This appeal is from the decision of the Patent Office Board of Appeals sustaining the examiner’s rejection of all the claims in appellants’ application,1 on the basis of 35 U.S.C. § 103.
The Invention
Appellants’ invention concerns a phosphate-containing composition suitable for coating metal surfaces and serving as a base to which paint will adhere. The specification states:
At the present time * * * many articles which are to be phosphate coated are made of more than one metal, and, thus, may have surfaces of any two or all three of iron, zinc and aluminum. Additionally, it has now become a more common practice to handle iron, zinc and aluminum articles at the same coating line location, so that the development of a coating solution and process which may be used satisfactorily to coat all three of these metals is highly desirable. Unfortunately, up to the present time it has not generally been possible to coat steel, zinc and aluminum in the same * * * phosphate coating bath, so as to provide an amorphous coating on the surface of all three of these metals which will permit the deformation of *1319the painted metal surface without adversely affecting the paint film.
Appellants solve this problem by using a coating composition such as that set forth in claim 11:
11. A composition suitable for coating surfaces of zinc, steel and aluminum which consists essentially of an acidic aqueous solution, substantially free of alkali metal ions, containing an ammonium primary phosphate, fluoride ions and an accelera-ator selected from water soluble compounds of tungsten and molybdenum, said fluorides being selected from water soluble fluorides, fluoroborates and silieofluorides.
The preferred composition contains primary ammonium phosphate, ammonium flouride, either ammonium molybdate or ammonium tungstate, and phosphoric acid. Appellants also claim a coating method, exemplified by claim 6:
6. A method of forming a paint-base coating on the surfaces of zinc, steel, and aluminum, which comprises contacting the surface to be treated with the coating solution as claimed in Claim 11, and maintaining the solution in contact with the surface for a period of time sufficient to form the desired coating thereon.
The References
The primary reference is a patent to Ross et al. (Ross)2 Ross deals with phosphate coating compositions for ferrous metal surfaces. Ross discloses coating compositions containing an alkali metal phosphate, e. g., sodium dihydrogen phosphate, and also containing sodium silicofluoride and sodium molybdate. The molybdate is added merely to impart a bluish color to the coating, and not for its properties as an accelerator. In fact, one of the objects of the Ross invention was to provide a phosphate coating composition which would not require accelerators.
Amundsen et al. [Amundsen]3 is directed to surface coatings for “metals, such as iron, steel and zinc.” Amundsen states that by using “mono alkali metal or mono ammonium phosphates in conjunction with alkali or ammonium fluorides, or bifluorides, coating solutions can be made up for iron, steel or zinc, which * * * form uniform films in a very short time.”
Dodd et al. [Dodd]4 discloses phosphate coatings for the surface of “metals such as ferrous metals and other metals such as aluminum, zinc, magnesium and template5 * * The Dodd composition combines a primary alkali metal or ammonium phosphate with a molybdenum or tungsten accelerator compound and a small quantity of a phenolic substance. Among the compounds mentioned as suitable accelerators are the ammonium, sodium and potassium molybdates and tungstates. Dodd acknowledged that phosphatizing baths using water soluble molybdenum compounds as accelerators had previously been known, but stated that such baths tend “to have poor adherence to the metal surface and have an undesirable soft or ‘spongy’ nature rendering them inefficient as rust retarders and of poor quality in forming a strong bond with paint, enamel and other organic finishes.” Dodd states that he solved that problem by the addition of his phenolic material, which is preferably a tannin or a tannin derivative.
Stapleton6 discloses compositions for forming phosphate coatings on iron or steel. Stapleton’s invention is aptly sum*1320marized in the following exerpt from that patent:
It has been discovered that excellent, fine grained, uniform, tenaciously adherent iron phosphate coatings may be formed on the alkali treated metal surface after rinsing with water by treating the metal surface with a phosphate coating solution consisting essentially of alkali metal dihydrogen phosphate containing an accelerator which consists essentially of alkali metal nitrate combined with a small proportion of alkali metal molybdate. * * -X-
In referring herein to “alkali metal” dihydrogen phosphates, “alkali metal” nitrate, and “alkali metal” molybdate, it is of course intended that the sodium, potassium and ammonium compounds of those radicals be included, together with lithium and the other well known alkali metals.
The Rejection
The examiner rejected the claims over Ross in view of Stapleton, Amundsen and Dodd. The examiner stated that “Ross is fully responsive to applicant’s claimed invention except that alkali metal cations are recited in lieu of ammonium ions.” The examiner took the position that it would have been obvious to substitute ammonium for alkali metal in Ross’s composition, since all three of the other references teach that they are equivalent. The examiner stated:
It would also be obvious to coat any of the substrates recited in applicant’s preamble because all the references teach that ferrous substrates can be coated and Dodd teach [sic] that all three substrates are equivalent.
In order to overcome the examiner’s rejection, appellants submitted an affidavit of one Karim I. Saad, under Patent Office Rule 132. Mr. Saad had received a B.S. in chemical engineering in 1959, and from that time until December, 1968, the date of the affidavit, he had worked in the area of phosphate coatings. In view of the importance of the affidavit to our decision, it appears in major part below:
THAT, under his direction, the following experimental work was carried out:
Three aqueous coating baths were formulated containing the following components in the amounts indicated:
Bath A — Components Grams/Liter
Sodium dihydrogen orthophosphate 12.5
Sodium flouride 3.4
Phosphoric acid (75% H3PO4) 4.1
Sodium molybdate (Na2MoO<t.2H20) 1.35
Bath B — Components Grams/Liter
Ammonium dihydrogen ortho phosphate 12.0
Ammonium fluoride 3.0
Phosphoric acid (75% H3PO4) 4.1
Ammonium molybdate (54% Mo) 1.0
Bath C — Components Grams/Liter
Sodium dihydrogen ortho phosphate 7.5
Sodium dihydrogen pyrophosphate 6.0
Sodium silico fluoride 0.75
Sodium molybdate 0.022
Igepal CA 630 wetting agent 0.75
Each bath was made up in a total bath volume of nineteen liters. The above bath concentrations and a bath temperature of 150°F were mentioned [sic. maintained?] while 4-inch x 12-inch panels of cold rolled steel, hot dip galvanized and aluminum, alloy 3003, were processed through the bath, six at a time, two each of steel, zinc and aluminum. The contact time of the bath with each of the panels was one minute. After processing a total of about forty-one square feet each of steel, zinc and aluminum, the last panels of each metal treated were examined to determine the characteristics of the coatings obtained.
This examination showed that, while all three baths produced a visible, uniform coating on the steel, only baths A and B produced a visible coating on the hot dip galvanized and only Bath B produced a uniform, dark coating on the aluminum.
On the aluminum, Bath B produced a good, uniform, dark-colored coating, while Bath A produced a mottled, nonuniform coating which was very light in color and barely visible and Bath C did not produce a visible coating. Samples *1321of the panels evaluated are attached as an Exhibit.
THAT, based on the above results, he has concluded as follows:
1. Although all three baths produce a satisfactory, visible coating on steel, only Baths A and B produce such a coating on hot dip galvanized (zinc) and only Bath B produces a uniform, dark colored coating on aluminum.
2. The results obtained, in terms of the coatings produced on aluminum, are not the same with all three baths.
3. The dark colored coating produced on aluminum with Bath B is more uniform and because of its dark color, more satisfactory from a commercial standpoint because of the ease of making a visual inspection of it.
4. It was surprising and unexpected to find that this much darker and more uniform coating could be produced on aluminum by using ammonium ions in the bath (Bath B) rather than sodium ions (Baths A or C).
While the affidavit says nothing about why baths of the particular concentrations used were chosen, the following appears in a letter from appellants’ attorney, which accompanied the affidavit:
In this Affidavit, three baths were formulated, Bath A being a bath comparable to those of the present invention except that sodium cations were substituted for the ammonium cations. Bath B is a bath formulated with ammonium cations in accordance with Example 1 of the present application and Bath C is a bath formulated in accordance with Example 1 of the Ross et al patent 3,060,066.
The examiner’s response to the affidavit and the arguments accompanying it was as follows:
If one accepted applicant’s argument that a showing of unexpected results would permit him to obtain a patent [,] applicant would not be able to obtain a patent because the affidavit only shows that at 1 minute exposure and at 150 °F applicant obtains a difference in coating each metal. The parameters for coating the substrates are not recited in the claims.
The board sustained the rejection, stating:
All the claims stand rejected as un-patentable over Ross et al. in view of the remaining references (35 U.S.C. 103). We have carefully reviewed Appellants’ arguments, but we agree with the rejection. Appellants apparently concede that the combined references would render the claimed process obvious to workers of ordinary skill in this art for coating an iron surface, but not for aluminum. They urge that the unobvious aspect of their process and composition is that their one bath can coat the three metals (Fe, Al and Zn), whereas the reference baths will coat only iron, or in some eases only iron and zinc. Assuming that this is adequately shown by the Rule 132 affidavit of record, the claims cannot be held to be patentable. The composition is obviously the same, whether iron or aluminum is to be treated. The process claims cover the reference process for treating iron. Clearly, Appellants are not entitled to a monoply that includes obvious variations of the prior art.
Opinion
Appellants argue that the teachings of the references are incompatible for various reasons, and that the claimed invention would not have been obvious even if the references were properly combinable. They point to the Saad affidavit as establishing that their composition has the unexpected ability to coat all three metal surfaces — aluminum, zinc and iron. The solicitor, on the other hand, refutes the existence of the infirmities found by appellants in the references. The solicitor further takes the position that the claimed invention is obvious for coating iron, and thus would have been obvious within the meaning of *1322section 103 of the statute, notwithstanding any alleged “unexpected results.” That was essentially the position taken by the board. The solicitor argues that the Patent Office position is fully supported by In re Mod, 408 F.2d 1055, 56 CCPA 1041 (1969) and In re de Montmol-lin, 344 F.2d 976, 52 CCPA 1287 (1965).
In order for us to ignore appellants’ evidence that their composition has un-obvious properties, it would be necessary for us to hold that whenever it is determined that the prior art provides any justification or suggestion for making a compound or composition, one of two things must happen. Either (1) the un-obvious properties of the compound or composition must somehow cease to be evidence that the compound or composition itself is unobvious, or (2) the case for obviousness of the compound or composition must become unassailable — -in effect allowing any indication of desirability of a compound or composition to create an irrebutable presumption that it is obvious. Nothing has been placed before us which would justify either holding.
We have recently pointed out that the rule derived from Mod and de Mont-mollin by the Patent Office is. not supported by those cases. In re Murch, 464 F.2d 1051, 1055-1056, 59 CCPA 1257 (1972). In the de Montmollin case, the court said:
Although appellants cite no decision in support of their statement that “There is no basis in law for ignoring any property in resolving an issue of ‘obviousness,’ ” that proposition is expressed in haec verba in In re Papesch, 315 F.2d 381, 50 CCPA 1084 [(1963)].
We do not agree with appellant that a single variance in the properties of new chemical compounds will necessarily tip the balance in favor of pat-entability where otherwise closely related chemical compounds are involved. * * * It goes without saying that all evidence of record tending to establish obviousness or unobviousness must be considered.7
This reasoning was cited with approval in Mod, 408 F.2d at 1057, 56 CCPA at 1044. As in Murch, supra, we are of the opinion the board and the solicitor were in error in effectively ignoring, and urging us to ignore, the evidence of unobviousness presented in the record.
In order to determine whether the board’s decision was correct (as opposed to the opinion supporting that decision), we must proceed to examine the basis upon which it was concluded that the claimed subject matter was not patentable under 35 U.S.C. § 103, to see whether it is supported by the facts disclosed by the art relied on. In re Heldt, 433 F.2d 808, 811, 58 CCPA 701, 706 (1970). If the evidence that the claimed subject matter is strong enough to establish a prima facie case, i. e., one which would prevail in the absence of rebuttal evidence, we must go on to examine the evidence of non-obviousness before the legal conclusion called for by section 103 can be reached.
1. The case for obviousness
We have no difficulty in concluding that the prior art relied on strongly indicates that the claimed phosphatizing composition and process would have been obvious to one of ordinary skill in the art at least as of the effective date of the Ross reference, the most recent one relied on. The claims would be met by the Ross composition made with ammonium salts instead of alkali metal salts, and the use of ammonium salts instead of alkali metal salts, for both the phosphates and the accelerators, is clearly shown by all of the secondary references. Indeed, in that part of Stapleton quoted above the patentee goes so far as to set up an artificial definition of “alkali metal” which includes the ammonium radical.
The main contention that appellants urge with regard to the Ross reference *1323is that Ross deals exclusively with ferrous surfaces, while “ [appellants’ claims require that all three surfaces must be treated.” Appellants further state:
How then can the board hold * * * that “the process claims cover the reference process for treating iron.”? Applicant’s claims do not cover a process useful only for the treatment of iron. Applicants’ claims only cover the situation where all three metals are treated simultaneously as in complicated multimetal parts or when one processor treats all three metals.
Assuming, arguendo, that this is one of those cases in which a statement of intended use constitutes a real limitation on a composition claim, see In re Szajna, 422 F.2d 443, 57 CCPA 899 (1970); Kropa v. Robie, 187 F.2d 150, 38 CCPA 858 (1951), we must still agree with the Patent Office that the appellants’ composition claims reasonably cover compositions for the treatment of surfaces consisting solely of steel, the surface material with which Ross deals.
Appellants’ ease regarding their process claims is even weaker. Appellants concede that the claims are broad enough to cover the situation where “one processor treats all three metals.” That situation is described in appellants’ specification as the “common practice [of handling] iron, zinc and aluminum articles at the same coating line location * * Thus appellants’ claims reasonably cover the treatment of steel and if appellants’ composition or process would have been obvious to one desiring to coat steel articles, the claims are at least prima facie unpatentable because they are too broad in the sense of 35 U.S.C. § 103. In re Landgraf, 436 F.2d 1046, 58 CCPA 929 (1971).
Moreover, even were we to accept appellants interpretation of their claims, that would not destroy the Patent Office’s case. As the examiner pointed out, there is evidence here that the art considered the three claimed substrates equivalently, or at least alternately, phos-phatizeable. Dodd uses a molybdate- or tungstate-accelerated phosphating composition for use on ferrous metals, aluminum, zinc, magnesium and terneplate. As appellants point out, Dodd’s improvement over previous phosphate baths with molybdate accelerators was to require the presence of a phenolic compound, but the ability of his bath to coat various metals was never tied to the presence or absence of that compound. Furthermore, the facts that Stapleton and Ross only specifically mention iron and steel and Amundsen only specifically mentions iron, steel and zinc are not equivalent to teachings that their compositions would not work on any metals other than those specifically named. In fact, all of those references broadly state that their inventions pertain to phosphate coatings for “metals” or “metallic articles.”
Appellants advance other criticisms of the secondary references, but point to nothing which would detract from their disclosures that ammonium salts may be substituted for the alkali salts in such phosphate coating baths.
In sum, we think that the evidence in this record that appellants’ composition and its use would have been obvious is very strong, much stronger than the minimum necessary to establish a prima facie case.
2. The rebuttal evidence
Appellants contend that by using ammonium salts rather than alkali metal salts, one obtains a composition which, unlike Ross’ composition, gives an acceptable coating on all three metals— steel, zinc and aluminum. Appellants argue that that result is. unexpected, is substantiated by the Saad affidavit, and is not suggested by any of the references. The examiner was of the opinion that the affidavit was insufficient to establish patentability of the instant claims, because “the affidavit only shows that at 1 minute and at 150° F. applicant obtains a difference in coating each metal.”
*1324In order for a showing of “unexpected results” to be probative evidence of non-obviousness, it falls upon the applicant to at least establish: (1) that there actually is a difference between the results obtained through the claimed invention and those of the prior art, In re Klosak, 455 F.2d 1077, 59 CCPA 862 (1972); and (2) that the difference actually obtained would not have been expected by one skilled in the art at the time of invention, Id.; In re D’Ancicco, 439 F.2d 1244, 58 CCPA 1057 (1971). We think that those requirements would have been met in this ease if appellants had established either (1) that their invention resulted in an unexpected improvement over the Ross invention; or (2) the broad proposition that substituting ammonium salts for alkali metal salts, the aspect of the invention for which the secondary references were relied on, unexpectedly gives a bath suitable for all three metals. We find appellants’ evidence to be insufficient to support either proposition.
We turn first to the comparison of the present invention, bath B in the affidavit and the prior art of record, bath C, which is allegedly the same as the bath used in Ross. Ross states the following with regard to the way in which his composition is to be applied:
The dry composition is dissolved in water to give concentrations of from about 1 to about 5 ounces per gallon of water, and preferably about 2 ounces per gallon of water, thereby producing a coating solution having a pH in the range of from about 3.8 to about 5.2. The desired phosphate coating is then produced by spraying a ferrous metal surface with this solution at a temperature in the range of from about 140°F. to about 180°F., and preferably at about 160°F. A coating is normally effected in about 1 to 2 minutes.
In the tests referred to in the Saad affidavit, the metal plates were apparently dipped in the test bath for a period of one minute at a temperature of 150°F. While the one minute contact time would appear at first blush to correspond to the lower limit expressed by Ross, it should be remembered that Ross specifies spray application of his coating. That the method employed in the affidavit adversely effect the amount of time needed to obtain a suitable coating as compared to a spray application is indicated by appellants’ own specification, wherein it is stated:
As is known in the art, the contact times for obtaining the desired coating will vary widely, depending upon the application techniques used, suitable contact times normally being within the range of several seconds to 10 or more minutes. Typically, coatings are obtained when using about 10 seconds to 2 minutes in spray applications, and in about 1 to 5 minutes when the metal surface is immersed in the coating solution.
No explanation is given why an immersion technique was chosen for the comparison in spite of the fact that Ross called for spraying. Nor is there any explanation given as to why the contact time was not lengthened to compensate for the change in coating techniques. There are other discrepancies which, by themselves, would probably not fatally taint the comparison, but nevertheless tend to weaken it. For example, no reason is given why 150° F. was chosen for the comparison. That temperature is within Ross’s range, but Ross’s preferred temperature is 160°F., and both of the examples in appellant’s specification were run at 70°C. (158°F.) While the record shows that these temperatures are not critical, a test of Ross’s composition under his preferred conditions would have carried more weight. In summary, we find appellants’ evidence insufficient to establish that the properties of appellants’ compositions are expectedly better than those of Ross’s compositions.
The situation with regard to the comparison between bath A and bath B is somewhat different. Here we have two coating baths having the same concentrations of phosphate, fluoride and *1325molybdate anions, bath A containing sodium cations and bath B containing ammonium cations; with identical metallic materials undergoing identical treatment. While bath A did produce a coating on aluminum, the coating produced by bath B is much darker. We think that this is some evidence that substitution of ammonium for the alkali metal cations would have a real effect on the composition’s ability to coat aluminum. We consider that this evidence, together with Saad’s statement that the darker coating obtained through bath B was “surprising and unexpected,” is probative on the question of obviousness, and that it gives some indication that the substitution of ammonia for alkali metal ions would not have been obvious. However, considering the qualitative nature of the results, the fact that the testing was performed at the lower limit of the contact times specified in appellants’ specification for immersion coating techniques, and the fact that the prior art contains indications that similar phos-phating compositions would have been expected to operate effectively on various metals including aluminum, we conclude that appellants have not established as a general proposition that the substitution of ammonium for alkali metal ions results in a composition which gives unexpectedly improved coatings on aluminum. The results obtained using bath C on aluminum lend but little support to appellants’ ease, for, as noted above, that composition was used in a manner other than that specified by Ross, and there was no comparison made between bath C and a bath identical to bath C except for the use of ammonium salts instead of the sodium salts. Thus appellants’ evidence, though probative, is insufficient to rebut the case for obviousness established by the Patent Office.
Finding that the case for obviousness of the claimed subject matter has been neither rebutted nor shown to be based on erroneous reasoning or misinterpretation of the references, we affirm the decision of the board.
Affirmed.
. Serial No. 574,578, filed August 24, 1966.
. U.S. Patent No. 3,060,036, issued October 23, 1962.
. U.S. Patent No. 2,665,231, issued January 5, 1954.
. U.S. Patent No. 2,502,441, issued April 4, 1950.
. “Terneplate” is defined in Webster’s Third New International Dictionary at page 2359 (1961) as “sheet iron or steel coated with an alloy of about 4 parts lead to 1 part tin.”
. U.S. Patent No. 2,839,439, issued June 17, 1958.
. 344 IT .2d at 978, 52 OOPA at 1289, emphasis supplied in part.