APPENDIX 4: VISIT TO NPL
1. Members of the Sub-Committee visited NPL at
Teddington on 6 March 2002, to see something of the technologies
involved in the development and fabrication of computer microprocessors
and to discuss issues with NPL experts.
2. The visiting party consisted of Lord Wade
of Chorlton (Chairman of the Sub-Committee), Lord Flowers, Lord
Freeman, Lord Hunt of Chesterton, Lord Methuen, Lord Oxburgh,
Lord Patel and Baroness Wilcox. The party was supported by the
Sub-Committee's Specialist Adviser (Professor Steve Furber) and
Clerk (Mr Roger Morgan), and the Select Committee's Specialist
Assistant. Dr Sarah Pearce of the Parliamentary Office for Science
and Technology was also present.
Introduction
3. The Committee was welcomed to NPL by Dr Bob
McGuiness, Managing Director, and Dr Kamal Hossain, Director of
Science and Technology. Dr McGuiness outlined NPL's role as the
United Kingdom's national laboratory for physical standards. As
such, its work was in the field of metrology the science
of measurement. International co-operation was vital if there
were to be agreed world-wide standards. Alongside NIST in the
US and PTB in Germany, NPL was one of the most respected national
standards laboratories in the world.
4. NPL was founded in 1900 as a traditional government
laboratory. In 1995, it became a GOCO a government-owned,
contractor-operated institution. 65% of its income came from government
(via the DTI); the remaining 35% was competitively earned. Dr
McGuiness said that the amount and high standard of the construction
work on site (state of the art buildings specifically designed
for metrology, with humidity and temperature control and vibration
stability) was a testament to NPL's past success and future ambition.
Presentations
5. Graham Peggs outlined the progress of technology
in the semiconductor industry. Consumer demand was the main driver
for more powerful computers, with the main attention on wireless
technologies, real-time graphics manipulation, games and gadgets.
6. Computer chips were immensely complex structures
containing tiny components. The tolerances of error and variation
were very small; metrology was a vital element in the manufacturing
process. Measurement techniques had to stay ahead of the required
manufacturing precision measurements needed to be about
ten times as accurate as the performance tolerances.
7. The semiconductor industry was unusual in
producing a forward projection of targets and problems for the
industry as a whole. This ITRS "Roadmap" was published
and updated with a remarkable degree of consensus 2,500
firms were involved in the production of the 2001 edition. The
ITRS set specific, quantified manufacturing objectives to allow
the industry to keep pace with Moore's Law but anticipated that
the physical limits of current approaches to metrology would prevent
the industry making progress beyond 2011. Radical breakthroughs
would be needed if progress were to be maintained.
8. John Gallop gave an introduction to computing
and disruptive technologies. He outlined the three limits constraining
the ultimate performance of microprocessors:
a. Thermal Speed could be increased
as the electrical energy required to write each bit of information
was reduced. However, to avoid random errors, that energy had
always to be significantly greater than the temperature-related
energy of the environment.
b. Quantum Continuing to reduce the energy
involved in writing each bit would eventually result in the flow
of so few electrons that the uncertainties inherent in quantum
physics came into play. The onset of such conditions would mean
that the time taken to write a bit could be reduced further only
if the energy required to do so were increased.
c. Power dissipation Increasing either
the density of components on a processor or the speeds at which
operations were carried out increased the amount of waste heat.
To avoid damage through overheating, the energy required to write
each bit had to be reduced as density and speed of operation increased.
9. Dr Gallop then described some areas of science
that could lead to dramatic alterations in computer processor
technology:
a. single electron transistors that would
reduce the energy involved in each processor operation;
b. spintronic devices that could use the properties
of both charge and spin of electrons to make logic gates;
c. molecular scale electronics that would allow
very small circuits to be constructed by self-assembly rather
than lithography;
d. biological computing (using protein chemistry
to store and process information) might produce compact but relatively
slow processors; and
e. quantum information processing, which was
attracting world-wide research interest, although as much for
cryptography as for general computing.
10. Such technologies were in their infancy,
and would require enormous investment before they could have a
substantial impact on computer processors. It was probable that,
while new technologies would be excellent for certain aspects
of information processing, they would have drawbacks for others.
The most likely path of development was to continue with the standard
CMOS technology at the heart of a processor, but with different
components based on alternative technologies built in
making a system on chip.
Tour of Laboratories
11. The Sub-Committee then visited various laboratories
to see some of the highly specialised equipment, some of which
had been made at NPL, that was needed for very small-scale metrology.
During the tour, there were five brief presentations.
12. Dr J-T Janssen showed how reducing the diameter
of wires essential to miniaturisation of components on
a processor changed the behaviour of electrons. At diameters
over one micron the behaviour was classical, demonstrating the
familiar properties of current and voltage dictated by Ohm's law.
As diameters reduced, these classical properties began to break
down. At a diameter of 0.1 micron approaching the wavelength
of electrons quantum effects began to emerge, and it became
possible to conduct electrons one by one.
13. Chris Hunt discussed the role of interconnectors
in microprocessors. As the density of components on a chip increased,
so did the density of input/output (I/O) interconnectors. The
physical barriers to getting more I/O connections on a chip using
current approaches would soon act as a limit to the rate of increase.
Another problem was the dissipation of waste heat from working
microprocessors. New substrate materials were being developed
to stop processors overheating as the density of components continued
to rise.
14. Martin Seah gave a presentation on analytical
metrology at the limits, which outlined some of the techniques
employed to measure surfaces and structures on the nanometre scale.
Auger Electron Spectroscopy, X-ray Photoelectron Spectroscopy,
Secondary Ion Mass Spectroscopy all provided excellent results,
but the ITRS indicated that the limits of their resolution would
be reached over the next 10 to 15 years. At present, many of the
instruments (many made in Germany) had markets in both semiconductors
and biotechnology. More refined or new instruments might be needed
to address the post-CMOS technologies and this was a factor which
UK manufacturers would watch closely. The potentially small demand
for the refined or new measuring devices would make this a tough
market for existing let alone new manufacturers.
15. Graham Peggs and John Gallop then expanded
on some of the themes from their earlier presentations with demonstrations
of some of the experimental facilities at NPL. The results of
experiments using scanning tunnelling microscopy (STM) and atomic
force microscopy (AFM) to measure and manipulate individual molecules
were shown, and some of the properties and potential applications
of carbon nanotubes were discussed.
16. Although it was not physically possible directly
to measure distances smaller than the wavelength of the light
or other energy illuminating the object, NPL had developed ways
of using X-ray interferometry (XRI) to measure even smaller distances.
Working together with the German PTB, NPL had developed a combined
STM/XRI nano-scale metrology tool with a wide range of potential
applications in the future development and manufacture of microprocessors.
Discussion
17. In a general discussion over lunch with Dr
McGuiness, Dr Hossain, the various presenters and other NPL colleagues,
the following main points were noted.
a. The ITRS noted a wide range of metrology
issues that, starting in 2003, currently had no solution.
b. Metrology was essential not only for manufacture,
but also throughout the innovation and development stages.
c. Maintaining the momentum of Moore's Law would
require radical new approaches on various fronts. (The term disruptive
technologies had undesirable negative overtones.)
d. NPL had some of the leading expertise in relevant
areas of metrology, although this was found in a variety of rather
separate projects rather than in a co-ordinated effort applied
to microprocessing. This reflected the position in the country
as a whole where there seemed to be no mechanisms to draw together
the relevant expertise.
e. The US was the dominant country in terms of
research in relevant areas. Taken together, European countries
played a strong role almost matching US research effort.
In the United Kingdom, there were some signs of lack of coherence
in research efforts applied to microprocessor development, and
better mechanisms for encouraging industrial exploitation were
needed.
f. In addition to Dr Hossain's involvement in
the 13 March seminar, the NPL team would be happy to discuss these
issues in an on-the-record public hearing[112]
if that would be helpful.
18. Members endorsed the Chairman's thanks to
Dr McGuiness and his NPL colleagues for their hospitality and
for a most stimulating and informative day.
112 They did so on 29 May 2002. Back
|