BY GLENN RINNE
 |
The most common modern use for lead is in solders for electronics manufacture because of lead's low melting point and corrosion resistance. Historically, lead has been used in a wide variety of ways for thousands of years. However, many health and technological problems have been attributed to lead. Therefore, there are ongoing efforts by engineers to eliminate the technological difficulties, and efforts by regulators and lawmakers to restrict its use through regulation. Lead occurs naturally in four stable isotopes, and as many as 20 different radioisotopes have been identified. During the past 20 years, there has been growing pressure to eliminate a particular isotope of lead from electronic solders: the radioisotope 210Pb.
To give an overview, the problem with 210Pb in solder is that is comes from the radioactive decay of polonium-210 (210Po), which releases an energetic alpha particle (helium nucleus). Although it is easy to remove the 210Po from the solder, the radioisotope 210Pb will decay with a half life of 22.3 years to 210Bi, which, in turn, decays with a half life of five days to 210Po, so the threat returns. This point, where 210Po is being replaced at about the same rate that it decays, is called “secular equilibrium” and is reached approximately two years after refinement.
Alpha Particles
The discussion of radioactive decay goes back some time in our industry. In the mid 1970s, engineers at IBM traced a particularly troubling random error to the emission of alpha particles from solder on their flip chip devices. Nuclear physicists working with precise measurement of infrequent decay events were aware of the alpha emissions of lead and determined that it is necessary to eliminate both the 210Po and the 210Pb to permanently reduce the alpha emission rate.
Ordinary lead has an alpha particle emission rate in the range of 1 to 30 alphas per hour per square centimeter (alphas/hr/cm2) or, more commonly, 1 to 30 counts per hour per square centimeter (cph/cm2), depending on the source mine. For current generation integrated circuits, a goal of 0.01 cph/cm2 is typical, although there are some particularly sensitive circuits, such as dynamic logic, that require 0.002 cph/cm2. Therefore, the concentration of 210Po must be reduced by a factor of anywhere from 100 to 15,000. This can be accomplished by either letting the 210Pb decay, or removing the 210Pb.
Depletion Through Decay
It might seem that lead found in the earth (which is about four billion years old) should be old enough that all the 210Pb should have decayed. Unfortunately, new 210Pb is continually created through a chain of radionuclides that begins with uranium-238 (238U). Lead ore must first be refined (smelted) to remove the uranium and its nuclides before the 210Pb decay and 210Po can proceed to completion. Then the wait begins. With a half-life of about 22.3 years, the concentration of 210Pb will be reduced by a factor of 15,000 in log2 (15,000) = 13 half-lives or 310 years. Of course, this will result in time-to-market problems.
Antique lead (lead that was smelted centuries ago), has been used for many years as a low-alpha emission shielding for nuclear particle research and is now available for the microelectronics industry. Lead ballast from shipwrecks more than 250 years old is now being recovered for use in microelectronics. Testing has shown activity levels of less than 0.01 cph/cm2 in these materials and some wrecks have yielded lead ballast with activities below 0.002 cph/cm2.
Unfortunately, ships from the 15th century were not large by modern standards and contained only about 50 tons of ballast. To put this in perspective, one shipwreck will supply enough lead for about three years of flip chip production at a wafer bumping service. Clearly, the industry will consume the ballast of all the known shipwrecks in a short time. Thus, there is a pressing need to find a method to purify contemporary lead.
Depletion Through Removal
Because 210Pb is almost chemically identical to its stable isotope siblings, removal of the 210Pb is a fairly difficult task requiring techniques developed for separating radioactive isotopes for nuclear weapons and reactors. To achieve an activity of 0.002 cph/cm2, the concentration of 210Pb must be reduced to about one part in 1017, a level of purity uncommon in the electronics assembly industry. A Russian method of isotope separation vaporizes the lead in an evaporator, ionizes the 210Pb using a dye laser, and separates the 210Pb from the vapor using a strong electrostatic field. The purified vapor is then condensed. This process is currently producing enough 0.002 cph/cm2 lead for approximately 120,000 solder-bumped wafers per month. This capacity will need to increase by a factor of at least 10 to meet the projected needs of the microelectronics industry.
The Green Movement
Throughout history, we have concentrated and distributed lead throughout our living spaces. We have spread it on our walls as paint, sprayed it on our food as a pesticide and burned it into the air as anti-knock compounds in motor fuels. It is not surprising that such use has resulted in many unfortunate consequences. (As early as 100 BC, Greek physicians reported toxicological problems with lead. In 1786, Benjamin Franklin linked illness in typesetters to the lead used to make moveable type. Only in the past 50 years, however, has the toxicity of lead been intensively investigated.) It is for these reasons that legislation has been enacted in Europe to restrict or eliminate the use of all isotopes of lead in electronics.
It is often assumed that lead-free solders will have no alpha activity, so conversion to lead-free solder solves both problems. In practice, lead-free solders do have alpha activity high enough to be of concern in microelectronics. The primary source of these alpha particles is 210Po, as in lead-based solders. Lead is a common contaminant in most commercial tin sources, often to a level of 100 ppm or more. For sensitive circuits, then, it is necessary to specify a low-alpha activity lead-free solder.
Silver is an increasingly popular choice for use as an alloy to create lead-free solder. However, one of the major sources of silver is in lead ore deposits. Therefore, silver must be sufficiently refined to reduce the 210Pb to acceptable levels. This is not a concern for Sn3.5Ag solder, because the silver concentration is fairly low, unless very low alpha activity is required. In such cases, high purity silver should be used. Bismuth and antimony also contain trace amounts of lead and should be specified carefully, as well.
In a Nutshell
Although lead is responsible for many medical and technological problems, it is a naturally occurring metal and has been an integral part of human existence for centuries. It has been extensively distributed throughout our environment by way of a wide variety of human activities. Currently, it performs an important function in some of our most critical technology. Although we can attempt to reduce the inherent health risks and eliminate the technological problems lead presents, we will not be able to completely eliminate either by legislation.
AP
Glenn Rinne is vice president of research and development at Unitive Advanced Semiconductor Packaging, P.O. Box 14584, Research Triangle Park, NC 27709-4584; 919-941-0606; Fax: 919-941-5097; E-mail: [email protected].
Testing Low-alpha Lead
Radioactive decay is a probabilistic phenomenon; an atom of a radioisotope placed on your desk may decay today, tomorrow or a billion years from now. We can't predict it in advance. By using statistics, however, we can say that, of a large population of such atoms, a certain percentage will decay during some time period.
Any such predictions have a margin of error, and this error gets smaller as the sample size gets larger. If we want to test a sample of low-alpha solder, how long must we test it and how large is the margin of error?
Here's an example: Suppose we want to measure the alpha emissions from a 1cm2 sample. Suppose, also, that we want to be 90 percent confident that the solder has an alpha activity no greater than the number we measure. Statistics tell us that the error of a measurement (E) of random events is equal to one divided by the square root of the sample size (N).
E = N-0.5
And, in this case, the margin of error should be:
E = 1 – 0.9 = 0.1
Solving for N we have:
N = (1/E)2
N = 100
Therefore, we need to count until we have measured 100 alpha particles to be certain within the margin error of 10 percent what the rate really is.
Now, if our solder is supposed to be 0.01 cph/cm2 activity, we would have to measure for 10,000 hours. Clearly this is unrealistic. To reduce the test time, we could increase the sample area to 1,000 cm2, which would require only 10 hours of testing to get 100 counts.