SEMICONDUCTING MATERIALS ENGINEERING PROGRESS

I. In microelectronics, the steady reduction of IC feature¹ sizes, accompanied by high current densities and increasing de­mands on electrical performance, has focused the attention of technologists on newer materials which exhibit² characteristics such as low contact resistance, reduced vulnerability³ to elec-tromigration, and processibility⁴ at low temperatures.

Over the years, the device size has been reduced tremen­dously. Improvements available⁵ in materials technology have allowed integration of more and more devices on the same chip, resulting in increased area. According to the theory of scaling, the smaller dimensions of a MOS transistor should enhance⁶ its speed. As a first-order approximation, therefore, this should pro­portionally increase the circuit speed. Indeed, for smaller circuits it does happen. However, for large circuits, the time delays⁷ asso­ciated with the interconnections can play a significant⁸ role in determining⁹ the performance of the circuit.

As the minimum feature size is made smaller, the area of cross section of the interconnection also reduces. At the same time a higher integration level¹⁰ allows the chip area to increase, causing the lengths of the interconnections to increase. The net¹¹ effect of this "scaling of interconnections" is reflected into an ap­preciable¹² RC time delay. For a very large chip with extremely small geometries, the time delay associated with interconnections could become an appreciable portion of the total time delay, and hence the circuit performance could no longer be decided by de­vice performance.

Thus, as the chip area is increased and other device-related¹³ dimensions are decreased the interconnection time delay becomes significant compared to the device time delay and dominates the chip performance. These are dominant factors limiting device performance.

Performance is the obvious goal of VLSI; reliability is a more subtle¹⁴ one. Therefore, new materials are required for VLSI in­terconnections.

The design¹⁵ of any machine or a device has always been limited by the materials available. The problem in question was that materials could be designed and tailored¹⁶ for any new structures.

Semiconductors are used in a wide variety of solid-state de­vices including transistors, integrated circuits, diodes, photodiodes and light-emitting diodes.

Several elements in and around group IV of the Periodic Table show intrinsic¹⁷ semiconductor properties but of these Ge and Si (and to a lesser extent Se) alone have shown chemical and electrical properties suitable¹⁸ for electronic devices operating near room temperature.

Germanium and silicon were the first semiconductor materi­als in common¹⁹ use.

A great contribution²⁰ to the study of semiconductor physics has been made by the prominent Soviet scientist A.F.Yoffe. It was in 1930 when Academician A.Yoffe and his co-workers started a systematic research in the field of semiconductors.

The diffusion theory of rectification²¹ on the boundary of the two semiconductors was elaborated by B.I.Davydov, a Soviet physicist, in 1938. Experimental support of his theory was of great importance in the investigation of processes occurring²² in p-n junctions.

Right after World War II, physicists John Bardeen, Walter Brattain and William Shockley, and many other scientists, turned²³ full time to semiconductor research. Research was cen­tered on the two simplest semiconductors — germanium and sili­con.

Experiments lead to new theories. For example, William Shockley proposed an idea for a semiconductor amplifier²⁴ that would critically test the theory. The actual device had far less am­plification than predicted. John Bardeen suggested a revision the­ory that would explain why the device would not work and why previous experiments had not been accurately foretold by older theories. In new experiments designed to test tie new theory they discovered an entirely new physical phenomenon — the transistor effect. In 1948 W.Shockley patented the junction transistor. Junction transistors are essentially solid-state devices having three layers of alternately²⁵ negative or positive type semiconductor material.

The early history of modern semiconductor technology can be traced²⁶ to December 1947 when J.Bardeen and W.H.Brattain observed transistor action through point contacts applied to poly-crystalline germanium. Germanium has become the material in common use. It was realized that transistor action occurred within the single grains²⁷ of polycrystalline material.

G.K.Teal originally recognized²⁸ the immense²⁹ importance of single-crystal semiconductor materials as well as for providing the physical realization of the junction transistor. Teal reasoned³" in 1949, that potycrystalline germanium's uncontrolled resistances and electronic traps³¹ would affect³² transistor operations in uncontrolled ways. Additionally,³³ he reasoned that polycrystalline, material would provide inconsistent product yields and thus be costly. He was the first to define chemical purity,³⁴ high degree of crystal perfection³⁵ and uniformity of structure as well as controlled chemical composition (i.e. donor or acceptor³⁶ concentration) of the single-crystal material as an essential foun­dation for semiconductor products.

The next decade witnessed³⁷ an ingermanium and the "universal" semiconductor material, silicon. Silicon gradually gained³⁸ favour over germanium as the "universal" semiconductor material.

Silicon is to the electronics revolution what steel was to the Industrial Revolution.

II. Silicon has been the backbone (основа) of the semiconductor industry since the inception of commercial³⁹ transistors and other solid-state devices.

The dominant role of silicon as a material for microelectronic circuits is attributable⁴⁰ in large part to theproperties of its oxide.

Silicon dioxide is a clear glass with a softening⁴¹ point higher than 1,400 degrees C. If a wafer⁴² of silicon is heated in an atmo­sphere of oxygen or water vapour,⁴³ a film of silicon oxide forms on its surface. The film considered is hard and durable⁴⁴ and adheres⁴⁵ well. It makes an excellent insulator. The silicon dioxide is particularly importarit in the fabrication of integrated circuits because it can act as a mask⁴⁶ for selective introduction of dopants.⁴⁷

Silicon's larger band⁴⁸ gap⁴⁹ permitted⁵⁰ device operation at higher temperatures (important for power devices) and thermal oxidation of silicon produced a non-water-soluble stable oxide (as compared to germanium's oxide) suitable forpassing p-n junctions, serving as an "impermeable⁵¹ diffusion mask" for common dopants, and as insulator coating⁵² for conductor overlayers.⁵³

Oxygen concentration present influences many silicon wafer properties, such as wafer strength, resistance to thermal warping (скачок), minoritу carrier lifetime and instabilitу in resistivity.

The presence of охуgen contributes to both beneficial and detrimental⁵⁴ effects. The detrimental effects can be reduced if the oxygen is maintained⁵⁵ at less than 38 ppms. Thus, the oxygen range⁵⁶ of the wafer present should be controlled. The results achieved with silicon are great.

However, although the silicon wafer clearly is a fundamental ingradient in the fabrication of an integrated circuit, the silicon materials specification⁵⁷ may not be critical element in doveloping a successful new 1C product strategy. If silicon material is to remain the semiconductor device material for the next ten years efforts must continue to reduce crystallographic defects, grown-up impurities introduced during device fabrication.

Large-scale integration (LSI) of devices has put great demands on electronic-grade single-crystal material. The semiconductor industry now requires high purity and minimum point-defects concentration in silicon in order to improve the component yield per silicon wafer. These requirements have become increas­ingly stringent⁵⁸ as the technology changes from large-scale integration (LSI) to very large-scale integration (VLSI) and very high speed integrated circuits (VHSIC).

The yield (or circuit performance) of a device and the intrinsic and extrinsic materials properties of silicon are interdependent. The silicon wafer substrate must be practically defect-free when the active device density may be as high as 10^s to 10⁶ per chip.

To increase further the speed of semiconductor devices re­quires not only refinements⁵⁹ in present designs and fabrication techniques, but also new materials that are inherently⁶⁰ superior to materials presently being used, like germanium and silicon. New material under consideration is gallium arsenide.

Gallium arsenide has a much higher electron mobility than germanium and silicon. The opportunities⁶¹ present are as fol­lows: it is potentially much faster; it has a larger band gap, permitting operation at higher temperatures; it is chemically and me­chanically stable. Mobilities in this high-purity gallium arsenide are about twice those of germanium and four times those of sili­con.

The potential of high-purity gallium arsenide was first ex­plicit⁶² in a new gallium arsenide-germanium hetero-junction diode. The hetero-junction device has a potential for much faster switching than conventional⁶³ p-n junction diodes. Its calculated switching time is on the order of a few picoseconds (trillions of a second).

However, the difficulty of producing gallium arsenide of suf­ficient⁶⁴ purity has limited its application.

Yet, gallium arsenide is far from the end of the story. Any searching for an answer makes contributions. This is the way of developing better materials and devices.