Reading time ( words)
Stimulated by the visit to Airbus Defence and Space, relaxed and refreshed by an excellent conference dinner in downtown Munich and a good night’s sleep, delegates were bright-eyed and attentive for the second day of the EIPC Winter Conference. The programme comprised two sessions of technical presentations, one on copper cleaning and advanced material solutions, and the other on advanced imaging and solder mask, with a final panel session on how to make PCBs smart and ready for Industry 4.0.
Professor Martin Goosey moderated the first session, which opened with a presentation from Martyn Gaudion, CEO of Polar Instruments in the UK and expert on PCB signal integrity and impedance design. He discussed how modelling and measurement could work together to ensure reliable prediction of multi-GHz transmission-line performance, with particular reference to the effects of copper surface roughness. After briefly reviewing the main resistive and dielectric contributors to insertion losses, he focused on the “skin effect”—the tendency of current to flow progressively closer to the surface of a copper conductor at high signal frequencies, to the extent that at 10GHz, the current was carried in a skin depth of less than 1 micron.
If theoretical loss calculations assumed a smooth conductor surface, they could give seriously inaccurate results in the case real conductors. Copper surfaces were deliberately treated during foil manufacture and multilayer PCB fabrication to enhance their adhesion to resin, and these bonding treatments generally resulted in surface roughening, to greater or lesser extent depending on the application. The surface roughness made a significant contribution to insertion loss, and needed to be taken into account in the loss calculation. “Remember: What you think you made may not be what you actually made!” Gaudion illustrated this with microsections and explained how a technique known as the Hammerstad Method could be used to measure roughness and derive correction factors.
The results of three measurement methods for insertion loss, Vector Network Analysis (VNA), Short Pulse Propagation (SPP) and Single Ended TDR to Differential Insertion Loss (SET2DIL) had been compared and it had been concluded that, with good coupon design and accurate microsectioning, excellent correlation between modelled and actual values could be achieved.
Gaudion’s presentation set the scene perfectly for the next presentation, from Wim Ongenae, sales and marketing director at MEC Europe in Belgium, entitled “Impact of Copper Topography on Signal Loss in High Frequency Applications.” Ongenae introduced a new chemical bonding pre-treatment for copper which offered an alternative to traditional surface-roughening processes. This process resulted in a chemical, rather than mechanical, bond to laminating resin, and maintained an effectively smooth conductor surface. The process sequence began with a non-etching acid cleaner followed by the deposition of copper-tin-nickel alloy coating only 50–100 nanometres thick. This was then treated with a silane anti-tarnish, present as a monomolecular layer. The silane had a similar function to the bonding treatment on glass fabric in that it acted as a chemical coupling agent, forming a covalent bond between the metal surface and the laminating resin and giving high adhesion without roughening the surface.
Adhesion testing with a range of low-loss prepregs had shown excellent results, superior to those obtained with conventional roughening treatments even after multiple reflow, and as a consequence of the smooth surface the transmission loss at 50 GHz was equivalent to that of untreated material and about 10dB/M less than could be achieved with other bonding treatments.
Dr. Peter Amann from KIV PCB ProfiChem in Germany described a family of new universal sprayable processes for copper pre-treatment and activation that could be used at various key stages of PCB manufacture to improve first-pass yield and reduce cost. He gave details of a formulation based on iron-sulphuric, replenishable with hydrogen peroxide, for cleaning and micro-etching copper prior to dry film photoresist application, which gave a finer surface texture than sodium perslphate, and one based on peroxide-sulphuric that was successful with liquid photoresist. For cleaning and activating prior to multilayer bonding, he recommended a nitric-peroxide formulation, which gave very reliable bonding results. Similar chemistry was recommended, but in a two-stage configuration, for copper preparation prior to solder mask application. In each case he gave operating conditions and replenishment schedules. For small-volume manufacturers who could not justify a separate pre-treatment line for every operation, he suggested a versatile multi-purpose line, with a series of process options and alternative turn-out stations.
Stefan Hotz from Atotech in Germany introduced a recyclable and environmentally friendly etchant for copper pre-treatment. The chemistry was based on iron and was compatible with both stainless steel and titanium equipment. A closed-loop electrolytic regeneration system simultaneously re-oxidised the iron from ferrous to ferric and electrodeposited copper as saleable metal. The process gave a low-profile grain-boundary micro-etch and the resulting surface gave excellent solder mask adhesion. And the closed-loop concept eliminated the need for feed-and-bleed replenishment so that no copper-rich waste water was generated. Final testing was taking place, with a view to releasing the process for sale in Q1-Q2 2015.
And now for something completely different, although it still involved copper! Martin Cotton, Director of OEM Technology for Ventec International Group, looked at thermal management of PCBs from a designer’s point of view. Traditionally, component-level thermal issues had been addressed with external or internal heat-sinks. He specifically mentioned “coins”—slugs of copper embedded in recesses in the circuit board under the hot-spots of critical components. The design rules were many and complex, particularly with respect to dimensioning and tolerancing when multiple coins were required. Several alternative concepts had been explored, either as separate sub-assemblies of fabricated as part of the PCB, and using thermal via holes for heat conduction. But recent developments in insulated metal substrate (IMS) materials offered a more straightforward solution, and overcame many of the issues associated with localised heat sinks by spreading the thermal load over a wider area. Many grades of IMS were available and Cotton stressed that, whatever the nominal thermal conductivity in W/mK, the significant parameter was the actual thermal impedance, which took into account dielectric thickness as well. “To reduce the thermal impedance by half, double the thermal conductivity or halve the thickness of the dielectric.”
From thermal impedance, the dialogue reverted to the subject of electrical impedance, with a presentation from Alexander Ippich of Isola Group in Germany. He reported the results of a designed experiment to observe the influence of copper foils on impedance, DC line resistance and insertion loss. A test coupon had been designed with five different line widths and routed on two layers, three coupons per panel, and three panels were manufactured for each combination of four different copper foil types and two different oxide-replacement bonding treatments. Standard mid-loss laminate and prepreg were used, in an 8-layer stack-up with the 1oz copper foils under evaluation on the signal layers 3 and 6.
The foils evaluated were: matte-side treated very low profile, shiny-side treated very low profile, matte-side treated ultra-low profile and matte-side treated ultra-low profile with adhesion promoter. The bonding treatments were standard oxide replacement and low-etch oxide replacement. Insertion loss, impedance and DC line resistance were measured and the coupons were subjected to repeated reflow and solder shock testing.
Ippich described the electrical test methods and discussed the results in detail. There was a significant difference in insertion loss between matte side treated very low profile foil and reverse treated very low profile foil, and insertion loss could be significantly reduced by going from very low profile to ultra-low profile foils. The addition of adhesion promoter to the ultra-low profile foil has no negative effect at 5 GHz . The use of a low-etch oxide replacement instead of a standard oxide replacement was seen to be beneficial regarding insertion loss. This was true for standard copper foil and also for lower profile copper foils.
Thermal stress testing included six-times repeated reflow at 230°C as received, solder float testing 3 x 10 seconds at 288°C after preconditioning for four hours at 150°C, solder float testing 6 x 10 seconds at 288°C after preconditioning for four hours at 150°C, followed by cross-sectioning to check for degradation. All samples survived solder float testing with no delamination observed. Likewise, reflow simulation testing at 230°C showed no irregularities and no delamination.
The session on advanced imaging and solder mask was moderated by Oldrich Simek, owner of PragoBoard in the Czech Republic, and his first presenter was Uwe Altmann from Orbotech in Belgium. Well-known for his knowledgeable introductions to new developments in imaging technology, Altmann reviewed the development history of laser direct imaging from the late 1990s to the present day, as a background to his description of Orbotech’s latest innovation, the multi-wave laser. The market demand was for the capability to image any resist, with high throughput, whilst maintaining the best in image quality and registration. The limitation of previous systems was that diode-pumped solid-state lasers could only produce a single UV wavelength. Primary imaging resists and solder masks had photoinitiator systems with different spectral responses, so a single wavelength could not necessarily achieve optimum photospeed over a range of products. The new multi-wave laser integrated 375 nm and 405 nm sources into the same head, with the ability to control their relative intensities to match a particular photoinitiator. Altmann explained that these two UV wavelengths affected the polymerisation mechanism in different ways, one tending to work downwards from the top surface of the resist whilst the other was reflected from the copper surface and worked in the opposite direction, resulting in very uniform polymerisation with less energy consumed. And he demonstrated this with SEM photographs showing perfectly straight sidewalls on resist 100 microns thick. Orbotech’s latest direct imager could achieve 18 micron resolution at a throughput of 150 double-sided panels per hour.
Next to speak was Don Monn, Taiyo America’s European Sales Manager, with a presentation on solder mask developments in the digital age. “Before I tell you what I’m going to tell you, I’m going to tell you what I’m not going to tell you!” Speaking from his many years’ practical experience of applying and imaging solder mask, he began by listing the limitations of the original technique, screen printing, then ticking them off one-by-one as technology advanced. His list included: registration issues, artwork, bleed, skips, exposed copper and uneven mask deposit. Liquid photoimageable resolved some of the issues, but imaging still depended on artwork, and coating skips were still possible. Laser direct imaging resolved the artwork issue, but some coating issues remained although coating techniques had progressively improved. “What’s next?” he asked, keeping his audience in suspense whilst he tempted them with another list: “It won’t require artwork, it won’t require most typical processing equipment, it will create lots of space on the shop floor, and it will eliminate the three biggest challenges in the solder mask operation: registration, mask in holes and lost or missing dams.” With a theatrical flourish, he unveiled a video showing solder mask being ink-jet printed. Although other suppliers’ prior attempts at solder mask ink-jetting had been unsuccessful, Monn was confident that Taiyo, in cooperation with several manufacturers of ink-jet printing equipment, had formulated a product that would be rapidly adopted by the PCB industry.
Returning to the topic of photoimageable solder masks, Sven Kramer from Lackwerke Peters in Germany gave an overview of today’s material capabilities and limitations, with a particular focus on their ability to withstand thermal stress. “The first law of material science: Everything can be destroyed by force!”
The most obvious effect of thermal stress on a solder mask was discolouration. What was an acceptable level, particularly for ultra-white materials used in LED applications? Was the ultimate degree of initial whiteness more important than the thermal stability of whiteness over time? He explored the failure mechanisms of solder mask under heat-ageing and thermal cycling conditions. The ageing process could result in loss of volatiles, oxidation, continuation of polymerisation, chemical separation of low-molecular-weight components or hydrolysis of polymers in the presence of moisture. The effect of copper surface preparation on the thermal ageing of solder mask had been extensively studied by Peters, particularly in respect of adhesion, and chemical methods had been shown to give more consistent long-term adhesion than pumice scrubbing or mechanical brushing.
Long-term ageing testing had demonstrated that specialised liquid photoimageable solder masks now commercially available were capable of operating at 175°C, with adhesion and insulation performance at a similar level to standard coatings rated for 150°C long term resistance. Substrate pre-treatment had become increasingly important to achieve the desired results, and it was necessary to use high-Tg substrates to avoid cracking resulting from thermal mismatches. It was not yet possible to offer products with long term 200°C thermal resistance. Although good adhesion to substrate could be maintained, insulation properties were degraded after long term exposure to 200°C, and the mechanism was not yet fully understood.
All this talk about solder resists—what about the formalities of gaining UL recognition for a solder resist and adding it to your UL-recognised PCB? With clear and straightforward explanations, Emma Hudson, Lead PCB engineer at Underwriters Laboratories, guided the way through the UL system, pointing out whether additional testing would be necessary within the Solder Resist Recognition Programme, designed to make it easier for the PCB manufacturer to add new resists to their UL-recognized boards. Ms Hudson made it clear that the primary concern with solder resists was the flammability hazard—did they add fuel to a fire? UL recognition showed a commitment to safety and provided the customer with added reassurance that an independent party had tested and would continue to verify the material on a regular basis.
Provided certain conditions were satisfied, the Permanent Coating Program in UL 796 allowed PCB manufacturers to add recognised resists to existing board types without any additional testing. And UL recognised solder resists could be cheaper and quicker for a PCB manufacturer to start using than unrecognised materials. She explained the requirements for gaining recognition of a solder resist, the requirements for improving recognition of an existing solder resist, considerations when adding a new solder resist to an existing rigid PCB and whether additional testing was required. With the re-classification of FR4 laminates, re-testing would be required if moving from FR4.0 to FR4.1.
More resists were being recognised in multiple colours, but it was not mandatory to test every colour to gain all-colour recognition, only the “worst-case” examples: no-colour, black, white and red. Reduced test programmes could be used to extend recognition of an existing resist to additional UL/ANSI grades or to improve other parameters. Ms. Hudson suggested that solder resist suppliers and PCB manufacturers should consider partnering when proposing to add a resist to a different UL/ANSI grade.
The final session of the conference was an open discussion on Industry 4.0 and Smart Manufacturing, led by a panel of industry experts and moderated by Alexander Schmoldt from Murata in Germany, who explained that Industry 4.0 - effectively the fourth industrial revolution - was about the technical incorporation of cyber physical systems in production and value chain logistics, and the integration of the Internet of Things concept into industrial processes. It would have important consequences in value creation and business models and would significantly affect the organisation of the work force. A basic paradigm change would be required: production in the “Smart Factory”: through “Smart Automation” interacting with “Smart Products,” each with a digital ID to enable new cloud-based services and business models.
Schmoldt considered the example of a Smart Factory manufacturing electronics, where an obvious way to make products “Smart” was to add an RFID function to their printed circuit boards, so that they each had a unique communication address and memory. The panel discussed the options for adding RFID to PCBs, by embedding an RFID into the board, or by adding an RFID module as a component during the assembly process. Various types of RFID devices were compared, and actual case studies were used to illustrate how the added value of smart PCBs benefited different stakeholders over the product life cycle.
Panel members Stephan Kunz from Schmoll Maschinen, Norbert Heilmann from ASM Assembly Systems and Torsten Bethke from Micronex each gave short presentations: Stephan Kunz demonstrated automated equipment for the embedding of UHF RFID chips in the edge of circuit boards, which was already being used commercially. Norbert Heilmann discussed the benefits and limitations of PCBs with embedded RFID in the assembly process, from the viewpoint of a manufacturer of assembly equipment. Torsten Bethke gave the views of an EMS provider, who saw serious challenges to human resources and the inevitability of a demographic change, where unskilled workers would be the losers. And there was concern from some members of the audience that the Internet of Things could lead to too much information being available to too many people and organisations, and that personal privacy was increasingly under threat.
In his closing summary, Schmoldt concluded that Industry 4.0 inevitably required smart products, and that RFID was recognized as a standard means to create smart objects. RFID in electronics 4.0 means source tagging of electronic products and RFID embedded into the PCB is the ideal means for a cradle to cradle approach, making cyclical and sustainable business models possible. The deployment of RFID infrastructure in electronics production had just begun, and presented business opportunities for systems integrators, hardware and software manufacturers. Many successful user-examples already existed in the industry, and had demonstrated the benefits of RFID to SMEs as well as large OEMs.
Alun Morgan brought proceedings to a close, once again acknowledging the support of the sponsors and thanking the presenters for sharing their knowledge, the delegates for their attention and, in particular EIPC Executive Director Kirsten Smit-Westernberg and Event Manager Sonia Derhaag for their faultless organisation and coordination of another hugely successful conference.
I am grateful to Alun Morgan for allowing me to use his photographs.