SMART Group Seminar: Harsh Environments & Electronics
A full-house international audience assembled at the National Physical Laboratory in Teddington, London, England, July 2, 2014, for SMART Group’s seminar on Electronics in Harsh Environments, welcomed by Dr. Chris Hunt, leader of the electronics interconnection team at NPL, the national measurement standards laboratory for the United Kingdom, and the largest applied physics organisation in the UK.
Chair of SMART Group Technical Committee Sue Knight, from STI, gave the first presentation, an introduction from her perspective as a high-end electronics manufacturing engineer. Harsh operating environments were commonplace: She gave examples in telecommunications, oil exploration and power generation, aerospace instrumentation, sensors, motors and actuators, and industrial controllers and sensors. And although high temperature probably represented the area of most concern in maintaining the integrity of substrates, components and interconnections, stresses generally occurred in combination: Typical additional factors being mechanical shock and vibration as well as humidity and chemistry.
As an illustration, Knight described the extreme conditions endured by the electronics mounted in an intelligent, directable drill bit used in oil exploration. Historically, sensing electronics had been mounted at least 10 metres back from the cutting tip of the drill. Presently, to get precise, real-time feedback to monitor performance and improve positional control, complex electronics assemblies were mounted very close to the actual drill bit. Pressures of 25,000 psi and temperatures in excess of 150°C were typical, even though the assemblies were built with commercial-specification components. Current challenges were to operate at pressures up to 30,000 psi and temperatures approaching 230°C, under conditions of extreme shock and vibration as well as the presence of dirt, steam, and slurry. The cost of failure was significant, so reliability was paramount and the manufacturer was responsible for validating that the tool would perform to expectations, modelling and verifying build quality and sensor and tool operation under extremes of temperature and vibration.
What was the relevance to everyday electronics? Knight explained how research and development generated at high level could be subsequently applied to benefit the functionality and reliability of commercially-available equipment. And how could the challenges be overcome? “I have just outlined some of the issues--the other presenters will explain how to solve them,” she said.
Alun Morgan, Isola Group’s director of OEM Marketing and a leading authority on laminate properties and applications, has a particular aptitude for explaining complex technology in a clear and understandable style. Invited to make a presentation on base materials for harsh environments, he admitted being tempted to save time by offering two words of advice--“use polyimide”--and sitting down again! But polyimide was not necessarily a universal solution: If properties were correctly understood and materials intelligently selected to suit the application, many cost-effective and in certain cases technically superior alternatives existed.
Exploring the realities of the thermal endurance of laminates, in terms of operating temperature, decomposition temperature, and flame retardant impact, then considering thermal cycling in terms of temperature ranges and numbers of cycles, and the influence of glass transition temperature on material performance, Morgan began by defining laminate terminology: Glass transition temperature (Tg) was the point at which a polymer changed from a glassy solid to a rubbery state, and several properties changed as the Tg was exceeded, including the coefficient of thermal expansion (CTE), which was much higher above Tg and much more evident in the non-reinforced z-axis. Modulus also decreased significantly as Tg was exceeded. The higher temperatures of lead-free assembly resulted in more total expansion for a given material and several mature lead-free-compatible materials incorporated inorganic fillers to reduce CTE values. Decomposition temperature (Td) was determined by measuring weight loss from resin as a function of temperature and was typically defined as the point at which 5% of the original resin mass was lost to decomposition. Resin decomposition could result in adhesion loss and delamination. Peak temperatures in lead-free assembly could reach onset points of decomposition. A high Td did not guarantee performance and conversely, a low Td by the 5% rule was not necessarily bad if the onset temperature was high enough. Moisture absorption could be assessed by a water soak or a pressure and humidity test. Because the vapour pressure of water was much higher at lead-free assembly temperatures, absorbed moisture could volatilize during thermal and cause voiding or delamination. PCBs that initially passed lead-free assembly testing might exhibit defects after storage in an uncontrolled environment as a result of moisture absorption. Time to delamination was related to decomposition temperature and adhesion between material components, and measured as the time for delamination to occur at a specific temperature, generally 260°C (T260) or 288°C (T288). Thermal expansion and moisture absorption could also influence results and in multilayer PCBs, the bonding treatment of the internal copper surfaces was also critical.
Morgan explained the basic chemistry of resins: How thermosets differed from thermoplastics, and how phenolic crosslinking agents had improved the thermal stability of FR4 resins and made them more compatible with lead-free processing. Next-generation FR4 materials had completely different curing systems, and offered 200°C Tg and 370°C Td, which approached the properties of polyimides.
Flame retardants tended to have an adverse on laminate Td, primarily because they were designed to decompose at high temperatures and release flame-suppressing species. For example, UL94V0 rated polyimide had a Td of 390°C whereas its UL94HB equivalent had Td of 416°C. For this reason, PCBs for the U.S. military were all specified no-flame-retardant polyimide.
Returning to an examination of the effects of glass transition temperature on thermal cycling performance, he demonstrated how Tg dramatically influenced thermal expansion and compared the characteristics of standard FR4, high-Tg FR4 and polyimide, commenting that certain OEMs were now setting extremely severe specifications for thermal cycling performance. He also emphasised the point that multiple thermal excursions occurred in manufacture as well as in operation, and that a lot of the useful life could be taken out of material and components before an assembly ever got into service.
In summary, Morgan concluded that although traditional material choices for harsh environments and high temperature operation had been effectively limited to polyimide materials, the use of novel systems of chemistry had now allowed the circuit designer a much greater choice of laminates, based on many additional performance parameters such as Dk, Df, moisture absorption, ease of processing, and, ultimately, the total cost of ownership.
Moving one step along the supply chain, it was the turn of Dennis Price, quality director at Merlin Circuit Technology, to discuss high-temperature PCBs and solderable surface finishes from a printed circuit fabricator’s viewpoint. With reference to the critical laminate parameters previously detailed by Alun Morgan, he described three methods for measurement of glass transition temperature: Differential scanning calorimetry (DSC), thermal mechanical analysis (TMA), and dynamic mechanical analysis (DMA). All gave different results, sometimes quite significantly so, as demonstrated by examples he showed, and it was important to define which method had been used when quoting Tg values.
Price commented that the continuous maximum operating temperature of a laminate was difficult to quantify meaningfully. The UL relative thermal index (RTI) was derived from a mathematical extrapolation of data from accelerated ageing measurements and defined a temperature at which the material would operate for 100,000 hours and still retain at least 50% of its original physical or electrical properties. In his opinion, the RTI might not represent a PCB designer’s ultimate material temperature choice, but it was a good starting point. He went on to describe the tests used to establish the various UL flammability classifications for vertical and horizontal burning, which gave a preliminary indication of their suitability for a particular application.
Price then discussed the coefficients of thermal expansion, in-plane and through-plane, of various laminates in the context of thermal cycling reliability and described the principles of the interconnection stress test (IST), an accelerated method creating uniform strain from within a multilayer daisy-chain pattern of holes and tracks by DC electrical heating and forced convection cooling. The technique identified and assessed the severity of post separation and barrel cracks in plated through holes.
So much for the thermal characteristics of laminates--he turned to the practicalities of managing and removing heat from real PCBs and handed round many examples of boards with external and internal heat-sinking and expansion-restraining components, from simple bonded-on heat ladders, through copper-invar-copper constructions, carbon cores and modern insulated metal substrates.
He concurred with Morgan that polyimide was the most popular high-temperature laminate choice and materials were available from a range of suppliers. But it had certain shortcomings--brittleness leading to chipping on final-stage drilling and routing and a tendency to high moisture absorption. The ultimate high-temperature laminate choice was ceramic, either low-temperature co-fired (LTCC) or high-temperature co-fired (HTCC), but these technologies were outside the scope of mainstream PCB manufacture.
Regarding solderable finishes for high-temperature PCBs, electroless nickel immersion gold (ENIG) was the popular choice, although variants such as ENEG and ENEPIG were sometimes specified. Solder mask was an issue with high-temperature PCBs: Standard materials were epoxy based and would oxidise at prolonged temperatures above 150°C, resulting in a colour change from the usually green original to dark brown. There was a polyimide-based solder mask available but it was not photoimageable and had limited technical capability, with Tg quoted at only 165°C.
“Solving the Interconnection Challenge” was the presentation from Martin Wickham, from NPL’s Electronics Interconnection Group. Expanding upon the subtitle, “Sintered Silver for Interconnect,” he described how high-temperature electronics, for which there was an increasing demand, were pushing the boundaries for solder. Few solder alloys were available for operation above 200°C. Alloys with sufficient headroom were limited and of those, alloys not containing lead were very limited. Examples such as gold-tin were prohibitively expensive. RoHS recast proposals were due to go open-scope from 2019 onwards, although current RoHS exemptions might stay “until alternatives become available,” when those without alternatives would be left behind.
Sintered interconnect metals offered an alternative to high-lead solder alloys for high temperature applications and there was considerable interest both in silver-based materials for die bonding applications and in copper-based materials for forming conductors. It was known that nano-particle silver could be sintered at temperatures of 275°C or lower and that once sintered, the material would perform close to the melting point of metallic silver, 961°C. Wickham shared the results of current research at NPL on sintered silver interconnect technology with different component types and PCB finishes. Early trials with nano-powders had shown that it was possible to create basic sintered silver joints at 250°C with reasonable attachment strength using parameters consistent with a surface mount assembly process.
The aim of current work was to create circuit assemblies fabricated using sintered silver interconnects. Four different components had been investigated together with three different board finishes and two commercial silver pastes. The degree of densification of the sintered metal was dependent on factors such as uniformity of dispersion, particle size and particle size distribution, temperature, heating time, and the effect of the surrounding die material. A major issue with nano sized particles was their tendency to agglomerate, so dispersants had to be added, along with binders and solvents, to yield a printable paste, and all of these added components were detrimental to the sinter process. Principal ongoing challenges to the use sintered metals as high-lead solder replacements for PCB assembly were their compatibility with existing processes and components, particularly the finishes on substrate and component terminations, and the minimisation of sintering pressure. The mechanical performance and reliability of sintered assemblies had not yet been fully characterised.
Silver sintering was one of several approaches being considered by the High-Lead Solder Replacement Program, discussed by Henkel’s Global Technical Specialist Richard Boyle. High-lead alloys had traditionally been used for die-attach and as solders for high-temperature environments. Under current RoHS legislation they could continue to be used in selected electronics and automotive applications where no practicable or proven alternative existed. But if and when suitable alternatives were found then it was likely that the legislation would be amended to prohibit all use of lead-containing materials.
“The great thing about lead is that it works! It’s worked for a long time, and it’s cheap,” said Boyle. He commented that the market would have to change, and the way the electronics manufacturing industry looked at the market would have to change as well. The aims of the High-Lead Solder Replacement Program were to find suitable replacements in terms of electrical, thermal and mechanical performance, to develop products in compliance with RoHS and ELV standards, and ideally to produce drop-in alternatives to existing paste, wire and preforms. Additionally, recognising the changing needs in automotive and IGBT applications, to improve upon high-lead solder especially with regard to power and thermal cycling performance. The programme was taking a multi-faceted approach, looking at different solutions for sub-segments of the market, in respect of application, cost and performance.
Boyle summarised work to date on organic-based, metallurgic-based, and combination organic-metallurgic techniques including high silver-content organic die attach, epoxy-solder paste based on tin-antimony alloys, transient liquid phase sintering, and low-pressure or no-pressure silver sintering. Developments were ongoing, but his opinion was that no single drop-in solution would emerge and that different methodologies would be required for individual market segments, which might involve radical changes to production processes. Boyle reminded the group, “We’ve got five years left..”
“A changing environment” was the scenario for David Greenman, non-executive director at HumiSeal Europe, to make then and now comparisons. In the past, electronic assemblies had been based on relatively simple design rules with wide spaces between conductors, and had generally been expected to work in relatively controlled environments. Worst case, they might be required to withstand thermal cycling, damp heat, electro migration, mould growth, and salt mist. Modern day electronics, with ever-increasing packing density and decreasing conductor spacings, were expected to work in all environments: Examples of additional stress factors included thermal shock and thermal cycling to high temperatures, cyclic damp heat, mechanical shock and vibration, corrosive liquids, salt spray fog, and tin whiskers.
Conformal coatings had historically been applied for protection against these conditions, but the question remained whether conformal coatings and their associated test standards had kept up with the conditions of modern day harsh environments. Although purchase specifications could call-up a range of international, national, government and trade association standards, more and more in-house company standards were being published with increasingly severe environmental conditions, and sequential testing was becoming frequently specified for both unloaded test boards and production boards. A typical company sequential test procedure might include: thermal shock, followed by damp heat cyclic, followed by first thermal cycling, followed by corrosive gases, followed by damp heat constant, followed by second thermal cycling, followed by salt spray test, followed by long-term high temperature, followed by final inspection.
How were the suppliers of conformal coatings responding to these new challenges? New coatings were being developed and Greenman quoted examples of UV curable coatings that looked and behaved like silicones, but were not silicone, synthetic rubber coatings with extreme flexibility and high-temperature performance, and new UV cure silicones that overcame the high slump ratio of thermal cure silicones. And these coatings were being tested to the extremes of all of the specifications currently published or proposed.
The seminar had provided, in logical succession, the expert views of the laminate supplier, the PCB fabricator, the researcher into novel joining methods, the solder supplier and the conformal coating supplier. It remained for Ian Fox from Rolls Royce Controls and Data Services to tie it all together from the point of view of an avionics OEM whose safety-critical products were required to work in close proximity to aero engines. His wrap-up presentation was entitled "Harsh Environment Electronics: Materials and Making it Work." Typical examples of harsh environment conditions were high or extremely low temperatures, high humidity, and corrosive atmospheres. He first considered the effects of high temperatures on components: The maximum storage temperature for conventional semiconductors was 150ºC, although many would operate up to 175ºC with some possible degradation in performance and reduction in lifetime. To achieve useful life above 200ºC, devices based on silicon-on-insulator were required, and for temperatures above 300ºC there were very few options. One was to use devices based on silicon carbide but these were very expensive and of limited availability.
For soldered assemblies at operating temperatures above 150ºC, the properties of eutectic tin-lead were significantly degraded, the solder was extremely plastic and its fatigue resistance was low. Changing the alloy to tin-silver (SnAg4) or tin-antimony (SnSb5) gave satisfactory performance at temperatures up to 175ºC. Modern phenolic based FR4 printed circuit laminates with 180ºC glass transition temperature were acceptable for operation at 150ºC, but higher temperatures required the use of polyimide. For operating temperatures in the range 175-225ºC, high-lead solder alloys Pb95Sn5 and Pb93.5Sn5Ag1.5 were commonly used, although their wettability was limited. At high soldering temperatures, flux degradation was very rapid and inert or reducing atmospheres were recommended. Ian considered that vapour-phase soldering was preferable to convection reflow.
Post-assembly cleaning was essential and a conformal coating was required to prevent corrosion and oxidation, although presently only a limited choice of suitable coating materials was available.
The upper temperature limit for laminate-based PCB designs is considered to be 225ºC, and ceramic substrates offered a better option, with hermetic microelectronics approach considered the most robust solution. A critically-important consideration was to ensure that the materials and atmosphere enclosed in the package would not outgas in service, so they were vacuum-baked before sealing and sample sealed packages were subjected to residual gas analysis by mass spectrometry.
The stated objectives of SMART Group in promoting the advancement of the electronics manufacturing industry through the education, training, and notification of its members were emphatically achieved in this exceptional seminar. The audience had the benefit of the most up-to-date information from leading industry experts and the opportunity to engage in interactive discussion and to network with their peers. Special thanks are due to National Physical Laboratory for hosting the event and particularly to Dr. Chris Hunt for his efforts in organising and coordinating the programme.
