6.3.3 Consideration
137 For the reasons set out below, I reject the various constructions propounded by Hanwha. As the parties' submissions were primarily directed to claim 1, and neither submitted that this aspect of claim 9 should be understood differently, I address the construction of both claims 1 and 9, commencing with claim 1.
138 Claim 1 is for a method. After the step of providing a silicon substrate, the next step is depositing a first dielectric layer of aluminium oxide on the substrate.
139 The skilled reader would understand the "silicon substrate" to be the silicon wafer the surface of which requires passivation. It was common general knowledge at the priority date that silicon was a semiconductor, the function of which in a solar cell is to convert incident light into energy by exciting valence electrons into the conduction band. Silicon is doped in order to promote directional flow. Typically, the silicon substrate was a wafer which had been subject to several processing steps, including the diffusion of dopants into it to form a pn junction before depositing dielectric layers onto it. Armed with this understanding, the person skilled in the art would most naturally consider the silicon substrate of integer 1(b) to be the silicon substrate itself together with dopant and incidental impurities.
140 Further, the words "depositing a first dielectric layer on a surface of the silicon substrate" in claim 1 are positional. They identify a location where the layer is to be placed. The Macquarie Dictionary defines "surface" as "outer face" (3rd ed, Macquarie Dictionary Publishers, 1997). "Substratum" is defined to mean "that which is spread or laid under something else". A "substrate" is something "which underlies or serves as a basis or foundation".
141 The natural and ordinary meaning of the contested words in integer 1(c) and integer 9(c) is that the first dielectric layer is to be placed on an outer face of the silicon substrate being the semiconductor material used in the solar cell, including dopant and may include trace or de minimis amounts of contaminants or impurities. The words "on a surface of" emphasise that it is not simply to be placed "on" the silicon substrate - perhaps like a shoe may be placed on a foot, despite the presence of a sock - but on the surface of the silicon substrate, like a sandal is placed on the foot, absent any intervening layer. It is true that the integer could perhaps be understood to require that the dielectric layer is to be located on any surface of the substrate, front back or sides, regardless of whatever else was on it, but if that was the meaning, the integer could have simply required that the first dielectric layer be deposited "on the silicon substrate". The preferable construction gives work to the words "on a surface of".
142 Furthermore, integers 1(d) and 9(e) require that the second dielectric layer be deposited "on a surface of the first dielectric layer" which tends to reinforce the positional significance of the steps set out in the method.
143 When one considers the disclosure of the specification this construction of integers 1(c) and 9(c) is reinforced.
144 First, the "silicon substrate" is described in terms as "a thin monocrystalline or multicrystalline silicon wafer or else a silicon thin wafer" (page 5 lines 21-22). No reference is made to any surface other than the silicon of the wafer onto which the first layer is to be placed. The specification provides no basis upon which the skilled reader might infer that the first dielectric layer is to be placed on a silicon substrate which has upon it or includes thermally grown or "native" silicon oxides, as Hanwha contends.
145 Secondly, when the specification refers to the surface of the silicon substrate, that reference is to the outer face of the silicon crystalline structure only. In one embodiment the specification instructs (page 6 lines 7-9):
Before the depositing of the first dielectric layer, the surface of the silicon substrate can be thoroughly cleaned, so that no contamination remain thereon that might disturb the subsequently deposited dielectric layer. In particular, the surface of the silicon substrate can be slightly etched away, for example in a solution which on the one hand contains an oxidising agent and which on the other hand contains hydrofluoric acid (HF) which etches away the oxidised silicon oxide. A suitable method known in the production of solar cells is for example what is known as RCA cleaning.
(Emphasis added)
The importance of thorough cleaning of the substrate was reinforced later at page 11 line 22-23. Although this passage quoted refers to the fact that the surface of the silicon substrate "can be" thoroughly cleaned, the expert evidence reveals that those in the field considered this to be a requirement.
146 It is apparent from this passage that cleaning of the "surface" includes the slight etching away of the silicon substrate. Two cleaning steps are mentioned; using an oxidising agent and hydrofluoric acid on the one hand and an RCA cleaning process on the other, both of which slightly etch the silicon surface to remove contaminants from it. The experts agreed that, in the discourse of the specification, oxides as well as metallic materials were being referred to and distinguished in this passage from the substrate or crystalline "surface" of the silicon wafer. The evidence demonstrated that cleaning was a standard process in research and in industry before the priority date and that such cleaning will remove, inter alia, any native oxide and regrowth of any native oxide would not be inevitable, as I discuss below. This was because it was known that the presence of native silicon oxide (and silicon oxide that resulted from the etching process) and other contaminants was generally undesirable because it could adversely impact the deposition of the dielectric layer.
147 Accordingly, in this aspect of the specification the reference to the "surface of the silicon substrate" was exclusively silicon, absent de minimis contaminants including silicon oxide. It tends against the construction proposed by Hanwha to the effect that silicon oxide, however formed, is to be understood to be part of the silicon substrate.
148 Thirdly, other parts of the specification also suggest that the dielectric layer is placed on the crystalline substrate, meaning in intimate physical contact with that substrate without any intervening material or layer.
149 Immediately after the passage quoted above, the specification describes one embodiment where the silicon substrate is firstly flushed with an aluminium-containing compound "so that an aluminium-containing layer is deposited on the surface" which can "cling to the silicon surface at the points at which it enters into contact with the silicon surface". The teaching is that a "chemical reaction with the silicon surface can occur; this is also referred to as chemisorption" and that "[i]n the best of cases this can lead to the formation of a monomolecular layer made up of molecules of the aluminium-containing compound". Later, an "essential advantage" of the ALD is said to be the fact that the entire substrate surface is coated uniformly which is said to be beneficial in particular in surface-textured solar cells. The "surface" to be textured here is the crystalline silicon.
150 Furthermore, earlier, in the passage on page 5 lines 10-17 (set out in section 5.1 above) the specification provides that the "key" to understanding the outstanding passivating effect and tempering layer of the invention "may be identified in the combination of the Si/Al2O3 interface, which is ideally atomically flat and is produced as a matter of course during the ALD process" (emphasis added). The word "interface" directs attention to a location where two "faces" of different materials - here silicon substrate and aluminium oxide layer - come into contact. There is no mention of a silicon/silicon oxide/aluminium oxide interface in the patent. Each of Professor Weber and Drs Glew and Ruby understood this to require an "abrupt" interface, thereby indicating no opportunity for an interface between aluminium oxide and any intervening material such as a metallic contaminant or oxide on the surface of the substrate. Whilst Dr Rentsch gave a qualified agreement that the passage teaches that the interface was only "ideally abrupt", in my view the teaching of the passage is plain, namely that one key to understanding the beneficial passivation effect of the two layered stack of the invention lies in the close interaction between the silicon substrate and the aluminium oxide first dielectric layer. I reject Dr Rentsch's qualification.
151 In the Construction JER, Professor Weber and Drs Glew and Ruby expressed the view that it was generally well known at the priority date that in heterojunction cells an abrupt interface was critical for good solar cell performance. I accept that evidence. Whilst a heterojunction cell is quite different to the invention described in the patent, the point is that the person skilled in the art would understand this passage, and the subsequent references to cleaning and the process of chemisorption, to reflect a teaching in the specification that in functional terms the passivation benefits of the invention are to be achieved by locating the first dielectric layer on the surface of the crystalline substrate. It was known that the surface of the silicon substrate could oxidise with native oxide and that it was difficult, but by no means impossible, for the surface in a processed solar cell to be oxide-free.
152 As Dr Ruby explained, "chemisorption" is a term referring to the formation of covalent bonds between molecules, in this case the formation of covalent bonds between the aluminium atoms in the precursor and the silicon atoms at the surface of the silicon substrate.
153 Each of the other construction experts agreed with this interpretation, with the exception of Professor Cuevas-Fernandez, who did not consider it sensible to expect that aluminium would form covalent bonds with silicon. Having regard to the views of the other experts, I consider their understanding to be more likely to reflect the understanding of the person skilled in the art.
154 These passages, although directed to preferred embodiments, tend to confirm that in the discourse of the specification and the claims, references to the "surface" of the substrate are references to the silicon substrate itself and indicate that silicon oxide is not part of that surface.
155 Fourthly, the disclosure of the patent is that silicon oxide is to be identified separately as material used in the second dielectric layer. This corresponds with the common general knowledge of those in the art at the priority date that silicon is a dielectric and not a semiconductor. It supports the submission advanced by REC Solar that silicon oxide does not form part of the surface of the silicon substrate.
156 That view is further supported by the evidence of the experts.
157 Dr Rentsch agreed in the context of a specific example posed to him in cross-examination that an interfacial oxide could be readily distinguished from the silicon substrate. When viewing a substrate with almost 20% oxygen, he considered "that this is not any more … the silicon substrate but … whatever kind of amorphous silicon oxide … definitely different to a silicon substrate".
158 Similarly, Professor Cuevas-Fernandez accepted that a silicon oxide layer is "certainly" distinguishable from the silicon substrate.
159 Dr Glew expressed his view to similar effect as follows:
I know that the silicon semiconductor of the crystalline substrate is no longer a crystalline substrate, it's now an amorphous material incapable of performing as a semiconductor, when the oxygen starts coming up, and I - and I now have this kind of messy dielectric varying composition and I no longer have a substrate. So I don't, you know, I think you can't - you can't neglect trying to understand the material, the layer that was there previously, and when it no longer acts like that layer.
160 Conversely, there was no general knowledge that an interfacial layer of silicon oxide may form during the ALD process of applying an aluminium oxide layer or that it may form during deposition by any other means, such as PECVD. The experts agreed that the first reported presence of such a layer was in the Hoex 2006 publication, which did not form part of the common general knowledge at the priority date.
161 In my view the specification did not inform or alert such a person of such an artefact.
162 In this regard, it is necessary to divert to address the lengthy evidence and arguments which were advanced concerning the disclosure of the Ritala publication.
163 Hanwha contends that, whilst it was not common general knowledge at the priority date that an interfacial oxide layer formed during the processing of the solar cell, the reference to Ritala in the patent makes clear that the Patent is not teaching that any interfacial oxide layer is to be avoided. Put affirmatively, Hanwha submits that the patent "permits", by its reference to Ritala, the existence of interfacial oxide layers upon thermal ALD deposition. As Hanwha put it in closing submissions, the reference in the specification to Ritala "underscores that the Patent is not concerned with the avoidance of an interfacial oxide, or indeed any silicon oxide forming at the surface of the silicon substrate", the consequence being that the invention as described is "indifferent" to whether or not silicon oxide is present prior to the deposition of the first dielectric layer of the invention. Hanwha submits that claims 1 and 9 should be construed accordingly.
164 In my view this convoluted argument must be rejected.
165 The Ritala disclosure is not expressed to be incorporated by reference into the specification. Its relevance should be understood by reference to the context in which it appears; Idenix Pharmaceuticals LLC v Gilead Sciences Pty Ltd [2017] FCAFC 196; 134 IPR 1 at [165] (Nicholas, Beach and Burley JJ); Merck Sharp & Dohme Corporation v Wyeth LLC (No 3) [2020] FCA 1477; 155 IPR 1 at [891] (Burley J).
166 Reference in the patent to Ritala appears after the passages to which I have referred above, where the key to understanding the disclosure of the invention is set out and the general description of embodiments (including references to cleaning and chemisorption) are made. Furthermore, immediately after the heading "Detailed Description of Embodiments", but before the reference to Ritala, the specification reiterates at page 11 lines 22-24 that the silicon water is "cleaned thoroughly" before being introduced into an evacuated coating chamber, and also repeats that "[c]hemisorption causes the molecule to be deposited on the silicon surface until the surface is saturated", thereby emphasising the points noted earlier in the specification.
167 The Detailed Description then provides additional information about the first embodiment described, which is plasma-assisted ALD. The specification provides (page 12 lines 17-25):
The variant of ALD described herein is referred to as "plasma-assisted ALD" and is well known from the literature; see for example C. W. Jeong et al., Plasma-assistant atomic layer growth of high-quality aluminium oxide thin films, Jpn. J Appl. Phys. 40, 285-289 (2001). Tests have shown that particularly good surface passivation can be achieved in that the plasma does not have direct contact to the substrates, as, in the event of such contact, ion bombardment can damage the substrate surfaces, but rather burns in a separate chamber from which the radicals are subsequently guided to the substrate surface. This variant of the method is referred to as "remote plasma-assisted ALD" and is described in US 7,410,671, for example.
168 Thus far, the skilled reader has no reason to doubt the matters to which I have referred regarding the surface of the silicon wafer. As I have noted, it was not common general knowledge at priority date that an interfacial oxide would or may be formed at the surface of a silicon substrate during the manufacture of silicon solar cells: the process of ALD in the context of solar cells did not form part of the common general knowledge and it was only after the priority date that an interfacial oxide became known to form in the process of ALD. Nor, as I have noted, could Hoex 2006 be relied upon for that purpose, because that publication was also not part of the common general knowledge.
169 It is only after this passage that Ritala is mentioned at pages 12-13 in the context of an alternative embodiment:
Alternatively, the Al2O3 thin layer 3 can also be deposited by means of thermal ALD, as described in the literature in M. Ritala et al., Atomic layer deposition of oxide thin films with metal alkoxides as oxygen sources, Science 288, 319-321 (2000), for example.
170 It is difficult to see how the context of this disclosure could demonstrate to the person skilled in the art reading the specification that this cross reference "underscores" that the Patent is not concerned with the avoidance of an interfacial oxide, or indeed any silicon oxide forming at the surface of the silicon substrate.
171 In my view the skilled reader would understand that, if they wished to refer to Ritala, they would gain an understanding of a particular method of performing ALD. The cross-reference can not be expected to provide an insight into the nature of the invention disclosed in the specification as claimed, a point that is perhaps emphasised by the fact that none of the expert construction witnesses gave evidence in chief that, in seeking to understand the disclosure of the patent, they would have gone to Ritala. None described reading Ritala when explaining the disclosure of the patent. Professor Cuevas-Fernandez in his oral evidence considered that reference to it was "not really necessary" to understand the patent or its applicability. Dr Rentsch only went to the Ritala publication in his sixth affidavit, when seeking to rebut a proposition put by Dr Ruby.
172 In context, there is no basis upon which one might assume that the skilled reader would go to the Ritala publication at all when seeking to understand the invention. It provides an "optional extra" for the skilled reader, in the form of a second means by which ALD might be performed. In my view this is a shaky foundation for an argument as to how the invention more broadly disclosed in the patent is to be understood.
173 Accordingly, in my view Ritala cannot be considered helpful in construing the meaning of "a first dielectric layer (3) on a surface of the silicon substrate" whether it is in relation to integer 1(c) or integer 9(c).
174 Finally, I note that the disclosure of the patent repeatedly emphasises that the invention disclosed is for the suppression of surface recombination losses (page 1 line 17 and elsewhere). The purpose of the invention is plainly to aid in the passivation of that surface. The construction that I prefer conforms with the function and purpose of the invention as described in the specification.
175 It is necessary to address a further argument advanced by Hanwha.
176 It contends that reading "depositing … on a surface" as requiring the first dielectric layer to be placed on the crystalline silicon of the substrate places a gloss on the language of the claim by inferring that no intermediate substance or step may be included (such as a thermally grown or natively appearing oxidation). Hanwha submits that this is an impractical approach that is not taught by the specification. In this regard, it submits that as native oxide is difficult to prevent from forming on any crystalline silicon exposed to oxygen and as the prevention of forming of it or its removal is neither taught in the specification nor required in order to achieve a passivation effect, favours its approach to construction ought to be favoured.
177 However, whilst it was known before November 2007 that silicon readily oxidises when exposed to air, such that "native" oxide regions or "islands" may form, it was also known that, in laboratory and also in industrial settings, a silicon wafer was typically cleaned with hydrofluoric acid before the deposition of dielectric layers. In that context, in the Construction JER, the experts agreed that it was not inevitable that the surface of the silicon substrate would be oxidised during the process of manufacturing silicon solar cells. The practice of cleaning is taught in the specification, as noted above, in the passage at page 6 lines 7-13 mentioned above. It was also, I find, part of the common general knowledge to clean the silicon wafer at that stage. Once cleaned, a native oxide would only form on a silicon substrate at a very low growth rate. Whether and how much native oxide grows is, the experts agreed, condition-dependent. Professor Weber gave evidence, in the context of heterojunction cells which were known as at the priority date, that it was possible to have an oxide-free interface, with the consequence that, when reading the patent, "one would not automatically assume that one might always have an oxide at the interface". I accept that evidence. I do not accept that a person skilled in the art reading the patent would assume that native silicon oxide growth would be an inevitable or even highly likely aspect of the making (whether in the laboratory or as a part of an industrial process) of solar cells using a silicon substrate.
178 There is no suggestion in the disclosure that there exists a pre-formed layer of silicon oxide prior to the deposition of the first dielectric layer.
179 What is more, there is force in the submission advanced by REC Solar that the various alternative arguments advanced by Hanwha run into a difficulty as a matter of logic. If the first dielectric layer may be applied not to the crystalline silicon substrate, but to a silicon substrate that has a layer of silicon oxide on it, one might ask rhetorically: how thick must that silicon oxide layer be before the first dielectric layer is no longer "on the surface"? The answer to that question given by Professor Cuevas-Fernandez and Dr Rentsch was that a 20 nm oxide would not be the "surface" of the silicon substrate but rather would amount to a dielectric layer. However, nowhere in the claims or the specification of the patent are parameters given whereby that thickness may be determined. This suggests that if the construction advanced by Hanwha were accepted, it would leave the scope of the claim unacceptably vague.
180 Finally, I do not consider that Hanwha's arguments based on the reasoning in Fresenius have merit. The argument advanced is to the effect that the inclusion of additional integers to a claimed combination does not necessarily avoid infringement if those additional integers are properly characterised as inessential or do not make a new working of the combination and all of the essential integers of the claimed combination are present. However, in considering infringement, one must at all times bear in mind that there is no infringement if the patentee has, by the form of the claim, left open what the alleged infringer has done; Fresenius at [49]. The fundamental rule is, as set out in Rodi & Wienenberger AG v Henry Showell Ltd [1969] RPC 367, that there will be no infringement unless the alleged infringer has taken all of the essential features or integers of the patentee's claim. As the High Court said in Radiation Limited v Galliers and Klaerr Pty Ltd [1938] HCA 17; 60 CLR 36 at 51 and as the Full Court emphasised in Fresenius at [50] in considering infringement, it is 'the substantial idea disclosed by the specification and made the subject of a definite claim' (emphasis added) that must be considered.
181 In the present case, I have found that one requirement of the invention disclosed and claimed in claims 1 and 9 is that the first dielectric layer be deposited on the surface of the silicon substrate. That requirement will not be met if a dielectric layer is deposited not on the silicon substrate, but on a layer of silicon oxide (itself a dielectric layer) that lies on the surface of the silicon substrate.