 |
Making "metallurgical" (=
"dirty") Simet is easy: Þ |
|
|
|
 |
A large scale
Simet production (> 1 Mio tons/a) exists for
metallurgical ("alloying") and chemical ("silicones")
uses |
|
|
A small amount of
Simet (some 20.000 to/a) is purified (factor
109 or so) to "semiconductor grade Si"
Þ |
|
|
|
Produce high-purity trichlorosilane
(SiHCl3) gas in a reactor and distill. |
|
|
|
Use SiHCl3 and
H2 to deposit Si on some Si core by a
CVD process |
|
|
The final result is ultra-high purity
(and expensive) poly Si (already doped if so desired) |
|
| |
|
|
|
|
Growing a "perfect" single
crystal from this poly-Si is not easy - but possible. |
|
|
|
|
The major crystal growth method is the CZ
(= Czrochalski) method: "Pull" the crystal from a crucible full of
molten Si. Þ |
|
|
|
Some (ususally < 300 mm diameter)
crystals are grown by the FZ (= float zone) method. Somewhat better
perfection, but more expensive than CZ. |
|
|
Major problem: Impurity segregation =
general tendency for most impurities (including doping atoms) to remain (=
enrich) in the melt. |
|
|
|
Segregation coeffcient =
ccryst/cmelt at interface, often
<< 1 and dependent on parameters like growth speed (usually a few
mm/min). |
|
|
|
+ Crystal is purer
than melt.
– It is practically impossible to grow a
crystal with a uniform impuritiy (including dopant!) concentration along its
length. |
|
| |
|
|
|
|
Produce wafers by cutting, grinding
and polishing |
|
|
|
|
Extreme precision for a mass product is
needed. |
|
|
|
"Flats" or "notches" (for
wafers > 200 mm) identify the crystallographic orientation and the
doping type. |
|
|
|
Beware! Flats are often custome specific and
different from the norm. Þ |
|
| |
|
|
|
|
| |
|
|
|
|
|
|
| Questionaire |
| Multiple Choice questions to
all of 6.1 |
|
| |
|
|
| |
|
|
© H. Föll (Electronic Materials - Script)
