As wafer [ weɪfə (r )] ( English for " waffle " or " wafer " ) circular or square, about one millimeter thick discs are referred to in microelectronics, photovoltaics and micro system technology. They are blanks, so-called ingots made of one or polycrystalline (semiconductor), and generally serve as a substrate ( base ) for electronic components, including integrated circuits ( IC "chip" ), micromechanical components or photoelectric layers. In the manufacture of microelectronic devices usually several wafers are combined into one lot and directly behind one another or processed in parallel ( cf. producing batches ).
A disc, in most cases, made of monocrystalline silicon, but also other materials such as silicon carbide, gallium arsenide, and indium phosphide is used. In microsystems technology glass wafers are used with a thickness in the 1- mm range.
The discs are manufactured in various diameters. The wafer diameter primarily used for time differ depending on the semiconductor material and the intended use ( silicon: 150 mm, 200 mm and 300 mm - 450 mm in the discussion; gallium arsenide: 2 inches, 3 inches, 100 mm, 125 mm and 150 mm - 200 mm is technically feasible ). The larger the wafer, the more ICs can be accommodated thereon. As with larger wafers, the geometric blending is smaller, the ICs can be produced more cost-effective (see yield ( Semiconductor Technology ) ). To maximize the yield, the wafers are produced in clean rooms.
The cost of manufacture of unstructured wafers depend on the diameter and the material ( silicon, germanium, gallium arsenide, etc.) and the manufacturing process (see below) from. The cost of processed wafers - so-called structured wafers - are rising sharply, were the more process steps are carried out. After the production of STI structures, the costs compared to unstructured wafers have at least doubled. Besides the number of processing performed, the costs also depend strongly on the structure size used. Computer chips on an average 200 - mm wafers with a feature size of 90 nm (90 nm technology ) by mid- 2008 € 850 per wafer. The cost of production of top quality products ( on 300- mm wafers ) in AMD graphics cards in 28nm technology, Intel processors in 22 nm technology, however, are significantly higher. Depending on the chip size can be produced on a wafer as a few dozen to a few hundred chips. Not included in these costs are expenses incurred after the chip fabrication, such as the packaging chips into housing.
Wafer fabrication begins with a block of semiconductor material, the ingot is referred to. Ingots can be constructed monocrystalline or polycrystalline and are usually made with one of the following methods:
- Zone melting method
- Czochralski method, including the liquid - encapsulated Czochralski method ( LEC method )
- Bridgman -Stockbarger method
- Vertical gradient freeze (VGF )
- Ingot casting method or Bridgman method with a controlled melting and cooling schedule, for polycrystalline silicon
All of these methods provide, in effect, more or less cylindrical or square single or polycrystals, must be sawn transversely to its longitudinal axis into slices, the wafers. In order to optimize the precision for this particular section in as little waste, separating the inner hole has been developed. The blades wear while cutting teeth (possibly cutting diamonds) on the inside of an inner hole that needs to be slightly larger than the diameter of the blank. Meanwhile, however, has also the wire saw, which was originally developed for solar wafers, more and more established.
In the literature there are special wafer labels that specify, among other things, that the production process was used. For example, wafers that have been produced by the Czochralski method, referred to as a CZ wafer. Similarly, the term FZ wafer, wafer, with the zone melting method ( engl. float zone ) were prepared, used.
The surfaces of the wafer must be mirror polished optically, for most applications. To the first lapped wafer and then treated by means of a chemical-mechanical polishing, until the required surface roughness ( a few nanometers ) is reached. Other important geometric parameters of wafers are the global thickness variations (English total thickness variation, TTV ), the type and size of warping ( engl. wafer warp ) or bending ( engl. wafer bow ) and much more.
Since the exact position of the machining machine is important for the processing of the wafers, the wafers are characterized (in the case of gallium arsenide to 125 mm in diameter, wherein the silicon to 150 mm in diameter) with a so-called " flats " (English for " flattening "). It is indicated by means of a primary and possibly a secondary Flats, which angular orientation is present and which crystal orientation, the surface ( see figure). With larger wafers ( for silicon from 150 mm in diameter) called notches ( notches ) are used instead of the Flats. They offer the advantage of more precise positioning and mainly cause less waste.
Nowadays also a unique wafer marking is written as a bar code, OCR-readable text and / or data matrix code by laser to a spot near the notch on the edge of the wafer base.
In the photovoltaic
In photovoltaics, two types of wafers can be distinguished in general: polycrystalline (also called multi-crystalline ) and monocrystalline wafers. The production takes place for both types by sawing corresponding ingots. Polycrystalline ingots are made of square-shaped polycrystalline silicon blocks, from which the shape of the wafer yields ( usually square ). Monocrystalline wafers are, however, cut from cylindrical monocrystalline ingots, as they are also used for microelectronic applications. They generally have a " pseudo- square " form, i.e., with rounded corners. In contrast to square cut wafers falls in the generation of the round monocrystalline ingots in less waste. Inefficient waste- rich method are to increase costs and worsen the environmental balance. In addition, the waste is contaminated by the cutting tool and the wire abrasion ( and forms a suspension ) and can be recovered only with difficulty again. Other methods such as " Edge - defined Film - fed Growth " ( EFG ) of Schott Solar, or " String Ribbon " (SR ) of the company Evergreen Solar makes it possible to draw very thin wafers directly from the melt. The waste water, energy, and waste-intensive wire saws omitted here. The wafer thickness is usually much thinner than in microelectronics, about 200 microns in the current mass production. It does not use polishing process. From the wafers in several subsequent processing steps solar cells and solar modules manufactured therefrom again.