OPINION AND ORDER
Plaintiffs Janssen Pharmeceutica, N.V. and Janssen Pharmaceutica Products, L.P. (collectively “Janssen”) bring this action for patent infringement against Eon Labs Manufacturing Corporation, Inc. (“Eon”) pursuant to 35 U.S.C. § 271(e)(2). Jans-sen is the holder of a composition patent entitled “Beads Having a Core Coated with an Antifungal and a Polymer,” U.S. Patent No. 5,633,015 (“the ’015 patent”). Janssen markets this composition in a capsule under the name SPORANOX®. Eon, a generic drug manufacturer, filed an Abbreviated New Drug Application (“ANDA”) pursuant to Section 505(j) of the federal Food, Drug and Cosmetic Act, 21 U.S.C. § 355(j), seeking Federal Food and Drug Administration (“FDA”) approval to make and sell a generic version of the SPORANOX® capsule. The ANDA procedure for drug approval, created by the Drug Price Competition and Patent Term Restoration Act of 1984, Pub.L.No.98-417, 98 Stat. 1585 (1984) (codified in sections of titles 21, 35 and 42 U.S.C.) (“the Hatch-Waxman Act”), allows a generic manufacturer to avoid the costly and time-consuming safety and efficacy studies that were required of the brand name drug, so long as the generic drug is the bioequivalent of the FDA-approved drug. In its ANDA, Eon certified, pursuant to 21 U.S.C. § 355(j)(2)(A)(iv), that the ’015 patent was “invalid or will not be infringed by the manufacture, use or sale of the generic drug for which the ANDA is being submitted.” After Eon notified Janssen of its ANDA filing, Janssen filed this suit for patent infringement on April 13, 2001, and a 30-month automatic stay of FDA ap*266proval for Eon’s ANDA was imposed pursuant to 21 U.S.C. § 355(j)(5)(B)(iii).
On March 7, 2004, the extended stay of FDA approval expired.1 On March 16, 2004, I granted a temporary restraining order (“TRO”) to prevent Eon from marketing its generic product, should it receive FDA approval to do so, pending determination of Janssen’s motion for a preliminary injunction.
In an order dated April 6, 2004, I construed the only independent claim in the ’015 patent, Claim 1, and denied Eon’s motion for partial summary judgment. On April 8, 2004, Magistrate Judge Marilyn D. Go issued a report recommending that Janssen’s motion for summary judgment rejecting the affirmative defenses of invalidity and unenforceability be granted and that Eon’s motion for summary judgment based on the defense of invalidity be denied. Since the filing of objections and responses to Judge Go’s report was not completed until shortly before the trial, decision was withheld and the parties permitted to further develop facts at trial. The ’015 patent
The patent in suit concerns “a novel composition of antifungal agents which have low solubility in aqueous media, a process for preparing said composition and pharmaceutical dosage forms for oral administration comprising the novel composition.” ’015 Patent, col. 1, lines 10-15. This composition and the pharmaceutical dosage forms allow poorly soluble itracona-zole molecules to be delivered to a patient in an oral dosage form to treat fungal infections.
Claim 1 of the patent, the only independent claim, reads:
1. A bead comprising:
a.a central, rounded or spherical core;
b. a coating film of a hydrophilic polymer and an antifungal agent selected from the group consisting of itraconazole and saperconazole, and
c. a seal-coating polymer layer, characterized in that the core has a diameter of from about 600 to about 700 <fm (25-30 mesh).
’015 Patent, col. 6, lines 16-24. The specification states:
The particular size of the cores is of considerable importance. On the one hand, if the co[res] are too large, there is less surface area available for applying the drug coating layer, which results in thicker coating layers. This raises problems in the manufacturing process as an intensive drying step is needed to reduce residual solvent levels in the coating layer. The intensive drying conditions may adversely affect drug dissolution from the beads and should therefore be controlled extremely well during the manufacturing process. On the other hand, small cores have a larger total surface available for coating resulting in thinner coating layers. Consequently a far less intensive drying step can be used to decrease residual solvent levels. Co[res] which are too small, e.g. 30-35 mesh cores, however, have the disadvantage of showing considerable tendency to agglomerate during the coating process. Therefore, 25-30 mesh co[res] represent the optimum size where neither agglomeration nor an intensive drying step unduly constrain the manufacturing process.
’015 Patent, col. 1-2, lines 58-8.
In the April 6, 2004 Order I construed “a bead” to mean one or more beads, and found “characterized in that the core has a diameter of from about 600 to about 700 *267<tm (25-30 mesh)” to mean that a core with a diameter of 600 to 700 microns was claimed. I rejected Eon’s assertion that the parenthetical “(25-30 mesh)” superced-ed the micron diameter as defining the size core claimed and concluded that “(25-30 mesh)” simply explained to one of ordinary skill in the art where cores of the size claimed are likely to be found. I also concluded that the term “about” did not expand the range of core diameter sizes claimed beyond 600 to 700 microns.
THE TRIAL
A combined hearing on the preliminary injunction motion and bench trial was held from May 17, 2004 through May 20, 2004, in accordance with Fed. R. Civ. Pro. 65(a)(2). Set forth below are the findings of fact and conclusions of law required by Fed. R. Civ. Pro. 52(a). Janssen presented testimony from two of the inventors of the ’015 patent, Roger P. Vandecruys and Paul Gilis; experts Dr. Saul J.B. Tendler and Dr. Roland Bodmeier; and Dr. Bruce L. Moskowitz, former Director of Clinical Research for Infectious Diseases at Jans-sen and currently its Executive Director of Medical Affairs. Eon presented testimony of its expert, Dr. Harry Brittain; Dr. Siya-wosh Moghaddam, Eon Director of Analytical Research and Development; Sadie M. Ciganeck, Eon Vice-president of Regulatory Affairs; and Dr. Bernard Hampl, Eon CEO and President. The credible evidence at trial established the following:
The Development of the Patented Composition and the Generic Product
Janssen patented the drug itraconazole, a broad spectrum antifungal, in 1981, U.S. Pat. No. 4,267,179 (“the 179 patent”). Itraconazole is used to treat serious systemic fungal infections that can cause serious lung, brain or other organ damage and, if left untreated, even death. In 1984, Janssen scientists in Belgium began experimenting with formulations to make an oral dosage form of the drug. This posed difficulties because itraconazole is nearly insoluble. The first attempt, which employed a tablet as a delivery method for the drug, was unsuccessful. Janssen scientists then experimented with applying the' drug, mixed with a hydrophilic polymer and a solvent, over many microscopic sugar spheres, and then filling a capsule with the coated spheres. This research led to the development in 1987 of the F5 formulation: a hard gelatin capsule filled with beads comprised of 18-20 mesh sugar spheres as cores; a drug coating layer of itraconazole and hydroxypropyl methylcel-lulose (“HPMC”) mixed with a methylene chloride/ethanol solvent; and a seal coating layer of polyethylene glycol (“PEG”), also mixed with a solvent. Each capsule contained approximately lOOmg of itraco-nazole.
The Mareh-April 1988 issue of the Phar-macopeial Forum, the journal containing revision proposals for the United States Pharmacopeia-National Formulary standards, proposed limits on residual organic volatile impurities, such as the methylene chloride solvent used in the F5 formulation. In September of 1989, in anticipation of these limits being adopted, Janssen conducted experimental studies on coated beads using different drying conditions and different sugar core sizes to determine if those differences would affect residual solvent levels. Janssen found that using a smaller core size, of 25-30 mesh, reduced the. residual solvent levels, as did increased drying. November 1989 experiments showed that changing the solvent/ethanol ratio also had a significant effect on initial solvent levels, before drying. Nevertheless, Janssen did not change the F5 formulation that used 18-20 mesh cores for its clinical tests.
Janssen began clinical trials of the F5 formulation in 1987 and added a compassionate use protocol to its Investigational *268Drug Number for these clinical studies. Under the compassionate use protocol, a doctor requested permission from Janssen to use SPORANOX®, which had not yet received FDA approval, on a patient who had failed therapy with approved medications or was intolerant to them. The doctor explained the patient’s condition to Janssen and requested permission to use SPORANOX® under a strict protocol, at no cost to the doctor or the patient. The protocol informed the doctor of how he or she was to make the diagnosis, how to administer the drug, at what dosage, and what information to track on an ongoing basis to determine whether the drug was safe to use in the patient. That information was returned to Janssen. The doctor was provided a form on which to keep records and to provide ongoing information to Janssen to confirm that the patient should be sent additional drugs. Providing this information was a condition for receiving the drug. All drugs shipped under the protocol were for use only on the designated patient and all unused drugs were required to be returned. The doctor was not informed of the formula for SPO-RANOX®, other than the fact it contained itraconazole. The label on the boxes of capsules shipped under the tests indicated that the drug was for compassionate clearance use restricted to a certain patient, usually identified by initials. The labels also stated that the drug was limited by federal law to investigational use. Typically, the clinical tests took place in a hospital setting. The patient was required to sign an informed consent form, which advised the patient that the patient was using an investigational drug and that the patient’s name would be kept confidential. The patient was not under any express obligation of confidentiality to the doctor or to Jans-sen.
On April 16, 1990, Janssen applied for trademark registration for the name SPO-RANOX® for the itraconazole capsules. The trademark was published on December 4, 1990. On May 30, 1990 Janssen filed a New Drug Application (“NDA”) with the FDA seeking approval to market the F5 formulation. In November 1990, the FDA advised Janssen of its concern over the limit that Janssen had set for residual methylene chloride levels for the capsules. In January through March 1991 Janssen confirmed, by large scale tests, the effect of bead size reduction and solvent ratio change on residual solvent levels in the capsules. Janssen experimented with 25-30 mesh and 30-35 mesh sugar cores in the formulation and found that the “optimal” size was 25-30 mesh. The biggest problem with the smaller 30-35 mesh cores was the tendency towards agglomeration. The result of these experiments was the F12 formulation, that contained 25-30 mesh cores and used a 60/40 solvent to ethanol ratio. In March and April of 1991 Janssen discussed revising its proposed limits for residual methylene chloride with the FDA, and, on April 26, 1991, Janssen withdrew its NDA'on the F5 formulation.
Studies conducted from June through August 1991 showed that the F12 formulation is bioequivalent to the F5 formulation, so that whatever test results had been generated on safety and efficacy with the F5 formulation were transferrable to the information on the F12. On October 10, 1991, Janssen filed an NDA for SPORA-NOX® capsules containing 25-30 mesh cores, the F12 formulation. From November 1991 through March 1992 Janssen worked with the FDA on production and test methods for the F12 formulation, and in March the formal process validation studies commenced.
The European priority patent for the F12 formulation was filed on September 3, 1992. In drafting the European patent application for what was to become *269the ’015 patent, the words that follow “characterized in that” were considered the novelty, the inventive step. The international regulations at the time the European patent application was filed required that claims of an invention be defined in metric measurements, such as microns, and 25-30 mesh is not an accepted metric measurement.
On September 11, 1992, Janssen received FDA approval to market the F12 formulation of SPORANOX® in the United States. The Patent Cooperation Treaty (“PCT”) application was filed on August 27, 1993. Janssen failed to file the U.S. patent application within the required three-year time period of the PCT filing, but was granted permission to extend the time to file. The ’015 patent was issued on May 27,1997.2
Eon Labs began development of a generic version of SPORANOX® in 1996, when it noted that the ’179 patent on itraconazole was expiring in 2000. Attempts to create a bioequivalent capsule using 25-30 mesh sugar cores were unsuccessful. After learning of the ’015 patent in 1997, and Janssen’s use of 25-30 mesh cores, Eon began experimenting with SOBS mesh cores and 20-25 mesh cores in its formulation. Bioequivalency tests on the formulation with the 20-25 mesh cores were eventually successful.
In both SPORANOX® and Eon’s generic product, microparticle sugar spheres, the “cores” characterized in Claim 1 of the patent, are the carriers of the itraconazole. The process that both Eon and Janssen use to coat the particles is generally the same. The cores are placed in a cylindrically shaped fluidized bed granulator, which keeps the particles in circulation in the cylinder with air pressure. A spray insert atomizes the drug coating solution into the cylinder where it coats the cores. Much of the solvent added to the drug coating solution is dried out of the solution before it hits the particle; thus the coating is not a “wet” solution, but a “solid” one. More of the residual solvents are removed while the cores are circulating in the fluidized bed granulator. The particles are then placed in vacuum dryers to cause further evaporation of the residual methylene chloride solvent. After this drying step, a sealing layer of polymer is added, and the particles are dried in the fluidized bed granulator.
There are many possible process variables, including drying temperature, air volume, the composition of the spray material, the equipment used, and the thickness of the coating on the core, that affect the amount of vacuum drying needed to reduce residual solvent levels before the sealing layer is applied. Eon uses less methylene chloride in its drug coating solution than the amounts that Janssen suggests are acceptable in its example, but the residual solvent levels in Eon’s ANDA, *270although within the FDA limit set for SPORANOX®, are higher than the actual levels in the SPORANOX® capsules.
This is so because the 20-25 mesh cores used by Eon require a thicker drug coating layer than the 25-30 mesh cores claimed in the ’015 patent. As described in the ’015 specification, when larger cores are used, the drug layer must be thicker, since it is spread over fewer cores, and the residual solvent levels are higher. The active drying time specified in Eon’s ANDA is within the time given in the ’015 patent example, but is based on drying smaller batches, around 6.7 kg per sub-batch dried, than Janssen’s specification example, 41.74 kg. See ’015 Patent, col. 5, line 18. Eon does not use methylene chloride in its seal coating layer, which Jans-sen does.
The Diameter of a Core
According to Webster’s Third New International Dictionary (1993), a “diameter” is the length of a straight line through the center of an object, or its thickness. If an object is perfectly round, it has only one diameter measurement. That is not the case with sugar spheres. The sugar spheres that are used as the cores in the invention are not true spheres, but tiny multi-sided particles. Since sugar cores have various shapes, there are multiple ways to define particle size and to measure the diameter of a single core.
Because cores used in pharmaceutical compositions are very small, the industry has created a system of using metallic filters or wire screens, referred to as “mesh” sieves, to separate and classify different size sugar spheres or other micro-particles. The “mesh number” of a particular sieve corresponds to the number of square openings per linear inch, which, in turn, refers to the length of the side of the individual square openings in the sieve, as measured in microns3 or millimeters.
The actual dimensions of the openings of a given size sieve vary within certain tolerances. United States Pharmacopeia (“USP”) in conjunction with the National Formulary (“NF”) requires that sieves used for particle classification conform to the standards of the American Society for Testing and Materials (“ASTM”). The ASTM sets the standard or nominal opening of the apertures in a 20 mesh sieve at 850 microns, a 25 mesh sieve at 710 microns, and a 30 mesh sieve at 600 microns. For example, a 20 mesh sieve contains 20 square openings per inch, each opening being approximately 850 microns in length as measured along the side of the opening. A 25 mesh sieve contains 25 openings per inch, each opening being approximately 710 microns long. Thus, the higher the mesh number, the more openings there are on a given sieve, and the smaller those openings are. A group of particles that falls through a 25 mesh sieve, but stays on top of a 30 mesh sieve, will be labeled “25-30 mesh” or “600 — 710 (pm.” Similarly, a group of particles that falls through a 20 mesh sieve, but that stays on top of a 25 mesh sieve, will be labeled “20-25 mesh” or “710 — 850 <p m.” At the time of manufacture and packaging for pharmaceutical use, 100% of the cores in a labeled product have fallen through the sieve with the larger openings and remained on top of the sieve with the smaller openings noted on the label.
To determine whether the particle distribution in the labeled product meets USP specifications, analytical sieving, that is, using test sieves to verify the particle size range, can be performed. This is done by placing a group of particles on a “nest” or stack of sieves with graduated size openings, with the size that is larger than the *271named size of the material on top, the next smaller size underneath, and the smallest named size opening underneath that. The nest is agitated for a period of time and the distribution of the particles on the sieves is noted. To be within ASTM standards, the distribution of particles in a given lot may contain up to 10% cores larger than the largest named sieve and up to 10% cores smaller than the smallest named sieve.
The micron range and the mesh range are used interchangeably in the art; in fact, the ASTM tables refer to the “standard” sieve designation in microns and the “alternative” designation in inches or mesh number. The “alternative designation” mesh number is also referred to as the “U.S. No.” According to ASTM specifications for sieves for testing purposes, it is preferred that sieves with openings smaller than 1/4 inch, or 6.3mm, be identified by the standard designation in millimeters or microns, rather than the mesh number.
Since the particles used as cores are multi-sided, each core has many possible diameter measurements, a “spectrum of diameters” according to Janssen’s expert, Dr. Bodmeier; that is, there is no single diameter measurement. At the very least, each core, when measured in two dimensions, has a minimum and maximum diameter which may differ by as much as 300 microns. In other words, it is possible that a particle could have one diameter that is between 600 and 700 microns, and one diameter that is not.
Furthermore, there is uncertainty in the measuring process itself. An actual micron measurement of any of the diameters of a single core necessarily requires an approximation through one of various types of microscopy including optical and scanning electron microscopy. The minimum diameter measurement as measured by microscopy is also called the “effective sieve diameter” because it is the shortest distance across the particle, as viewed in a two dimensional aspect, and it is considered to be the diameter that will allow the particle to pass through a particular size sieve opening. It is not possible to say conclusively whether the diameter of the third dimension is the same as the others.
The “true' sieve diameter” is the three dimensional diameter that allows the particle to pass through a sieve of a certain size. Neither Dr. Tendler, Janssen’s expert, nor Dr. Brittain, Eon’s expert, could measure.the true sieve diameter by microscope, because neither one could measure in the third dimension. Thus, the microscopic measurements the experts made are not necessarily indicative of whether a particle passed through a certain size sieve.
There is also a diameter that is defined at the time when all the sugar particles are separated by sieves and classified for sale based on a “mesh cut,” that is, the time at which the particles are obtained by a drug manufacturer. ' As described above, a mesh cut defines the range of particle sizes within the labeled product by noting the sieve through which all of the particles passed and the next smaller sieve through which none of the particles passed. Drug manufacturers select cores by mesh cut for pharmaceutical use. No manufacturer measures the size of individual cores. For quality control, a manufacturer performs analytical sieving on a sample of the cores it intends to use to make sure the sieves have produced a product which conforms to the allowable specifications.
The core size which was used during the development of the F12 formulation was selected by, and supplied to, the inventors by mesh cut. None of the Janssen inventors measured any individual sugar core in the F12 formulation to determine its size. Indeed, there.is no evidence that any person at Janssen ever measured an individu*272al core used in the F12 formulation either prior to filing the patent application or after.
The experts, Dr. Tendler for Janssen and Dr. Brittain for Eon, testified that, even when they were selecting cores to measure by microscope to determine whether an individual core diameter was between 600 and 700 microns, they first sieved the cores through a 710 micron (25 mesh) sieve to collect sub-samples that would have at least one diameter smaller than 710 microns.
According to Eon’s ANDA, upon analytical sieving, between 5.4 and 6.8 percent of Eon’s ANDA batch (20-25 mesh) cores fell through a 25 mesh sieve. This is within NF tolerances, as described above, that allow up to 10% of particles in a labeled range to be smaller than the smallest named sieve and larger than the largest named sieve on analytical sieving. Dr. Tendler’s tests on a sample of the cores provided in Eon’s ANDA batch showed that approximately 1.8% passed through a 710 <t m sieve and were retained on a 600 Sm sieve, and about half of those measured between 600 <t m and 700 <pm under an electron microscope. Dr. Tendler was of the opinion that between .9% and 1.8% of Eon’s ANDA cores had a diameter that measured between 600 and 700 <tm.
DISCUSSION
In the claim construction of April 6, 2004, I concluded that the ’015 patent claimed a bead comprised of a core with a diameter between 600 and 700 microns. That the evidence at trial showed that there are many possible diameters of a core, and that a diameter may be measured by different methods and at different times, makes it necessary to look not only at the ordinary definition of diameter, but also to the intrinsic and extrinsic evidence produced at trial and then to explain how the diameter size claimed is to be determined. See Texas Digital Systems, Inc. v. Telegenix, Inc., 308 F.3d 1193 (Fed.Cir.2002).
As noted above, the core size used during the development of the F12 formulation was selected by, and supplied to, the inventors by mesh cut. Even the trial experts, highly skilled in the art, when selecting cores for microscopic measurement, selected their samples by sieving. The specification mentions the word “diameter” only once, when it recites the exact language that is used in Claim 1. ’015 Patent, col. 1, line 51. The core size is referred to as “25-30 mesh.” without including any reference to micron range, six times. There is also one reference in the example to “25-30 mesh (600-700 <t m) sugar spheres.” ’015 Patent, col. 5, line 18. The cores are also defined as suitable if they have “appropriate dimensions (about 25-30 mesh).” ’015 Patent, col. 2, lines 11-12.
Considering all of the facts, I conclude that one of ordinary skill in the art would understand that the patent teaches a composition of one or more beads containing a core with a diameter between 600 and 700 microns as determined at the time the core is classified for use. At that time, there is a labeled range that identifies the diameter as classified by the sieve opening to which it corresponds. In Claim 1, the parenthetical “(25-30 mesh)” notes the mesh number known in the industry to include the 600 to 700 diameter cores. It is not necessary that the cores be selected from a 25-30 mesh process, because there may be a time in the future when particles are classified and sold by a different process. But, whatever that process may be, it is the diameter size as determined at the time the cores are acceptable for use in the industry that will determine whether or not they are claimed in the ’015 patent. Thus, a diameter of 600 to 700 microns *273refers to the diameter as measured at the time the cores are separated by sieves, classified, and made available for sale to the drug manufacturer. This is so even though, on later analytical sieving, another smaller diameter of the core may allow it to pass through the screen upon which it was previously retained.
Put another way, Janssen cannot be claiming something that Janssen itself never measured. It can only be understood as claiming cores that are viewed in the industry as 600 to 700 microns at the time they are classified and obtained for use, even though some of such cores, within the NF standards, prove, on analytical sieving, to have at least one diameter that is smaller than 600 microns or larger than 700 microns.
In sum. I adhere to my ruling that the core size is critical to the invention and that the number range of 600 to 700 microns defines the diameter of the core size claimed. However, I clarify my construction, as to how the size diameter claimed is determined, to mean the diameter as determined at time of manufacture, that is, the time at which people practicing the patent would obtain the sugar spheres, and the time at which the particles are classified and labeled. To the extent that this clarification conflicts with the earlier construction, the earlier construction is amended.
INFRINGEMENT
Literal Infringement
Janssen claims that Eon’s ANDA literally infringes claims 1, 2, 5 and 6 of the ’015 patent.4 To find literal infringement, every limitation in those claims in the ’015 patent must be present in Eon’s ANDA. The properly construed claim must read on the accused device exactly. Amhil Enters., Ltd. v. Wawa 81 F.3d 1554, 1562 (Fed.Cir.1996).
Eon’s ANDA product will be made using cores from a 20-25 mesh cut. These cores are defined as having a diameter between 710 and 850 microns at the time they are classified for use and obtained. All of the 20-25 mesh cores were retained on a 710 micron sieve: none of the particles fell through the 710 micron sieve. By contrast, Janssen’s patented composition claims cores defined as having a diameter between 600 and 700 microns at the time they are obtained: all of the particles would have fallen through a 710 micron sieve. Thus, since none of the diameters of Eon’s cores are literally between 600 and 700 microns, Eon’s ANDA does not read on Claim I and does not literally infringe the ’015 patent. That later analytical sieving, in which cores were placed on a nest of sieves and agitated so that a smaller diameter of a certain number of cores allowed these cores to pass through the smaller apertures of a 25 mesh sieve, or that microscopy showed that some of the diameters of Eon’s cores selected from a 20-25 mesh cut are in the 600 to 700 micron range, does not change this conclusion.
Doctrine of Equivalents
Janssen claims that, even if there is no literal infringement of claims 1, 2, 5 and 6, those claims and claims 7, 8, 11, 12, 13, 14, 17, and 18 are infringed under the doctrine of equivalents. Infringement under the doctrine of equivalents requires that the “accused product or process contain elements identical or equivalent to each claimed element of the patented invention.” Warner-Jenkinson Co. v. Hilton Davis Chemical Co. 520 U.S. 17, 40, 117 *274S.Ct. 1040, 137 L.Ed.2d 146 (1997). A doctrine of equivalents analysis must be applied to the individual claim limitations, not to the invention as a whole. Id. at 29, 117 S.Ct. 1040. The doctrine of equivalents prevents an accused infringer from avoiding infringement by changing only minor or insubstantial details of a claimed invention while retaining the essential functionality of the details. However, the doctrine is limited so as not to conflict with the importance of the claims in determining the scope of a patentee’s exclusive rights. The application of the doctrine requires resolving the conflict between preventing “fraud on a patent” and the primacy of the claims in defining the scope of a patentee’s exclusive rights. See Sage Products, Inc. v. Devon Industries, Inc. 126 F.3d 1420, 1424-25 (Fed.Cir.1997).
Eon makes several arguments as to why its ANDA does not infringe under the doctrine of equivalents. First, it argues that, under Sage, the use of 20-25 mesh cores was foreseeable to the person who drafted the claims in the ’015 patent and, since that size was not claimed, Janssen cannot expand the scope of the claim to cover what was foreseeable at the time. Alternatively, it argues that the difference in core size is not insubstantial. Finally, it argues that the 20-25 mesh core does not perform substantially the same function in substantially the same way with substantially the same result as the 25-30 mesh core in the ’015 patent. See Graver Tank & Mfg. Co. v. Linde Air Prods. Co., 339 U.S. 605, 608-10, 70 S.Ct. 854, 94 L.Ed. 1097 (1950).
The foreseeability analysis under Sage depends upon whether a skilled patent drafter would foresee the limiting potential of the words “characterized in that the bead has a core of from about 600-700 <t m (25-30 mesh).” See Sage, 126 F.3d at 1425. “For a patentee who has claimed an invention narrowly, there may not be infringement under the doctrine of equivalents in many cases, even though the pat-entee might have been able to claim more broadly. If it were otherwise, then claims would be reduced to functional abstracts, devoid of meaningful structural limitations on which the public could rely.” Id. at 1424.
Although I have held that the prosecution history did not estop Janssen from asserting infringement under the doctrine of equivalents, see Order of April 6, 2004 at 9, the evidence at trial established that other size cores were known to, and even tried and rejected by, Janssen. As early as 1989 Janssen knew that the problem of high residual solvent levels could be addressed by using cores smaller than those obtained from an 18-20 mesh product as was used in the F5 formulation, and it tested 25-30 mesh and 30-35 mesh cores.
If Janssen desired broad patent protection for any size core that performed a similar function, it could have claimed a larger size range of core diameters than 600-700 <f m (25-30 mesh). Had it done so, the novelty of the larger range could have been scrutinized by the Patent and Trademark Office. However, rather than claiming a broader range of core sizes, or even suggesting that a broader range was claimed, Janssen explained in the specification that the size of the core was of particular importance and that cores that are “too small” or “too large” were not “optimal.” There was a specific reason that the core size of 600-700 $ m was claimed: it allowed the bead to be manufactured so that problems of agglomeration and high residual solvent levels were avoided. Furthermore, Janssen specifically disclosed cores of 30-35 mesh, the next smaller standard mesh size, that contains cores from 500 to 600 <t m, and stated that these were “too small.” ’015 Patent, col. 2, line 4. Although it did not give an example by *275micron number or mesh cut of cores that were “too large,” it did disclose cores said to be “too large.”
The next larger standard mesh size is 20-25 mesh, which contains cores from 710 to 850 <t m in diameter. That Janssen did not_specifically disclose 20-25 mesh cores does not allow Janssen to claim that size now. If Janssen desired patent protection for the 20-25 mesh cores, and meant to exclude only 18-20 mesh cores that proved unsuitable in the F5 formulation as “too large,” it could have characterized the core size claimed with a broader range. “As between the patentee who had a clear opportunity to negotiate broader claims but did not do so, and the public at large, it is the patentee who must bear the cost of its failure to seek protection for this foreseeable alteration of its claimed structure.” Sage, 126 F.3d at 1425.
Similarly, in light of its emphasis on the importance of the particular core size in solving residual solvent problems and its characterizing the bead by its core size, Janssen cannot now claim that the difference in core size between the ’015 patent and Eon’s ANDA is insubstantial. See Tanabe Seiyaku Co., Ltd. v. United States International Trade Commission, 109 F.3d 726, 732-33 (Fed.Cir.1997). Dr. Bod-meier’s testimony that a core up to 100 microns smaller (30-35 mesh) is substantially different from a 25-30 mesh' core, but that a core up to 140 microns larger (20-25 mesh) is not, is unpersuasive. To conclude that the difference in core size is insubstantial would, read out the very characterization of the bead that Janssen sought to claim. See Novartis Pharmaceuticals Corp. v. Eon Labs, 363 F.3d 1306, 1312 (Fed.Cir.2004).
One of ordinary skill in the art, reading the ’015 patent, would not think that a 20-25 mesh core was interchangeable with the “optimal” size core nor that it was equivalent. One of ordinary skill in the art would conclude that the only size core taught, and claimed, by the ’015 patent was a core of 600-700 (pm (25-30 mesh) and that smaller or larger mesh cut cores were not claimed. See Tanabe, 109 F.3d at 732.
Janssen waited to patent its invention until it found a formulation that reached acceptable residual solvent levels for FDA approval. That it chose to wait until it found the precise core size for the formulation that resulted in FDA approval further supports that Janssen dedicated all smaller and larger core sizes to the public. Having found that the difference in core size is not insubstantial and that the cores claimed in the ’015 patent and Eon’s ANDA cores are not equivalent, I need not address function/way/result analysis. See Warner-Jenkinson, 520 U.S. at 39-40, 117 S.Ct. 1040.
In sum, Eon’s ANDA does not infringe any of the claims of the ’015 patent either literally or under the doctrine of equivalents. Since all of the claims that are alleged to be infringed are dependent upon Claim 1, and I find that Claim 1 is not infringed either literally or under the doctrine of equivalents, I need not address every other claim individually. None of the claims are infringed.
AFFIRMATIVE DEFENSES AND COUNTERCLAIMS
Prior to trial, Eon and Janssen each moved for summary judgment on Eon’s affirmative defense of invalidity, and Jans-sen moved for summary judgment on Eon’s defense that the ’015 patent was unenforceable because of inequitable conduct before the Patent and Trademark Office and because Janssen filed this lawsuit as an anti-competitive act with no basis to believe that Eon’s ANDA infringed the ’015 patent.
*276 Unenforceability
Magistrate Judge Marilyn D. Go recommended that Janssen’s motion based on unenforceability be granted. Eon did not file any objections to that recommendation, nor did it present any evidence at trial to show that the patent was unenforceable on the grounds alleged. Therefore, Eon’s affirmative defenses and counterclaims based upon unenforceability are dismissed.
Public Use
Judge Go also recommended that I grant summary judgment to Janssen on Eon’s claim of invalidity based upon public use. As noted earlier, because of the preliminary injunction motion and the expedited timing of the trial, the responses to Judge Go’s report were held for consideration until after the court allowed counsel to develop additional facts at trial. The evidence at trial confirmed the correctness of Judge Go’s conclusion.
Eon claims that the ’015 patent is invalid because it was in public use, or on sale, prior to the “critical date” under 35 U.S.C. § 102(b), which is the date one year prior to the date of the United States patent application. “Whether a patent is invalid due to public use under § 102(b) is a question of law based upon underlying questions of fact.” Smithkline Beecham Corporation v. Apotex Corp., 365 F.3d 1306, 1316 (Fed.Cir.2004). “Public use under 35 U.S.C. § 102(b) includes any use of the claimed invention by a person other than the inventor who is under no limitation, restriction or obligation of secrecy to the inventor.” Netscape Communications Corp. v. Konrad, 295 F.3d 1315, 1321 (Fed.Cir.2002).
As a preliminary matter, the court notes that the compassionate clearance tests, which Eon claims were a public use, were performed on the F5 formulation, which, since it used cores of 18-20 mesh, is not the patented composition. For that reason alone, the public use defense as to the compassionate clearance tests fails.
In any event, the evidence introduced at trial established that all of Janssen’s compassionate clearance tests that were performed on the F5 formulation, as well as the clinical trials conducted from June to August 1991 with the F12 formulation, were confidential and controlled by Jans-sen. As to the compassionate clearance tests on the F5 formulation, the evidence shows that the doctors were under a strict protocol, controlled by Janssen; that they had to follow the protocol or the drug would not be shipped to them; that they were under a confidential arrangement with Janssen; that they were not informed of the formula for SPORANOX®; and that the labels on the boxes of SPORA-NOX® stated that it was an investigational drug not for commercial use. That the patient, who did not know any information about the drug other than its name and that it was an investigational antifungal, was not under a confidentiality agreement with Janssen, does not negate the confidential and controlled use of the F5 formulation.
As to the clinical tests performed on the F12 formulation between June and August 1991, a confidentiality statement in the protocol for those studies states that the information in the document contains trade secrets and commercial data that are privileged or confidential and may not be disclosed unless required by law. The evidence supports Janssen’s position that the use was confidential and controlled by Janssen. Thus, Eon has not shown by clear and convincing evidence that any of the uses of the F5 or F12 formulations of SPORANOX® capsules prior to the critical date were public uses.
On Sale Bar
“Patents are presumed to be valid and an accused infringer challenging *277the validity of a patent under the on-sale bar must demonstrate by clear and convincing evidence that there was a definite sale or offer to sell more than one year prior to the application for the subject patent, and that the subject matter of the sale or offer to sell fully anticipated the claimed invention or rendered it obvious.” Elan Corp. v. Andrx Pharms., Inc. 366 F.3d 1336, 1340 (Fed.Cir.2004). (Internal citations omitted.) “Only an offer which rises to the level of a commercial offer for sale, one which the other party could make into a contract by simple acceptance ..., constitutes an offer for sale under § 102(b).” Id. at 1341.
During discovery, the parties accepted Judge Go’s adoption of Janssen’s proposed applicable critical date of September 3, 1991, one year prior to the European Patent application priority date noted on the ’015 patent. Judge Go also relied on that date in her report and recommendation on the parties’ motions for summary judgment on the public use affirmative defense. Neither party objected to the use of that date in the report and recommendation. On the eve of trial, Eon filed a Memorandum asserting, for the first time, that the correct critical date is August 27, 1992, one year prior to the Patent Cooperation Treaty filing date. Relying on that date as the critical date. Eon raised an “on-sale” bar invalidity defense at trial, claiming that an August 17, 1992 letter purportedly announcing the upcoming availability of SPORANOX® capsules constituted an offer for sale prior to the critical date.
This attempt to change the applicable critical date the day before trial began was untimely. Janssen had no opportunity, given the lateness of the claim and Eon’s previous representations that it would not raise an on-sale defense, to offer any witnesses to explain the purpose and meaning of the letter. It now seeks to reopen the trial to present witnesses on its behalf if the court accepts the letter. In fairness, were I to consider the letter offered by Eon sufficient to establish that it is an offer for sale, I would allow Janssen to reopen the trial to rebut. However, since it is Eon’s burden to prove that the invention was offered for sale by clear and convincing evidence, and since Eon failed to do so, there is no need to reopen the trial.
The letter, dated August 17, 1992, appears to be signed by the Director of Trade Relations at Janssen and is addressed to a senior pharmaceutical buyer at a wholesale company announcing the launch of SPORANOX® antifungal capsules. The letter states that Janssen expects final FDA market approval in the near future and that the product will be sold in cases containing 24 boxes of capsules; it also states the list price per box and per case, and the size and weight of the case. According to the letter, a recommended opening order, based upon the wholesaler’s sales of a competitive antifungal product, has been attached and all orders received through October 30, 1992 will receive an extra 8%% discount, plus an additional 60 days dating. Finally, the letter states that “your cooperation with this recommended opening order is greatly appreciated, and we ask that you forward the order by September 4, 1992 to our distribution center. This will help ensure a rapid launch of this new product.” The attached sheet lists a certain number of cases for five distribution centers and a dollar figure that represents the amount of sales of the competitor’s product in that location.
It is far from clear that the only thing that any recipient of this letter had to do to form a binding contract was to “accept” the recommended order quantity. The letter is clear that the product is not yet *278available and is phrased in terms of what will occur at some point in the near future. Eon has not provided any evidence that shows whether or not the letter was ever sent to, or received by, any person, and no evidence of what the custom or practice in the industry was in regard to a letter of this type. Thus, even if I were to accept Eon’s proposal to change the critical date to August 27, 1992, the letter, standing alone, does not show by clear and convincing evidence that Janssen made a commercial offer for sale of the invention prior to that date.
In sum, I find that the ’015 patent is not invalid based upon public use or because it was the subject of an offer for sale.
CONCLUSION
For the reasons discussed above, I find that claims 1, 2, 5 and 6 of U.S. Patent 5,633,015 are not infringed literally or under the doctrine of equivalents. I also find that claims 7, 8, 11, 12, 13, 14, 17 and 18 are not infringed under the doctrine of equivalents.
The temporary restraining order is hereby lifted and the preliminary injunction motion is denied. The Clerk of Court is directed to enter judgment for Eon that its ANDA does not infringe the ’015 patent. The Clerk of Court is also directed to enter judgment for Janssen dismissing Eon’s claims of unenforceability and invalidity.
SO ORDERED.