Daesang Corporation & Anor v Ajinomoto Co Inc

This case is concerned with the validity of a patent for a very simple, or apparently simple, invention. It involves the production on an industrial scale of a substance known as aspartame, a low calorie sweetener. The patent is European Patent EP 0 091 787 (“the patent”) and the proprietor is the defendant, Ajinomoto Co. Inc. (“Ajinomoto”). Its priority date is 12 April 1982. The claimant, Daesang Corporation (“Daesang”), seeks revocation on the grounds of anticipation, obviousness and insufficiency. Anticipation and obviousness are alleged over a single piece of prior art, Japanese Laid-open Patent Specification 167267 (“JP ‘267”). There is no counterclaim for infringement because the question of infringement in this country is being considered by the courts in the Netherlands in conjunction with the issue of infringement of the Dutch equivalent of the patent.

At the heart of the patent is a process of purification which involves crystallising aspartame from a concentrated solution. The patent claims that the product is easy to recover in high yields and high purity. Before considering the various issues in the case, some understanding of crystallisation will be useful.

Crystallisation is an important step in the production of many industrial chemicals. It is a way of separating a product in substantially pure form from an impure solution. It also has the advantage of delivering the product in solid form. Solids are generally concentrated and easy to package and transport.

A common experiment used to demonstrate the growth of crystals to school children, is as follows. Copper sulphate is dissolved in water. A hot strong solution is made. It is poured into a beaker. A single crystal of copper sulphate is then suspended in the solution. The beaker is left undisturbed on a shelf or in a cupboard and the solution is allowed to cool and evaporate a bit. After a few days it will be found that a larger crystal is now suspended in the solution. This demonstrates the growth of a crystal in a static solution.

There is a limit to the amount of a substance which can be dissolved in a solvent. For example if one takes ordinary cooking salt (sodium chloride) and stirs it into a pot of water, a point will come at which no more will dissolve. The solution is said to be saturated. Normally, it is possible to dissolve more by increasing the temperature of the solvent (i.e. the water). So, a litre of water at 30^oC will dissolve less than at, say, 70^oC. Figure 8 of the patent, which is reproduced as Illustration 1 below, illustrates the solubility characteristics of aspartame in water. It is a fairly conventional solubility curve. It shows that at about 70^oC, it is possible to make a solution which contains about 5.5% by weight of aspartame. At 20^oC, the solution can only contain about 0.6% by weight of aspartame. If you take a saturated solution at 70^oC and cool it to 20^oC, the amount of aspartame in solution must drop. The cooling solution which has “too much” aspartame in it is said to be supersaturated. The excess aspartame will come out as crystals. On the other hand, if additional crystals are put in a 1% solution at 20^oC, they will not dissolve because the solution is already saturated. But if they are put into a 1% solution at, say, 70^oC, it is far from being saturated and the added crystals will dissolve. At 70^oC an aqueous solution can contain nearly 56 gms of aspartame per litre, at 60^oC it can contain about 40 gms per litre and at 10^oC it can contain about 10 gms per litre. It follows that if a saturated solution of aspartame at 70^oC is cooled to 60^oC, about 16 gms of aspartame must come out of solution (i.e. 56 gms – 40 gms). If it were cooled from 70^oC to 10^oC, 46 gms would come out of solution leaving behind a saturated solution containing only 10 gms per litre of aspartame. At 10^oC, just under 20% of the original aspartame is left in the liquor. It would be wasteful and therefore expensive to throw this away. An efficient system would try to recover this valuable chemical from the liquor.

Basic concepts in the field of crystallisation are set out in the expert reports of Professor Garside, for Daesang, and Professor Hounslow, for Ajinomoto. Just before the trial each considered the other’s reports and indicated which paragraphs he agreed with. I suspect that there would have been even greater agreement had they been given more time to carry out this task. Nevertheless much was agreed. The following section of this overview is based on those agreed passages.

In any crystallisation process, two things happen. First the crystals must be born. Second they must grow. The first of these is called nucleation. It is the process of producing new centres of crystallisation (“nuclei”) upon which new crystalline material is deposited to form crystals. The more nuclei there are in a solution the more crystals there will be but these crystals will tend to be smaller than those formed in a similar solution but having a smaller number of nuclei. In the latter case, the amount of solid coming out of solution is spread amongst a smaller number of crystals. The growth rate is the rate at which the crystals increase in size. The kinetics leading to the processes of nucleation and growth are very different from each other and in developing a system to produce large crystals the expert is balancing the nucleation process with the growth process.

Nucleation can be split into two types, primary and secondary nucleation. Primary nucleation is the spontaneous formation of crystals unmediated by the presence of other crystals. Secondary nucleation is the process of formation of new crystals mediated by the presence of other crystals. In the latter process, small “seed” crystals are added to a system to form a foundation on which larger crystals can grow. This is what is happening in the child’s example set out above.

The driving force for both nucleation and growth is supersaturation. The solubility curve set out in paragraph 5 above, partitions the figure into two regions. Solutions having a composition and temperature above the curve are said to be supersaturated. Those below the curve are undersaturated. The further a system is away from the curve in the top left part of the figure the more supersaturated it is. The higher the degree of supersaturation, the higher the impetus for crystal growth and for nucleation. The difference between the amount of solute (e.g. aspartame) in a solution at a particular temperature and the amount it should have at that temperature if the solution is saturated is sometimes referred to as its  C . The larger the  C the greater the drive to crystallise. The higher the  C , the greater the impetus to crystal growth and the higher the rate of primary nucleation. The product will include a large population of small crystals. These crystals are often so small that the end result is termed a slime. This is what happens if one starts with a saturated solution and rapidly decreases its temperature. One ends up with a high degree of supersaturation (i.e. a high  C ). As Professor Garside explained, if a saturated solution is crash cooled (cooled as fast as possible) then a very large number of nuclei form. On the other hand, if the saturated solution is cooled slowly, the rate of nucleation is much lower and this encourages the growth of existing crystals rather than the formation of new nuclei. A smaller number of big crystals is produced. Once again the school example set out above illustrates the effect of slow cooling.

Supersaturation not only affects the rate of growth but also the quality of the crystals produced. Conditions of high supersaturation (i.e. high  C ) often produce crystals of a less desirable shape, for example with rough surfaces, holes, etc. These crystals are undesirable because they adsorb more solvent making solid/liquid separation more difficult.

Nucleation is extremely sensitive to the local environment over small distances. If a saturated solution is subjected to uneven cooling, different parts will be at different temperatures. In this state the solution is said not to be isothermal. Therefore different parts of the solution, even though they are close to each other, will experience different levels of supersaturation, and the rate of primary nucleation and crystal growth in different parts of the solution will be different. In those areas where solution has cooled a lot, the degree of supersaturation will be high and this will tend to produce a very large number of small crystals; the slime referred to above. On the other hand immediately adjacent to that part of the system may be a volume of solution which has only cooled a little bit. Here the degree of supersaturation will be small. This will tend to produce a small number of large, well-formed crystals. Overall, the system will contain a mixture of all the crystals produced in the various locations within it. This can result in crystals of poor quality - many small and some relatively large. In this case it is said that the crystals have a wide size distribution. This is to be contrasted with a well-controlled process which would produce crystals having a relatively narrow distribution of sizes, still showing some relatively smaller and some relatively larger crystals, but nothing like to the same degree as is found with poor control. The effect of uneven cooling of a solute can be illustrated by reference the same Figure 8 in the patent. It is set out below showing three different  C s produced if there is uneven cooling of a solution which is supersaturated at 70 ^o C.

The crystals produced in those parts of the system where the  C is 10 ^o C will be large but comparatively few in number. The crystals produced in those parts of the system where the  C is 60 ^o C will be very small but large in number.

The design of a crystallisation process usually involves attempting to make the initial temperature and concentration as high as possible (subject to constraints such as thermal stability of the molecules), and making the final temperature as low as possible (subject to operating constraints, such as the temperature of the available cooling medium and the freezing temperature of the solution). In this way the maximum change in concentration, and so the maximum amount of precipitation is assured. Therefore the maximum amount of desired product is recovered as crystals and the least amount of it is left in the cool solution. On the other hand one also wants to have crystals of good quality and of narrow size distribution. Therefore, although the temperatures at the start and finish of the crystallisation process can be far apart, the process will need to be controlled so as to obtain crystals of suitable size, shape and quality.

Finally, crystals in a supersaturated solution sometimes do not grow, or grow as large, as expected. The reason for this usually relates to the type and amount of impurities in the solution and/or crystals. It was well known in 1982 that the rate at which crystals grow can be strongly dependent on the presence of trace levels of impurity.

All of the above would have been known to the skilled worker in the field at the priority date of the patent.

The patent is concerned with the purification of aspartame (which it refers to as “APM”) by using a static system to produce large crystals. It explains that

It goes on to report that such agitated crystallisation produces a product which is unsatisfactory. It says that it consists of

Some of the difficulties in trying to recover these crystals are described. The solution to these problems and the heart of the alleged invention consists of making use of static crystallisation. In essence, the invention consists of taking a supersaturated solution of aspartame and cooling it without agitation so that crystals are produced which are easier to filter. If this process is used, a most unusual change takes place. If one considers a supersaturated solution of aspartame at, say, 70^oC and one cools it to, say, 40^oC, it will be seen from Illustration 1 above that the amount of aspartame in solution falls from 5.5% by weight to 2% by weight. That means that about 3.5% by weight turns into crystals. The solution which is left contains 98% by weight of water. Notwithstanding the fact that most of the system is still liquid, it sets into what the inventors call a “pseudo-solid”. This is described in the consistory clause and the immediately following passage in the patent:

The patent sets out one other feature of the process which, it claims, is unexpected. When operating on an industrial scale, one objective is to obtain the desired product as quickly as possible. For this and other reasons, there is forced cooling of the supersaturated solution. This involves the use of cooling surfaces supplied with coolant to reduce the temperature of the solution. Even though the contents of the crystalliser solidify, they do not stick to the cooling surfaces. Furthermore the pseudo-solid exhibits a property which Professor Hounslow described as astonishing, namely that when agitated, for example with a spatula, it turns back into a liquid. These characteristics are referred to in the patent:

The patent describes various types of equipment which can be used to put the process into effect. It also contains two examples to illustrate how the invention can be put into operation.

There is one other passage in the description which deserves mentioning in the light of one of the arguments advanced in this case. The patent includes what it describes as a “comparative example”. This involves running an aspartame purification process but using agitated crystallisation. The supersaturated solution is stirred while it is cooled from 25^oC to 10^oC. The product of this example is then compared with the products obtained by the patented static crystallisation process. Unsurprisingly, the purpose and effect of this example is to illustrate how the key step of static crystallisation is essential to secure the desired product.

Ajinomoto asserts independent validity for claims 1, 4, 5 and 6. Claim 1 reads as follows:

Mr Kitchin QC, who appears for Daesang, argues that some of the wording in this claim is open to criticism. In particular he says that the expressions “industrial scale”/”industrial crystalliser”, “sherbet-like pseudo-solid phase”, “without effecting forced flow” and “bundle-like crystal aggregates” are all ambiguous. However he accepts that, in the current state of the law, he cannot rely on ambiguity as a ground for invalidity. Nevertheless he wishes to keep the point open for consideration at a higher level. In addition, he says that these expressions, and in particular the reference to “bundle-like crystal aggregates” gives rise to an insufficiency attack. He does not put this at the forefront of his arguments but rather advances it as a squeeze against a too narrow construction of the prior art.

Although it will be necessary to consider the meaning of “industrial scale” below, it is convenient to say at this stage that there is nothing in the allegation of ambiguity. It is true that each of the expressions complained about is imprecise but there is no suggestion that anyone in the art would have any difficulty in understanding them, the specification or the claims. Daesang’s central argument is that, if one crystallises aspartame from a static aqueous solution, inevitably one will see the solution turn into a sorbet-like solid and that is because, inevitably, it contains crystals which the patentee describes as bundle-like aggregates. The end point is so clear that there can be no doubt if you reach it. In context, the limits of the claim are also clear.

There is also nothing in the insufficiency attack. The method for obtaining the pseudo-solid, which inevitably means obtaining the correct type of crystals, is simple in the extreme. Mr Kitchin says it amounts to nothing more than crystallisation without forced flow, that is to say static crystallisation. He says that if the system is static, the inevitable cooling must be by conduction. One will inevitably obtain the pseudo-solid. It must and can be broken up into a slurry for filtering and, as the patent states and is not in dispute, this is easy to do. Inevitably one has thereby obtained the type of crystals with which the patent is concerned. There is nothing in the drying step. I do not think any of this was in dispute. Implementing the process covered by the claim is, therefore, extremely simple and insufficiency is unarguable.

Furthermore, this analysis also throws light on the inventive concept in the claim. Most of the integers in it consist of no more than a description of the inevitable result of carrying out the crystallisation. Even the requirement that the precipitated solid phase formed after cooling is about 10g or more per litre of solvent adds nothing. It sets a limit on recovery from the crystallisation process which is so low as to be virtually meaningless. In the result I think that Mr Kitchin is right to say that the inventive concept in this claim amounts to no more than crystallizing aspartame from an aqueous solution without forced flow (i.e. statically) on an industrial scale. It is very simple.

Before turning to consider the issues of substance, namely anticipation and obviousness, it is convenient to consider the witnesses and the correct approach to the prior art. Only a small number of witnesses were cross examined. None of them struck me as being other than truthful. Mr Kitchin criticised Dr Kishimoto, a senior Ajinomoto employee, as being an advocate for his employer’s case. I think that is an unduly harsh criticism. He had been employed by Ajinomoto for a long time and, inevitably, he recounted the history of its discovery and exploitation of the patented process from the perspective of someone on the inside. I accept the accuracy of his evidence. Its relevance is another matter.

The major witnesses were the experts. Daesang called Professor John Garside. He is Principal and Vice Chancellor of the University of Manchester Institute of Science and Technology (“UMIST”). Prior to this he was Professor of Chemical Engineering at UMIST. He has had a long and distinguished academic career but he has also had commercial experience, more recently as a consultant, as well. Ajinomoto called Professor Michael J Hounslow, the Head of the Department of Chemical and Process Engineering at the University of Sheffield. He is younger than Professor Garside but also has had a distinguished academic career and has had considerable commercial experience as well. Both the expert witnesses were clearly honest and knowledgeable. If there is any criticism of them it is that they both were much more able and knowledgeable than the notional skilled addressee of the patent and prior art. The areas of dispute between them represented bona fide differences of opinion. Professor Hounslow was the more exuberant of the two. He appeared to enjoy the intellectual stimulus of being cross-examined and, at times, seemed concerned to work out the underlying purpose of Mr Kitchin’s questions. I think a little time would have been saved had he concentrated exclusively on answering the questions put to him rather than speculate on what was likely to come next. Mr Kitchin said he was lacking in objectivity. I do not agree. He is probably a good lecturer who can deliver his message with some dash. His evidence was given in this style. However I do not think it was in any respect untruthful or unfair. When he expressed disagreement with Mr Kitchin or Professor Garside, he did so courteously but quite forcefully. Professor Garside was quieter, but no less firm in his views. Both of them were well able to explain to me the technology which would have been known to the skilled addressee and both expressed views as to how that addressee might have understood the prior art. I did not feel that in giving evidence either of them stepped outside areas of technology familiar to them.

One of the major areas of dispute between the parties is as to the meaning of the only piece of prior art relied on, JP ‘267. In the end the meaning of that document is for the court. However the differences in interpretation voiced by the Professors reflects the submissions of the parties and, perhaps, throws light on its clarity or otherwise.

In relation to obviousness, courts have frequently warned against the dangers of hindsight. Once an invention has been made and understood it can be easy to arrive at it from the prior art by a series of logical and apparently obvious steps. The smaller the invention, the easier it is to analyse, explain and discredit it. However the misleading effect of hindsight is not limited to the issue of obviousness. When trying to understand the disclosure of a piece of prior art it is just as essential to view it through the eyes of the skilled addressee at the time without knowledge of the invention. After-acquired knowledge may result in the reader getting more out of a document than he would have had he read it at its date of publication or at the priority date of the patent.

Daesang alleges both anticipation and obviousness in the light of JP ‘267. In this judgment references to JP ‘267 are to the translation of it used at the trial. Daesang also pleads obviousness over common general knowledge. In opening, Mr Kitchin did not argue that case although he did not expressly abandon it. In reply, I understood him to put rather more emphasis upon this attack. As indicated above, it is necessary to determine what that prior art teaches without regard to the existence of Ajinomoto’s patent. For ease of reference, and because there are no line numbers on the pages of the translation, each of the extracts set out below will be numbered.

The area of technology at which JP ‘267 is directed is indicated by the opening paragraph of the section entitled “Detailed Description of the Invention”. It reads as follows:

The specification then refers to known methods of making aspartame and notes that during production certain groups of chemicals are always made and appear as impurities. It refers to certain known methods for removing these impurities but says that they are not satisfactory for various reasons. The objective of the patent is then stated:

I should mention, in passing, that none of the witnesses suggested that they had any difficulty in understanding what was meant by this statement and, in particular, the reference in it to “an industrial method”. The authors then describe briefly their investigations:

Extract 3 is significant. It concentrates on trying to use the anion-exchange resin to reduce the impurity level. It says nothing about the method of crystallization used. Indeed, its reference to crystallization appears to be deliberately vague, hence the use of the words “such methods as cooling crystallization” in the first paragraph. The results were unsatisfactory.

The passage also explains the relationship between the quality of the crystals produced and the recovery rate. As explained at paragraph 5 above, even if one crystallises from a very hot saturated solution by cooling to a low temperature, a considerable amount of the aspartame remains in the cool liquor at the end of the crystallisation process. Jettisoning that would be wasteful. The first paragraph in Extract 3 is directed to increasing the “recovery rate”, that is to say reducing the amount of aspartame wasted by the process. The suggestion in that paragraph is that the aspartame-containing liquor left after the crystallisation is concentrated and recycled. No doubt the impurities in the liquor will also be concentrated and recycled as well. Because of that, as the second paragraph explains, the yield and quality of the recovered crystals of aspartame falls. The third paragraph describes an attempt to minimise this problem, in other words to avoid excessive wastage of aspartame but, at the same time, not to cause too great a deterioration of the quality of the crystals recovered. This proposal involved only recycling part of the liquor. As the authors point out, to make this work, that is to say to maintain the quality of the crystals, it was necessary to jettison a “considerable amount” of the liquor and this pushed down the rate of recovery of aspartame to levels which were not industrially acceptable.

The solution to these difficulties is set out in the next paragraph as follows:

What the authors are doing is replacing the simple recycling of the liquor which was described in Extract 3 by a process which involves recovering the aspartame in it in a more purified form for recycling. This recovery procedure involves concentrating the liquor and crystallising out the aspartame (which should then be more pure than the liquor itself) followed by redissolving this aspartame and recycling it so that it is passed through the anion-exchange resin. It is apparent that the main purpose of this procedure is to try to leave most of the impurities behind, rather than recycling them.

It should be noticed that in this Extract, once again, no details are given of the crystallization process. On the contrary, essentially the same general words “crystallized by such methods as cooling” are used as in Extract 3. What is identified as being the reason for turning failure into success and which was only discovered after the claimed intensive investigation is the modified recycling procedure. There is nothing to suggest that the method of crystallisation was investigated, let alone identified as a significant factor. This passage is immediately followed by the consistory clause:

Once again, in this passage there is no suggestion that any particular type of crystallisation has been or has to be used. The difference between the purification system described in Extract 3 which is said not to work and the one in accordance with the consistory clause (Extract 5) which it is claimed does work can be represented in the form of the flow diagrams set out below in which the non-working system is on the left. The flow diagram on the right has been marked to show each of the four steps (A) to (D) referred to in Extract 5.

The specification then states that the invention will be explained “in detail” process by process. The description of step (A) takes up almost all of page 4 of the translation. The types of solvent which can be used are discussed as are the suitable anion-exchange resins. Ways of feeding the solution past the resin are also described, together with acceptable temperature ranges for that part of the process. This passage also considers the concentration of the α-APE solution, the amount of resin to be used and the residence time of the solution within the resin. In relation to the amount of resin to be used, the authors state that it must be enough to adsorb the impurities in the α-APE and, in relation to residence time, they say that it is to be sufficient to complete adsorption. The teaching of this part appears to concentrate on ensuring that the impurities in the solution are reduced to a minimum.

The description of process (B), that is to say the crystallisation stage, is very short. It reads, in its entirety, as follows:

There are also descriptions of stages (C) and (D) but they are not relevant to any of the issues in this case. The authors describe the benefits of their process as follows:

The document ends with two worked examples in which the process of the patent is compared with the unsuccessful recycling process described in Extract 3. These examples only involve the collection of a few tens of grams of α-APE. They are laboratory scale experiments, not industrial scale. They are about a thousandth of the scale of the examples in the patent in suit. Example 1 of JP ‘267 is the process which is said to produce the beneficial results. For present purposes it is only necessary to set out the description, again brief, of stage (B). It reads as follows:

The authors operated the cycle 5 times. They give results which show that the purity and yield of the α-APE recovered was as high at the beginning of those cycles as it was at the end. The crystallization yield was stated to be about 80%. As for purity, this was said to be 100% which, the authors explain, means that compounds other than α-APE were below the detection limit or could not be detected by high-speed liquid chromatography and thin-film chromatography. It also states:

The example of the non-working system, called the “Comparative Example” in the translation, is described on pages 6 and 7 of the specification. It is stated that stages (A) and (B) were carried out in the same way as in example 1. However, in this case the second purification stage was not used (see the flow diagram at paragraph 37 above). On the first run through, the crystallisation yield was 79.5%. The specification does not say anything about the size of the crystals at this point. 75% of the aspartame-containing liquor which is left after the extraction of the crystals is mixed with new aspartame-containing liquor and introduced to the anion-exchange resin. As explained in relation to Extract 4 above, this presumably is an attempt to reduce the quantity of impurities which are returned to the top of the anion-exchange resin. This recycling procedure is done five times. By the fifth cycle, the crystallization yield had dropped to 58%, the size of the purified α-APE crystals was 80 μm or less and impurities were detected by thin-film chromatography.

There is no dispute that the general teaching in JP ‘267 tells the reader nothing about how crystallisation is carried out. Daesang argues that the most relevant disclosure on this issue is Extract 8 above relating to the examples. This is said to disclose, at least by inference, that static crystallisation was used in that experiment. This is explained by Professor Garside as follows:

Based on this he says:

Professor Garside’s view that the document discloses static crystallisation is the opposite of the conclusion of Professor Hounslow. Among other things, the latter said:

Furthermore he said that, whatever may have been the process conditions employed for the examples, there is nothing to suggest that static crystallisation should be used if the purification system was to be applied on an industrial scale. This is not only referred to in paragraph 75 of his First Report which is quoted above, but in other passages as well:

The question of what JP ‘267 teaches can be split into two parts (i) what does it teach about the method of crystallisation used in the examples and (ii) what does it teach about the method of crystallisation which should be used in any industrial application?

The description of the examples in JP ‘267 must be read in the context of the document as a whole. There is nothing in the general teaching before the section dealing with the examples which suggests that the authors had any interest in different types of crystallisation. The references in the text to “crystallized by such methods as cooling” reinforces the point. From the evidence given before me, it appears that Professor Garside was not convinced that the authors knew much about crystallisation. Furthermore, the patent claim at the beginning of the document places no emphasis on the type of crystallisation employed. If the particular method of crystallisation had been important to securing the alleged better quality product of JP ‘267, it would be expected that the authors would have demonstrated that by running examples of the process with and without the relevant crystallisation method. This would have shown how failure to use the particular method resulted in a poor product. One of the notable features of the patent is that the authors do include comparative examples to highlight the factor which they assert changes failure into success. But that factor is not the type of crystallisation used. It is whether or not impurity-containing liquor from the output of the filter should be recycled to the input of the anion-exchange resin, as in the unsuccessful process, or cleaned of impurities by steps (C) and (D) in the successful process. As noted above, the crystallisation step at stage (B) – which is the relevant one for present purposes – is identical in the two examples.

In my view, the authors are concentrating on a new way of reducing impurities in order to obtain a better crystallised product at high yield. This is consistent with the common general knowledge referred to at paragraph 14 above that impurities can adversely effect crystallisation.

This is the context in which the teaching relating to the examples should be construed. The same indifference to the method of crystallisation is shown as in the rest of the teaching. One is left in the dark as to the precise method used. As Professor Hounslow explained, the authors could have been referring to one of a number of ways of carrying out the crystallisation. Professor Garside appears to have read the absence of reference to agitation or any other more sophisticated method of crystallisation as meaning that a simple static system was used. But the better view is that the description of the method of crystallisation is short, not because the method was simple, but because the authors were not interested in describing in detail a step in the process which was of little interest to them. I do not think that the skilled addressee who knew nothing of Ajinomoto’s patent, would have read this document as Professor Garside does.

The lack of teaching of the use of static crystallisation manifests itself in another way. For reasons set out below, even now it is impossible to be certain that the authors actually used static crystallisation. If they did not use it, it is hardly surprising they did not describe it.

The authors of JP ‘267 were aiming to make a pure product by isolating pure crystalline aspartame from an impurity-containing liquor. The crystals have to be recovered, for example by filtration. This can be seen in the flow diagram in paragraph 40 above. They say that small crystals do not work as well. They were seeking large crystals with low impurity levels. It is not in dispute that, if the authors had carried out the static crystallisation, they would have obtained the pseudo-solid described in the Ajinomoto patent. The filtration step described in relation to example 1 (see Extract 8 above) therefore would have involved filtering a solid mass or the liquifaction of that mass before filtration. None of this is described.

There was no dispute between the Professors that the creation of a pseudo-solid would have been a strange and unexpected event. In his First Report, Professor Hounslow said, in a passage agreed with by Professor Garside:

Professor Hounslow expanded upon this under cross-examination. He said that the pseudo-solid would have appeared very unusual to the skilled team in 1982. In a passage in his cross examination when he was responding to a suggestion that scaling up would not have required any special techniques he said:

Later he said:

No doubt Professor Hounslow’s language is more colourful than would be used by Professor Garside but I did not understand the underlying message to be in contention. The creation of a pseudo-solid would have been a surprise. Indeed it is apparent from the evidence in the trial that, on the small scale being used in the example in JP ‘267, the flask containing the aspartame would not just have turned solid, the contents would not fall out if you turned it upside down. The ease with which it could be returned to a liquid was unprecedented. None of this is referred to by the authors of JP ‘267.

The failure to report this is particularly noticeable in relation to the comparative examples. There the differences between the patented and non-patented process were being highlighted. The authors were paying particular attention to the structure of the recovered aspartame, hence the disclosure of the size of the crystals. This means that those crystals must have been subject to examination under a microscope. If Daesang is right and the authors of JP ‘267 used static crystallisation, the difference between the product of the two processes would have been impossible to miss. Furthermore, on this hypothesis, when the authors of the document refer to the second stage of the patented process and report that “the crystals precipitated were collected by filtration”, they ignored the fact that to do that they needed to liquefy a pseudo-solid.

Furthermore the way in which the authors describe the non-patented comparative example is particularly surprising. It will be recalled that the first run of that example produced gave a 79.5% crystallisation yield and, because for that run there was no inclusion of recycled, impurity-containing liquor, it would have produced good quality crystals. The purpose of the experiment is to demonstrate that continued recycling of the aspartame containing liquid which is left after the crystallisation stage – something which is necessary to avoid wastage of aspartame – causes the build up of impurities which affect crystal size. Therefore the purpose of the experiment is, amongst other things, to demonstrate the difference between the type of crystals obtained on the first run through with a clean new aspartame-containing solution and the type of crystals obtained once there had been repeated recycling of the liquor after filtration. Once again, if Daesang is correct and static crystallisation was used, the product of the first run must have been a pseudo-solid which would have needed to be liquefied in order to extract not only the crystals but to release back the aspartame-containing liquor for recycling. By comparison, the later runs which contained the increased levels of impurities would not have produced the pseudo-solid. This difference would have been unmissable. It is not referred to by the authors.

In my view it is by no means clear that the authors of JP ‘267 obtained a pseudo-solid in any of their experiments. Had they done so, the likelihood is that they would have reported that fact. Since Daesang’s case is that use of static crystallisation would inevitably have produced a pseudo-solid, and it ran experiments to prove as much, it must follow that it is not possible to be confident that the authors of JP ‘267 used static crystallisation. If they did not use it, it is hardly surprising that they did not describe it.

Even if, contrary to the above conclusions, JP ‘267 does describe laboratory scale experiments which included static crystallisation, does it describe or teach the use of that type of crystallisation on an industrial scale? It does not. On the contrary, as pointed out above, the general description and the claims in the patent pay no attention to the particular type of crystallisation method used. There is no suggestion that any particular type of crystallisation should be used if an attempt were to be made to carry out the recycling process on an industrial scale.

There is no dispute as to the law. Daesang must prove that JP ‘267 contains clear and unmistakable directions to carry out a process falling within the claims of Ajinomoto’s patent. The examples in JP ‘267 do not describe such a process, not least because it is carried out on a laboratory scale. Furthermore, for the reasons set out above, it does not contain clear and unmistakable directions to carry out the examples using static crystallisation. For the reasons set out in the last preceding paragraph, there is also no teaching of using such crystallisation in an industrial scale process. The allegation of anticipation fails accordingly.

This leaves the question of obviousness. There are two topics which should be considered first, namely the meaning of “on an industrial scale” and the relevant common general knowledge which the skilled person in the art would have had in 1982 as to the use and application of static and agitated crystallisation.

Mr Kitchin argues that this expression is imprecise, which it clearly is. He says that it is ambiguous but concedes that, on the law as it stands at the moment, that is not a basis for attacking the validity of the claims. In any event, the evidence does not support the suggestion that this term is ambiguous in the sense that the reader does not know what it means. On the contrary, although the exact boundary between industrial and non-industrial scale is unclear, none of the witnesses appeared to have any difficulty in understanding the term. For example Professor Garside refers throughout his report to “industrial scale” and “industrial”. He states that his own areas of expertise include teaching, speaking at conferences and writing articles about industrial crystallisation issues. His comment on this expression in the patent is as follows:

Neither he nor any other witness criticised the use of this expression. The use of the expression “industrial scale” serves to distinguish commercial production from the sort of activity which is carried on in a laboratory. Within reason, the scientist carrying out a laboratory experiment does not care how expensive the experiment is nor how long it takes to run it. His primary concern is to get the product he is after. Running a process on an industrial scale is quite different. Someone wishing to run a process on such a scale will want it to produce large quantities of product, speedily, at reduced unit cost and with uniform and predictable quality. No doubt each of these qualities is imprecise in the sense that they involve relative values such as “large”, “reduced” and so on. A weight of product which would be called large if the product were an ingredient for a pharmaceutical would, no doubt, be insignificant if it were steel ingots. This is why Professor Garside pointed out that 380 litres was on the industrial scale where the food or pharmaceuticals industries were concerned. For any particular product, those in the art would have little difficulty recognising an industrial scale. The borders of “industrial scale” may be imprecise, but that is all. In the end I did not understand Mr Kitchin to disagree with this analysis (Footnote: ¹). “Industrial scale” incorporates at least those requirements of bulk production, speed, reduced costs and uniform and predictable quality. Someone wishing to turn a laboratory scale process into an industrial one must address, or be hopeful that he will be able to address, all of those requirements.

Although there was a great deal of evidence given in relation to the state of knowledge of agitated and static crystallisation and as to what the skilled team’s approach to selecting a suitable crystallisation process for the industrial production of aspartame would have been in 1982, in the end there was much common ground.

As mentioned above, someone designing a plant for the industrial production of any chemical, including aspartame, would be attempting to recover the final product at speed, in bulk, at reduced cost and at uniform and predictable quality. The crystallisation process chosen would be designed to meet those requirements.

As mentioned above, at a late stage before the commencement of the trial, the two Professors read each other’s reports and marked up those passages with which they agreed. Some of those agreed passages are of particular importance to this issue. They include the following:

From Professor Garside’s Report:

From Professor Hounslow’s First Report:

Together these passages indicate that, subject to avoiding excessive agitation, an agitated crystallisation gives better crystals and greater uniformity of product. The reason for this can be explained by reference to the discussion of supersaturation set out towards the beginning of this judgment. Assume that one has decided to use cooling crystallisation to produce pure crystals and that the starting material is a saturated solution of a substance, such as aspartame, at 70^oC. Also assume that the objective is to get as high a yield of crystals as possible in as short a time as possible, while still getting a uniform product. The objective of high yield means that one will want to cool to as low a temperature as possible so that most of the target chemical comes out of solution. The objective of high speed will mean that the cooling has to be done as quickly as possible. Assume that, to meet these requirements an attempt is made to use static crystallisation. The solution is put in a crystalliser which is fitted with cooling surfaces maintained at, say, 10^oC. Because there is no agitation, the temperature of the charge in the crystalliser will vary from place to place. Those parts of the charge which are immediately in contact with the cooling surfaces will be rapidly cooled to 10^oC or a temperature close to it. Those parts which are furthest from such surfaces will be at a much higher temperature, up to 70^oC. Their cooling will be slower. The temperature of the charge will vary from location to location. The charge is not isothermal. As explained in paragraph 11 above, the result will be that there will be a wide distribution of crystal sizes because the rates of nucleation and crystal growth will vary from location to location within the charge. In an agitated system, variations in temperature of the charge are minimised. It is for this reason that Professor Hounslow said, and Professor Garside agreed, that the normal approach is to use programmed cooling so that the whole crystalliser will be reasonably isothermal, and that that “necessitates” agitation. It is also why Professor Garside said, and Professor Hounslow agreed, that the simplest way to keep the system uniform is to use agitation. The resulting product will have a narrow size distribution and, as Professor Garside said, this permits more efficient solid-liquid separation.

This explains why, on an industrial scale, a better quality crystalline product is obtained by using an agitated crystalliser. That this was, at the priority date, the thinking of those skilled in the art can hardly be denied. Reference can be made, for example, to a standard textbook; The Chemical Engineer's Handbook, Fifth 5th Edition, edited by Perry and Chilton, McGraw Hill, 1973 (“Perry”). Professor Hounslow described this as the chemical engineer’s Bible. Professor Garside agreed that it was often referred to in that way. He said that it would perhaps be the first book that a chemical engineer would look at if he was coming new to a particular process. Perry contains a section dealing with static crystallisation:

Professor Hounslow said in relation to these passages that the discussed difficulties inherent in using un-agitated vessels accorded with his own experience. Such systems were difficult to operate and produced a very irregular product. He confirmed that the commercial production of aspartame is not cost-effective on a “very small scale” nor would it be produced in “primitive areas by the application of copious cheap labour”. It therefore falls outside the group of specialized applications in which Perry says static crystallisation might be used notwithstanding its known disadvantages.

These passages were also put to Professor Garside in cross-examination. He said that what Perry said about static crystallisers was broadly true. He agreed that there are perceived to be a number of “serious disadvantages” with them and they certainly would not be the first choice for crystallization processes. They would not be the thing you would look at first. On the other hand he suggested that they would be “one of a suite of possibilities”. However the latter suggestion should be viewed in the light of other evidence he gave under cross-examination. For example he was asked how the practical engineer would view the options:

Professor Hounslow was firm on this subject. In his First Report he said that in relation to solution crystallisation, that is to say the sort of crystallisation at issue here, one would “invariably” choose an appropriate agitated design (Footnote: ²). He went on to say this:

Under cross-examination he was equally firm:

Professor Garside did suggest that static crystallisation might be used in some processes. However even here, the areas where such an application would be appropriate was very limited:

Needless to say, this does not apply to aspartame where the form of the crystals to be produced is of importance. Some idea of how wedded those in the art were to the use of agitated crystallisation was given by Professor Garside when he was challenged on how long a skilled worker would persist with trying agitated systems:

On the evidence it appears that over the wide field in which crystallisation has been used in industrial processes, static crystallisation has virtually never been used. It is not suggested that it has ever been used where a consistent product with narrow size distribution of the crystals was required. The reader of JP ‘267 would not find any indication in it that static crystallisation should be used if the examples or teaching were to be applied on an industrial scale. There is nothing in what is written which would suggest that any benefit it promises could not be achieved, and indeed achieved better, by employing agitated crystallisation. Unless the reader of JP ‘267 was taught, or had reason to suspect, that an agitated system would not work, for compelling technical reasons he would automatically use such a system. In the light of these considerations, the teaching of JP ‘267 does not render claim 1 of the patent obvious. However that is not the end of the story.

In his reply speech, Mr Kitchin approached the issue of obviousness from a different angle. The argument can be summarised as follows. The sort of chemist who would carry out simple laboratory scale experiments on aspartame would carry out a static crystallisation. He is likely to leave a pot of solution unstirred for crystallisation, for example in a cupboard or in a fridge. He would find and examine the pseudo-solid phase. He would determine solubility curves. He would look at crystals under the microscope. Laboratory scale crystallisation experiments would be carried out to examine the effect of crystallisation conditions on crystal size, size distribution, habit and the effect of seeds (that is to say artificially added nuclei) and the effect of these crystal characteristics on filterability. Different rates of cooling, whether or not to agitate and different agitation regimes would be investigated. He would find that the pseudo-solid breaks up and does not leave a residue. If an agitated system was tried, filterability problems would probably have been found in the lab, and certainly on a pilot scale. He would end up using static crystallisation for the industrial process. It is said that all of this would be done in the course of routine investigations.

This approach can be applied to the teaching in JP ‘267. The routine investigations referred to above would be applied as of course to the repetition of the examples in the patent. The sequence of investigations would be followed and one would end up using a static crystallisation process. It will be appreciated that, for this argument, the detailed teaching (or lack of it) in JP ‘267 is irrelevant. Daesang’s case is that these routine investigations would be carried out by any team which is trying to purify aspartame. In reality, JP ‘267 is a distraction. Any company trying to make aspartame would carry out these investigations. In substance, Daesang argues that the invention is obvious in the light of common general knowledge.

At the outset it must be said that this argument is not well served by history. Aspartame was discovered some 16 years before the priority date. A number of large chemical companies became involved in trying to find a commercial product to put on the market. If Daesang’s argument is right, all of them should have stumbled on the pseudo-solid at a very early stage and, from there, have arrived at the use of a static crystallisation process. If that happened, no one seems to have published the fact. Even the discovery of the pseudo-solid is not reported. The authors of JP ‘267, who on this analysis, should have found the pseudo-solid and from there worked out that it was static crystallisation which was the important step, seem to have failed to do so.

The argument proceeds along a series of small logical steps which is very easy to construct after the event. It is the kind of hindsight analysis which can be unfair to patentees. As Mr Prescott argues, even if one found the pseudo-solid, there is a significant prospect that it would be regarded as an irrelevance or worse. Professor Hounslow suggested that the last thing one would want to try to scale up would be something which set solid in the crystalliser. This would be a significant deterrent to going down this route. That would be particularly so unless it was known or reasonable to predict that the creation of the pseudo-solid would be better than the product of a normal agitated system. If the products were likely to be equivalent, there would be obvious technical reasons for not going down the pseudo-solid route. Why create a pseudo-solid which might give problems of encrustation and variability in product and which would need to be liquefied before it could be filtered, when a liquid could be made directly by known agitation processes? As far as the evidence goes, there was no reason to believe or suspect that the pseudo-solid route would produce the better crystals. A static system would be expected to produce just those differences in supersaturation which have commonly led to unpredictable products with wide size distribution of crystals. There is nothing to suggest that aspartame in a static system used on an industrial scale would have been any different. The point can be put another way. Assume that the skilled worker discovered that a satisfactory product could be obtained on the small scale with static crystallisation, notwithstanding the fact that it inevitably suffered from uneven temperature distribution and, therefore, uneven levels of supersaturation. Even in those unpromising conditions, a promising filterable product could be obtained. There is nothing to suggest that the skilled worker would believe that the same or better results would not be obtained by using the well known, more controllable and faster agitated system in which crystal size distribution would confidently be expected to be narrower with the result that filterability would be better.

It seems to me that the skilled worker’s experience would suggest to him that he would obtain technically better results by using the agitated system. If that was his view, he would not choose or contemplate a static system for commercial production because it would fail to deliver those benefits of speed, predictability, uniformity and cost saving required in such a process. If, on building a pilot plant or going into full scale production he found there was a problem with separating the crystals from the liquor, he would then have to try to find out what had caused it. Perhaps after two years he would have given up trying to use agitated crystallisation. Professor Hounslow gave a list of possible routes he would have examined to try and solve the crystal/liquid separation problem. It is not necessary to set them all out here. He said in re-examination that, after running through a number of possible options he would “probably cast about rather wildly” looking for something else to change.

Professor Hounslow expressed the view that the Ajinomoto process was counterintuitive and far removed from the accepted wisdom of the day. I accept that evidence. Whether one starts from JP ‘267 or common general knowledge, claim 1 is not obvious.

Ajinomoto only asserts independent validity in relation to claims 4 to 6. In view of the findings above, all the subsidiary claims survive. However I can deal briefly with each of claims 4 to 6 in case the finding of validity of claim 1 is overturned on appeal. None of them was the subject of spirited defence by Mr Prescott.

Claim 4 provides:

Professor Garside said that this is a very broad range and determining the preferred coolant temperature is merely a matter of carrying out straightforward trials on the equipment and the cooling device used. I accept that evidence. If Claim 1 were invalid for obviousness, this claim would be also.

Claim 5 provides:

Professor Hounslow acknowledges that for liquids and simple suspensions, the choice of the dimensions would be relatively straightforward. He says that it is not so straightforward in the case of a pseudo-solid. Professor Garside says that the preferred design of cooling arrangements is something that can be determined by straightforward calculations and trials. I accept the latter evidence. I would only add that it would be obvious in a non-stirred system where temperature differences could be a problem, to ensure that the distance between all parts of the charge in the crystalliser and cooling surfaces are kept to a minimum. That, in effect, is the inventive concept of this claim.

The only other claim to consider is claim 6. This provides:

As Mr Kitchin points out, Ajinomoto only attempts to support the validity of this claim if the words “and/or” means “and”. Mr Prescott does not suggest otherwise nor did he argue the point of construction. In my view, the words “and/or” mean what they say. They are not limited to cases of “and”. Professor Garside says that cooling alone would give supersaturation and that that is a usual way of causing crystal precipitation. This evidence was not challenged. If claim 1 were invalid for obviousness, this claim would be also.

In the result, Daesang’s claim fails. The patent is valid.