IIa Technologies Pte Ltd v Element Six Technologies Ltd [2023] SGCA 5

The parties are competitors in the production of synthetic diamonds grown using chemical vapour deposition (“CVD”). At its simplest, under this process, a substrate (which is also known as a diamond seed) is placed in a reactor containing a mixture of gases, including methane (CH4) and hydrogen (H2). Upon exposure to high energy, the gaseous molecules break up into plasma containing carbon (or “C”) atoms. The C atoms are then deposited on the substrate, growing the synthetic diamond layer by layer.

The appellant, a company incorporated in Singapore, claims to be a major manufacturer of CVD diamonds and has a diamond-growing facility in Singapore. The respondent, a company incorporated in the United Kingdom, is part of the Element Six Group. The Element Six Group is a member of the De Beers Group, which is itself a subsidiary of the Anglo American PLC. The De Beers Group is an international diamond business and a diamond producer, while the Anglo American PLC is a global mining company. According to the respondent, the Element Six Group is a global leader in the design, development and production of “synthetic diamond supermaterials”. The respondent claims that its CVD diamonds have potential technical applications in various industries, including optics, semiconductors and sensors.

In HC/S 26/2016 (“Suit 26”), the respondent claimed that the appellant had infringed two of its patents registered in Singapore, Singapore Patent No 115872 (“SG 872”) and Singapore Patent No 110508 (“SG 508”). To prove infringement, the respondent relied on three samples of diamonds, which were labelled Sample 2, Sample 3 and Sample 4 (collectively, the “Samples”), allegedly purchased from the appellant or the appellant’s related entities or distributors. It claimed that the Samples were made from CVD diamond material synthesised by the appellant and that the appellant must have used the method of growth taught in Claims 62 to 71 of SG 872. In its defence, the appellant denied infringing the patents and, in any case, disputed their validity. The appellant also sought the revocation of both patents by way of a counterclaim. On 7 February 2020, the trial judge (the “Judge”) declared that SG 508 was invalid and revoked it on the basis that it was neither novel nor inventive (Element Six Technologies Ltd v IIa Technologies Pte Ltd [2020] SGHC 26 (“Judgment”) at [291] and [478(d)]–[478(e)]). This was a complete defence to the respondent’s claim for infringement in so far as SG 508 was concerned (Judgment at [416]). However, the Judge found that SG 872 was valid and declined to revoke it. She further found that the appellant had infringed specific claims in SG 872. In this appeal, although both parties proceeded on the basis that Claims 1ii), 1iii), 57, 58 and 62 of SG 872 were infringed, it is not entirely clear whether the Judge confined her finding of infringement to specific limbs of Claim 1 (Judgment at [9], [416], [443], [454], [456] and [478(a)]). But this difficulty need not detain us. Even taking Claims 1ii) and 1iii) as the point of departure for our analysis on invalidity, our conclusions affect the entirety of Claim 1 and the outcome of the appeal will not change even if the Judge had found the whole of Claim 1 to be infringed.

On 6 March 2020, the appellant filed CA/CA 41/2020 (“CA 41”) appealing against the Judge’s rejection of its defence that SG 872 is invalid and its counterclaim to revoke the entirety of SG 872. It also challenged the Judge’s finding that Claims 1ii), 1iii), 57, 58 and 62 in SG 872 had been infringed. The respondent has not cross-appealed against the Judge’s findings in respect of SG 508.

This judgment deals only with CA 41.

Background

Procedural history and case management on appeal

On 17 August 2020, the appellant filed CA/SUM 87/2020 (“SUM 87”), seeking leave to admit further evidence from Dr Werner Kaminsky (“Dr Kaminsky”, the appellant’s expert) for the purpose of challenging the evidence of Dr Anthony Michael Glazer (“Dr Glazer”, the respondent’s expert) on his gap theory (“Gap Theory”) (see [36] below for relevance of the Gap Theory in this appeal). We dismissed SUM 87 on 16 September 2020.

As appears below, the technical background to SG 872 – including the physical properties of diamonds, how SG 872 seeks to enhance the quality of CVD diamonds grown and the measurement or evaluation of the improvement brought about by SG 872 – is highly complex. The following steps greatly assisted us in coming to grips with the difficult material:

(a) On 19 April 2021, we directed the parties to prepare a “Primer” (with a 50-page limit) setting out their points of agreement and divergence on topics including the common general knowledge (such as different methods of growing diamonds and types of defects in diamonds), the state of the art (including other patents) and the inventive concept of SG 872.

(b) We also convened a “Technology Tutorial” that was conducted on 2, 3 and 4 August 2021. Each party was allowed to nominate up to two presenters from among their expert witnesses to address these topics as well as provide an introduction to SG 872 and the Metripol system. The parties filed and exchanged Powerpoint slide decks prepared by their presenters seven days before the Technology Tutorial.

In the section that follows, we outline the salient aspects of the background to SG 872 as presented in the Primer and Technical Tutorial, which will set the context for our analysis of its validity.

Following the Technology Tutorial, CA 41 was heard over three half-day hearings on 20, 24 and 25 January 2022.

Subsequently, on 14 September 2022, we wrote to the parties to clarify whether each product claim in SG 872 covers a class of products as opposed to an individual product. The significance of this point, as well as of the parties’ responses on 21 September 2022, will become clear when we analyse whether SG 872 should be revoked in its entirety.

Technical background

We first set out the technical background to SG 872. As we later elaborate, SG 872 discloses a new single crystal CVD diamond material, at least 0.5mm in thickness, which possesses one or more properties stated in the patent. One such property is low “optical birefringence” (henceforth referred to as “birefringence”). It is the respondent’s case that without the new CVD growth process taught within SG 872, the single crystal CVD diamond material manufactured would not display the properties defined in the product claims of SG 872. It is said that these properties are indicative of a high-quality diamond.

Synthetic diamonds and CVD

Diamonds are a form of solid C (meaning carbon) in which each C atom is bonded to four neighbouring C atoms to form a tetrahedral structure. Many of these tetrahedral structures come together to form a diamond lattice. A “single crystal” diamond, which is claimed in SG 872, is one in which the crystal lattice of the entire sample is continuous and unbroken. In contrast, a polycrystalline diamond consists of many single crystals.

Diamonds may form naturally within the Earth’s crust or be manufactured in a laboratory. CVD is one of the main ways of manufacturing diamonds, the other of which involves applying high-pressure-high-temperature (“HPHT”) on a C source mixed with a catalyst.

Impurities, defects and strain in diamonds

In reality, no diamond (natural or man-made) is an array of equally spaced atoms which are purely C. Diamonds will contain impurities and defects. We make this point to provide context to what the respondent/patentee sought to achieve through the new growth processes taught in SG 872. As will become clear, the type and extent of defects in a diamond bear on its physical properties, such as its strength, and hence its potential industrial use(s) and application(s). The respondent sought to produce CVD single crystal diamonds with certain qualities which were required for specific industrial applications. Allegedly, such diamonds could only be produced using the processes claimed in SG 872. What follows elaborates on the nature of defects or impurities arising in diamonds and the relationship between these imperfections and a diamond’s physical properties.

Impurities refer to non-C atoms bonded into the diamond’s lattice structure. Common impurities include nitrogen (N2) and boron. The concentration of nitrogen and boron is expressed in parts per million (“ppm”) or parts per billion (“ppb”). Based on its purity, a diamond can be categorised into two main types: Type I and Type II. These can be further sub-divided into Types Ia, Ib, IIa and IIb. Type I diamonds contain relatively large amounts of nitrogen – up to 0.3% (5ppm to ~ 3000ppm). Most of the nitrogen atoms in Type Ia diamonds are found in clusters, but this is not the case in Type Ib diamonds. Type II diamonds contain nitrogen below 5ppm. The only impurity in Type IIa diamonds is nitrogen, whereas impurities in Type IIb diamonds consist of nitrogen and boron.

When impurities (meaning non-C atoms) enter the diamond’s structure, a point defect is created in the lattice structure. Extended defects, on the other hand, are structural defects caused by disruptions to the diamond lattice that extend across the crystal. One example of an extended defect is a stacking fault, which occurs where the relative positioning of two adjacent layers (or planes) of crystal structure does not conform with the rest of the crystal:

Another type of extended defect is a dislocation. Dislocations occur where there are atoms out of position in the crystal structure that, among other things, form an extra line of atoms within the structure (that is, an edge dislocation):

Extended defects can be visualised using X-ray topography images where stacking faults appear as dark planes (Figure 3) and dislocations appear as dark lines (Figure 4). Thicker and relatively darker lines show bundles of dislocations:

Dislocations cause strain, or distortion to the position of atoms, in the diamond structure. The strain extends across the full length of the dislocation, which typically runs all the way across the diamond. Strain weakens the chemical bonds, which then can be broken with lower energy. This reduces the strength of the diamond to withstand external pressure. According to Dr Christoph Erwin Nebel (“Dr Nebel”), the appellant’s principal expert, if there is strain, electrons between the atoms are no longer homogenously distributed, resulting in an electric field that may interact with light and increase birefringence in the diamond (a property we explore in some detail at [20]–[27] below).

Dislocation density refers to the number of dislocations per square cm or mm of surface. Dislocations extend across the diamond like a line (hence, “line defects”) and terminate only on the surface:

Dislocation termination points reveal themselves as etch pits. An optical evaluation of the number of etch pits across the surface area of the diamond will reveal its dislocation density:

Birefringence

Birefringence is a physical property of diamonds relevant to the key claims in SG 872. We shall therefore explore this concept in some detail.

Birefringence, stated generally, is related to how light interacts with a certain object. Light is made of waves of electric and magnetic fields. These waves oscillate in all directions perpendicular to the direction of the transmission of light. The distance travelled when an electromagnetic wave goes through one full oscillation cycle is called wavelength, which is denoted by λ. The wavelength of light determines its colour.

While normal light consists of electric fields oscillating in all directions in the plane perpendicular to the direction of propagation, a polariser filter may be used to only allow light containing electric fields oscillating in a single direction to pass through. “Polarised light” therefore has only one direction of oscillation.

The atomic arrangement in a material affects the speed at which light travels through it. The speed of light in an optical material is determined by the material’s refractive index. The denser the material, the higher its refractive index and the slower that light travels through it.

Where the internal structure of the crystalline material is perfect, in that it has no impurities or defects, the crystalline material is highly symmetrical. Such material is referred to as isotropic. In isotropic materials, the refractive index is the same in all directions. This means that light passes through an isotropic crystal at a single velocity.

But for anisotropic materials, the refractive index may vary in different directions. Although a diamond may theoretically be isotropic due to its highly symmetrical uniform lattice of C atoms, in reality, diamonds are anisotropic due to strain caused by dislocations. When polarised light is passed through a diamond with strain, the light splits into two relevant components which travel at different velocities. One component travels faster along the path in the material with the smallest refractive index (n”) and the other travels slower along the path in the material with the largest refractive index (n’). The difference between the smallest and largest available refractive index is called birefringence (Δn = |n’ – n’’|). In other words, strain in diamonds results in birefringence.

The gap between the faster and slower components of light when they emerge out of the material is defined as the optical retardation (also referred to as “retardation” or “path difference”). Light of any wavelength which passes through a sample of an anisotropic material having birefringence Δn, where the sample has a thickness of “L”, will display retardation of L·Δn. Dr Nebel depicted birefringence and retardation as follows:

For light of a particular wavelength (or λ), retardation can also be represented as phase shift “δ”:

Dr Nebel represented the components of the above equation, using the following diagrams of light components that have split up when passing through an object:

Salient features of SG 872

This appeal concerns SG 872, which was filed on 20 November 2003. In essence, the patent covers a new optical quality synthetic single crystal CVD diamond material and the method for its production. The patent contains both product and process claims. A product claim is one which “asserts exclusive rights over a new thing, such as new machines, items or articles of manufacture and composition of matter”. Although the patentee may describe only one use of his invention, he has the right to stop others from using it for other purposes (Susanna H S Leong, Intellectual Property Law of Singapore (Academy Publishing, 2013) (“Susanna Leong”) at para 18.002). We will henceforth refer to a diamond falling within any of the product claims in SG 872 as an “SG 872 Diamond”. On the other hand, “a process claim covers a new activity such as a process or method of manufacturing and logically, in a method or process claim, it is the process used and not the end result which must be new to justify patentability.” (Susanna Leong at para 18.003).

The pivotal product claim, which all subsequent claims refer back to, is Claim 1:

Single crystal CVD diamond material falling within Claim 1 has low strain and exhibits at least one of the optical properties defined in the eight limbs of Claim 1. Two of these limbs, limbs ii) and iii), assume significance in this dispute. Following the terminology used by the parties and the Judge, we will refer to these two limbs as “Claim 1ii)” and “Claim 1iii)” respectively. As the respondent pleads that Claims 1ii) and 1iii) were infringed (see Judgment at [417]), we elaborate on these. Claims 1ii) and 1iii) assert a monopoly over single crystal CVD diamond material with low birefringence. Low birefringence is quantified in Claim 1ii) as a sample (a) remaining in the first order (δ does not exceed ); and (b) with the modulus of the sine of the phase shift (that is, |sin δ|) not exceeding 0.9. Low birefringence is quantified in Claim 1iii) as a sample (a) remaining in the first order (δ does not exceed ); and (b) with the maximum value of Δn[average] (that is, the average birefringence across the thickness of the sample) not exceeding 1.5 x 10-4.

We highlight four key points pertaining to Claims 1ii) and 1iii) which the parties do not dispute. First, the term “first order” for the purposes of SG 872 (and Claims 1ii) and 1iii) in particular) has been defined in a specific way. The parties’ experts, when explaining certain technical concepts (see above at [27] and below at [191]), referred to each “order” as a multiple of 2π, with “first order” covering a range of values where δ does not exceed 2π (or again, expressed as an equation, δ ≤ 2π). However, SG 872 defines “first order” in a much more limited sense, namely, where δ does not exceed (or again, expressed as an equation, δ ≤ ) (the “SG 872 First Order”).

Second, it is undisputed that SG 872 directs the person skilled in the art (“PSA”) to use a Deltascan (which has been renamed as the Metripol since 2001) or a “similar instrument with similar resolution” to determine the value of |sin δ| and Δn[average]. The Metripol is an optical microscope-based imaging system, which was developed by Dr Glazer (the respondent’s expert), Dr Kaminsky (the appellant’s expert), and Dr Morten Geday (“Dr Geday”) (the appellant’s witness) in or around 1995. The Metripol has a polariser that rotates in fixed multiples of different angles (for example, by 30o, 60o, 90o) (“Rotation Angles”). At each angle, polarised light passes through the specimen and is captured by a camera. The Metripol calculates the intensities of light received by the camera at different Rotation Angles, and thereafter computes |sin δ|, which is a function of δ.

It will be recalled that δ is the optical retardation of light of a particular wavelength that passes through a material, and is related to the birefringence of the material by the formula set out at [27] above. As appears from the three graphs below (in particular, the green arrows in graphs (b) and (c)), |sin δ| is the amplitude of a sinusoidal curve (meaning half the distance from the highest point of the curve to its lowest point). This sinusoidal curve is derived by plotting the intensity of polarised light that passes through the diamond specimen on the y-axis, against the polariser’s Rotation Angle on the x-axis. Depending on whether and the extent to which the intensity of light varies when the polariser is angled differently, |sin δ| will assume a different value:

Third, after computing |sin δ|, the Metripol produces colour-coded images of the material, where the different colours represent different |sin δ| values at any place within the image (the “Metripol |sin δ| Map”). The Metripol is capable of refining the |sin δ| value at each pixel position in the Metripol |sin δ| Map. An illustration of the Metripol |sin δ| Map can be found in the following diagram, taken from the respondent’s Technology Tutorial slides:

The same Metripol |sin δ| Map can also be represented in greyscale. The patent specification of SG 872 states that the spatial variation of |sin δ| can be appreciated from these images. The patent specification also states that for each sample, sets of |sin δ| images are to be recorded for analysis.

Fourth, it is agreed between the parties that the Metripol data on its own does not tell the PSA whether the δ value of a diamond falls within the SG 872 First Order (the “Metripol Uncertainty Problem”). The Metripol does not calculate δ. As mentioned, it generates the value of |sin δ|. However, each |sin δ| value corresponds to many δ values, and only one of these δ values lies within the SG 872 First Order. From the |sin δ| value alone, the PSA cannot derive a corresponding δ value and will not know whether the δ value of a diamond is within or outside the SG 872 First Order. This point can be illustrated in Figure 13 below, where δ1 is the value that lies within the SG 872 First Order:

If the Metripol shows that the |sin δ| value of the diamond is 0.9, the corresponding δ value can be 1.12, 2.02, 4.26, 5.16, or any other higher value not reflected in Figure 13 above. δ will only be within the SG 872 First Order if it has a value of 1.12, as that is the only value that does not exceed (see above at [31]). The rest of the values lie outside the SG 872 First Order. From this range of possible δ values, the PSA relying solely on the Metripol reading cannot determine the δ value of the diamond, and accordingly, cannot ascertain whether δ is within the SG 872 First Order.

In this appeal, as they did in the proceedings below, the parties diverge on the issue of whether the PSA knew of solutions to overcome the Metripol Uncertainty Problem at the relevant time. The thrust of the appellant’s contention is that the PSA, faced with the Metripol Uncertainty Problem and with no knowledge of any workable solution at the relevant time, would not be able to determine whether the δ value of the diamond was within the SG 872 First Order and hence within the scope of the asserted monopoly. For this reason, the appellant contends that SG 872 is insufficient. In turn, the respondent’s case is that the PSA would have no difficulty ascertaining that the δ value of the diamond was within the SG 872 First Order by using various methods, such as the Gap Theory proposed by Dr Glazer. We will deal with this issue in our analysis later.

We now turn to Claim 62, the main process claim in SG 872 to which all other process claims in SG 872 refer. Claim 62 specifies a method of producing low-strain single crystal CVD diamond material that exhibits optical properties defined in the product claims in SG 872. Claim 62 states as follows:

Certain terms relating to the growth process, which are referred to subsequently, call for further explanation. The “source gas” in the CVD chamber comprises several gases, including a “C” source which is typically methane (CH4) and a source of atomic hydrogen which is typically hydrogen (H2). The parties agree that other gases, such as an inert gas (like argon) and a dopant gas (like nitrogen, boron or phosphorous), a source of impurities which is used to affect the electrical or optical properties of a diamond, are also used depending on the desired characteristics of the CVD diamond. Each input gas is controlled by a mass flow controller (“MFC”). The flow of gases across MFCs is measured as cubic centimetres of gas flowing through the MFC per minute (expressed as standard cubic centimetres per minute or “sccm”). The concentration of each gas is represented as a percentage, ppm or ppb. As for the “plasma etch” performed on the substrate prior to the commencement of the CVD process, this involves “exposing the substrate to a plasma formed by hydrogen (H2) and/or oxygen (O2) at high temperature in a CVD chamber.” The specification of SG 872 states that such in situ plasma etching minimises the surface damage of the substrate. We note, however, that the respondent disputes whether the use of a plasma etch to achieve this effect is common general knowledge to a PSA (we will explain the concepts of common general knowledge and the PSA at [63]–[75] below).

The respondent claims that the production of low-strain single crystal CVD diamond material disclosed in Claims 1 to 61 and 72 to 78 was not possible prior to the discovery of the process in Claim 62. The respondent argues that the growth process in Claim 62, specifically, the addition of 300ppb to 5ppm nitrogen to the synthesis atmosphere (the “Claim 62 Nitrogen Range”), is novel because it runs contrary to previous thinking that adding nitrogen to the synthesis atmosphere would have a deleterious effect on the quality of the diamond produced; nitrogen was regarded as an impurity. As the Judge noted, the common general knowledge at the material time was that the addition of nitrogen would increase birefringence. The Claim 62 Nitrogen Range is therefore the inventive concept asserted in Claim 62 (see Judgment at [253]). We agree. According to the specification of SG 872, the Claim 62 Nitrogen Range is not too high because it accomplishes the aim of “prevent[ing] or reduc[ing] local strain generating defects whilst being low enough to prevent or reduce deleterious absorptions and crystal quality degradation.” Nor is the range too low, because, according to SG 872, “material grown under conditions with essentially no nitrogen, or less than 300ppb nitrogen has a comparatively higher level of local strain generating defects”. Dr Philippe Bergonzo (“Dr Bergonzo”), one of the respondent’s experts, depicted the benefits of operating within the Claim 62 Nitrogen Range as follows (see the curve in black):

As against this, the appellant regards the purported benefits of using the Claim 62 Nitrogen Range as “illusory” and “false”. It questions why, in the 17 or so years since SG 872, not a single paper has been published discussing or evidencing this discovery. However, as we shall explain, even assuming that the purported benefits of the Claim 62 Nitrogen Range are scientifically true, there are other issues undermining the validity of the patent.

Decision below

Against the appellant’s arguments to the contrary, the Judge held that Claims 1 and 62 of SG 872 were valid. For brevity, we will only summarise the Judge’s reasons for rejecting the appellant’s contentions on insufficiency, and only in so far as they are relevant to the determination of this appeal.

The Judge observed that the appellant had relied on s 25(5) of the Patents Act (Cap 221, 2005 Rev Ed) (“Patents Act (2005 Rev Ed)”), which provided that claims shall be “clear and concise” in s 25(5)(b). Citing this court’s decision in First Currency Choice Pte Ltd v Main-Line Corporate Holdings Ltd and another appeal [2008] 1 SLR(R) 335 (“First Currency Choice”) at [72], the Judge held that the requirement of clarity was not a ground for revocation under the Patents Act (2005 Rev Ed). She emphasised that what was important was whether the invention has been disclosed sufficiently for a PSA to perform it (Judgment at [191]–[192]), and proceeded to consider the appellant’s contentions on why Claim 1 was insufficient. One of these contentions was that Claim 1 of SG 872 was insufficient because the Metripol could not identify whether δ was within the SG 872 First Order (Judgment at [193]–[196]). She eventually rejected that contention because she accepted Dr Glazer’s evidence that the PSA would know how to determine whether δ was within the SG 872 First Order using the Gap Theory (Judgment at [200]–[206]).

The Judge also found that Claim 62 had sufficiently disclosed the claimed invention. The appellant argued that the PSA would not know how to calibrate the other growth conditions to the chosen nitrogen concentration to grow an SG 872 Diamond. The Judge rejected this submission given Dr Bergonzo’s evidence that the missing details would be known to a PSA with a working knowledge of the research and development of CVD diamond synthesis (Judgment at [285]).

Another of the appellant’s submissions was that the insufficiency of Claim 62 was demonstrated by how the single crystal CVD diamond material grown in Examples 1 and 14 of SG 872 as well as Examples 4 and 6 of SG 508 had different qualities, even though these examples disclosed a growth process where the synthesis atmosphere had 300ppb to 5ppm nitrogen. The Judge rejected this submission, noting that the respondent had explained that the difference in other parameters of the growth process (such as gas flow and methane concentration in the synthesis atmosphere) between SG 508 and SG 872 resulted in a diamond with different qualities being grown. She also observed that these other parameters did not, in any event, constitute the inventive concept of SG 872 (Judgment at [286]–[287]).

Given her finding that Claims 1 and 62 of SG 872 were valid, the Judge did not revoke SG 872 and proceeded to find that both Claims 1 and 62 had been infringed (Judgment at [416]). In particular, she found that the Samples infringed Claim 1 on the basis that they fulfilled limbs ii) and iii) of Claim 1 (Judgment at [431]–[443]). The Judge therefore made the following orders, amongst others (Judgment at [478(a)]–[478(c)]):

The parties’ cases

Appellant’s case

The appellant’s attack on the Judge’s decision cuts across many areas of patent law, such as novelty, inventive step, and sufficiency. Much reliance has also been placed on concepts recognised in English law which have not been recognised locally, such as “Biogen insufficiency”. For the purposes of the present appeal, we only need to focus on two parts of the appellant’s case, both of which concern the non-fulfilment of the sufficiency requirement and are determinative of this appeal.

The first relates to the insufficiency of Claim 62 on the basis that the patent specification fails to teach the “novel” growth process clearly and completely enough for it to be performed by a PSA. In particular, the appellant faults Claim 62 for not disclosing “what ranges of temperature, gas flow rate or methane concentration will give diamond material within claim 1iii) or any other product claims for that matter.” We shall refer to these variables, and any other variables in the patented CVD growth process (other than the concentration of nitrogen), as the “Other Growth Conditions”. The appellant argues that it is not enough to point to the examples in SG 872 because these “do not support a claim of the breadth of claim 62.” As a result, the PSA is “left with a long and extensive research project to find out what ranges of these other process parameters are required to obtain claim 1iii) or 1ii) diamond material” and that, therefore, SG 872, is insufficient.

The appellant contends that the respondent’s own arguments under novelty establish insufficiency. Relevantly, the respondent argues that Examples 4 and 6 of SG 508, which teach the use of 2.5ppm and 3.8ppm of nitrogen in the synthesis gas mixture respectively, do not anticipate SG 872 because there are “a lot of differences” in terms of gas flow, temperature and methane concentration between those examples and Claim 62. The respondent asserts on this basis that Examples 4 and 6 in SG 508 do not anticipate Claim 62 because following those examples will not inevitably lead to the production of an SG 872 Diamond. But if this is correct, the appellant points out that it must mean “there are other ranges of essential parameters for the process in SG 872 to ‘work’” and that their omission from Claim 62 or the specification of SG 872 means that the invention is insufficiently disclosed.

Before proceeding further, we note the respondent’s submission that pursuant to O 57 r 9A(5B) of the Rules of Court (2014 Rev Ed), the appellant cannot impugn the clarity and completeness of SG 872 as this ground of insufficiency is not particularised in the Appellant’s Case (the “O 57 Objection”). Notwithstanding the respondent’s contentions, we are inclined to consider the appellant’s arguments on this issue (see Oxley Consortium Pte Ltd v Geetex Enterprises Singapore (Ptd) Ltd and another matter [2021] 2 SLR 782 at [45]; Global Yellow Pages Ltd v Promedia Directories Pte Ltd and another matter [2017] 2 SLR 185 at [22]). Although the Appellant’s Case focused on another type of insufficiency that we will turn to next, it foreshadowed the possibility of other insufficiency arguments arising “depending on how the Respondent trie[d] to avoid invalidity.” And in its Reply, the appellant developed its submissions on this point in direct response to the Respondent’s Case. The respondent also did not press the O 57 Objection in oral submissions and instead dealt substantively with the appellant’s arguments in question. There is therefore no procedural impropriety afoot and/or prejudice to the respondent that precludes the appellant from relying on this ground of insufficiency.

The second part of the appellant’s case which we will focus on concerns the insufficiency of Claim 1 arising from ambiguity or uncertainty. The appellant argues that the Judge erred in failing to consider this type of insufficiency, which arises where the PSA does not know whether he is inside or outside the claim. On the facts, it argues that the Claim 1 and SG 872 are insufficient because the PSA cannot tell whether the δ value of a particular single crystal CVD diamond was within the SG 872 First Order and hence within the scope of the asserted monopoly.

Respondent’s case

The respondent supports the Judge’s findings that SG 872 is not anticipated, obvious or insufficiently disclosed.

Its case is that SG 872 claims a new single crystal CVD diamond material with exceptionally low strain and a new method of producing said diamond material. By way of a letter dated 21 September 2022, which was sent in response to our query (see above at [10]), the respondent accepts that each product claim in SG 872 covers a class of products. The distinction between a single product and a class of products is relevant to the issue of insufficiency. If the claim is to a single product, then it is sufficient if it enables the making of that one product. If, on the other hand, the claim is to a class of products, that class of products is sufficiently enabled only if the skilled man can work the invention in respect of all members of the class (Regeneron Pharmaceuticals Inc v Kymab Ltd and another [2016] EWHC 87 (Pat) at [209]). We return to these principles at [108]–[112] and [184] below.

In response to the appellant’s submission that the patent is insufficient because it fails to teach the Claim 62 growth process clearly and completely enough, the respondent argues that the correct test is whether the patent’s specification as a whole discloses the invention sufficiently. It argues that the specification of SG 872 (in particular Example 1 read with Example 9) gives full disclosure of the Other Growth Conditions, including temperature, pressure and gas flow rates. It stresses that the burden is on the appellant to show that the invention is unworkable by the PSA who is “trying to give practical meaning to the patent specification.”

Next, as to the appellant’s submission that the Judge erred in failing to consider a further type of insufficiency brought about by ambiguity or uncertainty, the respondent submits that the Judge rejected the appellant’s argument on insufficiency arising from “ambiguity”, on the basis that in First Currency Choice we held that lack of clarity under s 25(5)(b) of the Patents Act (2005 Rev Ed) is not a ground for revocation. It further submits that the appellant has not explained how the Judge erred in following First Currency Choice, or why the court should depart from the position at law that lack of clarity is not a basis for revocation. On the facts, the respondent submits that the PSA can determine that δ is within the SG 872 First Order by using four different methods.

Issues to be determined

In light of the foregoing, the issues we shall address in this decision are as follows:

Foundational principles

Patents in general

To set the context for the analysis that follows, we first touch on some foundational principles that guide us, beginning with the objectives of the patent regime and how these are advanced. Where relevant, we also explain principles of patent law which are engaged in the reasoning of this judgment, or which provide context to our analysis.

To ensure that patents effectively incentivise innovation in furtherance of the first aim, the patent regime only grants monopolies over deserving inventions. An invention must be novel, contain an inventive step and be capable of industrial application, before it will be patentable (ss 13–16 of the Patents Act (2005 Rev Ed)). If any one of these requirements is not met, the patent may be revoked (s 80(1)(a) of the Patents Act (2005 Rev Ed)). We briefly introduce the concepts of novelty and inventive step as the parties have made submissions on them (although, as foreshadowed at [46] above, we will not analyse the parties’ submissions on these requirements in this judgment).

Novelty requires the invention to be new, in that it does not form part of the state of the art (s 14(1) of the Patents Act (2005 Rev Ed)). If directions forming part of the state of the art (meaning a piece of prior art) lead inevitably to something that would infringe the patent, the invention lacks novelty and the patent can be revoked (Mühlbauer AG v Manufacturing Integration Technology Ltd [2010] 2 SLR 724 (“Mühlbauer”) at [17]). The state of the art comprises all matter (whether a product, a process, information about either, or anything else) which has at any time before the priority date of that invention been made available to the public (whether in Singapore or elsewhere) by written or oral description, by use or in any other way (s 14(2) of the Patents Act (2005 Rev Ed)). By default, the “priority date” of an invention is the filing date of the patent application (s 17(1) of the Patents Act (2005 Rev Ed)). Under certain conditions, a patentee can depart from this default position by claiming that the priority date of the invention takes reference from the date of filing of an earlier patent application or applications (see s 17(2) of the Patents Act (2005 Rev Ed)), which has the effect of limiting the state of the art to what it was at that earlier date. On the other hand, a later priority date may widen the pool of material which may anticipate the invention.

Separately, the second aim of the patent regime, namely knowledge dissemination, is advanced by the “sufficiency” or “enabling disclosure” requirement, which provides that the patent specification must disclose the invention in a clear and complete manner for it to be performed by a PSA. This requirement is statutorily embodied in ss 25(4) and 80(1)(c) of the Patents Act (2005 Rev Ed), with the latter providing for the revocation of the patent if the patent specification is insufficient.

Another point of commonality is that the court, when undertaking the novelty, obviousness and sufficiency inquiries, as well as in construing the claims, dons the mantle of the PSA, that is, someone in the field of technology relevant to the invention in question and who possesses, amongst other characteristics, common general knowledge in the art (Mühlbauer at [18] (novelty); ASM Technology Singapore Pte Ltd v Towa Corp [2018] 1 SLR 211 at [78] (obviousness); First Currency Choice at [60]–[61] (sufficiency); Lee Tat Cheng v Maka GPS Technologies Pte Ltd [2018] 1 SLR 856 (“Lee Tat Cheng (CA)”) at [41] (claim construction)). Common general knowledge is information which, at the relevant date, is common knowledge in the art to which the alleged invention relates, so as to be known to duly qualified persons engaged in that art (British Thomson-Houston Co Ltd v Stonebridge Electrical Co Ltd (1916) 33 RPC 166 at 171; Terrell at para 8-61).

Further, the precise assumptions applicable to the PSA are not identical in all areas of patent law. For instance, in matters of claim construction and sufficiency, the PSA is assumed to have the patent specification in his hands whereas when considering the issue of obviousness, the PSA is deemed only to be considering the prior art without the patentee’s invention in front of him (Terrell at para 8-19).

The central issues concerned in our decision relate to claim construction and insufficiency. To reiterate, the former identifies the invention that is the subject of the claims, whilst the latter entails a consideration of whether the patent specification sufficiently enables a PSA to perform that invention. It is in this context that we now turn to examine the characteristics of the PSA relevant to our reasoning in the present judgment, and the common general knowledge he (or, the notional team comprising the PSA) would possess.

The person skilled in the art

The PSA, apart from possessing common general knowledge in the art, also has a practical interest in the subject matter of the patent and is likely to act on the directions given in it with the desire to make the directions in the patent work. He is a reasonably intelligent but unimaginative workman or technician who has the skill to make routine workshop developments, but not to exercise inventive ingenuity or think laterally (Ng Kok Cheng v Chua Say Tiong [2001] 2 SLR(R) 326 (“Ng Kok Cheng”) at [21]; First Currency Choice at [28]; Ng-Loy at para 30.1.12).

The PSA may be an individual person or a notional team of people. The need for a notional team may arise where it is clear that the patent specification engages more than one set of skills which in the real world would be possessed by more than one person. This may be the case where the art is one that concerns a highly developed technology. In such circumstances, the specification can be said to be addressed to a team, with each member contributing his or her individual skill which would in turn be employed in combination in interpreting and carrying into effect the instructions in the patent (Terrell at paras 8-35–8-36; The General Tire & Rubber Company v The Firestone Tyre and Rubber Company Limited and others [1972] RPC 457 (“General Tire”) at 483; Schlumberger Holdings Ltd v Electromagnetic Geoservices AS [2010] EWCA Civ 819 at [33]; Halliburton Energy Services, Inc v Smith International (North Sea) Ltd [2006] RPC 2 at 46; see also Institut Pasteur and another v Genelabs Diagnostics Pte Ltd and another [2000] SGHC 53 (“Genelabs (HC)”) at [194]). For the avoidance of doubt, the formation of such a notional team does not mean that the common general knowledge of the PSA is only that which is common to persons with these different backgrounds (Inhale Therapeutic Systems Inc v Quadrant Healthcare Plc [2001] All ER (D) 211 (Jun) at [39]–[42]; First Currency Choice at [31]).

As noted at [65] above, when considering the issues of claim construction and sufficiency, the PSA is assumed to have the patent specification in his hands and have the same attributes for both inquiries (see Terrell at para 8-19). The PSA, for the purposes of the insufficiency inquiry, is “trying to carry out the invention and achieve success” (Zipher Ltd v Markem Systems Ltd and another [2008] EWHC 1379 (Pat) (“Zipher”) at [366]). It is through the eyes of the PSA that the patent will fall to be interpreted, and the PSA must know how to perform the invention upon reading the specification in light of his common general knowledge.

Moving now to the PSA in the present case, the Judge found that the PSA, for the purpose of claim construction, would be a notional team of persons working in the field of growing CVD diamonds, with expertise in diamonds (including natural, HPHT and CVD diamonds) and the sciences (including applied physics, optical engineering and material sciences) (Judgment at [14]):

The parties have not challenged the Judge’s finding in that passage of the composition and qualifications of the notional team of persons. Further, as already mentioned at [69] above, the PSA for the purposes of claim construction and sufficiency will be the same. However, before us, the appellant seeks to impute the PSA with the specific purpose of growing single crystal CVD diamond material for use in the most demanding optical applications – “etalons and anvils”. The respondent rejects this imputation. Although this dispute between the parties mainly arises in the course of their submissions on novelty and obviousness, a proper identification of the PSA’s practical interest is also important in the context of claim construction and sufficiency.

As earlier stated, the PSA is trying to carry out the invention (see [69] above) and has a practical interest in the “subject matter of the patent” (see [67] above). What the subject matter of the patent is should be determined by reference to the patent specification. Having regard to the specification of SG 872, we are satisfied that the PSA’s interest would not be limited to producing etalons and anvils. The patent specification describes the invention in SG 872 in more general terms – a single crystal CVD diamond material which possesses certain desirable characteristics, and which can be used in connection with a wide range of optical applications which “include but are not limited to” optical windows, laser windows, optical reflectors, optical refractors and lenses, diffractive optical elements, etalons, ornamental use and anvils [emphasis added]. Accordingly, the PSA’s practical interest lies in using the growth process taught in SG 872 to grow low-strain diamonds for a range of optical applications.

The common general knowledge

As already noted, the PSA is imputed with common general knowledge in the art. He draws upon such knowledge to understand a patent and in his attempts to carry it into effect. As mentioned at [63] above, common general knowledge is information which, at the relevant date, is common knowledge in the art to which the alleged invention relates, so as to be known to duly qualified persons engaged in that art. The relevant art to which the alleged invention is to be gathered from the patent specification itself (First Currency Choice at [30]).

Common general knowledge is ordinarily established, where not agreed, by expert evidence which is usually supported by reference to textbooks or other reference works which the PSA would be expected to have access to and acquire his information from. The publication at or before the relevant date of such documents may be prima facie evidence tending to show that the statements contained in them were part of the common general knowledge, but that is far from complete proof as those statements may well have been discredited, forgotten or ignored (Terrell at paras 8-75–8-77). It is also not sufficient just to point out that it was said in a scientific paper that has been well circulated. A particular disclosure does not become common general knowledge merely because it is widely read, and still less because it is widely circulated. To prove that a particular disclosure in scientific paper has become common general knowledge, it must be shown that it is generally known and generally regarded as a good basis for further action by the bulk of those who are engaged in the particular art to which the disclosure relates (General Tire at 480–481).

In the present case, as identified by the Judge, the art relates to the growth of CVD diamonds (Judgment at [14]). In our view, this includes methods of measuring the physical properties of the CVD diamonds grown. Based on the Primer, the parties are agreed that the information set out at [11]–[27], [32]–[35] and [38] above constituted common general knowledge in the art of growing CVD diamonds at the relevant time. There are further points of agreement on what was part of the common general knowledge and they will be mentioned, where relevant, when we apply the principles of sufficiency to the facts of this case. But before going into the issue of sufficiency, it is pertinent for us to first consider how the claims, in particular the product claims, ought to be construed. It is to this issue that we now turn.

Patent construction: do the product claims in SG 872 cover a single product or a class of products?

The question here is whether the product claims in SG 872 cover a single product or a class of products. As we shall see, this affects what, in law, must be enabled by the specification of SG 872 (see [108]–[112] below), and whether the claims in SG 872 are in fact insufficient (see, eg, [184]–[185] below).

Whether the patent claims a single product or a range of products is a question of construction of the claim(s) in the patent (Regeneron Pharmaceuticals Inc v Kymab Ltd [2021] 1 All ER 475 (“Regeneron (SC)”) at [3]. Before setting out the principles on claim construction in detail, we touch on the relevance of the priority date(s) of the product claims in SG 872 to their interpretation.

In Lee Tat Cheng (CA) at [53] and [55], we held that the relevant date for claim construction is the date of the patent application (meaning its filing date) (see also Lee Tat Cheng (HC) at [49]). However, it does not appear that the patentee there claimed that its invention took priority from an earlier filing. If the patentee there had claimed priority, a conundrum noted by the authors of Terrell would have arisen. That conundrum arises from the fact that the English cases are divided over whether the relevant date for patent construction is the priority date or filing date of the patent application, or the publication date of the patent (see Terrell at paras 9-08–9-014). At present, the respondent does claim priority from GB Patent Application No 0227261.5 (“GB 261”), which was filed on 21 November 2002. The parties agree that Claim 1ii) takes priority from 21 November 2002 but disagree on whether Claims 1iii) and 62 enjoy the same priority. This specific complication did not present in Lee Tat Cheng (CA). However, nothing in this case turns on which of the three dates should be adopted in law and the parties do not submit otherwise. We therefore prefer to leave this issue for an appropriate case and proceed on the basis that the relevant date for patent construction is the filing date of SG 872 (that is, 20 November 2003). Consequently, any common general knowledge which may be relevant to patent construction must be shown to exist by this date. In addition, whether the asserted priority date is granted only assumes significance in this case in relation to the issues of novelty and obviousness. However, as stated at [46], we need not reach or address the latter issues in this judgment.

Turning to the substantive principles on claim construction, in Lee Tat Cheng (CA), we reaffirmed the correctness of construing a patent’s claims purposively (at [41] and [55]). This entails, among other things, viewing the claims through the lens of the PSA (at [41]):

The patent “specification”, referred to in the passage above, comprises the description of the invention, drawings (if any) and the claims (Halsbury’s Laws of Singapore vol 13(3) (LexisNexis, 2020 reissue) (“Halsbury’s Singapore IP”) at para 160.321). Examples illustrating how to carry out the invention also form part of the patent’s specification (see rr 19(3) and 19(5)(e) of the Patents Rules (1996 Rev Ed)). However, the claims serve a different purpose from the rest of the patent’s specification, as explained by Floyd LJ in Adaptive Spectrum & Signal Alignment Inc v British Telecommunications PLC [2014] EWCA Civ 1462 at [45]:

The parties initially appear to have approached this appeal on the basis that each product claim is a claim to a single product – a new CVD diamond material. For instance, the Appellant’s Case states that “Claim 1 is a product claim, so the eight alternates in Claim 1 disclose the same physical thing – a particular CVD diamond material, characterised in eight different ways”. The Respondent’s Case states that “SG 872 claims a new diamond material” [emphasis in original omitted; emphasis added in italics]. Both parties took similar positions in their written closing submissions in the second tranche of the trial.

We wrote to the parties on 14 September 2022 seeking clarification on this issue. The appellant confirmed its view that SG 872 teaches the PSA a new method to “grow one product, i.e. a CVD single crystal diamond material having one or more of the specified parameters.” It argued that the “different ranges and combination of properties in each product claims [were] therefore variations of the same product grown from the one process described in SG 872” [emphasis added]. In contrast, the respondent took the view that each product claim could be said to cover a class of products. In particular, the respondent noted that “given that the product claims of SG 872 claim, inter alia, a range of values (or depend on a claim which does), these would cover a class of products meeting the values within that range as opposed to an individual product.”

We agree with the respondent. Properly construed, each claim asserts a monopoly over a class of single crystal CVD diamond materials.

We turn first to Claim 1, which, as noted above, is the claim that all the other product claims refer back to. In our judgment, Claim 1 can be infringed by various types of single crystal CVD diamond materials, each with a different combination of the physical properties defined in the limbs of Claim 1. That the diamond materials falling within Claim 1 may possess different combinations of physical properties is clear from the words in the chapeau of Claim 1: “A CVD single crystal diamond material which shows at least one of the following characteristics” [emphasis added]. Our reading of the plain language of Claim 1 is fortified by the description of the invention in the specification of SG 872. The relevant portion of the description reads as follows:

The “above properties” referred to in the passage include optical homogeneity, effective refractive index, birefringence, optical absorption, optical scatter, laser damage threshold, thermal conductivity, parallelism, flatness and mechanical design strength. Thus, the description indicates that the invention covers a range of single crystal CVD diamonds possessing different combinations of physical properties.

Further, as noted in Lee Tat Cheng (CA) at [41(c)], the language of the claim must be construed through the eyes of a PSA. In this regard, we found the evidence of Dr Mark Edward Newton (“Dr Newton”), the respondent’s expert witness, to be instructive. As the respondent alluded to in its letter of 21 September 2022, Dr Newton testified that the various limbs in Claim 1 cover optical properties which “may be related but remain different” [emphasis added in bold italics]. He added that “[m]any are not directly derivable one from another”. It is not clear if these observations also apply to Claims 1ii) and 1iii), which both relate to birefringence. Dr Newton’s evidence nevertheless reinforces our understanding that diamonds which fulfil different combinations of limbs in Claim 1 are in fact different diamonds. Our conclusion would be different if the measurements in each limb of Claim 1 were different ways of quantifying the same property.

Although Dr Nebel’s first report dated 13 May 2019 (“Dr Nebel’s 1st Report”) stated that a diamond satisfying one limb of Claim 1 will also satisfy all remaining limbs, and that the properties mentioned in Claim 1 are “merely ... different measurement techniques for a common aspect of the diamond” [emphasis added in bold italics], we reject this. For one, he offered no credible evidence in his Reply Report dated 26 June 2019 or oral testimony to support this assertion. In his reply report, Dr Nebel asserted that measurements like “dislocation density, Raman peak width, X-ray rocking curve width and birefringence” measure “strain in diamond ... and any change in strain in a diamond reflects a proportional change in these measurement parameters.” But Dr Nebel did not explain nor has the appellant otherwise shown that the optical properties or measurements described in each limb of Claim 1, including optical homogeneity, free spectral range, contrast ratio, insertion loss and variation in refractive index, merely measure a common property or will necessarily be present if one limb of Claim 1 is fulfilled. Crucially, Dr Nebel’s evidence does not square with the specification of SG 872. The “Summary of the Invention” states clearly that the CVD single crystal diamond material invented “may not be endowed with all of the above properties in all cases”, but should show “at least one, preferably at least two, more preferably at least three, and even more preferably at least four” characteristics which are listed. The listed characteristics in the Summary of the Invention correspond to properties defined under the limbs in Claim 1, and include optical homogeneity (see Claim 1i)), birefringence (see Claims 1ii) and 1iii)), effective refractive index (see Claim 1iv)), free spectral range (“FSR”) (see Claim 1v)) and contrast ratio (see Claim 1vi)). Moreover, in its submissions below, the appellant did not appear to dispute the correctness of the portion of Dr Newton’s evidence that we have accepted at [85] above.

Turning now to the subsequent product claims, these likewise cover a range of diamonds. These claims may be placed into three categories based on their phraseology, all of which incorporate Claim 1. The first category of claims incorporates all preceding claims, including Claim 1. An example is Claim 6, which states: “A CVD single crystal diamond material according to any one of the preceding claims, which has a value of effective refractive index of 2.3964 to within an accuracy of +/-0.001” [emphasis added]. The second category of claims expressly incorporates a preceding claim, the latter of which incorporates Claim 1. An example is Claim 7, which states: “A CVD single crystal diamond material according to claim 6, which has a value of effective refractive index of 2.39695 to within an accuracy of +/-0.0005.” By referring back to Claim 6, Claim 7 likewise incorporates Claim 1. The third category consists only of Claim 72. It reads: “A CVD single crystal diamond material produced by a method according to any one of claims 62 to 71.” Claims 62 to 71 disclose a process that purportedly enables a PSA to produce a CVD diamond in one or more of Claims 1 to 61. By referencing all of these process claims, Claim 72 includes the range of diamonds defined under Claim 1.

Once a subsequent product claim incorporates Claim 1, it asserts a monopoly over a range of products. This is because any diamond within the class of products in Claim 1, which also fulfils the additional parameter expressed in the subsequent claim, will fall within that subsequent claim. It is irrelevant that the additional parameter in the subsequent claim is a specific or single value of a particular physical property (for instance, Claim 6, in so far as the additional parameter, “effective refractive index of 2.3964 within an accuracy of +/-0.001”, expresses a single value of an effective refractive index) or a binary condition that is or is not satisfied (for instance, Claim 50, which has an additional requirement that the CVD diamond be formed into a polished gemstone). For reference, these claims state as follows:

In these instances, the PSA would be satisfied that the patentee contemplated multiple diamonds in the Claim 1 range fulfilling the additional parameter in Claims 6 and 50. This is because the claim incorporates the entire range of diamonds in Claim 1 and there is nothing in the language of Claims 6 and 50, or for that matter in any other similar claim, to justify limiting the monopoly to a single product. We approach this on the basis that the respondent should be afforded the full extent of the monopoly that is supported by the language it has adopted.

It is even more the case that where the additional parameter in the subsequent claim is itself expressed as a range of values, the claim covers a range of products. One example of this is Claim 4, which states: “A CVD single crystal diamond material according to any one of the preceding claims, wherein the modulus of the sine of the phase shift, |sin δ|, for at least 98% of the analysed area remains in first order and does not exceed 0.4” [emphasis added in bold italics]. The phrase “does not exceed 0.4” indicates that Claim 4 monopolises a class of CVD single crystal diamond materials which have different |sin δ| values falling within the range of less than or equal to 0.4 (and which fulfil any preceding claim). By way of example, a CVD single crystal diamond material for which |sin δ| is equal to 0.2 as well as a CVD single crystal diamond material for which |sin δ| is equal to 0.3 could fall within the scope of Claim 4, subject to the diamond material also satisfying a preceding claim.

We find support for our construction of the product claims in the court’s analysis in Anan Kasei Co. Ltd and another company v Neo Chemicals and Oxides Ltd (formerly Molycorp Chemicals and Oxides (Europe) Ltd) and another company [2019] EWCA Civ 1646 (“Anan”). The patent contained a product claim for ceric oxide, which stated as follows: “A ceric oxide consisting essentially of a ceric oxide, and wherein said ceric oxide has a specific surface area of not smaller than 30.0 m2/g when subjected to calcination at 900˚C for 5 hours.” (at [8]). One issue which arose was whether a claim limited by reference to a desirable physical characteristic, namely high specific surface area, and specifying that it remained the same after being subjected to a high temperature, insufficiently described the invention and was invalid. In analysing this issue, Floyd LJ (with whom Lewison and Peter Jackson LJJ agreed) noted that the claim covered a class of products (at [53]):

Regeneron (SC) is also illuminating. The patent in the suit contained a product claim which read as follows:

Lord Briggs summarised the claim in these terms (at [15]):

Relevantly, the appeal in Regeneron (SC) proceeded on the basis that the claim covered a range of transgenic mice, which differed based on the amount of human antibody genes implanted into the genome of the mouse (see [7], [16]–[17] and [81]). However, it bears highlighting that this issue of claim construction was resolved by the courts below and was not an issue in the appeal before the UK Supreme Court (“UKSC”).

In the present case, by defining parameters using ranges of values or incorporating claims which do so, each product claim in substance claims or asserts a monopoly over a range of single crystal CVD diamonds, each with a different combination of physical properties.

Before leaving this issue, we note that while Dr Newton’s evidence assisted us in construing the product claims, the expert evidence is not entirely clear as to whether each product claim covers a class of products or not. Be that as it may, claim construction is ultimately a task for the court. In particular, while expert evidence is useful in explaining the technical terms and technical features of the invention, the nature of the invention for which a patent is granted must be ascertained by the judge and not an expert (Brooks v Steele and Currie (1896) 13 RPC 46 at 73 and Dyson Appliances Ltd v Hoover Ltd [2002] RPC 22 at [13], cited in Terrell at paras 9-182 and 9-184). For the reasons given above, we hold that each of the product claims in SG 872 covers a range of products.

Insufficiency

The law of insufficiency

The sufficiency or enabling disclosure requirement finds statutory expression in ss 25(4) and 80(1)(c) of the Patents Act (2005 Rev Ed). Section 25(4) of the Patents Act (2005 Rev Ed), in the context of prescribing the requirements of a patent application, states that:

Non-compliance with s 25(4) of the Patents Act (2005 Rev Ed) is a ground for revoking the patent, as provided for in s 80(1)(c) of the Patents Act (2005 Rev Ed):

It is this requirement that compels the inventor to tell the world how his invention works so that, after the expiry of the patent when his invention falls into the public domain, a PSA will have sufficient information to work the invention and build on it (see Ng-Loy at paras 29.1.3 and 30.3.1).

English case law may be instructive when interpreting the scope of the sufficiency requirement under our Patents Act (2005 Rev Ed). Section 25(4) of the Patents Act (2005 Rev Ed) is in materially similar terms as s 14(3) of the UK Patents Act 1977 (c 37) (“UK Patents Act 1977”), which reads:

Section 80(1)(c) of the Patents Act (2005 Rev Ed) is also materially similar to s 72(1)(c) of the UK Patents Act 1977, which states:

Although the word “enough” is omitted in ss 25(4) and 80(1)(c) of the Patents Act (2005 Rev Ed) (and, for that matter, the same provisions in the Patents Act 1994 (2020 Rev Ed)), the sufficiency requirement in Singapore is not stricter. Our courts have held that a patent specification suffices if it is “clear enough” and “complete enough”; absolute clarity and completeness are not uncompromisingly required. This is because under ss 25(4) and 80(1)(c) of the Patents Act (2005 Rev Ed), as long as the specification is sufficiently clear “for [the invention] to be performed by a person skilled in the art”, it does not matter that the specification does not state every single step required for performance. After all, the PSA does not have to be told what is self-evident or what is part of his common general knowledge (First Currency Choice at [73]; Ng Kok Cheng at [47] and [49]). In this regard, the local position is in line with the ss 14(3) and 72(1)(c) of the UK Patents Act 1977.

That said, not every aspect of English law on insufficiency is applicable in our context. The UK Patents Act 1977 was passed to give effect to the Convention on the Grant of European Patents, 5 October 1973, 1065 UNTS 199 (“European Patent Convention”). Section 130(7) of the UK Patents Act 1977 provides that certain provisions, including ss 14(3) and 72(1)(c) of the UK Patents Act 1977, were “so framed as to have, as nearly as practicable, the same effects in the United Kingdom as the corresponding provisions of the European Patent Convention”. The English courts have thus approached the law in these respects with the objective of striving for consistency between European and English patent law (Regeneron (SC) at [30]; Generics (UK) Ltd and others v H Lundbeck A/S [2009] 2 All ER 955 (“Generics v Lundbeck (HL)”) at [86] per Lord Neuberger; Merrell Dow Pharmaceuticals Inc and another v H N Norton & Co Ltd and another; Merrell Dow Pharmaceuticals Inc and another v Penn Pharmaceuticals Ltd and another (1997) 33 BMLR 201 at 205). In interpreting and developing the jurisprudence under ss 14(3) and 72(1)(c) of the UK Patents Act 1977, the English courts have therefore introduced certain concepts into English law from European patent law to give effect to this harmonising objective. Such concepts, however, are not necessarily applicable to ss 80(1)(c) and 25(4) of the Patents Act (2005 Rev Ed), and their applicability falls to be determined with reference to the text and legislative objective of ss 80(1)(c) and 25(4) as well as the Patents Act (2005 Rev Ed) as a whole.

We begin by noting the policy that undergirds the sufficiency requirement. Both English and local case law have recognised that the sufficiency requirement lies at the heart of what is sometimes called the “patent bargain”. As already alluded to above at [57], the inventor/patentee is rewarded with a limited-term monopoly over the claimed invention, in exchange for enabling a PSA to work the claimed invention and dedicating it to public use after the monopoly has expired (Rohm and Haas Electronic Materials CMP Holdings, Inc (formerly known as Rodel Holdings, Inc) v NexPlanar Corp and another [2018] 5 SLR 180 (“Rohm”) at [161]; Warner-Lambert Company LLC v Generics (UK) Ltd t/a Mylan and another [2019] 3 All ER 95 at [17]). The patent bargain breaks down if the patent does not sufficiently teach a PSA how to perform the invention, in which case it makes little sense to reward the inventor/patentee with a monopoly. It is for this reason that the failure to satisfy the sufficiency requirement is a ground for patent revocation.

We now turn to set out general principles applicable to the legal requirement of sufficiency in Singapore. References to English jurisprudence will be made where appropriate.

It is well-settled under Singapore law that the burden of proving insufficiency rests on the party challenging the validity of a registered patent (Ng Kok Cheng at [48]).

The assessment of whether the legal requirement of sufficiency is met proceeds in two steps. The first step involves identifying the invention and deciding what it claims to enable the PSA to do. The second steps asks whether the specification enables him to do it (Kirin-Amgen Inc and others v Hoechst Marion Roussel Ltd and others; Hoechst Marion Roussel Ltd and others v Kirin-Amgen and others [2005] RPC 169 (“Kirin-Amgen”) at [103] per Lord Hoffmann, cited with approval in First Currency Choice at [61]). Indeed, the court can only ascertain whether the disclosure has been sufficient after ascertaining what needs to be disclosed (meaning the invention).

Both steps of the inquiry require the court to don the mantle of a PSA possessing common general knowledge of the art. The first step involves the construction of the patent claims from the perspective of the PSA. Claim construction principles as set above at [79]–[80] are applied (see Towa Corp v ASM Technology Singapore Pte Ltd and another [2017] 3 SLR 771 at [66] and Main-Line Corporate Holdings Ltd v United Overseas Bank Ltd and another (First Currency Choice Pte Ltd, third party) [2007] 1 SLR(R) 1021 at [72]). At the second step, the PSA uses his common general knowledge to supplement the information contained in the specification in order to perform the invention (see Genelabs (CA) at [61]–[63]). The sufficiency inquiry at the second stage is undertaken with reference to the date of filing of the patent application (Biogen Inc v Medeva PLC [1997] RPC 1 (“Biogen”) at 54 per Lord Hoffmann, followed in Kirin-Amgen Inc v Transkaryotic Therapies Inc [2003] RPC 3 at 70). This is sensible, since s 25(4) of the Patents Act (2005 Rev Ed) imposes the requirement of sufficiency at the time the patent application is made (see also Genelabs (HC) at [202]; Susanna Leong at para 16.266). As an aside, since we are regarding the relevant date for construing the patent claims (see at [78] above) and assessing insufficiency as the date of filing of SG 872, and we do not intend to discuss the issues of novelty and obviousness, we need not rule on the respondent’s claim for priority from GB 261.

Where the invention is a process, enablement at the second step requires that process to be carried out by the PSA; where the invention is a product, enablement requires the PSA to be able to make that product (Generics v Lundbeck (HL) at [20]; Roughton, Johnson & Cook, The Modern Law of Patents (LexisNexis, 5th Ed, 2022) (“The Modern Law of Patents”) at para 4.37).

Regardless of the nature of the invention, the patent specification must enable the invention to be performed by the PSA over the full breadth of the monopoly claimed, as we explain further below. Lord Hoffmann, who delivered the leading decision in the House of Lords in Biogen, observed that this is a long-established principle in English law (at 48) (see also Idenix Pharmaceuticals, Inc v Gilead Sciences, Inc and others [2014] EWHC 3916 (Pat) (“Idenix (HC)”) at [468]; Regeneron (SC) at [3] and [80]). Later in Kirin-Amgen at [102], Lord Hoffmann expressly applied this principle to s 72(1)(c) of the UK Patents Act 1977, which embodies the concept of sufficiency in the context of patent revocation:

This court in First Currency Choice at [61] cited Kirin-Amgen at [102] with approval when setting out the applicable principles in relation to s 80(1)(c) of the Patents Act (2005 Rev Ed).

To illustrate what it means for a specification to enable the invention to be performed to the full extent of the monopoly claimed, Lord Hoffmann gave the following illustrations in Biogen at 48:

Subsequently, in H Lundbeck A/S v Generics (UK) Ltd and others [2008] RPC 437 at [34], Lord Hoffmann re-iterated that where a claim is to a class of products, the class of products is enabled only if the PSA can work the invention in respect of all members of the class. The specification can show that this is empirically demonstrated or disclose a principle which can reasonably be expected to apply across the class.

However, the patent need not set out every detail necessary for the performance of the invention across its entire scope. As mentioned at [100] above, the patentee can leave the PSA to employ his skill and common general knowledge to work out what needs to be done. But the PSA must not be expected or required to exercise inventive ingenuity or exert undue effort in order to perform the invention across its entire breadth.

The requirement of enablement across the full breadth of the claim can be understood with reference to the patent bargain. The inventor/patentee should only be entitled to the full extent of the monopoly claimed if he has fulfilled his end of the bargain by disclosing the full scope of the invention clearly and completely enough. Only if this is done can the second aim of the patent regime, knowledge dissemination, be achieved by way of allowing the public to benefit from the invention after it falls into the public domain. A patent system which allows the inventor/patentee to monopolise more than that which he has sufficiently enabled may unduly stifle research, contrary to its primary aim of encouraging innovation (see Regeneron (SC) at [23]; Lionel Bently et al, Intellectual Property Law (Oxford University Press, 6th Ed, 2022) at pp 607–608). Hence, even where a patent specification is clear and complete enough to teach a PSA how to perform part of the claimed invention, that does not justify a monopoly over the entire breadth of the invention and the patent may be revoked pursuant to s 80(1)(c) of the Patents Act (2005 Rev Ed).

Before leaving this point, we note in passing that the majority of the UKSC in Regeneron (SC) held that it is enough if the patent enables the PSA to make “substantially all the types or embodiments of products within the scope of the claim” (at [56(iv)]) [emphasis added]. This threshold appears to be satisfied where there remain a “tiny or inconsequential number of embodiments which are not enabled” such that the scope of non-enablement is de minimis (at [36] and [56(v)]). The standard of substantiality was endorsed by the majority because it was the position in EU law (at [31]–[32] and [36]), and s 130(7) of the UK Patents Act 1977 requires interpretative consistency between European and UK patent law. We are not bound by a similar obligation. Without the benefit of submissions, we shall defer consideration of this aspect of Regeneron (SC) to an appropriate case. In any event, as we shall see, the threshold of substantiality does not have a material bearing on this case.

Next, although insufficiency is a single ground of objection to the validity of a patent contained in s 80(1)(c) of the Patents Act (2005 Rev Ed), there are distinct ways in which a patent may be insufficient (see Zipher at [362]; Anan at [22]). As will become clear, the appellant’s case engages different aspects of the rule on patent sufficiency, two of which are determinative of this appeal. We now turn to consider the different ways a patent may be insufficient under English and/or Singapore law.

The first way in which a patent may be insufficient is where the patent specification is not clear and complete enough to enable the PSA to perform the invention across the whole breadth of the claim(s) without an undue burden (see Terrell at para 13-31). This objection, termed “classical insufficiency”, is well-established in English and local jurisprudence (see First Currency Choice at [60]–[62]). In evaluating the merits of this allegation, the court has to assess the steps which would be necessary for the PSA to take under the patent’s specification to carry out the invention (see Zipher at [363]).

Under English law, a patent specification which enables a PSA to perform the full breadth of the claim without undue effort (meaning that it is not classically insufficient), may nonetheless suffer from another type of insufficiency, known as “Biogen insufficiency”. This form of insufficiency arises where the scope of the claim exceeds the technical contribution to the art made by the invention (see Terrell at para 13-09), whereas classical insufficiency is concerned with the mismatch between the scope of the claim and the method of performance taught in the patent specification. The “technical contribution to the art” does not refer to the technicalities of performing the claimed invention, but, more generally, to how the invention has in a practical sense added to or advanced the state of the art (that said, we note that the precise concept of “technical contribution to the art” has been articulated in various ways (see Generics v Lundbeck (HL) at [95] per Lord Neuberger, at [45] per Lord Mance and at [30] per Lord Walker)).

Consider, for example, a claim to “a heavier than air flying machine” in a patent which only discloses how to make an airplane. Assume for the moment that an airplane is a new and non-obvious invention, and the patent specification is clear and complete enough for a PSA to make an airplane. The airplane is a machine that is (a) capable of flight and (b) heavier than air, and by teaching the PSA how to make an airplane, the patent specification has enabled performance of every integer (or “element”, see Rohm at [103] and [179]) of the claim. In other words, the patent specification has sufficiently enabled the full breadth of the claim and can avoid a classical insufficiency attack. Yet, because the language of the claim has defined the invention in such general terms, the monopoly asserted by the claim is capable of extending to all other forms of heavier than air flying machines that are not airplanes and which may be manufactured by a wholly different process. The breadth of the claim thus exceeds what the patentee has contributed to the state of the art, in this example, how to make an airplane. This gives rise to Biogen insufficiency (see Anan at [52], citing Biogen at 52).

At present, Singapore patent law has not recognised the concept of Biogen insufficiency under s 80(1)(c) of the Patents Act (2005 Rev Ed), and it is unclear whether it should. We observe that Biogen insufficiency was first recognised by Lord Hoffmann in Biogen at 54, in the light of the jurisprudence under the European Patent Convention:

As the present appeal can be resolved without considering the applicability of Biogen insufficiency in Singapore, we leave this question open for determination in a future case.

Apart from classical insufficiency and “Biogen insufficiency”, there is a third way in which insufficiency may arise under English law, that is, where the PSA does not know how to determine whether a particular product or process is within or outside the scope of the claim, even after employing the common general knowledge and applying the normal process of claim construction (see Generics [UK] Ltd (t/a Mylan) v Yeda Research and Development Co Ltd and another [2012] EWHC 1848 (Pat) (“Generics v Yeda (HC)”), approved in Generics [UK] Ltd (t/a Mylan) v Yeda Research and Development Co Ltd and another [2013] EWCA Civ 925 (“Generics v Yeda (CA)”) at [78]; Unwired Planet International Ltd & Ors v Google Commerce Ltd (2016) [2016] EWHC 576 (Pat) (“Unwired”) at [163]). English courts previously labelled this type of insufficiency as “ambiguity” (see for example, Zipher at [374] and Unwired at [149]). However, Floyd LJ and Lewinson LJ in Anan (at [24]–[25] and [101]) considered this is a misnomer, because “ambiguity” usually refers to a situation where words in a claim are capable of more than one meaning, but this in itself does not render the claim invalid if the normal process of claim construction through the eyes of the PSA can resolve this issue. In place of the term “ambiguity”, Floyd LJ and Lewinson LJ held that the term “uncertainty” more accurately described this third type of insufficiency.

Uncertainty is a distinct objection from classical insufficiency. The latter is concerned with whether the patent specification sufficiently teaches the PSA how to obtain the product or work the process that is the subject of the claim, while the former is concerned with whether the PSA, after following the teachings in the patent specification, can tell whether the product obtained, or process worked, falls within the scope the claim. This distinction is illustrated by the facts of Kirin-Amgen, the first House of Lords decision to recognise uncertainty as a species of insufficiency. There, claim 19 of the patent was to a recombinant erythropoietin (“rEPO”) made by a specified process with, amongst other characteristics, a higher molecular weight than urinary erythropoietin (“uEPO”) (at [14]):

The difficulty with claim 19 lay in identifying the uEPO to test against the rEPO. Different uEPOs had different molecular weights and depending on the uEPO selected as the benchmark, a particular rEPO could either be within or outside the scope of the claim. As the specification did not disclose exactly which uEPO to use, it was impossible to determine whether a particular rEPO fell within claim 19 and the PSA was left guessing which uEPO the patentee had in mind. And, because different uEPOs had different molecular weights, the PSA would not know in advance whether any given uEPO would bring a particular rEPO within the claim. On this basis, the House of Lords, per Lord Hoffmann, invalidated claim 19 for insufficiency pursuant to s 72(1)(c) of the UK Patents Act 1977 (at [121]–[125], [129] and [131]). In arriving at this conclusion, Lord Hoffmann pointed out that the choice of uEPO has “nothing to do with making the invention work” (meaning that it has nothing to do with classical insufficiency) but relates to the criterion against which one tests whether a particular rEPO falls within the claims (at [129]).

In a sense, the objection of uncertainty in Kirin-Amgen arose from the application of a test prescribed by the patent. The prescribed test for ascertaining whether the claim boundary had been crossed was whether the molecular weight of a particular rEPO was higher than the molecular weight of uEPO. The uncertainty resided in how that prescribed test was to be applied, because the PSA did not know which uEPO to use as the benchmark. However, uncertainty in which test should be used to determine whether a particular product or process meets the characteristics specified in the claim can also sustain an objection of uncertainty in the context of insufficiency. This can be seen in Glaxo Group Ltd and other companies v Vectura Ltd [2018] EWHC 3414 (Pat) (“Glaxo Group”). The validity of five patents was in issue in that case, but for the sake of illustration, we focus only on claim 1 of one of these patents, European Patent (UK) No. 1 337 240 (“Patent 240”). Claim 1 of Patent 240 claimed a method of making composite active particles. The method involved the milling of particles of active material in the presence of particles of an additive material, such that the “particles of additive material become fused to the surface of the particles of active material” (at [86]). The patent contained very little guidance as to how the PSA was to determine whether the specified process had produced composite active particles with additive particles fused to the surface of the active particles (at [177]). The patent proprietor’s expert gave evidence that the PSA could carry out electron dispersive X-ray spectroscopy (“EDX”) to cross-check if this was the case, but Arnold J rejected this for two reasons. First, he found that a patent is insufficient if the technique for determining whether a particular product or process falls within the scope of the claims is not mentioned in the patent specification and is not part of the PSA’s common general knowledge. On the facts, EDX could not save the patent from insufficiency because the relevant test was not mentioned in the patent specification and was not part of the PSA’s common general knowledge even though it existed at the priority date (at [180]). Second, EDX had limitations which made it unsuitable for determining whether the product or process in question had the characteristic called for by the claim. The patent proprietor had also failed to validate the use of EDX for this purpose (at [181]). Arnold J therefore found that Patent 240 was invalid for insufficiency on the basis that it did not enable the PSA to determine whether a process or product fell within the claim (at [176] and [181]).

Whilst uncertainty as a type of insufficiency is well-established in English patent law, it has not been expressly recognised under Singapore patent law. The possibility of uncertainty giving rise to insufficiency was only contemplated in passing in First Currency Choice at [72], where we said:

This proposition rested on two authorities: Simon Thorley et al, Terrell on the Law of Patents (Sweet & Maxwell, 16th Ed, 2006) (“Terrell 16th”) at para 7-106 and Kirin-Amgen at [124]–[129], both of which recognise uncertainty (which is referred to as “ambiguity” in Terrell 16th at para 7-106) as an objection against the sufficiency of a patent.

On the facts, the appellant’s case relating to the Metripol Uncertainty Problem is in essence an objection of uncertainty. The Judge considered and rejected the appellant’s factual contention that the PSA in this case would not know how to tell when the δ value of the single crystal CVD diamond remains in the SG 872 First Order (Judgment at [200]–[206]), but did not comment on whether Singapore patent law does or should recognise uncertainty as a distinct type of patent insufficiency. On appeal, the appellant urges us to hold, as a matter of Singapore patent law, that a claim is insufficient if it is uncertain such that the PSA will not know whether he is inside or outside the claim.

In our judgment, where a PSA does not know how to determine if a particular product or process is within the scope of the claim even after employing his common general knowledge and the normal claim construction process, the sufficiency requirement in ss 25(4) and 80(1)(c) of the Patents Act (2005 Rev Ed) is not satisfied. We refer to this as uncertainty. Insufficiency remains a single ground of revocation embodied in s 80(1)(c) of the Patents Act (2005 Rev Ed), but it may arise in at least two distinct ways: classical insufficiency and uncertainty. We arrive at this view for three reasons.

First, as a matter of logic, a PSA can only be said to know how to perform the invention, if he knows what steps he needs to take to arrive within the claimed invention and knows how to determine if he has in fact obtained a product or worked a process within the scope of the claimed invention. It is meaningless for the PSA to know the former, meaning what steps he must take, if he does not also know the latter, meaning whether he has successfully taken those steps and made the invention. As Birss J said in Unwired at [159], the problem in Kirin-Amgen was that there was “a failure to disclose the invention clearly enough for it to be performed at all because the [PSA] could never know if they were within it or not” [emphasis added in bold italics].

Second, the sufficiency requirement seeks to ensure that others would be able to work the invention and benefit from it when it falls into the public domain following the expiry of the patent. But what benefit is there to the public if even the PSA does not know whether he is working the invention? In such circumstances, the inventor/patentee has not fulfilled his end of the patent bargain and ought to be denied a limited-term monopoly over the claimed invention.

Third, a patentee filing a patent application is in essence asking the State to grant it a property right. This much is recognised in s 41(1) of the Patents Act (2005 Rev Ed):

It therefore behoves the patentee to properly define when that property right has been infringed so that others would know how not to trespass. As Lewison LJ in Anan said at [99]:

Similar sentiments were echoed by the High Court in Rohm at [102]:

A patentee has not properly defined the extent of his property right if a PSA does not know how to ascertain when a particular product or process constitutes the subject matter over which the property right is claimed. Much uncertainty would be engendered if the law were to grant or recognise the asserted monopoly in such circumstances.

We turn to consider the respondent’s submission that the Judge, on the basis that lack of clarity under s 25(5)(b) of the Patents Act (2005 Rev Ed) was not a ground for revocation, rejected the appellant’s argument on insufficiency arising from “ambiguity”. The respondent further suggests that the acceptance of the appellant’s argument on insufficiency by uncertainty requires a departure from the position at law that the failure to comply with the clarity requirement is not a basis for revocation.

We disagree with the respondent’s interpretation of the Judge’s reasoning. The Judge was aware that the appellant’s arguments on the Metripol Uncertainty Problem went towards the issue of sufficiency instead of clarity (see Judgment at [192]–[197]), and she did not dismiss the appellant’s arguments on the basis that lack of clarity is not a ground for revocation.

More importantly, the respondent’s submissions raise the need to clarify the distinction between insufficiency arising from uncertainty for the purposes of ss 25(4) and 80(1)(c) of the Patents Act (2005 Rev Ed), and lack of clarity for the purpose of s 25(5)(b) of the Patents Act (2005 Rev Ed). This distinction is important, because the former is a ground for revoking a patent once granted while the latter is not (see Ng Kok Cheng at [74] and First Currency Choice at [72]). The test for clarity is whether the PSA can understand the words used in the claims. In this regard, there is no need to remove all conceivable doubt as to the meaning of the claims, but the claims must be as clear as the subject matter reasonably permits (see The Modern Law of Patents at para 4.133, citing Chevron Research Company’s Extension [1975] FSR 1 at 13 and LG Philips LCD v Tatung (UK) [2007] RPC 21 at [20]). A lack of clarity in the claim’s language, however, does not per se amount to uncertainty in the context of insufficiency (see The Modern Law of Patents at paras 4.58 and 4.132). To result in uncertainty giving rise to insufficiency pursuant to ss 25(4) and 80(1)(c) of the Patents Act (2005 Rev Ed), the lack of clarity in claim language must leave the PSA unclear as to how to determine whether a particular product or process is within the scope of the claim even after drawing upon his common general knowledge or applying the typical claim construction process (see First Currency Choice at [72]; Martek Biosciences Corp v Cargill International Trading Pte Ltd [2011] 4 SLR 429 at [72]; Anan at [24]–[25] and [101]). As we have explained above, lack of clarity under s 25(5)(b) of the Patents Act (2005 Rev Ed) is not the issue we are concerned with when dealing with the issue of uncertainty.

In the light of the foregoing discussion, we turn to consider the substantive arguments on sufficiency in the context of this case. But before doing so, we should add that while the preceding discussion (including that at [56]–[75]) referred to the 2005 Revised Edition of the Patents Act, on which the parties’ claims and counter-claims are based, these views apply equally to the 2020 Revised Edition of the Patents Act 1994 because the differences between the relevant provisions in these editions are not material to the points made.

Whether any or all of the claims in SG 872 are invalid due to classical insufficiency

Whether Claim 62 is classically insufficient

The appellant submits that Claim 62 presents the PSA with “a classic undue burden” as he is faced with a research project to find out what Other Growth Conditions are necessary to obtain a specific type of SG 872 Diamond. To evaluate the appellant’s submission, it is necessary to understand when a PSA is saddled with an “undue burden” to work the invention.

(1) What is an “undue burden”

To re-capitulate, the patent specification need not set out every detail necessary for performance, including what is self-evident or part of common general knowledge, and can leave the PSA to use his skill to perform the invention (First Currency Choice at [62]; Ng-Loy at para 30.3.4). However, the PSA must be able to perform the invention “without prolonged research, enquiry and experiment” (Halliburton Energy Services Inc v Smith International (North Sea) Ltd and others [2006] EWCA Civ 1715 at [13]; see also Susanna Leong at para 16.273). An oft-cited passage from the judgment of Aldous J in Mentor Corporation and another v Hollister Incorporated [1991] FSR 557 at 562 explains the point as follows (see Terrell at para 13-18):

As we noted in First Currency Choice at [60], “[i]t would not be desirable, and, indeed, probably quite impossible, to lay down any hard-and-fast rule” as to the required degree of clarity and completeness to satisfy the requirement for sufficiency of disclosure. The amount of teaching required in the specification may vary from invention to invention (Mentor Corporation and another v Hollister Inc [1993] RPC 7 at 11, cited in Susanna Leong at para 16.281).

Further, the examples in the patent specification may be taken into account when determining whether there is sufficiency. This is self-evident from the Patents Act (2005 Rev Ed) and Patents Rules. Section 80(1)(c) of the Patents Act (2005 Rev Ed) provides that the “specification” of the patent must “disclose the invention clearly and completely”. And s 25(3)(b) of the Patents Act (2005 Rev Ed), read with rr 19(3) and 19(5)(e) of the Patents Rules, shows that the specification must contain a description of the invention, which may itself include examples where appropriate. It was also recognised in Eli Lilly and Co v Human Genome Sciences Inc [2008] EWHC 1903 (Pat) at [239] that sufficiency must be assessed “on the basis of the specification as a whole including the description and the claims” [emphasis added] (see also Pacific Biosciences of California, Inc’s Applications BL O/500/18 at [43], cited in CIPA Guide to the Patents Act (Paul Cole & Richard Davis eds) (Sweet & Maxwell, 9th Ed, 2022) at para 14.28).

(2) Does Claim 62 impose an undue burden on the PSA trying to work it?

In light of the foregoing principles, the question before us is whether Claim 62 enables a PSA to perform the entire breadth of the claim without an undue burden. In summary, we consider that Claim 62 does impose an undue burden on the PSA because:

(a) The Other Growth Conditions do affect the quality of the CVD diamond produced.

(b) The specification of SG 872 (including the examples therein) merely provides a starting point for an onerous research programme:

(i) The patent specification does not teach the PSA how to determine where within a range of values he should operate, in respect of each of the Other Growth Conditions, so as to produce an SG 872 Diamond with a particular property or combination of properties.

(ii) Calibrating the value of each Other Growth Condition is complex because these conditions and the appropriate nitrogen concentration in the source gas are interrelated.

(iii) The CVD growth process and the resulting diamond are sensitive to changes in the Other Growth Conditions.

(iv) The authorities show that providing discrete examples of sets of values of the Other Growth Conditions, and the corresponding quality of the diamond produced, does not provide enabling disclosure across the entire breadth of the product claims, each of which asserts a monopoly over a range of products.

(c) Experimental data, while not adduced in this case, is not necessary to ground a classical insufficiency challenge.

(d) The respondent has not proved that the manner of calibrating the Other Growth Conditions in order to produce a CVD diamond of a particular quality forms part of the common general knowledge which need not be taught by SG 872.

We elaborate on these points.

(A) Other Growth Conditions affect the quality of the CVD diamond produced

Claim 62 states that the process taught therein enables the production of “a CVD single crystal diamond material meeting the requirements of one or more of claims 1 to 61” [emphasis added in bold italics]. To provide sufficient enablement across its entire breadth, Claim 62, read with the specification of SG 872, must teach the PSA to produce every diamond falling within each of the product claims in SG 872 (which themselves cover a range of diamonds), or any combination of the product claims.

According to the appellant, the gap in SG 872 is that it fails to specify the Other Growth Conditions that should be applied in order to produce a diamond meeting the requirements of a particular product claim or a combination of product claims. The implicit premise of the appellant’s argument is that the Other Growth Conditions do affect the quality of the CVD diamond produced, and this in turn affects whether the single crystal CVD diamond falls into one or more of the product claims in SG 872.

We deal first with this implicit premise. In our judgment, the fact that the Other Growth Conditions do affect the quality of the CVD diamond produced is evidenced by both parties’ experts.

The respondent’s expert, Dr Newton, accepted that “under high gas flow conditions the gas flow could have an effect on the chemistry of the [growth] process” [emphasis added]. As noted at [47] above, “gas flow” is one of the Other Growth Conditions. Dr Newton’s oral testimony, when asked about whether a PSA following SG 508 and SG 872 will end up with the same diamond, is particularly illuminating. He said that while the nitrogen ranges taught in the two patents overlap “it is not sufficient just to look at one of the parameters. You must look at the methane, the hydrogen, the flow rates, the temperature, the pressure. It’s a sensitive process. You start changing one of those and you go into a different regime of growth” [emphasis added]. Dr Bergonzo, another of the respondent’s experts, also testified that higher gas flows could reduce the effectiveness of in-situ etches in removing subsurface damage and thus reducing strain in the grown layers. On the appellant’s side, Dr Nebel similarly recognised that “[t]he gas flow around the substrate as well as the plasma exposure affect the growth of diamond significantly.”

Aside from the expert evidence, it is the respondent’s own submission that following the growth process in SG 508 will not inevitably lead to a diamond fulfilling the parameters of Claim 1iii) because of, among other things, differences between the Other Growth Conditions in SG 508 and SG 872. In other words the respondent accepts that the Other Growth Conditions do influence the quality of the diamond produced. For context, SG 508 teaches the conversion of a coloured single crystal CVD diamond to another colour under heat treatment (referred to as annealing). As part of its case that Examples 4 and 6 of SG 508 anticipate Claim 62, the appellant submits that the growth processes taught in SG 508 and Claim 62 overlap. A table produced by Dr Nebel, and adapted by the appellant, helpfully summarises the overlap (emphases by appellant omitted):

In an attempt to distinguish the growth process in SG 508 from that in SG 872, the respondent, as stated at the start of this paragraph, argued that while the nitrogen concentrations used in Examples 4 and 6 of SG 508 (and the nitrogen range in SG 508, generally) fall within or overlap with the Claim 62 Nitrogen Range, there are “a lot of differences in terms of gas flow, temperature and methane concentration” and that therefore, following Examples 4 and 6 will not inevitably lead to an SG 872 diamond. This clearly shows the materiality of the Other Growth Conditions.

We therefore accept that the Other Growth Conditions do affect the strain in the diamond that is eventually grown and that it is necessary for the PSA to determine the precise values of the Other Growth Conditions to use in order to grow a CVD diamond of a particular quality.

(B) Specification of SG 872 merely provides a starting point for an onerous research programme

We now turn to consider the guidance the patent specification gives in respect of the Other Growth Conditions, to determine whether the PSA is faced with an undue burden of ascertaining the precise values to use.

We begin with some general observations on the process claims and relevant portions of the patent specification. None of the process claims provides directions on the precise values the PSA should use for the Other Growth Conditions, so as to produce diamonds with particular characteristics satisfying one or more of the product claims. As we shall see, this conclusion remains unchanged even after the process claims are read in the context of the entirety of SG 872, in particular the examples.

The description in the patent specification indicates ranges of values for some of the Other Growth Conditions:

Regarding temperature, the specification of SG 872 provides no guidance on the appropriate atmospheric temperature or substrate temperature to use during the growth process. It merely notes that a process called “annealing”, in which elevated temperature is used in a controlled manner to bring about a beneficial modification to any property of diamond, may be combined with the “diamond of the invention” to enhance specific properties. Annealing takes place between 1200oC and 2800oC. However, this temperature range in relation to annealing say nothing about the temperature conditions that should be applied in the process claims of SG 872.

Additionally, the patent specification identifies one main set of specific values for the Other Growth Conditions in Example 1 (the “Example 1 Other Growth Conditions”). The same values are used in the other 14 examples in SG 872, subject to modifications to the gas pressure in Examples 9 and 14 and the concentration of methane in Example 14.

In Example 1, 1ppm nitrogen was used together with these Other Growth Conditions:

To summarise, the Example 1 Other Growth Conditions teach the use of: (a) a 2.45 GHz reactor; (b) gas flow of 15/75/600 sccm for O2/Ar/H2, and 36 sccm for CH4; (d) pressure of 263 x 102 Pa; and (e) substrate temperature of 730oC during in situ etching, and 800oC during growth. The parties also agree that Example 1 teaches the use of 5% of CH4 in the source gas. SG 872 does not expressly state which claim(s) the resulting diamond satisfies, although it appears to fulfil at least Claim 1v). This is because the diamond grown “had a FSR of 1.6678±2x10-4cm-1” and Claim 1v) covers diamonds with a FSR which varies by less than 5 x 10-3 cm-1.

Example 9 introduced a variation to the gas pressure. It shows that using 5ppm of nitrogen with – (a) the Example 1 Other Growth Conditions or (b) the Example 1 Other Growth Conditions, save that the gas pressure is 210 x 102 Pa – yields an SG 872 diamond. While not expressly stated, it appears that the diamonds grown under Example 9 satisfy Claims 16i) and 17 of SG 872. The optical scatter of the diamonds falls within the ranges defined in these claims.

Example 14 shows that using the Example 1 Other Growth Conditions with modifications to either gas pressure or methane – (a) 5ppm of nitrogen with 330 x 102 Pa gas pressure or (b) 2.5ppm nitrogen with 3.5% CH4 – yields an SG 872 Diamond. The diamond grown under (a) is referred to as “E14.1”, while the diamond grown under (b) is “E14.4”. While not expressly stated, it appears that the optical absorption of E14.1 and E14.4 falls within the range defined in Claim 19.

With this context in mind, we now arrive at the heart of the issue – whether the need to determine the Other Growth Conditions places an undue burden on the PSA to work the entire breadth of Claim 62. Having reviewed the evidence, we answer this question in the affirmative for the following reasons.

(I) Guidance in patent specification (including examples) is limited

First, the guidance on the Other Growth Conditions provided in SG 872 – both the ranges of values provided in the specification (see [146] above) and Examples 1, 9 and 14 – is inadequate.

With regard to the ranges of values, these only concern two variables: “gas pressure” and “plasma power density”. There is no such guidance on other variables such as gas flow, temperature and the concentration of the remaining gases in the source gas. And even, for the ranges provided, there is no teaching on how to determine where within the range the PSA should operate so as to produce a CVD diamond with a particular characteristic or combination of characteristics.

The respondent argues that Example 1, read with Example 9, gives “full disclosure of the reactor conditions such as temperature, pressure and gas flow rates”. We disagree. The Example 1 Other Growth Conditions, even when read with Example 9, only disclose the Other Growth Conditions to be used with 1ppm and 5ppm of nitrogen to produce three variants of an SG 872 diamond. These three variants are differentiated by the growth conditions used to produce them: (a) Example 1 Other Growth Conditions with 1ppm nitrogen (Example 1); (b) Example 1 Other Growth Conditions with 5ppm nitrogen (Example 9); and (c) Example 1 Other Growth Conditions, save for a gas pressure of 210 x 102 Pa, with 5ppm nitrogen (Example 9). These examples in no way enable the PSA to know what values of the Other Growth Conditions to use to produce diamonds falling within all of the 68 product claims in SG 872 either individually, or in any combination of the product claims. This deficiency remains even when Example 14 is taken into account. This example merely introduces two other variations to the Example 1 Other Growth Conditions – a gas pressure of 330 x 102 Pa with 5ppm nitrogen, and 3.5% CH4 with 2.5ppm nitrogen.

(II) Calibrating the Other Growth Conditions is a complex process

Second, and crucially, the determination of the specific values of the Other Growth Conditions to use in order to produce a diamond of a particular quality is a complex process. The Other Growth Conditions and the nitrogen concentration in the source gas are interrelated. The specification of SG 872 reveals that several of the Other Growth Conditions affect the appropriate concentration of nitrogen to be used. SG 872 pertinently states that the concentration of nitrogen to be used in the growth process is “a sensitive function of the growth conditions, including temperature and pressure” [emphasis added]. The specification adds that the limits to the concentration of nitrogen to be used “are process dependent, such that they may vary according to the process conditions used, including the actual gaseous source of N, and also the specific material properties required, and are best illustrated by way of example” [emphasis added in bold italics]. Yet, there is no exposition on the nature of the relationships between the Other Growth Conditions and the appropriate nitrogen concentration, or how these relationships should be taken into account when determining the values of the Other Growth Conditions. The specification also recognises that the relationships between nitrogen concentration and the Other Growth Conditions are “best illustrated by way of example”, but, in our judgment, fails to deliver on that promise (for reasons given at [154] above). Instead, the PSA is left to undertake extensive research to uncover the relationships between the variables in the growth process depending on the physical properties sought in the diamond grown.

(III) Growth process is sensitive to changes in the Other Growth Conditions

Third, to compound matters, the entire growth process and the resultant diamond material are sensitive to changes in any of the Other Growth Conditions. It is not as though the PSA can vary the Other Growth Conditions without any material consequences to the quality of the diamond produced. This point is elucidated in Dr Newton’s oral testimony. He stated that the growth process is “sensitive” in the sense that a change to one variable will result in a “different regime of growth” altogether (see above at [141]). Dr Newton’s evidence coheres with the specification of SG 872, which states that the appropriate concentration of nitrogen is a “sensitive function of the growth conditions” [emphasis added].

The upshot of all of this is that the PSA is unduly burdened with the need to experiment with innumerable combinations of Other Growth Conditions across the entire nitrogen range in Claim 62, so as to determine which combination will result in a diamond meeting the desired product claim(s). This is an onerous research programme because there are many values for each variable to experiment with, all while balancing the (undefined) relationships between the Other Growth Conditions and appropriate nitrogen concentration. Without any adequate teaching of a principle of general application or unifying characteristic that can guide the PSA to determine the appropriate value of each of the Other Growth Conditions, the PSA must resort to a prolonged trial and error experiment, fraught with uncertainty and unpredictability, using arbitrarily selected values for each variable in the growth process.

(IV) Authorities highlight the insufficiency of providing a starting point for an onerous research programme

Based on the foregoing, Examples 1, 9 and 14, as well as the ranges of values for gas pressure and plasma power density in the description, at best provide a starting point for an onerous research programme.

The case law holds that an undue burden is imposed if this is all that the patent’s specification affords.

In Bayer Schering Pharma/Reach-through claim [2009] OJ EPO 516 (“Bayer”), the patent in suit was for the use of compounds, which are capable of “stimulating the soluble guanylate cyclase independently of the heme group in the enzyme, to manufacture medicaments for the treatment of cardiovascular disorders”. The patent was held to be insufficient under Art 83 of the European Patent Convention because “compounds” was not defined by any chemical structure, but solely by the functional ability to “stimulate guanylate cyclase”. Article 83, similarly to s 25(4) of the Patents Act (2005 Rev Ed), requires the invention to be stated “in a manner sufficiently clear and complete for it to be carried out by a person skilled in the art.” The Technical Board of Appeal (“Board”) held that trial and error on “every conceivable chemical compound” for the claimed capability was an undue burden. It also noted that “the simple structural identification of one suitable compound class of general formula” would not help the PSA. By analogy, the provision by Example 1 in SG 872 of a baseline for a research project cannot cure Claim 62 of insufficiency. We therefore agree with the appellant that the examples in SG 872 “cannot support a claim of the breadth of claim 62”. It is useful to reproduce the relevant portion of the Board’s reasoning in full (at [5.2]):

Aldous LJ’s obiter dictum in American Home Products Corporation v Novartis Pharmaceuticals UK Ltd [2001] RPC 8 equally emphasises the inadequacy of supplying a starting point for a research programme (cited in Terrell at para 13-25). The patent in suit there was for the “[u]se of rapamycin for the preparation of a medicament for inhibiting organ or tissue transplant rejection in a mammal in need thereof” (at [5]). A competitor produced a derivative, which gave rise to the issue of whether the derivative fell within the scope of the claim. The English Court of Appeal held that the claim did not extend to derivatives but considered the position of insufficiency under s 72(1)(c) of the UK Patents Act 1977 in the event it did. Aldous LJ opined that the patent would be insufficient because, as the judge had found below, the number of possible derivatives was “vast” and many would not exhibit the required immunosuppressant activity. As a result, “whether any particular molecule derived from rapamycin would work at all was impossible to predict with certainty” (at [42]–[43]):

So too is the case of Amorphous silica/INEOS T 1743/06 (“Amorphous Silica”) on point. The patent there was found to be insufficient as the process claim for producing amorphous silicas according to the product claims involved variables which the PSA was not taught to calibrate. Similar to the present case, the product claims for “amorphous silica” were characterised by a series of physical properties. For reference, the process claim stated as follows:

The patent contained two examples of specific amorphous silicas possessing characteristics that fell within the ranges of values defined under the product claim (at [1.2]). In respect of the process claim, the Board noted as follows:

One specific issue which vexed the parties was the absence of any teaching on the appropriate stirring speed to use during the preparation of the silica. The patentee argued that the PSA could determine the appropriate stirring speed by varying it while reworking the two examples in the patent specification. However, the Board held that this did not overcome the lack of teaching for the stirring speed, and other process parameters, in respect of the other amorphous silicas that were not contemplated in the examples but fell within the range of products claimed:

As a result, the Board in Amorphous Silica upheld the revocation of the patent. This reinforces our view at [160] that just providing discrete sets of process parameters in examples without a general or unifying principle to guide the PSA in calibrating these parameters to produce a desired product may not sufficiently enable the invention across its entire breadth. Indeed, in Claim 62, given the complexities described at [155]–[156] above, the Example 1 Other Growth Conditions and the modifications in Examples 9 and 14, do not avoid the need for an unduly prolonged research project.

Finally, Saint-Gobain Adfors SAS (a company existing under the laws of France) v 3M Innovative Properties Co (a company existing under the laws of Delaware, United States) [2022] EWHC 1018 (Pat) (“Saint-Gobain”) underscores the importance of providing guidance on how to vary process parameters to achieve a specific product in the range of products claimed in the patent, in so far as how to calibrate the process parameters is not part of the common general knowledge. The patent contained a product claim for dish-shaped abrasive particles with, among other integers, a sloping sidewall and a specific thickness ratio. The latter was described as follows: “a thickness ratio of Tc/Ti for the dish-shaped abrasive particles (20) is between 1.25 and 5.00, wherein Tc is the thickness at a corner (30) of the sidewall (28) and Ti is the smallest thickness of the interior of the first face (24)” (at [93]) [emphasis added in italics]. Abrasive products made from abrasive particles are used to abrade, cut, grind, finish or polish a variety of materials (at [13]). The “recessed or concave” face of the invention was said to allow the claimed products to remove more material, when being applied as an abrasive particle, than a flat abrasive particle (see [59]). Critically, the court held that the patent had to enable the production of particles without undue burden across the whole range of thickness ratios defined in Claim 1, namely 1.25 to 5.00 (at [196]). This also strengthens our construction of the product claims in SG 872 (see [83] above). As to the point on insufficiency, the court held that the patent imposed an undue burden to produce particles across the entire scope of the product claim (at [233]). Part of the patent specification stated that it was possible to produce particles with thickness ratios between 1.55 to 2.32 (at [68] and [211]). But the court was not convinced that the PSA could produce particles at the upper end of the defined range of thickness ratios (that is, approaching 5.00) without an undue burden. The patent provided no indication of how to adjust relevant process parameters to vary the thickness ratio across the range of the claim (at [212] and [233]). For instance, one step in the production process involved drying a wet gel that had been prepared (at [36]). While the patentee’s expert argued that in order to control the Tc/Ti ratio, the PSA would increase the temperature and hence the “drying rate” of the wet gel, the court held that the PSA would not have thought to do so (at [217]). Claim 62 is likewise deficient for failing to provide adequate guidance on how to produce diamonds across the entire range of each product claim.

(C) Experimental data is unnecessary

In the present case, it is also no answer for the respondent to point to the appellant having failed to perform experiments to support its case on insufficiency. In Novartis AG v Johnson & Johnson Medical Ltd [2011] ECC 10, which concerned a patent for extended wear contact lenses, the English Court of Appeal upheld the lower court’s finding that the patent was insufficient because it did not teach which materials within the specific families and examples described were suitable for the production of ophthalmically compatible extended wear lenses. Neither did the patent enable the PSA to predict whether any lens is likely to be opthalmically compatible over a period of extended wear (at 195). Crucially, the court held that it was “irrelevant” that the challenger had not performed any experiments. As observed by Jacob LJ, the more important point was that (at 196):

In much the same way, we do not need experimental data to conclude that the patent tells the PSA nothing about how to obtain the appropriate combination of the Other Growth Conditions and nitrogen concentration to grow a single crystal CVD diamond material with characteristics satisfying one or more of the product claims. The PSA attempting to perform Claim 62 may by happenstance stumble on the appropriate combination and grow the low-strain diamond desired, but such a result is not due to the teaching in SG 872. It is clear from the face of SG 872’s specification that any success the PSA has in performing Claim 62 will be due to his own luck and efforts. In these circumstances, it would be unfair to allow a patentee to take credit for something the patent does not sufficiently enable and which requires the PSA to expend undue effort to uncover.

(D) Calibrating the Other Growth Conditions is not common general knowledge

We turn to consider whether the PSA, imputed with common general knowledge, would know how to calibrate all the Other Growth Conditions. In so doing, we will deal with two points that were raised by the respondent and the Judge respectively.

The respondent argues that the PSA “would know to adjust the regular reactor conditions such as temperature and gas flow rates”. Knowledge that the Other Growth Conditions do affect the quality of the diamond grown may have been part of the common general knowledge. But that is quite different from saying that how to calibrate all the Other Growth Conditions so as to produce an SG 872 Diamond of a specific quality was common general knowledge at the filing date. There was simply no evidence at all to support this and all we have said about the lack of sufficiency explains why this is untenable.

The Judge, however, found that Claim 62 is not classically insufficient because the PSA would know how to adjust the Other Growth Conditions (Judgment at [285]). With respect, we find that she erred. According to the Judge, Dr Bergonzo’s evidence was that “the missing details, such as the geometry of the substrate holder, would be within the knowledge of the PSA skilled with working knowledge of the research and development of CVD diamond synthesis” (Judgment at [285]).

The portion of Dr Bergonzo’s evidence the Judge appears to have relied on is found in Dr Bergonzo’s report dated 26 June 2019 (“Dr Bergonzo’s Reply Report”), which was written in response to Dr Nebel’s reports. To understand Dr Bergonzo’s evidence in its proper context, we need to summarise the relevant portion of Dr Nebel’s 1st Report, to which Dr Bergonzo was responding.

Dr Nebel had stated that the PSA would know that the Other Growth Conditions which affect the appropriate concentration of nitrogen include the: “Geometry of reaction chamber, Pressure, Temperature, Gas flow rates, Composition of Source Gas, Microwave Power and quality of substrate used.” He noted that Claim 62 “fails to provide accurate process conditions at which [300ppb to 5ppm nitrogen] would result in the intended objective of the method.” His views on the Examples bear setting out in full:

Dr Nebel also mentioned that in the absence of a specific value of microwave power, the PSA would not be able to achieve the plasma power density required. He added that the growth of the diamond is “strongly affected by the geometry of the molybdenum substrate holder used”, but SG 872 “doesn’t teach anything about the geometry of the substrate holder or the placement of seeds on the substrate holder.”

Dr Bergonzo’s response was that the PSA would know how to achieve “the high plasma density required” and “what substrate holder to use with his reactor”. It is immediately clear that Dr Bergonzo only addressed two of the Other Growth Conditions and omitted to explain how the PSA would know how to calibrate the remaining variables so as to produce an SG 872 Diamond of a specific quality. Therefore, even taking Dr Bergonzo’s evidence in respect of the geometry of the substrate holder, on which the Judge appeared to focus (Judgement at [285]), at face value, this is no answer to the difficulty the PSA faces in determining all of the remaining Other Growth Conditions.

As for Dr Bergonzo’s evidence on the appropriate plasma power density, he does not claim that the PSA would know how to calibrate this parameter depending on the quality of diamond desired. Dr Bergonzo simply testified that the PSA would combine his knowledge of the typical microwave power with the gas pressure used in Example 1 (263 x 102 Pa) to “give … the high plasma density required”:

But knowing how to achieve a “high plasma density” is not the same as knowing how to vary the plasma density to achieve a specific diamond in the range of diamonds covered by SG 872’s product claims.

Dr Bergonzo’s first report dated 13 May 2019 is also of no assistance to the respondent. There, Dr Bergonzo asserts that the “description of SG ‘872 provides ... [a] detailed summary of the substrate conditions used ... includ[ing] sizes, thickness, orientation, surface roughness, plasma etch conditions, duration and temperature, etc. This can be found, for instance, under Example 1 of SG ‘872. ... I am unable to identify any more points or parameters that needed to be, or should have been, disclosed to a PSA” [emphasis added in bold italics]. However, as we analysed at [152]–[168] above, the specification of SG 872 does not provide enabling disclosure across the whole breadth of Claim 62.

Read in its proper context, Dr Bergonzo’s evidence does not address the heart of Dr Nebel’s concern: that “The ‘872 Patent fails to provide any teaching on these other parameters and their relationship with the claimed properties of CVD diamond.” We agree with Dr Nebel. Dr Bergonzo also does not appear to expressly assert that a method to calibrate all the Other Growth Conditions was within the PSA’s common general knowledge at the filing date. In so far as he does, he and the respondent have failed to back this up with evidence to show that this was “generally known and generally regarded as a good basis for further action by the bulk of those who are engaged in the particular art” (see [74] above) and we reject any such contention.

Therefore, having regard to the language of the claim itself, the relevant context in SG 872 and the expert evidence, we overturn the Judge’s finding that Claim 62 is not classically insufficient and hold it to be invalid in so far as the respondent’s claim for infringement is concerned.

We accept that a claim in a patent is not required to specifically describe “all possible ways in which the invention can be carried out” (Halsbury’s Singapore IP at para 160.369). The inventor cannot be expected to relieve the PSA from all obligation to take trouble in carrying into effect the description in the specification (First Currency Choice at [67]). Rather, what is lacking in Claim 62, and the specification of SG 872, is a teaching of a general principle that enables the PSA to (a) understand the relationship between the Other Growth Conditions and nitrogen concentration; and (b) determine how to calibrate the Other Growth Conditions to grow a diamond of a specific quality.

(E) The other process claims

In light of the foregoing, Claims 63 to 71, process claims which refer back to and narrow the process taught in Claim 62, are likewise classically insufficient. For example, Claim 63 monopolises “[a] method according to claim 62, wherein the synthesis atmosphere contains more than 500 ppb nitrogen” while Claim 67 monopolises “[a] method according to any of claims 62 to 66, wherein the density of defects is such that surface etch features related to defects is below 102/mm2.” All of these subsequent process claims require an understanding of how to calibrate the Other Growth Conditions in order to grow a diamond of a specific quality, but, like Claim 62, are utterly devoid of such guidance.

Whether the Product Claims are also classically insufficient

Since all of the process claims in SG 872 (meaning Claims 62 to 71) are classically insufficient, and since it is the respondent’s case that producing an SG 872 Diamond was not possible prior to the discovery of the process in Claim 62 (in particular, the Claim 62 Nitrogen Range) (see [39] above), the sufficiency of these product claims is called into question.

As a starting point, working a product claim entails making the product (see Regeneron (SC) at [23]). A product claim’s contribution to the art is “the ability of the skilled person to make the product itself ...” (Regeneron (SC) at [56(ii)]). The patentee cannot obtain a product monopoly without disclosing how to make the product because if it were otherwise, the public would get nothing of substance in return for the grant of monopoly and the patent bargain described at [57] and [102] would be not be satisfied (see Regeneron (SC) at [23]). And, a claim must enable the invention to be performed over its whole breadth (see [108]–[109] and [114] above; Saint-Gobain at [196]). We also explained at [83]–[93] that each product claim in SG 872 asserts a monopoly over a class of products. Hence, the question is whether the specification of SG 872 sufficiently enables the PSA to produce diamonds across the entire breadth of each product claim in SG 872.

Having found that all the process claims in SG 872 are classically insufficient, we are satisfied that all the product claims are invalid for the same reason. Claims 62 to 71 assert that the process taught therein enables the PSA to produce an SG 872 Diamond. Since these process claims are classically insufficient, the PSA is faced with an undue burden to grow diamonds across the entire range of products claimed in each product claim. This is not to say that the sufficiency or, more generally, the validity of a product claim is necessarily tied to that of a related process claim as a matter of law. This happens to be the case on the present facts because SG 872 does not teach an alternative method of producing an SG 872 Diamond besides the one found in Claims 62 to 71 (see also [39] above), and neither is it the respondent’s case that there is a method within the PSA’s common general knowledge to grow SG 872 Diamonds.

We return to our earlier reference to the substantiality threshold in Regeneron (SC) (see [112] above). In our view, even if all that is required is enablement substantially across the breadth of each product claim, that standard is not met. As we observed above at [154], the specification of SG 872 only teaches the production of five variants of an SG 872 Diamond. There are, however, innumerable combinations of values of Other Growth Conditions and nitrogen concentrations to use, and innumerable embodiments of an SG 872 Diamond disclosed in Claim 1 (which we take to be the broadest product claim in SG 872). Enabling five variants of an SG 872 Diamond is not substantial enablement.

Revocation

As all claims in SG 872 are invalid for classical insufficiency, we allow the appellant’s counterclaim and revoke SG 872 in its entirety under s 80(1)(c) of the Patents Act (2005 Rev Ed). Although this is sufficient to dispose of the appeal, we briefly consider one further difficulty.

Whether any or all of the claims in SG 872 are invalid due to uncertainty

Whether Claim 1 is uncertain

To recap, low birefringence is quantified in Claim 1ii) as a sample with (a) δ remaining in the SG 872 First Order; and (b) |sin δ| not exceeding 0.9. Low birefringence is quantified in Claim 1iii) as a sample with (a) δ remaining in the SG 872 First Order; and (b) Δn[average] not exceeding 1.5 x 10-4. The specification of SG 872 directs the PSA to use a Deltascan (which has been renamed as the Metripol since 2001) or a “similar instrument with similar resolution”. The question which arises for our consideration is whether the PSA would know what test he should apply to ascertain whether a particular single crystal CVD diamond satisfies the integer that δ is within the SG 872 First Order.

The respondent claims that there are three instruments which are similar to the Metripol – the Polscope, the Millipol and the Rotopol. We reject the respondent’s allegation that the Polscope is a similar instrument. According to Dr Geday, though the Polscope was available in 2002 and was considered a direct competitor to the Metripol, it was less precise and hence inferior to the Metripol. The respondent has not shown us any other material which contradicts Dr Geday’s evidence. In so far as the Millipol and the Rotopol are concerned, it is common ground that both these instruments are similar to the Metripol but were developed only by 2009. As such, they could not have formed part of the PSA’s common general knowledge at the relevant time and cannot be relied upon as enabling the PSA to test for the parameters in Claims 1ii) and 1iii). In these circumstances, the following analysis will focus only on the Metripol.

It is common ground between the parties that the Metripol, on its own, does not give the PSA data to determine whether the δ value of a single crystal CVD diamond material falls within the SG 872 First Order. This has been referred to as the “Metripol Uncertainty Problem”. The reason for this is that the Metripol does not calculate δ. The Metripol generates the value of |sin δ|, but each |sin δ| value corresponds to many δ values and only one of these δ values lies within the SG 872 First Order. A graphical representation of this can be found at [35] above. It is for this reason that the appellant submits that Claims 1ii) and 1iii), and SG 872 as a whole, is invalid for insufficiency by ambiguity, or, as we prefer to term it, uncertainty.

Against this, the respondent claims that Dr Kaminsky had testified that he “totally disagree[d]” that the Metripol Ambiguity Problem was “insoluble”. This is a mischaracterisation of Dr Kaminsky’s testimony. The cross-examination of Dr Kaminsky had proceeded on the basis that each “order” is a multiple of 2π, and “first order” was defined as δ not exceeding 2π. This is distinct from the SG 872 First Order which is specifically and more narrowly defined in terms of δ not exceeding . However, it was in the former context that Mr Alvin Yeo SC, counsel for the respondent, asked Dr Kaminsky if he thought that it was an “insoluble problem” to determine “which order” δ was in. Dr Kaminsky replied that he “totally disagree[d]” that it was an insoluble problem. Thus, when understood in context, Dr Kaminsky’s position, at its highest, is that it is possible to determine whether δ is between 0 and 2π. However, this does not necessarily mean that it is possible to determine whether δ falls within a more narrow range of values between 0 and (that is, the SG 872 First Order) (see above at [31]).

The respondent further submits that there are four solutions to the Metripol Uncertainty Problem which the PSA would know of. The first solution is to interpret the data obtained from the Metripol using the PSA’s knowledge of the nature and characteristics of single crystal CVD diamond material, with reference to Dr Glazer’s Gap Theory. If the PSA remained in doubt as to whether δ is within the SG 872 First Order, he could use a second method known as the Cross-Polariser Method. This involves the use of cross-polar microscopy and the Michel-Levy interference colour chart. The third entails the use of a standard compensating plate with white light. The fourth and final solution is to carry out the measurements using the Metripol at two or more wavelengths. The last two solutions will be collectively referred to as the “Glazer 1996 Solutions”, as they are found in “An Automatic Optical Imaging System for Birefringent Media” (1996) 452 Proc R Soc Lond A 2751 (“Glazer 1996”), a scientific paper referred to in SG 872.

In turn, the appellant disagrees that these solutions would enable the PSA to determine if δ is within the SG 872 First Order. We note in passing that although the appellant’s argument focuses on whether a particular solution is part of the PSA’s common general knowledge at the priority date of SG 872 (meaning 21 November 2002), it is not in dispute that there is no material difference in the common general knowledge of the PSA as at the filing date (20 November 2003) and the priority date (21 November 2002).

In our judgment, the appellant, by highlighting the Metripol Uncertainty Problem, has adduced sufficient evidence to suggest that the PSA would not have been able to determine if δ is within the SG 872 First Order for a particular single crystal CVD diamond. The evidential burden thus shifts to the respondent to show that the PSA would know of a workable solution to overcome this and would be able to ascertain whether δ is within the SG 872 First Order in so far as single crystal CVD diamonds are concerned. This means that the respondent must show that any of the four solutions set out at [192] above fulfil two cumulative criteria: (a) it is a workable solution for ascertaining whether δ is within the SG 872 First Order where single crystal CVD diamonds are concerned, and (b) it is a solution that would have been known to the PSA either because (i) it is taught in the patent or (ii) it is part of the common general knowledge (Glaxo Group at [180]–[181]). The four solutions raised by the respondent do not fulfil either or both of these criteria.

(1) The Gap Theory

We begin with the Gap Theory, which was accepted by the Judge as providing a solution to the Metripol Uncertainty Problem (Judgment at [200]–[204]).

The respondent, relying on the evidence of Dr Glazer and Dr Newton, submits that the PSA using the Gap Theory would be able to tell from the Metripol measurements whether the sample remained in the SG 872 First Order. According to the respondent and Dr Glazer, the Gap Theory works on the premise that the PSA would know the following (collectively, the “Premises”):

(a) A theoretically perfect diamond with no strain-causing defects is isotropic and would, if measured, show |sin δ|=0 and δ=0 (the “First Premise”).

(b) If the Metripol |sin δ| Map of the analysed area of a sample shows regions in which |sin δ| is close to 0 without significant spatial variations associated with defects causing strain, these regions correspond to δ being close to 0 (the “Second Premise”).

(c) Strain in the material varies continuously, and thus, δ and the |sin δ| measurements are continuous with no gaps (the “Third Premise”).

The respondent submits that with the knowledge of the Premises, the PSA, in respect of an analysed area without significant spatial variations associated with defects causing strain, would know that the analysed area remains in the SG 872 First Order (δ≤ ) if the |sin δ| values extend upwards from close to zero to a maximum value (|sin δ|max) that is not close to 1 (meaning that there is a gap between |sin δ|max and 1). Dr Glazer explained at the trial that this interpretation is based on the understanding that δ varies continuously. He said that it is not possible for there to be a gap between |sin δ|max and 1, and for there to be δ values outside the SG 872 First Order simultaneously. If there are δ values outside the SG 872 First Order, there must have been a continuous range of δ values which extends from 0 up to those values beyond the SG 872 First Order, in which case, there would be no gap between |sin δ|max and 1. The Gap Theory therefore determines whether δ remains in the SG 872 First Order by looking at whether there is a gap between |sin δ|max and 1. If this gap exists, the range of δ values must be within the SG 872 First Order; if this gap does not exist, the range of δ values must have extended beyond the SG 872 First Order.

Whether the Gap Theory is sound is a major point of dispute between the parties both at trial below and on appeal. Much of the dispute on this point turned on whether the Gap Theory has been disproved in the light of a small yellow streak in a cross-polarised image of Sample 2. However, it is unnecessary for us to resolve this factual dispute because even if we assume in the respondent’s favour that the Gap Theory is sound, the respondent has not proven that the Gap Theory would have been known to the PSA at the relevant time.

At the outset, a PSA reading the specification of SG 872 would not know of the Premises or know that he should use the Gap Theory to determine whether δ falls within the SG 872 First Order. At trial, Dr Glazer tried to rely on the following passage to claim that it teaches the PSA to use the Gap Theory:

The Gap Theory, including its Premises, must be part of the common general knowledge at the relevant date in order for it to be attributed to the PSA. In this regard, we agree with the appellant’s submission that the respondent has not established that the Gap Theory was part of the PSA’s common general knowledge. This evidential gap is fatal to the respondent’s case, given that it is the respondent which bears the evidential burden to show that the PSA would know, as part of his common general knowledge, a workable solution to determine if δ is within the SG 872 First Order (see above at [194]).

Dr Glazer and Dr Newton gave evidence that the PSA would know how to interpret the Metripol data with reference to the Gap Theory, but neither expert cited any publications, released at or before the relevant date, which referred to the Gap Theory. Proof that scientific publications have described the Gap Theory as a solution to the Metripol Uncertainty Problem does not necessarily demonstrate that the Gap Theory was part of the common general knowledge, but it would, at the very least, have been prima facie evidence that it was (see above at [74]). Indeed, while the First Premise can be inferred from what the parties agreed to be the common general knowledge (see above at [24]–[27]), the same cannot be said for the Second and Third Premises. Yet, the respondent, Dr Glazer and Dr Newton did not refer us to any publications which would establish these two premises as part of the common general knowledge.

To bolster the Third Premise, the respondent relies on a histogram prepared for each Sample. Each histogram graphically depicts the number of pixels which have a particular |sin δ| value as measured by the Metripol. These histograms were set out in Dr Newton’s first report dated 13 May 2019 (“Dr Newton’s 1st Report”), wherein he sought to prove that the Samples infringed Claims 1ii) and 1iii). We set out the histogram for Sample 3 for illustration:

According to the respondent, these histograms demonstrate that the continuity of strain in single crystal CVD diamonds results in |sin δ| measurements which are continuous. We find that these histograms do little to advance the respondent’s case. These histograms do not show the more crucial point that δ varies continuously with no gaps (see above at [197]). More importantly, neither Dr Glazer nor Dr Newton relied on such a histogram in their expert reports to show that the PSA knew that |sin δ| measurements of a single crystal CVD diamond would vary continuously in value because of the continuity of strain in the diamond. After all, these histograms are only prepared in respect of three samples and cannot be representative of a wider trend, or proof that this fact was well accepted by those skilled in the art.

The respondent notes that Dr Kaminsky did not challenge Dr Glazer’s and Dr Newton’s evidence that the PSA would be aware of the Gap Theory and its Premises. In our judgment, however, that alone cannot suffice to establish that the Gap Theory and its Premises were common general knowledge at the relevant time because no supporting material was cited. Dr Glazer’s and Dr Newton’s evidence, whilst unchallenged, remain bare assertions and do not justify a finding of fact that the Gap Theory and its Premises were common general knowledge.

This difficulty is enhanced when we have regard to a point made by the appellant in seeking to undermine the credibility of Dr Glazer’s and Dr Newton’s assertion that the PSA would know of the Gap Theory and its Premises. The Gap Theory and its Premises were first raised in Dr Glazer’s and Dr Newton’s second reports, after Dr Newton had concluded in the infringement analysis of his 1st Report that each Sample had a range of δ values within the SG 872 First Order. As pointed out by the appellant, Dr Newton did not use the Gap Theory or any of its Premises in coming to his conclusion on infringement in his 1st Report. For each Sample, Dr Newton observed that “the Metripol images show regions that clearly have a low density of strain-inducing defects”, and supports this by simply pointing to a Metripol |sin δ| Map (in greyscale and colour) of the entire Sample which is created by taking overlapping Metripol |sin δ| Maps and stitching them together using a software programme. No further elaboration was given. In respect of each Sample, Dr Newton then proceeded to identify the |sin δ|max value of the analysed area, and computed one maximum δ value (δmax) from |sin δ|max. We emphasise the fact that Dr Newton only derived one δmax value. This is curious, since it is common ground that as a matter of arithmetic, each |sin δ| value corresponds to many δ values (see above at [35]). Dr Newton’s 1st Report did not explain why he was able to derive only one δmax value from the |sin δ|max value. What is even more notable is that the δmax value Dr Newton arrived at happened to be within the SG 872 First Order. From this, Dr Newton concluded that the range of δ values is confined within the SG 872 First Order. At no point in his 1st Report did Dr Newton expressly rely on the Gap Theory, or make any statement which shows that he deduced that the δmax value must have been within the SG 872 First Order because there is a gap between |sin δ|max and 1. If the PSA would have known of and deployed the Gap Theory (because the Gap Theory was part of the common general knowledge at the relevant time), one would have expected Dr Newton to use it in his infringement analysis to prove that the δmax value of each Sample must have been within the SG 872 First Order. This was not done. Instead, the Gap Theory was only raised in Dr Glazer’s and Dr Newton’s second expert reports, after the appellant raised the objection that the PSA would not know how to determine when δ was within the SG 872 First Order.

The foregoing is further reinforced by the fact that Dr Glazer, the main proponent of the Gap Theory, simply stated in his first expert report that he agreed with Dr Newton’s infringement analysis without seeking to plug the gap in Dr Newton’s infringement analysis by referring the court to the Gap Theory.

For these reasons, in our judgment, the respondent cannot rely on the Gap Theory to resolve the Metripol Uncertainty Problem. The Judge therefore erred when she accepted the Gap Theory without first considering whether it was part of the common general knowledge at the relevant time.

There is a further issue with the Gap Theory, even if we were to assume that it was part of the common general knowledge. On the respondent’s own case, the Gap Theory only works on the condition that the sample will show regions in which |sin δ| is close to zero without significant spatial variations associated with defects causing strain. The appellant points out that not all single crystal CVD diamond samples will have this characteristic. The respondent derides the appellant for not adducing expert evidence to make good the assertion that not all single crystal CVD diamond samples will fulfil that condition, but in our judgment, it is the respondent who has the evidential burden of showing that single crystal CVD diamond samples will generally fulfil that condition. As noted earlier at [194] above, the evidential burden has shifted to the respondent to show that there exists a workable test to determine whether δ falls within the SG 872 First Order. The respondent points to the Gap Theory, which works only when the sample shows regions in which |sin δ| is close to zero without significant spatial variations associated with defects causing strain. To discharge its evidential burden of showing a workable test, the respondent must at least show that most, if not all, single crystal CVD diamond samples would satisfy the condition upon which the Gap Theory works. In our judgment, the respondent has not done so. Accordingly, the respondent has not proven that the Gap Theory is a workable test such that a PSA, analysing any single crystal CVD diamond sample in his possession, would be able tell if it fulfils the integer relating to the SG 872 First Order.

(2) Cross-Polariser Method

The Cross-Polariser Method, which involves the use of cross-polar microscopy and the Michel-Levy interference colour chart, was common general knowledge at the relevant time. Cross-polar microscopy works by shining a beam of white light on two polarisers, which are placed perpendicular to one another. If a birefringent material is placed between the first polariser and the second polariser, the polarised light passing through the material is slowed down (or retarded). Depending on the extent of retardation, different wavelengths constituting white light (400nm to 700nm) may interfere and display colour when the resulting beam is imaged on a plane. The cross-polarised image of an isotropic crystal (one with no birefringence) will be completely dark, but the cross-polarised image of a highly anisotropic crystal (one with high birefringence) will be colourful. The colours in the image are then compared against the Michel-Levy interference colour chart to ascertain the birefringence of the material.

Figure 16 depicts how δ values (within the range of 0 to 2π) can be derived upon locating the colours in the cross-polarised image in the Michel-Levy interference colour chart:

Where a particular region of the cross-polarised image contains dark magenta, δ is close to 2π. If a particular region is completely dark, δ is at or extremely close to 0.

According to the respondent, if the cross-polarised image of the diamond sample only shows black, white or grey contrasts, it indicates that δ is within the SG 872 First Order.

We are unable to accept the Cross-Polariser Method as a workable test which sufficiently helps a PSA determine whether δ is within the SG 872 First Order.

First, the Cross-Polariser Method is an inherently imprecise method as it requires a qualitative judgment of the shade of colour perceived in the cross-polarised image. Dr Nebel accepted that if a particular region shows up as black in the cross-polarised image, a PSA will know that the region is strain-free or near perfect. Accordingly, the PSA will know that δ is within the SG 872 First Order for that particular region. However, Dr Nebel pointed out that the presence of any other colour indicates that a particular region has some birefringence, but the interpretation of that colour will be “subjective”. This is in line with the respondent’s own acknowledgement that the Cross-Polariser Method is only a “qualitative assessment” that gives the PSA an approximation of whether δ is within the SG 872 First Order.

Second, the parties agree that it is common general knowledge that even if the cross-polarised image shows black, white and grey contrasts, and an absence of other colours, this is not a conclusive indication that δ is less than or equal to , even though it can indicate that δ is less than or equal to 2π. In particular, it is common general knowledge that grey contrasts may be a result of δ exceeding . This is illustrated in Figure 16 above at [209].

The respondent points out that Dr Newton, in the infringement analysis of his 1st Report, had cross-checked the Metripol data against the Cross-Polariser Method to confirm that the range of δ values of the Samples were within the SG 872 First Order. However, a closer examination of Dr Newton’s analysis suggests that the Cross-Polariser Method had not conclusively indicated that the range of δ values of the Samples were in fact within the SG 872 First Order. In his 1st Report, Dr Newton observed that cross-polarised images of the Samples “show only black/white/grey contrast” with no other colour. On this basis, he ruled out δ values near π and concluded that this was “further evidence” that the range of δ values was within the SG 872 First Order (δ does not exceed ) for 100% of the analysed area. With respect, it is unclear how Dr Newton can reach this conclusion given that it is common general knowledge that the presence of grey contrasts may be because δ exceeds , and the ruling out of δ values near π does not indicate that δ must be less than Dr Newton added that the absence of other colour “is consistent with δ not exceeding ”. This is true, but the absence of other colour is also consistent with δ exceeding . As such, the cross-polarised images do not conclusively show that δ is less than or equal to .

Thus, even though the Cross-Polariser Method was part of the common general knowledge at the relevant time, it is generally an imprecise and inconclusive method. Admittedly, the Cross-Polariser Method can conclusively show that δ is at or extremely close to 0 if the particular region under analysis only shows the colour black, in which case the region is strain-free or near perfect and δ must be within the SG 872 First Order (δ does not exceed ). However, the reliability of the Cross-Polariser Method in this narrow situation cannot save it from being a generally unreliable solution.

The respondent is aware of the drawbacks of the Cross-Polariser Method, and has sought to present the Cross-Polariser Method as a “confirmatory step” which can be used to cross-check the Metripol data. It suggests that the PSA can use the Cross-Polariser Method to ascertain whether δ is approximately within the SG 872 First Order, before going on to determine from the quantitative Metripol data whether δ is actually within the SG 872 First Order. We do not see how this helps a PSA conclusively determine that δ is within the SG 872 First Order. The Metripol data itself does not tell the PSA whether δ is within the SG 872 First Order (see above at [190]), and it is a misnomer to term the Cross-Polariser Method as a confirmatory step when any “confirmation” is not conclusive. In so far as the respondent is saying that the Cross-Polariser Method, coupled with an analysis of the Metripol data using the Gap Theory, can enable the PSA to determine whether δ is within the SG 872 First Order, this is impermissible since we have held that the Gap Theory has not been shown to be a part of the common general knowledge at the relevant time (see above at [200]–[206]).

(3) Glazer 1996 Solutions

The patent specification of SG 872 directs the PSA to Glazer 1996 for “[a]n explanation of how the Deltascan works”:

As mentioned, the Deltascan was later renamed as the Metripol. The only paragraph in Glazer 1996 which sets out the Glazer 1996 Solutions reads:

Both parties did not submit on whether an external document referred to in a patent specification is incorporated into the patent specification and hence relevant to the assessment of sufficiency. However, even if we assume in the respondent’s favour that Glazer 1996 can be treated as a part of the patent specification and that the PSA knows of the Glazer 1996 Solutions as a result, the respondent has not shown that the Glazer 1996 Solutions are in fact workable tests for the purpose of determining whether a particular single crystal CVD diamond material has a δ value within the SG 872 First Order.

First, Glazer 1996 itself does not indicate that the Glazer 1996 Solutions apply to single crystal CVD diamonds.

Glazer 1996 did not expressly or implicitly contemplate the applicability of the Glazer 1996 Solutions to single crystal CVD diamonds. By way of background, Glazer 1996 is a scientific paper which sought to introduce the Metripol (which was at that time a new optical microscope-based imaging system) and demonstrate that it is “a potentially important research tool for the study of optical anisotropy in many disciplines” and “any situation where birefringence is of interest”. It set out a general discussion of how the Metripol works, its usefulness as well as its limitations, including its inability to tell how many “periods” of the sine function have been passed through in the absence of other information supplied. It was in this general context that Glazer 1996 proposed the Glazer 1996 Solutions. Glazer 1996 was not concerned with the specific application of this system to single crystal CVD diamonds, and in fact made no mention of this particular class of diamonds. Although Glazer 1996 briefly considered the application of the Metripol to a synthetic diamond sample for the purpose of illustrating the versatility and use of the Metripol, Dr Glazer, a co-author of Glazer 1996, clarified in his second expert report that the synthetic diamond sample under consideration was a HPHT synthetic diamond. He also testified that Glazer 1996 was not published with single crystal CVD diamonds in mind.

Second, Glazer 1996 proposed the Glazer 1996 Solutions as methods to ascertain “how many periods (known as the order) of the sine function have been passed through” [emphasis in original], but it is unclear whether the word “order” refers to multiples of . The respondent’s own expert, Dr Newton, points out that the word “order” can have different definitions depending on the specific optics field in question. The word “order” may, in the alternative, refer to multiples of 2π with the “first order” covering a range of values where δ does not exceed 2π , in which case the determination of how many “orders” of sine function have been passed through using the Glazer 1996 Solutions would not help the PSA conclusively ascertain if δ falls within the more limited range covered by the SG 872 First Order (δ does not exceed ).

And, apart from Glazer 1996,there is no indication from the body of expert evidence that the Glazer 1996 Solutions can determine whether the δ value of a single crystal CVD diamond is within the SG 872 First Order. The respondent’s experts did not expressly mention that the Glazer 1996 Solutions can achieve this. Neither did they apply the Glazer 1996 Solutions when seeking to prove that the Samples infringed Claim 1.

In particular, notwithstanding that Dr Glazer was a co-author of Glazer 1996, Dr Glazer’s own expert reports did not rely on the Glazer 1996 Solutions to solve the Metripol Uncertainty Problem, even though he was aware, by the time of his second expert report, that the Metripol Uncertainty Problem was the subject of contention between the parties. Dr Glazer’s second expert report only contained a vague comment that “the PSA could look to Glazer 1996 for the details of the optical path”. When asked during cross-examination why he did not mention that the Glazer 1996 Solutions could solve the Metripol Uncertainty Problem in his expert reports, Dr Glazer did not give a satisfactory response:

Dr Glazer’s response is at odds with his first expert report, which stated that he was requested to opine on the use of the Metripol as a method of measurement for birefringence as set out in Claims 1ii) and iii) of SG 872. The scope of his instructions clearly covered the limitations of the Metripol in ascertaining whether δ is within the SG 872 First Order and solutions to overcome them. Dr Glazer’s response also does not sit well with his second expert report, which stated that he was requested to respond to Dr Kaminsky’s first expert report, which, amongst other matters, highlighted the Metripol Uncertainty Problem. It is troubling that Dr Glazer did not expressly point to or elaborate on the use of the Glazer 1996 Solutions, notwithstanding that he was a co-author of Glazer 1996 and it was well within the scope of his instructions to raise these solutions.

We further note that Dr Glazer testified in re-examination that the use of a standard compensating plate with white light is a “standard way of doing things” which “goes back to the 19th century”, but he does not go so far as to say this is a “standard” means of ascertaining whether δ in single crystal CVD diamonds is less than or equal to .

In fact, a comment made by Dr Glazer at trial raises doubts as to the feasibility of using the Glazer 1996 Solutions on single crystal CVD diamonds to ascertain if δ is less than or equal to :

The appellant, relying on Dr Kaminsky’s evidence, argues that the Glazer 1996 Solutions cannot determine if δ is less than or equal to in highly heterogeneous diamond samples. The respondent in turn submits that this limitation does not apply to single crystal CVD diamonds, which are homogeneous along the light path. In support of this, the respondent points out that the Judge had accepted Dr Glazer’s evidence that birefringence in single crystal CVD diamond mainly results from strain associated with dislocations grown into the material with a line direction that is close to parallel to the growth direction and the strain fields tend to show little variation through the depth of a sample (see Judgment at [207]–[210]). We see no reason to disagree with the Judge’s acceptance of Dr Glazer’s evidence in so far as this specific factual point is concerned. However, even if we accept that the Judge’s findings meant that the limitation identified by Dr Kaminsky is inapplicable, it does not go towards discharging the respondent’s evidential burden of proving that the Glazer 1996 Solutions could be used on single crystal CVD diamonds. The inapplicability of one possible limitation does not necessarily mean that the Glazer 1996 Solutions can be used on single crystal CVD diamonds. Apart from submitting that the appellant has no scientific basis for its assertion that the Glazer 1996 Solutions are inapplicable to single crystal CVD diamonds, the respondent has not drawn our attention to evidence which positively proves that they are in fact applicable.

Given the dearth of evidence establishing that the Glazer 1996 Solutions can be used to determine whether δ is within the SG 872 First Order (δ does not exceed ) for single crystal CVD diamonds, the respondent cannot rely on the Glazer 1996 Solutions to stave off an insufficiency attack.

(4) Sub-conclusion: Claim 1 is uncertain

For these reasons, we find that the PSA trying to perform the claimed invention would not know which test to apply to ascertain whether the δ value of a particular single crystal CVD diamond material remained within the SG 872 First Order. Although this uncertainty only affects one out of two integers in limbs ii) and iii) of Claim 1 respectively, the result is that the PSA would not be able to determine whether a particular single crystal CVD diamond material satisfies either limb ii) or limb iii). This bears some similarity to the facts of Sandvik Intellectual Property AB v Kennametal UK Ltd and another [2011] EWHC 3311 (Pat), where the uncertainty affecting one out of five integers of the claim in that case likewise caused the entire claim to be held insufficient (at [83] and [162]–[165]).

It follows from this that the PSA is unable to perform the invention across the full breadth of the monopoly in Claim 1. As we have found at [83] above, the monopoly asserted in Claim 1 covers a range of single crystal CVD diamond materials, each with a different combination of the physical properties defined in the various limbs of Claim 1. This necessarily includes single crystal CVD diamond materials which satisfy either limb ii) or limb iii), or both. Since the PSA would not know if a particular single crystal CVD diamond material has satisfied limb ii) or limb iii), the PSA would not be able to make a portion of the range of single crystal CVD diamond materials that Claim 1 seeks to monopolise. We therefore find that Claim 1 is invalid for insufficiency.

For completeness, we mention two other matters. First, we note that the non-enablement of two out of the eight limbs in Claim 1 does not result in insufficiency of a de minimis nature, such that it can nonetheless be said that substantially all products within the scope of Claim 1 have been enabled.

Second, we recognise that our finding that the PSA could not determine whether limb ii) or limb iii) has been satisfied raises the question as to how the Judge was able to conclude that all the Samples satisfied both limbs and hence infringed Claim 1 of SG 872. In our judgment, the Judge erred in coming to this conclusion.

In respect of Sample 2, the Judge reached her finding of infringement on the basis that the Gap Theory was capable of addressing the Metripol Uncertainty Problem (Judgment at [431]–[432] and [202]–[204]). With respect, the Judge’s analysis on infringement overlooks the issue of whether the Gap Theory was in fact used to prove that the δ value of Sample 2 was within the SG 872 First Order. Once that is considered, it is apparent from the description of Dr Newton’s infringement analysis (see above at [204]) that the Gap Theory was not relied upon. In fact, the Gap Theory as described by Dr Newton and Dr Glazer works on the condition that the sample must show regions in which |sin δ| is close to zero without significant spatial variations associated with defects causing strain, but neither expert expressly stated that Sample 2 fulfilled this condition. Dr Newton stated in his 1st Report that “[t]he Metripol images show regions that clearly have a low density of strain-inducing defects”, but it is unclear if this satisfies the condition for the applicability of the Gap Theory. Further, the Judge did not evaluate the reasoning behind Dr Newton’s conclusion that the δ value of Sample 2 was within the SG 872 First Order. As we have indicated at [204] above, in our judgment, Dr Newton did not give a clear and satisfactory explanation of how he reached this conclusion. We therefore overturn the Judge’s finding that Sample 2 infringed Claim 1.

Apart from Sample 2, the Judge also found that Samples 3 and 4 infringed Claim 1 of SG 872 on the basis that they satisfied limbs ii) and iii) of Claim 1 (Judgment at [435]–[443]). With respect, the Judge arrived at this conclusion without considering whether the δ values of Samples 3 and 4 were within the SG 872 First Order. As with Sample 2, Dr Newton’s analysis on infringement did not rely on the Gap Theory, nor did it set out a satisfactory chain of reasoning proving that the δ value of Samples 3 and 4 were within the SG 872 First Order. Accordingly, we overturn the Judge’s findings that Samples 3 and 4 infringed Claim 1.

Whether the other claims in SG 872 are also uncertain

The appellant, both here and below, argued that if Claim 1 is found to be insufficient due to the Metripol Uncertainty Problem, SG 872 as a whole should be found invalid for insufficiency.

We agree with the appellant. The failure to enable part of the monopoly in Claim 1 resulted in the same in respect of all other claims in SG 872. This is because all other claims, be they product or process claims, have defined the scope of their monopoly by reference to Claim 1. We elaborate.

As we have noted at [87] above, each subsequent product claim in SG 872 relies on Claim 1 to demarcate the scope of its asserted monopoly. Each subsequent product claim covers any diamond within the class of products in Claim 1, which also fulfils the additional parameter embodied in the subsequent claim. This will include diamonds which possess the characteristics listed in limbs ii) and iii) of Claim 1. However, the making of these diamond materials is not enabled because the PSA cannot tell when limbs ii) and iii) of Claim 1 are satisfied. Hence, each subsequent product claim is not enabled across its full scope.

In respect of the process claims in SG 872, Claim 62 defines the claimed process as one which would produce a single crystal CVD diamond material meeting the requirements of any one of the product claims, including Claim 1. In other words, Claim 62 monopolises a process which produces a range of single crystal CVD diamond materials, including those which possess either or both characteristics in limbs ii) and iii) of Claim 1. Part of this monopoly is not enabled given that the PSA cannot tell when limbs ii) and iii) of Claim 1 are satisfied. The same issue arises in respect of each of the other process claims, which defines its monopoly by reference to Claim 62 such that it also covers a process which produces a range of single crystal CVD diamond material satisfying any one of the product claims.

For completeness, we mention in passing that even if all that is required for the sufficiency requirement is for there to be enablement of substantially all the subject matter covered by a claim, this threshold is not met in respect of the subsequent product claims and process claims as the non-enablement in each of these claims is not de minimis.

Revocation

As each claim in SG 872 has not been enabled across the full scope of its monopoly due to the uncertainty affecting limbs ii) and iii) of Claim 1, this is another ground for allowing the appellant’s counterclaim and revoking SG 872 in its entirety under s 80(1)(c) of the Patents Act (2005 Rev Ed).

Before concluding, we should add that it is unnecessary for us to rule on the correctness of the observation in Sunseap Group Pte Ltd v Sun Electric Pte Ltd [2019] 1 SLR 645 (“Sunseap”) that a patent should be revoked if “all the independent claims in a patent have found to be invalid” because “it follows that the dependant claims must also fall” (at [70]). While the appellant appears to rely on this portion of Sunseap, the outcome of this appeal does not turn on its application. We leave further consideration of this observation in Sunseap to an appropriate future case.

Conclusion