Environmental Engineering Reference
In-Depth Information
bandgaps are needed. This is because as temperature increases the thermal energy
becomes sufficient to excite valence band electrons into the conduction band with
the result that the device becomes intrinsic and not semiconducting. For higher
energy band materials, the intrinsic concentration of carriers decreases, effective
mass increases, and dielectric constant of the effective mass decreases. Materials
more suited for elevated temperature applications include GaAs, SiC, GaN and
diamond. Table 6.6 summarizes the energy bandgaps of these semiconductors.
Table 6.6 Energy bandgaps of high temperature semiconductors
Material
Bandgap (eV)
Silicon (Si)
1.15
Gallium arsenide (GaAs)
2.15
Silicon carbide (SiC)
2.20
Gallium nitride (GaN)
3.45
Diamond
5.45
Because of the high bandgap leakage, current of these devices is lower for the
higher bandgaps, making their usefulness at elevated temperatures greater.
Threshold shifts in MOSFET and IGBT structures is lower for higher bandgap
materials and breakdown voltages are higher.
Thermal design is crucial for power electronic applications, particularly for
high power modules [19]. Long operating life and high reliability are attained
through minimization of thermal cycling, reduction in ambient temperature expo-
sure, and through proper design of the transistor stack. Thermal cycling causes
fatigue of the material interfaces in the transistor stack due to CTE mismatch of the
layers. This mismatch causes cracking of the solder attach, degradation of the
thermal resistance and eventual thermal overstress due to hot spots. Some brief
comments on reliability are covered in Section 6.9.
6.9 Reliability considerations
In broad terms, reliability is a subset of quality, and a metric that sets some
acceptable level on system dependability. In aerospace, the Federal Aviation
Administration describes the dependability of flight critical systems as having an
extremely remote probability of failure. In this context, the term 'extremely remote'
is interpreted as a rate of one failure in one billion hours of operation (1
10
9
).
According to Hammett and Babcock [20], achieving 10
9
dependability requires a
system with excellent first failure detection coverage. As insurance coverage against
first failure detection, a function must have at least dual redundancy and some means
of voting in the event one of the systems is not responding as anticipated. Aerospace
commonly requires triple redundant systems and various voting schemes to
ensure first detection and continued operation. The ability of redundant systems to
