LCoS manufacturing incorporates both traditional and novel packaging methods
BY JULIA GOLDSTEIN, DEAN BUI AND RONG HSU
For those accustomed to packaging chips for use on printed circuit boards (PCBs), microdisplay technology is something new. Liquid crystal on silicon (LCoS) microdisplay imagers represent one of the newest technologies in the field of liquid crystal displays, and the manufacturing process combines both familiar and novel packaging concepts. Manufacturing of an LCoS device begins with standard semiconductor processing, but the path to a finished imager requires innovative thinking from the packaging industry. The key difference is that the liquid crystal material cannot withstand the elevated temperatures common to semiconductor packaging. Standard processing temperatures used during die attach and encapsulation would ruin the device, so alternatives are needed.
Interconnection is made not to a ceramic or plastic package, but usually to a flexible printed circuit (FPC), requiring other process changes.
What is a Microdisplay?
Microdisplays are a fairly recent outgrowth of the flat panel and liquid crystal display (LCD) industries. As their name suggests, microdisplays are small LCDs, typically with a diagonal measurement of one inch or less. Microdisplays can be either transmissive or reflective (Figure 1). In transmissive microdisplays, a layer of liquid crystal is sandwiched between two glass plates and incident light is transmitted through the display. Thin film transistors (TFTs) are built into one of the glass plates, similar to the method used for large flat panel displays. There are two categories of reflective microdisplays: microelectromechanical systems (MEMS) and LCoS.
 Figure 1. The path of incident light for transmissive and reflective front projection systems. |
LCoS microdisplays consist of a liquid crystal sandwiched between a silicon chip and a glass plate. The silicon chip is manufactured using standard complementary metal oxide semiconductor (CMOS) technology, which provides a much higher degree of circuit integration than TFT microdisplays. Incident light passes through the glass and liquid crystal, and is then reflected off an array of micro-mirrors that are fabricated in the top metal layer of the silicon. The reflected image is then processed through an optical system and projected.
LCoS microdisplays are packaged and combined with optics to produce real or virtual images for one of two categories of applications. The first type is for use in projectors that display images from a computer, magnified onto a large screen. The second type is for handheld devices, such as pagers and cellular phones, to ultimately display full-color Web pages in a readable format. These handheld devices project a virtual image that is up to five times the size of the microdisplay itself. Initial LCoS prototypes were first reported in the literature more than a decade ago, but as recently as 1996 the systems did not produce sufficient resolution and contrast for today's applications.1-3 It has only been in the past few years that improvements in resolution, brightness, contrast and manufacturability have allowed LCoS microdisplays to become commercially viable.4
LCoS Manufacturing Process
The core of an LCoS microdisplay is the cell, which consists of the silicon chip and glass plate bonded together with a gasket and filled with liquid crystal material. The cell is then packaged to produce an LCoS imager (Figure 2). The processes involved in manufacturing LCoS cells are beyond the scope of this article. Instead, the focus here will be on the assembly process required to make LCoS imagers from the cells. One important feature to mention about the cell is that the glass plate is coated on the inner side (the side that comes in contact with the liquid crystal) with a layer of indium tin oxide (ITO) that functions as an electrode. The LCoS cell is manufactured with an offset between the silicon and the glass (Figure 3). This provides exposed ITO-coated glass on one edge for electrical connection of the electrode to the FPC. On the other side, the offset exposes bond pads on the silicon for wire bonding to the FPC.
 Figure 2. Drawing of an LCoS imager. |
The construction of an LCoS imager is illustrated in Figure 4. An FPC is used as the basic package for the cell and to make electrical connection to the optical system inside which the cell is to function. The FPC itself is bonded to a metal substrate, and discrete components and a surface mount connector are attached to make the FPC assembly. The LCoS assembly process begins by attaching a plastic or metal border on top of the FPC assembly to protect the cell from mechanical stress during handling. Next, the cell is attached and the ITO is electrically connected to the FPC assembly using a conductive adhesive. Wire bonding is then performed to make electrical connection from the bond pads on the chip to the bond fingers on the FPC. Finally, encapsulation of the exposed wire bonds is performed to complete the imager-assembly process. The photograph in Figure 5 shows an example of a completed LCoS microdisplay imager.
Materials Selection
For the most part, materials selection for packaging an LCoS cell is driven by performance requirements of the liquid crystal material inside the cell. The liquid crystal must be kept below its nematic-to-isotropic phase transition temperature. Above the transition temperature, the material is a liquid. In the nematic phase, the material flows like a liquid but the molecules are aligned, similar to the type of long-range order observed in solid crystalline materials. It is this alignment that allows the LCD to function. Liquid crystal materials chosen for LCDs are in their nematic phase at room temperature. Their performance varies with temperature, but ranges of operation from -40 to +85°C have been demonstrated.5
 Figure 3 (left). Drawing of an LCoS cell, showing top, side and edge views. |
Performance of an LCD is also highly dependent on the gap or spacing between the silicon and the glass within the LCoS cell. Variability of this spacing throughout the display area causes variability in the response time of the liquid crystals with respect to the applied voltages, the direct results of which are poor contrast and ghost images in the displayed picture. Stresses, introduced mainly by mismatches in coefficients of thermal expansion (CTE) of the packaging materials and by normal handling of the imagers, must be minimized to keep the cell gap carefully controlled and uniform throughout the cell area. Material selection must therefore be made with the minimization of both these sources of stress in mind. For example, an FPC that is too rigid can introduce stresses into the LCoS cell when the tail of the FPC is flexed.
Adhesives are used during many steps of the assembly process, namely FPC assembly, border attach, cell attach, ITO connection, and encapsulation. The limitations imposed by the liquid crystal necessitate the use of low-temperature, low-stress adhesives. Adhesives used for die attach and encapsulation are typically cured at temperatures of 125°C or higher, much too high for the liquid crystal. Similarly, solders cannot be used, because even low-melting specialty solders have melting points well above 100°C. Thermally or electrically conductive adhesives that can be cured at room temperature are one choice. Silicone materials are also a good option because they cure at room temperature in the presence of moisture, and remain resilient after curing. Another option is UV-curable adhesives, but one concern with these is degradation of the liquid crystal material if exposed to an ultraviolet light source. For steps during which an electrically conductive adhesive is required, room temperature curable, silver-filled epoxies are available.
Processing Considerations
Foundries that assemble LCoS imagers are usually accustomed to working with plastic and ceramic packages. Examining the assembly of an LCoS imager step by step, there are various issues that come up that are not encountered in traditional semiconductor packaging. These will be addressed one by one.
 Figure 4 (right). Typical construction of an LCoS microdisplay imager. |
As mentioned previously, a protective border can be attached to the FPC before or after the cell is attached. There are advantages and disadvantages to each approach. If the border is attached first, a wider variety of adhesives are available, because the temperature is limited only by the requirements of the FPC. In this case, curing at 150°C is perfectly acceptable. This saves time, because elevated temperature curing is typically complete within two hours, whereas room temperature-curable adhesives can require at least 24 hours of curing time. Because the cell is the most delicate and expensive part of the assembly, it may make sense to attach it as late in the process as possible. However, cell attach and ITO connection are more difficult with the border in place. There are clearly tradeoffs between handling issues, manufacturability and processing speed.
The next step of assembly is ITO connection. There is a vertical gap of at least 500 µm between the underside of the glass and the FPC that can be spanned with electrically conductive adhesive, but the mechanical integrity may not be sufficient. Also, the silver in the adhesive tends to segregate at the bottom, decreasing the electrical conductivity of the connection. One solution is to first attach a metal spacer to the ITO before assembly. In this case, the spacer is attached with conductive adhesive and cured at the beginning of the assembly process. Electrical connection to fill the much smaller gap is later completed with additional conductive adhesive after the cell is attached to the FPC assembly. This process provides a proven reliable connection.
Wire bonding from a silicon chip to an FPC can present a challenge. Ultrasonic aluminum wedge bonding is commonly used, which avoids the elevated temperatures required for thermocompression or thermosonic bonding. Still, standard wire bonding process parameters, such as those that work well for bonding to a ceramic package, tend to cause unreliable bonds. The choice of bonding wedge can be crucial in obtaining a reliable bond. Parameters, such as ultrasonic power, wedge force and bond time, must be also carefully adjusted to optimize bond pull strength and reliability.
 Figure 5. Examples of LCoS imagers. |
Encapsulation of the wire bonds is usually the final process in LCoS assembly. The silicone encapsulation material is a pourable liquid in its uncured state, so precise control of encapsulation volume is not feasible. Fortunately, some variability in encapsulant volume is tolerated. The primary requirements are that the encapsulant must completely cover the wire bonds, and that it can't extend above the top surface of the glass.
There also needs to be a mask that surrounds the active area of the cell (the region of the silicon containing the CMOS circuit). The function of the mask is to keep stray light from entering the cell and affecting image quality, and therefore the mask material must be non-reflective and non-transmissive. The mask should come as close to the active area of the cell as possible without obscuring any part of it. There are several possible approaches to masking, each of which occurs in a different point in the assembly process.
One solution is to have the mask be a separate piece attached on top of the cell after all other assembly is complete. This can work well, but attachment of this mask is a tedious operation, requiring manual placement under a microscope. Another option is to integrate the mask into the border placed on top of the cell. This simplifies the process flow but complicates border attach, because obtaining sufficiently precise border placement is extremely difficult. A third possibility is to put the mask directly on the glass by sputtering or evaporating a non-reflective coating and then patterning it. This is the easiest option from the point of view of assembly. Although a separate piece is still required on top of the assembly, as in the first option, the opening is much wider and precise positioning is not required. The problem is that it introduces difficulties in aligning the glass to the silicon during cell fabrication. In the standard cell fabrication process, the glass does not have any patterning on it, so precise alignment is not required.
Design Rules
The design rule requirements for assembly of LCoS imagers are different from those for conventional semiconductor packaging. There are critical dimensions at several different interface areas, and most of these affect the wire bonding process. As mentioned earlier, the silicon and glass are bonded together by an epoxy gasket. The gasket extends beyond the edge of the glass toward the bond pads on the silicon. The width of this gasket is determined by the degree of planarity of the wafer, the epoxy application method and the epoxy trace width. The critical dimension for wire bonding is the spacing between bond pad edge and gasket edge. The minimum spacing depends on the glass thickness, bonding sequence and the bonding wedge used, and is on the order of a millimeter. The minimum pad size and pad pitch must be chosen properly to minimize bonding yield loss and consider the possibility for rework. These requirements, along with scribe line width, define the necessary size of the silicon chip outside the active area.
Reliability Testing
Reliability testing and qualification of packaged semiconductors is typically done according to Mil-STD-883. The extreme temperatures mandated by this standard are, however, not suitable for LCDs. Instead, JEDEC Standard 23 (JESD23), “Test Methods and Character Designations for Liquid Crystal Devices,” provides criteria for high- and low-temperature storage, high-temperature/high-humidity storage, temperature cycling, thermal shock and high-temperature operating life (HTOL) testing. For example, the condition for HTOL testing is specified as 1000 hours at 50°C. Although this standard exists, an informal survey of LCD manufacturers showed that there is a wide variability in the testing conditions they use, and none use all the criteria specified by JESD23. Among the manufac-turers surveyed, HTOL tests were run anywhere between 50 and 85°C, all for 500 hours or less. Any drive to standardize reliability testing is complicated by the fact that manufacturers may use a variety of liquid crystal materials, each with a different nematic phase transition and maximum operating temperature.
Summary
LCoS microdisplays are still a fairly new technology, but the technology has matured to the point where commercial production is a reality. Technical issues have been resolved through the appropriate choice of materials, process flow and processing conditions. Exact manufacturing specifications and qualification requirements are still evolving, and it will be interesting to see what happens in LCoS microdisplay technology over the next few years.
AP
References
- D. McKnight, D. Vass and R. Sillitto, “Development of a Spatial Light Modulator: A Randomly Addressed Liquid-Crystal-Over-nMOS Array,” Applied Optics, Vol. 28, No. 22, 1989, pp. 4757-62.
- T. Baur, L. Pagano, M. Derks and D. Irwin, “High Performance Liquid Crystal Device Suitable for Projection Displays,” Projection Displays II, M. Wu, ed., Proc. SPIE, Vol. 2650, 1996, pp. 226-28.
- D. Banas, H. Chase, J. Cunningham, M. Handschy and M. Meadows, “Ferroelectric Liquid-Crystal Spatial Light Modulator for Projection Display,” Projection Displays II, M. Wu, ed., Proc. SPIE, Vol. 2650, 1996, pp.229-32.
- R. Melcher, “LCoS-Microdisplay Technology and Applications,” Information Display, Vol. 16, No. 7, 2000, pp. 20-23.
- F. C. Luo, “Active Matrix Liquid Crystal Displays,” Liquid Crystals: Applications and Uses, Vol. 1, B. Bahadur, ed., World Scientific Publishing, 1990.
For more information, contact JULIA GOLDSTEIN, owner, at Julia L. F. Goldstein Consulting, 4031 Twyla Lane, San Jose, CA 95130; 408-376-3987; Fax: 408-376-3987; E-mail: [email protected]. DEAN BUI was formerly a product/process development manager and RONG HSU was formerly director of operations at Aurora Systems Inc., San Jose, CA.