1 
Full title:   Decarbonizing the Global Energy System 
 
Author: Steve Fetter 
 School of Public Affairs 
 University of Maryland 
 College Park, MD 20742-1821 USA 
 sfetter@wam.umd.edu 
 301-405-6355; FAX: 301-403-8107 
 
Key words:  Climate change, nuclear energy, renewable energy, carbon sequestration 
 
Running title: Decarbonizing the Energy System 
 
Abstract   To stabilize greenhouse gas concentrations at an equivalent doubling, CO
2
 emissions 
must be limited to 5 PgC y
?1
 in 2050, compared to 8 PgC y
?1
 today. This will require the 
decarbonization of world energy supply, in which fossil fuels, which today account for 85% of 
energy supply, are replaced by carbon-free sources. Only five sources are capable of supplying a 
substantial fraction of the required carbon-free supply: biomass, fission, solar, wind, and 
decarbonized fossil fuels. Other sources are either too limited, too expensive, or too unproven to 
make a substantial contribution by 2050. Each of the major alternatives has significant economic, 
technical, or environmental handicaps. Biomass can supply affordable portable fuels, but would 
require vast areas of land, in competition with agriculture and natural ecosystems. Fission is a 
mature technology, but suffers from public-acceptance problems related to the risks of accidents, 
waste disposal, and proliferation. Solar is environmentally benign but expensive and would 
require massive storage or transmission. Wind is economically competitive at windy sites, but 
attractive sites are limited. Fossil fuels are cheap and abundant, but the cost of CO
2
 capture and 
disposal may be high and the environmental impacts unknown. 
 
CONTENTS 
 
Introduction 
Limits on Fossil-fuel Emissions 
Requirements for Carbon-free Energy 
Sources of Carbon-free Energy 
 Sources Unlikely to Make a Major Contribution 
 Biomass Energy 
 Fission Energy 
 Solar Energy 
 Wind Energy 
 Decarbonized Fossil Fuels 
Conclusions 
 2 
INTRODUCTION 
 
In response to concerns that increasing concentrations of greenhouse gases?in particular, CO
2
?
might lead to harmful changes in climate, the Framework Convention on Climate Change was 
negotiated in Rio de Janeiro in 1992. The objective of the Convention, as stated in Article 2, is to 
achieve ?stabilization of greenhouse gas concentrations in the atmosphere at a level that would 
prevent dangerous anthropogenic interference with the climate system? (1). Although the 
Convention established stabilization as a goal, the ?level that would prevent dangerous 
anthropogenic interference? remains undefined. 
 Most studies of climate change and its impacts focus on a doubling of the CO
2
 concentration 
from the pre-industrial level of 275 ppmv to 550 ppmv. According to the Intergovernmental 
Panel on Climate Change, the scientific body established to advise parties to the Convention on 
such matters, a doubling would, over the long term, increase the global-average surface air 
temperature by 1.5 to 4.5?C, with a best estimate of 2.5?C (2). The wide range is due largely to 
uncertainties about how cloud cover, ocean currents, and vegetation would change as the 
atmosphere warmed. More important than changes in average global temperature, but even more 
difficult to predict, are regional changes in seasonal temperature, precipitation, and soil moisture, 
and in the frequency of extreme events such as storms and drought. An appreciation of the 
magnitude of the expected changes in climate can be gained by noting that an increase in global 
temperature of 1.5?C would exceed natural fluctuations over the last ten thousand years, and an 
increase of 4.5?C would rival the increases that occurred during shifts from glacial to interglacial 
periods over the last two million years (3, 4). 
 The European Union has argued that the increase in global average temperature should not 
exceed 2?C, and therefore that stabilization at less than an equivalent doubling of the CO
2
 
concentration should guide global limitation and reduction efforts (5).
1
 Although this is 
reasonable based on current estimates of climate change and its impacts, stabilization below an 
equivalent doubling would require reductions in emissions far beyond existing commitments or 
proposals. In particular, it would require a fundamental transformation or ?decarbonization? of 
the global energy system during the next half century, in which traditional fossil fuels are 
replaced by energy sources that emit little or no CO
2
. Here I review the scale and the timing of 
this transformation and prospects for various carbon-free energy sources. 
 
LIMITS ON FOSSIL FUEL EMISSIONS 
 
Fossil-fuel burning is the most important source of greenhouse gas emissions, but it is not the 
only source. To translate a stabilization goal into limits on future fossil-fuel burning, one must 
take into account greenhouse gases other than CO
2
 and emissions of CO
2
 other than fossil-fuel 
burning.  
 An equivalent doubling is any combination of greenhouse-gas concentrations (above 
preindustrial levels) that would have the same effect on climate as a doubling of the CO
2
 
                                                 
1
 Although not stated explicitly, it is clear from the context that the EU was expressing support 
for stabilization at less than an equivalent doubling (i.e., including greenhouse gases other than 
CO
2
) rather than an actual doubling of CO
2
. 
 3 
concentration. Since a doubling of the CO
2
 concentration produces a radiative forcing
2
 of 4.4 W 
m
?2
, an equivalent doubling is any combination of gases that produces a combined radiative 
forcing of 4.4 W m
?2
.  
 Greenhouse gases other than CO
2
 include methane, nitrous oxide, and halocarbons.
3
 
Anthropogenic emissions of methane and nitrous oxide are due primarily to agricultural and 
waste-disposal activities. Strategies exist for controlling these emissions (6, 7), but it is unlikely 
that global emissions can be reduced significantly taking into consideration expected increases in 
population and per-capita consumption. Moreover, natural emissions of these gases may increase 
as a result of climate change (8, 9). If rates of emission of methane and nitrous oxide remain 
constant at today?s levels, the combined radiative forcing of these gases would increase from 
0.65 W m
?2
 today to about 1.0 W m
?2
 (10). Although emissions of many halocarbons will be 
phased out in accord with the Montreal Protocol and its Amendments, substitute compounds are 
greenhouse gases. Scenarios developed by the IPCC result in a radiative forcing of 0.3 to 0.4 W 
m
?2
 for halocarbons in 2100, compared to about 0.3 W m
?2
 today (10). Gasses other than CO
2
 are 
therefore likely to contribute a combined long-term radiative forcing of about 1.3 W m
?2
. For 
stabilization at an equivalent doubling, CO
2
 would be limited to a forcing of 3.1 W m
?2
 and a 
concentration of about 450 ppmv.
4
 
 In order to stabilize the CO
2
 concentration at 450 ppmv, total emissions must be reduced to 
5?6 PgC y
?1
 by 2050 and about 3 PgC y
?1
 by 2100 (10?12). The rate at which stabilization is 
achieved has little effect on this result. This includes emissions from all anthropogenic sources, 
not just fossil-fuel burning. In 1996, cement manufacture released 0.2 PgC y
?1
 (17); this can be 
expected to increase to about 0.5 PgC y
?1
 by 2050.
5
 Schimel et al (10) assume that land use 
                                                 
2
 ?Radiative forcing? is the change in the energy balance of the climate system that would result 
from an instantaneous change in greenhouse-gas concentrations. In equilibrium, the Earth 
radiates infrared energy to space at the same average rate as it absorbs solar energy. If the CO
2
 
concentration were suddenly doubled, infrared radiation initially would be 4.4 W m
?2
 less than 
solar absorption. Over time, the climate system would adjust (e.g., surface temperatures would 
increase) until the balance between infrared radiation and solar absorption was re-established. 
3
 Aerosols can be ignored in this context. Although aerosols may have a significant cooling effect 
that partially offsets the warming due to greenhouse gases, their residence times in the 
atmosphere are very short compared to greenhouse gases (days versus decades). Thus, unlike 
greenhouse gases, any effect of aerosols on climate is regional and depends on the current rate of 
emission in that region. Reductions in coal burning would lead to immediate reductions in 
aerosol concentrations (and the associated cooling effect) while having little effect on CO
2
 
concentrations for many decades. Moreover, efforts to control air pollution and acid deposition 
will lead to reductions in aerosol emissions independent of reductions in coal burning as 
pollution-control technologies diffuse to developing countries.  
4
 The radiative forcing, ?F, associated with a CO
2
 concentration C is given by ?F = 6.3 
log
e
(C/C
0
) W m
?2
, where C
0
 is the preindustrial concentration (275 ppmv). The concentration 
that produces a forcing of 3.1 W m
?2
 is given by C = 275 e
3.1/6.3
 =  450 ppmv. The uncertainty in 
these estimates is on the order of ?0.4 W m
?
? or ? 30 ppmv. 
5
 Per-capita cement production increased at an average rate of 3.5% y
?1
 over the last 50 years and 
2% y
?1
 over the last 25 years (17). Even if per-capita production grows at only 1% y
?1
 over the 
next half century, total emissions (calcination only) would reach 0.5 PgC y
?1
 by 2050. 
 4 
changes resulted in net average emissions of 1.1 ? 0.7 PgC y
?1
 during the 1980s, but a more 
recent evaluation by Houghton (13) gives net emissions of 2.0 ? 0.8 PgC y
?1
 over the same 
period. Future emissions are a matter of speculation; values in the literature range from a net 
release of over 2 PgC y
?1
 to a net uptake of over 2 PgC y
?1
 in 2050, depending rates of 
deforestation and reforestation and the carbon content of vegetation (14, 15).
6
 Changes in climate 
might cause large releases of carbon during the next century if mature forests die before they are 
replaced by new forests, if higher temperatures promote the decay of dead organic materials at 
high latitudes, or if drier conditions increase the frequency of forest fire. For an equivalent 
doubling, it is estimated that such processes could produce net emissions of 0.1?3.4 PgC y
?1
 
during the next century (8, 9, 16).
7
 For simplicity I will assume that, in the context of 
stabilization at an equivalent doubling, net emissions from land-use change and climate change 
will be roughly zero by 2050, but with a large uncertainty. 
 Thus, in order to stabilize greenhouse gas concentrations at an equivalent doubling, fossil-
fuel carbon emissions must be limited to 5 ? 2 PgC y
?1
 in 2050 and 2.5 ? 1 PgC y
?1
 in 2100, 
equivalent to an energy consumption of about 300 ? 120 and 150 ? 60 EJ y
?1
, respectively.
8
 For 
comparison, 1997?98 fossil-fuel carbon emissions and energy consumption were 6.4 PgC y
?1
 and 
340 EJ y
?1
, respectively (17, 18). Fossil-fuel combustion can continue to increase for another 
decade or two, but after that point it must begin a long and steady decline. 
 
REQUIREMENTS FOR CARBON-FREE ENERGY 
 
World energy consumption is expected to grow substantially over the next century, driven by 
increases in both population and per-capita consumption in developing countries. Figure 1 shows 
several scenarios of future energy consumption (19?22). These are ?reference? scenarios?that 
is, they assume no special policies to decrease energy consumption or carbon emissions (but they 
do take into account expected improvements in energy efficiency and price increases caused by 
the depletion of oil and gas resources). With the exception of the ?IS92c? scenario, which 
assumes essentially no population growth and limited oil and gas resources, total primary energy 
consumption is expected to double or triple over the next fifty years, from about 400 EJ y
?1
 in 
1998 to 750?1250 EJ y
?1
 in 2050. If fossil-fuel consumption was limited to 300 EJ y
?1
 in 2050 to 
                                                                                                                                                             
Opportunities to reduce calcination releases per ton of cement produced (e.g., by using waste 
lime) are limited. 
6
 Note that some of these estimates use different base year (1990) emissions. When they are 
normalized to the same base year emissions (1.1 PgC y
?1
), the range is of ?2.2 to 2.5 PgC y
?1
 in 
2050. 
7
 This is the net release of CO
2
 due to climate change, after subtracting the increase in carbon 
storage due to fertilization from increase CO
2
 concentration. The increased uptake due to 
fertilization is included in the calculations of emission pathways that lead to stabilization (10-
12). 
8
 About 25, 20, and 14 gC are released per kJ of thermal energy released from coal, oil, and gas, 
respectively. For the current mix of fossil fuels (30 percent coal, 45 percent oil, 25 percent gas), 
the average is about 19 gC/kJ of fossil energy. This might fall as low as 17 gC/kJ over the next 
several decades as users switch from coal to natural gas. Thus, 5 PgC is equivalent to about 300 
EJ. 
 5 
permit stabilization at an equivalent doubling, then carbon-free energy sources would have to 
supply the difference: 400?900 EJ y
?1
, at prices not much higher than the expected prices for 
fossil fuels at that time. 
 Energy consumption, and requirements for carbon-free energy, could be reduced by various 
interventions not included in the reference scenarios. These could take the form of taxes on 
energy or fossil fuels, emission quotas or tradable permits, subsidies for energy-efficiency 
improvements, or energy-efficiency standards. Economists generally regard a tax as the most 
efficient mechanism. Edmonds et al (23) calculate that a tax of $100 tC
?1
, escalating to $325?450 
tC
?1
 in 2050 and $750?1200 tC
?1
 in 2100, would be needed to stabilize CO
2
 concentrations at 
450 ppmv, depending on the prices of various energy-supply alternatives. The ?ecologically 
driven? scenarios developed by Nakicenovic et al (20), which result in a CO
2
 concentration of 
about 450 ppmv in 2100, assume a carbon tax of $150 tC
?1
 in 2050 (in addition to energy taxes 
amounting to 100 percent in developing countries and 300 percent in developed countries) 
increasing to $400 tC
?1
 in 2100 (20; L Schrattenholzer, personal communication). Total energy 
consumption in these scenarios is about 600 EJ y
?1
 in 2050, with 250?300 EJ y
?1
 supplied by 
carbon-free sources.  
 Taxes of this magnitude are very high by current U.S. standards. Existing energy taxes are 
equivalent to $30 tC
?1
 in the United States (24); a tax of $100 tC
?1
 would more than triple the 
current price of coal delivered to U.S. utilities and increase the retail price of coal-fired electricity 
by about 30%.
9
 Although such taxes need not have strong negative economic effects if they are 
phased in slowly and the revenues are recycled efficiently (26), polls consistently indicate that a 
large majority of Americans would be unwilling to accept taxes of this magnitude to address the 
climate change problem (27). Developing countries are likely be even less receptive to such 
taxes. 
 Thus, stabilization at an equivalent doubling will require roughly 600 EJ y
?1
 of carbon free 
energy by 2050 at prices comparable to those of fossil fuels, or 300 EJ y
?1
 at prices two to three 
times higher (with substantial carbon taxes on fossil fuels). For comparison, in 1998 carbon-free 
sources supplied less than 60 EJ y
?1
 (18). Carbon-free energy supply must therefore grow by a 
factor of five to ten over the next 50 years (an average grow rate of 3?5% y
?1
 over this period), 
from 15 percent of total commercial supply to 50?75 percent in 2050. 
 The transition to carbon-free sources will be the third transformation in world energy supply. 
The first shift, from firewood to coal, took place from 1850 to 1900. The second shift, from coal 
to oil and gas, occurred from 1925 to 1975. In these first two shifts, it took 50 years for the 
emerging source to go from 10 to 60 percent of total supply. The third major shift, from fossil 
fuels to carbon-free sources, will occur from 2000 to 2050?if we decide to take seriously the 
goal of stabilizing greenhouse gas concentrations at a reasonable level. 
 
                                                 
9
 Since coal is 75% carbon, a tax of $100 tC
?1
 would add $75 t
?1
 to the price of coal and $0.026 
kWh
?1
 to the price of coal-fired electricity (assuming a heating value of 29 GJ t
?1
 and an average 
net conversion efficiency of 35%). For comparison, in 1997 the average price of coal delivered to 
U.S. utilities was $29 t
?1
 and the average retail price of electricity was $0.085 kWh
?1
 (25). 
 6 
SOURCES OF CARBON-FREE ENERGY 
 
Only two sources of carbon-free
10
 energy?hydropower and nuclear fission?currently produce a 
significant fraction of world supply, with each accounting for about 27 EJ y
?1
 or 7 percent of 
commercial primary energy in 1998 (18). All other carbon-free sources?geothermal, wind, 
solar, and commercial biomass?together supplied only about 4 EJ y
?1
 in 1998 (18). Traditional 
biomass fuels are not included in this accounting; although they may provide up to 60 EJ y
?1
, 
much of this is fuelwood that is harvested in a unsustainable manner, resulting in a net release of 
CO
2
 (28). Carbon-free energy production has been growing recently at only about 2% y
?1
?much 
less than the 5% y
?1
 rate needed to stabilize greenhouse-gas concentrations at an equivalent 
doubling without resort to very high taxes. Moreover, most of the recent growth is due to an 
expansion of nuclear and hydro capacity, which is expected to taper off in the coming decades. 
 The list of potential carbon-free energy sources is long; in additional to those listed above, 
there is fusion, various forms of ocean energy, and ?decarbonized? fossil fuels. Unfortunately, 
each of these sources has significant technical, economic, and/or environmental drawbacks that 
must be overcome if it is to supply a substantial fraction of world energy supply. Although it is 
impossible to predict which source or combination of sources will prevail, it is possible to say 
which will not. As discussed below, hydro, geothermal, ocean, and fusion energy almost certainly 
will not supply a large fraction of world energy before 2050. The sources with the greatest 
potential in this time period are nuclear fission, solar photovoltaic, decarbonized fossil fuels, and, 
to a lesser extent, wind and commercial biomass. Table 1 summarizes the current and potential 
contributions of various carbon-free energy sources. 
 
Sources Unlikely to Make a Major Contribution 
 
Hydropower.  Hydropower currently is the largest carbon-free source of commercial energy. In 
1998, hydro produced about 2600 TWh
11
 of electricity?19 percent of global electricity 
production and 7 percent of primary energy (18). Global hydroelectric production experienced 
strong growth from 1900 to 1970, but growth has slowed to about 2 percent per year over the last 
decade. Future expansion is limited by the availability of economically attractive sites and, 
increasingly, by concerns about the environmental and social impacts of dams (29). Scenarios of 
future energy supply assume that hydro will contribute less than 5000 TWh in 2050, and that its 
share of total energy supply will remain about the same or decline (20). 
                                                 
10
 The term ?carbon-free? here refers to energy production with very low net emissions of CO
2
 to 
the atmosphere. Of course, no energy source can be truly carbon-free if it involves structural 
materials (steel, cement, etc.) or fuels that are processed using energy derived from fossil fuels. 
This ?embedded energy,? as it is sometimes called, is usually a small fraction of energy produced 
by the renewable and nuclear energy plants during their lifetimes. 
11
 A terawatt-hour (TWh) is a billion kilowatt-hours or 10
12
 watt-hours, and is equal to 0.0036 
EJ. The primary energy content of electricity from non-thermal sources, such as hydro and wind, 
is the energy needed to produce the same amount of electricity in a thermal plant. At today?s 
average efficiency (33%), 1 TWh = 0.011 EJ
p
; at the average thermal efficiencies expected 
several decades from now (40%), 1 TWh = 0.009 EJ
p
. 
 7 
 Without regard to environmental or economic constraints, global hydroelectric production 
potential is estimated at 15,000 to 19,000 TWh y
?1
 (30). The historical experience in the United 
States, Europe, and Japan, where hydroelectric production has leveled off, indicates that 40 to 65 
percent of this technical potential ultimately could be exploited. Thus, hydro ultimately could 
produce no more than 10,000 TWh y
?1
 or 100 EJ
p
 y
?1
. 
 
Geothermal energy.  An enormous amount of heat?nearly 10
13
 EJ?is stored in the Earth?s core 
from its formation 4.5 billion years ago and from the decay of radioactive isotopes in the core. 
More than 10
7
 EJ lies within a within a few kilometers of the surface and is theoretically 
accessible using current drilling technology. Because of the low thermal conductivity of rock, 
heat flow to the surface is very small?about 1000 EJ y
?1
 or 0.06 W m
?2
. The temperature of 
accessible rock generally is below the boiling point of water, making it difficult to extract heat 
energy economically. However, near tectonic plate boundaries molten rock from the core comes 
much closer to the surface, making the overlaying rock and any water trapped therein much 
hotter. Regions of concentrated, high-temperature water and steam (?hydrothermal? reservoirs) in 
shallow rock are far more easily exploited for electricity production, but they represent less than 
0.1% of the total resource. 
 Geothermal energy experienced rapid growth in the early 1980s; during the 1990s, however, 
production has grown at only 2% y
?1
. In 1998 geothermal contributed about 0.6 EJ
p
 to world 
energy supply?41 TWh of electricity and about 0.15 EJ of direct-use heat (18, 31). Nearly all of 
this was extracted from high-temperature hydrothermal reservoirs.  
 Because heat is withdrawn from the surrounding rock much faster than it is replenished by 
conduction from below, geothermal energy is an exhaustible resource. The total amount of heat 
that could be extracted from high-temperature hydrothermal reservoirs is on the order of 5000 
EJ
p
?less than oil or gas resources.
12
 Only a fraction of this could be extracted economically. 
Thus, hydrothermal energy is unlikely to be an important global energy source. 
 The amount of heat stored in hydrothermal reservoirs is tiny compared with the amount 
stored in hot, dry rock. The problem is delivering that energy in a useful form and at an 
acceptable price. The basic concept is to drill to parallel wells several kilometers deep into the 
rock and to fracture the rock between the wells. Water injected down one well is forced through 
the fissures in the hot rock and pumped to the surface via the other well. The technology is in the 
experimental stage and commercial feasibility is far away. Drilling to the required depths is 
expensive, but the most difficult problem is to create a stable fracture network of the proper size 
and porosity (31, 33). Otherwise pumping requirements or water losses can be unacceptably high 
or the rock can cool off too quickly. Even if these technical problems can be solved, long-term 
tests would be required before commercialization could begin. For these reasons, it seems 
unlikely that hot-rock geothermal could supply a significant fraction of world energy demand by 
2050.  
 
                                                 
12
 The accessible high-temperature (>150 ?C) hydrothermal resource in the United States is 
estimated at 4000 to 6000 EJ (32); based on this, the global resource is roughly 40,000 EJ. If one-
fourth of the accessible resource could be extracted and used to produce electricity at half the 
average efficiency for thermal power plants, the contribution to world primary energy would be 
roughly 5,000 EJ
p
. The amount that could be extracted economically would be smaller. 
 8 
Ocean energy.  Large amounts of energy are stored in the oceans in tides, waves, heat, and 
salinity gradients. Ocean energy is hampered by high capital costs, by the difficulty of 
maintaining equipment in corrosive marine environments and protecting it from storms, by low 
energy densities, conversion efficiencies, and capacity factors, and by geographic constraints that 
put most of the resource far from population centers. For these and other reasons, the oceans are 
unlikely to become a significant source of commercial energy for the foreseeable future. 
 Tidal energy is harnessed by building a dam across an estuary having a large tidal range. 
Because of its similarity to hydropower, the technology is fairly mature. Several small tidal-
power facilities currently are in operation, producing about 0.6 TWh y
?1
 of electricity (34). The 
total amount of energy dissipated by tides worldwide is over 200 EJ
p
 y
?1
, but only 5 to 10 EJ
p
 y
?1
 
occurs at sites that are technically exploitable (i.e., with a mean tidal range greater than 3 m). Of 
this, perhaps 10 to 50 percent could be exploited at reasonable cost. The desire to avoid adverse 
impacts on the ecology of estuaries could further limit the development of tidal power. 
 Technology to extract energy from ocean waves is still in the experimental stage. Although 
the total resource is comparable to that of tidal energy, there are no locations where wave energy 
is naturally concentrated. Most of the wave-energy resource is located offshore in deep water, but 
the estimated cost of electricity from offshore devices is two to three times higher than for 
shoreline devices (35). Capital costs are likely to be very high, as would be the cost of insuring 
against storm damage.  
 The temperature difference between warm surface water and cold deep water, which in the 
tropics is as high as 20?C, can be used to produce electricity. The total resource is on the order 
10,000 EJ
p
 y
?1
, but the amount that could be exploited (economics aside) is less than 100 EJ
p
 y
?1
. 
Although the feasibility of ocean thermal energy conversion (OTEC) was demonstrated in the 
1930s, the engineering difficulties of deploying the technology on a commercial scale are 
immense. The small temperature difference results in conversion efficiencies of only 2.5%, 
which in turn requires very large flows of water and huge pumping requirements. Because OTEC 
is restricted to deep, tropical waters, electricity would either have to be transmitted via long 
undersea cables to tropical countries, or used to produce electrolytic hydrogen. Preventing 
corrosion and storm damage to the plant also would be challenging.  
 Energy also is stored in the form of salinity gradients. The difference in salinity between the 
Earth?s river flow and the oceans is equal to 200 EJ
p
 y
?1
. Available technologies to convert this 
energy into electricity are extremely expensive, however. 
 
Fusion energy.  Nuclear fusion?the joining of light nuclei to form more-stable heavy nuclei?is 
the energy source of the stars. The energy potential of fusion is virtually unlimited. Using the 
fuels that are easiest to ignite, the current rate of global energy consumption could be sustained 
for 10 million years. Achieving the controlled release of this energy has proved extraordinarily 
difficult, however. For fusion to occur, nuclei must be brought very close together?close enough 
to overcome the strong repulsive force of the positively charged nuclei. The two main approaches 
are inertial and magnetic confinement. In first scheme, pulsed lasers or particle beams are used to 
squeeze a tiny pellets of fusion fuel, triggering a series of small nuclear explosions. In the second 
scheme, nuclei are held in a magnetic ?bottle? long enough, and at sufficiently high temperatures, 
so that there is a significant probability that fusion will occur. After the expenditure of tens of 
billions of dollars over more than forty years, both approaches are on the threshold of 
 9 
demonstrating ?break-even?: the release of more energy by fusion reactions than is consumed in 
squeezing or confining the fusion fuel.  
 After break-even is achieved, several additional decades of research and development would 
be needed to yield a device suitable for commercial energy production. The most optimistic 
researchers agree that a demonstration reactor will not operate before 2025. Fusion may one day 
prove to be society?s ultimate energy source, but it is unlikely that it will be available in time to 
contribute to the stabilization of greenhouse-gas concentrations. 
 
 This leaves five carbon-free energy sources that could potentially make a substantial 
contribution to world energy supply in 2050: biomass, fission, solar, wind, and decarbonized 
fossil fuels. Below I review the theoretical and practical potential of each of these sources, and 
explore the technical, economic, and other obstacles that would have to be overcome if they are 
to become major sources of energy.  
 
Biomass Energy 
 
Biomass?wood, crop residues, dung, and other combustible wastes?is the main source of 
energy for a majority of the world?s population. Because most biomass fuels are not traded on 
world markets, consumption is highly uncertain. Estimates range from 15 to 65 EJ y
?1
, or 4-15% 
of world energy consumption (20, 28, 36-37). 
 The source of all biomass is photosynthesis, in which plants use solar energy to produce 
carbohydrates from CO
2
 and water. The burning of biomass does not lead to a net emission of 
CO
2
 so long as biomass is grown at the same rate as it is consumed. Unfortunately, this is not the 
case today. About 60% of biomass energy is supplied by fuelwood, most of which is harvested in 
an unsustainable manner, resulting in deforestation, loss of natural wildlife habitat, and a release 
of CO
2
 into the atmosphere. Roughly 200 Mha would be required to supply this much fuelwood 
in a sustainable manner?twice as much as now exists in all forest plantations. Moreover, 
biomass typically is burned inefficiently, resulting in high levels of indoor and outdoor air 
pollution. 
 Biomass energy can, however, be a modern, environmentally benign energy source. In the 
United States, biomass supplied about 3 EJ
p
 in 1998, including about 60 TWh of electricity (18, 
38). Most of this was supplied by wood waste, and, to a lesser extent, agricultural waste, solid 
waste, landfill gas, and about 5 billion liters of ethanol produced from corn. Brazil produced 
about 13 billion liters of ethanol and 10 TWh of electricity from sugar cane in 1998 (18, 39).  
 Biomass has several advantages over other carbon-free energy sources. First, biomass is 
versatile. Biomass can be used to produce solid, liquid, and gaseous fuels as well as electricity; 
its ability to provide transportation fuels is particularly important. Second, the technology for 
producing biofuels is mature and is available even in the poorest countries. Third, relatively 
modest advances in production or increases in fossil fuel prices could make biofuels 
economically competitive without carbon taxes. Biomass can be produced at estimated delivered 
costs of $1.5-3 GJ
?1
 (40-42), compared to prices of $1-2 GJ
?1
 for coal and $2.5-5 GJ
?1
 for oil and 
natural gas over the last decade (25, 43). Using biomass feed at $2.5 GJ
?1
, ethanol or methanol 
can be produced today at an estimated cost of $0.25-0.3 L
?1
, which would compete with gasoline 
derived from oil at $30-40 per barrel (44, 45). Further technical improvements could make 
biomass-derived alcohol competitive at oil prices below $25 per barrel.  
 10 
 The energy potential of biomass is large. Plants store energy at a rate of about 3000 EJ y
?1
. 
Two-thirds of this productivity is on land, half of which is concentrated in the tropics. Humans 
already actively manage more than half of the useable land area for the production of food and 
fiber; cropland, pasture, and managed forests store about 600 EJ y
?1
. Some of this productivity is 
manifested as wastes that could be diverted for energy production, and some exists in the form of 
fallow or degraded cropland and pasture that could be converted to the production of energy 
crops. 
 The energy value of all biomass wastes?crop residues, dung, wood waste, solid waste, and 
sewage?is about 130 EJ y
?1
 (46). About one-quarter of this could be recovered for energy. The 
remainder is either uneconomical to collect, transport, or convert to energy, or is necessary to 
maintain soil quality, prevent erosion, and provide habitat for natural species. Production of 
recoverable residues should increase to roughly 50-80 EJ y
?1
 in 2050 (47). 
 In addition to wastes, energy crops could be grown on plantations. The amount of energy that 
could be supplied would depend on the amount of land and the average yield of the crops. Crops 
under consideration for temperate climates include woody plants, such as poplar and willow, as 
well as herbaceous plants, such as sorghum and switchgrass. Today, average net yields in 
experimental plots are 150-250 GJ ha
?1 
y
?1
. In tropical and subtropical regions, the leading 
candidates are Eucalyptus, with an average yield of 150-350 GJ ha
?1
y
?1
, and sugarcane, with an 
average yield of about 600 GJ ha
?1
y
?1
 (46). Here I assume that average net yields of 200 GJ ha
?1
 
y
?1
 can be achieved by 2050 over hundreds of millions of hectares of surplus and marginal land. 
 More difficult to estimate is the amount of land that realistically could be devoted to energy 
crops. In 1997, about 1500 Mha were classified as ?arable? (i.e., cultivated in the last five years), 
of which about 1000 Mha were harvested (48). Estimates of potentially arable land?land on 
which rain-fed crops could achieve reasonable yields, in addition to those currently under 
cultivation?range from 500 to 2500 Mha. Most of this land is in sub-Saharan Africa and Latin 
America. The wide range of values reflects incomplete knowledge of soil and climate conditions, 
differing evaluations of the potential of poor soils or steep terrain to support crop production, and 
differing views about the desirability and feasibility of converting natural forests and swamps 
into cropland. If conversion of natural lands is ruled out, 500-1000 Mha of potentially arable land 
would be available.  
 The availability of land for energy crops will depend on the balance between future growth in 
crop yields and grain consumption. If crop yields increase at a rate greater than consumption, the 
area harvested will shrink and large areas will be available for biomass plantations. If, on the 
other hand, increases in yields do not keep pace with increases in consumption, cropland will 
increase and little land may be available for energy crops.  
 Past trends are encouraging: between 1961 and 1996, world production of cereals increased 
by 140%, while the area harvested increased only 9%. This was made possible by large increases 
in average cereal yield, from 1.4 t ha
?1
 in 1961 to 3.0 t ha
?1
 in 1997 (37). It is unclear whether 
growth in yields will continue to keep pace with growth in consumption. Cereal consumption is 
expected to increase at an average rate of 1-2% y
?1
 over the next half century, driven by increases 
in population and per-capita grain utilization.
13
 
                                                 
13
 Population is expected to increase 30-100% by 2050 and per-capita consumption of cereals 
will increase 20-40% as diets improve and meat consumption rises. At per-capita incomes below 
$10,000 y
?1
, per-capita grain utilization has increased by about 90 kg y
?1
 for each doubling of 
 11 
  How much grain yields will increase in the future is the subject of much debate (49). 
Optimists point to the high yields that have been achieved in developed countries as evidence 
that the world average can increase substantially. Cereal yields in France and the United 
Kingdom are more than twice the world average, and China has attained yields 60 percent higher 
than the world average (37). Biotechnology holds the promise of further increases. Pessimists 
note that most of the increase in yields was achieved before 1984; since then, yields have 
increased at an average rate of only 1.3% y
?1
. Much of the past growth in yields was due to 
increased use of fertilizer, pesticides, and irrigation, but further increases in these inputs are 
problematic because of diminishing returns, environmental impacts, and water shortages. 
Pessimists also point to the steady loss of productive cropland, at a rate of about 10 Mha y
?1
, due 
to erosion, salinization, desertification, and urbanization (50, 51). Climate change, and associated 
changes in temperature, soil moisture, the frequency of storms and drought, and the range of 
pests and plant disease, adds further uncertainty to projections of future crop yields. 
 If increases in yield keep pace with increases in consumption, then 500-1000 Mha would be 
available for energy crops and the energy potential would be 100-200 EJ y
?1
. If grain 
consumption increases 1% y
?1
 faster than average yields over the next 50 years (e.g., 
consumption increases 2% y
?1
 while yields increase 1% y
?1
),
14
 the amount of land available for 
energy plantations would decrease by 700 Mha and the energy production potential would be 0-
50 EJ y
?1
. If yields increase 1% y
?1
 faster than consumption, an additional 500 Mha would be 
available and the energy production potential would be 200-300 EJ y
?1
. Including wastes, 
commercial biomass could supply 50-400 EJ y
?1
 by 2050. For comparison, scenarios developed 
by Naki?enovi? et al (20) assume that modern biofuels would supply 50?120 EJ y
?1
 in 2050 and 
160?300 EJ y
?1
 in 2100. 
 A major uncertainty is whether very large quantities of biomass can be grown and harvested 
in a sustainable and environmentally acceptable manner. There is no question that this could be 
done in principle, but whether it can be accomplished in practice depends on a wide variety of 
economic, social, institutional factors. The history of agriculture, which has been characterized 
by widespread land abuse, is not encouraging. 
 
Fission Energy 
 
Of the carbon-free sources that could make a major contribution to future energy supply, fission 
is the only one that is deployed commercially on a significant scale today. In 1998, fission 
reactors supplied over 2300 TWh of electricity?17 percent of world electricity generation and 
over 6 percent of commercial primary energy (18). 
Like wind and solar photovoltaics, fission supplies only electricity; unlike wind and solar, 
however, fission can produce electricity at a steady rate. Although in principle fission can supply 
                                                                                                                                                             
per-capita GDP (48). Per-capita GDP is expected to grow at a rate of 1-2% y
?1
 from 2000 to 
2050; per-capita grain utilization would therefore be expected to increase by 70-140 kg y
?1
, or by 
a factor of 1.2 to 1.4. 
14
 The use of average growth rates here is a mathematical convenience, and does not imply that 
that consumption or yields grow exponentially with time. In fact, both probably will follow a 
more S-shaped growth curve, as population growth declines and per-capita consumption 
saturates, and as natural limits to yield growth come into play. 
 12 
heat for industrial and residential use, accident concerns and other siting considerations have 
limited this application. And although the possibility of producing hydrogen from nuclear-
generated electricity is sometimes mentioned, electrolytic hydrogen would be considerably more 
expensive than hydrogen produced from biomass or decarbonized fossil fuels.
15
  
Fission?s energy production potential depends on the fuel-cycle technology used and the size 
of exploitable uranium resources. It is estimated that 15-125 Mt of uranium could be extracted 
from terrestrial ores at a cost of less than $260 kg
?1
 (53, 54). The type of reactor in widest use, 
the light-water reactor (LWR), requires about 200 t GW
e
y
?1
 if operated on a once-through fuel 
cycle, in which the spent fuel is treated as waste (53). In nuclear-intensive energy scenarios (20, 
55, 56) generating capacity rises from about 350 GW
e
 today to 1100-1700 GW
e
 in 2050, with 
nuclear supplying 70-110 EJ
p
 y
?1
 and 30-40% of world electricity supply. If this energy is 
supplied entirely by LWRs operating on a once-through fuel cycle, cumulative consumption of 
uranium would be 5-7 Mt; the reactors then in existence would require another 4-6 Mt for the 
remainder of their operating life. Thus, conventional uranium resources can easily support a 
high-growth scenario for at least 50 years using a once-through cycle.   
Over the longer term, heavy reliance on nuclear energy would require a transition to fuel 
cycles that use uranium more efficiently or that exploit unconventional uranium resources. The 
traditional solution is to recycle the unburned plutonium and uranium in breeder reactors. In this 
way, it is possible to decrease uranium requirements by a factor of 100, so that 25 Mt could 
provide over 10
6
 EJ
p
. Recycling plutonium raises concerns about the possible diversion of this 
material for weapons, however (see below). Less discussed is the possibility of using 
unconventional uranium resources, particularly the 4500 Mt that is dissolved in the world?s 
oceans. Recent studies indicate that uranium could be extracted from seawater for as little as 
$100 kg
?1
 (57; Seguchi and Foos, personal communications), in which case plutonium recycling 
and breeder reactors would remain economically unattractive indefinitely. 
Although fission?s technical potential is substantial, its near-term prospects are not very 
favorable. Forecasts range from a substantial decrease to a modest increase in installed capacity 
over the next 20 years, with fission?s share of world electricity production falling to less than 10 
percent by 2020 (53, 58). The only region expected to experience significant growth in the near 
future is East Asia. 
The main factor limiting the growth of fission is high capital cost and unpredictable 
construction times. In the United States, the average cost of nuclear-generated electricity in the 
early 1990s was nearly twice that of gas- or coal-fired electricity, due mainly to high construction 
and non-fuel operation and maintenance costs (59-64). The best U.S. nuclear plants, however, 
produce electricity at lower cost than the best coal-fired plants (61). In countries with well-run 
nuclear plants and expensive fossil fuels, such as Japan, nuclear is on average somewhat less 
expensive than fossil-generated electricity. New plants should have lower operating expenses. 
                                                 
15
 According to Ogden & Nitsch (52), hydrogen can be produced from biomass gasification at a 
cost of $6-9 GJ
?1
, from natural gas at a cost of $7-10 GJ
?1
, and from coal at a cost of $8-9 GJ
?1
. 
Carbon sequestration would add $0.2?1 GJ
?1
 to the cost of H
2
 produced from gas and $1?6 GJ
?1
 
for H
2
 produced from coal (assuming sequestration at $10?60 tC
?1
). The cost of electrolytic 
hydrogen, assuming nuclear-generated electricity at a cost of $0.03-0.07 kWh
?1
, would be $15-30 
GJ
?1
.  
 13 
Studies indicate that in many countries new nuclear plants should be economically competitive  
with new coal- and gas-fired plants, producing electricity for $0.03?0.07 kWh
?1
 (63, 65-66). 
Economic considerations aside, the future of fission energy is clouded by concerns about 
accidents, waste disposal, and the spread of nuclear weapons. Below I review the prospects in 
each of these areas. 
 
Accidents. The most serious accident outside of the former Soviet Union occurred at the Three 
Mile Island (TMI) reactor in 1979. Although the reactor core was seriously damaged, the amount 
of radioactivity released into the environment was too small to harm the surrounding population. 
This was the only accident in about 9,000 reactor-years of operation in which the reactor core 
was damaged.  
The accident at TMI triggered numerous improvements in reactor safety. Calculations 
indicate that the probability of core damage is less than 10
?4
 per reactor-year for current U.S. 
LWRs, and that the probability of a large release of radioactivity is about ten times smaller (67). 
Although these probabilities are low, they are not low enough. At this rate, accidents resulting in 
core damage would occur once per decade in a world with 1,000 nuclear reactors. 
New LWRs should be considerably safer. Calculations indicate that General Electric?s 
Advanced Boiling Water Reactor and Combustion Engineering?s System 80+ pressurized water 
reactor would have core-damage probabilities lower than 10
?6
 per reactor-year for internally 
initiated accidents (59). If rates this low could be achieved in practice, a very large expansion in 
nuclear capacity could occur over the next century with little chance of a serious accident. 
It will be difficult, however, to demonstrate that extremely low levels of risk have been 
achieved. Even advanced LWRs depend on the proper operation of equipment, such as pumps 
and valves, to prevent accidents. Safety also depends on proper operation and maintenance, and it 
is difficult to estimate the likelihood of operator errors that could trigger or exacerbate an 
accident. For these reasons, a substantial expansion of nuclear power may require the 
development of ?inherently safe? or ?passively safe? reactors, which place less reliance on the 
proper functioning of equipment and human operators. For example, a cooling system that relies 
on natural circulation is safer?and its safety is easier to demonstrate?than a system that relies 
on pumps. Design concepts have been put forward for passively safe LWRs, gas-cooled graphite-
moderated reactors, and liquid-metal-cooled fast reactors, which would shut down automatically 
and prevent core damage for several days or longer without operator intervention or off-site 
electricity. Although passively safe reactors would be more expensive than conventional LWRs, 
shorter licensing and construction times, higher investor confidence, and reduced public 
opposition would provide offsetting advantages. Thus, it seems plausible that nuclear fission 
could supply a large fraction of future energy consumption in ways that would be safe?and 
would be perceived as safe. 
 
Waste disposal.  Nuclear reactors generate radioactive wastes that must be isolated from the 
biosphere for many millennia. A number of solutions to this problem have been proposed over 
the years, ranging from disposal in deep sea beds to launching the waste into the sun. Most 
countries have adopted deep geological disposal in a mined repository, but no wastes have 
disposed of so far. Although spent fuel and vitrified wastes can be stored safely in interim 
facilities for 50 to 100 years or more, the continued accumulation of wastes in the absence of a 
proven, permanent repository is a barrier to the expansion of nuclear power in many countries.  
 14 
Cost is not a major issue; geological disposal is expected to add only $0.001 kWh
?1
 to the 
price of nuclear-generated electricity in the United States. The availability of land also is not an 
issue?all nuclear wastes that would be generated worldwide this century (and beyond) could be 
stored in an area one-tenth the size of the Nevada Test Site (the site of over 800 underground 
nuclear explosions).
16
 The main difficulty is selecting a site and certifying that, over many 
thousands of years and under almost any conceivable scenario, people would not be exposed to 
unacceptable risks. Even if there is a high level of scientific confidence that this can be achieved, 
it nevertheless may be difficult to overcome public opposition.  
There is, of course, considerable uncertainty about what might happen to nuclear wastes 
thousands of years after they are placed in a repository, and even more uncertainty about how 
humans might become exposed to the wastes. Calculations show that waste packages would 
remain intact for 500 to 1,000,000 years, depending on the design of the package, the thermal 
loading of the repository, the nature of the surrounding rock, and precipitation in the area (59). 
After the packages leak, it would take 1000 to 1,000,000 years for the most soluble radionuclides 
to reach the biosphere; the most hazardous radionuclides (plutonium and other transuranic 
elements), which are much less soluble, would take 100 to 1000 times longer to reach the 
biosphere (68). Natural analogues, such as natural reactors and uranium ore bodies, indicate that, 
at least in some geologies, the most hazardous radionuclides would be contained extremely well 
in the surrounding rock, and would decay to harmless levels long before they could come into 
contact with living things (69). 
The U.S. National Academy of Sciences (70) and regulatory bodies in several countries have 
recommended that the radiation standards that currently are used to protect the public should 
apply to future individuals. These standards are very stringent: in the United States, the dose to 
an individual from all nuclear facilities must be less than 0.25 mSv y
?1
?about one-tenth of the 
average dose rate from natural background radiation and about half the average dose rate from 
medical x-rays. Calculations for proposed repositories in Belgium, Canada, Finland, France, 
Japan, and Sweden indicate that the maximum dose to an individual would at all times be far 
below current limits (71, 72; Jean-Paul Schapira, personal communication). Although similar 
calculations show that maximum doses from the U.S. repository at Yucca Mountain would be a 
factor of 100 or more below suggested limits during the first 10,000 years (the time horizon 
proposed by the EPA for evaluating the performance of the repository), doses well in excess of 
such limits are possible after 50,000 years (73, 74). Whether this will prove to be a barrier to the 
licensing of Yucca Mountain remains to be seen. 
Currently, every country is expected to dispose of its own nuclear wastes?even small 
countries such as Belgium, Netherlands, Switzerland, and Taiwan, whose combined areas are 
less than the area of Indiana. This practice is inefficient, uneconomical, and potentially risky. 
Countries should be encouraged to accept nuclear wastes from other countries, provided that 
their repositories meet an international standard comparable to the most restrictive national 
standards.  
Because it is likely that geologic disposal will continue to be problematic in some countries, 
                                                 
16
 In a high-growth scenario, 3-4 Mt of spent LWR fuel would be discharged during the next 
century. Assuming a heat output (at time of emplacement) of 700 W per ton of spent fuel and a 
repository loading of 7 W m
?2
, the waste would occupy 300-400 km
2
. For comparison, NTS has 
an area of 3500 km
2
, and Manhattan Island has an area of 60 km
2
. 
 15 
research on other methods of disposal should be revived. The most promising alternative is sub-
seabed disposal, in which waste canisters would be placed in the thick layer of fine, sticky mud 
that exists on the ocean floor (75, 76). Vast areas of the seabed have been undisturbed for tens of 
millions of years, and it is estimated that radionuclides would move through the mud at a rate of 
only about one meter per million years. If radioactivity somehow leaked into the water at the 
bottom of the ocean, there are no pathways by which humans could receive a measurable dose. 
Although sub-seabed disposal currently is prohibited by international treaty, this could be 
changed if additional research shows that it is safe and if geologic disposal proves unworkable 
(77). 
It is sometimes claimed that reprocessing?separating and recycling the uranium and 
plutonium in spent reactor fuel?greatly reduces the cost and risk of waste disposal. Although 
reprocessing reduces the mass and the volume of high-level wastes by about a factor of five, the 
capacity of a repository?and therefore the cost of disposal?is limited by the heat output of the 
wastes, not by their mass or volume. Because most of the heat is produced by fission products, 
reprocessing would not reduce the cost of waste disposal by more than a factor of two. Likewise, 
the risks of waste disposal are dominated in most scenarios by long-lived fission products, such 
as technetium-99 and iodine-129, which are far more soluble in water than are plutonium and 
other transuranic elements.  
It has also been suggested that separating radionuclides with long half-lives and transmuting 
them into short-lived or stable nuclides would greatly reduce waste-disposal risks. Transmutation 
would be accomplished in a reactor or accelerator. Although the amount of long-lived waste 
could be reduced, it is unlikely that the small reduction in waste-disposal risk in the very long 
term (which is already very small) would outweigh the high costs and increased accident and 
proliferation risks associated with separation and transmutation in the near term (78). 
 
Proliferation.  All nuclear fuel cycles involve weapon-usable materials that can be separated 
using a relatively straightforward chemical process (79). Although fresh LWR fuel cannot be 
used for weapons purposes, spent LWR fuel is 1% plutonium. This ?reactor-grade? plutonium 
contains a higher percentage of undesirable isotopes than does the ?weapon-grade? plutonium 
used in stockpiled nuclear weapons. These undesirable isotopes emit heat and radiation, 
complicating weapon design and leading some observers to argue that reactor-grade plutonium is 
unsuited for weapons. In fact, any group that could make a nuclear explosive with weapon-grade 
plutonium would be able to make an effective device with reactor-grade plutonium (80, 81). 
Because access to weapons-usable material is the principle barrier to the acquisition of nuclear 
weapons, the plutonium discharged from civilian reactors should receive the same degree of 
protection from theft or misuse as assembled nuclear weapons. 
Under the Non-Proliferation Treaty, all but a handful of states have promised not to acquire 
nuclear weapons and have agreed to accept safeguards on peaceful nuclear activities to verify that 
nuclear materials are not being diverted or misused. As long as the fuel remains intact, it is 
relatively easy to detect diversion of the plutonium-bearing spent fuel, because international 
inspectors can simply tag and count the number of fuel assemblies. Spent fuel also is very 
difficult to steal, both because of its unwieldy size and because it is highly radioactive. A spent 
fuel assembly from a typical LWR is 4 m long, has a mass of 650 kg, and would deliver a lethal 
dose of radiation to an unprotected person in a few minutes (80). A single assembly contains 
enough plutonium for a nuclear weapon, but because of the high radiation field the spent fuel is 
 16 
said to be ?self-protecting.? The United States adopted the once-through fuel cycle in the 1970s 
primarily because it maintains nuclear materials in forms that are relatively invulnerable to 
misuse. At current and foreseeable uranium prices, it is also the least expensive fuel cycle.  
The main alternative to the once-through cycle involves the separation and recycling of the 
plutonium and uranium in the spent fuel. In contrast to spent fuel rods, which are easy to count 
and track, precise measurement of plutonium inventories in a reprocessing plant is difficult. The 
amount of plutonium in the spent fuel is uncertain and inventories are difficult to measure, 
leading to inevitable differences between the estimated amounts entering and exiting the plant. In 
a large plant, this ?inventory difference? can amount to many bombs-worth of plutonium per year 
(82). Although material accounting can be improved, it does not appear that one could detect 
with high confidence and in a timely manner the diversion of a significant amount of plutonium. 
The fabrication, transport, and storage of plutonium fuels provide additional opportunities for 
theft or diversion. 
Separation and recycle would, however, decrease the availability of plutonium to future 
generations, who might otherwise mine stores of spent fuel for plutonium after radioactive decay 
has rendered the fuel much easier to handle. But it is not clear that mining buried spent fuel 
would be simpler or less expensive than producing or diverting fresh plutonium or high-enriched 
uranium, and it is even less clear that the reduced availability of plutonium in the very long term 
would outweigh the increased near-term risks of theft and diversion associated with recycle. In 
any case, this risk could be minimized by centralizing repositories in a few countries and by 
designing long-term safeguards to detect intrusion into repositories (83, 84). 
 If nuclear power grows substantially, recycle may become necessary or economically 
attractive. In this case, additional technical and institutional barriers could be introduced to deter 
and detect theft or diversion. This could include novel reactor concepts, such as lifetime cores; 
new reprocessing techniques that do not involve the separation of pure plutonium; and fuel 
cycles that minimize the production of high-quality plutonium, such as the thorium fuel cycle 
(85, 86). Diversion also could be inhibited by internationalizing certain parts of the nuclear fuel 
cycle. A serious shortcoming of the current regime is that non-nuclear-weapon states are 
permitted to own and operate facilities capable of producing plutonium and HEU, and can 
produce, stockpile, and use these materials so long as they are under safeguards. But safeguards 
may be unable to detect the diversion of significant quantities of these materials in a timely 
manner from facilities that handle the materials in bulk form, such as reprocessing and fuel-
fabrication plants. It would be far easier to deter or detect diversions by states if such activities 
were managed directly by an international agency. Similar arrangements could be extended to the 
storage and use of fresh plutonium fuels, or even spent fuel. National reactors might be permitted 
to burn only low-enriched fuels, with the spent fuel turned over to international reprocessing or 
storage centers; reactors burning plutonium fuels would be managed by an international 
authority. 
 
Solar Energy 
 
Sunlight is the ultimate source of many of the forms of energy discussed above: biomass and 
fossil fuels, hydro, wind, wave, and ocean thermal energy. Here ?solar? refers only to the direct 
use of sunlight to produce heat or electricity. In 1998 solar produced about 3 TWh of electricity 
and perhaps 0.5 EJ of heat in solar thermal collectors. 
 17 
 The solar resource is huge. About 500,000 EJ falls on the continents each year. The resource 
is spread fairly uniformly, at least on an annual basis. Sunny areas, such as the southwestern 
United States or southern Spain, receive up to 9 GJ m
?2
 y
?1
, while cloudy, northern areas, such as 
the northwestern United States or the United Kingdom, receive as little as about 4 GJ m
?2
 y
?1
 (87, 
88). 
 As with other diffuse sources, the challenge is to capture and deliver solar energy 
economically. In temperate climates, properly designed and oriented buildings can be partially 
heated and lighted with solar energy at costs that are competitive with current U.S. energy prices 
(89). Today, however, less than 1% of new homes built in the United States incorporate 
significant passive solar features. The turnover of the building stock is very slow; even if passive 
solar design became far more popular, it would not contribute more than 1% of total U.S. energy 
demand in 2050.  
 Roof-mounted collectors can be used to heat air or water for residential or commercial use in 
existing buildings. To produce solar heat at $5 GJ
?1
 (the current retail price of natural gas in the 
United States), installed costs must be less than $200 m
?2
 in sunny areas and less than $100 m
?2
 
in less-sunny areas such as New York or London. Although the collectors themselves are 
produced in the United States for about $150 m
?2
 (90), installed costs are several times higher 
(88). The economics of solar heat are even less favorable for industrial users, who require higher-
temperature heat and who pay lower prices for conventional fuels. The potential for lowering the 
cost of solar heat is limited; the technology is mature and uses common materials. If energy 
prices double or quadruple, however, solar could provide a substantial fraction of the energy used 
for heat?perhaps 10 percent of total energy demand.  
 The technical feasibility of generating electricity with solar heat has been demonstrated in 
multi-megawatt facilities, both with distributed parabolic-trough collectors and with central 
?power-tower? receivers illuminated by hundreds of sun-tracking  mirrors. The cost of electricity 
from advanced devices located in very sunny areas is estimated at about $0.08-0.16 kWh
?1
 (91). 
With additional improvements in efficiency and cost, solar thermal electric plants might compete 
favorably with new nuclear plants in sunny locations. 
 The solar technology with the greatest potential is photovoltaics. Photovoltaic cells convert 
sunlight directly into electricity. They require no focusing or tracking mechanisms (although 
these may be used), boilers, turbines, or cooling water; they generate no waste products, heat, or 
noise. Photovoltaics are highly reliable, have long lifetimes, and require very little maintenance. 
Photovoltaic cells can be wired together to form units of any size, from a fraction of a watt to 
hundreds of megawatts. They can be integrated into the design of exterior building surfaces.  
 The cost of photovoltaic modules has decreased tremendously, from $100 per peak watt in 
1975 to as low as $4 W
p
?1
 today for large purchases. The installed cost per peak watt of net AC 
output to the grid, including support structures, inverters, and so forth, is roughly double the cost 
of the photovoltaic modules (92, 93). At this price, photovoltaic electricity remains far too 
expensive for widespread use. At an installed price of $1 W
p
?1
, photovoltaic systems would 
produce electricity at a cost of $0.04-0.1 kWh
?1
, depending on location, in which case they would 
compete favorably with other sources of electricity. It may not be easy to reduce the price of 
photovoltaic systems to $1 W
p
?1
, which corresponds to an installed price of $50-100 m
?2
 for 
photovoltaic modules. The cost of the raw materials alone is unlikely to be less than $30 m
?2
 
(93), and the price of installing common building materials, such as shingles or siding, is about 
$30 m
?2
 (94). 
 18 
 Even if prices fall to levels that would be economically competitive with other sources, solar 
would be limited to 10-20% of total electricity production unless large-scale, inexpensive storage 
or intercontinental transmission of electrical energy could be achieved. For the storage 
technologies available today?pumped hydro, compressed-air storage, and batteries?storage 
would increase the cost of electricity 40-200%. As mentioned above, the production of hydrogen 
is often mentioned as a means of storing and distributing solar energy, but solar electricity would 
have to be very inexpensive?less than $0.02 kWh
?1
?for electrolytic hydrogen to be cheaper 
than hydrogen produced from the gasification of biomass or fossil fuels. In the longer term, 
storage rings or transmission lines using high-temperature superconductors may provide an 
efficient and affordable means to store solar electricity or transmit it from sunlit to nighttime or 
overcast areas. 
 Some have suggested that large arrays of solar cells could be placed in geosynchronous orbit 
around the earth, with the power transmitted in microwaves to fixed receiving antennae on earth. 
Because the array would receive sunlight at a constant rate, without interference from the 
atmosphere or clouds, a photovoltaic module in orbit would on average produce electricity at 
about five times the rate that it would at the sunniest locations on the earth?s surface. This 
constant and predictable supply would, moreover, eliminate the need for energy storage. 
Although conceptually appealing, these advantages are unlikely to compensate for the enormous 
costs of placing and maintaining equipment in orbit. At current prices, launch costs alone would 
amount to $100 to $500 W
p
?1
.
17
 Putting aside questions about the overall technical feasibility of 
such a project, launch costs would have to drop by a factor of twenty or more for this concept to 
be economically competitive with ground-based generation. 
 
Wind Energy 
 
Wind power has been harnessed by humans for millennia, but only in the last decade has wind 
generated significant amounts of electricity. In 1998, wind produced 13 TWh, mostly in the 
United States, Germany, and Denmark (18). Installed capacity increased from 1 GW
e
 in 1985 to 
12 GW
e
 in 1999?an average growth rate of nearly 20% y
?1
. 
 Today, electricity is produced at a cost of $0.05-0.08 kWh
?1
 at sites with average wind power 
densities greater than 250 W m
?2
 at a height of 10 m (97, 98). About 5% of the earth?s land area 
has wind power densities this high; in theory, about 160,000 TWh y
?1
 (1400 EJ
p
 y
?1
) could be 
generated with wind machines distributed over this area. Advances in technology might make it 
possible to generate electricity economically at off-shore sites or at sites with lower wind power 
densities. The use of sites with power densities between 150-250 W m
?2
 would expand the 
production potential by a factor of three. 
 The amount of wind electricity that could be generated in practice is considerably lower. 
Much of the wind resource is located very far from population centers (e.g., in northern Canada 
and Russia), where the costs of transmission and maintenance would be excessive. 
                                                 
17
 Photovoltaic cells weighing only 5 g W
p
?1
 (0.12 mm thick) have been produced for use on a 
solar-power aircraft (95). Launch costs currently range from $20 to $100 per gram for 
geosynchronous orbit (96). Thus, launch costs would amount to $100 to $500 W
p
?1
, or about $75 
to $370 W
?1
 in orbit. Assuming that launch costs are amortized over 30 y at a rate of 10% y
?1
, 
launch costs would add $0.8-$4 kWh
?1
 to the price of electricity.  
 19 
Environmental constraints, such as the presence of existing forests and protected areas, would 
further limit the siting of wind turbines, as would public-acceptance considerations. All things 
considered, only about one-tenth of high-wind areas?mostly cropland and pasture?may be 
suitable for electricity production. Moreover, because of the intermittent and unpredictable nature 
of wind power, wind?s contribution to regional electricity supply would be limited to perhaps 
20%, unless large-scale energy storage or transmission is provided. Thus, the practical potential 
of wind electricity is about 12,000 TWh y
?1
. A realistic upper limit for wind production in 2050 
might be 4000 TWh y
?1
 (40 EJ
p
 y
?1
)?roughly 10% projected world electricity production.  
 Although wind is unlikely to become a dominant energy source, it has the potential to 
contribute a substantial fraction to total energy demand. In the short term, at least, it the most 
promising renewable energy source. 
 
Decarbonized Fossil Fuels 
 
Recoverable, low-cost resources of conventional oil, gas, and coal are sufficient to meet world 
energy needs for at least another one hundred years. Moreover, enormous quantities of 
unconventional fossil fuels?methane hydrates, oil shales, and tar sands?could be extracted at 
somewhat higher prices or with improved technology. If one could safely and inexpensively 
?decarbonize? or remove and sequester the carbon contained in fossil fuels, they could continue 
to serve as a basis for world energy supply even while greenhouse gas concentrations are 
stabilized.  
This option has the advantage of relying on well-established industries and technologies, offering 
the potential of a smooth transition to carbon-free energy production.  
 
Capture.  There are two main approaches for removing the carbon from fuels. The first is to 
capture the CO
2
 gas after combustion. This is practical only for large, centralized sources of CO
2
, 
primarily coal-fired power plants. The technology for capturing CO
2
 from flue gases using 
chemical solvents is mature but expensive. It is estimated that carbon-dioxide capture would 
increase the price of electricity from a traditional coal-fired power plant by 40-120% ($0.02-0.06 
kWh
?1
), equivalent to $100-260 per ton of carbon emission avoided (99-101). The costs would 
be greater for a gas-fired power plant, due to the lower carbon content of the fuel. 
 The second approach is to chemically convert fossil fuels into hydrogen and CO
2
. Hydrogen 
is produced from natural gas and gasified coal on a commercial scale today for the manufacture 
of ammonia and other chemicals; the cost per unit energy of the hydrogen product is about 70% 
greater than that of natural gas and five times greater than that of coal (102). Even at these high 
prices, hydrogen could be an attractive fuel in the long term because it can be converted 
efficiently in fuel cells into electricity with virtually no pollution. Coal also can be converted into 
hydrogen-rich fuels, such as methane or methanol, that are easier to transport and store than is 
hydrogen. The cost of such chemical conversions is very high, however?equivalent to $150-500 
per ton of carbon emissions avoided. 
 Perhaps the most attractive decarbonization concept is based on the integrated coal-
gasification combined-cycle (IGCC) power plant, in which the combustion of fuel gas derived 
from coal is used to drive a gas turbine, with the waste heat used to drive a steam turbine. In this 
case, the CO
2
 would be separated from the fuel gas before combustion, generating a stream of 
almost pure hydrogen. Although the cost of electricity from an IGCC plant is estimated to be 
 20 
somewhat greater than that of a traditional coal-fired power plant, the incremental cost of 
capturing the CO
2
 is smaller because of the high concentration of CO
2
 in the fuel gas. Even so, 
carbon-dioxide recovery is estimated to add $0.013-$0.026 kWh
?1
 or 25-50% to the price of 
electricity, or $65-160 per ton of carbon emissions avoided (99).  
 None of these techniques would eliminate carbon emissions completely. About 10 percent of 
the carbon contained in the fuel would be emitted into the atmosphere as CO
2
. This reduction 
would be sufficient, however, to allow stabilization at or below an equivalent doubling even if 
fossil fuels continued to be the dominant energy source throughout the next century. 
 
Disposal.  In order for decarbonization to contribute significantly to world energy supply over the 
next century, several hundred billion tons of carbon would have to be sequestered in ways that 
would prevent its release into the atmosphere for at least several hundred years. Such huge 
quantities of CO
2
 could be sequestered at reasonable cost only in natural geological formations or 
in the oceans. Other options, such the manufacture of solids or industrial chemicals or storage in 
engineered facilities or mined cavities, are too limited or too expensive to make a major 
contribution (103).  
 Oil and gas wells are probably the least expensive and the most reliable option for the storage 
of CO
2
. Exploration and drilling costs would be low, and the prior existence oil and gas deposits 
would ensure that CO
2
 could be stored for millions of years if the original pressure of the 
reservoir is not exceeded. Total world capacity is estimated at 150-500 PgC, based on estimates 
of oil and gas resources. A small fraction of this storage potential (10-15%) could be used to 
enhance the recovery of oil and gas remaining in active wells, thereby lowering the costs of 
sequestration. Carbon dioxide was injected into oil wells in the United States on a small scale in 
the late 1970s to enhance oil recovery, when oil prices were much higher than at present. Natural 
gas often contains CO
2
, which today is separated and vented to the atmosphere; injecting this 
CO
2
 is an obvious application of sequestration. In 1996, Statoil of Norway began injecting CO
2
 
from a gas field into an aquifer beneath the North Sea. 
 Storage in oil and gas wells alone would be not sufficient, however. A large fraction of fossil-
fuel use occurs in areas such as Japan, western Europe, and the northeastern United States, where 
the cost to transport CO
2
 to oil and gas wells would be very high. Disposal costs could be 
minimized by producing electricity or hydrogen close to oil and gas wells, but the savings would 
be more than offset by the high costs of transporting electricity and hydrogen over very long 
distances. Baring technical breakthroughs, such as inexpensive, long distance superconducting 
electrical transmission systems, storage sites would be located closer to areas of energy 
consumption.  
 One option is store CO
2
 underground in deep saline aquifers. In the United States, for 
example, 65% of power-plant carbon emissions occurs close to deep aquifers (104). Storage sites 
would be located at depths greater than 800 m, in order to maintain the CO
2
 in a dense 
supercritical phase, and under an impermeable layer to prevent the escape of CO
2
 or mixing with 
shallow aquifers used for drinking water or irrigation. The injected CO
2
 would displace and 
partially dissolve in existing water, and would react chemically with certain types of rock, 
particularly those rich in calcium and magnesium, to form solid compounds.  
 The potential storage capacity of underground aquifers is highly uncertain; estimates range 
from less than 100 PgC to more than nearly 3000 PgC (105). The wide range is partly due to a 
lack of basic geological data, such as the volume, porosity, and permeability of aquifers, and 
 21 
partly due to assumptions about how much CO
2
 could be stored by unit volume and about what 
types of aquifer structures would be provide long-term storage. The transport and storage of CO
2
 
on land raises concerns about public safety and environment impact from pipeline or well 
failures, but these should not be more difficult to address than those associated with the handling 
of natural gas. 
 Another option is to inject CO
2
 into the deep ocean. Since most of the CO
2
 emitted into the 
atmosphere would dissolve in the ocean eventually, one could think of this as simply accelerating 
a natural processes that would result from the burning of fossil fuels. The carbon sequestration 
potential of the oceans is huge?at least 1000 PgC. In contrast to underground aquifers, which 
can sequester carbon for millions of years, a significant fraction of the CO
2
 injected into the deep 
ocean would return to the atmosphere over period of several hundred years.  
 The rate of return of CO
2
 to the atmosphere would be determined primarily by depth of 
injection. At depths greater than 3700 m, the density of the CO
2
 is greater than that of seawater 
and the CO
2
 would sink to the bottom of the ocean, creating a CO
2
 ?lake? on the ocean floor. In 
this case, about 15% of the injected CO
2
 would return to the atmosphere over a period of roughly 
1000 years. Pipelines have not been laid at depths greater than 1000 m, but there may be other 
ways of achieving much greater depths. For example, long vertical pipes might be suspended 
from a tanker or offshore platform, or a dense plume might be created that would fall naturally to 
the ocean floor or become entrained in downwelling ocean currents. The fraction and rate of 
return can be significantly greater for CO
2
 dispersed at depths of less than 2000 m, depending on 
ocean currents and topography near the point of injection, leading to higher atmospheric 
concentrations after 100 to 200 years. Careful site-specific studies would have to be completed to 
assure that the environmental benefits of reduced CO
2
 concentrations would outweigh the costs 
and risks of ocean disposal. 
 The environmental impact and legal status of ocean disposal are uncertain. Sequestration will 
increase the acidity of seawater; depending on the dispersal mechanism, the decrease in pH could 
be biologically significant over large volumes of water. For example, the injection of 10 TgC y
?1
 
(corresponding to the carbon from half a dozen large coal-fired power plants) in a dense plume 
would reduce the pH below 7 (the level at which mortality is observed in some marine 
organisms) over about 500 km
3
; the corresponding volume for disposal via a towed pipe or a 
deep seabed lake is only 1?5 km
3
 (106). Environmental effects should be minimal as long as CO
2
 
is injected at depths greater than 1000 m, since nearly all marine life is found above this level. In 
any case, dumping of wastes in the oceans is regulated by international law, and issuance of the 
required permits would take into account possible effects on deep-sea marine life and the 
availability of land-based disposal alternatives. 
 The cost of disposal itself?that is, the cost of injecting CO
2
 deep underground or into the 
ocean?is small compared to the costs of capture; estimates range from $1-30 tC
?1
 (101, 105). 
More significant may be the cost of transportation to the disposal site. The most straightforward 
option is to transport the CO
2
 via pipeline at high pressure as a liquid or supercritical fluid. For a 
large pipeline carrying 5-30 TgC y
?1
 (equivalent to the CO
2
 emitted by 3 to 20 large coal-fired 
power plants), transport costs would be $0.01-0.04 tC
?1
 km
?1
 for either underground or ocean 
disposal (107-109). Transport and disposal by tanker is possible for ocean disposal, and may be 
cheaper at longer distances (110). Depending on transport distance, total disposal costs would 
range from about $10-60 tC
?1
. 
 22 
 Thus, the capture, transport, and disposal of hundreds of billions of tons of carbon is unlikely 
to be accomplished at an average cost of much less than $100 tC
?1
, which would represent a 
substantial increase in the price of coal or coal-fired electricity. Even so, decarbonized coal could 
be economically competitive with other carbon-free energy sources. 
 
CONCLUSIONS 
 
 Only five energy sources are capable of providing a substantial fraction of the carbon-free 
energy required to stabilize greenhouse gas concentrations at an equivalent doubling: biomass, 
fission, solar, wind, and ?decarbonized? fossil fuels. Other potential sources are either too limited 
(hydro, tidal power, and hot-water geothermal), too expensive (ocean thermal and wave energy), 
or too immature (fusion and hot-rock geothermal) to make a substantial contribution by 2050.  
 Unfortunately, each of the five major alternatives currently has significant technical, 
economic, and/or environmental handicaps. Biomass has the potential to supply low-cost 
portable fuels, but large-scale use of energy crops could compete with food production and the 
preservation of natural ecosystems. Fission, which is the only one deployed on a large scale 
today, suffers from public-acceptance problems related to the risks of accidents, waste disposal, 
and the spread of nuclear weapons. Solar is benign but very expensive, and would require 
massive energy storage or intercontinental transmission to supply a large fraction of world 
energy. Wind is economically competitive in certain areas today, but most of the resource is far 
from cities and would, like solar, require expensive storage or long-distance transmission to 
achieve a large fraction of the energy market. Fossil fuels are cheap and abundant, but the cost of 
capturing, transporting, and disposing of the CO
2
 could be high and the environmental impacts 
are largely unknown. A broad-based program of energy research and development is needed to 
address these concerns, and ensure that abundant, affordable, and acceptable substitutes for 
traditional fossil fuels will be available worldwide in the coming decades (111).
 23 
LITERATURE CITED 
1. United Nations Framework Convention on Climate Change, May 1992, 
http://www.unfccc.de 
2. Kattenberg A, Giorgi F, Grassl H, Meehl GA, Mitchell JFB, et al. 1996. Climate models?
projections of future climate. In Climate Change 1995; The Science of Climate Change, ed. 
JT Houghton, LG Meira Filho, BA Callander, N Harris, A Kattenberg, K Maskell, 285-357. 
Cambridge: Cambridge Univ. Press. 572 pp. 
3. Folland, CK, Karl TK, Vinnikov, KYA. 1990. Observed climate variations and change. In 
Climate Change: The IPCC Scientific Assessment, ed. JT Houghton, GJ Jenkins, JJ 
Ephraums, 195-238. Cambridge: Cambridge Univ. Press. 365 pp. 
4. Nicholls N, Gruza GV, Jouzel J, Karl TR, Ogallo LA, et al. 1996. Observed climate 
variability and change. See Ref. 2, pp. 133-192 
5. Ad-hoc Group on the Berlin Mandate, Implementation of the Berlin Mandate: Comments 
from Parties, addendum, FCCC/AGBM/1997/MISC.2/Add.1, 27 April 1997, 
http://www.unfccc.de/resource/docs/1997/agbm/misc02a1.htm 
6. Cole V, Cerri C, Minami K, Mosier A, Rosenberg N, et al. 1996. Agricultural options for 
mitigation of greenhouse gas emissions. In Climate Change 1995; Impacts, Adaptations 
and Mitigation of Climate Change: Scientific-Technical Analyses, ed. RT Watson, MC 
Zinyowera, RH Moss, 745-772. Cambridge: Cambridge Univ. Press. 879 pp. 
7. Levine MD, Akbari H, Busch J, Dutt G, Hogan K, et al. 1996. Mitigation options for human 
settlements. See ref. 6, pp. 713-743 
8. Lashof DA, DeAngelo BJ, Saleska SR, Harte J. 1997. Terrestrial ecosystem feedbacks to 
global climate change. Annu. Rev. Energy Environ. 22:75-118 
9. Melillo JM, Prentice IC, Farquhar GD, Schulze ED, Sala OE. 1996. Terrestrial biotic 
responses to environmental change and feedbacks to climate. See Ref 2, pp. 445-481 
10. Schimel D, Alves D, Enting I, Heimann M, Joos F, et al. 1996. Radiative forcing of climate 
change. See Ref. 2, pp. 65-131 
11. Enting IG, Wigely TML, Heimann M. 1994. Future emissions and concentrations of carbon 
dioxide: key ocean/atmospheric/land analyses. CSIRO Division of Atmospheric Research 
Technical Paper 31, Victoria, Australia 
12. Wigley TML, Richels R, Edmonds JA. 1995. Economic and environmental choices in the 
stabilization of CO
2
 concentrations: choosing the ?right? emissions pathway. Nature 
379:240-243 
13. Houghton RA. 1999. The annual net flux of carbon to the atmosphere from changes in land 
use 1850-1990. Tellus. 51B:298-313 
14. Alacamo J, Bowman A, Edmonds J, Grubler A, Morita T, Sugandhy A. 1994. An 
evaluation of the IPCC IS92 emission scenarios. In Climate Change 1994; Radiative 
Forcing of Climate Change and An Evaluation of the IPCC IS92 Emission Scenarios, ed. 
JT Houghton, LG Meira Filho, J Bruce, H Lee, BA Callander, et al., 237-304. Cambridge: 
Cambridge Univ. Press. 339 pp. 
15. Brown S, Sathaye J, Cannell M, Kauppi P. 1996. Management of forests for mitigation of 
greenhouse gas emissions. See Ref 6, pp. 773-797 
16. Kirschbaum MUF, Fischlin A, Cannell MGR, Cruz RVO, Galinski W, et al. 1996. Climate 
change impacts on forests. See Ref. 6, pp. 95-129 
17. Marland G, Boden TA, Andres RJ, Brenkert AL, Johnston CA. 1999. Global, Regional, and 
 24 
National Fossil Fuel CO
2
 Emissions, Carbon Dioxide Information Analysis Center, Oak 
Ridge National Lab., Oak Ridge, Tenn.,  http://cdiac.esd.ornl.gov/trends/emis/em_cont.htm 
18. Energy Information Administration. 2000. International Energy Annual 1998. U.S. 
Department of Energy, Washington, DC 
19. Leggett J, Pepper WL, Swart RJ. 1992. Emission scenarios for IPCC: an update. In Climate 
Change 1992: The Supplementary Report to the IPCC Scientific Assessment, ed. JT 
Houghton, BA Callander, SK Varney, 69-95. Cambridge: Cambridge Univ. Press. 200 pp. 
20. Naki?enovi? N, Gr?bler A, McDonald A, eds. 1998. Global Energy Perspectives. 
Cambridge: Cambridge Univ. Press. 299 pp 
21. Shell International Ltd. 1996. The Evolution of the World?s Energy Systems. London: Shell 
International. 
22. Schmalensee R, Stoker TM, Judson RA. 1998. World carbon dioxide emissions: 1950-
2050. Rev. Econ. Stat. 80:15-27 
23. Edmonds J, Dooley J, Kim S. 1998. Long-term Energy Technology: Needs and 
Opportunities for Stabilizing Atmospheric CO
2
 Concentrations, American Council for 
Capital Formation, Washington, DC. 
24. Baron R. 1996. Economic/fiscal Instruments: Taxation. ?Policies and Measures for 
Common Action? working paper, OECD, Paris 
25. Energy Information Agency. 1999. U.S. Energy Prices. 
http://www.eia.gov/emeu/steo/pub/4tab.htm 
26. Repetto R, Austin D. 1997. The Costs of Climate Protection: A Guide for the Perplexed. 
Washington, DC: World Resources Institute, Washington, DC. 
27. Kull S. 1998. Americans on Global Warming: A Study of U.S. Public Attitudes. Program in 
International Policy Attitudes, Univ. Maryland, College Park, MD. 
28. Hall DO. 1991. Biomass energy. Energy Policy. 19:711-737. 
29. March PA, Fisher RK. 1999. It?s not easy being green: environmental technologies enhance 
conventional hydropower?s role in sustainable development. Annu. Rev. Energy Environ. 
24:173-88 
30. Moreira JR, Poole AD. 1993. Hydropower and its constraints. In Renewable Energy: 
Sources for Fuels and Electricity, ed. TB Johansson, H Kelly, AKN Reddy, RH Williams, 
73-119. Washington, DC: Island Press. 1160 pp.  
31. Palmerini CG. 1993. Geothermal energy. See Ref. 30, pp. 549-591 
32. Muffler PLJ, 1979. Assessment of Geothermal Resources of the United States. Circular 790, 
US Geological Survey, Washington, DC. 
33. Brown G. 1996. Geothermal energy. In Renewable Energy: Power for a Sustainable Future, 
ed. G Boyle, 353-392. Oxford: Oxford Univ. Press, 479 pp. 
34. Cavanagh JE, Clarke JH, Price R. 1993. Ocean energy systems. See Ref. 30, pp. 513-547. 
35. Druckers D. 1996. Wave energy. See Ref. 33, pp. 315-352. 
36. Tillman DA. 1991. The Combustion of Solid Fuels and Wastes. New York: Academic 
Press. 
37. Food and Agriculture Organization. 1995. FAOSTAT-PC. Rome: FAO 
38. Energy Information Agency. 1998. Renewable Energy Annual 1998. U.S. Department of 
Energy, Washington, DC 
39. Rosillo-Calle F, Cortez LAB. 1998. Towards proalcohol II: a review of the Brazilian 
bioethanol programme. Biomass Bioenergy 14(2):115-24 
 25 
40. Carpentieri AE, Larson ED, Woods J. 1993. Future biomass-based electricity supply in 
north-east Brazil. Biomass and Bioenergy 14:149-173 
41. Turhollow A. 1994. The economics of energy crop production. Biomass and Bioenergy 
6(3):229-241 
42. Graham RL, Lichtenberg E, Roningen VO, Shapouri H, Walsh ME. 1995. The economics 
of biomass production in the United States. Proc. Biomass Conf. of the Americas: Energy, 
Environment, Agriculture, and Industry, 2
nd
, Portland, OR. 
http://www.esd.ornl.gov/bfdp/papers/bioam95/graham3.html 
43. BP Amoco. 1999. BP Amoco Statistical Review of World Energy 1999. 
http://www.bpamoco.com/worldenergy 
44. Wyman CE. 1999. Biomass ethanol: technical progress, opportunities, and commercial 
challenges. Annu. Rev. Energy Environ. 24:189-226 
45. Wyman CE, Bain RL, Hinman ND, Stevens DJ. 1993. Ethanol and methanol from 
cellulosic biomass. See Ref. 30, pp. 865-923 
46. Hall DO, Rosillo-Calle F, Williams RH, Woods J. 1993. Biomass for energy: supply 
prospects. See Ref. 30, pp. 593-651 
47. Johansson TB, Kelly H, Reddy AKN, Williams RH. 1993. A renewables-intensive global 
energy scenario. See Ref. 30, pp. 1071?1142 
48. U.S. Department of Agriculture, World Agriculture: Trends and Indicators. 
http://jan.mannlib.cornell.edu/data-sets/international/89024 
49. World Resources Institute. 1996. Food and agriculture. In World Resources 1996?97, 225-
246. Oxford: Oxford Univ. Press. 365 pp. 
50. Evans M. 1994. Agriculture and environment: a review, 1972?1992. Ambio 23(3):194 
51. Kendall HW, Pimentel D. 1994. Constraints on the expansion of the global food supply. 
Ambio 23(3):200. 
52. Ogden JM, Nitsch J. 1993. Solar hydrogen. See Ref. 30, pp. 925-1009. 
53. Nuclear Energy Agency, International Atomic Energy Agency. 1998. Uranium 1997. p. 60. 
54. Holdren JP, Pachauri RK. 1992. Energy. In An Agenda of Science for Environment and 
Development into the 21
st
 Century. Cambridge: Cambridge Univ. Press 
55. International Atomic Energy Agency. 1995. Nuclear Power: An Overview in the Context of 
Alleviating Greenhouse Gas Emissions. Rep. IAEA-TECDOC-793, IAEA, Vienna. 
56. Nuclear Energy Agency. 1998. Nuclear Power and Climate Change. Rep. NEA#0846, Paris 
57. Hiraoka T. 1994. Nuclear Electricity Generation by Seawater Uranium. J. Atomic Energy 
Soc. Japan 36(7):644-645 
58. Energy Information Administration. 2000. International Energy Outlook 1999. U.S. 
Department of Energy, Washington, DC. http://www.eia.doe.gov/oiaf/ieo99/home.html 
59. Bodansky D. 1996. Nuclear Energy: Principles, Practices, and Prospects. Woodbury, NY: 
American Institute of Physics. 396 pp. 
60. National Research Council. 1992. Nuclear Power: Technical Options and Institutional 
Options for the Future. Washington, DC: National Academy Press. 215 pp. 
61. Energy Information Administration. 1990. Electric Plant Cost and Power Production 
Expenses, 1988. Rep. DOE/EIA-0455(88), U.S. Depart. of Energy, Washington, DC. 
62. Energy Information Administration. 1995. Analysis of Nuclear Plant Operating Costs: A 
1995 Update. U.S. Depart. of Energy, Washington, DC. 
63. Delene JG, Williams KA, Shapiro, BH. 1988. Nuclear Energy Cost Data Base. Rep. 
 26 
DOE/NE-0095, U.S. Depart. of Energy, Washington, DC. 
64. Mark Gielecki and James G. Hewlett, ?Commercial Nuclear Electric Power in the United 
States: Problems and Prospects,? Monthly Energy Review (Washington, DC: Energy 
Information Administration, August 1994). 
65. Organization for Economic Cooperation and Development. 1989. Projected Costs of 
Generating Electricity from Power Stations For Commissioning in the Period 1995-2000. 
OECD, Paris 
66. Organization for Economic Cooperation and Development. 1993. Projected Costs of 
Generating Electricity: 1992 Update. OECD, Paris 
67. U.S. Nuclear Regulatory Commission. 1990. Severe Accident Risks: An Assessment for Five 
Nuclear Power Plants. Rep. NUREG-1150, Nuclear Regulatory Commission, Washington, 
DC.  
68. National Research Council. 1983. A Study of the Isolation system for Geologic Disposal of 
Radioactive Wastes. Washington, DC: National Academy Press 
69. Knight JL. 1998. Use of natural analogues in waste disposal. Interdisciplinary Science 
Reviews 23(3):233-241 
70. National Research Council. 1995. Technical Bases for Yucca Mountain Standards. 
Washington, DC: National Academy Press 
71. McCombie C. 1997. Nuclear waste management worldwide. Physics Today 50(6):60-62  
72. Sumerling T, Smith P. 1998. Disposal of nuclear fuel waste. Interdisciplinary Science 
Reviews 23(3):228?230 
73. US Dep. Energy, Off. Civil. Radioactive Waste Manage. 1998. Viability Assessment of a 
Repository at Yucca Mountain, Dec. DOE/RW-0508. http://www.ymp.gov/va.htm 
74. Craig PP. 1999. High-level nuclear waste: the status of Yucca Mountain. Annu. Rev. Energy 
Environ. 24:461-86 
75. Hollister CD, Nadis S. 1998. Burial of radioactive waste under the seabed. Scientific 
American 276(1):60-65 
76. Nadis S. 1996. The sub-seabed solution. Atlantic Monthly 278(4):28-39 
77. Waczewski J. 1997. Legal, political, and scientific response to ocean dumping and sub-
seabed disposal of nuclear waste. Journal of Transnational Law & Policy 7(1). 
http://www.law.fsu.edu/journals/transnational/issues/7-1/wacz.html 
78. National Research Council. 1996. Nuclear Waste: Technologies for Separations and 
Transmutation. Washington, DC: National Academy Press 
79. Hebel LC, Christensen EL, Donath FA, Falconer WE, Lidofsky LJ, et al. 1978. Report to 
the American Physical Society by the study group on nuclear fuel cycles and waste 
management. Reviews of Modern Physics 50(1):S1-186. 
80. National Academy of Sciences, Committee on International Security and Arms Control. 
1994. Management and Disposition of Excess Weapons Plutonium. Washington, DC: 
National Academy Press. 288 pp. 
81. Mark JC. 1993. Explosive properties of reactor-grade plutonium. Science & Global Security 
4(1):111?128 
82. Miller MM. 1990. Are IAEA Safeguards on Plutonium Bulk-handling Facilities Effective? 
Nuclear Control Institute, Washington, DC. http://www.nci.org/mmsgrds.htm 
83. Peterson PF. 1996. Long-term safeguards for plutonium in geologic repositories. Science 
and Global Security 6(1):1-26 
 27 
84. Peterson PF. 1999. Issues for detecting undeclared post-closure excavation at geologic  
repositories. Science and Global Security 8(1):1-39 
85. Galperin A, Reichert P, Radkowsky A. 1997. Thorium fuel for light water reactors?
reducing the proliferation potential of nuclear power fuel cycle. Science and Global Security 
6(3):265-290 
86. Kasten PR. 1998. Review of the Radkowsky thorium reactor concept. Science and Global 
Security 7(3):237-269 
87. Boes EC, Luque A. 1993. Photovoltaic concentrator technology. See Ref. 30, pp. 361-401 
88. Everett B. 1996. Solar thermal energy. See Ref. 33, pp. 41-88 
89. Brower M. 1992. Cool Energy: Renewable Solutions to Environmental Problems. 
Cambridge, MA: MIT Press 
90. Energy Information Administration. 1998. Renewable Energy Annual 1997, Vol. I. 
Washington, DC: U.S. Dep. Energy 
91. De Laquil III P, Kearney D, Geyer M, Diver R. 1993. Solar-thermal electric technology. See 
Ref. 30, pp. 213-296. 
92. Boyle G. 1996. Solar Photovoltaics. See Ref. 33, pp. 89-136 
93. Kelly H. 1993. Introduction to photovoltaic technology. See Ref. 30, pp. 297-336 
94. Consumer Reports. 1997. Shingles and siding. Consumer Reports 62(8):27-33 
95. Piellisch R. 1991. Solar powered flight. Sunworld, March-April, pp. 17?20. 
96. Hopkins JB. 1994. Orbital and planetary launch services. 
http://www.ksc.nasa.gov/facts/faq13.txt 
97. Energy Information Administration. 1997. Renewable Energy Annual 1996. Washington, 
DC: U.S. Dep. Energy 
98. Gipe P. 1999. Overview of Worldwide Wind Generation. http://rotor.fb12.tu-
berlin.de/overview.html 
99. Herzog HJ. 1998. The Economics of CO
2
 Capture. Fourth International Conference on 
Greenhouse Gas Control Technologies, Interlaken, Switzerland. 
http://web.mit.edu/energylab/www/hjherzog/economics.PDF 
100. International Energy Agency. 1994. Carbon Dioxide Capture from Power Stations. IEA 
Greenhouse Gas R&D Programme, Stoke Orchard, UK. 
http://www.ieagreen.org.uk/sr2p.htm 
101. Herzog H, Drake E, Adams E. 1997. Carbon Dioxide Capture, Reuse, and Storage 
Technologies for Mitigating Global Climate Change. Massachusetts Institute of 
Technology, Cambridge, MA. 
http://web.mit.edu/energylab/www/hjherzog/White_Paper.pdf 
102. Blok K, Williams RH, Katofsky RE, Hendriks CA, Hydrogen production from natural gas, 
sequestration of recovered CO
2
 in depleted gas wells and enhanced natural gas recovery. 
Energy 22(2/3):166 
103. International Energy Agency. 1995. Carbon Dioxide Utilisation. IEA Greenhouse Gas R&D 
Programme, Stoke Orchard, UK. http://www.ieagreen.org.uk/sr4p.htm 
104. Bergman PD, Winter EM. 1996. Disposal of carbon dioxide in deep saline aquifers in the 
U.S. US/Japan Joint Technical Workshop, State College, PA. 
105. International Energy Agency. 1995. Carbon Dioxide Disposal from Power Stations. IEA 
Greenhouse Gas R&D Programme, Stoke Orchard, UK. 
http://www.ieagreen.org.uk/sr3p.htm 
 28 
106. Caulfield JA, Auerbach DI, Adams EE, Herzog HJ. 1997. Near-field impacts of reduced pH 
from ocean CO
2
 disposal. Energy Conversion and Management 38:S343?S348 
107. Skovholt O. 1993. CO
2
 transportation system. Energy Conversion and Management 34(9-
11):1095?1103 
108. Summerfield IR, Goldthorpe SH, Williams N, Sheikh A. 1993. Costs of CO
2
 disposal 
options. Energy Conversion and Management 34(9-11):1105?1112 
109. Hendriks CA. 1994. Carbon Dioxide Removal from Coal-fired Power Plants Dordrecht, 
Netherlands: Kluwer Academic Press 
110. Golumb D. 1997. Transport systems for ocean disposal of CO
2
 and their environmental 
effects. Energy Conversion and Management. 
111. President?s Committee of Advisors on Science and Technology, Panel on Energy Research 
and Development. 1997. Report to the President on Federal Energy Research and 
Development for the Challenges of the Twenty-first Century, Washington DC, Office of 
Science and Technology Policy 
 
 29 
 
TABLE 1   Current and potential contributions of carbon-free energy sources to world primary 
energy supply. 
Primary Energy Production (EJ
p
 y
?1
) 
Energy Source 
1998 
Potential 
by 2050 
Long-term 
Potential 
Natural flow or 
resource (EJ
p
) 
Hydroelectric 29 40?60 60?100 400 y
?1
 
Geothermal 0.6 5?10 5?20 10,000,000 
Ocean 0.006 0?1 1?5 2,000,000 y
?1
 
Nuclear fusion 0 0 ? 4,000,000,000+ 
Biomass 4
*
 50?150 50?500 2,000 y
?1
 
Nuclear fission 25 70?150 500+ 10,000,000 
Solar 0.5 50?150 500+ 3,000,000 y
?1
 
Wind 0.14 20?50 100?250 40,000 y
?1
 
Decarbonized fossil 0 150+ 500+ 250,000 
*
Commercial biomass only; traditional biomass is variously estimated at 15?65 EJ y
?1
. 
 
 30 
Figure 1   Scenarios of future world commercial primary energy consumption by the Intergovernmental Panel on Climate Change 
(IS92), the World Energy Council (WEC), Shell Oil, and Schmalensee et al. 
 
0
500
1000
1500
1900 1950 2000 2050 2100
World Primary Commercial Energy Consumption (EJ/yr
)
actual
IS92a
IS92b
IS92c
IS92d
IS92e
WEC A
WEC B
Shell Sustained Growth
Shell Dematerialization
Schmalensee et al mean
Schmalensee et al 5%
Schmalensee et al 95%
 31