Methodology for high aspect ratio pulse plating: periodic reverse pulse tanks were once seen as a current density capacity enhancement for "standard thickness" boards. Now they churn out PCBs with aspect ratios of 15:1.

The potential of periodic reversed pulse (PRP) copper plating couldn't be ignored in 1999, when IBM Endicott began its design of a new copper plating line for manufacturing. It was decided that the line should incorporate pulse plating, but it was premature to commit all production to it. The ultimate design incorporated multiple smaller tanks, all of which could readily be converted to pulse, but only a few tanks were fitted upfront with the (more expensive) PRP rectifiers.

Tank design. Field studies provided no clear view of the best tank design--or the best source. Most data from suppliers had come from retrofitted DC tanks with whatever designs the customers had. Often the only changes that could be made were the cabling, rectifiers and bath chemistry, without any clear understanding of what "best practices" or optimal design would be. Some recommendations supported eductors; others pushed air agitation. Some promoted short anode-cathode distances; others suggested conventional spacing instead. Multiple inputs on cabling, rectifier suppliers and debates on square vs. complex waveforms had to be sorted out. Absent a clear consensus, experience and best-engineering assessments were the guides. Adding some level of flexibility into the design to accommodate some of the options also provided some insurance. The final results:

Agitation       Rack agitation (adjustable frequency).
                Flightbar vibration mounted on the flightbar.

                Air and solution agitation within a "floating shield"
                assembly.

Cabling         Twisted pairs, matched cable length, both ends of
                anodes.

                Located as close to the plating cells as possible.

Rectifiers      DC or PRP operable; Supplier: PE.
                Square wave; 6 Volt; +1200 Amp/-3600 Amp.
                Forward-pause-reverse-pause with programmable ramp.

Anode-Cathode   Set at a maximum of 12" for thicker boards.
                Adjustable up to a minimum of 6".

Chemistry. Supplier and customer visits and sample board processing failed to highlight a clearly superior process, so for project convenience (our plater was being built by Atotech) we opted for Atotech's Cuprapulse S4 Leveler system. Since this is a two-component system--Brightener (accelerator) and Leveler (suppresser)--two separate dosing support systems were required with traditional dosing based on amp-hrs.

The interaction of chemistry concentrations, rectifier parameters, current density and board attributes cannot be overstressed. The initial boards targeted for evaluations were 0.110" thick with 0.010" diameter drilled rood-sites, a standard line monitor test vehicle board used for process monitoring in Endicott. With a base assumption of operation around 20 asf for this type panel, Atotech's input for starting bath composition was:

* Copper: 19.0 grams/liter

* Sulfuric acid: 250 grams/liter

* Chloride: 65 ppm 

* Brightener: 0.02% by volume

* Leveler: 3.5% by volume.

Critical Decisions

One of the critical decisions at the onset was that bath aging, via dummy plating, would be deployed prior to running any test panels. The intent was to ensure that any results that were generated would be "real" and repeatable. This was a luxury that could be afforded due to the number of DC plating cells providing sufficient plating capacity on the new line. Testing an aged bath helped insure against "one month fall-out" scenarios and to provide reproducibility in results.

The testing of PRP presents many variables for a new user. In DC trials, current density and additive concentrations tend to be the sole variables, with all other variables fairly well defined and set to preferred values. PRP, being a complete unknown, required a plan to assess an array of new variables, primarily forward pulse width (time), reverse pulse width and forward-to-reverse current density ratio, while still taking into account average current density and the additives. Even with these wide ranges to choose from, it seemed that every supplier, user and paper on the subject of PRP recommended forward pulse widths of 10, 20 or 30 milliseconds with a fixed ratio of 20:1 for forward:reverse pulse widths. The typical current density ratios were reverse current densities of 1.5 to 3 times the amplitude of the forward.

At this point, a decision had to be made: either narrow the scope of the variables to attempt to define an operable range for manufacturing, or try to explore a more complete range of each component and include more different variables in more of a development effort. With the scope of activity so large, and the time already consumed by bath aging, we opted to define a workable set of parameters first (additional testing and optimization could be conducted later). By accepting the limited ranges on pulse widths, staying with the fixed 20:1 forward:reverse time ratio, and limiting the pulse waves to just the basic forward then reverse and repeat wave form, testing could be done quickly on a relatively small number of boards.

Test Results

Simple matrices were developed and run, looking at varied pulse widths, current densities and current density ratios. The bath chemistry was controlled to the predefined set points so as to not be an additional variable. Multiple test boards were run across multiple days using cross-sections and PTH copper thickness as the basic metric. These test runs resulted in three statistically significant "enhancers" to throwing power:

* Increasing the pulse widths.

* Increasing the reverse current density ratio.

* Increasing the average current density.

Enhanced performance with the wider pulse widths and increased reverse current density ratios were not surprising, but the improvement, although statistically less significant, with increased average current density was unexpected. The findings were welcome as they showed enhanced productivity. In retrospect, this result was most likely due to setting the initial target brightener concentration too high, thus favoring results at higher currents. Along with throw monitors from cross-sectioned boards, thermal cycling test vehicles were also processed through the various conditions. All runs showed very acceptable results, so it was concluded that process windows were fairly robust.

Using the best conditions as defined from the matrices, the ranges of the individual chemical concentrations were then tested and varied for production use. This helped to define proven conditions and specification ranges for the processing of customer acceptance product. It was obvious that these were not likely the most optimal parameters because the best conditions were found to be the extremes of what had been tested. But manufacturing conditions were identified and documented, setting a path for production to begin. Other test vehicle panels and production boards were processed at these conditions and although they had not been fully optimized for PRP, the comparative results vs. DC plating of the same product was dramatic (TABLE 1).

This allowed production of 10:1 aspect-ratio boards to begin. This also provided the benefit of more regular loading and use of the plating bath to occur, and additional testing of more-challenging boards and aggressive pulse parameters to commence. The trend of better throw with increased pulse widths continued beyond 30 ms pulse widths. Due to program limitations, forward pulse widths were only permitted to be two digits, so nothing greater than 99 ms forward could be explored. Even at the most extreme, 99 ms forward and 5 ms reverse, the increased throw performance was observed, but the grain structure of the resulting deposit was not very aesthetically pleasing. Visually comparable large-grained structures plated from DC baths in the past had been very susceptible to solder shock failures and very poor physical properties. But all solder shocks and other thermal cycling tests showed acceptable results. Still, even though all product test results were positive, we had exceeded our comfort level and settled on narrower pulse widths for production use.

To begin evaluation of thicker boards quickly, existing test boards with small clusters of thermal cycling PTH test sites were tested next. The boards were 0.220" thick with 0.014" diameter drilled PTHs and were run with only 90 minutes plating time, yielding over 0.001" of plating in the PTHs! It was noted on these and several other panel designs that the plating thickness in the smallest signal vias was the same as larger connector holes. This made parameter selection just based on specified aspect ratio impossible. Hole quantity and the resulting plateable areas were already accounted for in the definition of each part number, but not the variation in hole density or their effect on plating distribution and/or throw. Parameters used to plate test parts with a sparse number of small PTHs could not necessarily be used to plate the same aspect ratio boards with mod-sites. Often, maintaining the same parameters and adding some plating time could compensate for this. The impact of this effect finally couldn't be overcome simply by plating "a little more copper" when attempting to plate a board 0.174" thick with approximately a 3.5" square mod-site of 0.012" drilled PTHs.

Mod-sites create distribution challenges on the same order as dense lines versus isolated features in pattern plating. The thicker the board, the more tightly packed the holes, and the larger the mod-site, the greater the impact. At a thickness of 0.174", full grid 0.012" PTHs increase the plateable area 3X--that's 3 [in.sup.2] of plateable area for every square inch of board per side. With small mod-sites, this increase in plateable area can be overcome. Large rood-sites, such as the 3.5" one, do not permit system recovery and act as a big black hole. The surface copper thickness within the rood-site dropped off instantly, typically around 0.001". It was time for new parameters.

Using the big rood-site boards as the test vehicle, dramatic improvement was gained by moving to wider pulse widths (FIGURE 1). Reduction in current density showed initial improvements, but below 20 asf, PTH roughness appeared, getting more severe as the current density was further decreased (FIGURE 2). The general roughness continued to degrade to "spiky" plating at lower current densities or when brightener concentrations drifted higher. Roughness was first seen in the center of the small PTHs (FIGURE 3), but as the severity increased the roughness extended further up the barrel of the hole and began to affect the larger-sized holes, too.

[FIGURE 1-3 OMITTED]

To continue exploring lower current densities, it was clear that a separate bath with lower brightener concentration would be required. The existing production process would be negatively impacted if the brightener concentration of the bath were significantly reduced. To complement these current density and brightener reductions, it seemed appropriate to adjust the copper and other bath constituents at the same time.

To minimize any "new bath" effect, the existing bath was halved, half for the new bath chemistry and half for the existing process. New bath parameters were somewhat arbitrarily selected, focusing on the brightener reduction as the most significant change. Just as in DC plating applications, when reducing operating current density significantly, it is best to reduce the copper concentration and brightener while increasing the suppressor additive content.

With the inorganic and organic bath constituents set for lower current density plating, the most aggressive pulse parameters were retested with the average current density halved, to 10 asf. Instantly, enhanced throwing power was observed. The results were so good that little additional optimization work was needed. The only additional adjustments required were to reduce the dwell time to minimize the amount of PTH overplating that was occurring. These same parameters have also been successfully tested on other high-aspect-ratio boards with varied sized mod-sites, with no other optimization work done (TABLE 2).

Still, these results might not be optimal. Additional testing is still required to evaluate several individual and interactive variables. Boards or test vehicles that are sufficiently difficult to evaluate the parameters are also required. (Further studies are investigating 0.196"-thick boards with the same 3.5" mod-site of 0.010" PTHs.)

Reoptimization

Insights gained from the new bath regarding reducing current densities and widening the pulse widths merited revisiting the "high copper bath." The investigation of these variables in the old bath chemistry provided new capabilities and opportunities. Although reducing the current density did impact productivity, the resulting performance gains were obvious (TABLE 3).

Conclusions

The options for process combinations available today are nearly overwhelming, making it hard to claim a process is "fully" optimized. The good news is, one can still achieve significantly better results than what were previously possible in DC plating.

Some of the most basic lessons learned about PRP in the last two years at EI are:

* Clearly define the goals and product types being targeted.

* Highest throw.

* Highest productivity.

* Surface finish.

* Significantly different product types or goals will require different bath configurations.

* Lab bath or new bath results will not be attainable long-term in manufacturing.

* Bath stability and performance is degraded with intermittent use.

* Don't underestimate the capability of the process.

There are many tank design and process variables that have yet to be evaluated. Therefore, their relative importance or impact on results is still unknown. Track the base conditions: if the process is redeployed in another area, any differences should be accounted for even if their relative impact is unknown at the time.

Another significant unknown is whether the process parameters deemed best from full-panel plating would be of use in a pattern-plate mode. The addition of microvias to a high-aspect ratio has already been seen to disallow use of some of the more aggressive rectifier conditions. Pattern plating need only be treated as another board type and have new "optimal" parameters defined for it.

ISO 9001:2000 Made Easy, Cheap For the Printed Circuit Industry

Finally a firm of ISO professionals have come up with various packages that dispels the myths and confusion that surround the implementation of ISO systems while at the same time saving businesses tens of thousands of dollars on ISO certification. Unlike other ISO companies our goal is to increase your sales overnight (which to us means ISO) so we also have available various marketing packages that guarantee you will increase sales by providing you leads, contracts, sales letters & web sites with top search engine placements to make your ISO system a profitable and quality experience. Our packages have been used by Chevron, TRW and many other fortune 500 companies as well as 2-man mom and pop shops. We have many packages and payment plans to suit every budget.

Our 3 most popular package5 are:

1. Starting at only $1,995 (normally $3,995) for our complete ISO 9001:2000 software package that contains everything you need to get ISO certified including software that contains policy manual, procedures manual, forms and work instructions with 6 months technical support. 2. Our most popular package is the ISO 9000 USA.COM PACKAGE which also includes a quality policy manual custom written for your company, which makes you ISO compliant, software containing all procedures, forms and flow charts, personalized training course and 2 years unlimited technical support by the most experienced lead auditors. This package is normally priced at $6,995 but for the next 7 days it is only priced at $3,995. Both of these packages have been tried and tested by thousands of companies and are extremely simple and easy to implement with the majority of the work being completed for you and approximately 20% change is only required where you basically fill in the blanks through our step by step instructions. Both packages are in word format and just typing time alone will save you thousands of dollars. These packages are so effective that over 50% of Global QA customers to date have already had some sort of quality system in place and have purchased the package to help simplify and make more practical their existing quality system. No previous quality experience is required to successfully obtain ISO certification as our manuals are so easy to follow and we give you the most professional technical support.

3. For those companies that are too busy to complete the work themselves we have a complete A to Z turnkey pack age where we do absolutely everything for you including writing your entire policy and procedures manuals, forms, work instructions, on site training and guarantee you certification all for only $9,900 for up to 50 employees, (normally priced at $16,990). For 50 employees plus, call for individual pricing on our complete turnkey package. You will also receive an ISO 9001:2000 certificate from Global QA upon your successful completion and we guarantee you will be certified or your money back. Global QA has been in business for 8 years and we are the number one ISO firm in the USA and we also have payment plans available from only $665 per month.

Nobody in the USA can match Global QA's price, service, support and we guarantee we will beat any competitor's price and service. We have hundreds of testimonials from satisfied customers in hundreds of industries and we have the best marketing packages in addition to increase your profits and new customers.

Supplies are very limited so call Global QA now to obtain these great end of year discounts.

Global QA 800-500-7190 (All Other Standards Available)

TABLE 1. Production Comparison, DC vs. PRP Plating

                               DC PLATING     PRP PLATING

Current density                 19 asf        27 asf
Plating time                   115 min.       75 min.
Average surface Cu thickness    45 [micro]m   40 [micro]m
Minimum PTH Cu thickness        20 [micro]m   22 [micro]m
Average thermal cycles          25 cycles     43 cycles

0.140" thick production board with 0.014" diameter drilled PTHs (10:1
aspect ratio)

TABLE 2. Results of Lower Current Density Plating

BOARD THICKNESS    DRILL     ASPECT      BOARD      THROW *
                  DIAMETER    RATIO   DIFF FACTOR

147                  10       14.7       2161         90%
174                  12       14.5       2523         75%
250                  16       15.6       3906         70%
365                  26       14.0       5124         59%

* Throw was determined by dividing the average of the center PTH
reading over the total average surface plating.

TABLE 3. Results of Reoptimized Baths

                               DC PLATING     "OLD" PRP     "NEW" PRP

Current density                 19 asf        27 asf        22 asf
Plating time                   115 min.       75 min.       87 min.
Average surface Cu thickness    45 [micro]m   40 [micro]m   38 [micro]m
Minimum PTH Cu thickness        20 [micro]m   22 [micro]m   27 [micro]m
Average thermal cycles          25 cycles     43 cycles     56 cycles

0.140" thick production board with 0.014" diameter drilled PTHs (10:1
aspect ratio)

Ed: This article is adapted from a presentation at the IPC Fall Meeting, Oct. 2003.

ACKNOWLEDGMENTS

I would like to thank Dr. Robert J. Day, for his pioneering efforts in evaluating PRP in Endicott; the many members of the Copper Plate manufacturing department, without whom this work could not have been performed or evaluated; and Atotech's field and corporate engineering, for support and guidance.

BOB EDWARDS is a copper plate manufacturing engineer at Endicott Interconnect Technologies (endicottinterconnect.com). He can be reached at [e-mail].