Opinion
Patent Appeal No. 8603.
February 24, 1972.
Jack Oisher, Briarcliff Manor, N. Y., attorney of record, for appellant.
David I. Caplan, Arthur J. Torsiglieri, Murray Hill, N. J., attorneys of record, for appellees.
Appeal from the Board of Patent Interferences.
Before RICH, ALMOND, BALDWIN, and LANE, Judges, and RAO, Judge, United States Customs Court, sitting by designation.
This is an appeal from the decision of the Patent Office Board of Patent Interferences, adhered to on reconsideration, awarding priority to appellees, Joseph J. Kleimack, Howard H. Loar, and Henry C. Theuerer (hereinafter Kleimack).
Appellant, Tummers, is involved on application serial No. 349,709 filed March 5, 1964 as a division of application serial No. 84,923, filed January 25, 1961, and accorded the benefit of a January 29, 1960 foreign priority date. The Tummers' application is assigned to North American Philips Co., Inc. The sole count in the interference was copied from the Kleimack patent, No. 3,165,811, issued January 19, 1965 on an application filed June 10, 1960. The Kleimack patent is assigned to Bell Telephone Laboratories, Inc.
Now patent No. 3,217,214 issued November 9, 1965.
Based on Dutch application No. 247,902.
The invention in issue relates to a method of making a diffused junction transistor. It is best understood from the following analysis of the count, which was provided as an appendix to appellees' brief:
Drawing of the Device Formed by the Process Described in the Count of this Interference After Each Step Thereof (Assuming an NPN Transistor)
1. The method of making a junction transistor comprising the steps of forming by vapor deposition on a surface portion of a semiconductive body an epitaxial layer of higher specific resistivily than the specific resistivily of said surface portion, diffusing into a surface portion of the epitaxial layer a conductivity-type impurity for forming in said layer a first region of a conductivity type opposite that of said surface portion of said semiconductive body, while leaving a high specific resistivily portion of said epitaxial layer between said first region and the original surface portion of said body, and introducing into a limited surface portion of the first region a conductivity-type impurity of the type opposite that diffused into the first region for forming within said first region a second region of the opposite conductivity type, the second region serving as the emitter region, the first region serving as the base region, and the semiconductive body including the collector region of the junction transistor. [4] The primary issue in this case centers around the first step of the count which recites the vapor deposition of an epitaxial layer of high resistance on a substrate. Both parties agree that "epitaxial" means at least that it is (a) a single crystal with (b) a crystal orientation determined by that of the substrate. The evidence also supports the board's finding that "epitaxial" implies that the substrate for the layer is also a single crystal since single crystal layers do not ordinarily grow on polycrystalline substrates.After the interference was declared, Kleimack moved to dissolve the interference on the grounds that Tummers could not make the count, (1) because the first step of the count is not disclosed by Tummers and (2) because the Tummers' disclosure is not enabling. In the alternative, Kleimack concurrently moved for leave to take testimony on these issues should the motion to dissolve be denied. The primary examiner denied the motion to dissolve, but the board granted the motion for leave to take testimony. Both parties took testimony and submitted exhibits. After briefs were filed and oral arguments heard, the board awarded priority to Kleimack on the ground that Tummers was not entitled to make the count. This decision was adhered to on reconsideration.
The only issue before us is whether Tummers' application supports the first step of the count. More particularly, the issue is whether the disclosure of a vapor-deposited epitaxial layer is inherent in the Tummers' disclosure which does not explicitly mention epitaxial layers, but discloses vapor depositing a germanium layer on a silicon layer in a "known manner."
There is a considerable amount of evidence in the record pertaining to the state of the art as of January 1960 (the filing date of Tummers' Dutch application), and, contrary to appellant's arguments, we think the sufficiency of the Tummers' disclosure must be determined as of that date. We will not attempt to set forth all that the evidence shows in that regard. However, in general we think it fair to say that vapor deposition of epitaxial layers in homoepitaxial growth, for example silicon on silicon, was well known in 1960. In fact, the testimony indicates that all commercial bipolar transistors were homoepitaxy and that such transistors were widely used in 1960. Heteroepitaxial growth, for example germanium on silicon, on the other hand, was difficult, if not impossible, to achieve in 1960 and it certainly was not commercially feasible. There is a four percent mismatch in the lattice constants (size of a unit crystal cell) between germanium and silicon and, at least partially for that reason, all reported efforts to grow epitaxial germanium layers on silicon resulted in the formation of microscopic islands of epitaxial growth or unoriented polycrystalline aggregates.
See generally, Janicke, Patent Disclosure:
Some Problems and Current Developments, Part III, 53 J.P.O.S. 3 (1971). We have previously found it unnecessary to decide this issue. See, e. g., In re Barrett, 440 F.2d 1391, 58 CCPA 1155 (1971).
The board, after making findings of fact much more thorough than the brief summary above, found that Tummers had not sustained his burden of showing that his disclosure inherently supports the count and, therefore, held that he could not make the count. The board found that the processes of depositing a germanium layer on a silicon substrate available to those skilled in the art in 1960 would not inevitably produce an epitaxial layer as required by the count, but could result in polycrystalline growth or result in the formation of unusable germanium islands.
Appellant takes exception to every aspect of the board's decision, but in the main contends that the board erred in placing the burden on appellant and in failing to recognize that there were two principal issues in the case. That is, appellant asserts that the question is not whether anyone following known techniques without any deposition to grow epitaxial layers would automatically get epitaxy, but (a) whether one skilled in the art reading Tummers' specification would understand it to mean epitaxy and (b) whether he would then know how to accomplish that objective. Tummers takes the position that if one wanted to grow an epitaxial layer, then he could do so with Tummers' teachings augmented by the existing art knowledge, and anyone skilled in the art following Tummers' teachings would want to grow a single crystal epitaxial layer because nobody has ever made commercial transistors of polycrystalline material.
As to whether appellant has the burden of showing that he can make the count, we think clearly he does. Appellant not only copied the claim from appellees' patent but also relies on inherency of the disclosure, as supplemented by the prior art, for support of the count. Under such circumstances appellant has a twofold burden. As this court said in Dreyfus v. Sternau, 357 F.2d 411, 53 CCPA 1050 (1966):
First, one copying a claim from a patent for the purpose of instituting interference proceedings must show that his application clearly supports the count. * * * There must be no doubt that an application discloses each and every material limitation of the claims and all doubts must be resolved against the copier. * * * Second, where support must be based on an inherent disclosure, it is not sufficient that a person following the disclosure might obtain the result set forth in the counts; it must inevitably happen.
This doctrine, which the board followed in the instant case, has been reaffirmed by this court in several recent cases. See, Reed v. Tornqvist, 436 F.2d 501, 58 CCPA 864 (1971); Noyce v. Kilby, 416 F.2d 1391, 57 CCPA 1156 (1969), and Gubelmann v. Gang, 408 F.2d 758, 56 CCPA 1013 (1969). We find absolutely no merit to appellant's contention that the examiner's decision denying appellees' motion to dissolve switched the burden to appellees. The examiner's decision merely left the issue for the board to resolve after testimony was taken.
We also agree with the board that appellant has failed to meet his burden of proof in this case. Although we see nothing wrong with the board's approach, since appellant considers two main issues to be presented, we will discuss them as appellant has presented them.
First, appellant asserts that one skilled in the art reading the Tummers' disclosure would know epitaxial layers were inherently intended because all known bipolar transistors were built with single crystal layers. While we have no doubt that one of ordinary skill in the art would have thought of epitaxial layers if appellant had disclosed homostructures, such as silicon on silicon or germanium on germanium, we cannot say the same about the germanium on silicon heterostructures disclosed. The evidence clearly supports the view, in conformity with the board's findings, that in such situations one of ordinary skill in the art might very well think Tummers intended to disclose polycrystalline or amorphous layers. As mentioned above, if there is any doubt whatsoever about whether the limitations of the count are inherent in Tummers' disclosure, it must be resolved against Tummers. Reed v. Tornqvist, supra; Noyce v. Kilby, supra; Gubelmann v. Gang, supra; Dreyfus v. Sternau, supra.
Secondly, we think it even more persuasive of the holding that Tummers cannot make the count that one of ordinary skill in the art would not have known how to deposit epitaxial germanium layers on silicon even if he knew that epitaxial layers were intended by Tummers. The Tummers' disclosure states that the germanium is vapor deposited on the silicon "in known manner, not essential to the invention." The evidence of record clearly indicates that there was no known manner of depositing layers by heteroepitaxial growth. While the various concepts which could be combined to achieve such results were possibly available, they were scattered throughout the art. We are not convinced that how to select and combine the various teachings of the art (some of which dealt with heterostructures, some homostructures) to achieve successful heteroepitaxial growth of germanium on silicon would have been apparent to one of ordinary skill in the art in 1960, let alone convinced that such growth could be done in a "known manner."
It is noted that in 1964 reports were published by appellee's assignee stating that success has been achieved in epitaxially growing germanium on silicon.
What does seem to have been known by those of skill in the art in 1960 was that such heteroepitaxial growth could result in the formation of microscopic islands. The record shows, however, that such islands were too small to accommodate a spaced pair of the minimum size electrodes then in use in order to form a transistor. These islands cannot be considered "an epitaxial layer" as called for in the count.
Since only island type or polycrystalline aggregate type growth could be achieved in heterostructures in 1960, we fail to see how one of ordinary skill in the art reading Tummers' specification would know either that epitaxial layers were intended or how they were to be made. Therefore, the Tummers' specification does not inherently disclose epitaxial layers, Tummers cannot make the count which recites such layers, and priority of invention was properly awarded to appellees.
The decision of the board is affirmed.
Affirmed.