The obviousness issue
65 Arrow contends that the primary judged erred in concluding that the invention as claimed in claims 1 and 3 of the patent involved an inventive step as required by s 18(b)(ii) and s 7(2) of the Patents Act 1990 (Cth). The error is said to have arisen in two ways. First, Arrow contends that the primary judge's division of the Cripps question into two separate parts (what the hypothetical skilled addressee would have done and the expectation of the hypothetical skilled addressee in so doing) caused the primary judge's analysis to miscarry, in particular by the primary judge imposing too high a standard on the hypothetical skilled addressee's expectation of success. Second, Arrow contends that the primary judge considered obviousness not by reference to the claimed invention, as required, but by reference to the invention as described in the specification but not claimed.
66 I have decided that Arrow's cross-appeal should succeed.
67 Before the primary judge both parties relied on the Cripps question as the relevant test for obviousness. In the context of the present invention, this involves asking whether the hypothetical skilled addressee would have been led as a matter of course to try the steps leading to the invention in the expectation that it might well produce the invention, a useful alternative to or better thing than the prior art, or something useful for any other purpose. It is common ground that this is not the same as asking if the steps would be perceived by the hypothetical skilled addressee as "worth a try" or "obvious to try" (Aktiebolaget Hässle v Alphapharm Pty Ltd [2002] HCA 59; (2002) 212 CLR 411 at [50]-[53] and [60]-[76]; Astrazeneca AB v Apotex Pty Ltd [2015] HCA 30; (2015) 257 CLR 356 at [15]). The dispute is about how the primary judge posed the question (in two separate parts) and how he approached the concept of usefulness for the purposes of the required expectation.
68 In summary, I have decided that the primary judge conceived of the outcome Nichia sought to achieve or the invention as useful only if it involved a white LED which was "satisfactory" in terms of all qualities identified in the specification, of durability, colour tone and luminance, when the invention understood in the context of the claims, the specification and the prior art did not need to meet all of these requirements to be useful. While I do not accept one theme of Arrow's argument that, as the claims contained no integers concerning any qualitative aspect of the white LED, an expectation in respect of any white LED (whether useful or not) would suffice to establish obviousness, it is not necessary to go so far to conclude that the primary judge's reasons disclose an error in approach. It is sufficient that it is clear from the specification that the outcome Nichia sought was to obtain a white LED which met only one of the identified objects of durability, colour tone or luminance so that it must follow that a white LED of that kind is useful.
69 The Cripps question is adapted from the formulation in Olin Mathieson Chemical Corporation v Biorex Laboratories Ltd [1970] RPC 157 at 187-8. In Astrazeneca, a drugs case, French CJ re-framed the question in these terms (at [15]):
Would the notional research group at the relevant date, in all the circumstances, which include a knowledge of all the relevant prior art and of the facts of the nature and success of [the existing compound], directly be led as a matter of course to try [the claimed inventive step] in the expectation that it might well produce a useful alternative to or better drug than [the existing compound]?
70 In Olin Mathieson itself, where the prior art included the drug chlorpromazine, Graham J expressed the relevant expectation at 188 as:
… the expectation that it might well produce a useful alternative to or better drug than chlorpromazine or a body useful for any other purpose?
71 Also at 188 Graham J noted that it would not be right to include a limitation to therapeutic qualities in the question because claim 1 made no reference to such qualities. This does not mean, however, that Graham J proceeded on the basis that the Cripps question should be posed by reference to an expectation the steps in question might well produce a result which was not a useful drug or useful alternative.
72 The observation of Lord Hoffmann in Conor Medsystems Inc v Angiotech Pharmaceuticals Inc [2008] RPC 28 at [19] is not to the contrary. The claimed invention was a stent coated with taxol. As Lord Hoffman put it at [17]:
As in the case of many product claims, there was nothing inventive in discovering how to make the product. The alleged inventiveness lay in the claim that the product would have a particular property, namely, to prevent or treat restenosis …So the question of obviousness was whether it was obvious to use a taxol-coated stent for this purpose.
73 It was in this context that Lord Hoffman continued:
18 Mr Thorley, however, sought to avoid this question by watering down the claimed invention by reference to what he said were inadequacies in the specification. It did not contain information about human or animal tests which showed that it would work or provide enough information about doses and so forth to enable the skilled person to work it. It was therefore nothing more than an idea that taxol might work and any skilled person would have known that.
19 In my opinion, however, the invention is the product specified in a claim and the patentee is entitled to have the question of obviousness determined by reference to his claim and not to some vague paraphrase based upon the extent of his disclosure in the description. There is no requirement in the EPC or the statute that the specification must demonstrate by experiment that the invention will work or explain why it will work.
74 In other words, in that case obviousness was concerned with the use of a taxol coated stent for the (useful) purpose of preventing restenosis. It was not concerned with the mere idea that trying a taxol coated stent might determine whether restenosis might thereby be treated.
75 In Eli Lilly and Co Ltd v Apotex Pty Ltd [2013] FCA 214; (2013) 100 IPR 451 Middleton J at [527] expressed the relevant version of the Cripps question for the invention of the compound olanzapine in these terms:
Would the notional research group at the relevant date, based on their common general knowledge, directly be led as a matter of course to try olanzapine, in the expectation that it might well produce a useful alternative to or a better drug than known antipsychotic drugs, or a compound useful for any other purpose?
76 In so doing, Middleton J was aware of the fact that the claims involved any use of olanzapine rather than use as an antipsychotic drug (at [529]). Middleton J rejected a proposed formulation of the Cripps question in which the expectation was expressed as an expectation that the steps leading to the invention might well produce "a useful alternative to or a better drug than flumezapine [a known compound] or a body useful for any other purpose" (at [528]). Middleton J, referring to Olin Mathieson, in which the claims were also not limited to any therapeutic uses, said:
532 However, the absence of any limitation in the claims to therapeutic uses did not cause Graham J to disregard the utility of the compounds altogether. To the contrary, Graham J's reformulation of the Cripps question makes it clear that the utility of the compounds is an important part of the test. This is the correct approach to the Cripps question in the present case.
533 It is impermissible to divorce olanzapine from its context and subject it to an obviousness test that completely disregards its utility as an antipsychotic drug.
77 In Generic Health Pty Ltd v Bayer Pharma Aktiengesellschaft [2014] FCAFC 73; (2014) 222 FCR 336 the Full Court noted a similar issue at [25]:
The second issue of principle was whether the expectation of the hypothetical formulator (assuming it to be relevant) must be an expectation of the production of the invention or an expectation relating to "some other useful result" (Alphapharm Pty Ltd v H Lundbeck A/S (2008) 76 IPR 618, at [180]), and if so, what is meant by the words "some other useful result". The primary judge rejected the suggestion that "some other useful result" could be the obtaining of information to ascertain whether there was any bioavailability problem with an immediate release formulation of DRSP. The primary judge rejected the GH parties' contention, and held that "some other useful result" in the context of obviousness must be understood as the claimed invention or at least something very like the claimed invention.
78 There was no challenge to this second issue of principle in the appeal because the patent holder accepted that "we're trying to get bioavailable drospirenone" (at [26]).
79 In Generic Health at [71] the Full Court said this:
We do not think that the plurality in Alphapharm were saying that the reformulated Cripps question was the test to be applied in every case. Rather, it is a formulation of the test which will be of assistance in cases, particularly those of a similar nature to Alphapharm. The plurality did not reject as an alternative expression of the test the question whether experiments were of a routine character to be tried as a matter of course (The Wellcome Foundation Limited v VR Laboratories (Aust) Proprietary Limited (1981) 148 CLR 262, at 280-281, 286, per Aickin J). We do not think there is a divide here in terms of whether an expectation of success is relevant between a test which refers to routine steps to be tried as a matter of course and the reformulated Cripps question. It is difficult to think of a case where an expectation that an experiment might well succeed is not implicit in the characterisation of steps as routine and to be tried as a matter of course. On the other hand, we think a test formulated in terms of worthwhile to try was firmly rejected by the High Court in Alphapharm (see also Pfizer, at 476, [287], per French and Lindgren JJ [Pfizer Overseas Pharmaceuticals v Eli Lilly and Co (2005) 225 ALR 416]). The fact (if it be the fact) that the position in the United States may have shifted does not affect the binding nature of what the plurality said in Alphapharm.
80 It will be apparent from these decisions that the prior art and the claimed invention in the context of the specification as a whole are relevant to the appropriate formulation of the Cripps question in a given case. A "useful alternative" is one which is useful compared to the prior art in the field of the invention. A thing "useful for any other purpose" is also to be understood in this context.
81 The primary judge said that he considered the issue of obviousness by reference to "the outcome that is sought" (at [279]), which is an orthodox approach. As Nichia noted, a:
…useful result is "the final result which it was the object of the inventor to achieve": Alphapharm at [21], quoting with approval Technograph Printed Circuits Ltd v Mills & Rockley (Electronics) Ltd [1972] RPC 346 at 362.
Seen in its proper context, the "result" may be, for example: (1) a useful or better alternative to the prior art: cf Alphapharm at [53]; (2) solving a problem in the art: see eg AstraZeneca AB v Apotex Pty Ltd (2015) 257 CLR 356 at [115]; GlaxoSmithKline Consumer Healthcare Investments (Ireland) (No 2) Ltd v Apotex Pty Ltd [2016] FCA 608; (2016) 119 IPR 1 at [965].
82 One aspect of Arrow's submission was that as the claimed invention contains no quality integers the Cripps question is to be posed by reference to a white LED of any kind or quality, implicitly whether useful or not. I am unable to reconcile that submission with the established principles under which the claimed invention or alternative or other result is necessarily assumed to be useful. As explained below, this does not answer Arrow's case because whether the primary judge applied the orthodox approach he identified is in dispute, as are other issues of principle in his approach to obviousness.
83 It is convenient to consider the usefulness aspect of the Cripps question at a later point in the context of Arrow's challenge to the primary judge's rejection at [290] of its submission that Nichia had achieved the claimed invention by 30 May 1996 (a challenge which I have concluded should be accepted). Arrow's other contentions of error may be dealt with immediately.
84 In the present case, light emitting diodes or LEDs are identified in the specification as part of the prior art. These LEDs are identified as being used for a variety of purposes including displays, back lights, traffic signals, illuminating switches, and indicators. The specification identifies that various attempts had been made to make white light sources using LEDs but there had been a problem with generating white light of the desired tone, as well as other problems so that "a satisfactory light source capable of emitting white light … has not been obtained so far" (p 2.10-12). The invention is said to be "an alternative to known light emitting devices and displays" (p 4.9-10). At p 6 the specification says that:
The present invention aims to alleviate at least one of the problems described above and to provide a light emitting device which experiences only extremely low degrees of deterioration in emission light intensity, light emission efficiency and color shift over a long time of use with high luminance.
The present applicant completed the present invention through researches based on the assumption that a light emitting device having a light emitting component and a fluorescent material preferably meets the following requirements to achieve the above-mentioned object.
(1) The light emitting component is preferably capable of emitting light of high luminance with light emitting characteristic which is stable over a long time of use.
(2) The fluorescent material being provided in the vicinity of the high-luminance light emitting component, preferably shows excellent resistance against light and heat so that the properties thereof do not change even when used over an extended period of time while being exposed to light of high intensity emitted by the light emitting component (particularly the fluorescent material provided in the vicinity of the light emitting component is exposed to light of a radiation intensity as high as about 30 to 40 times that of sunlight according to our estimate, and is required to have more durability against light as light emitting component of higher luminance is used).
(3) With regard to the relationship with the light emitting component, the fluorescent material is preferably capable of absorbing with high efficiency the light of high monochromaticity emitted by the light emitting component and emitting light of a wavelength different from that of the light emitted by the light emitting component.
85 The examples of the invention in the specification include descriptors of luminance about 40% lower for one example (example 5) compared to other examples of the invention, "slightly yellow-greenish white light" being emitted (example 6) and "low luminance" (examples 7 and 8), as well as "good weatherability" (such as examples 5, 6, 7 and 8).
86 Arrow emphasised the following parts of the primary judge's reasons which should be set out in full:
278 Given the applicant's acceptance that YAG:Ce was part of the common general knowledge (having regard to my finding on the identity of the person skilled in the art), and given the other matters of common general knowledge referred to in the primer, I am satisfied that YAG:Ce was certainly an option and that the person skilled in the art would have been directly led as a matter of course to try it in order to produce a white LED. YAG:Ce was a known phosphor with known spectral properties.
279 But it is not enough that the person skilled in the art would consider it worthwhile to try YAG:Ce in combination with the blue GaN LED. More needs to be established in order for it to be concluded that the invention claimed in claim 1 and claim 3 is obvious in light of the common general knowledge. This is provided by the requirement in the reformulated "Cripps question" that the person skilled in the art must be led to try the posited solution "in the expectation that it might well produce" the outcome that is sought.
280 The expression of expectation is important. It signifies the likelihood, as anticipated by the person skilled in the art, that an event will occur or that a result will be achieved. On the other hand, as Jacob LJ explained in Saint-Gobain PAM SA v Fusion Provida Limited, Electrosteel Casting Limited [2005] EWCA Civ 177 at [35]:
… Mere possible inclusion of something within a research programme on the basis you will find out more and something might turn up is not enough. If it were otherwise there would be few inventions that were patentable. The only research which would be worthwhile (because of the prospect of protection) would be into areas totally devoid of prospect.
281 In my view, it is at this point that the respondent's primary case on obviousness, based on the common general knowledge, fails. I am satisfied that the person skilled in the art would have considered it worthwhile to explore the use of YAG:Ce in combination with the blue light of the GaN LED, but I am not persuaded that he or she would have done so with the requisite expectation of success.
282 Based on Blasse and Bril I and Blasse and Bril II, the person skilled in the art would have known that YAG:Ce could be used in flying-spot CRTs, especially for colour television. But, at the priority date, YAG:Ce had not been used in an application that had operating conditions comparable to those of the blue GaN LED.
283 It certainly would not have followed from the fact that YAG:Ce could be used in flying-spot CRTs that it could also be used satisfactorily with a blue GaN LED. As Professor Meijerink pointed out in concurrent evidence, the power levels in the two applications are "extremely different". Moreover, in flying-spot CRTs the laser contacts the phosphor for only a short period of time.
284 By way of example, Professor Meijerink pointed out that, even though ZnS phosphor is extremely stable in a CRT, the applicant's own experiments showed that it was insufficiently stable when used with a blue GaN LED. Relatedly, Dr Butcher gave evidence that quite often the luminescence of a given material will change depending on the level of intensity of the power source that is used to create that luminescence.
285 Further, Blasse and Bril I teaches that, in the CRT application, the use of YAG:Ce with efficient blue phosphors is required: see below at [297]. The person skilled in that art would not have known, and would not have been able to predict, whether a similar mixture would be required when YAG:Ce was used with a blue GaN LED. I do not accept that the person skilled in the art would have dismissed this possibility simply because the GaN LED itself emitted blue light.
286 Thus, there were a number of unknowns concerning how YAG:Ce with the blue GaN LED would perform, including whether it would produce white light at all. I accept Professor Meijerink's evidence that YAG:Ce's stability and performance with a blue GaN LED could not have been predicted, particularly having regard to the high operating temperatures, high photon flux and long operating times of LEDs. These matters would need to be investigated and ascertained by experimentation.
287 Evidence of the patentee's own experiments is admissible on the question of obviousness and was admitted without objection in the present case: Wellcome Foundation Limited v V.R. Laboratories (Aust.) Proprietary Limited (1981) 148 CLR 262 280-281. The applicant's own work supports my conclusion concerning the absence of the requisite expectation of success.
288 There can be no doubt that the applicant possessed expertise in the field of inorganic phosphors and other fluorescent materials, such as organic and inorganic dyes and pigments. It had carried out work using such materials with LEDs. Its own work led to the invention of the blue GaN LED. Yet even armed with this expertise and know-how, it was necessary for it to carry out a research programme in order to find a satisfactory way to make a white LED using the GaN LED.
289 I would add that its work in this regard went beyond simply using YAG:Ce as the fluorescent material. The description in the specification, and the claims of the patent, are significantly broader than simply providing a light emitting device using a blue GaN LED with YAG:Ce.
290 Thus, I do not accept the respondent's submission that the invention was made no later than 30 May 1996. Nor do I accept the respondent's submission that the applicant arrived at the invention in a short period of time that was indicative of no more than routine work and testing to arrive at the white LED.
291 The applicant selected YAG:Ce because it was worthwhile trying. But even with its own considerable expertise, it simply did not know whether it would work with the blue GaN LED to produce a satisfactory white LED. The level of its expectation of success may be measured by the fact that it concentrated its first efforts on ZnS-based phosphors. Along the way, it experienced failure. It was only after Mr Ogawa's instruction to test all the yellow light-emitting phosphors that the applicant undertook work on YAG:Ce. It was only through its research work testing YAG:Ce that it arrived at the invention. If the applicant had thought that YAG:Ce had greater prospects of success, then its selection and pursuit of ZnS-based phosphors as the preferred route was nonsensical. I do not accept that the applicant would have acted in such a way.
292 For these reasons, the respondent has not established that, at the priority date, the invention claimed in claims 1 and 3 was obvious and lacked an inventive step by reference to the common general knowledge.
87 Importantly, the primary judge found at [278] that the skilled addressee would have been directly led as a matter of course to try YAG:Ce to produce a white LED. To the extent Nichia suggested that the primary judge meant that the reference to "white LED" in [278] was to be understood as a non-useful white LED, in contrast to a useful white LED as referred to in one way or another in [283], [286], [288] and [291], I disagree. The primary judge plainly meant that the skilled addressee would have been directly led as a matter of course to try YAG:Ce to produce a white LED of the kind described in [283], [286], [288] and [291], being a "satisfactory" or useful white LED (an issue discussed further below). Further, the primary judge's description of YAG:Ce as "an option" in [278] is but one part of the sentence. It does not qualify the part which follows which is a finding, not otherwise challenged by either party, that the person skilled in the art would have been directly led as a matter of course to try YAG:Ce to produce the white LED of the invention. Subsequent references to the lesser requirement of "worthwhile to explore" (at [281]) and "worthwhile trying" at [291] also do not undermine the meaning or force of the primary judge's finding at [278] that the person skilled in the art would have been directly led as a matter of course to try YAG:CE in order to produce a (useful) white LED. It is clear that the primary judge was deliberately expressing himself in these terms in [278] as means of explaining that the first part of the Cripps question was satisfied but the second part, concerning the expectation in which the skilled addressee would have taken those steps, remained to be answered. The balance of his reasons at [279]-[292] then explain why the second part of the Cripps question (the expectation in which the steps were taken) was not satisfied.
88 The Full Court's observation in Generic Health at [71] that it "is difficult to think of a case where an expectation that an experiment might well succeed is not implicit in the characterisation of steps as routine and to be tried as a matter of course", indicates that care is needed in evaluating the character of the expectation separately from the steps the skilled addressee would have taken. There are indications that in evaluating the expectation the primary judge strayed from the test that the steps might well produce the invention or a useful result. Thus, at [280] the primary judge referred to the expectation as signifying a likelihood that an event "will" occur or a result "will" be achieved. In [285] the primary judge said that the person skilled in the art "would not have known, and would not have been able to predict, whether a similar mixture" (of YAG:Ce with efficient blue phosphors) would have been required and that the skilled person would not have "dismissed" the possibility that such a mixture would be required. In [286] the primary judge refers to "a number of unknowns" and that YAG:Ce's stability and performance with a blue GaN LED [gallium nitride LED] could not have been predicted. In [291] the primary judge said that Nichia "did not know" that YAG:Ce with the blue GaN LED would produce a "satisfactory white light".
89 To my mind, these references suggest that the primary judge strayed from the test of steps taken in an expectation that they might well produce the invention or a useful result towards a test of an expectation of knowing that steps will produce a useful result based on predictive capacity. This impression is reinforced by other parts of the primary judge's reasons.
90 One issue is the use the primary judge made of Nichia's work with ZnS-based phosphors in [291] to support his conclusion of lack of obviousness. As noted, the primary judge there said:
The applicant selected YAG:Ce because it was worthwhile trying. But even with its own considerable expertise, it simply did not know whether it would work with the blue GaN LED to produce a satisfactory white LED. The level of its expectation of success may be measured by the fact that it concentrated its first efforts on ZnS-based phosphors. Along the way, it experienced failure. It was only after Mr Ogawa's instruction to test all the yellow light-emitting phosphors that the applicant undertook work on YAG:Ce. It was only through its research work testing YAG:Ce that it arrived at the invention. If the applicant had thought that YAG:Ce had greater prospects of success, then its selection and pursuit of ZnS-based phosphors as the preferred route was nonsensical. I do not accept that the applicant would have acted in such a way.
91 The primary judge said at [259]:
…the alleged obviousness of an invention is not judged according to whether there was an equally or more obvious solution within the sight of the person skilled in the art at the priority date. Once again, the question is the obviousness of the invention in suit, not some other "invention", failed or otherwise.
92 The primary judge also said:
276 However, the question of obviousness is not directly concerned with choices of this kind. The question is whether the pathway chosen by the inventor is one which the person skilled in the art, equipped with the common general knowledge before the priority date, would have been directly led as a matter of course to try. If obvious, a solution does not become less obvious simply because the person skilled in the art might have been presented with other possibly effective solutions of the kind I have described: see, for example, the related comments made by Gageler and Keane JJ in AstraZeneca AB v Apotex Pty Ltd (2015) 323 ALR 605; [2015] HCA 30 at [115]. Still less will that be the case where these other possibly effective solutions are said to be more obvious.
277 The question thrown up in the present case is whether the person skilled in the art would have been directly led as a matter of course to try the combination of the blue GaN LED (which, on the respondent's and applicant's respective cases, is a given) with YAG:Ce to produce a white LED. Other pathways, which might be more or less "obvious", can be put to one side.
93 These are orthodox statements of principle. In Brugger v Medic-Aid Ltd [1996] RPC 635 at 661 Laddie J said:
… if a particular route is an obvious one to take or try, it is not rendered any less obvious from a technical point of view merely because there are a number, and perhaps a large number, of other obvious routes as well. If a number of obvious routes exist it is more or less inevitable that a skilled worker will try some before others. The order in which he chooses to try them may depend on factors such as the ease and speed with which they can be tried, the availability of testing equipment, the costs involved and the commercial interests of his employer. There is no rule of law or logic which says that only the option which is likely to be tried first or second is to be treated as obvious for the purpose of patent legislation.
94 Accordingly, the fact that Nichia first tried ZnS phosphors does not mean that the use of YAG:Ce was not obvious. This is particularly so in the present case in which the primary judge's findings indicate that Nichia first tried ZnS phosphors confident that they would work. In this regard, it must be understood that it was agreed that the common general knowledge included the following (at [169] read with the primer annexed as a schedule to the primary judge's reasons):
19. A phosphor is a substance that exhibits a luminescence process when suitably excited.
20. Phosphors often have 2 components. The first is the luminescent centre (sometimes called the dopant or activator) and the second is the host matrix or host crystal lattice.
21. The activator is usually a particular element, which will often be in an ion form in the host matrix.
22. The general role of a phosphor when used with a light emitting diode is to change the overall emission wavelengths emitted from the integrated device. The phosphor changes the overall perceived colour of light emitted by absorbing some (but not all) of the light emitted by the semiconductor chip. It achieves this by the activator absorbing photons emitted by the LED semiconductor to produce excited states of the activator. The activator subsequently relaxes back to its original ground state by emitting photons of a different energy and hence a different colour to that absorbed. A key requirement of the phosphor is having an activator with absorption transitions that match the LED emission, and having emission transitions at the energies corresponding to the desired new colours.
…
33… GaN LEDs can emit blue light.
…
56. YAG:Ce can absorb specific wavelengths of light that correspond to exciting cerium electrons to higher energy states. One wavelength for absorption in YAG:Ce is blue light of 450 nanometres.
…
66. Various colours of visible light can be combined to produce white light. For example:
(a) red, green and blue; and
(b) blue and yellow.
95 The primary judge also found that, by the priority date of the patent (29 July 1996):
(1) the use of phosphors in lighting was ubiquitous, including fluorescent lighting (at [154]);
(2) YAG:Ce was known and was part of the common general knowledge of the person skilled in the art, including that YAG:Ce absorbed wavelengths of light that correspond to exciting Ce electrons to higher energy states, including blue light of 450 nm and after being excited by the absorption process the Ce electrons return to a lower energy state by the complementary process of light emission, with the emitted light being of a different wavelength than the absorbed light (at [182]);
(3) "the person skilled in the art would have known that YAG:Ce was a phosphor with garnet structure that absorbs blue light and converts it with high efficiency into a yellow emission" (at [187]); and
(4) both the blue GaN LED and YAG:Ce as a known phosphor with known spectral properties were part of the common general knowledge (at [277] and [278]).
96 In respect of Nichia's development process, the primary judge found that, using a GaN LED and fluorescent materials, Nichia had developed "the world's first high-brightness blue LED using a GaN semiconductor" in 1993 (at [117]). Nichia thereafter developed a green LED (at [119]) and other LEDs using red, yellow and orange organic fluorescent materials (at [119]-[122]). Nichia is also the world's largest manufacturer of inorganic phosphors (at [123]). At the beginning of 1996 Nichia's President, Mr Ogawa, directed that Nichia develop a white LED, leading to the creation of the White LED Development Team (at [124]-[125]). Mr Shimizu was the team leader. He reported to Mr Shinomiya, to whom Mr Ogawa had given the direction for the development of a white light LED (at [125]). Mr Shinomiya and Mr Shimizu had the most experience with inorganic phosphors within Nichia (at [127]). Because synthesising a phosphor took up to a year they focused on Nichia's existing phosphors (at [129]-[130]). Mr Shinomiya knew that zinc cadmium sulphide phosphor activated by silver (ZnCdS:Ag) emits a yellow light when excited by an electron beam, Nichia having used this phosphor in monochrome cathode ray tube (CRT) display applications (at [131]). As such, the ZnCdS:Ag phosphor was included in about 20 candidates of inorganic phosphors that had a body colour (on the basis that a "a phosphor which is excited by blue light should show some body colour under natural light") (at [130]-[132]). As a result of this testing the ZnCdS:Ag phosphor was identified as the preferred candidate (at [133]). Although they knew that this phosphor was sensitive to water, they believed that as it had good stability in the CRT applications stability could also be achieved in the LED application because the phosphor could be sealed with resin (at [133]). As the primary judge then said:
134 Specific testing was done using ZnCdS:Ag phosphor. It was found that while the phosphor produced a white colour that was acceptable, it degraded within one day.
135 Mr Shinomiya said that, given its high reliability in the CRT environment, it was very surprising to him that ZnCdS:Ag phosphor would degrade so quickly in the LED environment simply because moisture was present. Mr Shimizu's observation, as reported to Mr Shinomiya, was that, under strong light and heat, the phosphor seemed to be blackened by the very tiny amount of water that either remained in the resin or entered from outside the resin.
136 Further development work was carried out using a ZnCdS:Cu,Al phosphor. The LED also emitted white light of an acceptable colour. This LED is, in fact, used as Comparative Example 1 disclosed at page 49, line 18 to page 50, line 9 of the specification:
(Comparative Example 1)
Formation of a light emitting diode and life tests thereof were conducted in the same manner as in Example 1 except for changing the phosphor from (Y0.8Gd0.2)3Al5O12:Ce to (ZnCd)S:Cu, Al. The light emitting diode which had been formed showed, immediately after energization, emission of white light but with low luminance. In a life test, the output diminished to zero in about 100 hours. Analysis of the cause of deterioration showed that the fluorescent material was blackened.
…
137 The "life test" referred to in this disclosure involves testing the product under harsher conditions than normal, for example at a temperature of 60°C and 90% humidity. The purpose is to test how the phosphor performs under conditions which accelerate its life cycle. In his affidavit, Mr Shinomiya gave the following evidence:
The problems the White LED Development Team experienced with the ZnS phosphor highlight one of the fundamental challenges involved with phosphor research. Even when a phosphor can be successfully employed in one operating environment, it cannot be assumed that it will have sufficient durability in another environment. This means that any candidate must be tested to ascertain whether or not it will actually have the required luminance and durability in a new environment. The light, heat and moisture in the LED environment can be particularly challenging in terms of durability.
97 In other words, Nichia selected the ZnCdS:Ag phosphor because it was known to have a body colour which could be combined with the known blue GaN LED to try to produce white light. It is apparent from the evidence referred to by the primary judge that Nichia undertook its work with the ZnCdS:Ag phosphor expecting it would work. It did produce white light but, to Nichia's surprise, the phosphor was not stable, despite having been stable in the known CRT applications. Accordingly, Nichia did not approach its tests with the ZnCdS:Ag phosphor expecting them to fail because of "the fundamental challenges involved with phosphor research." If this had been the case Mr Shinomiya would not have found the lack of durability of the ZnCdS:Ag phosphor "very surprising". While the white light produced was acceptable, the required durability could not be achieved (at [136]-[137]). As Mr Shinomiya said at [135] the lack of durability of the white light LED involving ZnCdS:Ag was "very surprising to him" because, it must be inferred, based on the CRT applications he assumed that stability could be achieved in the LED device.
98 Given Mr Shinomiya's evidence I consider it plain that Nichia tested ZnCdS:Ag phosphors expecting that this might well (indeed, would) work both to produce a white light LED of acceptable colour (which it did) and to be durable (which it was not). In common with YAG:Ce, the ZnCdS:Ag phosphor was itself a "known phosphor with known spectral properties" (to use the primary judge's phrase at [278]). It must follow that the primary judge's finding at [278] that "the person skilled in the art would have been directly led as a matter of course to try" YAG:Ce also applies equally to the ZnCdS:Ag phosphor. In the evidence, moreover, the person skilled in the art would have been directly led as matter of course to try the ZnCdS:Ag phosphor with a blue GaN LED to produce a white light LED expecting that it would work, as indeed Nichia did. That is to say, on the primary judge's approach, it must be the case that using the ZnCdS:Ag phosphor was obvious.
99 I do not consider the primary judge's reference to the fact that it was not known that YAG:Ce with the blue GaN LED would produce white light (at [286] in particular, but see also [394]) undermines this analysis. This is because the relevant test is not knowing that steps will or would or even may well work, but merely expecting that the steps may well work.
100 Given this, I disagree with [291] of the primary judge's reasons. If, as is the fact, Nichia was confident that the ZnCdS:Ag phosphor would work because it was known to have a body colour which the skilled addressee expected may well (or would) produce white light when combined with the blue GaN LED (as is the case here) and had been proven to be durable in well-known applications (as is also the case) and Nichia found it "very surprising" that the ZnCdS:Ag phosphor was not stable then the level of Nichia's expectation of success of YAG:Ce could not be measured from the fact that Nichia tried the ZnCdS:Ag phosphor first. The critical point, given the other findings, is also not that Nichia experienced failure but that it was confident of success with the ZnCdS:Ag phosphor. As a result, the question whether Nichia "thought that YAG:Ce had greater prospects of success" does not arise. Given the facts as found and undisputed evidence, there could not have been an occasion for Nichia to consider YAG:Ce before testing proved the lack of stability of the ZnCdS:Ag phosphor. Accordingly, Nichia's work with the ZnCdS:Ag phosphor cannot provide support to the conclusion that the selection of YAG:Ce was not obvious.
101 A further issue is that the primary judge rejected Arrow's submission at [290] that Nichia made the invention no later than 30 May 1996, after a short period of time of no more than routine work and testing.
102 The primary judge found that after the failure with the ZnCdS:Ag phosphor Mr Ogawa said that "there should be many other phosphors that can emit yellow light like the ZnS-based phosphors" and asked if the team had "tested all of them" (at [138]). The primary judge continued in these terms:
140 The day after the conversation with Mr Ogawa, Mr Shinomiya and Mr Shimizu went to an inspection room located in the applicant's Inspection Building, where the applicant's phosphor reference samples were stored. Mr Shinomiya said that he and Mr Shimizu carefully observed the body colour of several hundred sample phosphors, one by one. Mr Shinomiya gave this evidence:
The samples were stored in a variety of ways; some were stored in plastic bags, others were in transparent bottles. If a phosphor was in a black plastic bag or in a bottle with a cover to prevent exposure to light, we opened them so the phosphors were exposed to light during our observation. We were in the inspection room for more than two hours to observe all of the samples there. We did not look at any specification sheets, just the phosphor samples themselves.
141 Mr Shinomiya said that one of the phosphor samples that he and Mr Shimizu observed was YAG:Ce. Its body colour was slightly yellowish in appearance. Mr Shinomiya said that, at this time, he knew of this phosphor, but he did not know until then that its body colour was slightly yellow (or that it had a body colour other than white). YAG:Ce was available at Nichia because it had been used in electron beam excitation applications to emit green light, and for flying spot scanners. Mr Shinomiya said that, at that time, he had "never seen a real visible light emission of YAG phosphor in a flying spot scanner".
142 Mr Shinomiya said that, at this time, he also knew that YAG:Ce was used as an optical crystal for a laser oscillation. In cross-examination he accepted that, at that time, he knew that the phosphor was very stable when used in that application.
143 Mr Shinomiya said that not all of the phosphors owned by the applicant were stored in the inspection room. Mr Shinomiya gave an instruction that other phosphor samples held by the applicant were to be collected for inspection.
144 Mr Shinomiya said that, following his and Mr Shimizu's inspection of samples in the inspection room, YAG:Ce was identified as the preferred candidate for further testing because of its slightly yellow body colour. In late May 1996, the White LED Development Team produced LED samples using YAG:Ce and managed to create an LED that emitted a weak greenish white colour. This confirmed that the light emission had a broad emission spectrum when excited by a blue light. It also meant that there was a possibility that a white LED suitable for various applications could be obtained. At this juncture, it should be recalled that YAG:Ce falls within the definition of the phosphor of claim 3 of the patent.
145 Mr Shinomiya said that while YAG:Ce was stable in a laser application, it was not known whether the phosphor powder, having a diameter of only several micrometres, would have sufficient durability when moved to the LED environment. He said that this was particularly so in light of the team's experience with ZnCdS. Experiments were carried out on a test sample. The experiments showed that the phosphor had good durability in the LED environment.
146 I note that various test reports are in evidence, including a test report dated 30 May (known to be in 1996) and 6 June 1996 that concern the testing and comparison of YAG:Ce and ZnCdS phosphors.
147 Once YAG:Ce had been confirmed as the lead candidate, the remaining problem was to modify its greenish white colour to achieve a more desirable tone of white. Mr Shinomiya said that this could be achieved by reducing the colour temperature of the phosphor. Mr Shimizu requested that experiments be performed by the applicant's Phosphors Group. This stage of the work commenced in early June 1996. It was carried out by Mr Noguchi. Mr Shimizu's idea was to create a more desirable white LED by adding a red light component. He proposed introducing a co-activator which emitted red light, such as Ce with Sm3+, Ce with Eu3+ or Ce with Mn. The work with the coactivators was not successful.
148 Mr Shinomiya said that Mr Noguchi took it upon himself to prepare YAG:Ce with some of the Y substituted with Gd in order to make the wavelength of the emitted light longer. Mr Noguchi's plan was to change the matrix of the phosphor and adjust the content of Ce as an activator. The test with the modified YAG:Ce showed very good results. It produced desirable white light and had the flexibility of shifting the colour temperature in a broad range.
149 In June 1996, a decision was made by the applicant to file a patent application in respect of the work undertaken by the White LED Development Team. The patent in this proceeding derives from one of a number of applications that were made.
103 This summary (based on Mr Shinomiya's affidavit) does not deal with the fact that the Nichia experimental note of the 30 May 1996 experiment, to which the primary judge refers at [146], shows two different YAG:Ce phosphors, one at a ratio of 30:100 and another at a ratio of 50:100. The note recorded the former as having produced "whitish" (I interpolate, light) of "good reputation" (as I understand it, a description of the appearance of the colour) and the latter as "a little bit greenish" (light). This raises the question of when the invention was created. If the primary judge is correct in [290] then the inventive step was not in using YAG:Ce with the blue GaN LED to produce a "whitish" light, the inventive step must have come later than 30 May 1996, presumably in establishing the durability of the YAG:Ce phosphor by testing and improving its colour by modifying the YAG compound as described by the primary judge in [147]-[148]. Yet this is not the invention claimed. If the inventive step was in using YAG:Ce with the blue GaN LED to produce a "whitish" light then, contrary to the primary judge's conclusion at [290], the invention was complete as at 30 May 1996. This conundrum is not resolved by the primary judge having noted in [144] that Mr Shinomiya said the 30 May 1996 experiment results meant that "there was a possibility that a white LED suitable for various applications could be obtained" because the work done after 30 May 1996 is after the creation of the invention.
104 In its oral submissions on the cross-appeal Nichia accepted, as I think it had to, that by 30 May 1996 Nichia had made something that fell within claim 3, that is Nichia had the invention by that date. This, it accepted, accorded with the primary judge's observation at the end of [144] that it "should be recalled that YAG:Ce falls within the definition of the phosphor of claim 3 of the patent." In my view, it necessarily follows that [144] and at least the first sentence in [290] of the primary judge's reasons cannot stand together. Nichia sought to answer this by submitting that, as at 30 May 1996, Nichia did not know that the white light produced by YAG:Ce would be "satisfactory", in the sense the primary judge uses that term (including luminance, colour tone and durability). But this does not assist because the subsequent steps taken to prove durability and improve colour tone are not the steps involved in the Cripps question.
105 It will be recalled that, as the primary judge put it at [44], referring to p 6 of the specification:
The specification says that the invention aims to provide an alternative to known light emitting devices and displays, which alleviates at least one of the described problems to provide a light emitting device which experiences only extremely low degrees of deterioration in emission light intensity, light emission efficiency and colour shift over a long time of use, with high luminance.
106 As noted, the examples of the invention in the specification include a range of luminance and colour descriptions such as "incandescent lamp colour" with a luminance about 40% less than example 1 for example 5, "yellow-greenish white light" for example 6 and "low luminance" for examples 7 and 8. Further, only some of the examples are described in terms of durability, by references such as "good weatherability". It follows that the reference in the specification to the invention alleviating at least one only of the problems known in the prior art must be given effect. A white LED satisfying one only of these requirements, in the context of this invention, is necessarily useful. As a result, it would be wrong to conceive of the subject-matter of the requisite expectation in the Cripps question as a white LED satisfying all of the objects of durability, colour tone and high luminance. This explains, in part at least, why Nichia's concession that it had the invention by 30 May 1996 was unavoidable. At that time, Nichia had a white LED with "good reputation".
107 Once this is accepted the problem with the approach in [278]-[292] becomes apparent. Arrow submitted that, in effect, the primary judge "required the expectation to be whether the result would be commercially viable". This submission has force. It is apparent that in this part of his reasons the primary judge's approach involved a "satisfactory" white LED and that, by this, the primary judge did not mean merely useful in the sense involved in the Cripps question formulated in the context of the prior art, the specification and the claimed invention. The various references in this part of the reasons indicate that the primary judge conceived of the invention as one in which all of the objects, durability, good colour tone and high luminance were achieved. But this is not the invention described in the specification, nor the invention claimed. Nor, given this context, can the concept of usefulness in the Cripps question involve a device which meets all of these requirements. A white LED meeting any one of these requirements is the invention and is useful on the terms of the specification.
108 Thus, the concern is durability in [283], durability and luminance in [284], efficiency in [285], "stability and performance" in [286], and whether the white LED is "satisfactory" in [288] and [291]. Read fairly, it is difficult to escape the conclusion that the primary judge considered that unless Nichia expected that the steps it was directly led to take as a matter of course based on the common general knowledge would (as opposed to might well) produce a white LED which was durable, had good colour tone and high luminance, then the invention was not obvious. As explained, I consider a different approach necessary because, having regard to the claims and the specification, it is apparent that a white LED meeting any one of these requirements was a useful invention.
109 The fact that a finding about obviousness is a result of an evaluative process, which calls for restraint in appellate review of the kind described in Idenix Pharmaceuticals LLC v Gilead Sciences Pty Ltd [2017] FCAFC 196; (2017) 134 IPR 1 at [146], is immaterial in the present case given the conclusions above. The cross-appeal succeeds on issues of principle, as most recently discussed in Aldi Foods Pty Ltd v Moroccanoil Israel Ltd [2018] FCAFC 93; 133 IPR 375 at [2]-[10]. The consequence is that the appellate court must decide the question of obviousness for itself. In so doing, however, it is important to note that Nichia does not challenge the primary judge's finding at [278] that "the person skilled in the art would have been directly led as a matter of course to try it [YAG:Ce with the blue GaN Led] in order to produce a white LED". Nichia maintains that the primary judge's conclusion, that the skilled addressee would not have been so led with an expectation that this might well produce a useful white LED, was sound in any event.
110 There are four reasons why I have reached a contrary conclusion.
111 The first reason is that, as the Full Court said in Generic Health at [71], it "is difficult to think of a case where an expectation that an experiment might well succeed is not implicit in the characterisation of steps as routine and to be tried as a matter of course." The Full Court was not saying that the expectation of the skilled addressee is immaterial. Nor was it making a statement of principle as each case must depend on its own facts, but as Arrow put it:
It is very difficult to see why a person skilled in the art would have been directly led as a matter of course to combine YAG:Ce with a blue GaN LED to produce a white LED unless there was an expectation that it "might well produce" a useful result. Indeed, a person skilled in the art would only have been directly led as a matter of course to combine YAG:Ce with a blue GaN LED to produce white LED because they had an expectation that it might well produce a useful result.
112 The second reason is that if the approach in Wellcome is adopted, which asks if the steps taken were of a routine character to be tried as a matter of course, I consider the answer based on the primary judge's unchallenged finding at [278] is that the invention was obvious. As discussed below, the steps Nichia took up to 30 May 1996 were routine.
113 The third reason is that I consider Nichia's own work supports the proposition that the invention was obvious. After the ZnCdS:Ag phosphor surprisingly failed in terms of durability the only thing that happened before 30 May 1996, when the invention was created, was Mr Ogawa directed the team to test all phosphors that emitted yellow light like the ZnS-based phosphors. While Mr Shinomiya did not know about the spectral properties of YAG:Ce (but did know that it was very stable in at least one other application, see at [142]), the primary judge found that the skilled addressee would have known that YAG:Ce was a garnet structure that absorbs blue light and converts it with high efficiency into a yellow emission. Indeed, as the primary judge said at [187] Nichia accepted that this was part of the common general knowledge. Mr Shinomiya and Shimizu then went to Nichia's phosphor samples and examined them under light to find those with a yellow body colour one of which was YAG:Ce. Because of the yellow body colour YAG:Ce was selected as the preferred candidate for testing. Exciting YAG:Ce in different ratios produced "whitish light" of "good reputation" on 30 May 2016, which Nichia properly accepted to be the claimed invention. Nothing in the period from Mr Ogawa's instruction to 30 May 1996 suggests anything other than routine trial in the expectation that a yellow bodied phosphor would produce satisfactory white light (satisfactory in the sense of acceptable colour tone and/or luminance), just as the ZnCdS:Ag phosphor did.
114 The fourth reason is that Professor Meijerink's evidence, on which the primary judge relied, is also concerned with the production of a white LED which meets all requirements necessary for commercial viability. For example, at [262] the primary judge records Nichia's submission that:
…even if the person skilled in the art would have been directly led as a matter of course to use YAG:Ce, it does not follow that he or she would have done so with an expectation of success. The applicant referred to Professor Meijerink's evidence that it was impossible to know whether luminescent material could be used for spectral conversion in any given application under any given conditions, unless extensive research and testing of the material had confirmed its suitability. The applicant also referred to Professor Meijerink's evidence that, given the challenging and unique operating conditions in a blue LED, it would have been extremely difficult before the priority date to predict the stability and performance of any given luminescent material in the blue LED.
115 To my mind, this exposes a conception of the claimed invention and a useful white LED as a device meeting all requirements for commercial viability. To adopt Arrow's submissions, the effect of assessing obviousness by reference to an expectation that steps may well produce a white LED meeting (unstated) stability and performance requirements was to narrow the boundaries of the claimed monopoly. It also imposed a construct of usefulness for the purpose of the Cripps question not supported by either the claims or the specification.
116 I do not accept Nichia's submissions to the contrary.
117 The primary judge did say this at [64]:
Further, the specification makes clear that an object of the invention is to provide a light emitting device with high luminescence, and which experiences only extremely low degrees of deterioration in emission light intensity, light emission efficiency and colour shift over a long period of use.
118 However, read as a whole the specification discloses that the invention may meet only one of those objects and, as Arrow emphasised, the claims do not require anything other than a white LED made as described (which I accept must be useful). And this the primary judge also recognised at [62] when he noted that:
…claim 3 does not, in terms, require the creation of white light, nor white light of any particular nature, quality or colour rendering. Nor does it require that the light emitting device be suitable for any particular application; there are no requirements of stability, durability, efficiency or performance expressed as essential features of the invention.
119 The concept of a useful result, for the purposes of the Cripps question, cannot exceed that which the claims and the specification as a whole in the context of the prior art identifies as useful. Thus, the primary judge's observations at [66] about the invention being "directed to" a white LED that "should attain" certain desirable attributes (which the primary judge recognised were not integers of any claim) did not answer the requirements of the Cripps question (and, in that part of his Honour's reasons, was not intended to). Nor, for the purposes of the Cripps question, was the primary judge bound to proceed on the basis he did because Arrow did not challenge the utility of the invention. It is true that Arrow must be taken to have accepted that the white LED of the claims was useful, but given the prior art, the claims and the specification in the present case this does not mean that the expectation must relate to every desirable quality of the invention mentioned in the specification.
120 It follows that I do not accept that the fact that the Cripps question is directed to a useful result leads to the conclusion that the invention was not obvious unless the person skilled in the art would have taken the routine steps as a matter of course which Nichia took with an expectation that those steps may well produce a white LED which is durable, with good colour tone and high luminance. This involves too high a standard given that which is claimed and the terms of the specification.
121 Nichia's references to those parts of the primary judge's reasoning about other options which also would have been worthwhile to try do not assist Nichia. For one thing, it is apparent that the primary judge knew that "worthwhile to try" was not the relevant test in Australia. For another, it is apparent he knew that, as a matter of principle, other options do not necessarily answer the question of obviousness.
122 It is also not the case that Nichia necessarily possessed all of the common general knowledge that would be attributed the skilled addressee. For example, Mr Shinomiya did not know the spectral properties of YAG:Ce but the primary judge found (and Nichia accepted) that this was part of the common general knowledge (at [187]). Thus, while Nichia possessed expertise and know-how in the field of inorganic phosphors and other fluorescent materials, such as organic and inorganic dyes and pigments (at [119]-[121], [123], and [127]) the fact that Nichia first used the ZnS-based phosphors, as explained above, does not support the conclusion that the invention was not obvious. Similarly, the finding at [288] that "it was necessary for [Nichia] to carry out a research programme in order to find a satisfactory way to make a white LED using the GaN LED" appears otherwise to be based on the premise that Nichia did not have the invention by 30 May 1996, when it did have the invention by that date.
123 It is true that the Cripps question is not a statutory test. It may be accepted, as was noted in Alphapharm at [37], quoting Johns-Manville Corporation's Patent [1967] RPC 479 at 493-4, that there is "no verbal formula which is appropriate to all classes of claims." However, as Nichia also noted, the case below and the appeal were fought on the basis of the Cripps question.
124 It may be accepted that the primary judge rejected the evidence of Dr Bretschneider as "irremediably tainted by hindsight" (at [270]). As a result, no weight should be given to his evidence in the appeal. Arrow, however, did not need to challenge Professor Meijerink's evidence summarised at [286] because inherent in that evidence about predictability is a conception of usefulness which is inconsistent with the Cripps question as it should be applied in the present case. Accordingly, the evidence of neither expert provides any assistance. Otherwise, my conclusions depend on the primary judge's unchallenged findings of fact, at least to the extent that those findings do not involve a conception of usefulness that extends beyond that apt in the present case. This accords with the observation of the Full Court in Whitaker v Child Support Registrar [2010] FCAFC 112 at [2] that it "is not open to this court to determine issues of fact, as if the findings of the learned trial judge had not been made."