CitationDownload as .RIS
Emerald Group Publishing Limited
Copyright © 2004, Emerald Group Publishing Limited
Intellect Technical Workshop, 21 January 2004
Intellect Technical Workshop, 21 January 2004
CERN Microvia Technology
Keywords: Microvias, Conferences, Electronics industry
CERN, the European Organisation for Nuclear Research, presented a new, low-cost technique for microvia production at a workshop at Intellect's Russell Square offices on 21 January 2004.
Nathan Hill, who has responsibility for coordinating technology transfer for CERN in the UK, began with an overview of the structure and objectives of the CERN organisation, founded 50 years back and based in Geneva, with a one-billion Swiss Franc annual budget funded by 20 member states including the UK.
CERN's principal activity is the fundamental investigation of sub-atomic particles, with the world's largest particle physics laboratory. "Big physics presents big engineering challenges – not everyone can afford their own particle accelerator!" and its most significant current project is the construction of an underground accelerator 27 km in circumference, with four detectors each the size of an apartment block. When in continuous operation these will collect, record and analyse the equivalent of 3 million CD- ROMs-worth of data per second. Nathan remarked that an enormous commercial opportunity exists for the European electronics industry to supply processing and storage capacity to equip the system, the majority of the requirement being projected for the period 2006-2010.
The UK's contribution to CERN is funded by the strategic science investment agency PPARC, the Particle Physics and Astronomy Research Council, amongst whose objectives are the encouragement of multidisciplinary research and the dissemination of technologies to broader market areas. Technology transfer forms an integral part of CERN's principal mission of fundamental research. Recent UK successes include thermal management systems for semiconductor packaging, optical interconnects and medical scintillators.
CERN's activities are not limited to nuclear research; one of their engineering support departments is dedicated to the development of electronic modules, photolithography and microconnectics technologies, and has the capability to manufacture a wide range of printed circuit boards (PCB) and thick-film hybrids.
The section leader of their PCB workshop, Rui de Oliveira, explained how CERN's patented ChemicalVia process had originated from a project to improve the efficiency of manufacture of gas electron multiplier detectors, which require the creation of millions of fine holes in a metallised dielectric substrate.
A chemical procedure had been devised which would etch polyimide film without attack on copper, such that copper could be used as a conformal mask in a similar way to analogous processes employing plasma or laser ablation. This had been developed and optimised as a method for the creation of microvia holes in high-density- interconnects, which avoided the heavy capital investment and operational costs of plasma or laser equipment and could be straightforwardly integrated into existing process lines for the production of build-up multilayer circuit boards or multi-chip modules.
Rui took the example of a sequential build-up layer to illustrate the technique, and his explanation of the process prompted a very lively interactive session.
A commercially available single-sided adhesiveless copper-clad polyimide flexible material is laminated to the substrate PCB with a standard 7 μm epoxy bonding film, and normal photo-mech imaging and etching are used to create the via openings in the copper foil. These may be of any shape or size, but are typically 70 μm diameter round apertures in 5 μm foil.
The work is then placed in a chemical solution which etches the polyimide. No special equipment appears to be necessary – the work is suspended vertically in the tank like any normal process stage of a typical plating line. (The photographs showed process tanks enclosed in a cabinet with sliding doors, but no extra-ordinary safety precautions were said to be necessary.)
Although polyimide from different sources may etch at different rates, it typically takes about 10 min to etch to a depth of 50 μm. Graphs were presented which showed the etching rate to be almost linear. Certain materials have a greater tendency to undercut, but microsection photographs indicated that, under optimised conditions, the hole-wall profile is slightly conical with hardly any undercut. The material appears to be removed unidirectionally, and remarkably uniformly. In contrast, the equivalent geometry produced by plasma etching shows characteristically isotropic attack resulting in significant undercut.
Because the ChemicalVia process forms all of the via features in a single operation, the cost is effectively the same whatever the number, size or shape of holes – the more complex the design, the more cost-effective is the process.
Once the full thickness of the polyimide has been etched, a surfactant is used to clean degraded material from the hole walls and it remains to remove the thin epoxy adhesive layer to expose the copper target pad, using a sulphuric acid spray process which takes only a few seconds. Although the epoxy is dissolved isotropically, the glue-line is only 7 μm thick so undercut is insignificant.
The work is then ready to be metallised, plated, imaged and etched by normal PCB processes to produce the microvia-interconnected build-up layer.
The ChemicalVia process is limited to a single layer at a time – it is not practicable to form vias to the second layer beneath the surface, and aspect ratios are preferably less than about 0.7 (although 50 μm vias in 50 μm material are practicable), but sequential build-up is straightforwardly achieved.
Presently, it is not possible to stack microvias directly above each other because the thin epoxy bonding film is insufficient to fill the holes. However, it came out of the interactive discussion that current developments in via-filling electroplating processes should overcome this limitation.
Rui handed round a series of practical examples which illustrated actual HDI circuits made in his department by the ChemicalVia process. Included was one of the gas electron multiplier detector components upon which the original development had been focused. This had a very dense matrix of 50 μm through-holes in 50 μm material created by simultaneous double-sided etching. Microsection photographs showed a characteristic double-cone hole profile with minimal undercut.
Rui remarked that in his own case, the initial capital cost of implementation of the ChemicalVia process was approximately 6,000 euro, as compared with approximately 600,000 euro had he chosen the laser alternative.
Nathan Hill emphasized the cost advantages of the process whilst describing the mechanism by which CERN were prepared to licence the technology. Because the research had already been publicly funded, it was not intended to make substantial commercial profit, and the Technology Transfer agencies had a responsibility to make the technology available to even the smallest of PCB fabricators at nominal cost. Therefore, non-exclusive licences were available in the UK for approximately £15,000 plus a royalty of 2 percent on added value. Prospective licensees were welcome to visit CERN for further discussion, subject to non- disclosure agreement.
CERN's Technology Transfer agency in Poland had, in late 2002, successfully licensed the process to the Polish industry, and had cooperated in the building of a demonstration module which was now available commercially. Licensees could have technical training in CERN or in Poland, or could pay a consultancy fee to have training in their own factory.
Further information is available on the CERN Web site: www.cern.ch/Technologies/ChemicalVia
First point of contact in the UK is Nathan Hill: Tel: 01223 422405; E-mail: email@example.com.