By Dr. James A. Forster, Texas Instruments Interconnection Business, Mansfield, Mass. The development of the electronics industry traces its origins to the military and to space exploration-areas where uncompromising reliability is a must. The cost and inability to repair a satellite once launched was one of the earliest drivers for standardizing test processes to ensure reliable performance. One test method, burn-in, which was developed during those early days, has remained essentially unchanged. The infrastructure, the equipment, the test details and the strategies have changed, but the essential concept of stressing a component to identify the weak units, thereby causing them to fail, remains the fundamental principle behind this test. Burn-in today is an integral part of the BAT (Backend Assembly and Test) area of virtually all semiconductor producers. This part of the infrastructure, which includes testers, ovens, boards and sockets is about a $1 billion industry, according to many knowledgeable estimates. The Process Burn-in is a non-value-added process, and every semiconductor company employs a program to reduce or eliminate burn-in with ideas that range from BIST (built-in self-test) to wafer-level burn-in and test. Not all devices are burned-in; flash memory and many digital signal processing ICs, for example, might only be burned-in during the qualification process. Once the semiconductor processing technology is proven, and the reliability of the device has been shown to meet specific requirements, there is no longer a need for this process step. However, for ICs, such as DRAMs and microprocessors-which are at the leading edge of semiconductor fabrication technology-burn-in remains an essential part of the process. Rapid technological change and constant die shrinks contribute to a process that continues to evolve. This article provides some details on the issues of burn-in sockets and how socket suppliers serve this part of the infrastructure. What Is Burn-in? An IC is, at its most basic level, simply a series of gates, switches and transistors or amplifiers which, when arranged in a specific way, perform addition and subtraction. The beauty and elegance of semiconductor processing is the way in which these elements have been miniaturized, which enables designers to create circuits with millions of transistors. A natural consequence of such intricacy and miniaturization is the potential for a defect and its powerful negative impact on the functionality of an IC. The complexity of the IC manufacturing process is without equal. Unfortunately ICs are not "stand-alone" devices, but are assembled into products such as cell phones and computers. Because of the processes employed in the manufacturing process, some chips may be "fragile" and may fail after only a short period of time. It is essential that these devices be identified and scrapped before assembly into other products.
Failure Rate Vs. Time A typical reliability curve of failure rate versus time (which can be applied to many products), is shown in Figure 1. Because of the curve's shape, it is referred to as a "bathtub"curve. Fragile devices will fail soon after manufacture, as shown by the high initial failure rate, which is normally identified as "infant mortality." The failure rate is then flat as other types of random failures occur. After a period of time, when the product comes to the end of its useful life, the failure rate will increase as the product wears out. The problem faced by the manufacturer and reliability engineer is how to identify and remove devices that may fail. The process to identify and cull these "fragile" devices is known as burn-in, and involves stressing the part to force the failure of weak or fragile devices. Electrical and Thermal Stress Devices are stressed electrically and thermally; that is, they are placed in an oven at temperatures up to 150°C and voltages, which may be as high as 1.5 times the normal operating voltage, are then applied. At temperature time may be as short as a few hours or as long as 48 hours. Some space applications require two burn-in cycles, with total time at temperature of more than 240 hours. This stressing of the electrical paths within the chip, and the thermal expansion issues associated with heating the chip, has proven to be the most effective way to force early life failures. Burn-in Area The burn-in area of a backend assembly and test area is comprised of an assortment of custom ovens with test equipment attached (Figure 2). The ovens contain a series of slots that hold burn-in boards (BIBs).
These BIBs (Figure 3) are custom PWBs, which are designed to withstand numerous thermal cycles to 150°C and to provide the electrical interconnection between the tester and the DUTs. The socket is an integral part of the burn-in board and holds the device being burned-in. Sockets provide temporary electrical interconnection between the BIB and the IC's package. (A typical burn-in socket for a Tessera µBGA CSP is shown in Figure 4.)
Socket Types There are at least two dozen socket suppliers, servicing a worldwide market for sockets estimated at $400 million. The socket's purpose is to provide a temporary electrical connection between the BIB and the DUT. Each type of IC package is unique, and a socket must be developed and custom-built for each format, for example, TSOP or PGA packages. The essential parts of the socket are:
Each of these will be discussed below. Socket Body The socket body is the overall frame for the socket. It is normally made of a high-temperature glass-filled polymer material such as PPS (Poly-Phenyl-Sulfide) PES (Poly-Ether-Sulphone) or PEI (Poly-Ether-Imide). The body will be molded or machined depending on the volume of sockets required. The body holds all the components together and defines the footprint of the socket. If an alignment feature is part of the socket, it will be an integral part of the body or an insert. The development and growth of the different varieties of CSPs has challenged socket suppliers to develop ways to handle the different package outlines for given array formats. The package outline can be dictated by the chip size and this package format can have a variety of sizes for the same I/O format.
Socket designers must handle this variety of package sizes in a simple and cost-effective way. One way is the use of interchangeable adapters that individualize a socket to accept a specific package size. (An adapter is shown in Figure 5.) Contact The contact is the heart of the socket and everything else is there to allow the contact to fulfill its role of providing a reliable electrical connection between the DUT (or package under test) and the BIB. The contact is a spring that must be able to withstand the high temperatures used in burn-in. The most widely accepted contact material is beryllium copper, with the 172 alloy considered the standard of the industry. It is favored due to its excellent combination of high yield strength, excellent formability and good stress relaxation behavior. This material must be plated to prevent oxide formation and to ensure a low contact resistance. The most common system is gold with a nickel undercoat. However, some semiconductor companies have requested nickel-boron plating to eliminate any chance of the contact sticking to the solder ball. A number of different types of contacts have been developed to interface with the different I/O formats. For CSPs, a common I/O is a solder ball. For ball pitches of 0.75 mm and greater a pinch style contact similar to that shown in Figure 6 has found wide acceptance. The advantage of this contact type is that the probe (or witness mark) is above the equator of the ball and does not effect subsequent assembly operations.
Smaller Pitches The move to smaller pitches, such as 0.5 mm, has presented difficulties for a pinch-style contact, due to the amount of space between the balls for a movable contact. Socket suppliers are developing alternate contact systems such as the "spring" contact, which is shown in Figure 7. This spring is actuated vertically like a Pogo-pin-style contact but is significantly lower in cost. The major challenge for this contact is developing a low-cost method for making the springs consistently and developing a way to hold and assemble the springs in the socket. As pitches continue to be reduced below 0.5 mm, a new contacting technology will have to be found or developed. Actuation Mechanism There are two basic forms of sockets: open top and clamshell. An open top CSP socket is shown in Figure 4. Pressing on the cover actuates the socket and causes an internal component, the slider, to move sideways, which opens the pinch style contacts. With the contacts open, the package is dropped into the socket. As the package falls into the socket, the adapter guides it so that balls are located between the open jaws of the pinch-style contact. When the cover is released, the pinch-style contacts close and grip the sides of the solder ball, disrupting the oxide and establishing electrical continuity. Open-top sockets are preferred for automated loading and unloading of sockets. Clamshell Sockets A clamshell-style socket is shown in Figure 8. This socket comprises a frame or base and a cover. The cover is pivoted at the back of the socket and latches at the front. The lid contains a spring-loaded pressure plate, so that when the socket is closed the pressure plate pushes the package down onto the contacts. It is this pressure that causes electrical contact to be established between the spring and the solder ball. This socket contains the contact spring shown in Figure 7 and features a much simpler mechanism than the open-top variety Clamshell sockets are preferred for manual loading because the operator can easily latch and unlatch the clamshell to insert or remove the package being tested. Automated equipment is available for loading and unloading this type of socket. The clamshell, however, is principally a manual socket.
Interfacing with the BIB The type of contact employed defines the way in which the socket interfaces with the BIB. The pinch-style contact (shown in Figure 6), has a solder tail. This socket is assembled to the BIB and the solder tail passes through the burn-in board. The socket is permanently attached to the BIB by soldering. Sockets that fail can be removed by de-soldering the individual socket. This is not easily done, however, and there is some risk of damage to the BIB or to one of the circuits/interconnects. The clamshell socket shown in Figure 8 uses the compression spring contact. The socket in this case is located on the bib and held in place by four nuts and bolts. This socket is compression mounted onto the BIB by bolting the socket down. Should a contact be damaged or the socket fail, it can be easily removed and replaced with a new socket. There is a third mounting option, surface mount. While common in the assembly industry, this type of assembly process has not gained wide acceptance in the socket industry. In this form of mounting, the contact on the bottom of the socket is soldered in place on the BIB. The mechanical and electrical connection is the same and permanent. Again, individual sockets can be removed; however, this is not routine and the sockets or the PWB might be damaged. Summary This tutorial has reviewed the needs and methods of using sockets in the IC burn-in process. The type of contact and sockets used for CSPs has been described, including the materials and plating systems commonly used. The challenge facing socket manufacturers is to find ways to reduce the costs associated with the difficulty of making sockets as the pitch is reduced. A glossary of terms used in the burn-in socket industry accompanies this article.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2001