|
|
|
Wafer Preparation Processes:
|
|
1. Ingot
Growth
|
Two methods are presently used to grow single-crystal
silicon. These two methods are called
Czochralski and
float zone
crystal growth, respectively. (These two crystal growth methods are
often abbreviated as CZ and FZ) (Gise and Blanchard, 1979). Approximately 75 percent of
the single-crystal silicon used today is produced by the Czochralski
process of crystal pulling; 25 percent is produced by the float zone
refining technology (Veronis,
1979). |
Fig. 1.1 |
Czochralski crystal growth utilized a crucible in which pieces of
polycrystalline silicon have been heated to their melting point of 1415°C
(Figure 1.1). The crucible containing the silicon is made of quartz (SiO2)
and is heated by either induction (RF) or thermal resistance methods. The
crucible rotates during the growth process to prevent the formation of local
hot or cold regions. The atmosphere around the crystal-growth apparatus or
crystal puller is controlled to prevent contamination of the molten silicon.
Argon is often used as the ambient gas. When the temperature of the silicon
has stabilized, an arm with a piece of silicon mounted on the end is slowly
lowered until it comes into contact with the surface of the molten silicon.
This piece of silicon is called as the seed crystal. As the bottom of the
seed crystal begins to melt in the molten silicon, the downward motion of
the rod holding the silicon is reversed. As the seed crystal is slowly
withdrawn from the melt, the molten silicon adhering to the crystal freezes
or solidifies, taking on the crystal structure of the seed crystal. The rod
continues its upward movement, forming an even large crystal. The crystal
growth terminates when the silicon in the crucible is depleted. By carefully
controlling the temperature of the crucible and the rotation speeds of the
crucible and the rod, precise control of the diameter of the crystal is
maintained. The desired impurity concentration is obtained by adding the
impurities to the melt in the form of heavily doped silicon prior to crystal
growth.
Back to Top |
|
2. Wafer Type and Orientation
The silicon
crystals are first ground perfectly round (if necessary), and the rotational
orientation of the crystal is ascertained. The seed crystal has determined
which crystal face will be presented on the wafer surface, but the
rotational position of the rod determines other axes of the crystal. Since
the bar of silicon is one crystal, it has preferential break or cleavage
planes. It is critical for later device separation to align the circuits
precisely with respect to their cleavage planes. This precise alignment is
accomplished by grinding a flat along the crystal (prior to slicing) that is
used as a reference during all subsequent processing steps. X-ray
diffraction provides a fast and accurate method of determining the crystal
orientation prior to grinding the flat.
Fig. 2.1 |
The crystal is usually
doped p- or n-type while grown, depending on wafer specification. For
coarse wafer alignment and future identification, a notch and one or
more flats are ground along the length of the crystal. The primary flat
is typically located on a (01 )
surface and is used (with a secondary flat, when present) to quickly
identify the type and crystal orientation of the wafer. |
|
|
Back to Top |
|
3. Slicing, Lapping and Polishing |
|
The silicon
crystal is then sawed into thin slices of thickness 0.5-1 mm called wafers.
The process of cut the silicon ingot into thin slices is called slicing as
shown in Fig. 3.1. Extreme care is taken
to minimize the amount of the single crystal silicon that is lost in the
slicing process by using the inside diameter of a ring-shaped saw blade. The
blade is coated with diamond powder to enable it to cut through the hard
silicon. The slicing process leaves wafers with saw marks on both sides that
must be removed. A silicon etchant is used to remove the saw marks and any
accompanying damage from both sides of the wafer.

Fig. 3.1 Schematic diagram of the slicing process.
The wafers are then lapped, polished, and cleaned to remove the damage
caused by slicing. Care must
be taken to remove any crystal damage introduced by the slicing operation, or
the damage may prevent the successful fabrication of devices. The wafers are
next mounted on large circular polishing plates using either wax or a vacuum
to hold them. The polishing plates are mounted on a polisher, and one side
of the wafer receives a mirrorlike finish. The polishing operation uses a
polishing solution that simultaneously chemically etches and mechanically
polishes the wafers. The polishing pad must be tough and durable. When the
wafers have reached the proper thickness rang and surface quality, the
polishing plates are removed and the wafers are dismounted. The wafers are
thoroughly cleaned to remove any residual contamination, and inspected to
insure that wafers with imperfect surfaces are not shipped. Wafers that pass
the finial inspection are ready to start on their journey to become devices
(Gise and Blanchard, 1979).
Back to Top |
|
4. Dicing
After integrated
circuit (IC) fabrication and before chip package, the wafer with well-done
IC is diced into individual die with a diamond-tipped wheel, as shown in
Fig. 4.1. This process is called dicing. Dicing by grinding causes fracture
and cracks to the edges of the dies, which are to be removed by a chemical
etching process.

Fig. 4.1 Schematic diagram of the dicing process.
Back to Top |
|