EIPC Summer Conference 2016, Day 2: Strategies to Maintain Profitability in the European PCB Industry
Delegates awoke to a wet and gloomy Scottish morning on the second day of the EIPC Summer Conference 2016. One or two who maybe overindulged in the whisky on the previous evening had some difficulty in finding time for breakfast before commencement of the conference proceedings, but the atmosphere in the meeting room was a lot brighter than the weather outside, as Professor Martin Goosey introduced the day’s programme, which began with a session of four presentations on the theme of future technology in components, materials and processes.
Title of the first paper was “Consistent miniaturisation, from conventional assembly to high-performance device embedding,” and because Michael Weinhold had to attend a family funeral, Alun Morgan stood in at short notice to deliver his presentation.
Weinhold commented that increased performance and functionality, as well as cost reduction, were the key drivers in electronic device manufacturing, and although different market segments had different needs, they shared the fundamental requirements of predicted product life expectancy and of meeting international material regulations and manufacturing standards. Moreover, repair of electronics devices in the field was now considered a no-go option. How would the European electronics industry realise these requirements, and how would the global consumer electronics market impact different market sectors? Key markets in Europe were military, avionics, space, medical, industrial and automotive, and each needed to be serviced by a specific supply chain. He set out discuss how PCB fabricators, EMS companies and OEMs could redirect their efforts to manufacture high-performing PCBs at the expected quality and cost whilst still achieving realistic profitability.
Weinhold had just returned from the JPCA show in Tokyo, where he had seen many examples of state-of-the-art technology, including mechanical drilling of 30 micron holes, with coated drill-bits for highly-filled laminates, six-layer multilayers on IMS materials for high thermal dissipation requirements, advances in all-electronic PCB-based 76GHz long-distance radar for automotive applications, drones operated from smart-phones, and an increasing number of modules and PCBs with embedded components.
Discussing possibilities for adding value to the PCB manufacturing process, Weinhold noted that PCB fabrication technology was dominated by Asia, and that bare PCBs represented only between 4% and 10% of the value of assembled boards. Against this background, he believed that European PCB fabricators had growth opportunities in new product developments with embedded devices.
Asked Weinhold, “What would change over coming years?” He believed that wireless interconnection interfaces would drive the development of new products, especially in the context of the Internet of Things, and that chips would be provided with wireless communication interphases, with an impact on connectors and assembly technology. Power electronics and applications handling voltages higher that 500V would have a significant effect on design and manufacturing of PCBs and assemblies, and new high-voltage test methods would be needed to meet operational life expectation of the industry and the end-user. Comparative tracking index would be an important parameter in new product developments. High-frequency PCBs would be required for faster data transfer in automotive, military, aerospace, avionics and medical electronics, as well as in high-speed networks: local, urban and global, and materials and fabrication processes would need to be adapted accordingly.
How would these changes affect the PCB fabricator? Component development would continue to drive miniaturization, and miniaturisation would drive the demand for cost reduction. The fabricator would need to achieve high yields on HDI and microvia product, and would need engineering expertise along the total process from design, through fabrication, assembly and testing. Product development could continue to be done in Europe and, provided companies understood market trends and needs, there could be many new R&D and product-development jobs. Having partners in low-cost countries would enable profits to be made on repeat orders and mass production. The essence of Weinhold’s message was “Understand your competence and positioning, find the right partners, build on the strength of your company and sell what you have!”
How many valuable PCBs are scrapped in process for shorts and opens caused by resist breakdown in plating and etching processes? The days when such defects could be repaired by knife-and-fork methods are now far behind us. For some time, Orbotech have offered automated optical repair systems for short circuits—using laser techniques to ablate unwanted copper, but what about the repair of open circuits by deposition of copper? It appeared from Alfred Kaiserman’s presentation that a solution is now available. Preferring not to use the term “repair” because of historical negative connotations, he described a laser-based “3D shaping” system, recently introduced by Orbotech under the name Precise™ 800.
He stated that the system automated the shaping process by way of two proprietary technologies: 3D Shaping (3DS) Technology™ and Closed Loop Shaping (CLS) Technology™. The 3DS™ additive technique was based on a series of processes, including 3D defect analysis, 3D laser shaping and 3D visualization. By comparing the shape of the defect to real-time CAM data and simultaneously conducting 3D analysis, 3DS™ automatically identified where copper needed to be added. It then guided a laser to a proprietary metal carrier from which copper was deposited in a series of thin layers to build up an exact 3D replica of the missing conductor, by a process of laser-induced forward transfer. The results could be confirmed immediately by 3D visualization. Kaiserman illustrated the operation with a video simulation. Conductors restored by this technique had been exhaustively tested to industry reliability standards, and there was a high level of confidence in their integrity. The subtractive function, CLS™, used image analysis algorithms to make real-time comparisons between the actual image and CAM data in order to detect the precise location of shorts and opens. It then intelligently guided the laser to accurately ablate excess copper. The system was clearly capable of significantly increasing yield and offered a rapid return on investment.
Conductive anodic filamentation, CAF, is a failure mechanism that was first reported in the 1970s by Bell Laboratories, and has become a significant reliability concern with increasing circuit density and the rapid increase of the use of electronics in harsh environments and for high reliability and safety critical applications. Helmut Kroener, senior director OEM Marketing Europe at Isola Group, presented a paper on CAF experiences from an automotive product qualification. He explained the mechanism of CAF formation, an electrolytic corrosion effect which builds a conductive path along a glass filament - epoxy resin interface under conditions of high humidity and bias voltage, exacerbated by ionic contamination, and defined three typical time-dependent failure zones: infantile, where defects in the material led to copper migration after bias was applied, transition, where partial defects occurred and shorts could recover to isolation, and wear-out, where the true CAF capability of the material could be determined. Kroener showed many real examples, and demonstrated that in most cases the actual CAF filament was difficult to pinpoint.
Many factors could influence CAF, some due to the laminate manufacturer and some to the PCB fabricator. From the laminator’s point of view, the cleanliness of the glass fibres, the compatibility of the silane treatment, and the completeness of wetting of the glass by the resin were probably the most significant, and resin formulation was obviously another factor, particularly its thermal stability. To qualify a laminate for automotive applications involved extensive product and processing testing, and a separate qualification was required for each material category and each PCB factory location. CAF testing was one of the many requirements and the particular work reported by Kroener was part of the qualification programme for Isola 185 HR material at a tier-one automotive supplier, with PCBs manufactured at an Asian production location.
CAF testing was carried out in accordance with IPC TM 650 2.6.25 on six-layer HDI test coupons with blind and buried vias, preconditioned by 3 x 260°C reflow cycles. Test conditions were 85°C, 85% relative humidity, 100 volts, for 1000 hours. Material samples were supplied to the PCB fabricator for test vehicle manufacture and testing. CAF failures were reported by the PCB fabricator, all other tests having been successfully completed, and it was agreed with the OEM for the CAF testing to be repeated by an independent laboratory. Failures were again reported, and detailed failure analysis conducted. Agreed process improvements were made in material production and PCB manufacturing, and the test programme was repeated with successful results.
Kroener discussed in detail the failure analysis procedures, demonstrating the delicate microsectioning, microscopic and instrumental techniques used to identify the actual CAF defects, and the interpretation of the results. It was found that, of four failures reported, three had PCB manufacturing issues as the root cause and only one had been a true CAF issue. There was a clear need to understand the entire history, that both material production and PCB manufacture could impact the CAF end result, and that both had to be CAF-optimised to succeed. “A poor PCB manufacturer can turn a good CAF material into a bad one, but a good PCB manufacturer cannot turn a bad CAF material into a good one—both partners have to work at their respective processes to generate an optimum result and qualify for an OEM.”
Dr Despina Moschou, Research Fellow at the University of Southampton, gave conference delegates a fascinating glimpse into the field of microfluidics and biosensors with her presentation on Lab-on-PCB technology for bioanalytical applications.
Enormous progress had been made in the field of lab-on-a-chip (LOC) technology, a subset of micro-electro-mechanical systems (MEMS) devices, in which multiple laboratory functions had been integrated on a single chip to enable tasks such as automated medical diagnostic analysis. Microfluidics was the physics, manipulation and study of minute volumes of fluids. A microfluidic chip consisted of a set of micro-channels etched or moulded into glass, silicon or polymer, connected together in order to mix, pump, sort or otherwise handle the fluid, connected to the outside by inlets and outlets pierced through the chip. Integration of microfluidics and biosensors, the bringing together of these two technologies to enable the analysis of, for example, blood or saliva samples, was the subject of Dr. Moschou’s research.
Lab-on-PCB had been suggested as a solution in the 1990s, but the concept had been side-lined by easier microfluidic fabrication processes. More recently, PCBs had been recognised as potentially ideal integration platforms, particularly because the long-standing industrial infrastructure offered low-cost upscaling. University of Southampton was collaborating in the EPSRC-funded eμ-ELISA project, which aimed to develop low-cost, real-time detection devices by adapting well established PCB fabrication processes to produce bespoke functionalised electrodes coupled with micron-scale fluidic channels and chambers.
Dr. Moschou showed examples of sensors for DNA, lactate and glucose, and fluidic microvalves and micropumps made by PCB techniques, and demonstrated a 3-layer
PCB microfluidics device with reference electrodes on layer 1, sensing electrodes on layer 2 and microfluidics on layer 3, and also a 2-layer PCB sensing electrode structure. Prototype lab-on-PCB devices had now been successfully fabricated and the project was moving forward.
The session on solderable finishes and plating for PCBs was introduced and moderated by Paul Waldner. The first presentation was given by Chris Klok from MacDermid Enthone, who discussed the use of OSP finishes in automotive applications.
There had been interest in OSP in the automotive sector since 2010, the main driver being cost, along with concerns about whiskers, corrosion, electromigration and rework. Tier-one suppliers were driving the upcoming change, and the last challenge to be overcome was press-fit compatibility.
OSP was by far the most widely used solderable finish globally, but mostly in consumer electronics. Immersion silver was a popular finish in automotive, and immersion tin was gaining ground, but these metallic finishes were several times more expensive than OSP. Tier-one suppliers were moving away from immersion silver due to restrictions from several major European OEMs. HASL was gradually being phased out, and changes in soldering procedures favoured OSP. OSP was approved as solderable finish by the majority of tier-one suppliers of automotive electronics, but a generic specification was required for press-fit.
A German consortium of OEMs, tier-one suppliers, PCB manufacturers and press-fit pin manufacturers was doing research on press-fit capabilities on OSP-finished PCBs, carrying out optical analysis to IPC-A-610E, and mechanical and electrical testing and metallographic analysis to IEC-60352-5, with Eye of the Needle, Spring Shape and Cracking Zone pin types.
Insertion force measurements showed OSP to give similar results to immersion tin, and although a lot of qualification work remained to be done for different applications and pin types, and metallic finishes would still be required for specific applications and functionalities, press fit was no longer considered a road block for OSP in automotive and an increase in the use of OSP was to be expected
Still on the topic of preferred finishes for automotive electronics, Rick Nichols, Global Product Manager Final Finishing with Atotech, maintained that immersion tin was gaining market share, primarily due to the confidence of automotive OEMs. It offered maximum solder performance at a reasonable price and had corrosion resistance second to none. He discussed the practical and metallurgical aspects of soldering immersion tin and explained what defects could occur and how they could be avoided.
In the as-received state, the immersion tin finish on a PCB had a uniform white appearance, with a thin oxide layer at the surface and a thin Cu6Sn5 intermetallic layer at the interface, both of which were beneficial in Nichols’ opinion. After solder paste printing and during pre-heating prior to reflow of the first side, the oxide layer was removed by the flux. During the peak temperature phase of the reflow process, both immersion tin and solder were melted and there was rapid growth of Cu6Sn5 intermetallic, in the form of “scallops,” and slow growth of Cu3Sn intermetallic. After the first reflow, soldered pads were very well covered with solder and the non-soldered pads still appeared clean and white clean. During reflow of the first side, similar intermetallic growth occurred on the as-yet unprinted side, possibly reaching the surface, and the oxide layer increased in thickness. To compensate for this, it was recommended to print more paste on the second side. During pre-heating, the flux had more oxide to remove, and at peak temperature, because there was less free tin, there was reduced spread of solder. After the second reflow the non-soldered pads were still visually clean, but slightly darker in colour.
Nichols went on to discuss possible causes of de-wetting on the second side after successful first-side reflow: residues on the copper surface, contamination of the tin surface or evaporation of volatiles from solder mask during the first reflow. In his experience, solder masks were often not completely cured free from volatiles on as-received PCBs, and he strongly recommended precautionary UV bumping to give additional reliability.
A novel electroless nickel / immersion palladium / immersion gold universal finish for PCBs was described by Professor Karl Ryder from University of Leicester, a leading expert on applications of deep eutectic solvents in metal finishing. He explained that deep eutectic solvents were a class of ionic liquid in which organic cations combined with halide anions and complexing agents to yield a purely ionic material with remarkable solvent properties. The specific example used in the university’s research work was known as Ethaline 200, composed of ethylene glycol and choline chloride in 2:1 molar ratio, and this was relatively inexpensive and environmentally benign. To demonstrate the unusual solvation properties of deep eutectic solvents with metal ions, he showed a series of solutions of copper 2 salts, which would be blue in aqueous solution, covering a full rainbow spectrum of colours.
The benefits of deep eutectic solvents had been demonstrated in metal finishing applications such as electropolishing, electroplating and immersion plating, as well as in metal recycling and energy storage, and as fluxes that enabled soldering direct to difficult-to-wet surfaces such as electroless nickel. And a previous EU 7th Framework project called IONMET had demonstrated that silver could be successfully deposited on copper as a solderable finish for PCBs. A current project, MACFEST, co-funded by Innovate UK, aimed at producing a high-reliability solderable and wire-bondable universal PCB finish, with good planarity and long shelf life. Deep eutectic solvent technology was being employed to improve functionality and to reduce safety and environment concerns. The first 15 months of the 24-month project had been completed. Immersion palladium had been deposited onto a proprietary electroless nickel base layer from Ethaline at 80°C to a thickness of 70–100 nanometres in 30 minutes. The palladium deposit had been over-plated with gold from a second Ethaline-based formulation, with the gold present as chloride or thiosulphate, at 50°C for 9–15 minutes. Bright uniform deposits had consistently been achieved from chemistry free from acid and cyanide. This ENIPIG (electroless nickel, immersion palladium, immersion gold) finish had shown excellent solderability, with no evidence of the black pad or mud-cracking effects which could occur on the nickel surface when traditional aqueous chemistries were used for gold deposition.
Alex Stepinski is well-known for planning and engineering the first captive PCB manufacturing facility in North America in many years, at Whelen Engineering in Charlestown, New Hampshire. Their fabrication process is characterised by a high level of automation and the innovative re-thinking of many of its fundamental operating principles. His presentation focused on the reduction of chemical system costs, where fresh chemicals represented only about 40% of the total cost associated with a new chemical process. This percentage could be even less if equipment was used that had not been optimised for the process. Typical breakdown of the remaining 60% in the North American market was: regulatory costs and permits 15%, wastewater treatment costs 15%, drag-out losses, 10%, chemical maintenance costs 10%, energy costs 5%, and fume losses 5%. So where did he achieve savings?
Starting with emissions management: “Fumes are good chemistry being wasted in the process of generating system under-pressure to prevent ambient exposure of personnel and equipment to toxic chemicals.” Stepinski discussed various techniques both for minimising fume generation and for recovering chemistry, water and solvents from fume extraction systems. The ultimate solution was to totally mitigate emissions with a hermetically sealed system, which could give very rapid return on investment in the case of high total-dissolved-solids chemistries.
Turning to dragout management: “Chemical drag-out is good working chemistry being wasted by mixing it with water needed to rinse the panels for the subsequent process step”. He discussed drag-out minimization and recovery techniques and demonstrated that approximately 50% of dragout could be recovered simply by using concentrate rinses for make-up of the process bath. This could be increased to 80% if off-line thermal and membrane systems were used at the point source.
Finally, Stepinski considered developments in central wastewater technology: “Central wastewater treatment should only be used for purification of trace contaminants, and for the treatment of spent chemical concentrates. 90% of drag-out should never reach this process.” Traditional systems separated regulated contaminants from wastewater and discharged the decontaminated wastewater, with metal hydroxide waste as a by-product. Modern systems took the process a step further and deionised the wastewater for re-use. A hybrid reverse osmosis / high capacity ion exchange system was the most cost-effective solution currently available. Furthermore, thermal distillation of concentrate waste could result in a zero liquid discharge.
Stepinski made the point that closed loop systems effectively gave an unlimited water supply, so there was no need for water conservation, and much higher flow rates could be used for improved rinsing. The strategy then became one of budgeting contamination as opposed to budgeting water usage.
By putting all of these principles into practice, Stepinski had achieved a reduction of 80–90% in direct operating costs for chemical process systems compared with a conventional process, and the site was the first PCB production facility in North America to have all of its permits waived. The wastewater system was automated to the point of requiring only 10 hours per week of total labour. Total capital expenditures for all the chemical recovery systems associated with this project was $1.4 million, and costs would be lower in a brownfield site where existing equipment could be repurposed.
The final technical session was entitled “Tooling and fabrication experience with advanced technologies,” and was moderated by Martyn Gaudion. His first presenter was Mehul Davé, CEO of Entelechy Global, who made a convincing case for outsourcing CAM engineering. (Curious about the origin of the company name, I googled it and learned that “entelechy” means “the realisation of potential.”)
Why outsource CAM? In every PCB fabrication facility, there was increasing pressure on the front-end engineering department, more part numbers, more quick-turn requirement, more quoting activity, a greater focus on high technology and a customer expectation of fast response—all driving the need for off-shift pre-engineering and engineering capability.
A whole range of functions could be successfully outsourced, including quote-data processing, pre-CAM, CAM and post-CAM, and not just for the simple jobs - the right partner could handle complex high layer count HDI designs with controlled impedance and buried components, as well as flex and flex-rigid. There were six key areas of impact: on-demand capacity, improved automation, faster turnaround, reduced cost, improved quality and the ability to build redundancy in critical areas. The biggest benefit was flexible capacity, reducing overtime or the need to hire off-shift engineers. Automation was the key to efficiency, quality and reliability, and outsourcing reduced the need for dedicated staff to develop, support, and update automation. The time zone difference between India and Europe or North America was a benefit in that jobs sent in the afternoon would effectively be engineered overnight and be ready to go into production the following morning. Outsourcing gave access to higher quality talent, without employee overhead expenses or vacation costs, and reduced infrastructure costs. Outsourcing required documentation and systematic standardised processes, with consequent improvement in quality. Furthermore, having an offshore team ensured that “tribal” knowledge no longer resided in one or two key individuals who could leave the company and take the know-how with them.
The ubiquitous Alun Morgan came forward again, this time as European representative and project facilitator for the High Density Packaging User Group (HDPUG), to report the results of the HDPUG PWB Back Drilling Project. He explained that controlled depth back drilling of plated through holes was increasingly used in high speed designs to remove redundant copper which caused attenuation losses and made it difficult for a digital receiver to ascertain whether the received signal was truly a logic “one” or a logic “zero.” The critical parameter was the length of the remaining copper via stub from the targeted inner layer pad. Stub length design rules were driven by electrical requirements, not necessarily based on PWB reliability data or fabrication capabilities, and there appeared to be an increasing number of reliability issues attributed to back-drilling. So the project had set out to quantify the relative reliability of back-drilled holes to PTH holes, in terms of drill depth, hole and pad size and pitch, and to develop test coupons and measurement methods.
Some interesting observations had been made, particularly that normal PTH vias failed before back-drilled vias under thermal cycling, and there appeared to be a counter-intuitive hierarchy of design failures, with shallow back-drilled vias failing before middle and deep back-drilled vias. Morgan reverted to old-fashioned teaching aids and drew pictures on a flip-chart in response to requests from the floor to explain the failure mechanism, which in the event actually appeared quite logical.
A programme of IST testing in cooperation with PWB Interconnect Solutions had quantified the relationships between reliability and design rules, and verified the interconnection failure modes. Whereas normal plated-through vias tended to fail in tensile stress by barrel cracking, back-drilling effectively removed one outer-layer anchor and transferred thermal cycling stresses from tensile in the barrel to shear at the barrel-to-inner-layer-pad interface, resulting in bending of the copper pad and eventual cracking of the internal foil.
A non-destructive test for stub-length measurement, based on time domain reflectometry, had been developed in cooperation with Introbotix. This gave repeatable results that correlated well with microsection data, and the technology was applicable to both low-volume probing and robotic high-volume probing.
An example of cooperation between academia and manufacturing industry was demonstrated in the presentation from Anjali Krishnanunni, a KTP Associate/Project Officer at Coventry University and currently based at Stevenage Circuits. She explained how Innovate UK, the UK's innovation agency supported Knowledge Transfer Partnerships (KTPs) to assist businesses in gaining a competitive edge through better use of knowledge and technology.
Krishnanunni discussed current and future issues in PCB technology from the viewpoints of the designer, the fabricator and the assembler. Fine pitch component packaging demanded improved routability and higher interconnection density, using finer conductor geometries whilst keeping layer count to a minimum by cost-effective any-layer via-in-pad design rules. For maximising assembly yield, a critical factor was precision of solder paste application, and this was heavily influenced by the effect of PCB and solder mask topography on the consistency of contact between PCB feature and stencil. Established PCB fabrication techniques were approaching their capability limits and new methodologies were required. These were being explored in the Pitch Perfect KTP, a project with the objective of developing an ultra-high-density-interconnection fabrication technology compatible with a broad range of substrate materials, with low-cost interstitial vias and the potential for 25-micron lines and spaces, using low-impact surface modification techniques to minimise surface morphology. Although at the present stage of the project, IP considerations prevented her from disclosing practical details of process procedures, the technology would integrate straightforwardly with current process infrastructure, incurring minimal additional capital expenditure, and would be compliant with appropriate IPC standards.
Final presentation of the day came from Stuart Dalrymple, senior project manager at C-Tech Innovation, who reported the outcome of the REPRIME project, which had carried out research into the application of ultrasonics to replace poisons and explosives used in industrial metal finishing processes.
The project was funded by the UK Home Office, who wanted to find a technology-based instead of legislative solution to the perceived threat of chemicals held in relatively insecure conditions in small-to-medium companies being misappropriated and used to support terrorist activities. The objectives of the REPRIME project had been to overcome barriers to the adoption of cyanide-free technology, to demonstrate cyanide-free zinc and zinc-nickel plating on an industrial scale, to extend the work to cyanide-free copper, gold and silver plating, to reduce hydrogen peroxide use in the printed circuit industry and to ensure that the technology could be easily and cheaply retrofitted to existing equipment.
Referring to previous collaboration in the Susonence project, an EU Eco-Innovation Initiative, which had demonstrated that ultrasonics could be successfully used to replace chromic acid in the etching of ABS polymers, to improve the efficiencies of barrel plating in general metal finishing, and de-smear and copper etching processes in PCB manufacture. Dalrymple described how the use of ultrasonics had enhanced the deposition rate of cyanide-free zinc electroplating plating chemistries and improved coverage and distribution on complex shapes. It had also been shown that ultrasonics enabled the use of reduced concentrations of hydrogen peroxide in etchant solutions used in PCB manufacturing, and gave improved bath life with reduced frequency of replenishment and no adverse effect on downstream processing. The REPRIME project had been successfully completed, and was being rolled out to industry with continuing support from the Home Office, the Surface Engineering Association and the Institute of Circuit Technology.
In his closing remarks, Alun Morgan thanked the moderators and speakers for their contribution to an interesting, interactive and entertaining conference and networking experience, and delegates for their attention and involvement. He acknowledged the generous support of the sponsors, Viking, Isola, Polar and Ventec, and extended particular thanks to Kirsten Smit-Westenberg and Sonja Derhaag for their impeccable organisation and management of yet another highly successful EIPC event.
Click here for the slide show of the event. Read about Day 1 here.