T 0947/96 (Ceramic substrate/MITSUBISHI) 07-12-1999

1. The appeal complies with Articles 106 to 108 and Rule 64 EPC and is therefore admissible.

2. Amendments

2.1. Claim 1 contains the features of claims 1 and 2 as originally filed with the added restriction that the purity of the aluminum in the conductive islands and the heat sink is equal to or greater than 99.98 percent. Original claims 1 and 2 merely specify that the conductive islands and the heat sink, respectively, are made of aluminum or an aluminum alloy.

Referring to Tables 1 and 4 of the application as filed, specimens 9, 25, 36, and 38 have an purity of 99.98% of the aluminum in the conductive island and the heat sink; specimens 10, 27, 32, and 37 have a purity of 99.995%; and specimens 1, 6, 7, 24, 26, 29, 33, and 39. comprise "pure aluminum".

Thus, the end-points (99.98% and 100%), as well an intermediate point (99.995%) are disclosed in the application as filed. Moreover, figure 5 of the application as filed shows continuous graphs between the aluminum content of a substrate and the number of heat cycles applied to the substrate. It is evident from these graphs that the crack resistance of the substrate improves as the aluminum content is increased continuously from about 97.0% to just below 100%.

2.2. In the decision under appeal, the examining division argued that in the application as filed, no significance of the claimed range of purity was derivable, since it was evident from Tables 1 to 5 of the application in suit that there were several aluminum alloys which fell outside the claimed range but nonetheless solved the problems of decreasing the weight of the substrate and of improving its crack resistance in repeated thermal cycling, at least with respect to samples having conductive islands and heat sink made of copper.

In the Board's view, however, in the consideration of the requirement of Article 123(2) EPC, it is irrelevant whether or not there is any significance in selecting a narrow range or improvement over the selected range in relation to the disclosed broad range; what is relevant is whether or not the narrow range is clearly derivable from the application as filed.

Moreover, it is apparent from Tables 1, 2, 4, and 5 that the specimens 1, 6, 7, 9, 10, 24 to 27, 29, 32, 33, 36 to 39 having the narrow range of aluminum according to the invention provide an improved crack resistance in repeated thermal cycling and reduced weight ratio in relation to specimens having conductive islands and heat sink both made of copper, and thus solve the problems addressed in the application as filed.

Thus the limitation of claim 1 merely represents a reduction to a range already envisaged within the application as filed (cf. also T 571/89, unpublished, reasons 2.1; T 53/82, unpublished, reasons 2; T 2/81, OJ EPO 1982, 394, reasons 3).

In the Board's view, therefore the requirements of Article 123(2) EPC are met by the application as amended.

3. Novelty

3.1. Document D4, which the Board considers as the closest prior art, discloses a "metal member" bonded to a substrate of a non-oxide ceramic such as AlN (cf. page 3, lines 10 to 16 and 23 to 25). According to the document, the "metal member" may consist of "a simple body of copper, iron, chromium, nickel, molybdenum, silver, cobalt, or aluminum; or, alloys or mixtures thereof" (cf. page 5, lines 22 to 25). From page 9, lines 23 to 26, it is evident that the "metal member" corresponds to the conductive islands of the application in suit. Thus, document D4 discloses the alternative of having conductive islands made of aluminum bonded to a surface of an AlN substrate.

3.2. The device of claim 1 differs from that of document D4 in that (i) a heat sink is provided on the opposite surface of the conductive islands; and (ii) both heat sink and the conductive islands are made of aluminum having a purity equal to or greater than 99.98 percent. Document D4, on the other hand, neither mentions a heat sink nor specify any degree of purity for the metal used for the conductive islands.

Thus, the subject matter of claim 1 is new with respect to document D4 within the meaning of Article 54(1) and (2) EPC.

4. Inventive step

4.1. In the application as filed, the technical problems to be solved are two-fold: firstly to reduce the weight of the substrate; and secondly to enhance the crack resistance of the substrate against repeated thermal stress (cf. application as filed, page 2, last line to page 3, second paragraph).

4.2. In the application in suit, the above two problems are solved by providing the AlN substrate with the two features (i) and (ii) referred to in item 3.2 above. From the supplementary experiments provided by the appellant, it is evident that the technical effect of obtaining a high resistance against cracks due to repeated thermal stress is only obtained when both the conductive islands and the heat sink are made of aluminum having a purity of 99.98% or higher: Tables 2 and 5 of the supplementary experiments show the results of treating same specimens as described in the application as filed to between 400 and 700 heat cycles. The results showed that only specimens 1, 6, 7, 9, 10, 24 to 27, 29, 32, 33, 36 to 39, i.e., those falling within the scope of claim 1, withstood 700 heat cycles without cracks. Furthermore, specimens having only one of the conductive islands or the heat sink was made of aluminum having a purity of 99.98% or higher showed cracks after 400 heat cycles, although the same samples were shown in Tables 2 and 5 of the application as filed to withstand 200 heat cycles without cracks.

4.3. Document D4 on the other hand is concerned with the problem of adhesion of metal members bonded to the surface of a ceramic substrate and does not address any of the above problems. Heat sinks are also not mentioned in document D4. Nevertheless, as the problem of weight reduction is routinely encountered in the technical field of substrates for electronic circuits, the skilled person would consider this problem although this is not explicitly emphasized in the prior art document.

4.4. The skilled person faced with the task of decreasing the weight of a device according to document D4 and at the same time ensuring an adequate heat dissipation would in the Board's view consider the substrate made of AlN together with the "metal members" made of aluminum, since it is commonly known that AlN has superior heat dissipation properties and aluminum has low density.

It could be argued that while considering further improvement in heat dissipation, the skilled person would consider using a heat sink on a surface of the substrate opposite to that provided with the conductive islands, since the use of heat sinks is generally known in the art. Nevertheless, in the consideration of weight reduction of the substrate and the heat sink, the Board in agreement with the submission of the appellant, is of the view that the level of purity of aluminum would play an insignificant role, so that there was no reason for the skilled person to select the claimed range of purity of aluminum for both the conductive islands and the heat sink.

4.5. As discussed in section 4.2 above, the supplementary experiments clearly demonstrate that a combination of the heat sink and the conductive islands both made of aluminum having a purity equal to or greater than 99.98% remarkably improves the crack resistance of the substrate in relation to a combination falling outside the claimed range of purity. The Board finds no indications in the available prior art which would prompt the skilled person to investigate whether the crack resistance of the substrate against repeated thermal stress is influenced by the degree of purity of aluminum in both the conductive islands and the heat sink.

4.6. For the foregoing reasons, in the board's judgment, the subject matter of claim 1 is not obvious having regard to the cited prior art, and accordingly involves an inventive step as required by Article 52(1) EPC.

Claims 2 to 6 are dependent on claim 1, and therefore involve an inventive step.