27-02-2013, 04:48 PM
GaAs MMIC Reliability Assurance Guideline forSpace Applications
GaAs MMIC Reliability.pdf (Size: 1.32 MB / Downloads: 100)
Abstract
This guide is a reference for understanding the various aspects of monolithic
microwave integrated circuits (MMIC). There are special emphases on the reliability
aspects of MMIC devices. GaAs material properties and common device structures
along with the applicable failure mechanisms are addressed in detail. MMIC design
and qualification methodologies provide the reader with the means of developing
suitable qualification plans. Radiation effects on GaAs devices and packaging effects
on MMIC device reliability are discussed with supporting references.
INTRODUCTION
This chapter establishes a common reference for the varied backgrounds of the
readers. It discusses reliability and quality assurance in general and reviews the effects of
new technology on the failure-rate distribution of the product. It also gives the reader an
overview of why gallium-arsenide (GaAs) is used and a brief summary of the
development of the monolithic microwave integrated circuit.
Why GaAs is Used
Perhaps the primary benefit of GaAs comes from its electron-dynamic properties.
In equivalently doped n-type GaAs and silicon, the effective mass of the electric charge
carriers in GaAs is far less than that in silicon. This means that the electrons in GaAs are
accelerated to higher velocities and therefore transverse the transistor channel in less
time. This improvement in electron mobility is the fundamental property that enables
higher frequencies of operation and faster switching speeds.
While the principal reason for making transistors out of GaAs is greater speed in
performance, which is realized either as a higher maximum frequency of operation or
higher logic switching speeds, the physical and chemical properties of GaAs make its use
in transistor fabrication difficult. Most of the early development in solid-state electronic
devices centered on silicon- and germanium-based materials because of the relative ease
with which the material could be processed. Silicon and germanium are elemental
semiconductor materials, whereas GaAs is a binary compound. This is the root fact that
caused many technical obstacles in the use of GaAs. Other properties not in GaAs’ favor
for early solid-state device development included a lower thermal conductivity and a
higher coefficient of thermal expansion than silicon and germanium. However, as new
market applications demanded higher performances that could be achieved only with the
superior electron dynamics of GaAs, these obstacles have been overcome.
Hybrid and Monolithic Integrated Circuits
From 1930 to 1960, microwave or high-frequency technology consisted of
circuits manufactured using waveguide: rectangular hollow metal pipes that “guided” the
electromagnetic energy to its destination. The design was usually experimental and the
production was generally expensive and long. At that time, the microwave engineer was
known as a “plumber” and his tool of trade was a hammer. Around 1960, the
development of semiconductors in “planar” geometries and the production of cheap, lowloss
dielectric materials were the beginnings of the microwave integrated circuit (MIC).
This technology was later called hybrid microwave integrated circuitry because the active
devices (such as diodes and transistors) and some of the passive elements (resistors,
capacitors, and inductors) were discrete components mounted to a dielectric slab or
substrate. The MIC utilized metal transmission lines that were photolithographically
etched onto the substrate to guide the electromagnetic energy to various components of
the circuit. The performance approached the design prediction better than the waveguide
predecessor, but many perturbations in the line geometries and inconsistent material
properties caused much of the final circuit layout to be experimentally determined. Other
factors that made hybrid-circuit production difficult were the labor-intensive processes of
assembly and electrical performance testing. The assembly process required mounting
each individual discrete device on the substrate, and, because of variations in component
placement, the electrical test operation required labor to tune the circuit performance.
The attachment of devices to the substrate and the tuning techniques required to make
them perform became an art form and a hard process to control. Eventually, at higher
and higher frequencies, these processes became the limiting constraints to performance,
cost, yield, and reliability.
Reliability and Quality Assurance
For any application, the user of the part wants the assurance that the part will
continue to function correctly over a given time and under certain environmental
conditions. Part failure at any given time takes place when the combined effect of the
stresses imposed on the part exceeds the part strength. These statements allude to the
time dependency of both part reliability and user expectation. For example, an
expendable system might have a useful life of 1 minute while a satellite system must
have a predicted life of several years. Each user has a different expectation of part
reliability and a different level of commitment to pay for the assurance that the part will
meet the expectation. Traditionally, the procurement of highly reliable (hi-rel) parts
meant that the user of the component specified to the manufacturer additional
requirements to be met in the fabrication of the part. These specifications were usually in
terms of recording fabrication process steps, performing additional visual inspections,
and incorporating additional screens and burn-in tests. The user of the hi-rel part was
expected and usually willing to pay the cost for this increased reliability and quality
assurance.