Here is some additional info. about Gil Olachea. I am finding out he has also been involved in BGA ball grid array technology.
"Gil Olachea, VP of corporate marketing and communications at Amkor Electronics, will leave at the end of the month to join Intercell, a public holding company of other technology companies, as a corporate staff member. Mr. Olachea will be based in Phoenix, as he is now. Intercell is headquartered in Vancouver, B.C., but will be moving to Boulder, Colo. Mr. Olachea will be involved in sales, marketing, technology and manufacturing activities. "
For the geeks that enjoy technical information, I am including an article about BGA and its importance in the semi market. PIC has a working technology using diamond dust connectors that uses the BGA principals.
Ball Grid Arrays, Half Way There
"It is now four years since Ball Grid Arrays entered the popular lexicon. We are approximately half way through the typical seven year gestation period that a significantly new package needs for broad acceptance. There are still only four major users of ball grid arrays in volume production (Motorola, IBM, Compaq and Bosch).
However, there are more than 40 companies that are in various stages of implementation, ranging from limited production, through design wins to technology qualification (LSI Logic, Sun, Texas Instruments, Altera, Xylinx, Fujitsu, etc.). The applications are almost exclusively high leadcount ASICs and processors, with fast SRAMs the dominant low leadcount application (Motorolas 119 I/O SRAM being an example).
The proliferation of BGA design configurations and material sets has accelerated. The permutations and combinations are extensive and the industry is far from settling on a standard configuration and material set from which to derive an authoritative set of qualification and reliability data. The industry is still experimenting with alternative configurations and material sets to meet the constantly rising bar of cost, reliability and performance.
The electronics industry does not wait for the packaging engineers to do their work before acceptance of a new package, which is why the package acceptance process typically takes seven years with extensive concurrent development and deployment. In reality, industry wide acceptance of ball grid array technology is right on track for a new package and a tremendous amount of good work is being done to address the moving target of BGA issues. We address some of the more important of these issues in this bulletin.
Cost
The ball grid array is still an expensive package. At leadcounts in the 225 to 313 range it is approximately twice as expensive as the equivalent quad flat pack (plastic BGAs are currently selling for 1.9 to 2.2 cents per pin - PQFPs are typically 0.9 cents per pin). Even with the system level assembly yield benefits that BGAs provide (5 dpm vs. 100 dpm for 0.5mm pitch PQFP) this is an unacceptable premium for BGA packages unless device access time and/or high leadcount devices are mandatory.
The total electronics industry high leadcount requirements are still comparatively low at 0.4% of the die population (194M die in 1994), but are forecast to rise to 2.7% of the die population (2.9 Bn die in 2004). Perhaps the best advice is given by Gil Olachea, Marketing Manager of Amkor Electronics who states, "If you are not using high leadcount devices today you surely will be tomorrow. The ball grid array represents the best package alternative in terms of cost, performance, handling and assembly. Prepare for tomorrow. Get involved with BGAs today."
The material cost structure of a typical BGA provides an insight into why these packages are still substantially more expensive than a PQFP. Simply put, a leading edge technology printed circuit board or a cofired header is more expensive than a stamped leadframe until the leadcount and electrical performance takes the package into a domain where the PQFP cannot easily follow.
Even so, the electronics industry expects ball grid arrays to quickly progress to its package benchmark of a penny per lead. There is a strong probability that this will be achieved, and the focus of many development efforts is to reduce the high header cost -- the principal cost component in BGAs. Clues as to how this will be done exist in IBM's film grid array. Sheldahls recent relationship with Texas Instruments, developments at Hewlett Packard Palo Alto, and Olin's soon to be available metal ball grid array all point to single, thin dielectric film on metal arrays.
In all these approaches a single signal layer thin film or flex film fan out layer provides:
A TAB like cost structure (high volume TAB sells for approximately 70 per square inch versus $1.00 to $3.00 per square inch for high quality or double sided boards or multilayers). Commercially available chip access line widths/spaces down to 50/75 microns. A more consistent wire bonding surface when compared to laminate for higher assembly yields. A thin electrically predictable interface between die and board. A migration to a flip chip array configuration.
As the leadcount rises these advantages become more significant, and for these reasons we expect the enhanced plastic ball grid array with a thin insulating film (polyimide equivalent) to become the dominant ball grid array package. Today, BGAs are a more expensive package than the alternative. Tomorrow, there is really no alternative, and relative price differences between BGAs and alternatives will continue to decline. As we have seen so often in the past, "You can never beat an old technology on price but an old technology can never grow to create new headroom."
Device Reliability
There are two aspects to ball grid array reliability: (1) the package itself, and (2) the package to board interface.
All surface mount packages are sensitive to moisture absorbance. The surface mount reflow process is the most vicious experience of a package's life. Packaging and surface temperatures may exceed 220 C with temperature ramp rates of up to 10 C per second. The ball grid array package is particularly susceptible to delamination because this material set is not optimized to limit moisture absorption. Bad actors are the solder mask, the board laminate material, the die attach material and an unbalanced physical structure.
JEDEC standard A112 has six levels of package moisture performance. These levels are the length of time that the package may remain exposed on the assembly floor prior to reflow soldering. They range from unlimited atmospheric exposure before assembly to just six hours. Exceeding these assembly floor exposure times requires reheating for typically 24 hours at 125 C or prior storage in dry nitrogen. These added costs can be expensive, involving reprocessing tape and reel parts into high temperature baking trays and back again as well as the oven time. One advantage of ceramic ball grid arrays is that they are immune to these problems. Typical conventional plastic packages conform to levels 3 to 5 (JEDEC A112). Ball grid array packages are still inferior often requiring rebaking after 24 hours of assembly floor life exposure. The battle to reduce the moisture weight gain of the BGA package continues with a focus on the printed circuit substrate material, the solder mask, and the die attach material.
We have no doubt that BGAs will be proven to have equivalent delamination resistance to conventional plastic packages with supporting data over the next 12 to 18 months. Ironically, that is probably the time when new polymer film based material sets will be introduced for enhanced plastic ball grid arrays. We believe that this construction will further improve the reliability of the BGA structure. BGA to printed circuit substrate reliability is affected by a number of factors:
The use of a solder mask or lack of a solder mask at the ball pad interface Area array versus perimeter array The die size The rigorousness of the accelerated thermal cycling test which is used to simulate the end use environment The use of balls or columns in ceramic BGAs The metallurgy of the solder ball attachment The die thickness The modulus of the die attach material
Extensive tests have shown that maximum BGA to board reliability is achieved with a peripheral array package on a thick board substrate, with a thin die, attached with a low modulus adhesive, with the assembly using a non melting 90/10 ball and a eutectic interface, with a precisely defined solder pad. Not surprisingly this is the technology direction that the BGA users and suppliers are pursuing with qualification testing playing catch up to the combinational effect of every permutation as it is evaluated and incorporated.
How reliable are ball grid array assemblies? Not as reliable as the plastic quad flat packs that they are replacing but reliable enough for office and external communications environments. Automotive under the hood will take a little longer, perhaps a lot longer.
The Future and the Opportunities
Ball grid array usage is still in its infancy. The ball grid array is the preferred package for high leadcount devices and low leadcount fast access devices. Despite the apparently slow progress in solving the problems associated with BGAs, solid enquiry work is proceeding which will ensure that the BGA becomes the only cost effective reliable package in its domain. The enhanced plastic ball grid array with a metal backplate, film circuit and glob top will become the dominant package because of its combination of cost, reliability, die access and thermal performance.
Opportunities for suppliers are DIE ATTACH where the issues are well recognized and solutions are in hand, GLOB TOP where ongoing improvements proceed to molding compound reliability equivalency , FILM CIRCUITS as a die to board interface, in its infancy and possibly with density cost advantages, composite METAL HEADERS incorporating TCE matched ball attach areas. " |
