ICT 25th Anniversary Symposium

Citation

Cosslett, K. (1999), "ICT 25th Anniversary Symposium", Circuit World, Vol. 25 No. 4. https://doi.org/10.1108/cw.1999.21725dab.003

ICT 25th Anniversary Symposium

ICT 25th Anniversary Symposium

Keywords: ICT, Microvias Microvia Technology

This one-day symposium, organised by the Institute of Circuit Technology, was held at The Swan Diplomat Hotel, Streatley-on-Thames, UK on Tuesday, 20 April 1999. David Woodley, the Chairman of The Institute of Circuit Technology, welcomed delegates to the 25th Anniversary Seminar celebrating the 25 years since the founding of the Institute. The day was divided up into two distinct halves, the morning session devoted to the technical papers, whilst the afternoon session was dedicated to a works visit to the Panasonic assembly line at Thatcham. This report covers the talks given in the morning session.

The opening talk was given jointly by Sue Critcher of Inpaq and Brian Cosgrove of MCM and entitled "Designing with Microvias". Sue Critcher opened by reviewing the early stages of PCB manufacture. In the early days PCBs were made with 25 thou tracks, 25 thou apart, 25 thou diameter holes with 50 thou pads. At the design stage these were drawn large and then reduced photographically. This method of manufacture of through hole boards was satisfactory for assembling components on one side only but there were difficulties with assembling on the reverse side. Now we have 4 thou tracks, 25 thou pads and 12 thou holes and the overall track density is much higher. Components are now mainly surface mounted and this frees up internal layers.

With advancing technology allowing blind vias and buried vias there was also increased component trends. Sue Critcher showed a diagram of these starting with PLCC of pitch 1.0mm, followed by QFP of pitch 0.5mm, then BGA of pitch 1.27mm, then µ BGA, pitch 0.8mm, CSP, pitch 0.5mm, Flip Chip of 0.35mm and finally DCA (no pitch given).

Component Migration was also occurring and again she showed a bar chart of trends starting with 160 pin QFP of 1.3 square inches, followed by 169 pin BGA of 0.8 square inches and finally 188 pin CSP of 0.3 square inches. There were, also, smaller tracks and gaps. With these trends so costs started to rocket and she showed another bar chart of percentage increases in costs of routing with reducing track width.

Another diagram shown was of Pad diameter comparison:DIL Pad diameter ­ 63 thou, drill diameter 35 thou, SMD Pad diameter ­ 30 thou, drill diameter 16 thou Today's Pad ­ diameter 8 thou, drill diameter 4 thou Tomorrow's pad ­ diameter 4 thou, drill diameter ? thou.

Small holes cost more. Increases in hole drilling costs rise exponentially as hole size decreases. This was shown on a bar chart where the cost shown started with a drill size of 13.5 mil at about $2 per 1000 holes and a 4 mil drill cost $25 per 1,000 holes. The example given was that 6 thou holes are ten times the cost of 12 thou holes.

Why so expensive? Conventional methods are not compatible with the smaller hole sizes now required; hence increased costs.

Brian Cosgrove continued the talk on the manufacture of microvias and alternative drilling. Using plasma for the drilling part of the operation is a ten-stage process thus:

Surface clean*apply photoresist*via exposure*via develop*via etch*photoresist strip*plasma drill*surface etch*electroless plate*electroplate

Photo definable via process is a nine-stage process:

Surface clean*dielectric coat*via exposure*post-exposure bake*via develop*UV-bump*thermal cure*electroless plate*electroplate

However, a nine-stage process results when using laser drilling:

Surface clean*apply photoresist*via exposure*via develop*via etch*photoresist strip*laser drill*electroless plate*electroplate

Brian Cosgrove then went on to make the comparison with conventional design and microvia design. He showed slides of a conventional design with eight circuit layers and a comparable microvia design with four layers. He showed a micrograph of a microvia with a 0.3mm diameter drill and human hair laid across it, and superimposed was the area covered by a conventional pad which straddled three tracks. A microvia hole of 50 micron diameter was part of the microvia circuit. He next showed cross-sections of a conventional design and a microvia design.

Brian Cosgrove discussed the strategies of different microvia designs, the first with buried vias and the remaining two with no buried vias (Table I).

The speaker showed a number of slides of break-outs and made the point that BGA microvia break-out gives greater flexibility.

Brian Cosgrove then gave a résumé of the main issues and they were testing, particularly as components are downsized and pin size goes up, hole fill and process costs.

His final comments were that with vias, size is important and that the next idea to be developed would be buried capacitors.

The second talk of the morning was given by Mr Alun Morgan, Technical Director of Isola Werke UK Limited, on the subject of "Materials for Microvia Applications". Isola manufacture Glass Reinforced Materials, Aramid Reinforced Materials, Resin Coated Copper Foil and APL-D. The speaker showed a slide of comparative reinforcement properties as given in Table II.

Table IDifferent microvia designs

		Drill			Layer
		holes	Microholes		count		Route area
Conventional
design	5918			0		8	138 sq inches
Microvia
design 1	11594			9809		4	46.5 sq inches
Microvia
design 2	8758			7498		4	32.4 sq inches
Microvia
design 3	5812			4926		4	32.4 sq inches

Table IIReinforcement properties

	CTE ppm/°C		Specific gravity		DK 1Mhz
Non-woven
aramid		­4.4		1.44		4.0
E-glass		5.5		2.54		6.6
S-glass		2.6		2.49		4.5
Quartz		0.54		2.2		3.8

Table IIIThe three types of material

	Reinforcement	Pressed
Type	thickness	thickness
2.0N710	48 microns	51 microns
3.0N710	72 microns	76 microns
4.0N719	97 microns	102 microns

He dealt briefly with glass reinforced material before moving on specifically to talk about Du Pont Thermount. The next slide showed the three types of material available (Table III).

Alun Morgan showed micrographs of holes in single and also double layer microvias made with this material and another micrograph showing the cross-section of a micro chip module made with Du Pont Thermount.

The next part of Alun's talk was on Resin Coated Copper Foil, RCCF, for microvia applications. The product itself consists of:

1. 1.

   RCCF with FR-4 Epoxy, B-Stage only.

2. 2.

   Resin thickness 60/75 micron ± 5 microns at the coating.

3. 3.

   RCCF is smooth with good handling properties, i.e. resin system with minimum brittleness and no blisters.

4. 4.

   The resin surface after pressing on oxidised copper is between 50-60 ±10 microns.

It is available with copper foil thicknesses up to 35 microns, although 12 micron foil is not available until 1999. The preferred resin thickness is 75 microns plus or minus 5 microns.

This now led on to the next phase of the talk on APL-D, a new build-up technology for continuous multilayer production. The applications are, laser/plasma via build up ML-PWB: very thin ML-PWB: impedance controlled ML-PWB and standard ML-PWB. The process consists of taking:

1. 1.

   A conventional production of etched innerlayer.

2. 2.

   Bubble free embedding of the etched structure by silk screening with APL-5004 resin.

3. 3.

   Defined UV curing of the resin matrix to produce a tack-free surface.

4. 4.

   Double action hot roll lamination of resin coated copper foil APL-1103.

5. 5.

   Post-curing by defined temperature/time profiles.

6. 6.

   Cutting, drilling, PTH and assembling as usual.

The process advantages are: continuous, fully automatic process; no lay-up station; no ML press; no pre-preg and copper foil, handling; standard silk screen equipment used and standard hot roll laminator used.

The product advantages are: laser, plasma and mechanically drillable; even and uniform insulation layer; no image transfer of the glass fabric structure; good desmear properties of the micro vias; low dielectric constant and no tension.

Alun then showed a number of slides giving the characteristics of the undercoat resin, the characteristics of RCC foil APL-1103, the properties of the APL-D build up compared with FR-4. He then put up some flow sheets showing the sequences for production of four layer boards and six layer boards using APL-D technology as well as micrographs showing the cross-section of four-layer ML APL-D build up and a section of a through hole of six layer ML APL-D build up, finishing off with a micrograph of a six layer ML APL-D build up after thermal stress. The talk finished with a video, which showed a typical sequence of manufacturing.

After the refreshment break, the second part of the morning session was devoted to the "Microvia Manufacturer's Viewpoint". The first of the speakers was Mr Steve Jones of Via Systems Tyneside Limited. Steve explained that Via Systems was a large manufacturer of via systems and that the Tyneside plant, where he was presently working, was the largest plant of its type in Europe. They have to work to exacting standards because they manufacture large boards, up to 24 inches by 20 inches, with microvias and he has to control dimensions to 175 ppm. This talk would be concentrating on the Tyneside activities of the company. The plant was designed for high volume and low cost, but complications had arisen with the falling value of the Euro, the high pound and the collapse of the Asian economy.

Via Systems is the largest PCB manufacturer in the world and has to keep the volume market. The Tyneside factory is now supplying the mobile phone market but they are also looking at ways of using their technology in other areas. They have to make strategic decisions such as do they wait for orders or do they invest in further drilling capability now? Steve Jones showed the absorption curves of FR4, Matte copper and glass versus wavelength in microns. The curve showed high absorption of all three in the UV, low absorption for matte copper and glass, but still high for FR4, in the IR end of the spectrum. Steve showed a diagram of the heat affected zone of laser drilling and this corresponds to the formula:

L = (4Kt)1/2

where K = the thermal diffusivity, t = pulse width and L = length of the affected zone.

Four slides were shown of sequences of manufacture:

1. a.

   Laser drilling using conformal mask.

   • •

     Build inner layers and laminate.

   • •

     Image and etch via pattern.

   • •

     Laser drill.

   • •

     Electroless and electroplate.

   Advantage:

   1. 1.

      high throughput.

   Disadvantages:

   1. 1.

      extra process step;

   2. 2.

      registration of via image to next layer.

2. b.

   Direct laser drilling

   • •

     Build inner layers.

   • •

     Laser drill.

   • •

     Electroless and electroplate.

   Disadvantage:

   1. 1.

      limited drill speed.

3. c.

   Laser drilling of unclad dielectric

   • •

     Build inner layers and laminate and image outer layers.

   • •

     Apply dielectric coating on the core.

   • •

     Laser drill.

   • •

     Electroless and electroplate. Semi-additive.

   Advantages:

   1. 1.

      High throughput (300 holes a second);

   2. 2.

      Smaller lines and spacings through lower copper build up.

   Disadvantages:

   1. 1.

      Extra process step.

   2. 2.

      Additive process (copper adhesion).

4. d.

   Photorigs

   • •

     Build inner layers and laminate and structure outer layers.

   • •

     Apply photo-dielectric coating on the core.

   • •

     Expose dielectric with via pattern.

   • •

     Develop via pattern.

   • •

     Electroless and electroplate. Semi-additive.

   Advantages:

   1. 1.

      high throughput in working panels;

   2. 2.

      smaller lines and spacings through lower copper build up.

   Disadvantages:

   1. 1.

      copper adhesion on photo-electric;

   2. 2.

      limited material choice.

Steve Jones explained that of the four sites comprising Via Systems, North Tyneside was responsible for prototype production, conformal mask and direct laser drilling. Their objective at North Tyneside was to: Develop high volume, low cost, high definition interconnect technology; develop leadership in the HDI volume market; technology based on large area processing and develop capability for broad application range (CSP, Mobiles, Base stations, Switching, Datacomms).

Ultimately, it is their aim to drill 900 million holes a week by the end of 1999.

They have available:

• •

  ESI 5150 (UV-YAG)ESI 5200 (UV-YAG)Hitachi LCO (CO2)Lumonics GS 600 (UV-YLFICO2)

The Hitachi is a very good machine but the problem is that they have to use a conformal mask because Hitachi do not make UV lasers. The ESI machines are superb at drilling copper. The Lumonics machine drills copper quickly but unfortunately has some irritating habits: for example, it cannot alter the scale. The ESI machines and the Lumonics machine are in the assessment stage. Steve Jones then showed a number of micrographs of drilling as well as some SEM pictures of laser drilled holes.

He showed slides of bar charts showing microvia materials costs, but their preferred materials were Thermount and resin coated foils. These were not the cheapest. They have used aramid which they like to work with but unfortunately their customers do not like aramid so they do not use it. In the course of these developments they are conducting reliability and process stability studies. They have problems with resins causing blow holes in solder and plating. These faults were illustrated with slides. The aspect ratio of the holes has to be considered because when they exceed 1:1 plating thickness falls off at the bottom of holes. They also have problems with non-linear movements which they have difficulty in countering.

They have design rules to which they adhere quite strictly. In microvia design they consider it important that the copper area on all layers should be maximised in order to minimise distortion and to minimise thickness variation on HDI layers (RCC, Thermount, etc.). Steve drew his talk to a close by examples of applications and the development of "Coaxial lines" (patent pending). These give:

1. 1.

   an increase in signal/noise ratio;

2. 2.

   use of standard PCB technology;

3. 3.

   no additional steps when used in combination with microvias;

4. 4.

   high speed transmission in back planes.

His final slide was a cross-section through the "coaxial" line.

The final speaker of the morning session on the theme of Microvia Manufacturer's Point of View was Paul Comer, Technical Director ­ Graphic plc. Why do we need microvias? Paul suggested among other reasons that there were limitations to current design methods and limitations to current manufacturing methods. There were also requirements by customers and users for more performance in smaller volume. What is required is a graphic road map which will give the options on where we want to go and how we are going to get there.

His company looked at UK requirements for microvia PCBs in design development, exploitation of the technology, investigation of small to medium volume production, quick turnaround on delivery, the use of a variety of materials, and the combination of different technologies. They came up with what Paul described as the five "ers" that is Smaller, Lighter, Faster, Cooler and Cleaner. These are given in Table IV.

Table IVThe five "ers"

	Board design
er	requirement	Design feature
Smaller	More dense	Finer/denser lines/spaces
	circuitary	Smaller holes
	linterconnect	Fewer/thinner layers
		Buried
		capacitance/resistance
Lighter	Less material or	Alternative materials
	less dense	Fewer layers
	material	Thinner layers
		Thinner copper
Faster	Higher frequency	Alternative materials
	less power	Controlled impedance
		Shorter track routing
Cooler	Removing excess	Alternative material(s)
	heat	Thick copper plates?
	Controlling thermal
	expansion
Cleaner	More EMC	Alternative materials
	compliant	Extra layers
		3-D screening

It is unlikely that it is possible to get all the requirements on one board. It will be necessary to search for other technological solutions such as microvias, feature size reduction, material selection, buried resistance, buried capacitance, thermal management techniques.

Paul Comer then introduced three more "ers" which are Quicker, Better, CHEAPER. The commercial solutions to this would be: 24-hour/ seven day turnround and flexibility; continuous improvement; design for manufacture and applications engineering.

The choice of microvia formation method depends on a number of criteria:

1. 1.

   whether the manufacturer is a small or medium enterprise (SME)costreturn on investment (ROI)versatility.

2. 2.

   Product/customer sectors, e.g. militaryaerospacemedicaltelecomsindustrial controls and test equipment.

3. 3.

   Types of board materials, i.e.rigid, flexible, flexi-rigidFR4/glassPolyimide/glassPolyimide filmPTFENon-woven aramid.

4. 4.

   OptionsMechanical Capability/productionreliabilityPlasma Material restriction/costsPhotoimageableDielectric Material restrictions/long-term costLaser Cost/technology maturityLaser/Pl Option open.

5. 5.

   OptionsExcimer Slow/expensiveCO₂ Restrictions on materialsprocessing ­ does not cutcopperNd YAG Versatile! Most mature system for PCB machining

6. 6.

   Nd YAGCuts copper and dielectricNo imaging process to expose dielectric or form holesNo plating on to bare resin on surfaceNo hole cleaning after hole formation.

Paul Comer then showed a number of slides of laser equipment, the wavelengths of the laser type used, flow sheet of a laser rail and absorption curves of materials and various wavelengths from UV to IR. He also showed micrographs of laser drilled holes as well as cross-sections of laser drilled holes.

Paul then went on to discuss the Way Forward for Manufacturers (and suppliers).

The problems to solve are imaging systems, registration systems, materials and plating.

In the case of imaging the aim should be to:

• •

  reduce photo-resist thickness;

• •

  decide on liquid or dry resists;

• •

  try to eliminate phototools;

• •

  go for laser direct imaging;

• •

  reduce copper foil thickness and reduce copper foil roughness.

In the case of registration the aim had to be for:

• •

  material stability and consistency;

• •

  non-glass reinforced laminate/pre-preg;

• •

  use of pattern recognition alignment systems at all stages;

• •

  improved mechanical registration.

As far as materials were concerned the aim was for:

• •

  material stability and consistency;

• •

  non-glass reinforced laminate/pre-preg;

• •

  copper foils;

• •

  process compatibility between;

  • •

    mechanicallaserchemical.

The spin-offs are that it impresses the customers and forces process improvements in other areas. It is not just for drilling but also for cutting flexibles and creates process improvements on flexi-rigids and 3D circuit routing.

In his conclusions Paul Comer said that most manufacturers will eventually need microvia capability. Any methods chosen should be based on company size and status, customer/product sector, product type and materials used and versatility required from the system.

Do communicate with customers on design issues and seek guidance from PCB manufacturers on design rules and materials specified. Be aware of, but not afraid of, the potential problems of each method. And finally, look for spin-offs, have fun, and good luck because we will need it!

Ken CosslettReporter