. Hazardous properties of specialty gases** 

*2Department of Safety, Health and Environmental Engineering,* 

Eugene Y. Ngai1 and Jenq-Renn Chen2

*1Chemically Speaking LLC, New Jersey* 

*1USA 2Taiwan* 

J. Son and B. Javidi, "Three-Dimensional Imaging Methods Based on Multiview Images," J. Display Technol. Vol.:1 Issue:1, pp.125-140 (2005). **8** 

## Eugene Y. Ngai1 and Jenq-Renn Chen2

*1Chemically Speaking LLC, New Jersey 2Department of Safety, Health and Environmental Engineering, National Kaohsiung First University of Science & Technology, Kaohsiung 1USA 2Taiwan* 

## **1. Introduction**

164 Features of Liquid Crystal Display Materials and Processes

J. Son and B. Javidi, "Three-Dimensional Imaging Methods Based on Multiview Images," J.

The fabrication of TFT-LCD panel includes the growing and etching of a-silicon and silicon nitride films. These fabrication processes utilize significant amounts of silane, phosphine, ammonia, chlorine, boron trichloride, nitrogen trifluoride, fluorine, and hydrogen which are highly flammable, reactive, corrosive, and/or toxic. The amount used and scale of the supply system for these gases are far larger than for other tech industries such as the semiconductor and photovoltaic industries. Accidental leaks and fires of these gases are the major safety concern in the TFT-LCD fabs.

This paper first reviews the hazardous properties of the gases used in the TFT-LCD manufacturing processes. The best practices for handling these hazardous gases are then described. Finally, the past incidents and emergency response actions are also reviewed.

## **2. Hazardous properties of specialty gases**

The TFT-LCD manufacturing processes utilize significant amount of gases for thin-film deposition and etching. These include silane, phosphine, ammonia, and hydrogen for polycrystalline silicon and silicon nitride thin film deposition in a plasma-enhanced chemical vapor deposition (PECVD) reaction chamber. In addition, nitrogen trifluoride or fluorine is used in the PECVD chamber cleaning. Chlorine and sulfur hexafluoride are used in the dry etching of thin film. The hazards of these gases can be classified as pyrophoric, flammable and oxidizing gases. There are other bulk gases such as nitrogen, argon and helium which are used for inerting or purging and will not be discussed here owing to their low risks.

#### **2.1 Silane**

Silane or silicon tetrahydride (SiH4) is the most common silicon source used in TFT-LCD manufacturing. It is a highly flammable gas and has a very wide flammable range, from 1.37% to 96%. Silane is colorless and odorless although it has been reported to have a prudent odor and a reported time weighted threshold limit value (TLV-TWA) of 5 ppm by ACGIH. However, this data is based on analogy of another hydride gas Germane rather than the actual toxicological data. In fact, silane has no odor and its toxicity is low with a high median lethal concentration (LC50) of 9,600 ppm for rat at 4 hours exposure. The prudent odor comes from the impurity of trichlorsilane during early silane production.

Fig. 1. Critical exit velocity for delayed/prompt ignition as a function of vent diameter.

Fig. 2. Typical silane combustion powder.

Silane is used as a compressed gas. However, silane has a critical temperature of -3.4°C. It is possible that liquefaction may occur in cold storage or during expansion cooling from compressed sources. A silane supply system must be designed properly to prevent unexpected pressure surge from liquefaction and vaporization.

Silane is also a pyrophoric gas that normally ignites upon contact with air. Silane has a reported autoignition temperature of -50~-100°C. Autoignition of silane in air has also been reported down to -162°C, depending on the oxygen concentration of the mixture (Baratov et al., 1969). However, the autoignition temperature only denotes the temperature above which a given fuel-air mixture will autoignite in a closed vessel. It however does not refer to details of the autoignition process and how autoignition is affected by flow. The major potential hazard of silane is however not in its pyrophoricity but rather in its unpredictable ignition behavior in accidental releases. A silane release from a pressure source has been known not to lead to spontaneously ignition (Koda, 1992) and frequently delayed ignition occurs when the release is shut-off resulting in a "pop". In a semi-confined space with gas accumulation, the pop can lead to a gas explosion with significant damage (Ngai et al., 2007).

The mechanism of delayed ignition has been studied by Tsai et al. (2011) which demonstrated that silane release without prompt ignition was most likely caused by quench of the reactive kernel from the flow strain or scalar dissipation accompanied by the large velocity and concentration difference between the release gas and the ambient air. With diminishing release velocity, the flow strain reduces along with quench of the reactive kernel and ignition then occurs at a critical exit velocity. The ignition at reducing velocity may ignite the released gas and create a significant combustion or explosion. The critical exit velocity of indefinitely delayed ignition lies between 0.3 m/s to 4.3 m/s for vent diameter of 2.03 ~4.32 mm as shown in Figure 1. These velocities are at least two orders of magnitude smaller than the velocity from a pressurized source which in most cases reaches sonic velocity. Furthermore, Figure 1 also highlights that the critical velocity is also decreases with decreasing vent size.

There are factors other than vent size affecting the critical exit velocity. Among them, the temperature, moisture and silane combustion powder are the most noticeable factors. At higher temperature, the reactive kernel is stronger and thus the critical exit velocity is larger (Liang et al., 2010). Removing the moisture in the air also results in significantly higher critical exit velocity indicating an inhibitory role of moisture on silane ignition (Liang et al., 2010). The inhibitory action of moisture on the silane autoigniton in air is also in consistent with other fuel such as hydrogen and methane. Silane combustion produces white to brown powders as shown in Figure 2. These powders are also known to promote ignition but its exact role remains to be studied.

The presence of a critical exit velocity for prompt ignition of silane release has important implications in the safety of silane operations. First of all, almost all current silane uses in the semiconductor, TFT-LCD, and photovoltaic industries are supplied in the form of pressurized cylinders with pressure up 12.5 MPa. The high pressure silane is then regulated to 0.8 MPa for the supply tube into the cleanroom and then further regulated down to 0.3~0.45 MPa before feeding into the process chamber. Thus, almost all potential leak points in silane utilization have pressure and possible leak velocity higher than the reported critical velocity for prompt ignition. In addition, the potential leak points are the cylinder valve, tubing, and tube connections (Huang and Ngai, 2006). The potential release size will be much smaller than 2.03 mm, except for a catastrophic full bore tube rupture. Thus, delayed ignition should be considered as the usual case rather than a rare case in most silane operations.

Silane is used as a compressed gas. However, silane has a critical temperature of -3.4°C. It is possible that liquefaction may occur in cold storage or during expansion cooling from compressed sources. A silane supply system must be designed properly to prevent

Silane is also a pyrophoric gas that normally ignites upon contact with air. Silane has a reported autoignition temperature of -50~-100°C. Autoignition of silane in air has also been reported down to -162°C, depending on the oxygen concentration of the mixture (Baratov et al., 1969). However, the autoignition temperature only denotes the temperature above which a given fuel-air mixture will autoignite in a closed vessel. It however does not refer to details of the autoignition process and how autoignition is affected by flow. The major potential hazard of silane is however not in its pyrophoricity but rather in its unpredictable ignition behavior in accidental releases. A silane release from a pressure source has been known not to lead to spontaneously ignition (Koda, 1992) and frequently delayed ignition occurs when the release is shut-off resulting in a "pop". In a semi-confined space with gas accumulation,

The mechanism of delayed ignition has been studied by Tsai et al. (2011) which demonstrated that silane release without prompt ignition was most likely caused by quench of the reactive kernel from the flow strain or scalar dissipation accompanied by the large velocity and concentration difference between the release gas and the ambient air. With diminishing release velocity, the flow strain reduces along with quench of the reactive kernel and ignition then occurs at a critical exit velocity. The ignition at reducing velocity may ignite the released gas and create a significant combustion or explosion. The critical exit velocity of indefinitely delayed ignition lies between 0.3 m/s to 4.3 m/s for vent diameter of 2.03 ~4.32 mm as shown in Figure 1. These velocities are at least two orders of magnitude smaller than the velocity from a pressurized source which in most cases reaches sonic velocity. Furthermore, Figure 1 also highlights that the critical velocity is also decreases with

There are factors other than vent size affecting the critical exit velocity. Among them, the temperature, moisture and silane combustion powder are the most noticeable factors. At higher temperature, the reactive kernel is stronger and thus the critical exit velocity is larger (Liang et al., 2010). Removing the moisture in the air also results in significantly higher critical exit velocity indicating an inhibitory role of moisture on silane ignition (Liang et al., 2010). The inhibitory action of moisture on the silane autoigniton in air is also in consistent with other fuel such as hydrogen and methane. Silane combustion produces white to brown powders as shown in Figure 2. These powders are also known to promote ignition but its

The presence of a critical exit velocity for prompt ignition of silane release has important implications in the safety of silane operations. First of all, almost all current silane uses in the semiconductor, TFT-LCD, and photovoltaic industries are supplied in the form of pressurized cylinders with pressure up 12.5 MPa. The high pressure silane is then regulated to 0.8 MPa for the supply tube into the cleanroom and then further regulated down to 0.3~0.45 MPa before feeding into the process chamber. Thus, almost all potential leak points in silane utilization have pressure and possible leak velocity higher than the reported critical velocity for prompt ignition. In addition, the potential leak points are the cylinder valve, tubing, and tube connections (Huang and Ngai, 2006). The potential release size will be much smaller than 2.03 mm, except for a catastrophic full bore tube rupture. Thus, delayed ignition should be

considered as the usual case rather than a rare case in most silane operations.

the pop can lead to a gas explosion with significant damage (Ngai et al., 2007).

unexpected pressure surge from liquefaction and vaporization.

decreasing vent size.

exact role remains to be studied.

Fig. 1. Critical exit velocity for delayed/prompt ignition as a function of vent diameter.

Fig. 2. Typical silane combustion powder.

Department Agency for Toxic Substances and Disease Registry (ATSDR) sheet on

Transportation regulations prohibit the supply of pure phosphine in cylinders larger than 50 liters since it is highly toxic. As a dopant, phosphine is supplied to the process as a low concentration mixture in order to precisely control the diffusion rate for the TFT-LCD processes. As a mixture, the gas is no longer classified as highly toxic, therefore it is

Some gas suppliers are also supplying gas mixing systems which will dynamically blend the phosphine from a pure source cylinder with hydrogen. A single cylinder will supply 200

Hydrogen is a colorless, odorless, tasteless, highly flammable gas. It is also the lightest weight gas. It is much less dense than air and can disperse rapidly or accumulate in the

Hydrogen has a wide flammability range, 4% to 75% in air, and the small amount of energy required for ignition necessitate special handling to prevent the inadvertent mixing of hydrogen with air. Sources of ignition such as sparks from electrical equipment, static electricity, open flames, or extremely hot objects should be eliminated. Hydrogen and air mixtures within the flammable range can explode and may burn with an almost invisible flame. Although hydrogen is not pyrophoric like silane or phosphine, hydrogen release from a tube or a crack under high pressure may lead to spontaneous ignition. This is attributed to the low ignition energy, high blow-off velocity and the adiabatic compression

Gaseous hydrogen may be supplied in cylinders or in tubes that are designed and manufactured according to applicable codes and specifications for the pressures and temperatures involved. The pressure rating and internal volume of a container determines the quantity of hydrogen it can hold. Cylinders may be used individually or can be manifolded together to allow for a larger gas storage volume. Tubes are mounted on trucktrailer chassis or in ISO frames for transportation and are referred to as tube trailers or tube

Ammonia (NH3) is a toxic and corrosive liquefied compressed gas. It is used to grow a silicon nitride layer in the TFT-LCD industry. Pure ammonia is a white liquefied gas. It is a strongly irritating gas. It is a toxic gas with a PEL of 50 ppm and an LC50 of 7,338 ppm. Ammonia is lighter than air (specific gravity: 0.59) and is highly soluble in water. Ammonia

Ammonia has a very narrow flammable range of 16% to 25% and minimum ignition energy (MIE) 40,000 times (680 mJ) higher than that of hydrogen (0.017 mj). The common flammable gases such as propane have a MIE of 0.25 mJ which is still many times less than amonnia. The high lower flammable limit (LFL) and lighter than air vapor density makes it unlikely for an ammonia release to reach concentrations high enough to burn in open areas. It is thus classified as non-flammable for transportation purposes. The national building and fire

supplied in bulk high cylinders of 450 liters or larger.

from the shock wave of the release (Dryer et al., 2007)

regulations however classify it as flammable in storage and use.

cylinders of 0.5% concentration.

upper sections of enclosed spaces

modules, respectively.

**2.4 Ammonia** 

is thermally stable.

phosphine.

**2.3 Hydrogen** 

## **2.2 Phosphine**

Phosphine (PH3) is a highly toxic and pyrophoric liquefied compressed gas. Phosphine is used primarily as a phosphorus dopant (N-type dopant) in the TFT LCD industry. It is typically supplied as a mixture with hydrogen.

Pure phosphine is a colorless, pyrophoric liquefied gas. Although pure samples of phosphine are odorless, the gas often has an odor of garlic or decaying fish due to the presence of substituted phosphine and diphosphine (P2H4). It is a highly toxic gas with a PEL of 300 ppb, an LC50 of 20 ppm for rat at 1 hour exposure.

Phosphine has a low auto-ignition temperature <0° C so it may ignite spontaneously on contact with air. Autoignition of phosphine released into air is also subjected to effects by temperature, moisture, and flow strain just like silane. For example, autoignition of phosphine is reported to occur only in dry air but not in moist air (Brittion, 1990). However, there is still a lack of detailed studies on the exact conditions for these effects. It also has an extremely wide flammable range of 1.6% – 95% in air. It burns in air with an orange flame forming white phosphorous pentoxide as a byproduct as shown in Figure 3(a), which is a severe respiratory irritant. Under certain conditions, phosphine will form red phosphorus as a byproduct in a fire as shown in Figure 3(b).

Fig. 3. Typical phosphine combustion flame and powder. (a) Orange flame and white phosphorous pentoxide powder (b) Red phosphorus powder.

Acute exposures to phosphine cause respiratory tract irritation that attacks primarily the cardiovascular and respiratory systems causing peripheral vascular collapse, cardiac arrest and failure, and pulmonary edema. Inhalation is the primary exposure route, and there is no known dermal exposure route. Symptoms may include restlessness, irritability, drowsiness, tremors, vertigo, double vision (diplopia), ataxia, cough, dyspnea, retrosternal discomfort, abdominal pain, and vomiting. There is no antidote for phosphine exposures; treatment is supportive. Guidance on proper medical treatment is available from the US Health Department Agency for Toxic Substances and Disease Registry (ATSDR) sheet on phosphine.

Transportation regulations prohibit the supply of pure phosphine in cylinders larger than 50 liters since it is highly toxic. As a dopant, phosphine is supplied to the process as a low concentration mixture in order to precisely control the diffusion rate for the TFT-LCD processes. As a mixture, the gas is no longer classified as highly toxic, therefore it is supplied in bulk high cylinders of 450 liters or larger.

Some gas suppliers are also supplying gas mixing systems which will dynamically blend the phosphine from a pure source cylinder with hydrogen. A single cylinder will supply 200 cylinders of 0.5% concentration.

#### **2.3 Hydrogen**

168 Features of Liquid Crystal Display Materials and Processes

Phosphine (PH3) is a highly toxic and pyrophoric liquefied compressed gas. Phosphine is used primarily as a phosphorus dopant (N-type dopant) in the TFT LCD industry. It is

Pure phosphine is a colorless, pyrophoric liquefied gas. Although pure samples of phosphine are odorless, the gas often has an odor of garlic or decaying fish due to the presence of substituted phosphine and diphosphine (P2H4). It is a highly toxic gas with a

contact with air. Autoignition of phosphine released into air is also subjected to effects by temperature, moisture, and flow strain just like silane. For example, autoignition of phosphine is reported to occur only in dry air but not in moist air (Brittion, 1990). However, there is still a lack of detailed studies on the exact conditions for these effects. It also has an extremely wide flammable range of 1.6% – 95% in air. It burns in air with an orange flame forming white phosphorous pentoxide as a byproduct as shown in Figure 3(a), which is a severe respiratory irritant. Under certain conditions, phosphine will form red phosphorus as

C so it may ignite spontaneously on

**2.2 Phosphine** 

typically supplied as a mixture with hydrogen.

a byproduct in a fire as shown in Figure 3(b).

PEL of 300 ppb, an LC50 of 20 ppm for rat at 1 hour exposure.

(a) (b)

phosphorous pentoxide powder (b) Red phosphorus powder.

Fig. 3. Typical phosphine combustion flame and powder. (a) Orange flame and white

Acute exposures to phosphine cause respiratory tract irritation that attacks primarily the cardiovascular and respiratory systems causing peripheral vascular collapse, cardiac arrest and failure, and pulmonary edema. Inhalation is the primary exposure route, and there is no known dermal exposure route. Symptoms may include restlessness, irritability, drowsiness, tremors, vertigo, double vision (diplopia), ataxia, cough, dyspnea, retrosternal discomfort, abdominal pain, and vomiting. There is no antidote for phosphine exposures; treatment is supportive. Guidance on proper medical treatment is available from the US Health

Phosphine has a low auto-ignition temperature <0°

Hydrogen is a colorless, odorless, tasteless, highly flammable gas. It is also the lightest weight gas. It is much less dense than air and can disperse rapidly or accumulate in the upper sections of enclosed spaces

Hydrogen has a wide flammability range, 4% to 75% in air, and the small amount of energy required for ignition necessitate special handling to prevent the inadvertent mixing of hydrogen with air. Sources of ignition such as sparks from electrical equipment, static electricity, open flames, or extremely hot objects should be eliminated. Hydrogen and air mixtures within the flammable range can explode and may burn with an almost invisible flame. Although hydrogen is not pyrophoric like silane or phosphine, hydrogen release from a tube or a crack under high pressure may lead to spontaneous ignition. This is attributed to the low ignition energy, high blow-off velocity and the adiabatic compression from the shock wave of the release (Dryer et al., 2007)

Gaseous hydrogen may be supplied in cylinders or in tubes that are designed and manufactured according to applicable codes and specifications for the pressures and temperatures involved. The pressure rating and internal volume of a container determines the quantity of hydrogen it can hold. Cylinders may be used individually or can be manifolded together to allow for a larger gas storage volume. Tubes are mounted on trucktrailer chassis or in ISO frames for transportation and are referred to as tube trailers or tube modules, respectively.

#### **2.4 Ammonia**

Ammonia (NH3) is a toxic and corrosive liquefied compressed gas. It is used to grow a silicon nitride layer in the TFT-LCD industry. Pure ammonia is a white liquefied gas. It is a strongly irritating gas. It is a toxic gas with a PEL of 50 ppm and an LC50 of 7,338 ppm. Ammonia is lighter than air (specific gravity: 0.59) and is highly soluble in water. Ammonia is thermally stable.

Ammonia has a very narrow flammable range of 16% to 25% and minimum ignition energy (MIE) 40,000 times (680 mJ) higher than that of hydrogen (0.017 mj). The common flammable gases such as propane have a MIE of 0.25 mJ which is still many times less than amonnia. The high lower flammable limit (LFL) and lighter than air vapor density makes it unlikely for an ammonia release to reach concentrations high enough to burn in open areas. It is thus classified as non-flammable for transportation purposes. The national building and fire regulations however classify it as flammable in storage and use.

some cases it will cause flu-like symptoms. Guidance on proper medical treatment is available from the US Health Department Agency for Toxic Substances and Disease Registry

Chlorine is an oxidizer gas that is equivalent to a 60% concentration of pure oxygen or almost 3 times that of air. As an oxidizer chlorine is reactive with flammable gases such as silane or hydrogen. In some cases it can be explosive, even in concentrations as low as 3% when mixed with these gases. Pure chlorine has also autoignited flammable materials. Similar to ammonia, bulk chlorine is supplied in low or high pressure bulk containers (250

Fluorine is a highly toxic gas with extreme oxidation and corrosion potentials. It has a sharp, pungent odor that can be detected by most people at very low levels (0.1 ppm). It has a TLV-TWA of 1 ppm and an LC50 of 185 ppm (1 hr. rat). Chronic fluorine exposures at low levels may cause fluorosis or abnormal calcium accumulation in bone structure. Fluorine is highly reactive with moisture, forming hydrofluoric acid which is extremely corrosive to human

Medical treatment of exposure is specific and specialized. Releases into air can form varying concentrations of hydrofluoric acid, depending on conditions. When the aqueous hydrofluoric acid concentration is below 50%, it has delayed (up to six hours) symptoms of

Fluorine is the most reactive of all elements and the most powerfully know oxidizing agent. Fluorine is able to react with almost all elements and compounds depending on pressure and temperature, with the exception of lighter noble gases (e.g. helium, argon), inorganic fluorides of the highest valency, and perfluorinated organic compounds. It reacts with many organic substances even at room temperatures and often accompanied by combustion and possible explosion (European Industrial Gases Association, 2011). Therefore special steps, such as special cleaning and passivation procedures, must be

Fluorine reacts vigorously at ambient temperatures with most metals. The reaction intensity depends upon the surface area of the metal. Powdered metals or fine wires may react violently. Most inorganic materials react with fluorine; water forms hydrofluoric acid, and salts convert to fluorides. Contact with organic materials generally results in ignition or

Due to its extreme reactivity, national transportation regulations typically limit the total amount of fluorine in a 50 liter cylinder to 3 kg or to a pressure of 30 bar. Pure fluorine cylinders can be fairly reactive at 30 bar. Some users use a mixture which is less reactive. Testing by Air Products and Chemical Inc. (2004) has shown that fluorine is 2.5 times more reactive than oxygen, when it is mixed in a 20% concentration with nitrogen it will not ignite

Testing by Ngai (2005) has shown that pure fluorine even at a pressure as low as 1 atm (101 kPa) can cause 316 stainless steel tubing to burn. Pure fluorine did not react immediately with the raw chicken. To better simulate human skin, some human hair was placed on the chicken skin. This immediately ignited and started the skin on fire as shown in Figure 4. The reaction was so hot that it ignited the stainless steel tubing, which sprayed the chicken with molten stainless steel. This continued to react until the fluorine flow was stopped

tissue. Fluorine will have an immediate effect on the eyes and respiratory system.

exposure. In a fire, fluorine will react to form other toxic fluoride compounds.

(ATSDR) sheet on chlorine.

taken to protect systems.

a ¼" carbon steel rod at 13,790 kPa.

violent explosion.

(Ngai, 2005).

**2.7 Fluorine** 

liter - 950 liter). It is vaporized for use.

As a flammable gas, ammonia is reactive with oxidizer gases such as nitrous oxide or oxygen. Since ammonia has an LC50 higher than 5,000 ppm (threshold for toxic gas), the United States classifies it as a non-toxic non-flammable gas. All other countries classify ammonia as a toxic gas. Ammonia is a severe irritant with a sharp pungent odor that will have an immediate effect on moist tissues, eyes, upper respiratory system and skin. Guidance on proper medical treatment is available from the US Health Department Agency for Toxic Substances and Disease Registry (ATSDR) sheet on Ammonia.

Ammonia is very corrosive to low alloy carbon steel, causing stress corrosion. Ammonia is also corrosive to viton, zinc and copper. Bulk ammonia is supplied in low or high pressure bulk containers (250 liter - 950 liter). Some facilities use 40,000 lb ISO Modules that are sited like the nitrogen trifluoride or silane modules for continuous supply. It is vaporized for use on demand.

## **2.5 Nitrogen trifluoride**

Nitrogen trifluoride (NF3) is a colorless, odorless, nonflammable, oxidizing compressed gas. Nitrogen trifluoride is the most common reactor cleaning gases used in the TFT-LCD industry because of its advantages, such as high etching rates, high selectivity, carbon-free etching, and minimal residual contamination. It is first decomposed in a remote plasma chamber into a reactive fluorine before fed to the reaction chamber where it is used to react solid deposits on the reactor walls to form gaseous byproducts such as silicon tetrafluoride.

Nitrogen trifluoride is not toxic, it has an LC50 of 6,700 ppm (1 hr. rat). It is not hazardous by skin contact and is a relatively minor irritant to the eyes and mucous membranes. Exposure of rats to a 100 ppm concentration for 7 hrs per day, 5 days per week over a 19 week period caused minor pathological changes to the liver and kidney. No other effects were noted. The ACGIH established the current TLV-TWA of 10 ppm based on 1/10 of this value. Some countries classify nitrogen trifluoride as a toxic gas because it has a TLV-TWA <200 ppm.

Under ambient temperatures and pressures, nitrogen trifluoride is inert. However it can become an extremely reactive under certain conditions. It is a strong oxidizer that is violently explosive at elevated pressures with flammable gases such as hydrogen or silane. Purification of nitrogen trifluoride using a solid absorbent media such as molecular sieve is to be avoided. Fires and explosions have occurred.

Nitrogen trifluoride is reported to be a significant greenhouse warming gas with a Greenhouse Warming Potential (GWP) >10,000 as compared to carbon dioxide. Properly design remote plasma cleaning system will react 100% of the nitrogen trifluoride, so it has a GWP comparable to fluorine under these conditions.

#### **2.6 Chlorine**

Chlorine (Cl2) is a toxic and corrosive liquefied compressed gas. It is also an oxidizing gas. Chlorine is used primarily as an etching gas in the TFT-LCD industry. Pure chlorine is a yellow-green liquefied gas. It is a strongly irritating and highly toxic gas with a PEL of 0.5 ppm and an LC50 of 293 ppm. Chlorine is much heavier than air (specific gravity: 2.49) and is slightly soluble in water. Chlorine is thermally stable.

Chlorine is a severe irritant with a pungent odor that will have an immediate effect on moist tissues, eyes, upper respiratory system and skin. Chronic exposure will corrode teeth, in some cases it will cause flu-like symptoms. Guidance on proper medical treatment is available from the US Health Department Agency for Toxic Substances and Disease Registry (ATSDR) sheet on chlorine.

Chlorine is an oxidizer gas that is equivalent to a 60% concentration of pure oxygen or almost 3 times that of air. As an oxidizer chlorine is reactive with flammable gases such as silane or hydrogen. In some cases it can be explosive, even in concentrations as low as 3% when mixed with these gases. Pure chlorine has also autoignited flammable materials.

Similar to ammonia, bulk chlorine is supplied in low or high pressure bulk containers (250 liter - 950 liter). It is vaporized for use.

#### **2.7 Fluorine**

170 Features of Liquid Crystal Display Materials and Processes

As a flammable gas, ammonia is reactive with oxidizer gases such as nitrous oxide or oxygen. Since ammonia has an LC50 higher than 5,000 ppm (threshold for toxic gas), the United States classifies it as a non-toxic non-flammable gas. All other countries classify ammonia as a toxic gas. Ammonia is a severe irritant with a sharp pungent odor that will have an immediate effect on moist tissues, eyes, upper respiratory system and skin. Guidance on proper medical treatment is available from the US Health Department Agency

Ammonia is very corrosive to low alloy carbon steel, causing stress corrosion. Ammonia is also corrosive to viton, zinc and copper. Bulk ammonia is supplied in low or high pressure bulk containers (250 liter - 950 liter). Some facilities use 40,000 lb ISO Modules that are sited like the nitrogen trifluoride or silane modules for continuous supply. It is vaporized for use

Nitrogen trifluoride (NF3) is a colorless, odorless, nonflammable, oxidizing compressed gas. Nitrogen trifluoride is the most common reactor cleaning gases used in the TFT-LCD industry because of its advantages, such as high etching rates, high selectivity, carbon-free etching, and minimal residual contamination. It is first decomposed in a remote plasma chamber into a reactive fluorine before fed to the reaction chamber where it is used to react solid deposits on the reactor walls to form gaseous byproducts such as silicon

Nitrogen trifluoride is not toxic, it has an LC50 of 6,700 ppm (1 hr. rat). It is not hazardous by skin contact and is a relatively minor irritant to the eyes and mucous membranes. Exposure of rats to a 100 ppm concentration for 7 hrs per day, 5 days per week over a 19 week period caused minor pathological changes to the liver and kidney. No other effects were noted. The ACGIH established the current TLV-TWA of 10 ppm based on 1/10 of this value. Some countries classify nitrogen trifluoride as a toxic gas because it has a TLV-

Under ambient temperatures and pressures, nitrogen trifluoride is inert. However it can become an extremely reactive under certain conditions. It is a strong oxidizer that is violently explosive at elevated pressures with flammable gases such as hydrogen or silane. Purification of nitrogen trifluoride using a solid absorbent media such as molecular sieve is

Nitrogen trifluoride is reported to be a significant greenhouse warming gas with a Greenhouse Warming Potential (GWP) >10,000 as compared to carbon dioxide. Properly design remote plasma cleaning system will react 100% of the nitrogen trifluoride, so it has a

Chlorine (Cl2) is a toxic and corrosive liquefied compressed gas. It is also an oxidizing gas. Chlorine is used primarily as an etching gas in the TFT-LCD industry. Pure chlorine is a yellow-green liquefied gas. It is a strongly irritating and highly toxic gas with a PEL of 0.5 ppm and an LC50 of 293 ppm. Chlorine is much heavier than air (specific gravity: 2.49) and is

Chlorine is a severe irritant with a pungent odor that will have an immediate effect on moist tissues, eyes, upper respiratory system and skin. Chronic exposure will corrode teeth, in

for Toxic Substances and Disease Registry (ATSDR) sheet on Ammonia.

on demand.

tetrafluoride.

TWA <200 ppm.

**2.6 Chlorine** 

to be avoided. Fires and explosions have occurred.

GWP comparable to fluorine under these conditions.

slightly soluble in water. Chlorine is thermally stable.

**2.5 Nitrogen trifluoride** 

Fluorine is a highly toxic gas with extreme oxidation and corrosion potentials. It has a sharp, pungent odor that can be detected by most people at very low levels (0.1 ppm). It has a TLV-TWA of 1 ppm and an LC50 of 185 ppm (1 hr. rat). Chronic fluorine exposures at low levels may cause fluorosis or abnormal calcium accumulation in bone structure. Fluorine is highly reactive with moisture, forming hydrofluoric acid which is extremely corrosive to human tissue. Fluorine will have an immediate effect on the eyes and respiratory system.

Medical treatment of exposure is specific and specialized. Releases into air can form varying concentrations of hydrofluoric acid, depending on conditions. When the aqueous hydrofluoric acid concentration is below 50%, it has delayed (up to six hours) symptoms of exposure. In a fire, fluorine will react to form other toxic fluoride compounds.

Fluorine is the most reactive of all elements and the most powerfully know oxidizing agent. Fluorine is able to react with almost all elements and compounds depending on pressure and temperature, with the exception of lighter noble gases (e.g. helium, argon), inorganic fluorides of the highest valency, and perfluorinated organic compounds. It reacts with many organic substances even at room temperatures and often accompanied by combustion and possible explosion (European Industrial Gases Association, 2011). Therefore special steps, such as special cleaning and passivation procedures, must be taken to protect systems.

Fluorine reacts vigorously at ambient temperatures with most metals. The reaction intensity depends upon the surface area of the metal. Powdered metals or fine wires may react violently. Most inorganic materials react with fluorine; water forms hydrofluoric acid, and salts convert to fluorides. Contact with organic materials generally results in ignition or violent explosion.

Due to its extreme reactivity, national transportation regulations typically limit the total amount of fluorine in a 50 liter cylinder to 3 kg or to a pressure of 30 bar. Pure fluorine cylinders can be fairly reactive at 30 bar. Some users use a mixture which is less reactive. Testing by Air Products and Chemical Inc. (2004) has shown that fluorine is 2.5 times more reactive than oxygen, when it is mixed in a 20% concentration with nitrogen it will not ignite a ¼" carbon steel rod at 13,790 kPa.

Testing by Ngai (2005) has shown that pure fluorine even at a pressure as low as 1 atm (101 kPa) can cause 316 stainless steel tubing to burn. Pure fluorine did not react immediately with the raw chicken. To better simulate human skin, some human hair was placed on the chicken skin. This immediately ignited and started the skin on fire as shown in Figure 4. The reaction was so hot that it ignited the stainless steel tubing, which sprayed the chicken with molten stainless steel. This continued to react until the fluorine flow was stopped (Ngai, 2005).

For high flow rate greater than 50 L/min, large pressure drop will produce liquid silane mixture causing failure and leaks mostly in regulators. Thus, use of 2 regulators in series and heaters to compensate Joule-Thomson expansion cooling and prevent condensation.

As fluorine reacts with almost everything, corrosion of materials in contact with fluorine is dependent upon the fluoride formed with the material. If the fluoride is volatile or in the form of loose flakes, new material will constantly be exposed to the fluorine and will be continuously eroded until it burns through the material. For this reason material selection is extremely important. Materials that form a good fluoride coating will be protected from

Systems for fluorine service must not only be fabricated of the proper material, the design of

 Adiabatic compression - Adiabatic compression and heating is caused by the rapid pressurization of a system where the gas absorbs the energy and the gas temperature rises. This heating occurs at the point of compression or the point where the flow of gas is stopped, such as at a valve or regulator seat. Depending on the material in use where

 High flow – High flow velocity increases the potential for a particle to impinge onto a surface with sufficient force to ignite it. A common area where this occurs is at a

 Reactive materials - If the valve or regulator has a seat made of a nonmetallic material, it may be more prone to ignition because nonmetals typically have lower ignition

 Cleaning - to remove contaminants such as oil and particles from the system. These can be ignited easily generating sufficient heat to propagate a reaction to the system. Passivation – Fluorine passivation of metals is a pretreatment of the components in order to stabilize the wetted surfaces of the system by formation of a self-limiting protective fluoride film on the surface of the metal. The typical film thickness range from thickness of 5 Å (0.0005 microns) to 50 Å (0.005 microns) depending on the metal

 Friction - from a component that is malfunctioning or operating poorly can generate heat. Friction between two materials generates fine particles, which may ignite from the

Silane and fluorine represent the two extremes, the flammable and the oxidizer, of the specialty gases used in TFT-LCD manufacturing. All other gases bear similar but less extensive practices and protection measures as these two gases. For example, gas detectors

the hot gas impinges, the heat may be sufficient to ignite the material.

Drying – removal of moisture from the system surface is essential

Normally closed pneumatic shut off valves

High pressure test and He outboard leak test

the system must consider the following factors:

direction of flow change such as an elbow.

(Air Force Rocket Propulsion Laboratory, 1967).

temperatures than metals.

heat generated.

**3.3 Other specialty gases** 

Emergency remote shut down

**3.2 Fluorine** 

further attack.

Dedicated purge sources with automated purge cycles

Fig. 4. Reaction of fluorine and chicken skin. (a) The hair on the skin reacts with the fluorine and ignites the chicken skin. (b) Skin burns hot enough to ignite the stainless steel tubing, which is sprayed at the chicken. (c) The tubing continues to react until the flow is stopped.

Despite its toxicity and reactivity, fluorine is being increasingly used as a reactor cleaning gas owing to less energy requirement, lower Greenhouse Warming Potential, and faster clean time (Stockman, 2009). To supply pure fluorine, small generators are being installed and managed by gas suppliers. These limit the amount of fluorine stored. A typical fluorine generator is an electrolytic cell which dialysis hydrogen fluoride to form fluorine and hydrogen (Stockman, 2009).

## **3. Best practice for handling**

## **3.1 Silane**

The first step in safe handling of silane is to comply with relevant regulatory & industrial standards such as Compressed Gas Association (CGA), Semiconductor Equipment and Materials International (SEMI), National Fire Protection Association (NFPA), Factory Mutual (FM), European Industrial Gas Association (EIGA), Asian Industrial Gas Association (AIGA), Transportable Pressure Equipment Directive (TPED), Transportation of dangerous goods (ADR), SEVESO, ATEX, etc. The installation and operation practices are however not necessarily detailed in the above codes or standards. Rufin (2011) has presented a comprehensive list of best practices for silane supply system. The key features are summarized below.

From the perspectives of preventing silane explosion, the most effective measure is preventing the accumulation of unignited silane through the use of forced ventilation or open air storage. It is also important to detect an unignited release by installing a gas sensor. Flame (UV/IR) detector is also needed for detecting leaks with autoignition. Sprinklers are needed for cooling the cylinders and system. Operators must be also be protected by barriers and flame resistant protection suit.

Other minor but equally important issues include:


For high flow rate greater than 50 L/min, large pressure drop will produce liquid silane mixture causing failure and leaks mostly in regulators. Thus, use of 2 regulators in series and heaters to compensate Joule-Thomson expansion cooling and prevent condensation.

## **3.2 Fluorine**

172 Features of Liquid Crystal Display Materials and Processes

Fig. 4. Reaction of fluorine and chicken skin. (a) The hair on the skin reacts with the fluorine and ignites the chicken skin. (b) Skin burns hot enough to ignite the stainless steel tubing, which is sprayed at the chicken. (c) The tubing continues to react until the flow is stopped. Despite its toxicity and reactivity, fluorine is being increasingly used as a reactor cleaning gas owing to less energy requirement, lower Greenhouse Warming Potential, and faster clean time (Stockman, 2009). To supply pure fluorine, small generators are being installed and managed by gas suppliers. These limit the amount of fluorine stored. A typical fluorine generator is an electrolytic cell which dialysis hydrogen fluoride to form fluorine and

(a) (b) (c)

The first step in safe handling of silane is to comply with relevant regulatory & industrial standards such as Compressed Gas Association (CGA), Semiconductor Equipment and Materials International (SEMI), National Fire Protection Association (NFPA), Factory Mutual (FM), European Industrial Gas Association (EIGA), Asian Industrial Gas Association (AIGA), Transportable Pressure Equipment Directive (TPED), Transportation of dangerous goods (ADR), SEVESO, ATEX, etc. The installation and operation practices are however not necessarily detailed in the above codes or standards. Rufin (2011) has presented a comprehensive list of best practices for silane supply system. The key features are

From the perspectives of preventing silane explosion, the most effective measure is preventing the accumulation of unignited silane through the use of forced ventilation or open air storage. It is also important to detect an unignited release by installing a gas sensor. Flame (UV/IR) detector is also needed for detecting leaks with autoignition. Sprinklers are needed for cooling the cylinders and system. Operators must be also be protected by

hydrogen (Stockman, 2009).

**3.1 Silane** 

summarized below.

**3. Best practice for handling** 

barriers and flame resistant protection suit. Other minor but equally important issues include: Welding for all indoor distribution systems

Automation and remote control

Joule-Thomson compensation for high flow >50 L/min

Prevention of "domino effect" by steel plate or firewall between sources

As fluorine reacts with almost everything, corrosion of materials in contact with fluorine is dependent upon the fluoride formed with the material. If the fluoride is volatile or in the form of loose flakes, new material will constantly be exposed to the fluorine and will be continuously eroded until it burns through the material. For this reason material selection is extremely important. Materials that form a good fluoride coating will be protected from further attack.

Systems for fluorine service must not only be fabricated of the proper material, the design of the system must consider the following factors:


#### **3.3 Other specialty gases**

Silane and fluorine represent the two extremes, the flammable and the oxidizer, of the specialty gases used in TFT-LCD manufacturing. All other gases bear similar but less extensive practices and protection measures as these two gases. For example, gas detectors

gas detector reading below the maximum scale, normally about 50 ppm, may be considered as a small leak. For a larger leak, activate sprinkler or apply water spray from a distance to the leak point to disperse the gas cloud and reduce the extent of potential

If the leak is burning, cool the adjacent area with a water spray. Do not spray water directly onto the fire as this could put it out and create an explosive environment. The heat could melt the fusible metal in the pressure relief device (PRD). The PRD will not activate unless the internal pressure reaches 27.6 MPa (4000 psig). Water spray from the deluge system onto the tubes will prevent this from ever reaching this temperature. A leak on fire can plug from the solids formed. Solids can also trap silane which can still be reactive. Even after a fire and the cylinder has vented to zero gauge pressure, there is still one atmosphere of gas in the cylinder. Emergency response teams typically will dilute the remaining silane by pressurizing the leaking cylinder with nitrogen to 0.79 MPa (100 psig) and venting 3 times. This will dilute the silane concentration to 2000 ppm. At this concentration and pressure, the

The authors are aware of six hydrogen fires with one of the fires resulting in rupture of the hydrogen tube. The fire resulted from the detachment of an aging copper tube to the cylinder pack in a hydrogen filling station. The hydrogen ignited during tube detachment from a filling manifold and resulted in jet fire which impinged on a nearby tube trailer. The tube trailer was parked without protection of sprinkler system and two tubes were ruptured in less than 5 minutes of flame impingement. The most important emergency response after the tube explosion was continuous cooling and venting of hydrogen in the

There have been reported incidents of nitrogen trifluoride with flammable oils used in vacuum pumps. Systems for nitrogen trifluoride must be designed similar to the requirements for fluorine in 3.2. Adiabatic compression and heating can initiate the decomposition reaction of nitrogen trifluoride into reactive fluorine. Incidents have occurred with improperly designed or cleaned systems during routine activities such as pressurizing the system. Internationally the maximum fill limits for nitrogen trifluoride cylinders is 0.5 kg/liter. At 21°C the cylinder pressure will be 10.78 MPa. Quickly opening the cylinder valve into a system with a small dead volume will cause adiabatic compression and heating to 307°C. In the presence of a reactive nonmetal component or a contaminant, this can initiate the decomposition reaction. Given sufficient energy and pressure, the reaction can cause the metal system to also become involved. Figure 5 shows a burnt valve in a nitrogen

Chlorine is very corrosive to carbon steel and stainless steel. With the moisture in the air it forms hypochlorous (HClO) and hydrochloric (HCl) acids which will corrode the steel as shown in Figure 6. A corroded connection generally leads to leak. A good design suply system should pick up the leak by detector and shutdown the system. If the system is not

explosion. Always prepare for possible explosion.

cylinder can be shipped back as a non-hazardous package.

**4.2 Hydrogen** 

remaining tubes.

**4.4 Chlorine** 

**4.3 Nitrogen trifluoride** 

trifluoride line caused by adiabatic compression.

are needed for all gases except for nitrogen trifluoride. Flame (UV/IR) detectors and sprinklers are needed for all flammable gases. High pressure test and He outboard leak test with emergency remote shut down are also typical practices for all gases.

## **4. Incidents and emergency response**

## **4.1 Silane**

There are numerous silane leaks in the forty years of silane use (Huang and Ngai, 2006) mostly resulted from poor procedures, practices, and equipment. The reactions from the leaks were reported to be a fire, minor "pops" to explosions. The "pops" can occur when a small amount of Silane trapped behind the valve cap or pigtail is released rapidly without prompt ignition. In a few cases they have been severe enough to cause eardrum rupture if the amount of accumulated silane is significant. Thus, the past silane leaks do support the fact that a leak with indefinitely delayed ignition is commonly observed in silane operations.

Although the release with indefinitely delayed ignition were mostly small and had finite impact, the unignited release may turn into large release with significant impact. The following major incidents were known where silane was released without immediate ignition and exploded after a delay:


Among these incidents, the incident in Taiwan in 2005 has been described in detail by Chen et al. (2006), Chang et al. (2007) and Peng et al. (2008). The cause of leak is a failed silane cylinder valve with loosened valve stem retainer. Silane leak from such a loosened retainer has been shown to have delayed ignition by Ngai et al. (2007). The operator in this case did not notice the unignited silane release from the loosened retainer and continued to turn the valve handwheel until eventually the retainer, valve stem, and valve diaphragm were all detached. The rapid silane release with ignition at end of release resulted in a fatal vapor explosion.

Emergency response to a silane incident is the most challenging one among all gas incidents (Ngai, 2011). The greatest danger is a leak and no fire. As ignition is inevitable at the end of leak, a pop or explosion should be expected. Use surveillance camera to make initial assessment if possible. Remotely review the gas detector response over time on a graph to insure that it is not sensor drift or electronic spike. Evacuation should be done first if there is a gas detector alarm but without visual confirmation of flame. The responders should wear full personal protection equipment, including fire proof suit, glove, hardhat, ear plug etc. and approach the leak with great care. The leak could ignite as the pressure decays in the source cylinder or an air flow disturbance is created. Assess the leak rate with a portable gas detector through the gas cabinet duct. For a small leak, it might be safe to remotely shutoff the supply but expect ignition and a pop at shutoff. Slow shutoff of the valve will minimize the amount of silane when it ignites. A leak with a

are needed for all gases except for nitrogen trifluoride. Flame (UV/IR) detectors and sprinklers are needed for all flammable gases. High pressure test and He outboard leak test

There are numerous silane leaks in the forty years of silane use (Huang and Ngai, 2006) mostly resulted from poor procedures, practices, and equipment. The reactions from the leaks were reported to be a fire, minor "pops" to explosions. The "pops" can occur when a small amount of Silane trapped behind the valve cap or pigtail is released rapidly without prompt ignition. In a few cases they have been severe enough to cause eardrum rupture if the amount of accumulated silane is significant. Thus, the past silane leaks do support the fact that a leak with indefinitely delayed ignition is commonly observed in silane

Although the release with indefinitely delayed ignition were mostly small and had finite impact, the unignited release may turn into large release with significant impact. The following major incidents were known where silane was released without immediate

Among these incidents, the incident in Taiwan in 2005 has been described in detail by Chen et al. (2006), Chang et al. (2007) and Peng et al. (2008). The cause of leak is a failed silane cylinder valve with loosened valve stem retainer. Silane leak from such a loosened retainer has been shown to have delayed ignition by Ngai et al. (2007). The operator in this case did not notice the unignited silane release from the loosened retainer and continued to turn the valve handwheel until eventually the retainer, valve stem, and valve diaphragm were all detached. The rapid silane release with ignition at end of release resulted in a fatal vapor

Emergency response to a silane incident is the most challenging one among all gas incidents (Ngai, 2011). The greatest danger is a leak and no fire. As ignition is inevitable at the end of leak, a pop or explosion should be expected. Use surveillance camera to make initial assessment if possible. Remotely review the gas detector response over time on a graph to insure that it is not sensor drift or electronic spike. Evacuation should be done first if there is a gas detector alarm but without visual confirmation of flame. The responders should wear full personal protection equipment, including fire proof suit, glove, hardhat, ear plug etc. and approach the leak with great care. The leak could ignite as the pressure decays in the source cylinder or an air flow disturbance is created. Assess the leak rate with a portable gas detector through the gas cabinet duct. For a small leak, it might be safe to remotely shutoff the supply but expect ignition and a pop at shutoff. Slow shutoff of the valve will minimize the amount of silane when it ignites. A leak with a

with emergency remote shut down are also typical practices for all gases.

**4. Incidents and emergency response** 

ignition and exploded after a delay: - 1976, Germany, 1 fatality - 1989, Japan, 1 fatality & 1 injury - 1990, Japan, 1 fatality & 3 injuries


**4.1 Silane** 

operations.

explosion.

gas detector reading below the maximum scale, normally about 50 ppm, may be considered as a small leak. For a larger leak, activate sprinkler or apply water spray from a distance to the leak point to disperse the gas cloud and reduce the extent of potential explosion. Always prepare for possible explosion.

If the leak is burning, cool the adjacent area with a water spray. Do not spray water directly onto the fire as this could put it out and create an explosive environment. The heat could melt the fusible metal in the pressure relief device (PRD). The PRD will not activate unless the internal pressure reaches 27.6 MPa (4000 psig). Water spray from the deluge system onto the tubes will prevent this from ever reaching this temperature. A leak on fire can plug from the solids formed. Solids can also trap silane which can still be reactive. Even after a fire and the cylinder has vented to zero gauge pressure, there is still one atmosphere of gas in the cylinder. Emergency response teams typically will dilute the remaining silane by pressurizing the leaking cylinder with nitrogen to 0.79 MPa (100 psig) and venting 3 times. This will dilute the silane concentration to 2000 ppm. At this concentration and pressure, the cylinder can be shipped back as a non-hazardous package.

#### **4.2 Hydrogen**

The authors are aware of six hydrogen fires with one of the fires resulting in rupture of the hydrogen tube. The fire resulted from the detachment of an aging copper tube to the cylinder pack in a hydrogen filling station. The hydrogen ignited during tube detachment from a filling manifold and resulted in jet fire which impinged on a nearby tube trailer. The tube trailer was parked without protection of sprinkler system and two tubes were ruptured in less than 5 minutes of flame impingement. The most important emergency response after the tube explosion was continuous cooling and venting of hydrogen in the remaining tubes.

#### **4.3 Nitrogen trifluoride**

There have been reported incidents of nitrogen trifluoride with flammable oils used in vacuum pumps. Systems for nitrogen trifluoride must be designed similar to the requirements for fluorine in 3.2. Adiabatic compression and heating can initiate the decomposition reaction of nitrogen trifluoride into reactive fluorine. Incidents have occurred with improperly designed or cleaned systems during routine activities such as pressurizing the system. Internationally the maximum fill limits for nitrogen trifluoride cylinders is 0.5 kg/liter. At 21°C the cylinder pressure will be 10.78 MPa. Quickly opening the cylinder valve into a system with a small dead volume will cause adiabatic compression and heating to 307°C. In the presence of a reactive nonmetal component or a contaminant, this can initiate the decomposition reaction. Given sufficient energy and pressure, the reaction can cause the metal system to also become involved. Figure 5 shows a burnt valve in a nitrogen trifluoride line caused by adiabatic compression.

#### **4.4 Chlorine**

Chlorine is very corrosive to carbon steel and stainless steel. With the moisture in the air it forms hypochlorous (HClO) and hydrochloric (HCl) acids which will corrode the steel as shown in Figure 6. A corroded connection generally leads to leak. A good design suply system should pick up the leak by detector and shutdown the system. If the system is not

response actions are also reviewed. It is sincerely hope that the current summary may bring an integrated view on the complicated and delicate specialty gas supply system, and

Air Force Rocket Propulsion Laboratory (1967). Halogen Passivation Procedural Guide,

Britton, L. G. (1990). Combustion hazards of silane and its chlorides, *Plant/Operations* 

Chang, Y. Y., Peng, D. J., Wu, H. C., Tsaur, C. C., Shen, C. C., Tsai, H. Y., & Chen, J. R. (2007).

Chen, J. R. , Tsai, H. Y., Chen, S. K., Pan, H. R., Hu, S. C., Shen, C. C., Kuan, C. M., Lee, Y. C.,

Dryer, F. L., Chaos, M., Zhao, Z., Stein, J. N., Alpert, J. Y. & Homer, C. J. (2007).

European Industrial Gases Association (2011). Code of Practice: Compressed Fluorine and

Huang, K. P. P. & Ngai, E. Y. (2006). Silane Safety and Silane Incidents, *Silane Safety and ER* 

Koda, S. (1992). Kinetic aspects of oxidation and combustion of silane and related

Liang, R. L., Wu, S. Y., Tseng, N. J., Ku, C. W., Tsai, H. Y. & Chen, J. R. (2010). Effects of

Ngai, E. Y. (2005). Release Testing of Reactive Fluoride Reactor Cleaning Gases, *Semicon* 

Ngai, E. Y. (2011). Silane Emergency Response, *Silane Safety Seminar - Safe Handling and Use* 

Ngai, E. Y., Huang, K. P. P., Chen, J. R., Shen, C. C., Tsai, H. Y., Chen, S. K., Hu, S. C., Yeh, P.

mixtures. *Combustion, Explosion and Shock Waves*, Vol. 5, No. 4, (September 1969),

Revisiting of a silane explosion in a photovoltaic fabrication plant, *Process Safety* 

& Wu, C. C. (2006). Analysis of a silane explosion in a photovoltaic fabrication plant, *Process Safety Progress*, Vol. 25, No. 3, (September 2006), pp. 237-244, ISSN

Spontaneous ignition of pressurized release of hydrogen and natural gas into air. *Combustion Science and Technology*, Vol. 179, No. 4, (April 2007), pp. 663–94, ISSN

Mixtures with Inert Gases, European Industrial Gas Association, Globally

compounds, *Progress in Energy Combustion Science*, Vol. 18, No. 6, (November 1992),

Temperature and Moisture on the Ignition Behavior of Silane Release into Air, *8th International Symposium on Hazards, Prevention, and Mitigation of Industrial Explosions*,

H., Liu, C. D., Chang, Y. Y., Peng, D. J., & Wu, H. C. (2007). Field tests of release, ignition and explosion from silane cylinder valves, *Process Safety Progress*, Vol. 26,

Air Products and Chemical Inc. Safetygram #25: 20% Fluorine in Nitrogen, June 2004. Baratov, A. N.; Vogman, L. P. & Petrova, L. D. (1969). Explosivity of monosilane-air

*Progress*, Vol. 9, No. 1, (January 1990), pp. 16-38, ISSN 0278- 4513.

*Progress*, Vol. 26, No. 2, (June 2007), pp. 155-157, ISSN 1066-8527.

Technical Report AFRPL-TR-67-309, December 1967.

preventing future incidents from occurring.

pp. 592–594, ISSN 0010-5082.

Harmonized Document, IGC 140/11, 2011

*Seminar*, Kaohsiung, Taiwan, May 17-19.

Yokohama, Japan, September 6-10 , 2010.

*China*, Shanghai, China, March 14 & 15, 2005.

*of Silane Gas*, Taichung City, Taiwan, January 20, 2011.

No. 4, (December 2007), pp. 265-282, ISSN 1066-8527.

pp. 513-528, ISSN 0360-1285.

1066-8527.

0010-2202.

**6. References** 

Fig. 5. A burnt valve in a nitrogen trifluoride line caused by adiabatic compression.

Fig. 6. Green hypochlorous and hydrochloric acids from chlorine corrosion.

designed properly, the chlorine may continuous to leak and affect the cleanroom. The author is aware of a chlorine leak from a tube connection in an IC fab that was not detected until one hour later. The chlorine cloud speaded and significantly affected the cleanroom and production tools.

#### **4.5 Other specialty gases**

The authors are not aware of any phosphine, ammonia, or fluorine fatalities or fires related to semiconductor or TFT-LCD uses.

## **5. Conclusions**

The TFT-LCD manufacturing processes utilize significant amount of gases ranging from highly flammable to strong oxidizer gases. Special cares must be taken in handling these specialty gases to prevent fire, explosion and toxic release.

This paper has reviewed the hazardous properties and potential incidents of these specialty gases. The best practices for handling these gases are then described. The emergency response actions are also reviewed. It is sincerely hope that the current summary may bring an integrated view on the complicated and delicate specialty gas supply system, and preventing future incidents from occurring.

#### **6. References**

176 Features of Liquid Crystal Display Materials and Processes

Fig. 5. A burnt valve in a nitrogen trifluoride line caused by adiabatic compression.

Fig. 6. Green hypochlorous and hydrochloric acids from chlorine corrosion.

and production tools.

**5. Conclusions** 

**4.5 Other specialty gases** 

to semiconductor or TFT-LCD uses.

specialty gases to prevent fire, explosion and toxic release.

designed properly, the chlorine may continuous to leak and affect the cleanroom. The author is aware of a chlorine leak from a tube connection in an IC fab that was not detected until one hour later. The chlorine cloud speaded and significantly affected the cleanroom

The authors are not aware of any phosphine, ammonia, or fluorine fatalities or fires related

The TFT-LCD manufacturing processes utilize significant amount of gases ranging from highly flammable to strong oxidizer gases. Special cares must be taken in handling these

This paper has reviewed the hazardous properties and potential incidents of these specialty gases. The best practices for handling these gases are then described. The emergency


**9** 

Youn Jin Kim

*Korea* 

*Samsung Electronics Company, Ltd.* 

**Portable LCD Image Quality:** 

**Effects of Surround Luminance** 

Liquid crystal display (LCD) has become a major technology in a variety of display application markets from small sized portable displays to large sized televisions. Portable LCD devices such as smart phones and mobile phones are used in a diverse range of viewing conditions. We usually experience images on a mobile phone with a huge loss in contrast under bright outdoor viewing conditions; thus, viewing condition parameters such as surround effects, correlated colour temperature and ambient lighting have become of significant importance. (Katoh et al., 1998; Moroney et al., 2002) Recently, auxiliary attributes determining the mobile imaging were examined and the surround luminance and ambient illumination effects were considered as the first major factor. (Li et al., 2008) Surround and ambient lighting effects on colour appearance modelling have been extensively studied to understand the nature of colour perception under various ambient illumination levels (Liu & Fairchild, 2004, 2007; Choi et al., 2007; Park et al., 2007); thus, this study intends to figure out characteristics of the human visual system (HVS) in *spatial frequency domain* by means of analysing the contrast discrimination ability of HVS. In consequence, we propose an image quality evaluation method and a robust image enhancement filter based on the measured contrast sensitivity data of human observers

The former is to quantify the observed trend between surround luminance and contrast sensitivity and to propose an image quality evaluation method that is adaptive to both surround luminance and spatial frequency of a given stimulus. The non-linear behaviour of the HVS was taken into account by using contrast sensitivity function (CSF). This model can be defined as the square root integration of multiplication between display modulation transfer function (MTF) and CSF. It is assumed that image quality can be determined by considering the MTF of an imaging system and the CSF of human observers. The CSF term in the original SQRI model (Barten, 1990) is replaced by the surround adaptive CSF quantified in this study and it is divided by the Fourier transform

The latter is a robust image enhancement filter which compensates for the effects of surround luminance on our contrast perceiving mechanism. Precisely, the surround luminance adaptive CSF is used as a guide for determination of the adaptive enhancement

**1. Introduction** 

under various surround luminance levels.

of a given stimulus.

gain in the proposed algorithm.

