Shorts in MIBs can be either as-processed or latent-type
defects. Shorts are due to printing errors, pinhole voids in the dielectric,
or in the case of copper, protuberances and thin film plating on the surfaces
of the pores. Look to printing technique and screen or artwork problems to
correct printing errors. Copper film shorts are mostly due to improper drying
and firing conditions.
5-6 shows pictures of a copper growth called a protuberance. It is believed
that these defects are caused by the reduction of copper oxide in the pores of
the dielectric, leaving an essentially pure copper filling. The films are
similar but less developed buildups of copper. According to this theory,
four conditions must be present for the formation of these copper
precipitates. There must be a source of copper oxide; the temperature must be high enough
to make the oxide mobile in the glassy phase of the dielectric(870°C): the
affected location in the MIB must be subjected to reducing conditions; and the
time at temperature must be long enough to allow the build-up of the
precipitate to be significant.
Copper oxide is only mobile in the glassy phase of the
dielectric. Therefore, any time the dielectric gets fired into a dense,
low-porosity film, it is by nature going to be a greater problem for
as-processed shorts. The dilemma is that a dense dielectric is necessary for
reliable circuit operation under adverse environmental conditions. Oxygen
doping of the furnace atmosphere has been shown to densify the fired
dielectric film. Also, it has significantly increased long-term reliability
when ionically contaminated MIBs are subjected to a voltage bias at
temperature. The problem then becomes one of balancing conditions such that
acceptable yields and adequate dielectric density are obtained.
Opens can be due to skips on printing, poor screens and/or
insufficient paste on the screen, and contamination such as lint which on
burnout leaves an open path. A well-controlled process (and environment), good
screens, and thorough inspection for breaks after firing will generally
eliminate these opens. There are, however, some mechanisms peculiar to copper
processing which can yield either initial opens or latent opens.
Figure 5-7 and Figure 5-8 are
line opens caused by different but related mechanisms. The open of Figure 5-7 is
a result of large grains with impurities in the grain boundary
(possibly bismuth reduced from bismuth trioxide). This causes the line to be brittle
at that point and possibly to break during low temperature thermal stress.
Figure 5-8 shows a condition of poor sintering where hydrocarbons or
retained carbon prevented the intimate contact required to form a continuous
sintered layer. Both examples are caused by tiring conditions. That of Figure 5-7
is believed to be the result of reducing conditions at firing temperatures, while
that of Figure 5-8
is thought to be a result of inadequate burnout or possible redeposition
of hydrocarbonsat 400°C to500°C.
A continuity test following a
series of thermal cyclesfrom -55°C to+100°C has been somewhat effective in identifying
circuits with brittle copper. Sectioning and examination of the structure is
necessary to determine if the open is related to the conditions of Figure 5-7
or Figure 5-8 or if it is related to the printing and fabrication process. The
best method of controlling brittle copper defects is to determine furnace
conditions necessary to prevent them and then to tightly control the furnace
at that level. Figure 5-9 illustrates a normal grain structure.
Blisters and areas of low adhesion under conductors can
generally be traced to surface contamination and/or improper burnout of the
organics. Retained organics are mainly a copper problem and will cause a
blister if a dense film is fired over a layer with the poor burnout. Copper
conductor traces are quite impervious relative to the dielectric. Any gas
forming under a copper line will have an adverse effect on the adhesion of the
conductor to the dielectric at that point, possibly causing a blister. This is
because gas cannot permeate through the copper, and if it cannot diffuse
laterally around the line or pad, it will simply lift the copper.
Since copper oxide is necessary for the adhesion of the copper
to the dielectric, any condition which depletes the oxide layer at the copper
dielectric interface will reduce adhesion and possibly cause delamination.
Firing the circuits in reducing conditions is a primary example. Reducing
conditions cause so many potential problems with copper multilayer circuits
that this parameter should be most closely monitored and controlled. Practical
considerations such as varying oxygen content during burnout make this
difficult to do. Failure analysis for the detection of reducing conditions
must be constantly applied so any negative effects can be corrected
Occasionally, blisters also occur in the dielectric. Generally
not as serious as copper blisters, these are caused by gelled polymer vehicle,
entrapped organics, impurities, or incomplete drying.
Bow of MIBs can be a problem if it is excessive. Generally,
excess bow is caused by firing at temperatures in excess of the recommended
range. Any factor which increased dielectric density will also increase bow.
Excessively glassy paste formulations, the type of glass, and the amount of
oxygen in the furnace atmosphere all affect bow.
Discoloration of air-fired noble metal systems generally
indicates material contamination. Blushing is the most common discoloration
problem experienced with copper MIBs. Although not generally considered
detrimental to the product, it is an indication of the conditions under which
the circuit was processed. Firing in reducing conditions and incomplete
washing of a defective print will cause this