Molecular Construction Limits

by Robert J. Bradbury

Abstract


Some principles which we should consider: The concept of working with objects constructed at an atomic level was first explored by Feynman, in his now famous talk "There's Plenty of Room at the Bottom", in 1959.  The rules in this arena were explored in great detail by Eric Drexler in "Nanosystems", in 1992.  It is becoming common for Nobel prizes to be awarded for inventing machines which visualize and manipulate atoms, e.g. microscopes (1986),  ion traps (1989) and atomic traps (1997).

We can analyze historic trends in microelectronics, from the invention of the transistor, through the development of the integrated circut, to modern microprocessors and make some projections regarding future trends.  These projections may be combined with our current understanding of biology and compution to yield some estimates on computational limits.

It is important to remember that the pessimists have been wrong.  In 1983 paper by L. F. Thompson and M. J. Bowden titled "The Lithographic Process: The Physics", they stated:

"After consideration of all factors which limit resolution such as exposure hardware, resist systems, registration, alignment, and linewidth control, there is a general consensus that the useful resolution limit of photolithography may lie somewhere between 0.4 and 0.8 mm and depends on such factors as the implementation of short wavelength UV and the ability to accurately place images from a projection tool onto a silicon wafer."
These researchers worked at Bell Laboratories so we should consider this opinion well informed.  For many years Bell Laboratories has been viewed as favoring electron beam lithography over photolithography.  One is left to wonder if the purveyors of alternate technologies are those who tend to be the most pessimistic in their predictions.  Since the prediction clearly has been shown to be false with production lithography at 0.25 mm, soon to be 0.18 mm and research showing that optical lithography can be extended to at least 0.08 mm one is left with some doubt about the accuracy of pessimistic predictions.

Modern microelectronics is based on the ability to create patterns on the surface of semiconductor (Ge, Si, GaAs) materials.  These patterns are "drawn" using optical lithography whose trend is to increasingly smaller pattern dimensions.  From visible light (10 um scale) we have progressed into UV (.25 um scale) and will reach the .13 um scale by 2004.  EUVL methods recently developed at Sandia National Laboratories may take us even further to .01 um (100 nm).
 

Table 1. Lithography Trends
Year Lithography Method Source Feature 
Limit		 Comments
     
microns
  
1963
Photolithography
  
20 
  
1975
"
  
8.0
  
1977
"
  
5.5
  
1979
"
436 nm UV?
3.5
  
1982
"
350 nm UV
  2.25
  
 
"
      
1983
"
  
2.0
  
 
"
  
1.0
  
 
"
  
0.8
  
1993
"
  
0.5
 Production
1995
"
365 nm Mercury lamp
  0.35
 Production
1998
"
KrF 248nm laser
  0.13
 Production
1999
"
KrF 248nm laser
  0.10
 Production
1998
"
ArF 193nm laser
  0.08
 Uses phase-shift photomasks
~2003
"
ArF 193nm laser
0.13-0.10
 Production
~2007
"
F2 157nm laser with EPL
 0.10-0.07
 Production
~2013
"
Ar 126nm excimer
0.05 (0.03?)
 Experimental (see below)
2000+
EUV
13nm Xenon Gas laser
~ 0.08?
 Sandia EUVL Project
2000+
Ion Projection
He+ ions
0.04
 Ion Projection Lithography
2000+
X-ray
0.5-0.04 nm X-rays
0.05
2000+
E-beam
  
0.005
 SCALPEL system goes to 0.08 mm
Beyond these dimensions, there are at least three competing technologies, X-ray lithograpy, supported by IBM; e-beam lithography, supported by Lucent and nano-imprint lithography, developed by George Whitesides at Harvard and being explored in other laboratories now.  These technologies will allow the creation of patterns down to the .01 mm (100 nm) scale.  Currently it is unclear which of these technologies will be successful.  Economic theory would argue that competition will drive down their respective costs and evolutionary theory would argue that each technology may find a successful niche.  Below these dimensions, atomic force microscopes and variants thereof will allow us to manipulate individual atoms (~.2 nm).  It is worth noting that all living things, including humans, are testimony to the fact that macro-scale objects can be constructed out of small molecules and manipulations at the atomic scale.

Many people wonder if the historic trends in miniaturization can continue.  Economically it is difficult to see how they cannot. Few would argue of the benefits to society of:

Woe be to the company and its executives who allow themselves to stumble in these races.  Short will be the tenures of government officials who allow their countries to fall behind in these areas.  So while naysayers would argue that we must "hit-the-wall" soon, they in fact create the climate which negates their predictions.  Recognition and awards in our society are based on doing things which nobody else has done or succeeding where others have failed.  By defining the problem, the naysayers simply create another opportunity for creative individuals to achieve success.  How quickly those successes are achieved depends in large part on the size of the society and the fraction of its resources which can be devoted to investigating a problem.  As human population is still increasing and societies generally continue to grow wealthier, we should expect that increasing amounts of research will be done and the rate of problem solving will increase rather than decrease.

If current trends continue, we will reach the limits of 2-dimensional patterning of atomic scale structures between 2050 and 2060.  That means for a society our historic size, consisting of individuals with our innate intellectual capacity, it takes less than 150 years from the discovery of the electron to reach the limits of 2-D atomic construction.  The time scale may be shorter if the planet is larger, has more land mass, receives more energy from its sun or in some other way has a greater investigative carrying capacity.  The time scale may be longer if the investigative carrying capacity is smaller.  In either case is is likely that the time scale will be very brief from an astronomical perspective.

Once the 2-D limit is reached, the only way to continue progress is in the 3rd dimension.  We already build devices in 3 dimensions.  Current microelectronic devices may have dozens of steps of adding or removing material from the surface of the device.  Our most complex devices may have 6 layers of metal to connect the transistors.  Molecular beam epitaxy (MBE) is used to build devices which may have multiple layers of varying atomic thickness.  Once the 2-D limit is reached the pressure will increase to operate in the 3rd dimension if it has not previously been achieved due to the recongized benefits nanoassembly would provide.  So the evolution from not knowing the elements which compose matter to being able to build complex 3-D structures should take at most several hundred years for societies similar to ours.

So, we may assume that if a species evolves which begins to comprehend matter, the laws which govern its behavior and they have an interest in exploring and controlling it, there is rapid progress from a lack of knowledge to the ability to manipulate things at the limits imposed by nature.

References

See Also

Created: sometime in 1998
Last Modified: December 19, 2001
HTML Editor: Robert J. Bradbury