Material support: helping displays deliver higher performance
ANDREAS WEISHEIT Linde, Shanghai, China, and GREG SHUTTLEWORTH, Linde, Guildford, UK
Metal oxide transistors increase electron mobility by a factor of up to 40 compared to conventional technology, at a comparable cost base.
With the rapid development of advanced display technologies, such as OLED, high definition 3D and smart displays, continuing to deliver improved performance is challenging traditional display manufacturing techniques. New materials and manufacturing approaches will be required to deliver display products with the features and functionality demanded by customers but at market friendly prices and with a reduced environmental impact.
Conventional TFT-LCD display manufacturing is a sequence of complex processes, which can be classified in 4 categories: first forming of an array of thin-film transistors (TFTs) on the mother glass, second forming of a color filter, followed by the cell process where substrates are filled with liquid crystals, and finally the back-end assembly of the module. From a semiconductor and electronics materials perspective, we will focus on the thin-film transistor process (Fig. 1).
Figure 1. TFT-LCD display manufacture is a sequence of complex processes, which can be classified in the four categories shown.
Transistors control the pixels and switch them on and off. The type of transistors and the processes used for their manufacturing determine the performance of the display, the costs, and, to a large extent, the environmental footprint of the device. Today, two transistor types are common in mass production; amorphous silicon (a-Si) transistors are dominant (>95%), while low temperature polysilicon transistors (LTPS) have a niche position.
The manufacture of Ultra High Definition (UHD) 3D displays (3840 x 2160 pixels) is currently not possible using conventional amorphous silicon transistors, because when these are scaled down in size (in order to keep the pixel size small) they don???t conduct electrons quickly enough to maintain the fast frame rates required. Improving electron mobility in transistors is crucial in bringing higher resolutions and higher frame rates to TVs, mobile and computing devices (Fig. 2). Polysilicon transistors could deliver the required performance but would cost up to twice as much as the manufacturing process becomes more complex. A more promising approach is to use metal oxide transistors. This will increase electron mobility by a factor of up to 40 compared to conventional technology, at a comparable cost base as the process flow remains similar.
Figure 2. Summary of electron mobility required for future displays. 
Enabling UHD 3D is only one of the benefits of smaller transistors. They also allow more light to pass from the backlight through the backplane, reduce power consumption and increase battery life of mobile devices. Moreover, they also allow the higher currents needed to drive OLEDs. Consequently, the majority of leading display manufacturers have recently announced plans to launch products with metal oxide transistors. Samsung introduced a 55-inch OLED TV at the 2012 Samsung Premium TV Showcase in May of this year, and LG Display unveiled plans to launch a 55-inch OLED TV by the middle of 2012. Additionally, Sharp has announced that the company has started the commercial production of metal oxide displays in its Kameyama plant.
Metal oxide transistors
The channel of metal oxide transistors is mostly made from IGZO (indium, gallium and zinc oxides). The gate dielectric isolates the gate from source and drain. The passivation layer seals and protects the transistor (Fig. 3).
Figure 3. Cross-sectional of view of a metal oxide thin film transistor used in displays.
The formation of gate, dielectric, source drain and passivation layers is carried out through a sequence of deposition and etching steps to create the required patterns. This is where specialist gases and materials play a key role. The metallic layers (gate, channel and pixel layers) are typically deposited using physical vapour deposition (PVD). Here, argon atoms strike a solid target and sputter the material on to the substrate. Plasma enhanced chemical vapour deposition (PECVD) is used for the deposition of dielectric and passivation films. PECVD uses large amounts of silane (SiH4) as a silicon precursor and ammonia (NH3) as reactant to form silicon nitride (Si3N4). Hydrogen and nitrogen are used as carrier gases, while phosphine (PH3) is used for doping films. Etching of films is typically carried out using fluorinated gases which react with the films to create volatile compounds which can then be pumped away.
Since metal oxide channels are used to enable very high performance, changes in the dielectric and the passivation will be required as well. Both functional layers will change to silicon oxide (SiO2) which offers better stability and moisture protection than silicon nitride. To enable this change at production volumes, the PECVD process will require high volumes of high-purity nitrous oxide (N2O) as reactant rather than ammonia (NH3). While we are familiar with the use of this gas in dentistry, the volumes and purity required for displays present new challenges for the materials industry. The shift from silicon to metal oxide transistors would not be possible without a secure supply of high-purity N2O, which highlights the critical role that specialty gases and materials play in enabling the development of next-generation consumer electronic devices.
Deposition processes also deposits quantities of material on the inside of the process chamber, which must be periodically cleaned to maintain process cleanliness and production efficiency. Both these cleaning processes and etching steps described earlier use fluorine based gases, and given that process chambers need to accommodate glass substrates with an area of more than five square metres, the amount of gas required is very large.
Why is this important? The gases traditionally used have very high global warming potential (GWP). Sulphur hexafluoride (SF6) used for etching, has a GWP 23,900 times that of CO2, while nitrogen trifluoride (NF3) used for cleaning, has a GWP of 17,200. So with a typical Gen 8 facility using upwards of 300 tons of such gases per year, the potential environmental impact must be considered.
The case for F2
To minimize process emissions, most manufacturers have installed high performance scrubbing systems. However, there exists the risk of emissions during the whole life cycle of the material. Measurements by the Scripps Institute  have shown a rapid growth in the amount of NF3 present in the atmosphere, which correlates to a figure as high as 16% of NF3 produced ultimately escaping. There remains an incentive to consider materials with lower GWP to minimise the impact of emissions due to the manufacturing, transport and disposal of the materials, areas outside the control of TFT-LCD manufacturers. One strong contender is fluorine gas (F2).
The science goes like this. NF3 or SF6 gas is activated by a plasma to release fluorine radicals which then etch or clean the silicon films. The alternative, F2 gas, is the simplest molecule containing fluorine atoms and has the lowest bond energy. These properties of F2 provide significant benefits to the cleaning and etching processes; faster etching or cleaning, less tool down-time, reduced gas use, and less electrical power consumed by the plasma to break down the simpler F2 molecules.
Chamber cleaning using F2 rather than NF3 has been used in TFT-LCD production for some years now, delivering reductions in mass of gas required of 20% along with improved tool productivity due to a 30% reduction in cleaning time and almost 50% reduction in plasma power consumption. More recently, fluorine was evaluated as a direct replacement for SF6 for the etching process. Evaluation results on a Gen 4.5 size panel showed significant improvements in etch rate and etch uniformity when using fluorine for both silicon oxide and nitride films while maintaining the same feature taper angle.
Table 1: Potential CO2 equivalent savings from replacing NF3 with F2
Environmentally speaking, the big win is the zero GWP of F2. To highlight the impact, Table 1 shows the potential CO2 equivalent savings from replacing NF3 with F2 in a typical large scale TFT-LCD fab chamber-cleaning process.
It is thus likely that the replacement of NF3 and SF6 in the display manufacturing process can have a larger overall environmental benefit for TFT-LCD manufacturers than is achievable by any other means.
While you enjoy the stunning resolution and performance of the latest displays and reflect on the reductions in power consumption and increases in battery life, the materials industry will have played a key supporting role, from developing new materials for device research and development to delivering the materials at the quantities and purities required for pilot and mass production and for continuous improvement of the more established processes for better efficiency, lower costs and reduced environmental impact.
1. T. Kamiya et al., ???Present status of amorphous In-Ga-Zn-O thin film transistors,??? Science and Technology of Advanced Materials, 11 (2010)
2. R. F. Weiss et al., ???Nitrogen trifluoride in the global atmosphere,??? Geophysical Research Letters, vol. 35, October 2008
ANDREAS WEISHEIT is the Head of Global FPD and Asia Market Development at Linde and GREG SHUTTLEWORTH is an equipment product manager at Linde.
Solid State Technology, Volume 55, Issue 7, September 2012