At the beginning of Selecting a board finish, we developed three lists of requirements for a board finish from the viewpoint of fabricators, assemblers and designers. Before you read further, review from the evidences of the last two parts how HASL and ENIG stack up against the earlier wish-lists. Do you think there is room for improvement? And that these processes will retain dominance in the market?
Take at look at our comments and also review your answer as you continue reading.
You may well have reached the conclusion that, although HASL and ENIG have been with us for some time, a number of pressures are leading to change. In particular, the HASL process is likely to lose market share.
There are a number of drivers for this change:
Jim Reed of Texas Instruments put it this way as early as 1996: “Faced with a variety of issues, such as environmental (CFC and lead elimination), new packaging technologies (finer pitch devices, Chip-On-Board, flip chip, etc), design (co-planarity), cost (global competition), and metallurgical (solder wetting), future PCB surface finishes will most likely need to be different from the current tin-lead finishes”.
The industry has tackled these issues from a wide range of starting points, and devised a remarkable number of alternative finishes – not only are these different in kind, but there are significant variations in the implementations by different process vendors. However, we are starting to reach some consensus, with considerable interest in immersion tin and immersion silver as alternative plated finishes.
Ormerod, quoted in Selecting a board finish, emphasised the fabricator’s perspective; there is the alternative perspective from an assembler that the yield needs to be high. A SMART Group report showed very significant improvements at screen printing and component placement with most surface finishes as against HASL; there was more variability in soldering performance, but here HASL was beaten consistently by ENIG.
When we try and compare materials, which is the final task at the end of this session, we have to keep in mind the need to balance the interests of fabricators, assemblers and ourselves as designers. Inevitably a degree of compromise will be needed.
We are starting off our review by summarising what we know about tin-lead and nickel gold coatings, before looking in more detail at issues involved in organic coatings and alternative plated finishes. As you re-read this material, you might like to try listing against each of the finishes described some of their positive and negative aspects.
Finishes (or coatings) on the copper pattern of boards are applied to guarantee good solderability, even after prolonged storage, and to provide the required reliability during product lifetime. Ideally, a board finish should give extremely flat pads, a long shelf life without solderability deteriorating, and high solderability even after multiple soldering cycles, whilst being environmentally friendly and cost-effective to implement.
There are two main categories of finishes.
Metallic coatings, which form a wettable surface layer on the exposed copper of the PCB, and are chosen to be much less susceptible to oxidation than the copper on which they are deposited. This group is further divided into fusible coatings such as tin-lead which melt at soldering temperatures, in contrast to non-fusible coatings such as gold.
Organic coatings, which are designed to preserve the wettability of the existing surface by preventing oxidation and contamination.
Note that in this section we are considering two distinct types of coating. The first type, and the older of the two, is a relatively thick ‘varnish’ applied to the whole of a tin-lead finished board in order to preserve its wettability during storage. Remnants of the coating are generally removed by cleaning after soldering. [This type of coating must be distinguished from solder resist, a heat resistant organic coating, with openings at the solder lands, which is intentionally non-wettable by solder.]
The second type of organic coating, and by far the most common nowadays, is a vanishingly thin protective coating, applied just to the bare copper, which disappears during the actual soldering process.
There are many proprietary organic coating materials. As they must not obstruct wetting and the flow of solder to the soldering areas, they must either dissolve into or mix with the flux to a sufficient extent. A typical coating is based on modified colophony (the ‘rosin’ used to make fluxes), usually with an acrylate or epoxy ‘film former’ to prevents the colophony from crumbling, which would cause contamination of tooling such as punches or dies. However, on boards with plated through-holes, the use of a film former is less desirable because its presence in the holes tends to retard the filling of these holes with solder.
Such protective coats are applied by roller coating or dipping in a dilute solution, taking care that the surface is clean. The optimum thickness for such coatings is of the order of 0.5mm. This thickness provides adequate protection against fingerprints without hindering the action of the flux. Coating thickness can be determined by weighing a metallic test plate before and after coating, provided that the density of the dried solids of the coating solution is known.
The protective coat will in most cases reduce the wettability slightly, but, in the long term, the wettability is better retained. The qualitative effects of storage time on the wettability of boards, for coated, non-coated, and non-coated but packed conditions are shown schematically in Figure 1.
Reasonably good protection is provided if the boards are sealed in polyethylene bags together with silica gel (although polyethylene is not in fact gas-tight). Best results are obtained with laminated bags of polyethylene-aluminium-polyester, but in any case the wettability of the enclosed boards will slowly decrease. The maximum long-term storage life is debatable, but should not exceed two years.
Bare copper is only solderable when very fresh, and needs protection from oxidation. A number of organic ‘anti-tarnish’ finishes, or ‘corrosion inhibitors’, based on imidazole or triazole compounds, have been found to be effective against undesirable surface reactions on copper and copper alloys. Most are only a few atoms thick and need to be deposited on very clean copper.
OSP coatings come in many different formulations, although with similar basic chemistry. The differences include variations in coating thickness, the use of a water base rather than a solvent base, and whether some complex reaction products are involved. Trade names you will come across are ‘Entek’ (from Enthone) and ‘Cuprotec’ (from Shipley).
Key issues in the specification and process control for these coatings are:
A typical application process consists of several pre-cleaning and conditioning stages, followed by immersion in a dilute (0.1% by weight) solution of the OSP for 1–3 minutes. This forms a chemical layer on the surface of the copper 2–10nm thick, depending on the formulation.
Figure 2 shows a conveyorised horizontal implementation, but the process can also be used in a vertical mode, dipping the board successively into the solutions. The most critical part of the process is rinsing between the conditioner (‘topography enhancement’) and OSP stages – surface cleanliness is a concern with all OSP coatings, because contaminants can appreciably reduce the solderability of the board.
Proper control of the process is necessary to ensure a uniform and continuous coating of the correct thickness. The major variables affecting this are the acidity of the solution, the immersion time and temperature, and the concentration of the OSP chemicals. [It has been reported that, with some formulations, even slight over-treatment may reduce the wettability of the copper surface to such an extent that the material becomes unfit to use – the remedy being worse than the disease!]
How do these OSPs meet the requirements for a board finish?
Review your answer as you continue reading
Advantages of OSP compared with HASL are that:
From the designer’s point of view, OSPs give a very cost-effective and flat finish which is suitable for all types of components. OSPs can also be used in applications where the boards need copper pads, yet require other features plated with gold, silver, tin or solder. Water-soluble surface treatment agents based on some imidazoles are able to bond selectively to the copper, providing it with protection, without adversely affecting other metals present, or leaving any film on them.
The shelf life of OSPs is variously quoted, but is of the order of 6-12 months, depending on the environment. It is difficult to produce an adequate accelerated test1 for OSPs, because ageing in steam, or annealing at high temperature, oxidise the copper and degrade the organic coating in a way which is not representative of life. There is no tin-copper intermetallic compound to grow.
1 OSPs have been tested at 40°C/90%RH for 1000 hours to simulate two years of storage; others have tested at 65°C/95%RH for 24 hours and equated this to a shelf life of 12 months. Both predictions assume a model of the deterioration mechanism which is not justified!
Regardless of which finish is used on a board, the pad may not always be wetted by solder all the way up to the edges and corners. This is most commonly found when using no-clean fluxes with low activity. It is very difficult to spot this effect with solder-coated surfaces, but less than total wetting is readily visible with OSPs, because exposed copper can be seen at the perimeter of surface mount pads and through-hole annular rings. There has been concern that such exposed copper could promote copper corrosion, or reduce the SIR of the assembled board under conditions of elevated temperature, humidity and voltage bias. Results, however, show no evidence to support this concern.
Lucent reported2 that the biggest issue in using an OSP coating was not so much the coating itself, but the assembler’s inability to be sure that there had not been contaminants on the pad before OSP application. Pads that did not wet at all could be identified, and the joints scraped and repaired, but the possibility that there still remained partially-contaminated pads that appeared to have wet, gave concerns about joint integrity and consequent unreliability.
2 Robert Furrow (Lucent Technologies) postings to IPC TechNet 13 February 2001 and 18 April 2002
Whilst problems were only occasionally experienced, and solderability was poor on just a few pads per board, the cost of reject assemblies could wipe out the benefits of using OSPs. On most occasions, the problems were found not to be the OSP process, but rather things like unseen solder mask residues or incomplete removal of the tin etch resist. However, at least once, boards were shipped without the OSP coating applied.
Lucent also found quality issues when using OSP for via in pad (VIP) designs, where residues or water remaining in the small holes could compromise the OSP coating. Cost and quality issues have convinced several other companies that OSP is non-preferred, and that better results are obtained from a metal finish.Advice given3 is that
3 Postings to IPC TechNet on 13 February 2001 by Robert Furrow (Lucent Technologies) and Darrel Therriault (N-Cube)
One option which gets round these particular problems, although at the expense of more complex processing, is of course to use selective plating, limiting the use of ENIG to test points and ground frames, and protecting pads with an OSP. This prevents any possibility of black pad failures of the solder joints.
Contrast two different types of organic coatings used to help boards retain their wettability during storage. What are the issues that affect the application of thin ‘anti-tarnish’ coatings?
One of the problems with the OSP type of finish is that the copper surface is not flat on a microscopic scale: there are many fissures, predominantly at grain boundaries, where these finishes are not effective. Plated finishes are therefore highly attractive, although inevitably more expensive.
Tin can be plated to give a very wide range of results depending on the conditions: matt tin is a very pure material: deposited from an alkaline solution without brighteners, oxide films on the surface can easily be penetrated by probes. Bright acid tin, deposited from baths containing small quantities of organic materials, has greater hardness and wear resistance, and is cosmetically more attractive.
However, it has a much higher inherent stress, so is very subject to tin whiskers (see below).
A great deal of work has gone into producing tin finishes that will be a cost effective replacement for HASL. You will find mention of both grey tin and white tin in the literature. Both are flat uniform finishes, but the difference is more than colour: grey tin has large orthorhombic crystals; white tin has finer hexagonal crystals, giving a denser structure which is claimed to be more resistant to surface oxidation. The most commonly-found trade names for a tin finish are ‘Omikron white tin’ (Florida Cirtech Inc.) and ‘Stannatech’ (Atotech).
Typical process steps for immersion tin are:
4 Johal 2000 states that for a 4 hour exposure at 155°C (corresponding to 1 year of storage life), and with a thickness of <0.8µm of pure tin, the wetting angle drastically decreases compared with the fresh surface, indicating decreased solderability. However, the conditioning step greatly reduces the rate of deterioration
Immersion tin deposition is not a normal displacement process, although it is self-limiting and depends on copper from the foil being exchanged with tin, but can only take place in the presence of thiourea. Unfortunately, thiourea is a suspected carcinogen, with health and safety implications, so the process needs to be tightly controlled. [Another process issue is that the dissolved copper has to be removed from the bath.]
As a designer, you may well choose to use immersion tin in preference to HASL and other alternatives for applications which involve press-fit connectors, as the coating exhibits a useful degree of inherent lubrication, aiding the insertion process.
At low temperatures, tin deposits may suffer from what is called ‘tin pest’, which describes the modification of tin into a non-adherent powdery grey form known as ‘grey tin. This phase transformation can occur at temperatures between −40°C and 13°C, but typically only occurs below 0°C. This problem may be greatly reduced by alloying with very small quantities of antimony, bismuth or lead: an addition of 0.25% antimony is often used as a preventative measure.
Whiskering is a more serious concern than tin pest: whiskers are spontaneous filamentary growths which can occur on a variety of metals, including tin, cadmium, zinc, iron and nickel. They are single crystal structures that are reported to reach lengths of up to 9mm with diameters up to 5µm, and to be able to carry 10mA of current. Whiskering becomes a hazard when the whiskers become long enough to bridge across leads. In high voltage, high current circuits, any shorts caused by tin whiskers are quickly burnt out, and may not pose as serious a threat as in low power circuits, where they will cause intermittent failures.
“Tin whiskers are a fascinating and confounding subject” was Jay Brusse’s5 conclusion after four years reviewing the literature and carrying out research. Many conflicting test results have been reported, but the consensus is that whiskering is a result of stresses within the plated deposit. However:
5 Jay Brusse (QSS Group, Inc. at NASA Goddard) posting to IPC TechNet on 20 June 2002. The NASA public web site (http://nepp.nasa.gov/whisker) has a useful collection of reference materials and photographs.
Note that tin whiskers have an incubation period, so that a coating which has appeared whisker-free can develop the problem days, months or even years later. It is this incubation period that distinguishes whiskers from plating nodules, which may be roughly similar in appearance but will be present on the surface immediately after plating.
Whiskers are sometimes confused with electromigration, about which you will learn in the Failure mechanisms Unit, but have a number of key differences:
Tin whiskers will grow from almost all surfaces containing tin, depending on the environmental conditions.
6 It normally recommended that bright surface tin plated should be avoided because the co-deposited organics used to make the surface level and bright increase the internal stress, but there are drawbacks:
- A bright surface finish is cosmetically very attractive
- There is no easy non-destructive test for the co-deposited compounds
- It is not possible to distinguish visually between coatings which are naturally bright because of organics and those which have been reflowed
The key to reducing or eliminating whiskers is not to add lead (a widespread misconception) but to have a combination of the correct plating additives, good selection of materials, and stringent process controls and operating parameters.
Fortunately, reflowing the surface, and the high temperatures during board assembly, can alleviate stress in the finish and reduce the potential for whiskering.
Immersion tin has been generally favoured in Europe and North America, as against immersion silver (also referred to as ‘IAg’), which was mainly used in Asia. However, immersion silver is gaining popularity, partly because it does not need hazardous or toxic chemicals.
Unlike nickel, the silver layer (0.07−0.25µm thick) totally dissolves in the solder joint, leaving a homogenous tin-lead-silver alloy joint directly onto the copper surface. The procedure is similar to immersion tin: compared with ENIG, both processes are easier to control, have short process times (high throughput), and operate at moderate temperatures. [The coating thickness depends on immersion time and coating bath temperature, but the required thickness of silver can be deposited in around 3 minutes at a bath temperature of 50°C].
The competing systems which you may encounter are ‘Sterling’ immersion silver from MacDermid and the AlphaLEVEL process from Enthone. With some small differences between vendors, the process can be carried out in either vertical (batch) or horizontal (continuous) equipment and consists of a dual pre-clean, a pre-dip and a silver bath:
This last stage has the key process feature, which is the simultaneous deposition of an organic inhibitor. This both prevents the silver from migrating, and protects the silver from corrosion. Lucent reported7 that their experience in volume with immersion silver as a surface finish had been excellent and it had become the preferred finish at some of their manufacturing locations. The finish retained excellent solderability for over a year (if stored properly), handled multiple thermal excursions well and had superior hole fill characteristics compared to OSPs. Cost was intermediate between OSP and HASL finishes.
7 Robert Furrow posting to IPC TechNet 6 November 2000
They reported two relatively minor assembly issues:
There are some practical issues about which the designer should be informed:
Though immersion silver appears attractive, there is a major obstacle with a number of industry sectors because of the attitude of Underwriters Laboratory, a key US approvals organisation. Because silver which has been plated without the co-deposited organic compound is very prone to migration, there are different tests for silver products than for other surface finishes, and these translate into spacing restrictions that make the designer’s life very difficult.
This whole issue is currently up for debate, and some more rational approach is likely, reflecting the thinness of the silver layer involved and the fact that this is totally dissolved in the solder.
Which are the main alternative finishes to HASL, ENIG and OSP? What processes are involved in their use? Are there any potential reliability concerns for any of these coatings?
As a plated ‘gold finger’ on a circuit board8, a flash plating of soft gold or immersion gold would have a life only in the range of 1–5 insertions/extractions. However, adding small quantities of nickel or cobalt to the electrolytic plating process produces so called ‘hard gold’, which exhibits much lower wear in sliding operations. Using an under plate of nickel also improves wear resistance.
8 This section has benefited from postings to IPC TechNet on 2 August 2002 by Earl Moon and Don Vischulis
The number of cycles which fingers will survive depends on the mechanical design of the female connector contacts (geometry and mating pressure), the surface roughness and hardness of the mating surfaces, and the plating thickness. These variables are hard to characterize, so it is difficult to derive a definite relationship9 between their values and contact life. However, it has been found that increasing the gold and nickel thicknesses provides significant increases in connector life: typical values used are 1.5–2.5µm gold over 3–5µm nickel.
9 By comparison, connector manufacturers have insertion cycle relationships established for their products because they can control all of the variables
Tin-lead contacts require a different connector design because of the potential for surface oxidation. The most common solution is to have a hard wiping action, but this quickly cuts through the plating, limiting the number of insertions before damage is done to the underlying metal.
Some card edge connectors used to have ‘knife-edge’ connectors which were tin-lead plated and designed to be used with tin-lead plated board edge connections. This provided a ‘gas/air tight’ seal and gave a sufficient number of insertion/extraction cycles over flash coatings.
Note that the types of plating used for edge connection areas are likely to produce unreliable reflow soldered joints. The combination of maximum gold thickness and minimum solder volume should result in a joint containing not more than 1.4–1.8% of gold by volume (3–4% by weight).
Whilst separate switches give higher reliability, many products use keypad areas on the board which are shorted by pads of conductive elastomer. The on-resistance of each of these switches is relatively high, of the order of tens of ohms but this is generally not an issue, because the switches lead to high impedance inputs.
The key-pad areas on the board are most frequently ENIG, but immersion tin and silver are also used. Although their performance and image definition are not as good, screen-printed pads of carbon and silver-loaded epoxy are to be found on some product.
Apart from areas intended for soldering or wire bonding, what other coatings might you be expected to find on some boards?
HASL and ENIG are not the only finishes on the street!
When reviewing the options, you should realise the critical importance of working with both the end-user and the board fabricator to choose the most appropriate surface finish for the application:
The final finish needs to be selected based on design, fabrication issues, solder joint integrity/reliability, availability, and cost.
George Wenger (Celiant Corporation) posting to IPC TechNet on 19 April 2002
What we would like you to do is to draw up a table comparing the materials: this matrix should have the available materials as the heading for each column and the parameters which are important labelling the rows. For many of the entries you will have to make judgements of the sort ‘high’, ‘medium’, and ‘low’ – in other cases you may be able to put figures.
In order to populate your table, you should start by reading what we have to say about each coating and then look for other comments. For example, you could try searching Google for ‘white tin’ +PCB +process, which will give you a limited number of results, most of which relate fairly accurately to our requirement.
Compare the table that you populate against these.
You should not expect to find that all board suppliers are in total agreement. Part of the reason for this is a preference for certain finishes, based on the company’s experience, the level of local technical support and commercial factors. When you are selecting a board fabricator, then you need to be sure that the finish you ask for is within their repertoire, and is one that they can control – they should be aware of the quality issues and compromises involved and be able to advise you.