RICH, Circuit Judge.

Scaltech, Inc. (Scaltech) appeals from the judgment of the United States District Court for the Southern District of Texas, H-95-4190, granting a motion for summary judgment in favor of Retee/Tetra, L.L.C. (Retec), holding U.S. Patent No. 5,433,717 invalid because an embodiment of the claimed invention was “on sale” within the meaning of 35 U.S.C. § 102(b). We vacate and remand.

BACKGROUND

Scalteeh is in the business of recycling industrial waste produced during the refining of petroleum products. Retec is in the business of producing delayed coker quench streams for use in producing delayed petroleum coke. Their paths crossed when they both used waste products from the petroleum refinery process to produce coke, albeit apparently for different purposes — Retec for producing coke and Scaltech for disposing of waste.

Scaltech is the assignee of record of U.S. Patent No. 5,433,717 (the ’717 patent) issued 22 August 1995 on the application of Robert M. Scalliet, filed on 19 January 1993 and entitled “Recycle of Waste Streams.” The ’717 patent is generally directed to a process for the disposal of oil refinery waste in conjunction with a “delayed coking” process by which oil refiners produce coke. Coke is a porous solid that is produced as a by-product in the oil refining process. It is frequently burned as a fuel.

Production of coke by the “delayed coking” method involves heating the crude oil residue and introducing it into a vessel called a coker drum under specified conditions that result in transformation of the crude oil residue into coke. The coke is then cooled, or “quenched,” by the controlled introduction of an aqueous slurry of solids, or “quench stream,” into the coker drum. According to the ’717 patent, quantities of refinery waste may be successfully disposed of in the quench stream of the delayed coker unit if the waste introduced to the quench stream is comprised of greater than seventy percent of the solids having a particle size less than 15 microns. The ’717 patent teaches that this may be accomplished by the treatment of the waste before it is introduced into the quench stream to remove most of the free oil or mobile organic material and reduce the solid particle size. This leaves the de-oiled waste solids suspended in water, or an aqueous slurry of the waste solids, which is then introduced into the delayed coker quench stream.

Claims 1 and 6 are the only independent claims in the ’717 patent. Claim 1 reads:

In a process for producing delayed petroleum coke, wherein a liquid hydrocarbon feed stream is introduced into a delayed coking vessel under delayed coking conditions and the coke produced is quenched, the improvement comprising:

treating a waste stream containing water, organic compounds and solids so as to *1195 cause attrition of said solids to produce a delayed coker quench stream containing from about 5 to about 35% by weight solids, water and less than about 6% by weight mobile organics, said solids in said coker quench stream having a particle size distribution such that greater than about 70% of the total solids volume comprises solids having a particle size of less than about 15 microns; and

introducing said coker quench stream into said coking vessel during quenching. Claim 6, the broadest claim, reads:

A process for producing delayed coker quench stream for use in producing delayed petroleum coke wherein a liquid hydrocarbon feed stream is introduced into a delayed coking vessel under delayed coking conditions and the coke produced is quenched, comprising:

treating a waste stream containing water, organic compounds and solids so as to cause attrition of said solids to produce a delayed coker quench stream containing from about 5 to about 35% by weight solids, water and less than about 6% by weight mobile organics, said solids in said coker quench stream having a particle size distribution such that greater than about 70% of the total solids volume comprises solids having a particle size of less than about 15 microns.

Sealtech brought suit against Retec for infringement of the ’717 patent. After completing a modicum of discovery, Retec raised the affirmative defense of patent invalidity under 35 U.S.C. § 102(b), alleging that Seal-tech sold or offered for sale a process embodying the claimed invention of the ’717 patent more than one year before its filing date. On 24 February 1997, Retec filed a motion for summary judgment seeking a holding of patent invalidity.

The district court granted summary judgment in favor of Retee, holding the patent invalid under 35 U.S.C. § 102(b) as a result of an “on sale” bar. This appeal followed.

DISCUSSION

On summary judgment, the district court found the invention claimed in the ’717 patent was “on sale” within the meaning of 35 U.S.C. § 102(b). Summary judgment is appropriate when there is no genuine issue as to any material fact and the moving party is entitled to a judgment as a matter of law. Fed.R.Civ.P. 56(e). We undertake plenary review of a grant of summary judgment. See KeyStone Retaining Wall Sys., Inc. v. Westrock, Inc., 997 F.2d 1444, 1449, 27 U.S.P.Q.2d 1297, 1301 (Fed.Cir.1993).

The following facts pertinent to the alleged offer for sale are either undisputed or represent non-movant Scaltech’s version. See Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 255, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986). In 1987 Sealtech was treating refinery waste at Chevron’s refinery in Port Arthur, Texas. Sealtech was using a Guinnard DC-6 vertical disk centrifuge (the DC-6 centrifuge) to remove the oil from the raw waste, return the oil to the refinery, and press the wet solids into cakes for shipment to hazardous land fill locations.

While Sealtech was performing this contract, Chevron began testing an alternative solution for disposing of its refinery wastes involving injecting the waste into the coking process. Coke is made at the end of the refining process by injecting the unrefined hydrocarbons into large drums, called coker units. In the coker unit it is heated to a very high temperature in the “coking cycle.” During the coking cycle the hydrocarbon materials are converted to solid coke fuel. Pri- or to removal from the coker unit, water is introduced to cool the coke bed. The cooling process is called the “quench cycle.”

Injection of refinery waste into the coke bed during the quench cycle was initially suggested by Mobil Oil Company in 1975 and is described in detail in United States Patent No. 3,917,564 (the Meyers Patent). When waste injection rates began to exceed one to two pounds of solids per ton of coke, however, refineries frequently encountered unacceptable problems with the process. Chevron was no exception. It struggled with uneven solid formation in the coke, noxious *1196 odors, and soft spots in the coke bed. Thus, this method did not appear to be a viable method for the disposal of all its waste products.

When Chevron asked its waste disposal contractor, Scaltech, for suggestions to improve the process, the named inventor of the ’717 patent, Robert Scalliet, suggested that the above-outlined problems might be caused by excessive oil in the waste feed. He theorized that oil might be plugging the coke pores in the coker unit. Scalliet offered to provide a few loads of the de-oiled waste sediments, produced by processing the waste through a centrifuge, so that Chevron could test Scalliet’s theory. It was using the de-oiling method to prepare the hazardous waste cakes which were tucked into landfill. Chevron picked up one or two truck loads of the waste matter and apparently injected it into its coker unit. Scaltech did not actually participate in the injection. It only provided the de-oiled waste sediment that was part of its hazardous waste disposal process.

Scaltech’s contract to make hazardous waste cakes for Chevron was not renewed. In 1988, Scaltech heard third hand that Chevron was pleased with the results of the injection of de-oiled waste into its coke. Based on these promising comments, Scal-tech thought that it might have a method to overcome the one to two pound waste per ton of coke limitation posed by the Meyers patented process. It sought access to a coker unit such that it could test its theories. Between 1988 and 1991, Scaltech contacted six different refineries and proposed to use the DC-6 centrifuge to process refinery waste. In four of the proposals, it described the process as a test or a trial. In two, a 30 March 1988 proposal to Chevron and a 15 November 1988 proposal to Champlin, it did not expressly identify the procedure as experimental. The court relied on these final two proposals to invalidate the ’717 patent. In none of the proposals did Scaltech indicate that it would control either (1) the particle size to be injected in the coker unit or (2) the specific solids concentration in the material to be delivered to the coker unit. The only equipment described was the DC-6 centrifuge which it had been using to de-oil the solids at Chevron.

In December 1991 Scaltech entered into a contract with CITGO to separate the oil from its refinery waste and process the hazardous waste sediment. At that time CITGO opted to dispose of hazardous waste sediment in a cement kiln rather than in the hazardous waste cakes Chevron was producing. While performing this contract, Scaltech received permission, in early 1992, to test the process of disposing of the de-oiled waste by injection into the coker quench cycle.

It carefully monitored the oil content of the waste feed run into the DC-6 centrifuge and the waste solids slurry run into the coker unit. It determined that the waste solids slurry was virtually free of oil and, to their surprise, it discovered that the particles were also of very small size. The particle size was not being reduced by the DC-6 centrifuge itself, but rather by the ejection of the solid slurry onto a cast iron plate as it exited the centrifuge at high velocity. It concluded that the problems of the Meyer patent process were that the oil and the large particles wtere blocking the pores and preventing high absorption rates throughout the coke bed. By carefully monitoring the solid injection rates and gradually increasing the solid concentrations it was able to inject 18 pounds of solids per ton of coke. Having discovered the necessary conditions of small particle sizfe and high solids concentration, it then filed a patent application.

Ill

The question on appeal is whether an embodiment of the claimed invention was being offered for sale within the meaning of section 102(b). We conclude that, on the record before us, the district court erred by failing to address whether an embodiment of the claimed invention or, in the alternative, a substantially completed embodiment of the claimed invention with reason to expect that it would work for its intended purpose, was offered for sale.

Section 102(b) reads in relevant part: “[a] person shall be entitled to a patent unless ... the invention was ... in public use *1197 or on sale in this country, more than one year prior to the date of the application for patent in the United States.” 35 U.S.C. § 102(b) (1994). A claimed invention is considered to be on sale within the meaning of section 102(b) if, more than one year before the filing date to which the claim is entitled (the critical date), the claimed invention was embodied in or obvious in view of what was offered for sale, was operable, and the sale or offer for sale was primarily for profit (commercialization) rather than for experimental purposes. See King Instrument Corp. v. Otari Corp., 767 F.2d 853, 859-60, 226 U.S.P.Q. 402, 406 (Fed.Cir.1985); KeyStone Retaining Wall Sys., Inc., 997 F.2d at 1451, 27 U.S.P.Q.2d at 1302. Integral to our analysis are “[a]ll of the circumstances surrounding the sale or offer to sell, including the stage of development of the invention and the nature of the invention, [which] must be considered and weighed against the policies underlying section 102(b).” UMC Elecs. Co. v. United States, 816 F.2d 647, 656, 2 U.S.P.Q.2d 1465, 1471-72 (Fed.Cir.1987).

In this case we need only touch on the policies 1 because the district court’s error is found in threshold requirements that it failed to address — the determination of whether the subject of the barring activity was actually an embodiment of the claimed invention or, in the alternative, the determination of whether what was offered for sale was “a substantially completed invention, with reason to expect that it would work for its intended purpose upon completion.” Micro Chem., Inc. v. Great Plains Chem. Co., 103 F.3d 1538, 1545, 41 U.S.P.Q.2d 1238, 1244 (Fed.Cir.1997) (citing Seal-Flex, Inc. v. Athletic Track & Court Constr., 98 F.3d 1318, 1322 (Fed.Cir.1996)).

The “invention” which has been offered for sale must, of course, be something circumscribed by metes and bounds of the claim. See Graver Tank & Mfg. Co. v. Linde Air Prods. Co., 339 U.S. 605, 607, 70 S.Ct. 854, 94 L.Ed. 1097 (1950) (a claim in a patent provides the metes and bounds of the right which the patent confers on the patentee to exclude others from making, using, or selling the protected invention). Hence, the first determination in the section 102(b) analysis must be whether the subject of the barring activity met each of the limitations of the claim, and thus was an embodiment of the invention. In this case, claim 6, set forth above, is the broadest claim. It specifically requires that the “delayed coker quench stream [contain] ... about 5 to about 35% by weight solids ... less than about 6% by weight mobile organics [and the solids] ... having a particle size distribution such that greater than about 70% of the total solids volume comprises solids having a particle size of less than about 15 microns.” The record does not indicate whether an embodiment of the invention was offered, i.e., whether the limitations of small particle size and high solids concentration were met by what was offered in the 30 March 1988 proposal to Chevron or the 15 November 1988 proposal to Champlin. 2 We note that there is no requirement that the offer specifically identify these limitations. See, e.g., RCA Corp. v. Data General Corp., 887 F.2d 1056, 1060, 12 U.S.P.Q.2d 1449 (Fed.Cir.1989); Sonoscan, Inc. v. Sonotek, Inc., 936 F.2d 1261, 1263, 19 U.S.P.Q.2d 1156 (Fed.Cir.1991).

If an embodiment of the claimed invention was not offered for sale, we look to see if a substantially completed embodiment of the claimed invention was offered for sale and *1198 “there was reason for a high degree of confidence that it would work for its intended purpose,” Micro Chem., Inc., 103 F.3d at 1545, 41 U.S.P.Q.2d at 1244. On the record before us, the invention was not conceived until Scaltech entered into the contract with CITGO in December 1991 and had the opportunity to develop the claimed invention and discover the particle size and solids concentration limitations beginning in February 1992. This is after the critical date of 19 January 1992. In UMC Elecs. Co. v. United States, we indicated that

we do not intend to sanction attacks on patents on the ground that the inventor or another offered for sale, before the critical date, the mere concept of the invention. Nor should inventors be forced to rush into the Patent and Trademark Office prematurely.

If the inventor had merely a conception or was working towards development of that conception, it can be said that there [was not] ... any “invention” which could [have been] placed on sale.

816 F.2d 647, 656-57, 2 U.S.P.Q.2d 1465, 1471-72 (Fed.Cir.1987). This is consistent with the policy of preventing delay in the expiration of patent protection by delaying the filing of the application more than a year after substantial completion of the invention. Both the reasonable expectation that the claimed invention will work for its intended purpose and the prevention of the delay require that the inventor understand the nature of the invention in order to contemplate its sale. In other words, in order to have a reasonable expectation that he can reduce the invention to practice he must have conceived of the invention.

In the case before us, Scaltech offered to treat the waste prior to injecting it into the coker unit. It believed that de-oiling the waste would increase the amount of waste which could be consumed by the coke bed. It is only in hindsight that Scaltech recognized that two claim limitations, the particle size and the solid concentration, were critical to achieve this purpose and that those limitations might be met by practicing some of the process apparently offered in 1987-1991. Accordingly, on the record before us, the inventor was working toward the development of a conception, and there was not yet an “invention” which could have been offered on sale. See UMC Elecs. Co., 816 F.2d at 656-57, 2 U.S.P.Q.2d at 1471-72. We further note that an accidental or unwitting duplication of an invention cannot constitute an anticipation. See Tilghman v. Proctor, 102 U.S. 707, 26 L.Ed. 279 (1880); Eibel Process Co. v. Minnesota & Ontario Paper Co., 261 U.S. 45, 43 S.Ct. 322, 67 L.Ed. 523 (1923).

Finally, we note that neither the court nor the parties addressed what occurs if the subject of the barring activity did not embody the claimed invention, but was still rendered technical prior art for the purposes of a 102(b)/103 type obviousness analysis. 3 Nor did the court or the parties address whether the Chevron use in 1988 was an experimental use within the meaning of Lough v. Brunswick Corp., 86 F.3d 1113, 39 U.S.P.Q.2d 1100 (Fed.Cir.1996). 4

We have considered the parties’ other arguments and conclude that they are either *1199 unpersuasive or unnecessary for resolution of this appeal. Accordingly, we vacate the court’s ruling on summary judgment that the ’717 patent is invalid under the on sale bar and remand for further proceedings consistent with this opinion.

VACATE AND REMAND.

. We have identified the policies underlying the on sale bar as: (1) discouraging the removal from the public domain of inventions that the public reasonably has come to believe are freely available; (2) favoring the prompt and widespread disclosure of inventions; (3) allowing the inventor a reasonable amount of time following public use or sales activity to determine the potential economic value of a patent; and (4) prohibiting the inventor from commercially exploiting the invention for a period greater than the statutorily prescribed one year grace period. King Instrument Corp. v. Otari Corp., 767 F.2d 853 (Fed.Cir.1985). The policy underlying experimental use negation is that of allowing the inventor the opportunity to reduce the embryonic invention to practice; to produce an embodiment and' — at least for utility patents — to demonstrate that the invention will work for its intended purpose before filing a patent application. T.P. Lab., Inc. v. Prof'l Positioners, Inc., 724 F.2d 965 (Fed.Cir.1984).

. These are the offers that the district court relied upon in invalidating the claims of the '717 patent.

. Even in this context, the section 102(b) technical prior art (some apparatus or method that is less than the claimed invention) would not be subject to a reduction to practice requirement, because the reduction to practice requirement would not be applied to anything other than the claimed invention.

. Scaltech argues that its invention was still experimental at the time Scaltech was soliciting an opportunity to practice the invention. This argument fails because it is premised on the “experimental stage” doctrine which has been rejected by both this court and the Supreme Court. See Lough v. Brunswick Corp., 86 F.3d 1113 (Fed.Cir. 1996); City of Elizabeth v. American Nicholson Pavement Co., 97 U.S. 126, 24 L.Ed. 1000 (1877). The experimental stage doctrine finds its genesis in Eastman v. Mayor, Etc. of City of New York, 134 F. 844, 858 (3d Cir.1904) ("As soon as the invention is completed, viz.: 'in such a condition that the inventor can apply for a patent for it,' the [grace] period begins to run and the application must be made within this period”), and has been advocated sporadically since then. See, e.g., Pharmacia, Inc. v. Frigitronics, Inc., 726 F.Supp. 876, 885-86 (D.Mass.1989). However, the policy underlying experimental use negation is to give the inventor an opportunity to reduce the invention to practice, not the time to do so. City of Elizabeth, 97 U.S. at 135-36.