Smart Group Annual Conference 2011 The Oxfordshire, Thame, England 5 and 6 October 2011

Citation

Ling, J. (2012), "Smart Group Annual Conference 2011 The Oxfordshire, Thame, England 5 and 6 October 2011", Soldering & Surface Mount Technology, Vol. 24 No. 1. https://doi.org/10.1108/ssmt.2012.21924aaa.028

Smart Group Annual Conference 2011 The Oxfordshire, Thame, England 5 and 6 October 2011

Article Type: Exhibitions and conferences From: Soldering & Surface Mount Technology, Volume 24, Issue 1

SMART Group Chairman Keith Bryant, in his customary ebullient style, encouraged everyone to move forwards (or was that onwards?) and thanked everyone for coming along, especially those who have flown a very long way. A total of 17 exhibitors supported the event, which was, as usual, very well attended; not only by the SMART group stalwarts, but by many from the industries who are related to, and rely upon, the surface mount business (Figures 6 and 7).

Dave Hillman of Rockwell Collins shared his debate on bonding of BGA – should it be underfill, edge bond, or corner bond or what? It is all about reliability. Is underfill really necessary? Oh yes it is. It spreads stresses across the entire under surface and improves drop and vibration reliability. It also evens out CTE stress, and underfill adds robustness to help the product survive. A re-workable underfill has been hard to find at Rockwell Collins, and Dave illustrated the test vehicle that had been used over the years, it is a 16-layer 0.086 in. ENIG PCB with 0.5 mm 56, 0.8 mm 288 and 1.0 mm 256 I/O BGAs. Materials for edge or corner bond came from Henkel Loctite, Dymax, Zymet, Namics, and Scotchweld. Edge bonding is an art, as ideally only 50 per cent of the edge should be covered, 25 per cent of the edge length from each corner. Thermal cycle testing accords with IPC-9701, but at Rockwell Collins 2,000 cycles are standard. No underfill parts failed, but all the other bonding techniques failed below 1,000 cycles. Namics and Zymet products looked the best, but they too failed below 1,000 cycles. However, there was no thermal cycle advantage indicated between edge-bond and corner-bond. Whilst edge-bonding and/or corner-bonding is quicker than underfill, much depends on the material used, and how it flows out. All products are tin/lead, as end use is aimed at military applications. Non-reworkable underfills at Rockwell Collins work equally well on flip-chip as they do on BGAs, added Dave. The next stage is to look at drop shock testing, and vibration testing. Are underfills really reliable? Take 15-20 years, Dave said, as a yardstick of the time in service that underfilled components are expected to survive, and they do.

Figure 6 SMART speakers: day 1

Figure 7 Smart speakers

Working with MBDA Missile Systems is Charles Cawthorne, who spoke about the assembly of BGA packages from an end-user perspective. MBDA is a pan-European company with interesting ancestry, and the result of several marriages of convenience and mutual benefit. The development of missile systems is now very much shorter than it was, down to two years from ten. They use 1738 ball BGA devices 1.0 mm pitch amongst the range, and Charles described the inspection methods used for BGAs, which included endoscopy, X-ray, and inspection is 100 per cent initially. Inspection using X-ray initially is visual, looking via oblique angles for evidence of incomplete reflow, and an assessment of voiding. Incomplete reflow defects were clearly seen by this process of inspection, as well as solder draining on vias. Key aspects of BGA inspection include temperature profiling during reflow, voiding, and lead free. Other factors include PCB design and fabrication, solder resist design for BGAs in particular, and the use of lead-free BGAs in a tin-lead soldering process, etc. PCB manufacturing quality standards are now much advanced, and the preferred non-solder mask defined (NSMD) pads are available, and the precision needed for imaging on BGA boards is now reliable. Using lead-free BGAs in tin-lead assembly, and vice versa, is prohibited at MBDA for good reasons, and where conformal coating is used it is important that this does not penetrate under the device, and they have a sealant which can be applied to prevent this. They use underfill on a case-by-case basis, depending upon vibration levels in service, the air launching of missiles has the highest here. Sn/Pb BGAs can be successfully assembled using a tin-lead process, and used in high-reliability applications, but increasing success is being obtained with lead-free BGAs using a SAC 305 solder.

Measuring the reliability risk of PCB pad craters was the subject of a paper given by Karman Iqbal of Nordson Dage working with Universal Instruments. Pad cratering is the mechanical failure of a laminate under a pad caused by overstress. Manufacturing, handling and test can induce small cracks that do not show up in electrical test. Exposure to mild service stresses, these cracks propagate and cause premature failure. They use bond testing of individual pads to assess the risk of materials and put limits on manufacturing stresses. The failure is mechanical; so testing and/or inspection methods have to be specific. Angled pad testing is one, a pull test at 30°C, but measuring strength alone does not prove long-term reliability, so they have a pull test using heat which duplicates the cycling stresses involved. Fatigue failure testing was shown, here they apply 40-50 per cent of the pad strength as a fatigue load, and then kept pulling between 100 and 1,000 cycles to failure. They can also stress test to indicate likely failure using the Weibull method.

Kingston Computer Consultancy is co-ordinating a project entitled μBGA, funded by the EU under FP7, which is about creating IP on solder-ball production within Europe for members, including the SMART Group. The project manager is Simon-Peter Santospirito who informed us that solder ball production is essential with an increasing number of packages being made, including MEMS, SiPs, and having μBGA ball production in Europe would make the region independent of other sources. Non-European supply chain reliance is not attractive to the commissioners in Brussels and Strasbourg, so the project is looking at manufacture, placement, and application within Europe. Funding will be available for manufacture, allowing competition with the USA, China, and Japan, and by manufacture he indicated some 18-20,000 balls/s as a typical production speed, or 5 million per day. Solder ball size is aimed at 50 μm, production will be by ultrasonic solder jettison, a process which goes back to about 1879, but which was more properly developed in 1952. How a jet stream of solder is broken up into balls was described, jetting the spheres into an oxygen-free environment showed that a round ball is formed, but is usually co-joined in a dumbbell shape with another ball. By pulsing at 10 or 17 kHz, individual balls can be formed, so they then looked at electrostatic charging, and acoustic separation, and the latter looked very effective. The subject is complex; there is much work yet to be done, but it is a most interesting topic.

Dave Shaw had come down from Birmingham where he works for aero engine controls (AEC) to talk about cleaning, specifically the cleaning of electronic assemblies for aerospace use. Many low- to mid-volume manufacturers have a choice of multi-stage aqueous or semi-aqueous cleaning; aqueous is mainly in-line air systems, semi-aqueous uses spray under immersion batch systems, but environmental pressures have led to a change in thinking. AEC use aqueous spray in a single chamber, and an enclosed trichloroethylene system for cleaning reflowed PCBs. Both have advantages and disadvantages; for semi-aqueous single chamber systems, the latter including long process times, the need for batch processing, and residual rinse carry-over. However, on the up-side is cost-effectiveness, and the equipment occupies a small footprint. Solvent cleaning is quick, but it is an environmental problem, and is increasingly expensive. So, all in all, AEC are now looking at using aqueous cleaning methods for de-fluxing LCC and BGA assemblies after reflow. Dave detailed the work being done in this regard; the wash chemistry is the key, a balance between flux removal against board degradation, and part marking degradation. Process stability is critical, and it is a very fine balancing act.

The other day Dave came on again to end the day talking about voids. Not voids in our knowledge, as these have been filled by SMART group conferences, but macro voids, planar micro voids, shrinkage voids, micro-via voids, pinhole voids, and the Brontë-esque Kirkenhall voids. Dave Hillman asked if macro voids were good or bad. Would he be here if they were good? The effect of voids on solder joint reliability has been studied extensively, and Dave gave us a look at the work they have done on this. Using an ENIG fished FR4 board with all the bells and whistles fitted, they had 230,000 joints. The number of voids? 10.0.004 per cent! So, back to the drawing board, or TechNet, who made a large number of useful suggestions about creating voids? Taking 15 test vehicles, and using a water-soluble solder paste, tuning off the nitrogen, with contaminated pads, they went ahead and behold – voids. Dave has now written a paper entitled “The last will testament of voids”. As he reckons that they are now history. It is not the void that causes the cracking; it is where the void is. “25% of or less of ball X-Ray Image Area” says IPC, and that is based on data coming from Rockwell Collins. Confusing area and diameter is easy, but dangerous, and thus Dave proceeded to provide some fascinating details and magnified cross-sections showing defects, and how they had been caused –, i.e. blocked stencil, and it showed that a void with an area of 27 per cent did cause cracking. It is not the size of the void, it is the location. His conclusion is that you must not have voids greater than 35 per cent of the ball X-ray image anywhere in the BGA.

The other half of the Rockwell Collins team, Doug Pauls, set the tone on a damp second day by talking about humidity, the role of conformal coatings in protecting components, and the problem areas between these two topics. If you conformally coat a board then there is a risk of solvent entrapment. Conformal coatings do not reach their full potential for some time, maybe months, so their properties change with time. It is the CTE values of the coatings that can have a profound effect, and they vary depending on method of curing, or whether it is a one-pack, two-pack or 100 per cent solids coating. To prove the point they ran tests with two well-known conformal coatings, Humiseal 1B31, a single-pack acrylic, and 1H20AR1, a single-pack aqueous-based acrylic-urethane. The test environment was 55 to 125°C with 15-min dwell time, and cycles depending upon end use, aviation needing 2,000 cycles. The board was an eight-layer m/l board conforming to IPC standards, and the exhaustive test processing was described. The tests showed that the aqueous acrylic-urethane had no failures on BGAs, QFPs, and QFNs. The 1B31 coating showed that CTE had resulted in deformed solder balls in the BGAs, but with lower levels of solder joint failure for the SOICs, QFPs, and QFNs. So, if you do not want to have conformal coating as an underfill, you could use Parylene, or a proper underfill if the area under the component must be sealed off. But reworking has to be considered. So, back to edge bonding. See Dave Hillman above.

The very nice Mike Bixenman from Kyzen talked about IPC-CH-65B cleaning guidelines. He wondered if we had a couple of days spare so that he could be thorough, but sadly we had to settle for 45 min. How clean is clean? It depends! Cleaning is in fact a process, and what needs to be cleaned is now so small in size that the discipline of cleaning is now stricter than ever before. Cleaning helps to avoid failure, and whilst the human body may, inevitably fail, the advances in the medical field can prolong life with implants, and even these need cleaning. If cleaning is a cost factor, what might the cost be of not cleaning? Looking at PCB assemblies, what are the risks from post solder residues? Ionic residues are a known enemy of solder resist application, and definition of solder areas in a world of miniaturisation is now a vital aspect. How ionic contamination is created was explained, with electrons and ions attracting each other in varying degrees of strength, in a sort of chemistry dating agency. He also counselled against eating popcorn whilst handling circuit boards. Chloride is not a good element to have around. His use of the word soil was interesting – there are three categories, polar or ionic, then non-polar and non-ionic, and particulate. Salts, cream and dust, in simple terms. But the need to know what types of soil are to be cleaned is vitally important, so that the right cleaning agent, and application technique can be employed. Mike’s awesome knowledge of cleaning made it obvious that a map to good cleaning practice is readily available to all.

Vladimír Sítko came over from PBT in Brno, where they have, for the last 20 years, been making cleaning machines for electronics. Measuring the efficiency of a cleaning process relies upon simulation, in which all methods try to evaluate real assembly. He felt that there were four major factors – the cleaned subject, the chemistry used, the machine parameters, along with temperature and time. The test substrate used is ceramic chips on glass, with gaps of 25/35 μm 300 μm apart. The preliminary process involves glass surface activation, then saturation with flux, a visual evaluation, or inspection, then automatic optical inspection, and they monitor the situation after each cleaning stage. Cleaning in the chambers can be modified by changing the nozzles, likewise other process parameters such as flow, pressure, temperature and time, and they can map the entire chamber to show optimisation. The range of machinery available varies between models for stencil cleaning and for cleaning BGA boards, as well as boards with FPGA components and the like.

Doug Pauls had earned his lunch. He came in front of us as the avuncular figure of Rockwell Collins once again to talk about three case studies on conformal coatings. These were of course entirely theoretical, he hastened to add. The first was a study on the effects of organosilane primers mixed with water-based conformal coating. The second was about acrylic conformal coating cracking when subjected to highly accelerated stress screening (HASS) testing, and the last one was a chemical analysis of lots of conformal coating.

In the first case study “Environmental stress screening” had shown up small blisters which had been found in the conformal coating, intermittently, on the top of components. Why? A tortuous trail of tests was undertaken, and the chemical detective agency that is Mr Pauls’ department spent this summer seeking the answer to why there are problems with products for one customer, and not with the same products with another customer. It was all to do with residual stress and thermal stress. The second one concerned cracking in the coating in HASS testing. This is more of a thermal shock than a thermal cycle. The conclusions were that the cleaning method was a major contributor – the problem was much less pronounced when wash temperatures were higher, and when a pre-rinse to knock off the water soluble mask was implemented. Furthermore, the conformal coating was not being allowed to fully air dry before curing, so there were lots of bubbles. The third case study was resolved by the employment of the very bright young man who knew all about gas chromatography mass spectroscopy (GCMS) and how to use the equipment.

Martin Wickham from the National Physical Laboratory in Teddington also spoke about conformal coatings, where they had run a series of tests on adhesion with various substrates. Tin whiskers penetrating out from under coatings were found extensively in areas where there were coverage issues. Most coatings evaluated exhibited whisker growth from under coatings at plate edges, and coating coverage was an issue with a majority of the coatings evaluated. Where both plates of the samples were coated, six of the coatings evaluated did not exhibit electrical shorts at any time during the 150 days of testing. These were one acrylic, one paraxylene, and two polyurethane and two silicone coatings. Conformal coatings do inhibit the formation of whiskers. However, coating coverage is a significant issue and you get a generally improved performance where adjacent surfaces are both coated Martin informed the delegates that the on-going work at NPL on tin whiskers included – Sn whisker mitigation using conformal coatings – creating a databank of Sn whiskers in commercial components – Electrostatic attraction of whiskers – Whisker contact resistance measurement – whisker oxide thickness characterisation – current carrying capacity of whiskers – whisker oxide breakdown – electrical resistance of whiskers, whisker growth under electric fields, and whisker mitigation by coatings using an SOIC component test vehicle.

Graham Wilson of the Indium Corporation was keen to help companies by providing bucket loads of chemistry at entirely reasonable prices. Their customers have used conformal coatings for various reasons, one being because “they always have” a sort of belt and braces approach to the need for protection from harsh environments, moisture ingress, etc. With single-pack systems test vehicles are similar IPC test panels, he illustrated, for SIR and BONO tests. Whilst peel/adhesion is important, it seems that there are no international standards for coated residues over no-clean flux residues. So they ran some tests with other companies, following B24 SIR, and using various resin-based coatings – polyacrylate, modified acrylate, modified polyurethane, and co-polymerised polyurethane and polyacrylate. The acrylic-based coatings were the best, and tested for seven days at 85 °C 85 per cent Rh, and the SIRs under TM B24 all passed. Coating over no-clean flux residues is being successfully done.

Mike Konrad of Aqueous Technologies wondered why no-clean fluxes were the best selling products in an industry that spends a fortune cleaning them. About 55 per cent of companies clean no-clean. Before 1989, and the Montreal Protocol, most assemblies were cleaned using Freon or 1:1:1 Trichloroethylene, but then along came Multicore who launched a no-clean flux. Over 75 per cent of assembly is going to commercial products, but the military and medical sectors just carried on cleaning anyway. We are obsessed with cleanliness, and yet today we make only token acknowledgement of cleaning boards. No clean fluxes have problems; one is electro migration failure, and another is electrical leakage failure. But its not just flux, there is a whole host of reasons why cleaning is essential. There many factors which are relevant to cleaning.

During board manufacture there are etch residues, development chemistry, the quality of water used in rinses for both inner and outer layers, and HASL fluids; during component manufacture there are plating bath resides, water quality rinses, deflashing chemicals, and mould release agents; during assembly there is solder paste, flux, cored solder, fluxes for re-work or repair, cleaning chemicals, and pre-tinning flux residues. Phew. However, there are only five stages to the successful cleaning of an assembly:

1. 1.

   wash;

2. 2.

   rinse;

3. 3.

   verification;

4. 4.

   dry; and

5. 5.

   SPC.

Stage (1) is flux removal, where contact, heat and mechanical energy are needed. Rinsing is a function of fluid delivery systems – namely coherent sprays or jet/fan sprays; then comes hot air knives for drying, and finally documentation, and SPC data capturing is available on most processing machinery nowadays. How do you measure cleanliness? Ion Chromatography. How clean is clean? Zero is clean.

Attending a SMART Group Conference is like being at an annual reunion of distinguished alumni from a rather dedicated university chapter. There is a distinctive spirit to the event, a combination of experience and expertise that is reassuring given that their prime aim as an association is to ensure that things that really matter do not become detached from where they should be attached.

In an industry that has, over the years, had to cross several minefields laid by naïve people in Brussels, such field craft is totally admirable, and their strategic and tactical leadership is beyond reproach.

As their chairman said at the very beginning “Forwards!”

John Ling