Experts in wire bonding have written several books and articles on the subject, with a passing reference to plasma treatment as a way to impact the bonding process or the bond’s long-term reliability. However, there is not a lot of literature available on plasma applications in microelectronics packaging, especially in using plasma before wire bonding.
Figure 1. Bench-top plasma cleaning system for surface activation and adhesion improvement (AP600)
There are practical expectations of what plasma gives to both a wire bonding process performance and a packaged device long-term reliability. Also while developing a plasma process, a specific logic needs to be followed, even in case it is not hard and fast as some may desire.
Plasma benefits are found in two different areas:
- The wire bonding process – to be referred to as 'the statistics'
- The device’ long-term reliability – to be referred to as 'device reliability'
Even though these two areas are intertwined to a certain extent, they must be dealt separately to have more clarity.
Plasma and the "Statistics" of the Wire Bonding Process
Variety of Bonding Surfaces
Currently, “first” bonds are fabricated to an aluminium metallization on a semiconductor device and “second” bonds to gold. There is, however, a rapid change in this scenario due to the advent of copper and other wires and of new metallizations on both devices and the substrates or leadframes they are bonded to.
It is important to note that 50nm flash gold and 2µm of thick gold need to be treated differently. Also there is a difference in treatment of palladium or nickel metallization on an IC and on aluminium. In comparison with BGA substrates, metal leadframes undergo a separate treatment.
The semiconductor industry implements strict process control, which results in a foreseeable metallization, however the printed circuit board and plating industry processes show a higher variability.
Contamination on the Surface to be Bonded
Ideally, the bonding must be done onto clean metal surfaces on both the leadframe or substrate and the semiconductor device it is being bonded to. The possible exception is a thin oxide layer. Practically, there are several surface contamination sources, which influence the device reliability and the wire bond statistics.
- Inorganics, especially fluorine, on the IC pad that is formed during the wafer processes performed before singulation
- Organic contamination caused by bleeding and outgassing of the die attach adhesive. These are seen on both on the substrate or leadframe and the IC metallization
- "Atmospheric" contamination in the air that deposits onto the bond pads. This is largely organic, but there are also traces of inorganics from the atmosphere
- “Diffusion” products resulting from grain boundary migration of underlying metal layers: best known are nickel migration through flash gold and palladium migration.
- Excessive oxide on copper and aluminium surface
Wire bond statistics is mainly affected by the presence of oxides and organics while inorganic contamination affects long-term reliability. Nevertheless, there is an area of overlap.
Even though, in principle, plasma can provide a clean metal surface for wire bonding, it has been shown repeatedly that a metal surface gets contaminated with atmospheric organics within only a few hours of plasma treatment.
Effects of Surface Contamination on Wire Bond Statistics
Contaminants cover the surface to be bonded, preventing from obtaining a good bond. Furthermore, contaminant presence is non-systemic, non-uniformly distributed, appears and disappears and cannot be easily identified, located and quantified.
A bonding process can run for several days and hours without any problems and all of a sudden get out of control for no reason whatsoever. The immediate solution is to turn up ultrasonic power (change of a controlled parameter) with all the accompanying consequences.
The main issues include:
- Non-Stick on Pad (NSOP) Lifts
- Decrease in average shear strength
- Decrease in bonded area
- Decrease in average wire-pull strength
- Cratering and other damage caused by a too aggressive bond process
Influence of Plasma Treatment on Wire Bond Statistics
The benefits of plasma treatment include:
- Reduction or elimination of NSOPs
- Reduction or elimination of lifts
- Increase in bonded area
- Increase in average wire-pull strength
- Increase in average shear strength
- Increase in process window for the bonding process
- Reduction or elimination of cratering and other damage arising from a too aggressive bond process
Plasma and Device Reliability
Failure Modes during Reliability Testing
In spite of the variety seen in combinations of metallizations and bond wires, the failure modes seen during reliability testing are common to all metallurgical systems. In this case, monometallic systems can be considered as a “class within a class” instead of an exception.
The following are the most prominent reasons of failure:
- The failure of poorly formed bonds is more rapid when compared to well-formed bonds
- The ultimate failure of well-formed bonds will occur due to Kirkendall voiding (polymetallic systems), but normally well above the desired package lifetime
- Failures of "well-made bonds", i.e. cases in which the package reliability is compromised, is due to contaminant accelerated voiding (Horsting Effect) and/or corrosion due to contaminants
It is quite obvious that failure is connected to bond pad contamination. There are several other factors beyond the scope of this article, however the key solution is "clean your bond pad".
Intermetallics are formed by most common polymetallic systems. The gold-aluminium combination is the most extensively studied. Intermetallic formation is the first step in bond formation and intermetallics are formed at time of the production process, use of components and burn-in.
Intermetallic formation is one metal getting diffused into another and one of them is exhausted in the process; this finally results in failure. When the bond is contaminant-clean, the time to failure is much more than the component’s design lifetime and so intermetallic formation and the resultant Kirkendall voiding is not a problem. The intermetallics are electrically conductive and strong.
There are problems when the bonding surfaces are not clean but are contaminated with inorganic materials especially halogens. As a result of environmental contamination, wafer processes, halogens may be present or they may be in the molding compound used in the device package.
By a complicated technique known as the Horsting Effect and, like Kirkendall voiding, which is due to diffusion of one metal into the other, the halogens are concentrated into zones at the metal interface where they considerably improve the void formation process but also signify a site where corrosion will occur.
Halogens that are not included in the bond interface but are in contact with the metal surface, especially at the micro-weld perimeter can form electrochemical cells that result in rapid corrosion. A bond that is poorly formed having a low bond area and several micro-welds have a higher perimeter length in comparison to a well-formed bond and are more vulnerable to this corrosion mechanism.
Developing a Plasma Process
In an ideal scenario, plasma which sputters all the inorganic and organic contaminants from the bondpad is used and the perfect bond pad is obtained, which provides excellent "wire bond statistics" and "device reliability". However the challenge begins here.
In certain cases, this solution, which is high power argon direct plasma at relatively low pressure is used. It can be used on certain metal leadframes having power devices. Metal lead frames are not subject to overheating and re-depositing sputtered material becomes a problem only when it is considerable and causes a change in device performance or surface resistivity.
Sputtering of organic contamination has a significant impact on wire bond statistics, but it is much slower than removing it in a "chemical plasma" such as oxygen. Also oxygen plasma has no impact on inorganic contaminants.
So the first thing to consider is whether sputtering plasma or chemical plasma is needed. In most cases the process must include both effects – combination of argon and oxygen. For surfaces that are oxidizable, it is a combination of hydrogen and argon.
Chemical plasmas function best at higher pressures (250 -2000mT) whereas sputtering plasmas function at low pressures (150 - 250mT) in order to increase the mean free path of the energetic ions that perform the sputtering.
As organic contamination is the most common issue, first a higher pressure oxygen plasma is taken and it is made more aggressive by adding argon, lowering pressure and increasing plasma power.
Also, while dealing with considerable amounts of inorganic contamination or if it is observed that cycle time can be reduced and production output increased this aggressive plasma process is used.
For instance, in the case of flash gold increasing power significantly especially for organic substrates can cause excessive sputtering and overheating. So based on the components requiring wire bonding, organic substrate or lead frame, thin or thick gold, robust or sensitive components, large pitch or very fine pitch, the levers of the plasma process are manipulated.
The aim is to define a process window where the cleaning impact is maximised but the drawback of an over-aggressive plasma is avoided. Based on the number of variables, a suitable approach is the Design of Experiment (DoE). However, note that DoEs and SPC work similarly in a defined and predictable "cause and effect," which is normally free of contamination.
Going Into Production
From Process Development to Production
Contamination is defined as almost impossible to locate, identify and quantify, it obeys its own statistics remaining within specific ranges. By setting a plasma process on a restricted sample of parts, the consequence is that the sample average and process average will not match.
While moving from process development to production, the plasma must be re-positioned to agree with the worst case, which may be found in production.
Furthermore, while introducing a plasma process one will be dealing with a clean, normalized surface which is known and the same always. This needs a re-centering of the wire bond process.
In order to accurately determine whether the plasma is functioning perfectly, X-Ray photoelectron spectroscopy (XPS) alone or in combination with Time-of-Flight Single Ion Mass Spectrometry (tOf-SIMS) can offer insights into any kind of contamination.
For production QC, both techniques are not feasible and highly expensive. Further, contamination may not be the same so something that has been observed today was not available yesterday.
The method normally followed in setting a plasma process offers a significant level of security which in most cases is sufficient. The plasma process must clean the bond pads in the "worst case".
Without questioning what the contaminants are, the plasma process can deliver a normalized, clean surface. This surface will provide the wire bonding statistics established during process development.
All operators use wire bonding statistics as an indication of their plasma process functioning up to expectations. Plasma systems are developed to run a highly repeatable process with precise gas flow control, process time and pressure and RF pressure.
Practically it has been observed with the large number of machines running plasma before wire bonding that if the input to the process remains unaltered and the wire bonding process is stable, in that case wire bond statistics is highly stable.
Compatibility and Concerns
There are several doubts about the impact of plasma on the semiconductor devices, whether there are problems with device parameter changes, charging and ESD. One must note that almost every memory device and every microprocessor passes through a direct plasma prior to wire bonding (or in wafer level package processing) without any compromise in reliability or function. Plasma is effective and safe when used appropriately. There are, however, semiconductor device classes that show sensitivity to direct plasmas.
In certain cases a plasma which eliminates the RF field, the presence of energetic and charged particles, and the light inherent in the plasma process, can offer a solution. One such system is Nordson MARCH's "ion-free plasma".
Figure 2. High Throughput Plasma Treatment System (FlexTRAK)
Introducing the right plasma process before wire bonding will result in a cleaner bonding surface. Advantages of a plasma process are high device reliability, high wire bond statistics and prevention of excursions caused by non-systemic influences.
About Nordson MARCH
Nordson MARCH is the global leader in plasma cleaning and treatment equipment, and plasma applications technology. The company has been engaged in the plasma equipment business for more than 25 years and has broad experience in the areas of advanced semiconductor packaging and assembly, wafer level packaging (WLP), printed circuit board manufacturing, life science & medical device assembly, and various large-scale industrial applications.
Nordson MARCH was formed through the merging of March Instruments of Concord, California and Advanced Plasma Systems (APS) of St. Petersburg, Florida. Both March Instruments and APS were acquired in 1999 by Nordson Corporation. Nordson MARCH is a 100% wholly-owned subsidiary of Nordson Corporation.
Nordson Corporation is one of the world's leading producers of precision dispensing equipment that applies adhesives, sealants and coatings to a broad range of consumer and industrial products during manufacturing operations. The company also manufactures technology-based systems for curing and surface treatment processes. Headquartered in Westlake, Ohio, Nordson Corporation has direct operations and sales support offices in more than 30 countries.
Nordson MARCH headquarters are located in Concord, California, which also houses its product development staff, applications engineering, global technical service support and administrative functions. The company has its principal manufacturing locations in Vista, California and St. Petersburg, Florida. International sales, service and plasma applications staff are located around the world supported from facilities in Maastricht, Netherlands; Tokyo, Japan; Shanghai, China; Hsinchu, Taiwan R.O.C., Singapore, Seoul, South Korea and Chennai, India. Each international facility is equipped with a full complement of equipment for customer-specific applications support.
This information has been sourced, reviewed and adapted from materials provided by Nordson MARCH.
For more information on this source, please visit Nordson MARCH.