The Art of Probe Cards and Beyond
By Andrei Berar, consultant
I do not know how others are, but when I think of the early days of wafer probing, with single site, few pins, huge pitches, and hot probing at 85°C, my heart jumps for joy! The probe card industry has made tremendous progress in a relatively short period of time driven by two significant vectors: shrinking geometries and reducing cost. Figure 1 shows an overall size of the markets and market geometries requirements, combined with the drive for multi-sites in order to reduce cost.
Figure 1: An overall size of the markets and market geometries requirements, combined with the drive for multi-sites to reduce cost.
Probe card manufacturer's entropy is still very high with new players entering the marketspace, and old players re-inventing themselves, but most significant are the innovative technologies.
Innovation comes with a large R & D ticket price that is temporarily rewarded by the market, but is not sustainable, especially when combined with the "normal" cycles of the semiconductor industry. Figure 2 shows the return on capital research (ROCR) for four publicly traded probe card companies.
Figure 2 ROCR (Return on Capital Research = Gross Margin / Previous year R & D spending) for four major probe card companies that are publicly traded.
Sustaining a high return on R & D spending is difficult, and in the long run it settles towards a less than $2 return on each $1 invested in R&D. This low return is not enough to maintain the financial health of the company, considering also that most of the time the returns take much longer than one year to come to fruition and only the winning technologies are visible. Many unsuccessful developments also drain a lot of capital, with minimum to show for it. It should be noted that Cascade Microtec (CSCD) maintains an average higher ROCR, which probably is due to the fact that probe cards represent only 40% of their portfolio.
R & D spending is a symptom of two facts: the overall process of wafer probing is too complicated and the market size is not large enough. Probe cards are composed of parts or modules that can be more or less simple. However, complexity grows exponentially as more modules are combined to produce what we perceive the market wants. Adding more to an already crowded space does not leave much room for innovation. What is needed is the ability to simplify the complex and do it at scale.
It is also important to note that besides the development cost for the probe themselves, a significant development cost and time is allocated to the overall probe cad assembly, where many components require certain electro-thermal-mechanical properties and accuracy. This, sometimes overlooked cost, is very well described by Ira Feldman from Feldman Engineering, who presented a paper at IEEE Southwest Test Workshop in June 2011 entitled " Probe Cards Cost Drivers from Architecture to Zero Defects."
The invisible hand of the market has begun to recognize this conundrum and is providing guidance to address it. There are two vectors that could address the future health of the probe card market: First, simplify the wafer probe process to provide the ability to use standardized probing modules without losing the efficiency of multi-sides testing. Next, expand the market beyond probe cards, and take advantage of the strong and mature core R & D for collateral markets such as sockets and interconnects markets outside the semiconductors area (i.e. medical instrumentation, bio-tech)
Simplifying the Process
Implement a probing process that can take advantage of probing modules, a process that does not test the die on the wafer, although this may be counterintuitive. The die will then have to be singulated and positioned at a pre-defined center-to-center position. The process would be in line with the existing proposal for Test in Tray, where all the parts that fit in the tray will be tested in parallel with minimum index time (Figure 3). The trays are also an excellent carrier to perform tri-temp in one insertion.
Fig. 3 — Test 'n Tray singulated die and package (Courtesy of Centipede Systems)
Using this singulated die approach, the overall efficiency of probing becomes 100% since there are no double touches, stepping off the wafer or touching known bad die. The modules alignment is relative easy, since it is one module per die. The modules will be easily replaceable (like printing cartridges) one at a time with minimum interaction of the production process, or requiring the removal of a full probe card for repairing only a few probe tips.
Moving forward to the introduction of 450mm wafers, it will be practically impossible to handle this large size wafer without very costly automated equipment. Using singulated die probing ensures that most of the infrastructure is already in place, and therefore no major R & D will be required.
Considering the increasing popularity of wafer level chip scale packages, (WLCSP) having the die in trays will allow for a standardized handling process without removing the die from the trays, thereby minimizing any possible handling accidents. More users are using probe cards to test the packaged parts and also testing singulated die with modules of clustered probes.
Expanding the Market
The market has the potential to grow by expanding into the package test market, which is less R & D heavy and primed for innovation to address smaller geometries at higher speed and lower price/pin. Figure 4 shows the possible combined market size based only on test sockets. There are burn-in sockets and production sockets market that can be addressed as well.
Figure 4: Possible combined market size based only on test sockets.
Testing singulated die and packaged parts with the same probe card is already used in engineering environment (Figure 5). What will propel this solution to the next-generation process is again the concept of standardizing the back-end test process from wafer probe to shipped packages by using Test in Trays. The advantages are significant, as with any scalable simplified process as previously described. The same advantages for singulated die also apply to package testing.
Figure 4: Testing x8 singulated die (courtesy of BucklingBeam Solutions) and x8 packages
Test in tray will have a large impact on the ATE equipment (probers and handlers) and also on probe cards and sockets manufacturers. It will provide the ability to use technologically advance probe tips in high-volume, easy to manufacture modules that will address the smaller geometries, provide multi-DUT solutions limited only by the tester's resources, tri-temp testing in one insertion, and competitive ASP with high margins.
Probing Beyond Semiconductors
Addressing market opportunities outside semiconductors applications is less challenging from a technical perspective than from a business ability to penetrate this markets.
According to the technical roadmap for medical equipment, biotech applications are following a similar path towards smaller geometries and robustness. Making the move to new markets may require a new organizational infrastructure and new leadership working in parallel with the probe card developments. Microfabrica is one company that made the transition. After developing probe tips for probe cards and developing the operations capability for MEMS high-volume manufacturing while also investing in solutions for medical equipment, it has now dedicated itself to the medical market.
It is not a simple or easy road to get there, and many roadblocks will be challenging, but the movement has already started, and it must be on every probe card and socket manufacturer strategic road map for surviving.
We are in for a rewarding technology ride that could bring the original thrill back, as soon as the R & D investment brings significant returns.
Andrei Berar, consultant to the semiconductor industry, may be contacted at email@example.com.