Highly Engineered Glass Substrates for LCD Television: Why Reducing Value Is Incompatible with Consumer Expectations
With the explosion of LCD TV in the past few years and price pressures becoming larger factors, it stands to reason that some manufacturers may look for cheaper alternatives when it comes to the glass used in the making of LCDs. Does highly engineered, specialized glass still offer the best value for manufacturers?
by Peter L. Bocko and H. S. Lee
AS liquid-crystal-display (LCD) technology progresses to its full potential in the television market – with 50% penetration expected in 2008 – manufacturers will continue to experience pressure to reduce costs of the bill-of-materials: the backlight assembly, liquid-crystal materials, optical films, and, of course, the glass substrates. With this in mind, it is appropriate to re-examine the basic value propositions for these strategic materials. The substrate industry has been quite successful at providing highly engineered, high-value glass for panel manufacturers. But does this package of specialized properties and value still hold up?
The question is a timely one, given a late 2007 report that a commodity-type glass had been used in the fabrication of an LCD monitor. The industry response to this demonstration was intense and perhaps overwrought, but the question is still valid: Is it in the best interests of the industry to continue on the path of using highly engineered, high-value specialized glass compositions for LCDs?
Corning entered the LCD market when it supplied glass substrates using the fusion-forming process. In comparison to the float-glass method, in which molten glass is pulled onto a molten tin bath and allowed to solidify, using fusion-forming, molten glass is poured into a trough, or "isopipe." The glass fills the isopipe, flows evenly over both sides, and fuses at the bottom. It is then drawn down to form a continuous sheet of glass, which, because it is formed in air, is flat, pristine, and smooth on both sides. Moreover, the resulting substrate is free of any defects or scratches produced by grinding or polishing (in contrast, other techniques currently require additional steps in order to remove surface artifacts). Fusion-forming by its nature also engenders potential for tight control of the dimensional properties of large glass sheets.
The first glass employed in early trials of fusion-forming for LCD applications was Pyrex®, a composition containing alkali in the form of sodium. As shown in Table 1, Glass Property Comparison, the unique feature of Pyrex was its low thermal expansion – the relative increase in size caused by an increase in temperature – compared to a typical traditional soda-lime glass composition. Even at the onset of AMLCD technologies, the low coefficient of thermal expansion (CTE) was essential to reproducible mask positioning at the elevated TFT processing temperatures.
Although glassmakers recognized that sodium could degrade the electronic performance of thin devices, experiments seemed to indicate that perhaps a low level of alkali could be tolerated in LCDs, provided the glass had sufficient chemical durability.1 However, when it came to active-matrix technology, even a small amount of sodium was considered problematic. As glass compositions moved forward to match the ever-increasing demands on glass attributes, sodium as a mobile ion was reduced to trace levels to prevent numerous performance and reliability issues that could arise with ionic contamination of the liquid-crystal layer.
The first glass used on a commercial scale in AMLCD development was Corning 7059. Introduced in the early 1990s, 7059 had a simple composition with advantages in high-temperature durability and relatively low CTE. Nevertheless, this composition had its own limitations, which were exacerbated as the display industry sought to increase yield and make the LCD platform practical for high-volume manufacture. In those early days, manufacturers were using a harsh wet chemistry during substrate cleaning and in photolithography, and the simple barium boroaluminosilicate composition of Corning 7059 lacked the chemical durability required for this aggressive chemistry. Also, although the CTE had been lowered to 46 x 10-7/°C for high-throughput thermal processes and driver chip integration, concerns about expansion would only worsen as Gen sizes increased. Therefore, higher thermal and dimensional stability were critical.
The next step was Corning 1737, introduced in the latter half of 1996. Although developed in collaboration with a customer, this glass featured an attribute display manufacturers had not specifically requested: low density. Achieved by engineering the composition to lower levels of heavy, dense constituents, the low density of Corning 1737 reduced substrate glass weight in notebook computers (then the primary application), and also greatly facilitated larger substrate sizes, as the reduced weight eased automated handling. But an unanticipated benefit of the glass pulled it into the marketplace ahead of schedule. Customers were depositing highly stressed refractory metal films to reduce the electrical resistance of gate lines for larger notebook panels, and the intrinsic glass lattice strength of Corning 1737 created a more reliable surface for these films. This fortuitous circumstance set a recurring pattern in the LCD industry: highly optimized and perhaps (to some) over-engineered substrate compositions brought unanticipated benefits to panel manufacturers as they evolved their processes to achieve better display performance and greater manufacturing process throughput.
In the late 1990s, Corning began a structured collaboration with key customers to develop EAGLE2000®. The design team engaged with customer counterparts to create a joint vision of future display products and targets for LCD-manufacturing platform characteristics. The development targets for EAGLE2000® were driven by a continuing focus on key glass properties: density was reduced to 2.4 g/cm3 and thermal expansion to 32 x 10-7/°C. Of course, no one yet envisioned the three-meter-square platform of Gen 10, but customers did articulate their need for very large substrates capable of high-throughput processing. By design, EAGLE2000® had enough headroom in its basic specifications to meet unforeseen challenges, even though customers eventually scaled the platform well beyond their predictions.
That brings us to today's state of the art in glass substrates for LCDs: highly engineered, high-value specialized glass compositions, such as EAGLE XG™, an extension of the EAGLE family into an environmentally friendly glass substrate that contains no added heavy metals or halides.
Design Criteria for Glass Substrates
The basic design criteria for earlier display glasses have carried over to today's products: low density and low thermal expansion in a silica-rich composition.
• Low density offered low device weight for portable applications as well as more facile robotic handling during high-speed automation. Substrate gravitational sag was a special challenge in large sizes because equipment manufacturers had not yet learned how to support the substrate from the back side of the glass without inducing damage.
• Low CTE brought several benefits, including minimizing distortion during transient thermal steps and assuring the extremely tight tolerance requirements of high-aperture-ratio displays. Previously, the only way many notebook manufacturers could achieve an aperture ratio above 80% in commercial production was to employ a glass with the expansion coefficient of Corning 1737.
• Lastly, the trend has been toward increasingly silica-rich compositions with high-temperature attributes and mechanical reliability throughout all the stresses of display manufacture and device lifetime. High silica compositions can withstand extremes of chemical processing during panel manufacturing, as well as mechanical pressures generated by customers.
While design criteria have remained largely the same, much has changed in the LCD process. The mechanical engineering design for glass substrate handling has evolved significantly. In array technology, the use of compliant low-resistance metal films based on aluminum and its alloys has been mastered, reducing stress at the glass/gate metal interface. Advances such as color-filter-on-array could reduce alignment issues associated with the manufacture of high-aperture-ratio, high-resolution displays. Current substrate cleaning and etching processes contrast with the brute force of the typical Gen 1 and 2 lines. It is logical to wonder whether these advances have relaxed some of the stringent design criteria of current-generation AMLCD substrates.
The short answer is no. Even with these improvements, the simplistic value proposition deployed in the mid-90s for Gen 3 still holds up. For most customers, removing value from the substrate to achieve a lower bill-of-materials is an unattractive option, given the difficulties inherent in compensating for lower substrate performance in their process. In addition, LCD trends have further reinforced the value proposition of highly engineered, high-value specialized glass through less obvious yet equally vital process and application challenges.
4 to 12
Less than 0.05
Impact of Commodity Glass Attributes on AMLCDs
To understand the implications of using commodity-type substrates in a contemporary AMLCD manufacturing process, including required end-device performance, we will contrast the behavior of today's state-of-the-art substrate with commodity-type glass, focusing on a few increasingly important, though subtle, differences.
Figure 1 provides a concise summary of the key attribute issues associated with adopting the current AMLCD process and device structure from the current state-of-the-art substrate, such as Corning's EAGLE XG, to a hypothetical commodity glass. We emphasize "hypothetical" because an infrastructure to supply a commodity-type glass of the sizes, thinness, and extrinsic quality currently required for meeting even the most basic requirements of a modern LCD fab simply does not exist. In the past 20 years, the divergence of the AMLCD substrate platform's technology curve from that of conventional commodity glass manufacturing has produced a significant gap between their respective capabilities for dimensional attributes (large, thin, and flat) and intrinsic quality (surface and bulk).
Furthermore, in virtually all applications, commodity-type glass contains significant levels of alkali; therefore, glass properties typical of the soda-lime type have been applied in this analysis. Alkali, predominantly in the form of sodium and potassium oxides, has long been used to facilitate glass melting and remains an essential constituent in substantially all commodity glass manufacture. However, while high alkali content facilitates glass melting in a low-cost all-refractory brick melting apparatus at a high pull rate, it also impacts glass physical and chemical properties in a way that reduces thermal stability, increases the CTE, degrades mechanical properties, and increases ionic contamination risk. Historically, the presence of small, highly mobile ions such as sodium at the parts-per-billion level have led to degradation in performance. Key variables impacting this performance risk are the quality, composition, and location of barrier layers as well as device lifetime.
In typical soda-lime glass compositions, sodium is frequently batched as a glass component at between 4 and 12 wt.% of the glass. In sharp contrast, with highly engineered, alkali-free substrates, the sodium content is held to trace levels that are many hundreds of times lower than those in float-glass compositions – a 2 to 3 order-of-magnitude difference.
Fig. 1: Summary of some of the key attribute issues associated with the use of commodity-type glass compared to EAGLE XG™ in a modern LCD panel.
The most obvious impact on the glass is its reduction of thermal stability. Reducing film deposition temperatures in the thin-film-transistor (TFT) processes has been an active area of research, as many labs have attempted to fabricate TFT arrays on exotic substrates. While amorphous-silicon (a-Si) films have been deposited at temperatures below 200°C, there is a complex trade-off between the modified processes' deleterious effects upon the resultant film's electronic properties1 and the benefits of reduced temperature. Lower process temperatures can reduce but not remove the potential for increased substrate shape changes associated with commodity glass. TFT performance is typically negatively impacted by increased electronic trap densities, unless extensive process modifications are made, frequently in the form of extra process steps or increased process times. The practicality of such low-temperature approaches has been the subject of vigorous debate.
An increased thermal-expansion coefficient can exacerbate warp and in-plane dimensional distortion to the substrate when the substrate is exposed to a thermal gradient during the manufacturing process. These distortions can cause feature misalignment (in the TFT plane or between the TFT and color-filter plates) or handling problems. Thermal expansion for typical soda-lime glass is 90 x 10-7/°C – nearly three times that of highly engineered, alkali-free glass, such as EAGLE XG, at 31 x 10-7/°C.
Recently, optical retardation changes, which can be induced from panel design, manufacturing, and operation, have become a matter of concern as expectations for panel performance continue to advance. Figure 2 illustrates the results of a finite-element model: a numerical simulation of the impact of optical retardation on a panel's viewing performance. The conditions illustrated are for matched glass composition panels of EAGLE XG and typical soda-lime glass, with the two glass substrates in each panel assumed fixed at their edges and held stationary in a rigid frame. The TFT plate, closer to the backlight, experiences a higher temperature than the front plate, with a stress-producing temperature gradient of 10°C, back to front. Stress birefringence caused by the relative thermal expansions of the two compositions, along with their respective stress optic coefficients – a measure, higher for typical soda-lime glasses, of how much birefringence is caused in a material from a stress – produces a 5x greater retardation in the commodity-glass-based panel.
Brighter areas on the display screen indicate increased light leakage through the panel's polarizing films, causing unacceptable contrast irregularities on the LCD screen, which are more apparent with large-sized panels and worsen with more high-expansion glass content in the display. Some may think substituting a typical soda-lime color-filter plate in an LCD is straightforward compared to attempting a typical soda-lime TFT plate, but given this effect, this is far from the case. When a combination of a low-expansion backplane is paired with a high-expansion color filter, the backplane serves as an unyielding frame for the high-expansion color-filter glass, and the retardation persists.
Fig. 2: FEA numerical simulation of the optical retardation in 40-in.-diagonal panels. The simulated soda-lime glass display panel on the left shows significant retardation in the center of the panel – 3x the EAGLE panel on the right and 5x the retardation at the panel edges – thereby producing a panel with poor contrast.
Scratches and flaws on a glass surface dictate the breaking strength of the glass article under stress. A non-polished glass, such as EAGLE XG, has an advantage in panel strength due to fewer microscopic flaws on the fusion surface compared to polished glass. Commodity-type flat glass will need extensive grinding and polishing to reach the desired flatness, which for the LCD industry is measured in surface peak-to-valley variations on a nanometer scale. These exacting tolerances are necessary to fabricate displays of high contrast because variations in flatness cause differences in cell-gap thickness, which induce changes in the amount and path of light passing through the liquid-crystal cell. For commodity-type glass, the required grinding and polishing introduces microscopic flaws and increases manufacturing costs. Moreover, fatigue results when stress is applied to a flaw in the presence of water vapor, and a seemingly minor flaw can enlarge to become critical. Even among non-alkali AMLCD substrates, there are substantial differences in the dynamic fatigue properties.2 The fatigue constant of typical commodity soda-lime glass is such that the fatigue resistance of EAGLE XG is about a third higher. Long-term degradation in strength can occur if low-fatigue-constant substrates are employed in the panel. In addition, panel separation processes, always a source of yield loss, are more reliable with dynamic fatigue-resistant substrates such as EAGLE XG.
Impact of Ionic Contamination
If mobile ions, such as sodium, migrate into the liquid-crystal material, the electrical instability of the liquid-crystal layer will increase. Ions dissolved in the LC from peripheral materials have a direct and immediate impact on image quality through the development of a pathway for a leakage voltage.3 Thus, liquid-crystal suppliers have focused on the reduction of ionic contamination in the LC material. While the application of a barrier film on a commodity-type substrate may help minimize the risk of ionic contamination, this involves additional cost and is never foolproof.
Recently, image sticking on LCDs has been found to result from spatial differences of the photo leakage current in a-Si TFT-LCDs as well as from the parasitic capacitance induced by ionic impurities.4 Eliminating image sticking requires an optimized TFT device design and the removal of ionic contamination. A sodium-bearing commodity-type glass would be a move in the wrong direction.
Highly optimized specialty glass compositions have evolved to meet the needs of the LCD industry. Collaborations between glass suppliers and panel manufacturers have resulted in glass designs that brought substantial benefits to the LCD platform and contributed to LCD's leadership in the TV application. We have discussed some differences between a highly engineered, high-value specialized glass composition and a hypothetical commodity-type glass substrate in LCD process performance, display optical performance, and panel reliability. The risks these comprise far outweigh the potential benefit of a reduced bill-of-materials.
By "designing in" the glass benefits rather than "designing around" glass limitations, highly engineered glass substrates are actually the lower-cost solution. Reducing substrate value seems incompatible with consumer expectations for an excellent viewing experience and high panel reliability. With more complex content emerging and increasingly challenging requirements in almost every dimension of display performance, the highly competitive environment of information display will necessitate staying on the path of highly engineered, high-value specialized glass compositions for LCDs for some time to come.
1G. B. Raupp, et al., "Low-temperature amorphous-silicon backplane technology development for flexible displays in a manufacturing pilot line-environment," J. Soc. Info. Display 15(7), 445-454 (2007).
2S. T. Gulati. and J. D. Helfinstine, "Fatigue resistance and design strength of advanced AMLCD glass substrates," Display Manufacturing Technology Conference Digest of Technical Papers, 29-30 (1997).
3S. Shohei, "Liquid-crystal-material technologies for advanced display applications," J. Soc. Info. Display 8(1), 5-9 (2000).
4J. W. Lee, H. S. Hong, K. S. Cha, J. Y. Lee, B. W. Lee, and J. Yi, "A novel structure and process for improving image sticking," Proc EuroDisplay, 292-294 (2007). •