WEDNESDAY 8 MAY 2002
Lewis of Newnham, L.
Wade of Chorlton, L. (Chairman)
__________ PROFESSOR CYRIL HILSUM, Past President of the Institute of Physics and Independent
Corporate Research Adviser, PROFESSOR A MARSHALL STONEHAM, Former Vice
President of the Institute of Physics and Massey Professor of Physics, University
College, London, PROFESSOR IAN MARSHALL, Chair of the Institute of Physics’
Information Technology, Electronics & Communications (ITEC) Group and BT exact
Technologies, DR DOUGLAS PAUL, Research Fellow, Semiconductor Physics Group,
University of Cambridge and Former Editor of Technology Roadmap of European
Nanoelectronics, and DR W ANDREW PHILLIPS, Chair of the National Advisory
Committee on Electronic Materials & Devices and pyconsult Ltd, Institute of Physics,
Chairman 148. Good afternoon, everybody. It gives me great pleasure to welcome all our witnesses.
We are extremely grateful that such a distinguished group should come and give evidence to
us this afternoon and we look forward very much to hearing from you. May I also welcome
the guests. I do not know whether the guests are separate or a part of the team, but welcome
to you as well. You will have received a document which lists people’s interests and so these will not be spoken out during the meeting but will be put in the record at the end in the
normal way. Could I say to the witnesses that when we put our questions to you they will be
put to you as a whole and we will leave it to you to decide who answers the particular
question, and no doubt Professor Hilsum will allocate the questions as he thinks appropriate.
(Professor Hilsum) We have already done that.
149. There is no need for anybody else to take part in the answer to each question unless
they feel there is something to add. Would you be kind enough to introduce the team from the
Institute of Physics, Professor Hilsum, perhaps indicating your roles at the IOP and in your
(Professor Hilsum) I am Cyril Hilsum. I am Past President of the Institute of Physics. I
have spent much of my career in the Ministry of Defence as an active scientist, not as an
administrator, and then became Director of Research at the General Electric Company. Now I
work at University College, London, and I am also adviser to a small company, Cambridge
Display Technology, and a large one, Unilever.
(Dr Phillips) Andrew Phillips. I am a Fellow of the Institute of Physics. I spent 20 years
in the University Physics Department at Stamford and Cambridge. I then left to go into what
was GEC and then became Marconi where I was responsible for microwave development. I
am now an independent consultant.
(Professor Marshall) I am Ian Marshall. I have spent 20 years working for BT. I am
currently a long-term research programme manager. I am also part time Royal Society
Industry Fellow at University College. Within the Institute of Physics I am the Chair of the
Information Technology, Electronics and Communications Group and will become a member
of Council later this year.
(Professor Stoneham) I am Marshall Stoneham. For most of my career I was with the Atomic Energy Authority and I was for the last few years Chief Scientist of AEA Technology.
I moved to University College, London, where I am Professor at the Centre for Materials
Research. I am a former Vice President of the Institute of Physics and I run one of its major
journals. I also with my wife run a small technically based firm.
(Dr Paul) I am Douglass Paul from the University of Cambridge. I basically run all the
silicon and the silicon germanium research at Cavendish Laboratory. That involves industrial
contracts with IBM looking at the fundamentals of MOSFETs. I have also been involved at a
European level in writing a technology roadmap for UK Animatronics, a large part of which
is now in the International Technology Roadmap for Semiconductors 2001 edition.
Chairman: Thank you all for that. You have already had a list of the questions and we
have all read with interest your report which you kindly sent us and we will now proceed to
150. First of all, can I apologise to you all that I am not able to stay until the end of the
session because I have another commitment. My question is a two-part question. To what
extent are the activities of the Institute of Physics influenced by the ITRS? What steps do you
take to bring the generic problems identified in the Roadmap to the attention of your
members? These questions arise because we have been astonished at the detail of the ITRS
roadmap, which indicates many potentially fruitful areas for extracting maximum value from
the dominant CMOS technology.
(Professor Hilsum) I think a point should be made about that roadmap which is an amazing document and, though the UK has taken some part in its formulation, I think we
should point out that the UK is not represented on any of the major committees that form the
roadmap. That is a source of great embarrassment and sorrow to us but is indicative of what
has happened to the semiconductor industry in the UK.
151. Why is that?
(Professor Hilsum) I think that comes later on, if you do not mind, when we talk about why industry has not funded research. It is one of your later questions. I am quite happy to
answer it now. We will come back to it, I hope, because it is an important point about
government support and what government can do, particularly with industry.
Lord Lewis of Newnham 152. Could I ask the question the other way round? What are the criteria that are required
in order to be a member of that committee?
(Professor Hilsum) You have got to be active in research in microprocessor devices and applications, particularly in the devices and the technology of manufacture, but it also brings
in economics. You have to be a major player. A major player does not necessarily mean that
you have to be one of the major companies. You can make your own reputation but usually
you are backed up by some experience of production. I would have said that if the roadmap
had been formulated 20 years ago you would have seen a number of UK personalities
figuring in it. You might even have seen some of the people round this table – on your side.
The IoP’s strategy is not influenced directly by the roadmap because the technical activities
of the IoP are generally taking part in groups, so anything that came up centrally at the IoP
would be referred to the appropriate group and the appropriate group is ITEC and so
Professor Marshall is the appropriate person to answer that.
(Professor Marshall) Within the ITEC group, which I chair, this report is in fact very widely known and the members were very keen that we brought it to your attention. The
members of the group would like to address many of the issues it raises but they cannot really
do that through the Institute of Physics so that the Institute serves as a forum where they can
discuss issues and form collaborations and prepare projects but then those proposals have to
go through the EPSRC as you know and there does seem to be an issue if the EPSEC are not
able to provide enough funds to support substantial work in the area. That is the issue that
members have raised through me.
(Professor Stoneham) Might I just add a little extra there? I chaired the Board of
Directors of the Institute of Physics Publishing for a long time. It is the second largest
publisher of physics information in the world and of course its journals and magazines do
contain quite a lot of information which relates to this and which is drawn to the attention of
members, but of course it does not have any control over what they then do.
Lord Lewis of Newnham
153. I would like to follow up with a question to Professor Marshall. Can I be clear? Are you saying you made suggestions to the EPSCRC as to areas under which you ought to have
investigation and they were not able to do this?
(Professor Marshall) Not as the Institute of Physics. Members of the ITEC Group have
used group meetings as a forum for discussing joint proposals which they have made through
the EPSRC, some of which have been successful, some of which have not. As you know,
even successful projects are not very large, so they are not able to address many of the harder
issues raised in the report. It is a matter of regret for the members but as the Institute there is
not much we can do other than provide a forum.
154. Do you have an impression that there is a centre somewhere in government which tries to form an overview of the same field and is trying to use its influence to drive it ahead,
or is there no such body?
(Professor Hilsum) It is not my impression that there is such a body which you would
refer these problems to. Obviously the DTI has a general responsibility in the area and there
are various support schemes which we will deal with later. As you know, at one stage there
were initiatives, like JOAs(?) and ALVIs(?), support for innovation, a lot of schemes for
support and particularly the ALVI did deal with microprocessor type topics. This was more
or less at the same time when the Americans had a giant programme running from the
Department of Defence called VHSIC (very high speed integrated circuits) where specifically
the Department of Defence had an ambition to bring microprocessor work into the next
century more or less and a lot of funding went into that from the Department of Defence. At
that time we certainly had some initiatives and semiconductor work would have proved a
large part of it and we had our own Ministry of Defence programmes. In fact, of course, it is
not always appreciated that the first integrated circuits were made in the UK. There is some
controversy about that but I think that we had a good case for saying that and certainly it was
followed up and we were very active in the early days.
Lord Lewis of Newnham
155. For how much longer will it be possible to continue to improve the performance of microprocessors through making CMOS transistors smaller? What are the limiting physical
factors, and how will these interact with economic factors?
(Professor Hilsum) That is a good point. I have written down “smaller is not always
beautiful, particularly when it becomes too expensive”. The roadmap actually sets this out
pretty well and for the next ten years it shows that it does not see overwhelming problems in
reducing the feature size by what is almost an order of magnitude compared with what it is
today down to the low tens of nanometres, around 18 nanometres. That must be accompanied
by ways of exploiting devices better. This is the way you look at it because the roadmap
shows you the way in which you would go based on extrapolations of current technology and
every now and then it comes up against what it calls a road block and it says, “There are two
ways of getting round a road block. You either go through it or you go round it.” Sometimes
you try both and sometimes you find both work. It depends on which is the cheaper way forward. But always the driver is not actually to accomplish something that is faster but something which you can sell better and which will fulfil more of the function depending on whether you are going for the consumer market, the industrial markets, defence markets, or things like climate control. There is not really a physical problem I see of overwhelming importance for ten years. It so happens that in general they have been pretty conservative in their estimate of years with the last few roadmaps. This incidentally is the third roadmap and each time they have had to bring their timescales forward by about one year; in other words they are cleverer than they think they are. The answer is that there are ways of exploiting devices better and often it is cheaper to do that than it is to spend a lot of money on infrastructure and batteries to make things smaller, but you can do both and they probably will. You also ask about what the limiting physical factors are. There are a number of physical factors that do cause you problems that you have to tackle. They are the actual speed at which the carriers move through the transistors and other devices, the interconnect problem, insulating capacitances, interconnect resistances, inductances and capacitances. All of these things are important and they are all covered in the roadmap. There is nobody who thinks that you are going to come to a full stop. There are some people who feel it would be nice if they had some more money to tackle some of those problems, and I am sure you have seen and will see people who have got agendas in this, but I do not think that the roadmap sees that there is no way forward.
156. Are you saying that at the end of ten years there will be certain developments which
will allow you to go on using CMOS?
(Professor Hilsum) Yes.
157. Not that we will have to change to a different technique altogether?
(Professor Hilsum) No, except in niche markets. That is linked with another question you
have asked about alternatives to CMOS. What we would say is that silicon and CMOS will be predominant for many years. I say this with a little bit of sorrow because I spent some of my formative years working on alternatives to silicon and even at one stage was a founder member of we called the Anti-Silicon Society; the acronym you will notice is ASS. Unfortunately, though we did find some good other applications for gallium arsenide, we were never able to displace silicon and I am now convinced in the twilight of my years that this will not happen: silicon is going to be with us for many years to come. CMOS of course is a particular variety of silicon and that will stay for a long time, though there are some other things that can be done. Niche markets will develop and specialist devices can play an important subsidiary role.
(Dr Paul) You might want some numbers to put into all these things in terms of how
small you can make a transistor. I have seen a number of presentations by Intel, which is probably leading the world in terms of microprocessors at present. If you have a Pentium IV at the moment there are over 40 million 65 nanometre gate-leg transistors in that. Intel see no problems scaling that down to 30 nanometres which would be the next generation that they provide. Below that you start getting into some serious troubles if economics do not come in before then, and that is to do with things like power dissipation. Part of that is to do with leakage in devices and Intel is looking at new materials particularly for the gate insulator as it is called. That is one of the major problems that we face if we try and make things smaller. Intel research lines have demonstrated a 15 nanometre gate-leg MOSFET, really a conventional MOSFET, and AMD(?) have done a similar device. The shortest gate-leg transistor that has been fabricated and demonstrated is eight nanometres, although it does not switch off correctly, and theoreticians at conferences argue that the smallest gate-leg you can make is somewhere between four and six nanometres. Those are numbers that I hope will help give you some scales. Five nanometres is roughly ten times the latter’s constant with silicon to give you a rough idea of scale.
158. Thank you very much. In your written evidence you emphasise the significant role
that improved architectures, parallelism and software may play in exploiting the potential of CMOS to the full. To what extent may it be useful to consider the future development of microprocessors as an inter-disciplinary activity? If so, what are the relevant disciplines likely to be?
(Professor Hilsum) We have had a problem with this question because we have always
thought of microprocessor, indeed semiconductor, device development as being an interdisciplinary activity. It has always involved physics, material science (which brings in chemistry), engineering, software engineering and many other disciplines. We had a discussion here and there is even a point where philosophy has come into it, but I was hoping we would not get into that.
159. On the basis that it is, as you have described, technology that is going to flow on
from a lot of different areas in research centres, how do we pull it together to turn it into a focused role that can have results for both UK science and industry? If I could add to that, that really hits the basis of what this inquiry is about. I would be interested in getting your views on the role that this inquiry may have in this issue, and if that is the case what sort of role do you think we could play to ensure that we do have a more focused approach to this particular issue of post-CMOS technology that can bring benefits, as I say, for UK science and UK industry?
(Professor Hilsum) You are hitting at a very critical point of course because it goes back
to what I said about where we were. Assuming that we want to be present, and by “we” I
mean the whole of the UK, in an area which is of immense economic importance, then you look at the past and you say, “Where were we?”, and we are not there now so what did we do wrong? You can analyse what we did wrong. There are two things you then can do when you say, “We did something wrong and now the world has passed us by”. There is a great temptation to say that it is too late, you cannot do anything. You can just hang on to other people’s coat tails. I would like to feel that nothing is too late and it is never too late, that you can do some thing. I am encouraged by the Flanders Government. You might say that is a bit strange. How can anyone be encouraged by the Flanders Government except possibly the people in Flanders? In fact, an individual in Flanders some years ago said, “It is very late in microprocessors but we can do something.” He was a gentleman called Professor Roger Bannenstraden(?). He died, I think, last year. He almost single-handedly persuaded the Flanders Government and the Belgian Government, and I do not need to tell you how small they are compared with the UK, that they should set up an institution to revive industry in Flanders. This industry was wide but he specialised himself in things like solar energy and in silicon and they built an institution which now we use very largely. We do not have one of our own. I think it is very important to say that we can be ambitious. Yes, there are problems; yes, it is not going to be easy, but we can establish the UK with a role here. It does not come from academia. It has got to come from the whole country. You cannot fail to exploit a large part of your strength which is your industrial strength. Remember, it is science base. There are two words there. The “base” bit means there is something you put on it. We have got a very strong science base but we are not actually putting anything on that base. That gives you a vacuum and then other people come in and other countries exploit our science base to great extent. They exploit it by putting research contracts, they exploit it by coming and listening. They exploit it by taking many of the students. You can go to countries abroad and you find that they have got very many UK graduates, excellent people, who have not been used in this country. How can we exploit it? We have a concept and I should explain that there is a National Advisory Committee that came out of Foresight. Foresight was a wide exercise. There was a panel for defence and aerospace which I happened to be a member of, which faced the problem that DTI had that there were too many committees and they wanted a
series of specialised committees that they could go to for advice. One, which is the one of
interest here, is on electronic materials and devices. Professor Phillips is the Chairman of that
Committee and perhaps would like to say a word about one scheme which they have
proposed which could answer your question.
(Dr Phillips) The remit of this Committee was much broader than just silicon. It included
not just semiconductors, in other words not just compound semiconductors, but it also
included other non-semiconductor electronic devices. There was a common theme which was
to the effect that there are a great many resources and centres of expertise within the UK but
there is very little in the way of integration in the way that you see, for example, in the IMEC
example in Belgium. The recommendation really was that we try and impose some structure
on what is happening within the UK by setting up such a centre which has state of the art
equipment. One of the other difficulties, just to go back to an earlier question in terms of the
roadmap, in addressing some of those problems is that we do not have the state of the art
equipment centred in one place that allows you to make meaningful progress. You can work
on aspects of the pure science, if you like, but when it comes to practical realisation it
becomes much more difficult. The recommendation is that a centre be established and this
has a number of benefits. First of all, it would provide a centre of excellence, it would
provide if you like a certain weight which would allow us to get involved in collaborative
programmes within Europe. It would also act as a test bed because, although we are not
actually in silicon fabrication as such, we do have quite a strong industry providing
equipment for semiconductor fabrication, and one of the needs that that industry has is a
suitable place to provide a test bed, in other words to run their equipment under a
semiconductor production situation in order to provide data and so forth for marketing. It
would also provide the opportunity for relatively small pre-production runs and for allowing
people to test some of the basic ideas. It would cover not just silicon but also compound
semiconductors and indeed some of the other materials where you need to be able to process
160. Perhaps I could follow on what Dr Phillips has just been talking about in terms of semiconductor research and a testing centre. Which is the main candidate in your judgement
for location? What about order of cost? Could you comment on the relationship between
industry, government and the researchers in attempting to focus on the right areas and
produce the right result?
(Dr Phillips) I suspect some of the issues over location will probably reflect those that
have already been discussed in terms of the nanotechnology aspects. An obvious place would
be one big national facility such as Rutherford Appleton Laboratory. The precise way it was
organised would have to be through some kind of competition between different options. The
typical cost is going to be of the order of ?100 million in the initial establishment. In terms of
funding and as a consequence the way in which different users would interact I suspect there
ought to be core funding and then on top of that there ought to be specific projects in which
industry and academia become involved with EPSRC putting money in and industry putting
money in but with certain core running costs also provided centrally.
161. You said ?100 million but you do not mean ?100 million a year, do you?
(Dr Phillips) No. That would be the initial capital cost.
162. How much per annum thereafter for a while?
(Dr Phillips) Our estimate would be ?20 million.
(Professor Hilsum) Professor Marshall would like to say something on the extra things
that you want to do, but I want to stress that this is not an academic centre and this is not an
academic responsibility. We have the habit in this country where, when we need something,
we turn to one of the research councils and say, “You should set this up in a university”. That
is not what we are asking. Certainly academics should be able to use this facility but this is a
UK responsibility for industry, academia and government departments.
Lord Lewis of Newnham
163. I am slightly concerned and somewhat confused. In your excellent document you sent us you made the point quite clearly that in the University of Cambridge funding has been
predominantly through Hitachi and that this technology is now being taken back to Japan.
You also make the point that the training in semiconductor physics as it were is providing
good students for (by implication) other countries. What is wrong with industry? If we have
got this opportunity what you are saying to me is that it is not academia. We have got the
academic status, I assume, from what you are implying. It is an industrial problem. Why
should industry not take this on? If it has really got a problem with this particular venture and
it is going to provide the benefits for you which would otherwise pass you by, then it should
be doing something itself. Why does it need a crutch from us?
(Professor Hilsum) You keep on saying “why” and “should”, whereas I am interested in
what has actually happened. We should not be in this position because industry should have
taken care of it. It had these things. It did not. These things have been analysed. Everyone has
been over it in the last ten years and now the Government is beginning to do some things
about it. I will say that these initiatives are very welcome: the R&D tax credit for small
companies, the R&D tax credit for large companies, the corporate venturing scheme. All of
these things are an admission by the Government that you can say “why” and you can say
“should” but it has not happened. You have got the DTI R&D scoreboard. Excellent. They
thought seven years ago that the problem was that people did not understand what was
happening, so they set up this scoreboard, which was unique at this time, and it has been
copied in other countries now, which actually listed all of the major companies and this went
down to quite small ones that did research, said what their turnover was, how many people
they had working for them, went through all the facts and said how much R&D they did. To
their horror they discovered that the UK, except in the pharmaceutical industry which stands
out, actually did between one third and one half of what its competitors did, industry by
industry. The only industry that was difficult was food because our food industry is
dominated by Unilever so it was not easy to do a comparison. In every other industry you
could see the shortfall. The overall picture was confused a bit by the pharmaceutical industry.
The pharmaceutical industry stands out. It is competitive with every other country. It also
does one third (there are now three large companies; there were five when I did my analysis)
of the R&D in the country and, because that is at the same level of industry, when you do the
average over the whole UK it looks a bit better than if you exclude the pharmaceutical
industry. If you exclude the pharmaceutical industry our record is abysmal.
164. I do not want to disappoint you but if you actually look now at the pharmaceutical industry a large amount of that has now been taken out of this country and put in Switzerland
(Professor Hilsum) Please do not make me feel worse. I had a lecture by Sir Richard
Sykes and he did a splendid lecture on doing research and why you did research. At the end I
said to him, “It is a paradigm. You have explained why but you have not addressed other
companies, other industries. Why in the UK do they not?”, and he said, “I do not understand
165. Do you think this is due to the attitude of the City wanting short-term profits and this is a disincentive to industry to invest in longer-term things?
(Professor Hilsum) The answer must be yes to some extent but, having worked in
industry, I think it is much more the attitude of the managers in industry who actually do not
want to take these risks and they use analysts to some extent as an excuse because when you
talk to the analysts they say no, they are quite encouraging to research. Mind you, I have
never yet seen an example where a company announces it is going to do more R&D and its
shares go up in value. I did ask Carol Galley that question, as to had she ever given a
recommendation to buy a company’s shares because they were doing more R&D, and she did
not say yes and she did not say no. She hedged. It was fairly obvious that they do not think of
R&D as being an asset. Of course our standard structure of finance does not have R&D as an
asset. R&D, as you know, is written off at the end of the year so you are more or less saying
that it has got no value. If you buy some equipment you actually can depreciate that over
several years. You cannot depreciate R&D. I think that is all because Rolls Royce went off
the rails some 20 years and did depreciate their R&D but not fast enough.
166. While the UK no longer has any significant semiconductor manufacturing industry, it has a thriving microelectronics system design industry. However, Research Council funding
priorities seem to favour devices and materials rather than architecture and design. Do you
have a view on this apparent mismatch of UK industrial strengths and funding priorities?
(Professor Marshall) Certainly we do have a view. I would first of all say that within the
UK there is some semiconductor manufacturing going on. It is foreign owned.
167. It is a very small amount of manufacturing
(Professor Marshall) The linkage between the semiconductor manufacturing that we
have and any government-funded research is not very clear because a lot of it is foreign
owned and the R&D for those companies is done elsewhere. However, getting to the meat of
the question, I have recently been on the evaluation panel for EPSRC’s research programme
in this area and it is quite apparent, looking at that programme, that there is a lot of
investment going into devices and materials and not very much into systems and architectures.
EPSRC would like to change that. With provisos, I believe that they are right. They do need
to encourage more working systems and applications in the economic sector. However, the
reason the programme is the shape it is is that there are structural problems with the
electronics and primarily what I am trying to say is that if one wants to do state of the art
electronics research one needs to have access to high quality materials and device processing
which costs an awful lot of money. Within the limited budget that EPSRC currently have for
this sector they are really falling between two stools. They are trying to meet requirements for
providing device processing and are not actually able to invest enough in device processing
to have significant world class device research. It is fragmented across a very wide range of
materials – silicon germanium, silicon carbide, gallium nitrite, silicon at Southampton, and the fragmentation simply makes things worse and there is some duplication of materials
processing in all the different centres that have got specialist processing for specialist
material types. Because they are doing their best to support the cost of device processing they
are left with a very small budget for systems and applications which does not encourage the
academics to develop expertise in that area because they can see that there is not very much
money to get.
168. But if we look at the software development of this and architectures, this seems to be an area in which we are very good.
(Professor Marshall) Yes. On the industrial side we are very strong. On the academic
side that is less true, largely because there is a funding gap that the academics are not able to
169. Professor Hilsum presumably would say that this was not a job for the research councils anyway; this is an industry job.
(Professor Hilsum) It depends how deeply you are going into it. If you are simply doing
programming and making new software programmes, I would say that is correct. If on the
other hand you are analysing more carefully how you can exploit microprocessors I would
have said it is definitely a job for industry. I know some of the people in France have got
some very high level groups that are really trying to look at different operating systems, for example. Why use the straightforward operating system that may have developed and may have expanded from a base which itself was not optimum for doing complicated things that you probably want to do in the future? I would have thought there was certainly an academic role for this. Indeed, we envisage that there would be a role for this within the national centre.
(Professor Marshall) What I would say is that the way EPSRC is structured and the way
academia is structured, they are well set up to produce new ideas. In the electronics sector you might argue that if one put aside the cost of processing the EPSRC budget might be reasonable, and part of the problem I am trying to point out is that putting the cost of processing in there is skewing the research funding. If you said to EPSRC, “Your job is to fund research or new ideas in the academic sector and you should spend that money on new ideas and we will provide the facility that those new ideas can be tested on and ideas from industry that enables them to be exploited”, that would be a reasonable way forward. The academics sector is not really well set up to run a large facility. You get all these small facilities competing with each other. EPSRC are not well set up to run a large facility in partnership with industry, which is what is needed. Maybe a separate facility would free money in the EPSRC budget to allow them to spend more time doing research rather than building plans.
170. Before asking my question I have been pondering over something that was said
before, which was the short-termism of the City and whether that has been an effect. I am not one ever to defend the City, it is not my normal style, but I think when you look at Wall Street, for example, where public companies report every three months, not every six months, so you would expect them to be even more short-term than we are, yet that does not seem to be it. My own conclusion is that perhaps there are not enough scientists involved at the higher levels of our companies, and that may be a reflection of the companies themselves or a reflection of scientists, but I think there is not enough understanding of research and development in some of our companies.
(Professor Hilsum) Do you want me to address that?
171. Yes please.
(Professor Hilsum) I worked for some years, as I said, for GEC. I did not find that the
problem was actually at top level in GEC, which was Lord Weinstock. He was pretty difficult but if you knew what you wanted and you had the right words to express it, usually with a smile, -----
172. You are okay. He does not come here that often.
(Professor Hilsum) That is all right. I like him a lot. He was okay. It was the people
under him who actually were the problem. They are the people who often used to say that you had to be short term, and there is no question that if you had a good idea in GEC and you put it up you had to show a positive cash flow in three or four years at that, whereas the same question, when it was put to Siemens, was six or seven years. You can work your way through and see what is likely to happen. It does not happen suddenly, but if you analyse it over ten or 15 years you find you are doing the less risky things and there is no penalty for missing opportunities; that is the trouble. You have to have a penalty for missing opportunities. I can assure you there is one in Japan. There may be one in the US. I agree with you. I could never quite understand how you had similar financial systems and indeed a similar pattern of shareholding (it is quite different in the UK from Germany and Japan) between the UK and the US with finance houses holding quite a lot of shares. I could never follow quite why that was. I could only give you the numbers. In fact, the US has recently, two years ago, copied the R&D scoreboard so that we can do a very accurate comparison, almost company by company, industry by industry, and there is a factor of between two and three.
173. I would like to move on to the alternatives to CMOS because this is quite a
fundamental part of what this Sub-Committee is doing. In your evidence you did suggest some ways in which novel devices may begin to play a role in computing over the next 20 years. I would like you, if you could, to expand on these topics and in particular indicate which lines of research you believe are most likely to lead to devices that will have the capability to exceed the functional performance of CMOS. I think you have already given some indication on that. Where does this country have its greatest expertise?
(Professor Hilsum) Before I ask Douglas to answer that let me set this in context. We all
agree that niche markets will develop for a number of things, a number of other materials, a number of other devices, and the specialist devices will play an important role but it will be a subsidiary role. We do not want you to get the impression that clever research is going to lead to something which is going to push CMOS aside.
174. You have made that point.
(Professor Hilsum) Yes, but it is an important point because you will find some people
with an agenda are trying to persuade you that you ought to support some very exotic things that they work on and are going to change the world. They may do but I would not put my own money on it, so I would not like you to put anyone else’s on it.
(Dr Paul) To qualify starting this, we are talking about microprocessors, and every
microprocessor is made up of logic for memory. When you look at different and new technologies you have to look at each separately. In terms of memory, memory is actually quite advanced in terms of new ideas. That is mainly because DRAM will actually run out fairly quickly. You cannot scale it to below certain sizes because of the materials and inherent capacitance you find in those materials. People have been looking for new technologies which are memory economical. The ones that are being researched in government at the moment and are most advanced have single electron transistors on which, as has already been mentioned, Hitachi and the EU have funded quite substantial innovative research at Cambridge, most of which has now gone to research and development labs in Japan. The other technologies that are being pushed by a lot of countries are nematic technologies and that is really thinking of hard disk technologies and trying to get those on a silicon chip or some form of chip. In the UK in terms of innovation that is where we are very strong, but again we do not have any manufacturing base. Companies like Motorola have been pushing that quite substantially. If we go to logic, logic is far more difficult. We really come to the niche markets here. If all you are concerned about is the highest performance without cost being an issue, then superconductors are probably the best option to go, but they will only ever work at cold temperatures(?) and they will be very expensive to run. If you then start looking at alternative technologies, the real problems start to come with architecture. The architectures that are used really all go back to original designs and very little has changed. People are still almost doing nothing but scaling devices. They might have a few tricks in terms of parallel processing and other things on a chip, but everything is very similar and they do not cope with faults. The problem when you start going to new technologies, particularly things like molecules or III-V technologies, is that you do not have the yields and the accuracy that people have in silicon. There are ideas such as molecular self-assembly systems. The one that has actually been proven at room temperature and works in terms of logic is resonant diodes and they have at least been shown in III-V systems with small numbers of devices, about 250, to show an order of magnitude improvement over Siemens processors. There are some spintronics again using nematic ideas for switching. They are at a very basic level so far and have not been pushed very far. Again, there are problems with architectures in those. One area that the UK does have enormous expertise in is quantum optics. What we should say about this is that this is very far off and it really has very few applications at present, but those applications are really in defence and factorisation for breaking RSA codes, and also in security location. The UK has an awful lot of money from