Nanotechnology for Clean Energy and Resources

Revised for the Foresight Institute
October, 2002

Stephen L. Gillett, Ph.D.
Dept. of Geological Sciences
Mackay School of Mines
University of Nevada
Reno, NV 89557

Copyright (c) 2002 by Stephen L. Gillett
All rights reserved.  Used by permission.

TABLE OF CONTENTS

ABSTRACT	3
1. INTRODUCTION	5
1.1. The Promethean Paradigm	6
1.2. Beyond Prometheus	6
1.3. Applications of MNT to Resources	7
1.3.1. Vastly improved efficiency of energy usage	7
1.3.2. Information-intensive energy extraction	7
1.3.3. Solid-state energy generation	7
1.3.4. Molecular separation of elements and molecules	8
1.3.5. Change of materials mix	8
1.3.6. Space-based resources	8
2.  NANOTECHNOLOGY AND ENERGY	8
2.1. The Promethean Paradigm, I.	9
2.2.  Specific Technological Approaches:  MNT and Energy Efficiency.	10
2.2.1. MNT vs. Prometheus.	11
2.2.1.1. Fuel cells.	12
2.2.1.1.a. Catalysts.	12
2.2.1.1.b. Electrolyte.	13
2.2.2. Catalysis.	15
2.2.3. High-strength materials.	17
2.2.4. Energy storage.	17
2.2.4.1.  Portable power sources.	17
2.2.4.1.a. Batteries.	17
2.2.4.1.b. Capacitors.	18
2.2.4.1.c. Fuel cells.	20
2.2.4.2.  Electrosynthesis.	21
2.2.5. MNT, nanofabrication, and distributed fabrication.	21
2.2.6.  Passive energy handling and smart materials.	22
2.3.  Information intensive resource extraction.	23
2.3.1.  Ubiquitous sensing.	23
2.3.2. Diffuse resources.	23
2.4 Specific Technological Approaches:  MNT and Solid-State Energy Generation.
24
2.4.1. Heat engines:  thermoelectric power.	24
2.4.2 Piezoelectric power	25
2.5 Energy sources.	25
2.5.1. Conventional fossil fuels.	26
2.5.2. Unconventional fossil fuels.	27
2.5.2.1. 'Oil shale'.	27
2.5.2.2. Tar sands.	27
2.5.2.3. Clathrates.	28
2.5.3.  Applications of nanotechnology to fossil fuels.	28
2.5.3.1. Extraction of fossil fuels.	28
2.5.3.2. Catalysis and syngas.	28
2.5.3.3. Usage of fossil fuels.	29
2.5.3.4.  Thermoelectrics.	29
2.5.3.5.  Superstrength materials.	29
2.5.4. Nuclear fission.	30
2.5.4.1. Uranium.	30
2.5.4.2. Thorium.	30
2.5.5. Applications of nanotechnology to fission energy.	30
2.5.5.1. Mining and extraction.	31
2.5.5.2. Reprocessing of nuclear fuel and separation of nuclear waste.	31
2.5.5.3. Isotope separation.	32
2.5.6. Geothermal energy.	32
2.5.7. Applications of nanotechnology to geothermal energy.	33
2.5.7.1. Information-intensive exploitation.	33
2.5.7.2. Automated fabrication.	33
2.5.7.3. Thermoelectric materials.	33
2.5.8. Solar energy, indirect sources.	34
2.5.8.1. Hydropower.	34
2.5.8.2. Wind.	34
2.5.8.3. Wave energy.	34
2.5.8.4. Ocean-Thermal Energy Conversion (OTEC).	34
2.5.9. Applications of nanotechnology to indirect solar energy sources.	34
2.5.9.1. Materials implications.	34
2.5.9.2. Automated nanofabrication and diffuse resources.	35
2.5.10. Solar energy, direct sources.	35
2.5.10.1. Photovoltaics.	35
2.5.10.2. Artificial Photosynthesis.	35
2.5.10.3. Solar thermal.	38
2.5.11. Applications of nanotechnology to direct solar energy sources.	38
2.5.11.1. Nanofabrication.	38
2.5.11.2. Dispersed collection.	40
2.5.12. Tidal energy.	40
2.5.13. Nuclear fusion.	41
2.5.14.  Applications of nanotechnology to fusion energy.	41
2.5.14.1. Separation of nuclear fuels.	41
3. NANOTECHNOLOGY AND MINERAL RESOURCES	41
3.1 Minerals and Ores.	41
3.2 Element Separation	43
3.2.1. Fundamental costs of element extraction.	45
3.2.1.1. Solution extraction.	45
3.2.1.2. Mixing entropy.	46
3.2.1.3. Macroscopic 'entropy'.	47
3.2.2. The Promethean paradigm, part II:  Phase separation	48
3.2.2.1.  Beneficiation.	49
3.3. Molecular Separation.	49
3.3.1.  Solution separation.	50
3.3.1.1.  Semipermeable membranes.	51
3.3.1.2. Ion Exchangers.	52
3.3.1.3. Complexing agents and molecular recognition.	53
3.3.1.4. Switchable binding.	55
3.3.1.4.a. Redox-switchable binding.	55
3.3.1.4.b. Electrosorption.	58
3.3.1.4.c. Light-driven switching.	58
3.3.1.5. MNT and molecular separation.	61
3.3.1.6. Microbial-based  extraction.	61
3.3.1.7. Implications of biomimetic extraction.	62
3.4. Isotope Separation	62
3.5 Change of Materials Mix	63
3.5.1. The obsolescence of structural metals.	64
3.5.2. Carbon:	65
3.5.3. Terrestrial carbon resources, I.  The obsolescence of petroleum.	66
3.5.4. Terrestrial carbon resources, II:  The significance of carbonates.	66
3.5.5.  Silicate MNT and resources.	68
3.6. MNT and Rare Elements.	69
3.6.1. Thermoelectric materials.	69
3.6.2. Redox-active framework formers.	70
3.6.3.  Nuclear fuels.	70
4.  SPACE RESOURCES AND NANOTECHNOLOGY	70
4.1. Solar Power Satellites (SPS).	71
4.1.1. Nanotechnology and SPS.	72
4.2 Lunar Helium-3.	72
4.3 Near-Earth Materials.	72
5. CONCLUSION:  A VIEW FROM THE PALEOTECHNICAL ERA	72
5.1 Toward a New Stone Age?	73
6. REFERENCES CITED	73


LIST OF FIGURES

Figure 1:  Fuel cell	13
Figure 2:  Perovskite structure	19
Figure 3: Double layer	20
Figure 4: Band gap	36
Figure 5: Diatoms	44
Figure 6: Hydroxyoxime for Cu extraction	50
Figure 7: Electrodialysis	51
Figure 8: Crown ethers	53
Figure 9: Quinone <=> hydroquinone	55
Figure 10: Ferrocene	56
Figure 11: Azobenzene	58
Figure 12: Spiropyran photoisomerization	59
Figure 13: Limestone outcrop	67



ABSTRACT.
	Although resource issues have a somewhat musty feel, they have not gone
away:  in the case of the US and other industrialized countries, they've merely
been postponed by cheap foreign sources.  In fact, conventional technology is
exhausting its resource base at an accelerating rate, an acceleration
exacerbated by the revolution of rising expectations in the less-developed world
due to the global communications revolution.
	Nanotechnology is the only way to resolve this dilemma that has the prospect
of providing something like a sustainable First World standard of living for the
entire world.  Indeed, in the author's opinion this is most critical near-term
application of nanotechnology.
	Resource-related applications of nanotechnology include:
		 Greatly increased energy efficiency.  
			Beyond Prometheus.  Energy is used grossly inefficiently at present
because it is largely used as heat, both in Carnot-limited engines and in
thermal processing to manipulate matter via phase changes.  Fuels are "burned";
that's what fuels are for, a mindset that might be termed the "Promethean
paradigm."  However, burning a fuel wastes most of its energy.  Utilizing
chemical energy without thermalizing it, as organisms do, requires molecular
structuring.  A near-term technological example is fuel cells.
			Focused processing.  Conventional processing also creates many
by-products, which add to the costs both of purification and pollution control.
However, highly specific catalysts that would generate only the desired product,
in emulation of biological enzymes, require nanostructuring.
			Distributed fabrication.  When the slogan "matter as software"
begins to come true, the enormous present transportation infrastructure will
slowly become obsolete, as the massive importation of raw materials into
conventional factories, and their re-export as finished products, begins to be
supplanted by nanoscale fabrication from local materials--again, in emulation of
biological capabilities.
		 Information-intensive energy extraction
			Sensing.  Ubiquitous nanoscale sensing, an extension of
microtechnologies already being deployed on an experimental basis, will lead to
better management of oil and geothermal reservoirs, with significant savings.
			Diffuse sources.  In another elaboration of existing trends,
customized and ultimately distributed fabrication will make energy collection
from diffuse sources much more practical.  Examples include the exploitation of
small-scale mechanical sources such as low-head hydropower, tidal currents,
surf, and "at wellhead" geothermal power.
			More importantly, cheap large-scale fabrication of nanostructured
materials, which would eliminate moving parts, promise a suite of energy
applications including:
- Direct use of solar power, via photovoltaics or artificial photosynthesis.
- Thermoelectric materials to exploit small thermal gradients (e.g., geothermal
or marine, as in ocean thermal energy conversion); 
- Piezoelectric materials to convert mechanical stress directly into electric
potential.
			Efficient energy management.  Cheap nanofabrication will also lead
to such things as:
- Materials for passive energy management, such as electrochromic or
photochromic "smart windows"; 
- Efficient energy conversion devices such as non-thermal illumination sources
(e.g., "white LEDs");
- Electrosynthesis for fuel manufacture and electricity storage;
- Better electricity storage devices such as intercalation batteries and
high-performance "ultracapacitors."
			Superstrength materials.  As materials having strengths approaching
the ultimate limits set by chemical bonds become available, they will make
transportation considerably more efficient through savings in vehicle mass.
This will have a particularly pronounced effect on near-Earth space access.
		 Molecular separation
			Element separation, whether for pollution control or resource
extraction, is not intrinsically energy-intensive, as is shown by the
capabilities of biosystems.  The enormous energy costs of present-day
pyrometallurgy largely result from the application of heat to force phase
changes to exploit the partitioning of elements between phases.  Biosystems
achieve their efficiencies by eschewing phase changes in favor of direct
molecular separation via specialized molecular machinery.  Such "biomimetic"
molecular separation is an obvious goal even of primitive nanotechnological
systems.  It will have the effect of blurring the distinction between a
"pollutant" and a "resource."  Such technologies will also have energy
implications through making accessible low-grade sources of nuclear fuels (e.g.,
U, Li) and by vastly decreasing the cost and difficulty of fuel reprocessing,
although concerns about control of nuclear materials become more cogent.  In any
event, the 5000 year-old paradigm of digging up and "cooking" anomalous geologic
deposits to extract desired materials is coming to its end.
		 Change of materials mix
			As nanoscale fabrication makes accessible the ultimate materials
strengths set by covalent chemical bonds, the structural metals that dominate
present technology will become obsolete.  If carbon becomes the "ultimate
material", the carbonate rock that forms the bulk of the crustal carbon
reservoir becomes an important backstop resource.  Indeed, the very silicates
that make up most of a rocky planet become a valuable feedstock for a mature
nanotechnology.
		 Off -Earth resources
			By vastly decreasing vehicle and payload mass, nanotechnology is
likely to make near-Earth space access considerably more economic in the
relatively near term.  This may make off-Earth resources such as solar power
satellites and asteroidal metal more attractive, though nanotechnological "end
runs" such as molecular separation may make such sources obsolete even as they
become practical.


1.  INTRODUCTION
	Nearly 30 years ago Limits to Growth, one of the most influential books of
the 20th century, was published (Meadows et al., 1972).  It purported to
demonstrate, by means of a computer model, the imminent collapse of industrial
civilization due to the exhaustion of critical resources.  It was followed
shortly by the 1973 "oil crisis", which immediately lent credibility to such
scenarios.  However, when gas lines went away and oil prices dropped in the 80s,
many people concluded the crisis was "phony"--and by implication, resource
shortages were too.
	It wasn't, and they aren't.  US oil production had peaked in 1970--not
coincidentally, just before the "gas crises"--and suddenly we were at the mercy
of the vagaries of foreign supplies.  Despite this "wake up call", the US has
made a societal choice since then to continue depending on cheap foreign oil.
One result is that two numbers have been rising steadily since the early 1980s:
US oil consumption, and the percentage that is imported.  In 1999 the US
produced 6.45 x 106 bbl/day oil (Oil & Gas Journal Databook, 1999), and imported
9.61 x 106 bbl/day (Oil & Gas Journal, 2000).
	That cheap foreign oil is also limited.  Duncan (2001) has developed a good
case that global production of conventional oil will peak toward the end of this
decade.  When that happens, there will be no place left to import from.  The ~6
x 106 bbl/day of US domestic oil production ranks the US second among world oil
producers, behind only Saudi Arabia.  And it still isn't enough.
	Minerals resources, obtained from geologically anomalous concentrations
("ores"), are also largely obtained from abroad.  For example, the US produces
no primary chromium (Papp, 2001) or manganese (Jones, 2001), both critical to
steel manufacture.  Meanwhile, because of a suite of reasons (high labor costs,
high environmental costs, depletion of high-grade sources, an often hostile
public and regulatory environment) mining in the US continues to dwindle so that
the dependence on foreign sources continues to grow.  For example, primary US
copper production has dropped by ~25% since 1996, so that imports now account
for about one-third of US copper demand (Edelstein, 2001).
	Although the prospect of "imminent resource shortages" now has a somewhat
musty feel, it's simply been postponed by cheap foreign sources.  The author,
who has worked professionally in resources for over 20 years, is used to the
public's being out of touch with resource issues. They seem completely unaware
of the utter US reliance on imported energy and minerals.  But it's extremely
disturbing that so many of the technical community also seem unaware of the
degree to which the brave new high-tech future still relies on coal, oil, and
minerals all ultimately--and messily--dug out from the Earth.
	In addition, the global communications revolution is causing a worldwide
revolution of rising expectations, which puts further pressure on the resource
base.  If nothing else, exporters now have other markets for those resources
than the industrialized countries such as the US.  Eventually domestic political
concerns may make exporting resources politically impossible, as has already
happened in many of the industrialized countries.  The US no longer exports oil,
for example, although it was once the world's biggest oil exporter and is still
the world's second-biggest producer.
	Finally, resource issues are being exacerbated by environmental concerns.
As is well-known, present technology is polluting, and a major part of that
pollution stems from the production and consumption of resources.  Little of the
material excavated from a mine ends up in finished products.  Most is just
waste, which is discarded in vast heaps that (at best) must be expensively
reclaimed and revegetated.  Reaction of some of these materials with water and
air can cause further environmental degradation, as in acid-mine drainage.  The
extraction, transportation, and consumption of petroleum have familiar
environmental hazards ranging from oil slicks to photochemical smog.   Coal is
abundant and cheap, but byproducts of its combustion include acid rain and fly
ash, and its mining is both dangerous and environmentally disruptive.

1.1. The Promethean Paradigm. 
"The habit of neglecting or denying the possibility of technological advance is
a common problem.  ... Snug limits would simplify our future, making it easier
to understand and more comfortable to think about.  A belief in snug limits also
relieves a person of certain concerns and responsibilities.  After all, if
natural forces will halt the technology race in a convenient and automatic
fashion, then we needn't try to understand and control it. 
     "Best of all, this escape doesn't feel like escapism.  To contemplate
visions of global decline must give a feeling of facing harsh facts without
flinching.  ... Genuine courage requires facing reality, facing accelerating
change in a world that has no automatic brakes."
	K. Eric Drexler (1986, p. 166-7)

	But not only is present technology exhausting its resource base while
polluting the environment, it is also ludicrously inefficient and almost
gratuitously dirty.  No law of nature demands that things must be done this way.
	Both the inefficiencies and the pollution come about largely through an
overwhelming dependence on heat: on energy in its most disorganized and wasteful
form.  At least in Western culture, "technology" has been identified with "fire"
since the legend of Prometheus.  But fire turns out to be a clumsy and messy way
of manipulating the world.
	Other inefficiencies, and associated pollution, come about from what may be
termed the "(dis)economies of scale".  Because arranging matter with
conventional technology is difficult, matter-organizers ("factories") have
traditionally been large-scale, capital-intensive affairs to which raw materials
are brought, and from which finished products are exported.  This obviously
requires an enormous transportation infrastructure, one of course that now
subsumes the entire world.  For example, bauxite, a tropical soil, is mined in
Jamaica, sent to Norway for processing into aluminum (the electrolytic process
for extracting aluminum requires vast amounts of electricity, and Norway has
abundant cheap hydropower).  Then the aluminum metal is sent around the world to
be further fabricated into finished products.  If the promise of distributed
fabrication (section 2.2.5.) can be achieved--"matter as software"--this
energy-intensive transportation network will become largely obsolete.
	Another result of the present clumsy approaches to matter arranging is the
unintended manufacture of unwanted byproducts.  With little control over
occurrences at the nanoscale, conventional synthetic processes yield a suite of
molecular products, of which typically only one or two is desired.  The others
become waste that must be discarded or treated.

1.2.  Beyond Prometheus. 
	Molecular nanotechnology (MNT), the design and construction of macroscopic
materials at the molecular level, will play a major part of solving the issues
of both sustainable resource extraction and byproduct mitigation.  Indeed, in
the author's opinion this is by far the most critical near-term application of
MNT.  Furthermore, MNT is the only technology that holds promise for achieving
something like a sustainable First-World standard of living for the entire
world.  Thus forswearing MNT is simply not an option, despite naïve calls for
its "relinquishment". 
	The relevance of nanotechnology to resource issues falls into several broad
categories.  Some applications, involving proto-nanotechnology (e.g.,
nanostructuring by self-assembly) or even microtechnology, are already present
in embryonic form.  These can be expected to become more efficient and more
prevalent as nanoscale control over fabrication matures.  Other applications
will require a more mature level of nanotechnology, even up to full-blown
self-replicating molecular machines.  Thus there are serious timescale issues in
any discussion of MNT applications, and I have focused mostly on nearer term
because of its importance.  I also mention biotechnology where it seems
particularly germane.  After all, biotechnology merely involves exploiting
natural molecular machinery, and its development costs are thereby minimized.

1.3.  Applications of MNT to Resources.
	For energy, the broad categories of MNT applications include:

	1.3.1. Vastly improved efficiency of energy usage:
		 non-thermal energy usage, via nanostructured devices such as fuel
cells;
		 molecularly tailored catalysts, for heightened selectivity and
byproduct elimination;
		 high-strength materials, which will decrease transportation costs,
including access to space;
		 electricity storage and electrosynthesis, both for portable power
sources, and for chemical fuel generation;
		 distributed fabrication, to minimize transportation infrastructure.

	1.3.2. Information-intensive energy extraction:
		 extensive real-time sensing, for better conventional resource
extraction through (e.g.) reservoir management;
		 cheap nanofabrication, which will make practical distributed collection
from diffuse sources (e.g., artificial photosynthesis and low-head hydropower),
as well as better energy management through such devices as "smart windows" and
non-thermal light sources.

	1.3.3. Solid-state energy generation:
		 solar power, via photovoltaics and artificial photosynthesis;
		 thermoelectric conversion, the direct solid-state transformation of
thermal gradients into electricity;
		 piezoelectric conversion, for direct conversion of mechanical energy
into electricity.

	For materials, categories include:
	1.3.4. Molecular separation of elements and molecules:
		 non-thermal separation of solutes from solutions, be they leachates,
natural brines, or wastewater;

	1.3.5. Change of materials mix, as high-strength covalently bonded materials
replace metals in structural applications:
		 "waste as resource", such as biowaste replacing petroleum as a source
of reduced carbon compounds.
		 carbonate rocks (limestone), as the backstop resource of carbon on the
planet;
		 even silicates, the very stuff of a rocky planet, will become more
valuable than nickel steel.

	Finally, a separate note is appropriate on

	1.3.6. Space-based resources:
		 space-based power systems, through cheaper access to space and improved
on-orbit fabrication.
		 space-based materials, which will most likely be rare but important
elements such as
Pt.

	These various points will be elaborated and illustrated below, with
applications to particular energy sources sketched.
	Finally, the author has tried to avoid ideological bias among different
energy sources.  Nanotechnology has much to offer in making practical "green"
energy sources such as solar and wind.  It also has much to offer in making
practical sources such as nuclear power, which are considerably more
controversial in some quarters.  In the author's opinion, humanity can
ill-afford to forego any energy sources, particularly on the basis of a priori
ideological positions.  In the end, choices are likely to be made on the basis
of sheer economics, with due account taken of externalities.  

2.  NANOTECHNOLOGY AND ENERGY
     Free energy--the "capacity to do work", in the formal physics
definition--is the most fundamental resource.  With a sufficient supply of free
energy, almost anything can be carried out.  The first law of thermodynamics
states that energy cannot be created nor destroyed, but the critical constraint
lies in the second law of thermodynamics.  As usually expressed, the second law
states that in an isolated system a quantity called entropy, with units of
energy/temperature, must always increase in any spontaneous process.  It can
then be shown that entropy increase in turn implies that free energy cannot be
recycled fully.  Although total energy remains unchanged, less is available to
carry out useful work.  A certain amount is irretrievably lost, in practice
generally as low-grade waste heat, which on Earth is ultimately radiated to
space.  (Many texts on thermodynamics and physical chemistry discuss the
formulation of the thermodynamic laws at length.  See Moore, 1972, pp. 77-115,
for example.)
	Hence, ongoing sources of free energy--or just "energy," loosely--are
required for all biological and technological activity.   Sunlight is the source
for the biosphere.  In the case of conventional technology a whole suite of
sources is used, but the dominant source is fossil fuels.  These have an energy
density considerably higher than that of sunlight.

2.1. The Promethean Paradigm, I. 
	Nonetheless, energy could be used vastly more efficiently in technological
processes.  Present-day energy efficiencies lie anywhere from factors of several
to orders of magnitude below the thermodynamic limits.  In effect, the high
energy densities of conventional fuels are a "brute force" compensation for the
inefficiencies with which they are used.
	The reason, as foreshadowed in the Introduction, is that energy is largely
used as heat.   "Fuels" are "burned"; that's what fuels are for, at least in
conventional thinking.  Indeed, "energy" often means just heat--a generally
unexamined assumption that might be termed the "Promethean paradigm".  
	It turns out, however, that "burning" a fuel--i.e., converting its chemical
energy into heat--discards much of its free energy, the energy available to do
work.  For an example, let us look at the reaction of H2 and O2 to form water
vapor at standard temperature and pressure (STP, 1 atmosphere and 25C):
	
(R1)		H2 + 1/2 O2 ==> H2O (g).

For processes at constant pressure and temperature, typical of those at the
surface of the Earth, the maximum useful work that can be obtained (or
alternatively, the minimum energy cost of a thermodynamic transformation) is
given by the Gibbs free energy:

(E1)		G = H - TS = U + PV - TS.

Here H is the change in enthapy (change in internal energy U, plus any
pressure-volume work PV done), T is the absolute temperature, and S is the
change in entropy.  Obviously, G is a function of temperature.  Since H and S
are functions of temperature and pressure, G is a function of pressure as well.
	The TS term, which has dimensions of energy, represents the "tithe" paid to
the second law.  It is a measure of the energy that is lost.  Because the TS
term increases with temperature, the irreversible losses also are larger the
higher the temperature at which reaction is carried out.  Hence the maximum
thermodynamic efficiency is given by G/H.  For reaction (R1), for example, the
Gibbs free energy at STP is some -228.6 kJ/mol, whereas the enthalpy is -241.8
kJ/mol.  In the thermodynamic limit, therefore, some -228.6/-241.8 ( 95% of the
total reaction energy is available to do useful work. 	
	This is far more efficient than conventional processes.  Combustion heat
from this reaction could instead be used to drive a heat engine.  A heat engine
turns some of the flow of heat from a hotter body to a cooler body into useful
work, and the maximum efficiency at which it can do so is the Carnot efficiency:

(E2)		eC = (Th - Tc)/Th,

where Th is the absolute temperature of the hot body and Tc that of the cold.
The Carnot limit can be derived rigorously from the second law of thermodynamics
(e.g., Moore, 1972, p. 78-85). 
	For example, the heat from stoichiometric combustion of H2 and O2 to form
one mole of water could be used to form 3.8 moles of steam at 550C (823 K) at a
pressure of ~2.1 x 107 Pa (3000 psi), which are typical operating conditions for
a modern steam-turbine electrical generating plant (e.g., Peltier, 1995).  The
limiting Carnot efficiency, assuming discharge to a reservoir at 25C (298 K), is
then only:

		eC = (823-298)/298 = 63.8%.
		
In practice, however, the Carnot limit doesn't apply even theoretically.  For
engineering convenience the steam will be allowed to expand as nearly
isentropically as possible, as part of a so-called Rankine cycle, and this
expansion will do work by spinning the turbine blades.  The entropy of the steam
at high pressure is 177.0 J/mol K.  Its enthalpy (which is just its heat
content, as no chemical reactions are occurring) is 63.8 kJ/mol (Meyer et al.,
1979).  If this steam is expanded to a pressure of 3386 Pa (1 inch Hg, a typical
value) the enthalpy at the same entropy is 38.0 kJ/mol.  The limiting efficiency
is then only (63.8-38.0)/63.8 ~ 40%.  As can be seen, the Rankine cycle is
considerably less efficient than even the Carnot cycle.
	A great deal of the free energy potentially available has been sacrificed by
converting the chemical energy of the fuel into heat, and this conversion is not
required by the laws of thermodynamics.  These are limits set by particular
technological approaches.  We have had the "heat" crisis, not the "energy"
crisis.

2.2.  Specific Technological Approaches:  MNT and Energy Efficiency.
	Before sketching specific aspects of MNT applications to energy, some
generalizations on the technological issues are worth highlighting:

	 nanoscale structuring at an interface is fundamental to a great many
applications, including:
		- catalyst surfaces (section 2.2.2.);
		- large-area semiconductor p-n junctions, critical to thermoelectric
(section 2.4.1.) and	photovoltaic (section	2.5.10.1.) materials;
		- nanolayered structures, such as for high-performance capacitors
(section 2.2.4.1.b.),	thermoelectric materials (section 2.4.1.), multijunction
photovoltaics (section 	2.5.10.1.), piezoelectric stacks (section 2.4.2), and
sensitized photosynthetic	materials (section 2.5.10.2.).

	 solid electrolytes prove to be widely applicable:
		- critical in fuel cells (section 2.2.1.1.) and supercapacitors (section
2.2.4.1.b.)
		- relevant to intercalation batteries (section 2.2.4.1.a.), smart
windows (section 2.2.6.),	electrocatalysis (section 2.2.4.2.), and also to
ionic separation (section 3.3.1.4.a.).

	Such applications have no moving parts other than electrons and (in some
cases) individual ions.  They therefore seem accessible to very primitive
approaches to molecular structuring such as self-assembly and chimie douce*.  I
elsewhere have suggested that "near-term" nanotechnology is likely to involve
molecularly structured materials rather than full molecular machines (Gillett,
1996a).  Hence it seems fortunate that much energy-related nanotechnology seems
to fall in this category.  However, for bulk energy generation large quantities
of material are required, and this remains a significant challenge to make
cheaply.

	2.2.1. MNT vs. Prometheus. 
	As might be expected, transforming chemical energy without thermalizing it
requires controlling the reactions at the nanoscale.  (Nearly the only advantage
of fire is that it's simple.)  Thus MNT promises to make such transformations
considerably more practical.
	As usual, biosystems are far ahead of technology.  Even though organisms are
loosely spoken of as "burning" food for energy, biological systems are not heat
engines and so are not Carnot-limited.  Rather than converting the energy of
chemical fuel (i.e., food) to heat, and converting only some of that heat to
work as it flows to a cooler body, living things oxidize their fuel
non-thermally, via a cascade of molecular-scale mechanisms that approach the
reversible limit set by the difference in free energy.  Since they carry out
their chemical processes isothermally in the range 25-40C or so, the
irreversible losses are also much lower (that is, the TS term is much smaller). 
		2.2.1.1. Fuel cells.  Probably the best example of a technological
non-thermal conversion device is a fuel cell, which transforms chemical energy
directly into electricity (e.g., Kartha & Grimes, 1994).  Fuel cells first
attracted attention in the 19th century, but although practical fuel cells have
been around since the 1960s, only recently have they begun attracting serious
attention outside high-value niche markets such as spacecraft.  They have been
beset with two problems: high cost due to fabrication issues and limited
lifetimes, and restriction to hydrogen as the fuel. 
		In large part both problems stem from the limited current capabilities
in nanofabrication.  As might be expected, converting chemical energy to
electricity without thermalizing it requires nanostructuring.  In fact, the
failure of 19th century attempts to convert coal electrochemically into
electricity, as well as the long delay in commercializing fuel cells, probably
stem from the need for cheap nanofabrication.  
	The two main nanofabrication issues are first, better catalysts, which
ideally would be 
	 tolerant of common contaminants such as CO and sulfur species;
	 work at room temperature;
	 work with C-containing fuels, not just hydrogen;
	 cheap!
Unsurprisingly, no current catalysts fit these desiderata, most especially the
last.
		The second issue is better solid electrolytes.  As described below, fuel
cells require that ions be conducted between anode and cathode to compensate the
electron flow in the external circuit, and ideally this flow would occur in the
solid state.  
	These two issues will be elaborated below.
			2.2.1.1.a.  Fuel cell catalysts.   The nanotechnological prospects
for catalysts are overviewed in the next section.  Here just the issues relevant
to fuel cells are highlighted.  
			Fuel molecules in a fuel cell must be ionized to react.  Indeed, the
fuel cell can be thought of as intercepting the electrons transferred during the
oxidation reaction to do useful work.  Hence one catalyst must promote
ionization of fuel at the anode, e.g.:

		H2 ==> 2 H+ + 2 e-.

while at the cathode another must promote the reduction of dioxygen by electrons
with as low an overvoltage as possible:

		1/2 O2 + 2 e- ==> O=.

			Platinum-group metals (PGMs) are usually used at the anode because
of their high activity (Hamnett, 1996).  They obviously add significantly to the
cost, however.  Hence catalyst usage must be minimized, which at least involves
nanostructuring of the interface so that precious-metal atoms are deployed most
effectively.  Even better, of course, would be the development of alternative
catalysts not using rare elements.  The current catalysts are also susceptible
to "poisoning" (deactivation through impurities that bind to and block the
active sites), particularly CO and sulfur compounds.  For this reason even
current "low temperature" fuel cells run at temperatures significantly above
ambient (>70C up to several hundred C, depending on the electrolyte), as the
poisoning agents tend to desorb at higher temperatures.  However, not only is
high temperature operation considerable less convenient, it also imposes an
additional thermodynamic cost in lessened efficiency.
			At the cathode, cheaper catalytic materials (e.g., transition metal
oxides) can be used, but rapid reaction without the use of high overvoltage is
still a limiting factor, especially near room temperature (e.g., Hamnett, 1996).
			Because they use specific molecular mechanisms, non-thermal
converters like fuel cells are also much more "picky" about their reactants than
is combustion, as is indeed shown by their sensitivity to catalytic poisoning.
In particular, fuel cells that could use carbon-containing fuels (e.g., CH4,
CH3OH) would be much more adaptable to the current mix of fuels.  Liquid
carbon-containing fuels are also considerably easier to store and handle than
H2.  Hydrogen is particularly inconvenient for transportation applications
because H2 storage imposes a substantial weight penalty.  
			Alcohol fuels would be particularly attractive because they are
common products of bio- and technological syntheses.  They also are more
reactive than hydrocarbons, which makes the catalysis less demanding.
Furthermore, because they are water-miscible they are less convenient for
combustion engines.  Purifying ethanol derived from fermentation, for example,
is a major expense for ethanol to be used as fuel.  Indeed, if purification is
carried out thermally (!), by distillation, there is probably a net energy cost
(Giampetro et al., 1997).  This issue would vanish if diluted ethanol could be
used directly in a fuel cell.
			Unfortunately C-tolerant fuel cells are still experimental.  No
catalyst that can activate C-C bonds at room temperature has yet been found,
which precludes the use of conventional hydrocarbon fuels (Hamnett, 1996).
Poisoning by CO and H2S are also more serious issues.  
Hibino et al. (2000) demonstrated a "low-temperature" (773 K) solid-oxide fuel
cell that is CO tolerant and can operate on small multicarbon hydrocarbons such
as propane, but its "low-temperature" operation is merely relative.
			The C-H bond is also difficult to activate electrocatalytically, but
an experimental fuel cell using CH4 has been demonstrated (Bergens et al.,
1994).   However, methanol is both more reactive and, being liquid, more
convenient than methane.  Unlike ethanol, it also has no C-C bonds.  Hence the
so-called "direct methanol" fuel cell has received a great deal of attention
(e.g., Goodenough, 1992; Hamnett, 1997).   Unfortunately, with present catalysts
the most practical direct-methanol fuel cell requires methanol vapor, so that it
still operates at elevated temperatures.
			2.2.1.1.b.  Fuel cell electrolyte.   All fuel cells also require an
electrolyte between the cathode and anode to conduct ionic species so that they
can react, and fuel cells are commonly classified by the nature of this
electrolyte (Figure 1).  Liquid electrolytes have been used in "traditional"
fuel cells.  Examples include the "classical" alkaline cell, used on the Apollo
missions, in which the electrolyte is aqueous KOH operating at ~70ºC (Hamnett,
1996).  Phosphoric acid and molten carbonate fuel cells are not poisoned by CO2
as is the alkaline cell, but they also operate at considerably higher
temperatures.  



Figure 1:  Fuel cell.  Fuel cells convert the energy of a redox reaction
directly into electricity, in essence by intercepting the flow of electrons from
the oxidant to the reductant to drive an electrical circuit.  For illustration,
in this diagram the reductant (fuel) is H2 while the oxidant is atmospheric O2,
but conceptually any fuel would work similarly.  The practical issue lies in the
electrodes, which must catalyze the ionization of oxidizer and fuel without
significant degradation.  In any fuel cell, an ionic flow in the electrolyte
must also compensate the external electron flow.
 
	Although they have the virtue of simplicity, liquid electrolytes have
serious deficiencies.  Corrosion of the electrodes places major constraints on
their possible compositions, and high temperature operation also decreases
thermodynamic efficiency.  Containing hot, corrosive liquid also does not
decrease the engineering difficulties.
	All-solid-state fuel cells are thus attractive.  They require a solid
electrolyte, a substance that conducts ions while remaining in the solid state
(e.g., Rickert, 1982).  Present-day fuel cells operating near room temperature
that use H2 incorporate a proton-permeable membrane, which conducts protons from
anode to cathode.  The material used in commercial units is Nafion, a solid
fluorocarbon polymer.  It is hardly ideal, being expensive and subject to
degradation through dehydration.  As with any conventional polymer, too, there
is hardly nanoscale control in its synthesis.  The membrane pores are not
uniform in size or distribution, and the surface active sites involved in proton
binding and release are haphazardly exposed.  Thus there is much effort directed
toward finding alternative proton-conducting materials (Peled et al., 1998). 
	Other solid electrolytes are 3-D crystal structures.  These work by having
many ionic site vacancies: under the influence of an electric field an adjacent
ion can hop into the vacancy, leaving a new vacancy behind.  Thus in effect the
vacancy moves, analogous to the motion of a "hole" in a semiconductor.
Obviously (and unfortunately), however, this hopping works best at high
temperature, where the thermal motion of the ions makes them easier to move.
For example, one of the most practical solid electrolytes is presently doped
ZrO2, the basis of so-called "solid-oxide" fuel cells, but it is only effective
>600°C (e.g., Rickert, 1982).  The crystal fundamentally consists of
close-packed O= ions, which include the Zr cations in the interstices.
Substitution of (e.g.) ~10% Y3+, instead of Zr4+, however, leaves some O= sites
vacant to preserve charge balance.  Oxygen anions can therefore migrate toward
the anode.  In this case, too, the oxygen anions are obviously the mobile
species.
	Such high-temperature operation obviously severely limits potential
applications, and as mentioned decreases the ultimate thermodynamic efficiency
substantially.  However, a large number of void-bearing crystalline structures
is known, such as the zeolites, open frameworks based on corner-sharing SiO4 and
AlO4 tetrahedra.  Such "molecular sieve" materials, in which the crystal
structure is traversed by molecular or supramolecular-scale channels, may
provide alternative solid electrolytes for cations, especially for small cations
such as H+.  Unfortunately, currently accessible structures are impractical, as
they have conductivities several orders of magnitude lower than alternative
ionic conductors (e.g., Kelemen et al., 1989).  Part of the problem, however, is
that the zeolites tested were polycrystalline aggregates consisting of tiny (~1
m) unoriented crystals, so the bulk conduction properties probably reflect a
great deal of grain-boundary and surface effects.  Hydration of the cations can
also increase their effective size and thus decrease their mobility
significantly (Rees, 1978).  
	Perhaps more promising are open frameworks formed by redox-active elements
such as W.  These are already known to take up ions, including H+, in some
cases.  Reduction of framework ions draws in such small counterions to maintain
charge balance.  Such intercalation is the basis of lithium batteries (section
2.2.4.1.a.) and electrochromism (section 2.2.6.), and it seems directly relevant
to solid electrolytes.  Zeolites cannot work in this way, because neither Si nor
Al has oxidation states separated by a single electron. 
	In any case, nanoscale fabrication of solid electrolytes, such as megascopic
single crystals, is likely to be important for practical low-temperature fuel
cells. 

	2.2.2. Catalysis.
	Catalysts, of course, cause a thermodynamically favorable reaction to occur
under circumstances where it otherwise wouldn't.  They do so by decreasing the
activation energy to reaction via molecular-scale interactions.  
	Three parameters, which are commonly independent, describe catalysts:  their
activity, or how effective they are at promoting a desired reaction; their
selectivity, or the degree to which they promote the desired reaction only; and
their stability, or resistance to degradation while operating.    All are
important for energy efficiency.   Activity is important for minimizing the
amount of catalyst required.  Selectivity is even more important, because the
greater the degree to which the desired product is formed, the fewer the
unwanted by-products, and thus the less the purification and disposal costs.
Finally, a practical catalyst must remain active over millions of catalytic
cycles before requiring replacement or reactivation.
	Homogeneous catalysts are mixed in with the reactants.  Because they are
molecular, they can be highly selective, but they have the serious disadvantage
of being difficult to separate from the reacting fluid.  Heterogeneous catalysts
are solids, with catalysis taking place at the interface between the solid and
the fluid phase containing the reactants.  They are considerably easier to use
because the catalytic surface (which commonly is a high-surface area solid fixed
in place) is separate from the reactants and products.  Heterogeneous catalysis
is intrinsically more complicated, however, because the selectivity and activity
are intimately bound up with atomistic details of the surface structure.  Often
the very nature of the "active sites", where catalysis occurs, is uncertain.
Traditionally, therefore, catalysis is an "art" that relies on a host of
empirical observations with limited understanding of the actual molecular
mechanisms.
	This is changing, as understanding continues to improve at a molecular scale
(e.g., Somorjai, 1984, 1986, 1990; Bond, 1993).  For example, electronic
modification of the catalytic surface by adatoms, which is critical both in
poisoning and "promotion" of catalytic activity, has been modeled (e.g.,
Feibelman & Hamann, 1985; Maclaren et al, 1986), and the mechanisms of catalyst
activity are now the subject of detailed modeling at the molecular level (e.g.,
Catlow & Thomas, 1992).  Despite such improvements, the empirical basis of
practical catalysts has hardly vanished.  Schlögl (1993) called practical
catalysis "still magic," which occasioned a heated discussion (Thomas &
Zamaraev, 1994).
	Even if the catalytic mechanisms are understood, with conventional
fabrication techniques there's little nanoscale control over structures that
direct those mechanisms.  Hence conventional syntheses, although not quite at
the eye-of-newt-by-the-light-of-the-full-Moon level, are empirical recipes that
commonly are poorly reproducible.  In many cases one batch of catalyst differs
enough from the next that the chemical processes have to be adjusted for each
new batch (R. Phair, pers. comm., 1997).
	Zuburtikudis & Saltsburt (1992) used conventional microtechnological
techniques to prepare well-characterized catalytic surfaces, the idea being to
isolate the variables relevant to catalytic activity.  Although their study had
an experimental focus, being able to structure catalyst surfaces directly at the
nanoscale would obviously be extremely attractive.  This is another obvious
near-term application of nanofabrication techniques that fall short of full
molecular machinery.  Indeed, fabrication of "high technology" catalysts has
been proposed using extensions of conventional microtechnology techniques
(Ribeiro & Somorjai, 1994).  The structuring of catalyst surfaces could also be
an application of the primitive "molecular looms" I have previously suggested
(Gillett, 1996a).  Catalysts have a strong "multiplier effect" because each
catalytic site is used repeatedly.  Hence even a modest amount of
nanostructuring, such as might be carried out by ganged scanning-probe
microscopes, could yield a useful amount of product.  The short-term potential
thus seems especially high.
	One last issue of great practical importance is worth mentioning.  Most
complex compounds, such as pharmaceuticals, occur in "mirror-image" ("chiral")
forms, and generally only one "handedness" is biologically active.  Indeed, the
other so-called "enantiomer" is commonly toxic or otherwise undesirable.  Hence,
there is currently enormous interest in chiral-selective syntheses, an
especially high-value market since it is focused on pharmaceuticals. 

	2.2.3.  High-strength materials. 
	A result from elementary materials science is that macroscopic materials are
far weaker, typically by 1-2 orders of magnitude, than would be inferred from
the strength of the chemical bonds making them up.  The reason is that
macroscopic materials are dominated by microscale and nanoscale flaws, from
crystal dislocations to intracrystalline grain boundaries, and these flaws
control the macroscopic strength (section 3.5.1.).  One of the most promising
applications of nanofabrication is the construction of materials that are
essentially defect-free at the atomic level, such that their strength approaches
the limit set by chemical bonds.  (Such capabilities will also have profound
implications for the mix of desired raw materials, as discussed at length in
section 3.5.1.)  Because strong materials involve no molecular moving parts,
they are also likely to be accessible in the nearer term as their fabrication
need not be carried out by full-blown molecular machines.
	The stronger a material, the less required for a given application.  This
mass savings is likely to have its greatest effect on transportation by
decreasing the weight of vehicles.  Even though the sheer volume of
transportation, particularly for bulk commodities, is likely to dwindle as
nanotechnology matures, travel is unlikely to vanish.  At least in the
relatively near term, moreover, high-strength materials will also be useful for
such macroscopic moving parts as turbines.

	2.2.4. Energy storage.  The storage of electricity is a particularly
pressing problem.  It consists of two parts, of which the first is the less
important:
		2.2.4.1.  Portable power sources.  The inadequacy of current batteries
for powering the welter of new portable electronic (and electric) devices is
well known.  Nanotechnological fabrication can probably contribute here. 
			2.2.4.1.a. Batteries.  A battery is an electrochemical device that
generates DC through a coupled set of redox reactions.  The positive electrode
is reduced:

			Ox + e- ==> Red'

while the negative electrode is oxidized:

			Red ==> Ox' + e-,

the electron transfer being responsible for the generated current.  If the
battery is reversible, application of current in the opposite direction
regenerates the original species Ox and Red from Red' and Ox', respectively. 
			In the reversible limit, the energy obtained is the Gibbs free
energy for the reaction:

				Ox + Red ==> Ox' + Red'

which is also equal to the charge Q transferred times the change in potential,
in volts:

(E3)				G = QV.

For real substances V < ~4 V (e.g., Conway, 1991).
			Intercalation-based batteries, generally using the small Li+ ion,
have attracted a great deal of attention in recent years because of their higher
energy densities.  In such batteries, at least one redox-active electrode has an
open crystal structure with voids capable of intercalating Li+ ions.  In
so-called "rocking chair" batteries (e.g., Tarascon & Guyomard, 1993; Thackeray
et al., 1995; Exnar et al., 1997), both electrodes can take up Li+.  During
discharge Li+ is expelled from one electrode and taken up by the other.
			In commercial batteries, for example, oxidation of Co in LiCoO2
expels Li+, which is taken up in a graphite anode.  Recharge of the battery
involves re-reduction of Co with concomitant take-up of Li+.  In effect an
intercalation electrode greatly boosts the surface area at which redox reaction
can take place, which improves energy storage while also maintaining recharge
capability.
			Many fabrication issues arise with such batteries.  Obviously a
redox-active crystal structure capable of reversibly intercalating small ions is
required.  One mechanism for degradation of such batteries is by irreversible
changes in the crystal structure on deintercalation, through mechanisms whose
molecular details remain somewhat obscure.  The crystal structure must also be
fabricated with near-nanoscale precision.  LiCoO2 turns out to be accessible
with conventional fabrication methods, but others would be attractive if they
could be synthesized reproducibly (e.g., Koksbang et al., 1996).  Overall, it
seems that a capability of fabricating such structures with nanoscale control
would make such batteries cheaper and more practical.
			2.2.4.1.b. Capacitors.  A capacitor is, of course, an alternative
way to store electricity.  The energy of a capacitor is given by:

(E4)		E = 1/2 CV2 = 1/2 QV,

where C is the capacitance, V the voltage, and Q the charge.  Thus, for a given
charge and voltage, a capacitor stores half as much energy as a battery (Conway,
1991).  For a parallel plate capacitor, the capacitance is given by:

(E5)		C =  0A/d,

where 0 is the permittivity of free space,  is the dielectric constant of the
spacing material, A is the plate area, and d is their separation (all SI units).
Obviously, increasing capacitance requires increasing A and/or , and/or
decreasing d.
			Even though capacitors are less efficient than batteries at energy
storage at the same voltage, the voltage of batteries is limited by the
potential of redox reactions.  Capacitors are less limited:  by equation (E4)
the energy storage goes up as the square of the voltage.  However, the voltage
across a capacitor cannot increase arbitrarily because at some voltage the
capacitor "breaks down" through charge crossing the gap d.  Capacitors thus
involve a tradeoff between a larger d to maximize the breakdown voltage, and a
smaller d to increase capacitance.  In electrolytic capacitors, for example, a
thin oxide film is formed atop a metal foil by electro-oxidation, and this film
serves as the dielectric.  The thicker the film, the more resistant the
capacitor to breakdown, but the smaller its capacitance.  In any case, d must be
large enough that electron tunneling is unimportant. 
			Improved dielectrics provide both a way to boost the dielectric
constant and minimize d. Interest in better dielectrics was inspired decades ago
by ferroelectric materials such as BaTiO3.  This compound has a distorted
perovskite structure (Figure 2), which consists of apex-sharing TiO6 octahedra
with the large voids between the octahedra occupied by the Ba++ ions.  The TiO6
octahedra are sufficiently distorted that the Ti4+ ions can "rattle around",
which leads to very high polarizability (i.e., dipole moment), and hence to a
large dielectric constant as well as to ferroelectricity. 


Figure 2:  Perovskite structure.  An oxygen anion lies at each vertex shared
between the octahedra.  Small cations (e.g., Ti4+), denoted by the small open
circles, lie at the center of each octahedron.  The large site between the
octahedra is occupied by a large cation (e.g., Sr2+).  This site may be
unoccupied, as in stoichiometric WO3 (the so-called "ReO3" structure).
Generally the cubic symmetry of the ideal perovskite structure is distorted,
especially if the large site is empty or only partly occupied.

			More recently, a combination of very high surface area and very
small charge separation has been achieved with so-called "double layer"
capacitors.  A "double layer" is formed at the interface between a charged
surface and an electrolyte, because ions of opposite charge are attracted from
the electrolyte to form a compensating layer (Figure 3).  This acts like a
capacitor in which the distance d (equation (E5)) is of the order of tenths of
nanometers.  On discharge of such a capacitor, electrical current flows between
the electrodes while a compensating ionic current flows in the electrolyte.  In
conjunction with high-surface-area nanoporous electrodes, such as carbon
aerogels (e.g., Pekala et al., 1998), this leads to extraordinarily high
capacitances, on the order of farads per gram.  Unfortunately, however, the
voltage is limited by the breakdown voltage of the electrolyte (~1.2 V, in case
of aqueous solutions).  Alternative electrolytes, especially solid electrolytes,
promise to improve this value considerably.  Experimental double-layer
capacitors have been fabricated using a proton-conducting solid polymer
electrolyte (Peled et al., 1998).



Figure 3: Double layer.  A charged surface, such as an electrode, in contact
with an electrolyte attracts an "atmosphere" of compensating counter-ions from
the solution, forming the so-called "double layer."

			Finally, "supercapacitors" (Conway, 1991) or "ultracapacitors" (Liu
& Anderson, 1997) are capacitors that store additional energy through surface
redox reactions, a phenomenon termed "pseudocapacitance" (Huggins, 1996; Conway
et al., 1997).  Typically they are based on high-surface-area transition metal
oxides.  RuO2 has had the best performance, but cheaper alternatives show much
promise (e.g., Liu & Anderson, 1997; Lee & Goodenough, 1999ab).  Specific
capacitance 10-100 times higher than simple double-layer capacitors has been
observed.  They exhibit transitional behavior between capacitors and batteries,
and so seem reminiscent of the "batacitor" postulated by science-fiction author
Philip José Farmer (1971).  Indeed, the electrodes in intercalation-based
batteries also have some analogies (Conway, 1991). 
			Many issues arise in high-performance capacitors that could be
addressed by better control of fabrication at the nanoscale.  Carbon-based
capacitors are plagued by high resistance due to the vagaries of the grain-grain
contacts on which conduction relies.  Individual capacitor elements can be
"stacked" to yield high voltages (e.g., Bullard et al., 1989), but achieving
this requires building a nanolayered structure with repeated "sandwiches" of
electrode-electrolyte-electrode.  Of course, if such capacitors are not to be
restricted to niche markets, the nanofabrication must also be carried out
cheaply.
			2.2.4.1.c. Fuel cells.  Alternatively, perhaps batteries and
capacitors are a dead end except in niche markets.  Maybe laptop computers and
such will be better powered by small fuel cells, which potentially have much
higher energy densities.  One idea would be a metal hydride fuel pack that would
exsolve hydrogen in a controlled manner.  On its exhaustion, it would be swapped
out, just like a conventional battery, and could then be recharged with H2.  If
fuel cells that can use C-bearing fuels (e.g. methanol) can be perfected and
miniaturized (section 2.2.1.1.), portable electronic devices may instead be
simply fueled, rather as one today might fuel a cigarette lighter.
		2.2.4.2.  Electrosynthesis.  The second, related but perhaps more
important electricity-storage issue is the lack of a good way of storing large
amounts of electricity, the immediate output of a great many energy-conversion
systems.  It turns out that a related issue is the long-term synthesis of liquid
fuels.  Because of their ease of handling, owing to their high density coupled
with mobility, liquid fuels dominate transportation applications.  At present,
of course, they are largely obtained from petroleum.  In the future alternative
fuels will be required, even though they will no longer be "primary" energy
sources like petroleum but a merely a convenient form for storage and usage.
		Electricity can be used to drive chemical reactions "uphill", as in the
familiar example of electrolysis of H2O into H2 and O2.  More complex
transformations are termed electrosynthesis.  Electrocatalysis is the term
applied to catalysts activated by electric potential.  Properly it should not be
applied to reactions driven uphill, but this distinction does not seem to be
made consistently.
		A recent example is the ambient-pressure electrosynthesis of NH3
directly from N2 and H2 (Marnellos & Stoukides, 1998).  The current Haber
process for synthesizing NH3 is a paragon of the thermal paradigm, involving
direct reaction of H2 and N2 at high pressure and red heat.  Methanol
electrosynthesis in acetonitrile has also been demonstrated (Pierre & Ziade,
1986).  Perhaps reactions like those carried out currently thermally by
catalysis of syngas (section 2.5.3.2.) could instead be carried out at low
temperatures by electrosynthesis or electrocatalysis.
		Solid electrolytes are again especially important, because they allow
driving and controlling the reactions of ions to form new species.  For example,
in some cases reaction rates can be varied over orders of magnitude by varying
the applied electrical field through the medium of a solid electrolyte (e.g.,
Lintz & Vayenas, 1989; Gur et al., 1991).  Indeed, as electrosynthesis is
formally just the reverse of electricity generation in a fuel cell, it has a
broad synergy with fuel-cell research (e.g., Bebelis et al., 1993), both in
catalysis and in solid electrolytes.
		As discussed in section 2.2.2., the degree of selectivity of catalysts
is of great economic importance in maximizing efficiency and minimizing
byproducts.  The nanoscale modification--"functionalization"--of electrode
surface properties is also a focus of much recent work in electrochemistry
(e.g., Rolison, 1990; Murray, 1992).  One focus of such functionalization is to
induce highly specific binding of reactive species to the electrode.  This is an
alternative approach to highly specific electrocatalytic reactions.

	2.2.5. MNT, nanofabrication, and distributed fabrication.
	Since the onset of the Industrial Revolution, the organization of matter has
become progressively cheaper.  A more recent trend is miniaturization:  the
dwindling cost of organizing matter at smaller and smaller scales.  The progress
from vacuum tube to transistor to ever more densely packed microchips is a
familiar example.
	MNT will continue both these trends, and in those cases it is evolutionary
rather than revolutionary.  The trends are important to energy applications
nonetheless.  How cheaply nanostructured materials can be fabricated will be
crucial in exploiting diffuse energy resources, as emphasized in the discussions
above and below.  Conventional microtechnology, for example, is not nearly cheap
enough for such bulk applications.  A computer is relatively cheap only because
the chip at its heart is so small.  The present high cost of microfabrication is
the major problem that besets photovoltaics, for example (section 2.5.11.).
	A newer trend that MNT will also accelerate is "custom" rather than "mass"
production.  Not only is fabrication becoming cheaper, it is also becoming more
flexible.  This is due, of course, both to increased computer power and to
computer-controlled design and manufacture. 
	In the long term, probably the most important effect of MNT is what might be
termed "distributed fabrication":  the local construction of artifacts out of
local materials.  The idea can be summed up by the slogan "Matter as software".
Technologically this is a considerably more revolutionary advance, but it merely
replicates what biological systems do already.  A plant, for example, is not
assembled by sending off to the leaf factory, and the root factory, and the stem
factory, and so on.  These items are assembled from ambient sources according to
a molecularly encoded digital instruction set:  its DNA. 
	Distributed "fabrication" (or better, "distribution") is already happening
with information products.  Consider journal articles downloaded as .PDF files.
rather than mailed as paper copies, or music distributed as .MP3s rather than as
packaged CDs.  Even this embryonic sort of distribution has the potential of
significant savings in conventional resources through minimizing paper and
plastic usage as well as transportation costs.  Of course, further growth in
such distribution will be as much a function of legal and institutional
constraints as it is of technology.  
	Obviously such a capability will have a huge effect on the global
transportation network.  The enormous energy demand involved in simply moving
around partly organized matter will largely vanish.  Of course, undoubtedly
there will be a transition period.  For example, simple chemical feedstocks are
likely to be locally fabricated before complex machinery, much less foodstuffs.
But since the transportation of bulk commodities such as raw ores currently
accounts for a significant part of global energy costs, even primitive
distributed capabilities will have a non-negligible effect.

	2.2.6.  Passive energy handling and smart materials. 
	Passive energy handling is not so glamorous as the exploitation of raw
energy sources.  Nonetheless, it could make a significant contribution to the
energy mix.  Consider the 0.44 x 1018 J (0.42 x 1015 BTU) expended on household
air-conditioning in the US alone in 1997 (US DOE-EIA, 2001).
	A major part of the climate-control costs in modern buildings simply stem
from waste heat, mostly dumped by the lighting system.  Conventional
incandescent lights exemplify the inefficiencies of the thermal paradigm.  They
are considerably better heaters than illuminators.  A filament at 2500 K,
assuming blackbody radiation, radiates ~96% of its output in the infrared.  Even
conventional fluorescent lights dump a great deal of waste heat.  Thus, simply
converting electricity into visible light with greater efficiency would yield
energy savings, both directly in minimizing illumination costs, and indirectly
in minimizing the costs of dealing with waste heat.  Light-emitting diodes
(LEDs) that emit white light are attracting serious attention now as low-energy
alternatives to conventional lamps (e.g., Tamura, 2000), and again the main
issue in making such alternatives practical is cheap nanofabrication (cf.
Scheel, 2000). 
	Unwanted solar heating is of course also important, at least in summer.
Using electrochromic layers to darken windows has attracted a great deal of
attention (e.g., Baucke, 1985; Oi, 1986; Deb, 1992, 1995). Electrochromic
materials change color under an applied electric field.  Much attention has been
directed toward Li+ intercalation into WO3 (e.g., Passerini, 1989; Rauh & Cogan,
1993; Cogan et al., 1997; Forslund, 1997; Zhang et al., 1997), which has an open
distorted perovskite structure.  Electrochemical reduction of some of the
framework atoms from W6+ to W5+ puts the additional electrons into the
conduction band, where they become delocalized to yield typical metallic
behavior.  The material thus changes from transparent to reflective.  Charge
balance is maintained by intercalation of the Li+ ions.  Reversal of the applied
potential reoxidizes the W to W6+, with expulsion of the Li+ ions and loss of
reflectivity.  Obviously, construction such a window, particularly of
significant area, presents a nanofabrication challenge.
	In "photochromic" systems, the color change is driven instead by photon
absorption, a particularly attractive feature for a self-shading window, or
"smart window."  In the 1960s, swift self-darkening under the sudden intense
illumination of a nuclear explosion was also of military interest for protective
sunglasses (Bertelson, 1971a).  One type of photochromic glass relies on
reversible photodriven redox reactions of dispersed nanoparticles, typically a
silver halide (Araujo, 1997).  An elegant new approach, however (Kostecki &
McLarnon, 2000), combines electrochromic WO3/Li-electrolyte layers, described
above, with a layer of photoactive semiconductor.  The current generated by the
illuminated semiconductor provides the electric potential to drive intercalation
and darkening.  To be practical, however, such a system obviously requires
low-cost fabrication of large-area nanolayered structures. 

2.3.  Information intensive resource extraction.
	2.3.1.  Ubiquitous sensing. 
	When extracting from a subsurface natural energy source such as an oil or
geothermal reservoir, the details of the subsurface structure are obviously of
paramount importance.  Subsurface information, however, is also extraordinarily
difficult to get, particularly in the detail required.  Nanotechnology has
obvious applications here.
	The extraordinary decrease in the cost of computing over the last few
decades has already had a significant effect:  the cost of domestic production
since 1984 has dropped from $14 to $4/bbl, largely through information
technologies (Paul, 2001).  Most of this involves the processing and
manipulation of seismic data to picture subsurface structure, and this trend is
another that will only accelerate as information manipulation continues to
become cheaper.
	Another embryonic trend is cheap distributed sensors based on
microtechnology, such as downhole thermometry, from fiber-optic cables, and
autonomous micro-flowmeters, which are "throwaways" that can be disseminated in
the subsurface (Paul, 2000).  This trend will also accelerate as MNT comes on
line.

	2.3.2. Diffuse resources. 
	A great many energy sources (sunlight, geothermal power) are intrinsically
diffuse, and so practical collectors must be cheap as well as efficient.  This
obviously has major implications for their fabrication:  it must be as simple
and automated as possible.  Hence, another result of cheap nanofabrication will
be to make collecting energy from diffuse sources much more practical, with
enormous potential implications. 
	Distributed fabrication will most likely accelerate this tend.  Biosystems
already carry out distributed fabrication of energy-gathering systems; consider
photosynthesis by green plants. 
	
2.4 Specific Technological Approaches:  MNT and Solid-State Energy Generation.
	Technologies applicable to a range of energy conversion applications are
described in this section.  Technologies specific to a particular application
(e.g., photovoltaics) are described in the appropriate part of section 2.5.

	2.4.1. Heat engines:  thermoelectric power.
	After emphasizing that thermalizing energy is a really silly thing to do, it
now seems ironic to talk about using nanotechnology to extract work from heat.
Nonetheless, heat engines are not going to vanish entirely.  In some cases
(e.g., nuclear fission) a non-thermal approach is very difficult to envision.
In other cases (e.g., OTEC, geothermal power), a thermal gradient exists
naturally, and so one might as well exploit such a gradient as efficiently as
possible.  Science-fiction writers, for example, have long suggested personal
devices powered by body heat (e.g., Heinlein, 1956).
	The most efficient exploitation technologies will probably involve
nanotechnology.  Conventional heat engines use a "working fluid" (steam,
combustion products) whose thermal expansion is used to convert heat into
mechanical work.  Although such technologies are highly mature (consider steam
turbines and internal combustion engines), they are high-maintenance and prone
to catastrophic failure because of the abundance of macroscopic moving parts.
They also require a large temperature difference, and the range of that
difference must include a phase transition or other large volume change, such as
occurs in the reaction of gasoline vapor with air.  Alternative working
fluids--in particular, liquids with lower boiling points than water--are
sometimes used to exploit lower temperatures as in geothermal fields, but the
increase in complexity makes such approaches marginal at best. 
	Direct nanoscale conversion of a thermal difference into another form of
energy is likely to be considerably more reliable as well as more efficient.
Considerably smaller temperature differences are also likely to become usable.
Such devices already exist in embryonic form:  thermoelectric power generators,
which generate an electrical potential at a junction between two dissimilar
materials, typically a p-n junction between two highly doped semiconductors
(e.g., DiSalvo, 1999).  (If a current is applied in the reverse direction, such
junctions act as refrigerators.)  Like fuel cells and photovoltaics, these
devices have been restricted to niche markets, in this case due to both high
cost and low efficiency.  The best current conversion factors are only about 10%
of the Carnot limit.  Even a domestic refrigerator manages ~30% efficiency.
	A dimensionless figure of merit ZT, where T is absolute temperature, is
quoted to characterize thermoelectric materials (e.g., Mahan, 1998; DiSalvo,
1999).  For many years the best values have been ~1, but recently values of
~1.8 (Fleurial et al., 1999) and ~3 (Hagelstein et al., 2002) have been
reported.  A value of ZT ( ~4 would lead to a huge number of new applications
and might even revolutionize power generation. 
	Optimizing thermoelectric materials perhaps no better illustrates the
promise of nanoscale manufacture, because of the daunting number of parameters
that must be optimized, often in inconsistent ways.  For example, electrical
resistance and thermal conductivity must be minimized.  For most materials this
is an incompatible objective.  Approaches taken include attempting to synthesize
structures that minimize phonon transmission while leaving the electronic
structure relatively unaffected.  Nonetheless, even if a bulk substance can be
found that meets these criteria, it probably will be difficult to synthesize via
traditional "shake and bake" approaches, so that nanofabrication ability will be
critical.  Mahan (1998) and Mahan et al. (1997) discuss the issues in finding
better thermoelectric materials.  Hagelstein et al. (2002) describe a "thermal
diode" approach, in which the thermoelectric current is enhanced by thermionic
emission from an adjacent layer.
	Nanolayered structures are more promising theoretically, but their
manufacturing is even more difficult.  As DiSalvo (1999) notes, "The challenge
in making modulated films is to make very thin, atomically perfect layers..."
(emphasis added).  And, it might be added once again, for macroscopic
applications the even greater challenge is doing so cheaply.

	2.4.2. Piezoelectric power. 
	Piezoelectric substances develop an electric potential along particular
crystal axes when deformed; or alternatively, deform mechanically when an
electric field is applied.  Thus they potentially provide a way to transform
mechanical work into electricity without macroscopic moving parts such as
turbines. 
	Piezoelectric materials are widely used at present for oscillators and
sensors, most notably for nanoscale science in scanning-probe microscopes
(SPMs).  However, although their potential for power generation has been
discussed (Galasso, 1969, p. 105-6; Goldfarb & Jones, 1999), it has not been
practical as yet.  One study has shown that the maximum efficiency of
electricity production with conventional packages of PZT (lead zirconate
titanate, Pb(Zr,Ti)O3, another perovskite), a commercially available
piezoelectric material, is only ~10% (Goldfarb & Jones, 1999).  According to
this analysis, a great deal of energy is stored temporarily as strain without
causing electrical generation.  This energy is simply returned mechanically to
the environment.
	Quite apart from the potential of developing and fabricating new
piezoelectric materials, nanofabrication would help make even the present
low-efficiency materials more practical for power generation.  Power generation
would requires nanostructuring on a large scale, not only for the sheer volume
of material required, but because building up significant voltages requires
nanolayered structures, in which thin layers of piezoelectric material are
interlayered with electrodes of alternating polarity connected in series.

2.5 Energy sources.
	Energy sources, both conventional and unconventional, will now be reviewed
briefly, with notes as to the potential applications of MNT.

	2.5.1. Conventional fossil fuels. 
	These still supply some 90% of total energy consumption in the
industrialized world.  In the US, the current mix is about 22% coal, 40% oil,
and 28% natural gas (e.g., Skinner et al., 1999, p. 431).  Proportions in other
industrialized countries are similar, except that coal usage tends to be higher
in the US because of the abundance of US coal reserves.  Coal is not merely a
relic of the 19th century; in 1999 coal generated some 39.8% (311,560 MW) of
electricity in the US, the largest single share (USDOE-EIA, 2000).  (One result
of the 70s "oil shocks" is that petroleum was largely phased out in the US as a
fuel for the routine generation of electricity.)
	Fossil fuels are geologically processed reduced carbon compounds that
originally came from biological material.  They thus represent energy stored by
ancient photosynthesis*.  Hydrocarbons and carbon, because of their stability,
are the ultimate product of processing organic material with heat and pressure
under anoxic conditions.
	Because of their tremendous importance, fossil fuels have an enormous
literature on their formation, their characterization, the geologic settings in
which they occur, and exploration for them.  Oil in particular is the subject of
libraries.  Formation of oil occurs by thermal processing of buried organic
matter in the so-called "oil window" of temperature and pressure.  Too cool, and
the organic matter remains embedded in the rock as a tangle of high
molecular-weight, largely aliphatic polymers termed kerogen; too hot, and all
the long-chain hydrocarbons break down into graphite and natural gas (mostly
CH4).  Of even more importance, however, the rock "plumbing" must be such that
the oil, once formed, can flow into porous "reservoirs" where it can be
extracted conveniently.
	Oil has been of particular value because of the convenience of storing and
handling a high energy-density liquid.  Because of this demand, therefore, it is
not surprising that oil supplies are the most critical.  Indeed, conventional
oil production may peak this decade (Duncan, 2001).  Moreover, the current US
position in particular is precarious.  For 1999, total US oil consumption was
~5.9 x 109 bbl (1 bbl = 42 US gallons = 0.16 m3), of which some 3.5 x 109 bbl
(60%) was imported (calculated from Oil & Gas Journal Databook, 1999 and Oil &
Gas Journal, 2000).  Furthermore, the ~2.4 x 109 bbl that was produced
domestically amounts to the second-largest in the world, only Saudi Arabia
producing more oil.
	Supplies of other fossil fuels are larger, but they are considerably less
convenient.  Coal is environmentally damaging, not just in its mining but in its
usage.  Being nearly pure C, coal yields more CO2 per joule on its complete
combustion than any other fuel.  Combustion of impurities such as pyrite (FeS2)
also contributes SO2 to the exhaust gas.  This is a source of acid rain through
the atmospheric reactions:

(R2)	SO2 + 1/2 O2 ==> SO3; SO3 + H2O ==> H2SO4.

Coal mining is also extremely dangerous for the miners. 
	Natural gas (CH4) is much cleaner, but its domestic supply is also limited
(Duncan, 2001). 
Although considerably larger supplies remain available outside North America, it
is also difficult to transport.
	The basic problem is that conventional fossil fuels are highly limited
although they still account for some 90% of energy usage in the industrialized
world.  Moreover, they are being depleted even while demand for energy grows
worldwide, as people in developing nations aspire to a lifestyle like that in
the wealthy, industrialized countries.  One way or another, this trend will
change sharply in the near future, probably within the next decade.

	2.5.2. Unconventional fossil fuels. 
	As might be expected, conventional fossil fuels represent a tiny fraction of
reduced organic matter, ultimately of biological origin, that is held in
sedimentary rocks.  By far the bulk of this material is much too dispersed to be
realistically considered a fuel prospect*.  Several types of relatively
concentrated deposits, however, have excited interest as potential energy
sources over the years.
		2.5.2.1. "Oil shale".  Oil shale is a sedimentary rock (not a shale, but
a marl) that contains ~15% kerogen (Hinman, 1991).  As noted, kerogen is
"proto-petroleum"; heating kerogen causes depolymerization into shorter length
hydrocarbon chains that can then flow more freely.  Natural oil formation is
thought to involve the thermal degradation of kerogen in the "oil window" of
temperature and pressure. 
		Traditional attempts to process oil shale in essence carry out
"artificial maturation": the rock is heated to crack the kerogen into shorter
aliphatic chains, which can then flow out.  As might be expected, such processes
are dirty and clumsy.  The rock must be fractured extensively, which implies
mining it, and "cooked" in bulk.  This generates a great deal of waste.
Additionally, of course, the concentration of kerogen must be great enough to
make up for the energy expended in mining and heating.
		2.5.2.2. Tar sands.  These consist of a gummy mess of very high
molecular-weight aliphatics dispersed in unconsolidated sand or in sandstone.
They represent the "dregs" left behind when a conventional oil reservoir was
breached by erosion so that all the low-molecular weight compounds escape.
(This, by the way, is the ultimate geologic fate of all oil deposits.  Most oil
is <100 my old, which is quite young in a geologic sense.  Oil deposits are
ephemeral, geologically speaking.) 
		Tar sands in northern Alberta are currently being exploited.  They are
mined, heated to cause the thick oil to flow out, and then these long-chain
aliphatics are "cracked" (i.e., cleaved) catalytically to yield lighter
hydrocarbons typical of conventional oil.  Although the degree of cracking is
not so drastic as would be required for oil shale, the exploitation of tar sands
involves similar environmental issues. 
		2.5.2.3. Clathrates.  Also called gas hydrates, these consist of a form
of water ice having a very open crystal structure that can accommodate small
guest molecules such as methane (Kvenholden, 1993).  Clathrate deposits are
locally abundant beneath the sea floor in high latitudes, but they present
formidable problems for exploitation.  Deep water drilling is difficult and
expensive, and the degree of drilling needed to tap a deposit is uncertain.
Conversely, on depressurization some such deposits may undergo catastrophic
release, with potentially serious consequences.  Naturally triggered
catastrophic release has been speculated to have caused sudden warming events in
the geologic past (e.g., Nisbet, 1990; Hesselbo et al., 2000), as methane is an
excellent greenhouse gas.

	2.5.3.  Applications of nanotechnology to fossil fuels.
		2.5.3.1. Extraction of fossil fuels.  Probably the least glamorous but
most important application of nanotechnology to fossil fuel recovery in the near
term lies in "information intensive" extraction.  This is particularly the case
in "enhanced recovery", which involves techniques to mobilize the oil that would
otherwise remain in the ground.  (Conventional recovery typically leaves some
two-thirds of the oil in the ground (e.g., Montgomery, 1997, p. 305)).  For
example, one form of  "tertiary recovery" involves injecting steam into the
field to mobilize the oil.  In one case (Paul, 2001), monitoring reservoir
temperatures with downhole fiber-optic thermometers while tracking the flow of
oil and steam with ultra-miniaturized flow sensors inserted into the stream led
to substantial (~20%) savings.  Less injected steam was required, and it could
be targeted more effectively.  This is an example of how nanotechnology will
provide incremental improvements of techniques already carried out in embryonic
form with microtechnology.
		Bioprocessing may also be mentioned as an example of using natural
molecular machinery to carry out distributed processing.  Biotechnological
approaches have been under investigation for tertiary oil recovery (e.g., Yen,
1990), and are likely to be both much cheaper and far less disruptive
environmentally.  Indeed, bacterial strains capable of metabolizing kerogen have
recently been reported (Petsch et al., 2001).  Such organisms, or the enzyme
systems derived from them, may be able to carry out low-temperature cracking of
kerogen or of the long-chain aliphatics in tar and asphalt.  These could also
provide considerably cleaner and less energy intensive approaches to processing
of oil shale and tar sands.
		2.5.3.2. Catalysis and syngas.  Catalysis underlies most of the
petrochemical industry at present.  It involves far more than just cracking of
long-chain aliphatics into the C6-C8 molecules that make up gasoline.  Most
simple feedstock organic chemicals (e.g., CH3OH, CH3COOH, HCOOH) are now made by
catalytic processing using "synthesis gas" (syngas), a mixture of CO, CO2 and H2
(e.g., Klier, 1982; Sneeden, 1982ab).   CO and H2 can be made by the steam
cracking of methane (the usual route):

(R3)		CH4 + H2O ==> CO + 3 H2.

or by the water-gas reaction:

(R4)		C + H2O ==> CO + H2.

The latter is the basis of approaches to coal gasification.  It was much used
for fuel gas generation a century or so ago, but now has been largely supplanted
by natural gas.
		The CO/CO2 ratio in the syngas is varied by the water-gas shift
reaction:

(R5)		CO2 + H2 <==> CO + H2O.

Depending on the pressure-temperature conditions, the catalyst(s), and the ratio
of molecular species, a wide variety of simple one- and two-carbon molecules can
be synthesized.  However, catalyst activity, selectivity and reproducibility
would benefit greatly from greater synthetic control at the nanoscale (section
2.2.2.).
		Similar processes provide a way to "fix" hydrogen that might be
generated by other means, such as by artificial photosynthesis (section
2.5.10.2.).  One possibility is the synthesis of methanol,

(R6)		CO2 + 3 H2 ==> CH3OH + H2O, G (STP) = -9.1 kJ/mol,

which has the further advantage of CO2 fixation (Jessop et al., 1995).  Methanol
is not only usable in thermal engines; as a simple, storable, water-miscible
fuel it is attracting attention for next-generation fuel cells (section
2.2.1.1.).
		2.5.3.3. Usage of fossil fuels.  The obvious improvement here is to
release their energy in fuel cells rather than via combustion.  As noted,
however, fuel cells that can use hydrocarbons are in their infancy (section
2.2.1.1.), and are currently much too short-lived for practical applications.
Near-term fuel cell applications in vehicles rely on "reforming" the
hydrocarbons to extract only the hydrogen, which obviously increases complexity
and decreases efficiency. 
		As with any fuel cell, a practical hydrocarbon cell will involve
structuring at molecular scales, and its structure may even be more intricate to
deal with the more complex fuel.  Moreover, even when robust fuel cells are
developed there remains the issue of fabricating them cheaply enough to be
competitive.
		2.5.3.4.  Thermoelectrics.  The use of thermoelectrics to extract
electricity from waste process heat has been the subject of speculation (e.g.,
DiSalvo, 1999).  As already noted (section 2.4.1.), such uses imply not only
thermoelectrics with figures of merit considerably better than those available
at present, but extraordinarily cheap, high-volume fabrication of them. 
		2.5.3.5.  Superstrength materials.  Superstrength materials will have
many applications in conventional extraction.  Besides the obvious improvements
in efficiencies of transport, drilling will benefit particularly.  The "drill
string", the column of rotating pipe tipped by a diamond-studded bit for "making
hole", could be considerably lighter, with consequent energy savings in its
operations. 
		Additionally, much greater depths could become routinely accessible,
although the destruction of permeability at great depth makes the value of this
problematic.  Of course, the fabrication of cheap, megascopic diamond, which is
commonly envisioned as a result of MNT, is also going to make drill bits much
cheaper.
		In the far future, semi-autonomous burrowers, perhaps of microscale or
even nanoscale dimensions, may be able to rove at depth in the crust seeking out
valuable materials.  This prospect, however, seems sufficiently remote that it
does not now merit detailed discussion.

	2.5.4. Nuclear fission. 
	Despite its air of "political incorrectness" and the very real problems of
nuclear waste disposal, nuclear fission remains a major contributor to the
energy mix worldwide, contributing some 26 x 1018 J of electrical energy in 2000
(USDOE-EIA, 2000).  Its contribution, moreover, will almost certainly increase
in the next few years.  The recent electricity shortages in California, for
example, suggest that the political palatability of fission power is likely to
increase substantially if the alternative is blackouts. 
		2.5.4.1. Uranium.  All commercial power reactors use the isotope 235U,
which is fissionable by slow neutrons.  As is well known, it constitutes only
~0.7% of natural uranium, so fuel must be enriched in the fissionable isotope, a
cumbersome and capital-intensive procedure. 
		Most natural uranium consists of the isotope 238U, which is not
fissionable by slow neutrons.  As is well known, however, 238U will absorb a
slow neutron to yield 239Pu after a chain of short-lived  decays.  Plutonium-239
is fissionable by slow neutrons.  It has been extensively used in nuclear
weapons but not for commercial power generation, both because of safety and
nonproliferation concerns.  Since Pu is an element, it can be separated from its
environment by chemical means, and so making a concentration that would be
usable in nuclear explosives is relatively simple.  Enriching 235U, on the other
hand, requires isotopic separation, which is considerably more difficult and at
least heretofore has required an enormous infrastructure.  Pu-fueled power
reactors, however, have been thought likely to be necessary some day.  Indeed,
"mixed oxide" fuels, containing up to 5% Pu obtained from reprocessed spent
uranium fuel, are already used in a few countries (USDOE-EIA, 2000).
		2.5.4.2. Thorium.  A third nuclide of relevance for fission power is
232Th, the only long-lived isotope (t1/2 ~ 1010 y) of thorium.  Like 238U, it is
not fissionable by slow neutrons, but it will absorb a neutron to form 233Th,
which -decays over several days to fissionable 233U.  Although there has been
interest in thorium as an energy source for decades--for many years the US
Geological Survey had a "Branch of Uranium and Thorium Resources"--233U has not
been employed on other than an experimental basis.  Nonetheless it remains a
"backstop" energy resource, particularly since the crustal abundance of Th (7.2
ppm) is about four times that of U (Mason & Moore, 1982, p. 47).

	2.5.5. Applications of nanotechnology to fission energy. 
	Fission is obviously a thermal technology, and it's hard to see how it can
be otherwise.  The energies of nuclear reactions lie orders of magnitude above
those of chemical bonds, so even molecularly structured assemblages seem
unlikely to be able to control nuclear reactions at the atomic level.  However,
this does not mean MNT is irrelevant to fission.  Most obviously, thermoelectric
power generation may prove practical.
		2.5.5.1. Mining and extraction.  The big application of MNT toward
fission fuels, however, probably lies in their extraction and separation.
Fission fuels are "fossil", too, in a sense: they also are dug out of the Earth
and their supply is ultimately finite.  The considerations of concentration and
grade (section 3.1) apply to uranium deposits just as they do to
non-energy-related metals. 
		As described below, a major resource-related application of
nanotechnology lies in the low-energy molecular separation of elements, most
directly from aqueous solution (section 3.3.1.).  As for other elemental
commodities, therefore, nanotechnology could vastly increase the supply of U and
Th.  Leaching of low-grade U- or Th-bearing rocks, for hydrometallurgical
extraction, has already been carried out (e.g., Sayed et al., 1997), and
selective "uranophiles" (highly selective agents for complexing the uranyl ion,
UO22+) have been the focus of much research (Araki et al., 1993; Jacques &
Desreux, 1996; Nagasaki & Shinkai, 1991; Shinkai et al., 1989; Xu & Raymond,
1999).  Uranium also exists in natural brines; indeed, extraction of UO22+ from
seawater has been investigated (Aihara et al., 1992; Kanno, 1981; Kelmers, 1981;
Yamashita et al., 1981).
		A potential "down side" relevant to nuclear proliferation should be
mentioned.  Protactinium-231 (t1/2 ~ 3 x 104 y), which occurs naturally in small
amounts in uranium-bearing materials from the decay of 235U, could also be
extracted.  This isotope is the only naturally occurring one besides 235U with
possible utility in a nuclear device (Franka, 1994), and unlike 235U it could be
purified without the necessity of separating isotopes.
		2.5.5.2. Reprocessing of nuclear fuel and separation of nuclear waste.
The result of nuclear fission is a mishmash of fission products, the lighter
nuclei formed by the fission of the U atoms, mixed together with a set of
heavier U and actinide isotopes formed by absorption of stray neutrons.  The
fission products tend to be strongly -active because they are neutron-rich.  The
actinides are -active and typically long-lived on a human timescale (as with
239Pu).  A fair amount (~1% 235U, as well as ~5% Pu generated in situ;
USDOE-EIA, 2000) of unreacted fuel also remains, but it is unusable because of
competitive absorption of neutrons by the other nuclides.
		Obviously, it would be attractive to separate the components of nuclear
waste, for fuel recovery and also to recover potentially valuable radionuclides.
However, separating--"reprocessing"--nuclear waste is both difficult and
hazardous.  Indeed, such complex mixtures are difficult to deal with even when
not strongly radioactive.  The procedures involve dissolution in strong acid and
then separation via a long and complicated sequence of steps using
precipitation, ion exchange, solvent extraction, and so on, with the additional
difficulty that all reagents and materials used themselves become contaminated
with radioactive material.
		Separation from such solutions could alternatively be carried out
molecularly, by systems like those described in section 3.3.  Indeed, besides
the "uranophiles" described above, effort has been directed toward finding
actinide-specific binding agents, although the focus of these studies has been
therapeutic applications (e.g., Raymond et al., 1984; Kappel et al., 1985).
This is a more difficult problem, however, because of radiation damage to the
nanomechanisms.  They will have to be robust and ultimately probably
self-repairing.  If such systems can be developed, they will make the
reprocessing of nuclear waste considerable more practical.  Indeed, possibly
each reactor installation could reprocess its own waste. 
		Such technologies, however, will also have unfortunate political
ramifications.  They are likely to make it much easier for small groups to
extract nuclear materials from radwaste such as that from conventional power
reactors.  Separation of elemental plutonium (or 233U from irradiated thorium)
should be relatively straightforward, even if the lifetimes of the
nanomechanisms are limited due to radiation damage.  These concerns become
particularly cogent with the recent terrorist attacks against the US.
		2.5.5.3. Isotope separation.  Isotope separation is fundamental to
nuclear materials.  Unfortunately, separating isotopes is much more difficult
than separating elements, and at least in the near term there seem to be few
ways that MNT can yield major improvements, as described in section 3.4.  If MNT
approaches to isotope separation prove feasible, however, this is an obvious
application.  Unfortunately, if for example 235U could be readily separated from
238U it would lead to similar political ramifications as the cheap separation of
Pu.

	2.5.6. Geothermal energy. 
	The interior of the Earth is very hot, and so the thermal difference between
the interior and the surface is potentially a vast energy resource.
Unfortunately, it has been impractical in all but unusual situations because the
thermal gradient is generally so low.  Prohibitive depths would need to be
reached to obtain a thermal difference that could be exploited with conventional
technology.  The typical geothermal gradient in the crust is only ~10-15ºC/km,
so that even reaching the boiling point of water (at STP) lies near the limits
of current drilling technology.  Moreover, conventional geothermal power
generation relies on steam as a working fluid, and to get the volumes required
the steam must have access to a large volume of rock.  Thus the rock must be
reasonably porous, with cracks if nothing else, but the deeper the level, the
greater the degree to which any open space has been closed up by the confining
pressure.
	Therefore, for conventional geothermal power the geothermal gradient must be
considerably elevated, so that temperatures of 100-200ºC can be attained within
a kilometer or so of the surface.  In addition, the rock must both be permeable
and be thoroughly saturated either with steam, or with water sufficiently hot
that a significant fraction will vaporize on exiting to atmospheric pressure.  A
further engineering difficulty ensues because hot water in contact with rock
tends to be highly mineralized, so that scaling from the precipitation of solid
compounds is a serious problem.  The hot water is typically groundwater of
ultimate meteoric origin, so the degree to which it is mineralized tends to
reflect the average residence time of the water.  The longer recharge takes, the
more mineralized the water.
	The requirement of a working fluid leads to further difficulties.  First,
water loss by extraction is often a serious issue.  Hence many geothermal plants
replenish the water by reinjecting the spent, cooled water back into the field.
This has the added advantage of disposing of the water, which is commonly too
mineralized or saline to release to conventional wastewater treatment without
special pretreatment.  Second, many areas of high geothermal gradient contain
little groundwater: so-called "hot dry rocks".  There have been ongoing
experiments on drilling and fracturing such rocks, followed by injection of
water to make steam.  Los Alamos in particular has been conducting such studies
since the late 1970s.  However, the process is not yet practical.  Finally, many
geothermal fields are too cool to produce much steam.  If conveniently located,
these can be used as a source of low-grade heat for (e.g.) space heating or for
such applications as drying food.  Alternatively, in so-called "two-phase"
generation systems, the hot water is used to boil a working fluid with a lower
boiling point, which is used to drive a turbine.  Obviously, however, this both
increases complexity and decreases net power output.

	2.5.7. Applications of nanotechnology to geothermal energy. 
		2.5.7.1. Information-intensive exploitation.  As in so many cases
involving the vagaries of natural systems, geothermal energy is both diffuse and
additionally not very uniform.  Conventionally a geothermal "field" is exploited
somewhat similarly to an oil field.  Wells are drilled for steam production,
based on geologic and hydrologic models of the subsurface.  Hence there is a
great deal of practical similarity to oil production:  because of incomplete
information about subsurface structure, fracture connectivity, flow patterns,
and so on, drilling "dry holes" (wells producing insufficient steam to be
economic) occurs commonly and is a serious risk.  Thus, many of the ideas for
information intensive exploitation also apply to geothermal power. Extensive
downhole monitoring of temperature, to monitor reservoir conditions, is an
obvious first step, as are free-flowing microsensors for (say) determining the
fate of reinjected water.
		2.5.7.2. Automated fabrication.  There is commonly significant variation
in the temperature of steam even from different producing wells in the same
field.  Hence it would be most efficient to use the steam from each individual
well to run its own optimized turbine.  However, at present turbines are much
too expensive to have one atop every well.  Conventionally, therefore, the steam
from all the wells is combined in a manifold that feeds a single turbine.  The
additional expense of the plumbing, and the efficiency losses, are overwhelmed
by turbine costs.  Cheaper fabrication, probably also with superstrength
materials, is likely to change this.
		Finally, as described for oil (section 2.5.3.5.), superstrength
materials are likely to have a pronounced effect on drilling technology in the
near term.  Strong materials are even more important for geothermal drilling
because of the high temperatures encountered.  In the far term, of course,
developments such as automatic burrowers seem possible but do not warrant more
than a mention now.
		2.5.7.3. Thermoelectric materials.  Turbines may not even be used with a
mature nanotechnological approach to geothermal energy.  Perhaps the most
interesting application of thermoelectric materials lies in geothermal power.
In few other cases are the difficulties of a working fluid so evident:
temperature differences are often too modest, or span the wrong range, or else
the fluid is absent completely, as in the case of hot dry rocks.  The subsurface
"plumbing", both natural and artificial, is also critical but difficult to
ascertain, much less modify.  	Unfortunately, a working fluid probably cannot
be eliminated completely.  In the case of a geothermal system, the function of
the working fluid is not merely mechanical:  it also gathers heat from a large
volume, transferring it to a cooler environment so that a thermal gradient
becomes available.  However, better thermoelectric devices would allow replacing
the mechanical turbines used in the current conversion approaches.  Not only
would such devices minimize the problems with fouling, corrosion, and mechanical
mishap that plague conventional geothermal installations, they would allow
direct use of lower geothermal gradients, because the phase change necessary to
drive a turbine would no longer be required.

	2.5.8. Solar energy, indirect sources.
	Heating by the Sun drives the global systems of weather and ocean
circulation.  The fluid flows generated by this huge, albeit inefficient heat
engine represent a source of energy that can be tapped.  Some means of doing so
(e.g., windmills, water wheels) go back centuries.  To a much greater degree
than other natural sources, however, they suffer from what might be termed
"storm risk".  The vagaries of weather demand that power-extracting
infrastructure must be highly overdesigned to withstand the rare freak storms or
floods.  This obviously greatly increases the capital expenditure.
		2.5.8.1. Hydropower.  Electricity generation from large hydropower dams
is certainly a mature technology, and it has a lot of advantages: no emissions,
relatively low maintenance, and little start-up or shut-down delay.  Indeed,
hydraulic head in a reservoir is one of the few ways to "store" electricity at
present.  However, dams seriously disrupt river systems through profound and
generally deleterious effects on riparian ecosystems, and through interruption
of sediment transport.  They also permanently flood farmland, and in any case
reservoir lifetimes are finite, on the order of a few centuries at most, due to
ongoing sedimentation.  Finally, few sites remain for hydropower dams even if
building them were not politically difficult.
		Running water nonetheless still represents a large potential energy
resource.  In the US conventional hydropower accounts for <5% of energy
consumption, but this reflects only ~20% of the total potentially available
(Dowling, 1991).  Of course, the remaining untapped hydropower is so dispersed
that little could be tapped by conventional damming.  Over the last few decades
an alternative approach to hydropower generation has attracted attention:
"low-head hydropower" or "microhydro".  This ranges from turbines driven by low
dams ("weirs") to "run of river" systems.  Like wind power, the latter use the
natural currents to generate power, typically with small turbines or water
wheels (Fraenkel, 1999).
		2.5.8.2. Wind.  Direct electrical generation from wind is a maturing
technology, with a growing number of commercial installations.  The issues are
much as with distributed hydropower:  the energy is diffuse, with large capital
investments required, and with the added risk that not only the speed but the
very direction of the current to be tapped is variable. 
		2.5.8.3. Wave energy.  The surf raised by the wind contains a great deal
of energy, and tapping it has intermittently excited interest over the years.
It is another diffuse resource, however, and the "storm risk" factor is
particularly severe.
		2.5.8.4. Ocean-Thermal Energy Conversion (OTEC).  In tropical areas
there is a significant temperature difference between deep seawater (~4º C) and
the surface (~25ºC).  This, of course, is maintained by solar energy.  Deep
ocean water is cold because it sinks at the poles and flows along the bottom
toward lower latitudes, while the sun directly warms up the surface water.
Although the temperature difference is small, it has attracted much interest
because the reservoirs are both large and fluid.  Thus the issue of dealing with
cubic kilometers of unyielding rock, as in geothermal applications, does not
arise. 

	2.5.9. Applications of nanotechnology to indirect solar energy sources.
		2.5.9.1. Materials implications.  Superstrength materials have many
obvious applications.  OTEC, for example, would profit from long cables that
must hang under their own weight, to reach the sea floor, and yet still be
strong enough to withstand the storm risk.  Superstrength materials also have
obvious applications in turbine and windmill blades.
		Thermoelectric conversion also seems particular relevant to OTEC, given
the small temperature difference, but thermoelectrics capable of exploiting such
small temperature differences do not yet exist.
		Piezoelectric materials may also prove to be practical for power
generation.  A few studies have proposed piezoelectric-based arrays for
generating power from surf (Taylor & Burns, 1983; Hausler & Stein, 1983).  For
example, the change in tension of a float-tethered cable could be exploited as
waves raise the float as they cross it.  Similar systems might be employed for
low-head hydropower, to avoid such devices as turbines, paddlewheels, and others
that use macroscopic rotation.
		2.5.9.2. Automated nanofabrication and diffuse resources.  Exploiting
(say) the surge of surf with an array of piezoelectric cables obviously implies
that the cables are made extremely cheaply, as would an array of micro-turbines
in a "run-of-river" hydroelectric scheme.  Once again the value of cheap
fabrication at the nanoscale in exploiting diffuse resources becomes evident.

	2.5.10. Solar energy, direct sources.
	Solar energy has obvious advantages that have been recognized for nearly a
century (Ciamician, 1912).  It is free, ubiquitous, and overall amounts to an
enormous energy input.  Of course, its disadvantages are equally obvious:
sunlight is intermittent, unreliable, and diffuse.  Because of its diffuse
nature, moreover, the area that must be devoted to energy collection also
becomes an issue.  Land is a high-value commodity in most of the places where
energy is in high demand.
		2.5.10.1. Photovoltaics.  These are the most obvious and familiar
solar-energy technology, devices that when illuminated generate an electrical
current.  Restricted to niche markets for decades, costs have now dropped to the
point that they now seem to be crossing the threshold into large-scale
commercial applications. 
		Photovoltaic devices are based on semiconductors, usually elemental Si.
Absorption of a photon with energy greater than the semiconductor bandgap
promote an electron across the bandgap into the conduction band, leaving behind
a vacancy, or "hole" which acts like a unipositive charge in the valence band.
If this electron-hole pair is formed in the vicinity of a space charge, such as
at a p-n junction, charge separation takes place due to the electric field.
With the proper provisions for collecting the opposite charges, this generates a
potential that can drive an electric circuit.
		Of course, this bald summary leaves a lot of practical details, and
those details seem to be exactly where nanoscale fabrication can offer benefits,
as discussed below (section 2.5.11.).  The "storability" problem is also
particularly urgent.  As has been noted, the storage of electrical energy is
already an issue even with power sources less fickle than sunlight.
Transmission of electricity becomes another issue, because many of the places
where sunlight is most attractive as an energy source, such as deserts, are
remote from population centers.
		2.5.10.2. Artificial Photosynthesis.  This has a literature stretching
back to the early 1970s (!).  It's curious that "solar power" is so often
thought of exclusively in terms of photovoltaics.  Nature doesn't use solar
energy that way, and there is no reason we have to.  Via photosynthesis, green
plants store energy of sunlight directly into chemical bonds.  Indeed, as noted
this is the ultimate source of most of the fuels we depend on today.
		Using sunlight to make fuel solves immediately the storability and
transport issues that arise with making electricity.  Ironically, with such
technology the desert countries of the Mideast also could continue to export
fuel indefinitely.
		Practical technological photosynthesis probably will be based on
semiconductor-driven coupled redox reactions (photo-oxidation and
photoreduction) (e.g., Nozik, 1978; Gerischer, 1979; Bard & Fox, 1995; Lewis,
1995).  In such devices the electron-hole pairs formed by absorption of a
photon, instead of being used to drive electric current, are used as chemically
active agents to drive coupled redox reactions.  The hole is a strong oxidizing
agent, whereas the electron is a strong reducing agent, and they can bring about
reactions that are energetically uphill.  Ever since Fujishima and Honda's
(1972) pioneering work, most effort has been directed toward decomposing water
to H2 and O2, by reactions such as:

(R7a)		H2O + h+ ==> H+ + OH.;

(R7b)		H+ + e- ==> 1/2 H2

(R7c)		OH. + OH. ==> H2O2 ==> H2O +  1/2 O2, 

where h+ and e- are the photogenerated hole and electron, respectively (e.g.,
Tan et al., 1993).  Other investigations have looked at the reduction of CO2 or
inorganic carbonate to simple organic molecules (e.g., Goren et al., 1990;
Halmann et al., 1984; Heleg & Willner, 1994; Kuwabata et al., 1995; Raphael &
Malati, 1989; Thampi et al., 1987; Yamashita et al., 1994), the reaction of N2
with hydrogen-bearing molecules to yield ammonia (Schrauzer et al., 1983;
Ileperuma et al., 1989), and so forth. 
		Parenthetically, because the electron and hole are capable of forming
highly reactive species, such as the OH radicals in the equations above, or
superoxide ion (O2-) from reduction of atmospheric O2, they can also drive
"downhill" reactions that otherwise are kinetically inhibited.  Such species
react readily with many dissolved compounds, in particular organic pollutants
such as halogenated hydrocarbons, aromatic hydrocarbons, cyanide ion, and so on.
Such "photocatalysis" has spawned an extensive literature on using powdered
oxide semiconductors, primarily TiO2, for pollutant destruction (e.g., Matthews,
1988; Wold, 1993; Cunningham et al., 1994; Linsebigler et al., 1995; Riegel &
Bolton, 1995; Fujishima et al., 2000).
		Familiar semiconductors such as Si are not well suited for such
applications, for a couple of reasons.  One reason is bandgap energy:  the band
gap must be large enough, and its edges must be positioned properly, for the
desired reactions to occur.  For oxidation to occur, the energy of the top of
the valence band must lie below that of the species to be oxidized.  Similarly,
a species to be reduced must have an energy below the bottom of the conduction
band (Figure 4).  Silicon is also too chemically reactive to tolerate aqueous
solutions in the presence of air, much less in the case of direct water
oxidation.  It quickly develops a passivating layer of SiO2. 



Figure 4: Semiconductor band gap.  Any semiconductor contains a "valence band"
(V.B.) separated by an energy gap (the "band gap") from a higher-energy
"conduction band." (C.B.)   At absolute zero the valence band is completely
filled with electrons, while the conduction band is completely empty.  In the
immediate vicinity of an interface with another substance, electrons will flow
toward the substance with the lower Fermi level (highest filled energy level)
until the increasing electrostatic charge makes such flow energetically
unfavorable.  This so-called depletion layer (space-charge region, Schottky
barrier) then causes the energy levels of the bands to change, resulting in
"band bending" near the interface.
	Absorption of a photon with energy greater than the band gap promotes an
electron to the C.B. while leaving a positive "hole" in the V.B.  In the
space-charge region, the electric field there causes charge separation to occur.
This is the basis both of photovoltaic devices and of semiconductor mediated
photochemical reactions, as the hole is a powerful oxidizing agent while the
promoted electron is a powerful reducing agent.  A redox couple whose energy
falls in the band gap can be oxidized by the hole (R ==> O + e-), or reduced by
the electron (O + e- ==> R).

		Although efforts have been made to passivate Si and other semiconductors
with electroactive moieties so that they can be used in aqueous solution
(Bocarsly et al, 1980; Bolts et al., 1979; Wrighton, 1978), most work has been
carried out with wide-bandgap oxide semiconductors such as ZnO, SnO2, and
SrTiO3.  By far the majority of this work has employed TiO2, which is cheap,
nontoxic, and abundant, as well as stable indefinitely in contact with H2O in
the presence of O2.  In addition, one polymorph of TiO2, anatase, generates an
electron-hole pair energetic enough to completely decompose water.  The bandgap
of TiO2 is 0.512 aJ (~3.2 eV), which corresponds to a wavelength of ~390 nm.
This lies in the near ultraviolet (UV), which is why TiO2 powder appears bright
white.
		Semiconductor-based photosynthesis is fundamentally different from
natural photosynthesis, which relies on dye-based absorption in conjunction with
a nanostructured molecular chain to carry out charge separation.  The natural
version is both more complicated and less efficient (~1%).
		2.5.10.3. Solar thermal.  The Sun, of course, also heats things up, and
for completeness this aspect of solar power should be mentioned.  Solar heat can
be used not just for space heating, but also for low-grade process heating, some
of whose applications (e.g., drying food) date back to antiquity.   Even apart
from the vagaries of weather, controlling incident solar radiation is difficult
because of the changing position of the Sun in the sky.  Either the changing
incident angles must be lived with, or else complex mechanical arrangements of
mirrors and lenses are required.  An alternative method of control could be
another application of photochromic or electrochromic glass, which could vary
the reflectivity of panels as needed.
		Finally, solar heat could be used to drive thermal engines directly, as
in current experimental projects.  Should such direct use of solar heat prove
convenient, thermoelectric materials are likely to be more attractive than heat
engines using working fluids.

	2.5.11. Applications of nanotechnology to direct solar energy sources.
		2.5.11.1. Nanofabrication.  Present-day solar cells are expensive, and
that expense results from their fabrication costs.  The most efficient material
is crystalline semiconductor-grade silicon, because electron-hole recombination
is minimized, but this material is expensive.  The repeated cycles of
purification through remelting and recrystallization required to make such
material represent the thermal paradigm at an extreme.  Moreover, the ultrapure
silicon must be doped to yield a large-area p-n junction where the photons are
absorbed, as the electrostatic field at this junction (Schottky barrier) causes
the electron-hole separation.  Alternative fabrication approaches could make a
large difference in costs (Green, 1998), and nanofabrication techniques seem an
obvious approach.  
		Another route to cost minimization is through minimizing the material
used.  Electron-hole formation is a near-surface process, so the bulk of the
silicon goes unused.  This has motivated much research into using thin film
materials.  Unfortunately, amorphous films, such as are commonly obtained from
chemical vapor deposition, have very low efficiencies due to heightened
electron-hole recombination.  Also, if the semiconductor layer is too thin,
photon losses due to transmission become significant.  Si is an "indirect"
semiconductor in which promotion of an electron to the conduction band is
formally forbidden (Tan et al., 1993).  Such materials must be 10-100x thicker
than a direct semiconductor to have an equivalent chance of absorbing a photon.
		Optimizing the collection of the electrons and holes to driven the
external circuit also is a nanofabrication issue.  Ideally every electron and
every hole would be collected before it has a chance to recombine, and
approaching this ideal requires a dense network of fine wires (at present
fabricated by conventional microtechnology techniques) on the semiconductor
surface.
		There also is an intrinsic loss in the very nature of photon-induced
electron-hole pair formation in a semiconductor.  The energy of photons with
energy less than the band gap is completely lost because they are not absorbed,
while the energy of those with energy greater than the bandgap is partly lost
because the additional energy is thermalized, merely going into heating the
semiconductor.  Thus the bandgap value has a "leveling" effect, and the best
bandgap value for a given wavelength distribution of the incoming radiation
represents a tradeoff between these loss mechanisms.
		In principle, the efficiency of semiconductor collection can be improved
by using a stack of semiconductors with bandgaps of different energies, termed
multijunction solar cells.  Obviously, however, this efficiency is bought at the
cost of a great increase in complexity, and moreover complexity at the
nanoscale.  Again we are confronted with the issue of fabricating nanolayers
with near-atomic precision.
		For semiconductor-based photosynthesis, some issues are the same as with
semiconductor-based photovoltaics.  Large surface area remains an issue, and
opacity is also an issue because TiO2 and other oxide semiconductors are also
indirect semiconductors.
		However, a number of issues are considerably simpler or even irrelevant.
Junction fabrication is not required because a Schottky barrier is set up
automatically at the semiconductor interface with the aqueous solution, due to
the difference in Fermi levels between water and the semiconductor.  Micro- or
nanoscale wiring to collect the generated holes and electrons is also not an
issue.  However, it has been shown that dispersed surface deposits of a catalyst
such as Pt greatly improve photoredox efficiency, probably through facilitating
hole-electron separation as well as through direct electrocatalytic activity
(e.g., Disdier et al., 1983; Hope & Bard, 1983; Kobayashi et al., 1983; Ohtani
et al., 1997; Sakata et al., 1982).  This obviously increases cost and
complexity.
		An even more critical issue, however, is the utilization of the incident
energy.  A wide-bandgap semiconductor such as TiO2 absorbs <12% of the incident
sunlight, because relatively few solar photons lie in the UV.  One approach
toward improving the energy efficiency is by means of surface attachment of some
other strongly absorbing species, such as a dye.  If the energy difference
between the ground and excited states of the dye straddles the bottom of the
conduction band, an electron from the excited state of the dye can be injected
into the conduction band, where it becomes available for photoreduction
reactions.  Hence, more of the incident radiation becomes usable.
		Such "sensitized" semiconductors have an additional problem in that the
dye layer is too thin to have a large cross-section for photon absorption.  Most
photons, even of the proper energy, continue to be transmitted.  One solution to
this problem is to have a fractal surface, so that due to scattering a photon
has a large chance of being absorbed ultimately (O'Regan & Graetzel, 1991). 
		Finally, there remains the issue of what fuel to produce.  The
disadvantages of H2 storage and delivery remain, although being gaseous it tends
to escape automatically once synthesized in an aqueous solution and so
diminishes separation problems.  Direct, semiconductor-mediated synthesis of
C-bearing fuels such as methanol has been demonstrated, and could use such
attractive substrates as CO2 or inorganic carbonate (section 2.5.10.2.), the
latter abundant in rocks such as limestone (section 3.5.4.).  However,
water-miscible fuels such as alcohols are difficult to separate from the aqueous
synthesis environment, and pure hydrocarbon fuels are much more difficult to
synthesize because of the difficulty of sequestering oxygen-containing moieties
derived from the original water and/or CO2 molecules.
		Although cheap nanofabrication also is likely to be necessary to make
artificial photosynthesis practical, the issues seem less well-defined than they
are for photovoltaics, a reflection of the lesser maturity of the field.
Serious questions about the most advantageous technological approaches still
remain.  Nonetheless, in the author's opinion this seems a much more flexible
and powerful way of harnessing solar energy, because of its automatic solution
of three pressing problems:  the fitfulness of solar irradiation, the
storability of sunlight-derived power, and the imminent exhaustion of convenient
chemical fuels.  Of course, breakthroughs in the storage of electrical energy
might change this assessment, particularly if large new sources of electricity
(as with solar power satellites, section 4.1.) become available.
		2.5.11.2. Dispersed collection.  Sunlight is the sine qua non of a
dispersed energy resource, so not only do we need nanoscale fabrication, we need
cheap nanoscale fabrication.  This is particularly pressing because power
generation requires that substantial area be exposed to the sun.  Land is a
commodity in short supply in most of the industrialized world, particularly in
urban areas where electricity demand is highest.  In such areas, an obvious
source of underutilized area is rooftops.  Covering every roof with a solar
array would obviously go a long way toward resolving issues of both land and
power, although with photovoltaics the storage issues would remain.  For this
reason several countries have targeted rooftop installations, including Japan,
Germany, and the US (Green, 1998).  Roads are another potential source of area
that is underutilized at present, although they are a considerably more
demanding environment and would probably require self-cleaning materials,
possibly through photocatalytic coatings.
		Even solar arrays dispersed on roofs clearly imply considerably cheaper
fabrication than exists at present.  The form of the PV cells could also be
considerably more convenient than the present rigid blocks.  One "wishlist" item
would be a PV material that could be dispersed off rolls like carpeting.
Nanofabrication on such scales is particularly challenging in the case of even
more complex systems, such as multijunction arrays, intended to exploit a larger
percentage of the incident sunlight.  Self-maintenance and repair are a
desideratum, especially because urban roofs are not a very benign chemical or
physical environment.  At this point, however, we are moving beyond near-term
nanotechnology into the realm of molecular machines. 
		Semiconductor-based artificial photosynthesis, by contrast, seems
extremely promising for otherwise attractive solar-power sites, such as deserts,
that are remote from demand centers.  As mentioned, an irony is that such
technologies would allow the traditional oil-exporting nations of the Mideast to
continue exporting fuel indefinitely.

	2.5.12. Tidal energy.
	As the Earth spins on its axis, it rotates through the oceanic tidal bulge
raised by the Moon.  This leads to the well-known rise and fall of the tides,
roughly at 12.5 hour intervals, along the seacoast.  Although tidal currents
have been used to generate electricity on a small scale, they represent another
diffuse resource and thus their contribution has been negligible. 
	Although tidal energy is of course not derived from the Sun--the energy
comes from the rotation of the Earth--the same sorts of issues arise as for surf
power and low-head hydropower.  These were discussed in sections 2.5.8. and
2.5.9.

	2.5.13. Nuclear fusion.
	Controlled nuclear fusion, of light isotopes such as D and 7Li, has been the
"holy grail" of energy research for over 50 years.  Unfortunately, it still
seems remote.  Conventionally, fusion has been attempted by the "brute force"
technique of magnetically squeezing a hot thermal plasma.  An alternative using
high-powered lasers to compress pellets has also been the subject of much
interest.  In all cases the result of fusion will be heat, which would power a
heat engine.
	A novel approach to "aprotonic" fusion using electrostatic accelerators to
carry out reactions such as:

		1H + 11B ==> 3 4He

has been the subject of patents (Bussard, 1989).  To add to its conceptual
elegance, such fusion need not merely yield heat because the -particles (4He
nuclei) could be braked electrostatically, and thus their energy be converted to
electricity directly (Bussard & Jameson, 1995).  Evidently, however, this
approach has proved more difficult than expected.  Certainly no commercial
applications exist.

	2.5.14.  Applications of nanotechnology to fusion energy. 
	Whether nanotechnological fabrication would significantly help the
fabrication of fusion reactors is unclear.  Obviously, the same considerations
of relative energy differences between the nuclear reactions vs. those of the
chemical bonds in any confinement device apply just as strongly here as they do
in the case of fission energy.  Possibly, however, alternatives to the
"brute-force" fusion by thermal plasmas, such as laser ignition or electrostatic
confinement, will benefit from fabrication at nanoscale precision.
		2.5.14.1. Separation of nuclear fuels.  The obvious application of
nanotechnology to fusion, as for fission, again lies in the cheap separation of
nuclear materials.  The nuclides proposed as most promising for fusion (e.g.,
7Li, 11B) are all of rare elements.  On a per-atom basis, B makes up 0.0019% of
the crust; Li 0.0060% (calculated from Mason & Moore, 1982, p. 46-7).  However,
both have been enriched in certain natural brines, which would facilitate their
separation by the techniques described in section 3.3.1.  Indeed, natural brines
are already a source of both elements even with current technology.
		Also as with fission, nuclear reactions depend on the particular
nuclide, not on the element.  Hence isotope separation is again a critical issue
(section 3.4.)  The prospects of separating D from H seem best, because they
have the largest relative mass difference of any two stable isotopes.


3. NANOTECHNOLOGY AND MINERAL RESOURCES
	"If it can't be grown, it has to be mined"
		- a common bumper sticker in northern Nevada

3.1 Minerals and Ores.
"Rich mineral deposits are a nation's most valuable but ephemeral material
possession."
		-- T.S. Lovering, 1969

	Fundamentally, all non-agricultural materials (save water and air) that
humanity uses are obtained from geologically anomalous
enhancements--"ores"--that occur locally on Earth.  Ore is an economic term:  it
is a natural deposit "from which a desired material can be won at a profit".
Thus what constitutes an ore is a function of demand and technology as well as
of geology.  Ores are commonly worked for particular elements, as with most
metals (Fe, Ag, Au, Cu, Zn, etc.), but they include nonmetallic commodities as
well.  They need not even be elemental, but can be particular "pre-organized"
materials, such as diamonds or other gems, mineral fibers such as asbestos (a
valued "strategic material" until quite recently), or chemical compounds such as
salts or limestone.
	Ore deposits typically represent occurrences in which by happenstance some
geologic process, or set of processes, has run to an extreme.  There are three
dominant considerations in whether a deposit is "ore":
	 the grade (the concentration of the desired substance);
	 the quantity ("tonnage") of the deposit;
	 the access (which typically includes legal as well as physical
constraints).
Of these, the grade is nearly always the most important.  Aside from legal
proscriptions, the only exceptions are high volume, low-value commodities such
as sand and aggregate, for which local sources are required because
transportation costs become dominant.  With present separation techniques,
letting Nature do as much of the separation as possible is so cost-effective
that it is worth spending the effort--worldwide!--to seek out highly unusual
deposits in which the concentration of a desired element is as enhanced as
possible.  Such orebodies are also difficult to find because they are highly
localized, so that "exploration" for mineral deposits remains an important
endeavor.  "Prospecting" for ore is still carried out, and even though the
techniques are substantially advanced beyond a burro, pick, and gold pan, the
degree to which the principle remains unchanged is surprising.  Sheer luck still
plays a larger role than might be imagined*.
	Even crustally common elements such as Al or Fe are obtained from their own
ores, not from common rocks.  Aluminum is the most common metal in the crust,
and the third-most common element.  Common rocks typically contain a few percent
to a few tens of percent Al, but they are not commercial Al sources.  As
described in the Introduction, Al is currently extracted from a tropical soil
(bauxite) in which Al oxyhydroxides have been concentrated due to millennia of
leaching by rainwater.  This ore is refined into metal in areas having cheap
electricity, which commonly lie halfway around the world from where mining took
place. 
	Ore deposits are obviously limited, and the trend has been to exploiting
lower and lower grade sources over time.  For example, since the 1880s the
average grade of US copper ores has dropped from ~3% to ~0.5% (Gordon et al.,
1987, p. 42). 
	A further issue is looming here:  even though the desired metal (Cu, say)
makes up only 0.5% of the rock, in conventional ores it is nonetheless
segregated in specific phases--so-called "ore minerals"--that can be separated
from the host rock--the "gangue"Ý--by physical means.  In the case of copper the
typical ore minerals are sulfides, usually containing iron as well as copper, as
for example chalcopyrite (CuFeS2).
	There is a fairly well-defined threshold concentration at which these
separate phases do not form (e.g., Gordon et al., 1987, pp. 26-27).  In that
case the copper remains dispersed in the gangue minerals, into which it
substitutes for more common elements such as Fe and Mg, a phenomenon termed
"solid solution".  This "background" concentration is the form in which most
elements are present in the crust of the Earth, and it is a considerably more
difficult form to deal with.  No current technologies separate elements from
their background values.  
	Conventional exploitation of an orebody also leads to serious environmental
issues.  First, as is obvious from the example of copper, most of an ore does
not consist of the desired element, and the crushed piles of discarded gangue
minerals are at best an eyesore that must be expensively reclaimed.  At worst
they become a significant source of pollution