In this section we are considering some of the ways in which moisture impacts on the reliability of electronic assemblies. We look first at the effects of moisture on surfaces, and then on moisture transmitted into the bulk of polymeric materials. Inevitably, many of the faults we discuss will be related to semiconductors.
From general experience of life, we probably have some idea about what corrosion is and have experienced the higher levels of corrosion that occur in the presence of moisture and the hostile gas species, such as chlorine, ammonia, sulphur dioxide, hydrogen sulphide and oxides of nitrogen, that are often present in the atmosphere. Whether our experience is of tarnished silver cutlery or rotting exhaust boxes, what we have seen is corrosion, defined as ‘the destructive attack of a metal caused by either a chemical or an electrochemical reaction with the various elements in the environment’.
The phenomenon of corrosion involves reactions which lead to the creation of ionic species, by either loss or gain of electrons. Take the case of the rusting of iron, where metallic iron is converted into various oxides or hydroxides when exposed to moist air. The equations for this reaction are:
Because the iron loses electrons, chemists refer to this as an oxidation process. In this case, the resulting product is a mix of hydroxides; where copper reacts with acid residues creating chlorides and sulphides, this would equally (and somewhat confusingly) also be referred to as an oxidation process.
Corrosion of bare conductors will happen at a rate that varies substantially, depending on the conditions. Table 1 indicates some of the factors that affect such corrosion. Note that we are considering unprotected materials, and that the addition of an effective surface coating can substantially affect the outcome for the better.
|Conductors||Nature of the material or alloy
Nature of any reinforcement
Corrosive elements (type; concentration)
Another factor that affects the rate of progress of corrosion is the nature of the ‘corrosion product’. If the material produced by corrosion is insoluble and forms an impervious and tenacious layer, the corrosion reactions becomes self-limiting, as the corrosive medium can no longer diffuse through the corrosion product. A useful example of this is the oxidation of aluminium, which forms a thin protective layer of aluminium oxide; as ‘anodised’ aluminium, this is a good example of self-passivation.
If, on the other hand, the corrosion product is soluble or porous, corrosion will continue until the material is depleted, and no further reaction can occur. This is seen with the rusting of iron, where the oxide/hydroxide ‘rust’ has a different crystal structure from the iron, and creates only a porous, poorly adherent layer which does not protect against continued attack.
The corrosion situation is more complex if more than one metal is present within the system. Which metal will undergo oxidation is determined by the ‘standard electrode potential’, which is measured against an arbitrary zero created by a ‘standard hydrogen electrode’. Table 2 gives standard electrode potentials for many of the metals used in electronic packaging – the more positive the electrode potential, the more ‘noble’ the metal is, and hence the less prone to oxidise and hence corrode.
When dissimilar materials are connected together, the less noble metal will corrode relative to the more noble one. Thus, given a corrosive environment, one would expect the tin-lead of the joints to be corroded before the copper of the tracks, and this is indeed usually the case.
|Metal||Metal ion in equilibrium||Electrode potential at 25°C (V)|
Corrosion can occur with no external applied potential, and is merely dependent on the thermodynamic free energy of the reaction – if that free energy is negative (that is, heat is given out) then the reaction will take place spontaneously.
The rate of reaction is governed by a complex set of factors, but will always increase with temperature and depend on the concentration of the reacting species. For most metals of interest in packaging, corrosion rates below 30% relative humidity are too slow to be significant.
More than just moisture is needed for corrosion to take place in a semiconductor package. Moisture needs to be combined with ionic contamination. The problem of corrosion in an integrated circuit can thus be split into two parts:
External sources of ionic contamination can be the result of soldering, cleaning or contact with human skin. Within the integrated circuit, contamination can potentially come from both the die and the moulding. However, in modern plastics the ionic content is very low, so corrosion is mainly the result of reactions between moisture and:
The integrity of the die passivation is a major factor in building in resistance to corrosion, and experience with plastic packages has shown a relationship between the presence of cracks in these layers, and resulting corrosion. These micro-cracks may be caused by improper handling before moulding, or be present initially. Sources of micro-cracking during wafer fab include brittleness caused by incorrect deposition conditions and topographic reasons, such as poor step coverage.
Key requirements for reliability are therefore:
Electromigration is an electrochemical process where metal on an insulating material, in a humid environment and under an applied electric field, leaves its initial location in ionic form and redeposits somewhere else. Such migration may reduce isolation gaps and ultimately lead to an electrical short circuit.
The process begins when a thin continuous film of water has been formed and a potential is applied between oppositely charged electrodes.
Positive metal ions are formed at the positively biased electrode1 (the anode), and migrate toward the negatively charged cathode. Over time, these ions accumulate as metallic dendrites, reducing the spacing between the electrodes, and eventually creating a metal bridge.
1 Electromigration is closely related to corrosion, with the anode being attacked, but which circuit element is the anode is determined by the applied field rather than the oxidation potential of the metal used.
Whilst most often seen as a surface effect, vertical migration can also occur when moisture has penetrated into the bulk of a porous material.
The rate of electromigration increases with temperature and has four prerequisites – a mobile metal; a voltage gradient; a continuous film of moisture; soluble ions:
The move to finer pitch components, and consequently reduced spacings, makes electromigration more likely, because the voltage gradients between conductors increase and it becomes more difficult to eliminate the minute amounts of ionic contaminants which are sufficient to cause problems.
Three main methods have been used to reduce the electromigration problem:
What are the processes that lead to the failure of plastic-packaged semiconductor devices in a humid environment. What are the key factors in minimising failure from these?
All polymeric materials are permeable to moisture to a lesser or greater extent, and will absorb moisture when installed in a humid environment. Moisture absorption will thus have an effect on both packages and on boards – in fact any polymeric material used in assembly.
The amount of water absorbed, and the rate at which equilibrium is reached will depend on the type of package, the material, and the conditions under which the part is stored. The normal method of evaluating moisture take-up is by measuring the percentage weight gain. For example, for a PBGA exposed at 85°C/85%RH, equilibrium at about 0.5% weight gain occurs after 1 week (168 hours), but 70% of that absorption takes place in the first day.
If plastic packaged parts have been manufactured or stored under humid conditions, they can develop cracks on soldering, by whatever method this is carried out. Absorbed moisture turns to steam when heat is applied, building up a pressure of several atmospheres in the interior of the component, which causes the housing to ‘balloon’ and crack. The stages of this so-called ‘popcorn effect’ are shown in Figure 1. Ultrasonic microscope photographs confirm that delamination has occurred, providing an easy path to the chip and bond pads both for moisture and external sources of ionic contamination.
The sensitivity to moisture will depend on the construction: at the device level, delamination effects can be reduced by perforating lead-frames to improve adhesion, decreasing the particle size of fillers, and stamping lead-frames to eliminate burr formation sites that contribute to stress concentration.
The popcorn problem is most commonly seen with large flat devices, and has been a particular problem with Ball Grid Arrays. The effect can, however, be largely eliminated by ensuring that the components are manufactured and stored in a strictly dry environment, and only exposed to factory conditions for a very short period prior to solder assembly.
J-STD-020, a joint EIA/IPC standard, defines six levels of susceptibility to moisture in terms of the maximum allowable assembly floor life. The levels given in Table 3 are related to specific test conditions which are applied before reflow soldering and evaluation.
|5||£30°C/60%RH||24 or 48 hours|
Devices which pass the Level 1 test are classified as not moisture sensitive, and do not require dry packaging. All other parts should be fully dried by baking by the manufacturer and vacuum sealed in a moisture resistant bag, together with a packet of desiccant and a moisture indicator card. JEDEC specification JEP113B Symbol and Labels for Moisture-Sensitive Devices recommends a standard ID label, which incorporates the moisture-sensitive ‘three raindrops’ symbol.
After opening the bag, the moisture indicator card should always be checked to ensure the seal was not impaired. The parts then should be mounted within the floor life time given by the sensitivity level hours, depending on ambient conditions, or stored in an environment which prevents the total absorbed moisture from exceeding 0.1% of the total package weight.
A second ‘caution’ label on the bag should state the sensitivity level and the recommended heat treatment to be used to dry out packages suspected of having been stored in humid conditions. Typical conditions for re-baking are 192 hours at 40–45°C and <5%RH, or 24 hours at 120–130°C, the choice depending on the rated maximum temperature of the local packaging: reeled parts, for example, may not be able to withstand the higher temperature without the tape deteriorating.
Parts which will only pass Level 6 are regarded as extremely moisture-sensitive, with the dry pack providing inadequate protection. In such cases the customer must be advised of the classification at shipment, and a special warning label included to indicate that parts must be baked dry within the six hours preceding reflow. The minimum bake time and temperature for the device need to be determined from absorption studies.
What is meant by the popcorn effect. How can this be prevented?
The effects of moisture are not limited to the popcorning of active components; equivalent effects may be found with PCBs exposed to moisture and then soldered. Where damp boards are not thoroughly dried beforehand, they may delaminate during soldering, or at the very least there will be problems, such as solder balling on reflow and spitting and solder voids on wave soldering.
As with popcorning, most of the moisture-related issues are exposed during assembly. However, moisture can have an ongoing impact on circuit performance by affecting the electrical characteristics of the laminate. The higher the ambient relative humidity, the greater the moisture on the surface as well as in the bulk of the material. Ulbricht reported that the surface resistance expressed in ohms/square can vary as much as 4–5 orders of magnitude over a 30-90% relative humidity range. Dielectric losses also increase with increasing moisture content.
From this, one might suppose it better to keep parts dry, and assemble in a dry atmosphere. However, at humidities below 30%, static charging problems become severe. As a compromise, most fabrication and assembly shops operate at a relative humidity of around 50%.
We have seen already that moisture has an impact on plastic encapsulated micro circuits (PEMs) in the short term, before use, where absorbed moisture can result in failure. There is also a concern that moisture, in conjunction with sources of ions within the PEM, might result in unreliability. It was this concern that led to delays in the 1980s in converting from hermetic packages to PEMs for applications such as telephone exchange equipment.
Whilst hermetic packages can be assessed directly by leak testing, parts embedded in polymer have to be tested indirectly for moisture resistance by looking for the adverse effects of moisture penetration. This is done by subjecting parts simultaneously to high temperature and high humidity for an extended period, typically 1,000 hours, and is a destructive sample test to qualify the batch, rather than a 100% screen. Standard conditions start at 85ºC/85% relative humidity (RH), but more severe tests are frequently specified, in order to test the limits of the latest encapsulating processes.
Device failure will generally be accelerated by voltage as well as by temperature and humidity, and tests with applied bias are sometimes specified. However, heat generated by the chip itself will reduce moisture concentration at the surface and thus has some protective effect.
Special test chambers, often referred to as autoclaves because of their origin in medical sterilisation applications, apply elevated temperature and 100% RH simultaneously, but care has to be taken to ensure that moisture does not condense on the parts under test, especially when bias is applied to devices.
In hermetic packages, moisture can be excluded almost completely, but it can only be minimised in plastic packages. Not only are the intermolecular spaces of the plastic large enough to allow the diffusion of water, but additional paths for moisture penetration are found at the interfaces of different materials with badly optimised or weakened mutual adhesion. The most important interfaces are:
Stress is built up during shrinkage after moulding and by CTE mismatch between the different package materials. Unless precautions are taken, this stress will lead to adhesion failure and delamination, enhancing the diffusion rate of moisture at the interfaces. Stress can also damage the chip surface and create micro-cracks in the package, all of which facilitate the penetration of moisture, as shown schematically in. Figure 2
Important factors in minimising interface problems are:
How can you determine the long-term effects of moisture on a plastic-encapsulated integrated circuit, and how might the adverse effects be minimised?