Reliability of Electronic Components

Reliability is a relatively new concept, which rounds off the quality control and is linked to the study of quality itself. Simply explained, the reliability is the ability of an item to work properly; it is its feature not to fail during its operation. One may say that the reliability is the operational certainty for a stated time interval. This definition is however imperfect, because although containing the time factor, it does not describe precisely a measured size.

Today, the manufacturing of electronic components is the most dynamic process, because the great demands imposed on the performance specifications of modern devices determine a quick rate of change for these products. The electronic components and, especially, the semiconductor devices have always been thought as having the potential to achieve a high reliability and, consequently, the development of many quality and reliability techniques was made particularly for these devices. Consequently, the reliability researches on this field stand for the front line in the battle for the best products.

An understanding of how electronic parts fail’ is essential for the improvement of the reliability of devices as well as of the systems in which they are used. Such failure analyses help in identifying device failure modes, mechanisms, and stress factors that influence degradation [3.1]. The early failures (infant mortality, or failure during the burn-in or debugging period) occur at high initial failure rate λ- which is the number of failures of a part per unit of time. This failure rate decreases rapidly and stabilises at a time T_B, when the weak units have died out. Early failures (poor solder joints, wire bonds, or hermetic seals; poor connections, dirt or contamination on surfaces or in materials; chemical impurities in metal or insulation; voids, cracks, and thin supports in insulation or protective coatings; and incorrect positioning of parts) are characterised by a high but rapidly decreasing hazard rate² and are caused by many factors, including the built-in flaws of faulty workmanship, inferior materials and inadequate process controls, transportation and assembly damages, and installation and test errors.

The diodes and the rectifiers are bipolar components with non-linear constants which have a different behaviour, depending on the polarisation of the applied voltage [4.1]…[4.13]. The silicon is used almost exclusively as semiconductor material. The main constructive forms are planar diodes and MESA diodes. Among the important characteristics that can’t be exceeded, the reverse voltage, the forward current and the maximum junction temperature (including data concerning the thermal resistance at high temperature) may be mentioned.

The explosive development of the semiconductor technology imposes greater demands on the quality and reliability of the components. The problem of the failure rate during the life of a device or system is more and more important and the failure analysis helps to the improvement of the product quality.

The Silicon Controlled Rectifier (SCR), invented in 1958, in the laboratories of General Electric, is the most important member of the thyristor family of semiconductor components, including the triac, bi-directional diode switch, the silicon controlled switch (SCS), the silicon unilateral and bilateral switches (SUS, SBS) and light activated devices like the LASCR. Most recent members of the thyristor family are the complementary SCR, the programmable unijunction transistor (PUT) and the asymmetrical trigger [6.1][6.6]. As a silicon semiconductor device, the SCR is compact, static, capable of being passivated and hermetically sealed, silent in operation and free from the effects of vibration and shock. A properly designed and fabricated SCR has no inherent failure mechanism. When properly chosen and protected, it should have virtually an operating life without limits,even in harsh atmosphere. Consequently, countless billions of operations can be expected, even in explosive and corrosive environments. All components — including power semiconductors — have the potential of failing or degrading in ways that could impair the proper operation of such systems. Well-known circuit techniques (including fusing and self-checking) are available to protect against the effects of such phenomena. For any systems where safety is in question fault analysis is recommended.

Even from the beginning, the semiconductor industry was characterised by a high innovation rate. A spectacular moment was the appearance of the integrated circuits on the market, allowing high cuts of price and performance growth. The first integrated circuit (reported by Jack Kilby and Robert Noyce) was not a sudden discovery, being prepared by previous devices. Invented in 1958, the solid-state circuit was developed in 1959, when the planar technique arises. This was the milestone for subsequent development of the monolithic integrated circuits, containing bipolar and unipolar (mostly MOS) transistors, based on a silicon substrate. The global market for semiconductor devices increased with 15% per year in the last twenty years, reaching $ 140 billion in 1997.

Silicon technology was (and still is) the dominant technology of the semiconductor industry; silicon devices have more than 95% market share of the over $140 billions semiconductor business at the present time. Greater integration, higher speed, smarter functions, better reliability, lower power and costs of a silicon chip are the permanent goal in order to meet the increasing requirements of information technology. The industry progress has closely followed two laws. The first is the Moore’s law, the 1975 observation by Gordon Moore that the complexity of ICs had been growing experimentally by a factor of two every year. He attributed this to a combination of dimension reduction, die size increase, and an element which he called “circuit and device cleverness” — improved design and circuit techniques which allowed more function per unit area at a given lithography. With a slowing down of the rate of progress to a factor of two every 1.5 years, Moore’s law continues to hold well today. The second law is the law of π, a somewhat tongue-in cheek statement that memory chips, in a given generation, sell for about π dollars when they reach their peak shipping volume, and eventually reach a selling price of π/2 dollars. The law has not really held, though in constant dollars it is not too bad, but the point is that the cost of a chip has only gradually increased from generation to generation, held down by the ability of the industry to yield larger and larger chips while making them smaller on increasingly larger wafers. Device miniaturisation was the main trend (Fig. 9.1), and the silicon device technology progress followed the scaling-down principles and Moore’s law for the last three decades.

Visible light-emitting diodes LEDs (red, green, yellow, and blue) became indispensable as visual indicators. Combinations of LEDs — in a hybrid or monolithic form — are among the competitors for the lucrative visible alphanumeric display market. Reliability of such LEDs is now almost taken for granted; however the main emphasis of this chapter will be the understanding of degradation processes in LEDs and optocouplers. In Fig.10.1 a classification of optoelectronic semiconductor components is given.

Much work has been carried out in the past to study the various types of (low-frequency excess) noise sources as they commonly occur in silicon planar transistors used in monolithic integrated circuits. Some examples of such noise sources are presented in the following.

In the beginning, only metallic packages were used for transistor encapsulation. These type of packages seemed to be very reliable, both for military and civilian applications. In 1962, General Electric used for the first time plastic packages for transistors. Thus, the costs were significantly reduced, even with 90% in some cases [12.1]. First, plastic encapsulated transistors were developed for mass consumption, without taking into account the reliability or the environment. Therefore, the low cost of these new transistors called rapidly industry and army’s attention. Consequently, starting from 1964, their market increased appreciably. Almost immediately, the weaknesses referring to the reliability were revealed, especially in combined conditions of high temperature and moisture, when the failure rate increases dramatically compared with the metal encapsulated transistors’ one. This explains why, with rare exceptions, at the time, the plastic package was not accepted by the army.

Now, the electronic components are currently used in all activities. Technology growths so rapidly, that manufacturers and users must accommodate an increased importance to the components control. To make this, it is absolutely necessary to have computer-assisted equipment. For the acquisition of such very modern computerised equipment, high investments are needed, not only for the machines, but also for personnel instruction, drawing up of software control and managing programmes.

This delicate and, in the same time, interesting subject will be presented from the viewpoint of a simple user of components. The emphasis will be not on the solid physics, but on the adequate design of electronic systems or on their manufacturing on an industrial scale. The scanning electron microscope, for instance, will be mentioned only incidentally, while the habitual analysis means (such as: electrical measuring, optical microscopy and chemical procedures) will be currently used.

Designed and manufactured by the Swiss Technology Corporation Oerlikon Contraves AG, RAMTOOL++ is intended as an improved tool for Reliability, Availability and Maintainability (RANI) engineering. Proved since 1985 on daily operation, the tool had been successfully employed on the European and North American market on advanced programs. RAMTOOL++ was elaborated by a multinational and various engineering disciplines team (mechanics, electronics, and mathematics) and ensures Concurrent Engineering for RAMs.