14
Semiconductor Chip Protection as a Case Study
MORTON DAVID GOLDBERG
Intellectual property is old. Semiconductor chips are new. I have been asked to look at how the two have worked together, and what we can learn from the experience.
First, I review the history of the technology and of the intellectual property law that Congress custom-tailored to protect it. Then the provisions of that law and what some of us see as its shortcomings are discussed by focusing specifically on three areas: how the law defines the technology, its broad exceptions to proprietor's rights, and the difficulty of internationalizing protection.
Let us look first at a brief history of the technology, just as Congress had to do before it passed the Semiconductor Chip Protection Act (SCPA) of 1984.1 The semiconductor chip was invented in 1959;2 and the first microprocessor chip was developed in 1971.3 By the early 1980s, developers could fabricate chips containing more than 100,000 transistors.4 From the 1970s through to the 1990s, the chip has become so ubiquitous that it is found in products ranging from automobiles to refrigerators to personal computers and a vast variety of "personal electronics."
The story of course does not end there. Since the early 1980s, the pace of innovation in semiconductor technology has accelerated. Chips currently in production contain in excess of 1,000,000 transistors. Dynamic random access memory chips (DRAMS), which have set the pace of progress in the industry, have provided a fourfold increase in capacity every three years—even though each increase has required engineers and scientists to solve ever more complex problems, driving the technology to even greater heights.
The photolithographic process used to fabricate the vast majority of semiconductor chips is conceptually relatively simple. The manufacturer applies a layer of photoresist (a material that reacts to light and resists the action of certain chemical agents) to a wafer of material called a substrate. The photoresist is exposed with a predetermined pattern. After being "developed," portions of the photoresist are washed away, leaving the substrate exposed. The substrate is then treated with a chemical agent that may etch material away from the exposed part, deposit material on it, or permeate into it. The manufacturer removes the photoresist and then repeats the process for each of the multiple layers required to form the device.5
The photolithographic process just described has many applications beyond semiconductor chips. The substrate does not have to be silicon (or any semiconductor for that matter), and the product does not have to be electronic circuitry. Manufacturers can use photolithography with masks on a variety of substrate materials, such as glass, polycrystalline silicon, sapphire, ceramic material, superconducting material, magnetic domain material—the list goes on and on, and continues to grow.
Moreover, the resulting product does not have to be a "chip." It can be a flat-panel display, a miniature motor and gears, a thin-film recording head, or any one of a number of items that are not usually considered to be electronic circuitry. It is possible that within a few years, virtually every portion of computer hardware, from the display to the mass storage devices to the packaging for chips, will be fabricated by using some kind of masking process.
Each stage of the process, from preliminary design through fabrication, requires investment, skill, creativity, and just plain hard work. As the technology became increasingly important in the U.S. economy, additional legal protection at some stage of the process appeared to be necessary to protect this investment if innovation was to flourish.6 Beginning in the late
5 |
Alternatively. circuitry can be written directly on the surface of the substrate by using electron beams. This technique is not yet widely used commercially. |
1970s, innovative semiconductor manufacturers pressed Congress to grant additional intellectual property protection for this particular form of investment in monetary and intellectual capital. The result was the Semiconductor Chip Protection Act.
From the outset, the legislative effort to fashion protection for chips focused on the "mask works." These are the series of masks that bear the circuit designs used to expose the photoresist in the fabrication process. Each mask bears the information that dictates which areas are to be exposed and which are to be covered during a given step in the process. Together, the masks describe the entire three-dimensional topography of the finished product.
However, this focus was not inevitable. If the form of protection chosen for semiconductor chip products had been derived from patent law, more emphasis might have been placed, for example, on the fabrication process or on the product than on the intermediate masks.
Initial proposals for a chip protection law called for an extension of copyright law, declaring mask works to be pictorial, graphic, and sculptural (PGS) works, notwithstanding their utilitarian purpose.7 Ordinarily, the design of a "useful article" is protectable as a PGS work only to the extent that it incorporates aspects that are physically or conceptually separable from their utilitarian aspects.8 The proposals would have created an exception to this limitation for mask works.
These efforts failed, partly because of the limitations the drafters tried to place on copyright protection for mask works, and partly because it was unclear how provisions of the draft legislation did and did not relate to the Copyright Act as a whole. Mask works were narrowly defined by reference to the definition of "semiconductor chip product." They were to have an abbreviated term of protection—10 years—and were to be subject to special exceptions for reverse engineering. Later drafts of the legislation created a new category of works with a separate bundle of rights distinct from those generally accorded to works under section 106 of the Copyright Act.9 Because the legislation failed to provide appropriate integration of chip protection into the copyright statute,10 it was not adopted.
Instead, Congress chose to create a sui generis law outside of the copyright statute. The SCPA bears the basic features I have described: narrow subject matter, 10-year term of protection, and a broad exception for reverse engineering. Congress attempted to strike a balance, as the copyright and patent laws do, between innovators and imitators. It also attempted to allow a certain degree of flexibility within the narrow confines of a custom-tailored law drafted to meet the specific concerns of manufacturers of a specific product at a specific point in its technological development.
DEFINING THE TECHNOLOGY
It would seem axiomatic that in fashioning intellectual property protection for a new technology the first order of business is to determine what the technology is. The foundation of the SCPA was codified in one of its basic premises: that the technology was the fabrication of semiconductor chip products.11 This was not a surprising choice, given the state of the technology in the late 1970s and early 1980s when the bill was drafted, and the fact that the law was passed largely at the behest of the semiconductor industry. However, time and change have quickly eroded both foundation and premises.
For example, a more relevant technology might be the fabrication of products on any substrate using masking techniques. As noted above, making a chip does not differ very much conceptually from the making of many
other products. Considerable skill and creativity are invested in the design of the mask works that determine the topography of those products, but this design work is easily appropriated since, in essence, each copy of the product carries its own blueprint with it.
In addition to semiconductor chip products, many other products are being made, or will be made, by using masking techniques. They include not merely devices such as microprocessors and DRAMs that are traditionally thought of as chips. They also include, for example, thin-film heads, flat panel displays, micromechanical devices, chip packaging, magnetic bubble devices, magnetic mass storage devices, optical devices, and superconducting devices. Yet because they are not formed on a semiconductor substrate, or are not intended to perform electronic circuitry functions, it may be difficult to consider those items as "semiconductor chip products" under the act. Much of this important technology would thus not be given any of the special protection that the SCPA attempts to provide.
In the future, even devices such as transistors—the heart of what is commonly known as the computer chip—may be fabricated in a way that may deny them protection under the SCPA. For example, researchers have recently made a promising speed breakthrough in lateral bipolar transistors.12 Transistors of that kind, with their low power consumption, could power high-performance, low-power computers in the future. The devices are made of silicon deposited or "grown" in very thin layers on an insulator; they are not etched from or deposited on a semiconductor substrate.13 Consequently, chips made in this way might not constitute semiconductor chip products within the meaning of section 901(a)(1) of the SCPA; and even though they would perform the same functions as any other computer chip, they might not be protected under the act.
To a greater or lesser extent, each of these devices shares (from an intellectual property perspective) the salient properties of semiconductor chips: considerable skill and creativity are invested in the design of the masks, and the design can be appropriated easily because it is borne on the product's surface. There is no principled reason why they should not receive the same protection, but in its sui generis approach the SCPA was too custom-tailored so as to fit a specific technology.
PROTECTION-SCOPE AND LIMITATIONS
Protection for mask works is subject to several significant limitations, the most important of which is the reverse engineering exception. Section 906(a) of the SCPA permits
a person to reproduce the mask work solely for the purpose of teaching, analyzing, or evaluating the concepts or techniques embodied in the mask work or the circuitry, logic flow, or organization of components used in the mask work; or . . .[to perform such analysis and] to incorporate the results of such conduct in an original mask work which is made to be distributed.
As with the technology of chip design and manufacture, the technology (and cost) of reverse engineering has not stood still since the enactment of the SCPA. Companies specializing in chip analysis can now "peel" or "strip" away a chip's various layers and provide cross sections, topological layouts, and material analyses for $10,000 to $30,000.14 These same companies can also provide the "paper trail" that is a key element of a reverse engineering defense under the act. Computer programs can derive the logic diagram of a chip, and the logic diagram can then be used as the input for any one of a number of computer-aided chip design tools. The design can be modeled by using yet another software tool. A second comer can produce a "new'' chip in this way for a very small fraction of the original producer's typical development costs for an innovative product. The second comer can use computerized optimization and can introduce some degree of variation from the original design.15
So, while section 906 of SCPA apparently sanctions it, this practice allows free riders to profit easily and cheaply from the success of others who make the R&D design investment and take the risks in bringing a new and innovative product to market.16
It can be argued, of course, that this kind of activity is beneficial. After all, it does encourage making at least incremental improvements in the original design and making products available at a lower price. That is the
essence of imitative competition. We must bear in mind, though, that the goal of our intellectual property laws is to give the public the benefit of innovative competition. By granting Congress the power to enact intellectual property protection to "promote the progress of science and useful arts,"17 the framers of our Constitution intended to spur innovation, and innovation is what has made the United States a leader in high technology. Without it, there would be no high-tech products to imitate, and without meaningful intellectual property protection, innovation would lag.
The reverse engineering provisions of the SCPA are commonly mentioned in the industry as a primary reason that chip developers have brought few lawsuits under the act against copiers of their chip designs. Section 906(a) is widely viewed as the exception that swallows the rule of protection for mask works. Although difficult to prove, the perception strongly suggests that the SCPA may not provide meaningful protection.
INTERNATIONALIZING PROTECTION
Another major weakness of the SCPA is the absence of any workable effective international protection. Had protection for mask works or chip topography been integrated into an existing form of intellectual property protection—as noted above, copyright was the original candidate—it could have been internationalized by means of existing treaties. The sui generis treatment of chip protection under U.S. law, by contrast, has been extremely difficult to project into the international arena.
A sui generis law requires a sui generis treaty—a treaty that must be negotiated without any international consensus on what sort of regime for protection is appropriate. The primary multilateral effort to date was the Washington Treaty,18 prepared under the auspices of the World Intellectual Property Organization (WIPO). The process was strongly influenced by developing countries hostile to intellectual property protection generally, and it produced a treaty so flawed that not one single major chip-producing country could support it.19 Chief among the problems that the United States
found with the Washington Treaty were the inadequate term of protection, the lack of specific protection for mask works incorporated in a finished product, broad provisions for compulsory licenses, and excessively permissive treatment of so-called innocent infringers.
There has been an effort to cover protection of chip topographies in the language of the Trade Related Intellectual Property Rights (TRIPS) agreement proposed in the General Agreement on Tariffs and Trade (GATT).20 Section 6, article 35 of the TRIPS draft distributed by GATT Director General Arthur Dunkel on December 20, 1991, provides that TRIPS parties will provide protection in accordance with what are in effect the substantive provisions of the Washington Treaty, as supplemented by further provisions in section 6. The additional provisions endeavor to remedy the most significant shortcomings of the treaty.21 It is far from certain, however, that the TRIPS effort will come to fruition.22
The only means remaining under the SCPA for internationalizing chip protection is through reciprocity with countries that protect (or are taking steps to protect) U.S. mask works on substantially the same basis as the SCPA.
The SCPA provides two means for giving foreign nationals reciprocal treatment. The first is a presidential proclamation under section 902(a)(2). This provision permits the president to extend protection under the act to foreign nationals for mask works that are first commercially exploited abroad. This extension can be made only after a finding that in the foreign country, U.S. mask works receive national treatment or treatment equivalent to that given under the SCPA. To date, there have been no presidential proclamations under section 902(a)(2).
The second reciprocity provision, section 914, was an eleventh-hour amendment to the draft law.23 Section 914, International Transitional Pro-
visions, permits the Secretary of Commerce to issue temporary orders that extend the benefits of the law to nationals of countries that are "making good faith efforts and reasonable progress toward" entering a bilateral or multilateral treaty with the United States or enacting domestic legislation along the lines of the SCPA. The "transitional provision" was originally scheduled to expire in 1987, but has been extended twice and is currently scheduled to expire on July 1, 1995.24
Although originally meant only as a stopgap measure to be used until other countries met the requirements of section 902, section 914 has been the sole vehicle for internationalizing protection of semiconductor chips. Nineteen foreign countries (including Japan, Germany, Canada, and the United Kingdom) have been granted interim protection under section 914 orders issued by the Secretary of Commerce.
By employing this "arm-twisting" method of inducing foreign countries to adopt our own approach, the United States has achieved some positive results in internationalizing the protection of chip topography, but the means is hardly ideal for international comity. No sovereign nation appreciates another nation dictating its laws to it, and resentment of this approach may be one of the factors hindering cooperation in multilateral efforts to harmonize the protection of chip topography worldwide.
Moreover, it is unlikely that the United States can rely forever on temporary section 914 orders to induce countries to make permanent changes to their laws. These countries must recognize that the current status of indefinite temporary reciprocity cannot be permanent, that is, unless the legal fiction of perpetually "making good faith efforts and reasonable progress" is stretched to the breaking point.
CONCLUSIONS
The SCPA is a law that was custom designed to address narrow concerns. It freezes in the law definitions that relate to only a portion of a specific technology—as that technology existed when the legislative effort began, several years before the law's passage. It also freezes in the law a balancing between innovative and imitative competition that was based on a competitive environment vastly different from what exists today (e.g., as discussed above, the substantial decline in the cost of reverse engineering relative to the cost of developing an innovative product).
Apart from the narrowness of its subject matter, the protection afforded is thin indeed. Establishing a paper trail can be as simple as ordering from a catalog and can be a successful defense to infringement under the reverse
engineering provisions if a defendant introduces some—it is impossible to say how much, since we have no judicial guidance25—variation on the original design.
It is possible that, in time, Congress will enact a legislative "fix" that will remedy at least some of the law's shortcomings. However, as the technology continues to develop, will Congress, or the law itself, keep pace? Unlike the Copyright and Patent Acts, the sui generis SCPA has neither a flexible and expansive subject matter nor a historic body of principles and precedents for adapting to changes without repeated congressional action.26
Moreover, the solutions in the United States, be they judicial, legislative or administrative, may differ greatly from the solutions abroad, since for a sui generis scheme of protection there is no common body of copyright or patent principles or multilateral treaty to guide the courts, legislatures, and administrative agencies in other countries.
ACKNOWLEDGMENTS
I acknowledge with gratitude the substantial contributions to this chapter by Jesse M. Feder, Esq., my colleague at Schwab Goldberg Price & Dannay. In the technical discussion, I am heavily indebted to the expertise of Drs. Praveen Chaudhari, Tak Ning, and Webster Howard, of the IBM Corporation, for an appreciation of the present state of the semiconductor industry and its possible future directions. In all cases, however, I am solely responsible for the interpretation that appears in this chapter and for any error or omission it might contain.
25 |
Neither of the decisions rendered in Brooktree Corp. v. Advanced Micro Devices, Inc., 705 F. Supp. 491 (S.D. Cal. 1988) (order denying preliminary injunction) and 757 F. Supp. 1088 (1990) (order denying defendant's motion for judgment notwithstanding the verdict) illuminate this issue. |
26 |
Although it is beyond the limited scope of this case study, I think it is instructive to contrast the experience of chip protection under the SCPA with the experience of computer program protection under copyright law. Congress amended the Copyright Act in 1980 essentially only to confirm the explicit congressional recognition in 1976 of computer programs as copyrightable literary works. Pub. L. No. 96-517, 94 Stat. 3015, 3028. In 1990, Congress granted a rental right to computer program copyright owners, as it had already done to owners of copyrights in sound recordings. Pub. L. No. 101-650, 104 Stat. 5089, 5134, 5135. No further recourse to Congress has been necessary. Since the passage of the 1976 act, the technology of software engineering and the shape of the software industry have evolved considerably. Many new and difficult issues have arisen. Notwithstanding concerns expressed by some commentators (see Chapter 11) the courts have met these challenges successfully by applying traditional copyright principles. Moreover, there is a growing international consensus that computer programs are literary works protected under the Berne Convention. |