Issue



Green IC packaging


08/01/2001







Reinventing the integrated circuit is more than just a lead-free issue

BY JEFF CANNIS

Building a totally environmentally friendly integrated circuit (IC) is an elusive goal as long as semiconductor die are doped with arsenic and phosphorus. However, lead-free and halide-free packages for these die go a long way toward supporting current goals for environmentally safe electronics.

Lead is certainly on the way out of the electronics industry. Driven more by electronics OEMs than by government regulators, lead is beginning to disappear from electronic products, and it is being replaced faster than many expected. Of course, it is likely that the OEMs were impelled to action by the approaches regulators have taken in a number of countries. In Japan, some European countries and some U.S. states, regulators are pushing for buy-back programs under which OEMs would bear the responsibility for collection and recycling of electronic products, rather than have their customers toss them into the trash. The less hazardous materials those products contain (including lead), the easier this will be for the OEMs.

Worldwide Response

Japanese companies have been the quickest to reduce lead content and, because Japan supplies most of the world's consumer electronics, it is expected that this will have a global impact. This past year, all of Japan's biggest consumer manufacturers announced accelerated schedules for eliminating or sharply reducing the use of lead. Hitachi, Matsushita (Panasonic) and Sony have all planned for consumer products to be lead-free this year. Toshiba has announced plans to do the same for cellular phones by next year, and Fujitsu intends to produce all lead-free products by December 2002. By the end of 2002 or the middle of 2003, all major Japanese electronic OEMs are planning to be 100 percent lead-free.

There is more to this than promises. Last year, for example, NEC introduced computer notebooks with lead-free motherboards. In a follow-up, the company said that by 2002 its total lead usage will be half what it was in 1997.


Table 1. Compositions and melting points of lead-free solder formulations.
Click here to enlarge image

One factor spurring these companies to action was Japan's Home Electronics Recycling Law, which required OEMs to be prepared to collect and recycle certain major products by last April. Another Japanese law forbids companies from putting any kind of waste-leaching toxic elements into the environment.

In a strategic response, the manufacturers chose to make "lead-free" a product differentiator. Preliminary results seem to justify that tactic. After Panasonic placed a lead-free sticker on a minidisk player, it reportedly saw a 13-percent increase in sales.

Japan's regulators imposed nearer-term deadlines than regulators in the U.S. and Europe. While Europe has a patchwork of local initiatives in different countries, the broadest effort comes from several proposals under consideration by the EU European Council (EC). The Waste Electrical and Electronic Equipment (WEEE) directive addresses the collection, recovery and recycling of electronic products. The related Restriction of Hazardous Substances (ROHS) directive seeks to reduce the risk of use, treatment and disposal of the same products. These directives would effectively ban the use of certain listed materials, including lead, from circuit boards. As it stands now, the elimination date for these substances is January 2006. A related EC recycling regulation, the Electrical and Electronic Equipment (EEE) Directive, addresses end-of-life reclamation.

In the U.S., there are measures being considered in several states that would require recycling, and the U.S. Environmental Protection Agency has recently required industry to report discharges of lead into the environment at much lower usage levels (10 vs. 10,000 pounds) than in the past.

Lead-free IC Packages

In electronics manufacturing, board assembly represents a larger lead issue than IC packages, but packaging is a part of the total lead-free story. Recent efforts have been concentrated on the tin-lead plating on the lead frames in packages, such as quad flat packs (QFPs), small-outline packages (SOPs), and in the lead content of the balls on area-array packages.

On lead frames, nickel-palladium with a gold flash has historically been the most common substitute for solder lead plating. A 0.02 to 0.15-micron layer of palladium is deposited over a 1 to 3-micron plating of nickel, which is then coated with a 30 to 100 Å gold flash. The palladium protects the nickel from oxidation, and the nickel keeps the base metals in the lead frame from diffusing into the palladium. The gold improves wettability.

Factors discouraging the widespread adoption of nickel-palladium include reliability, supply and cost. On the one hand, Texas Instruments has done much to allay concerns about nickel-palladium joint reliability. TI pioneered the use of nickel-palladium in the late 1980s and has produced billions of components with this type of termination finish. They have frequently published research supporting the reliability of solder joints made to nickel-palladium.1,2,3

A large fraction (75 to 80 percent) of the world's palladium comes from Russia and South Africa, however, making supply a matter of some concern. Reflecting this, the price for one troy ounce of palladium has risen from less than $150 before 1997 to approximately $700 at the end of April 2001 (after an earlier peak around $1,050). Issues of cost and supply are of far more concern to the automobile industry, which uses most of the world's annual output of palladium in catalytic converters. Palladium metal cost is less of a factor to the electronics industry, where the raw-materials cost of a thin flash of the metal on an IC lead frame is not a large component of overall packaging cost. A number of lead frame suppliers offer nickel-palladium at competitive prices.


Figure 1. Representative reliability data.
Click here to enlarge image

For leaded packages, the simplest option would be to use a pure tin finish. A move in this direction would bring the industry back full circle to where it was decades ago, when tin plate predominated.4 Elemental tin does present a potential reliability problem - short circuits caused by "whiskers," which are elongated single crystals of pure tin that grow from the surface of the metal under stress. These crystals are described as whiskers because they're typically around 3 µm in diameter and up to 4 mm long - just right for shorting adjacent leads. In conventional tin-lead solder, the assumption was that lead keeps the tin from forming whiskers. In fact, it was thought that only a few percent of lead added to tin is sufficient to suppress tin whiskers from occurring. Thanks to the recent increase in whisker-related research, we are now finding that even tin-lead depositions can form whiskers under the appropriate conditions. It now appears that most high-percentage tin deposits are susceptible to whicker growth.5

The industry appears to be split on the issue of lead-free plating. Much of the Japanese market is focused on various tin alloys, most notably tin-bismuth (2 to 3 percent Bi). European countries, the U.S. and some Asian regions have a strong following for pure tin. Tin-silver and tin-copper also have scattered support but at a much lower level. The final chapter has yet to be written.

Options for Area Array Packages

For ball grid array (BGA) packages and the assortment of chip scale packages (CSPs) that use solder ball terminations, the industry effort has focused on identifying a lead-free alloy that provides good solder joint reliability at the lowest possible melting temperature. Pure tin is not the answer, however. Elemental tin by itself doesn't wet the surfaces being joined as well as tin-lead solder, and it melts at 232°C - too hot for the conventional organic encapsulating compounds used in semiconductor packaging. A vast array of exotic alloys has been under consideration,6 but the confusion a few years ago appears to be resolving itself. Table 1 summarizes the compositions and melting points of a number of lead-free solder formulations that have been evaluated.

Of the hundred-plus alloys previously under consideration, most companies are now focusing on the tin-silver-copper family of ternary alloys. The addition of small amounts of silver and copper improve wetting characteristics, lower melting temperature and increase solder joint reliability. A significant decrease in melting point can also be achieved by adding bismuth or zinc. But bismuth raises supply issues similar to those for palladium, because much of the commercially available bismuth comes from Bolivia and China. A percentage of bismuth production is a direct result of lead mining, which brings into question the environmentally viability of bismuth. Cost and available of zinc are not an issue, but the long-term solderability of zinc alloys is problematic. The oxidation of zinc necessitates the use of specialized fluxes and soldering process conditions (nitrogen blanket) to achieve acceptable solder joints. Despite the general migration to tin-silver-copper alloys, it is likely that standardization is still years away.

The tin compounds discussed can be used for the balls on the exterior of ball grid array packages in which die are wire bonded to a tape or laminate substrate inside the package. However, packages that employ flip-chip-in-package (FCIP) assembly, where die are attached to substrates using a solder-reflow process, introduce a complication. Ordinarily, the balls used inside the flip-chip package are eutectic tin-lead. However, specific high-end reliability applications call for the use of alloys with 90 to 97 percent lead. These materials have melting temperatures above 300°C. Lead-free solutions for both types of applications will have to be identified.

For lead-free FCIP, there are a number of possibilities. The most promising technique does away with solder balls inside the package entirely. Instead, the bond pads on the die are gold-bumped using standard wire-bonding techniques. Essentially, the wire bonder thermosonically attaches a gold wire to each bond pad on the chip, just as in conventional wire-bonding assembly, except that there is no package at this point in the process. Instead of being fastened to some package substrate at the other end, the wire is broken off and recycled, leaving a gold bump on each bond pad on the die. (This is in lieu of the ball-attach process that takes place in conventional FCIP assembly.)

After the die is gold bumped, high tin-content solder paste is applied to the substrate, chip and substrate are assembled, and the solder paste is reflowed. As with conventional FCIP-assembled products, the solder paste has a melting point substantially higher than the balls used on the exterior of the package and there is no further reflow when the package is attached to the system board. Gold stud bumps have the additional advantage of being able to be used at wire-bond pitches on die designed for wire-bond assembly, easing the transition into flip chip packaging.

Reliability

There is now a substantial body of data stating that in terms of second-level failures (between package and board), lead-free packages are as reliable or even somewhat more reliable than equivalent conventional packages.6,7,8,9 Thermal cycling tests are typified by Figure 1, which shows that solder joints made with Sn/4.0Ag/0.5Cu solder balls exhibit 25 to 200 percent longer life.7 Testing has been conducted on PBGA and 144-ball BGA packages (which use a flexible tape substrate to minimize mounting height).

Halide-free

Lead is only one aspect of what is becoming known as the "green" package. The other significant aspect is the use of bromine- and antimony-containing flame retardants in mold compounds and in area-array package substrates. As with lead, the larger issue affects system assembly - in this case boards and enclosures - but, again, packages are part of the total picture. These flame retardants are based on a combination of bromine and antimony oxide. Halogens, including bromine, are only weak fire retardants, and antimony oxide by itself is not a fire retardant, but they become very effective when combined.

The combination retards flames in two ways. During burning, antimony oxide promotes the formation of "char" (essentially carbon), which reduces the formation of volatile gases. At the same time, the heat of initial combustion promotes a cross linking between the organic compound and the antimony, which results in a more stable thermoset polymer. In addition, at temperatures above 315°C, bromine forms hydrobromic acid, which reacts with the antimony oxide to form antimony trihalides and oxyhalides that trap free radicals, inhibiting ignition and pyrolysis.

The objection to bromine/antimony flame retardants is what happens at the end of a product's life. According to the WEEE draft proposal, when electronics treated with bromine flame retardants are recycled, they can generate dioxins and furans. (Dioxins and furans are based on a common chemical "skeleton" onto which one to eight chlorine atoms can be attached in a variety of positions. The different combinations yield 75 distinct dioxins and 135 different furans.)

In the mid 1980s, studies showed that toxic polybrominated disbenso furans (PBDF) and polybrominated disbenso dioxins (PBDD) were formed during the extruding process, which is part of the plastic recycling process. There is also evidence that polybrominated diphenylethers (PBDEs) might act as endocrine disrupters, and high concentrations of PBDEs have been found in the blood of workers in recycling plants. Moreover, polybrominated biphenyls (PBBs) have been found in Arctic seal samples, which indicates a wide geographical distribution from both PBB manufacturing and waste dumps. Once PBBs have been released into the environment, they can reach the food chain.

On the packaging side, qualified suppliers can produce substrates that do not contain halides or antimony, so there is no problem meeting the EC requirements. In the larger world of complete systems, it is not so clear that there are easy substitutes for bromine and antimony.

A controversy arises because in the U.S., the State Fire Marshals Association has indicated it will not accept consumer electronics that contain fiberglass-reinforced circuit boards not protected with brominated flame retardants. In a recent press release distributed to European publications, the association indicated it will call for a trade embargo on any consumer electronics manufactured without brominated flame retardants. It remains to be seen how this will be resolved.

Summary

Lead-free interconnects and halide-free substrates are the essential ingredients of the "green" IC package. In the larger scheme of things, packaging makes only a very small contribution to the total amount of heavy metals and halides that are added to the environment when electronic equipment is scrapped. However, the semiconductor assembly and test segment of the industry, responding to the demands of its chipmaker and OEM customers, is doing its part by making lead-free and bromine-free packaging available for chip vendors that require it.

AP

References


  1. D.W. Romm, B. Lange, D.C. Abbott, "Evaluation of Nickel-palladium-finished ICs with Lead-free Solder Alloys," IPC Works 2000 Proceedings, 2000.
  2. D.C. Abbott et al., IEEE Trans. CHMT, 14(567), 1991.
  3. D.W. Romm, D.C. Abbott, "Lead-free Solder Joint Evaluation," SMT, March 2000.
  4. S. Winkler and B. Hom, "A Look at the Past Reveals a Lead-free Drop-in Replacement," HDI, April 2001.
  5. R.A. Schetty, "Minimization of Tin Whisker Formation for Lead-free Electronics Finishing," IPC Works 2000 Proceedings, 2000.
  6. "Lead-free Solder Project Final Report," National Center for Manufacturing Sciences, August 1997.
  7. A. Syed, "Reliability of Lead-free Solder Connections for Area-array Packages," APEX 2001.
  8. G. Swan et al.,"Development of Lead-free Peripheral Leaded and PBGA Components to Meet MSL3 260°C Peak Reflow Profile," APEX 2001.
  9. NCMS High Temperature Fatigue Resistant Solder Program.
  10. Bartelo et al., "Thermomechanical Fatigue Behavior of Selected Lead-free Solders," APEX 2001.

Jeff Cannis, senior manager of process engineering, can be reached at Amkor Technology, 1900 South Price Road, Chandler, AZ 85248; 480-821-2408 x5207; E-mail: [email protected].