R E V I E W : E N G I N E E R I N G

Advanced Technology Paths to Global ClimateStability: Energy for a Greenhouse Planet

Martin I. Hoffert,1* Ken Caldeira,3 Gregory Benford,4 David R. Criswell,5 Christopher Green,6 Howard Herzog,7 Atul K. Jain,8

Haroon S. Kheshgi,9 Klaus S. Lackner,10 John S. Lewis,12 H. Douglas Lightfoot,13 Wallace Manheimer,14 John C. Mankins,15

Michael E. Mauel,11 L. John Perkins,3 Michael E. Schlesinger,8 Tyler Volk,2 Tom M. L. Wigley16

Stabilizing the carbon dioxide–induced component of climate change is an energyproblem. Establishment of a course toward such stabilization will require the devel-opment within the coming decades of primary energy sources that do not emit carbondioxide to the atmosphere, in addition to efforts to reduce end-use energy demand.Mid-century primary power requirements that are free of carbon dioxide emissionscould be several times what we now derive from fossil fuels (!1013 watts), even withimprovements in energy efficiency. Here we survey possible future energy sources,evaluated for their capability to supply massive amounts of carbon emission–freeenergy and for their potential for large-scale commercialization. Possible candidatesfor primary energy sources include terrestrial solar and wind energy, solar powersatellites, biomass, nuclear fission, nuclear fusion, fission-fusion hybrids, and fossilfuels from which carbon has been sequestered. Non–primary power technologies thatcould contribute to climate stabilization include efficiency improvements, hydrogenproduction, storage and transport, superconducting global electric grids, and geoengi-neering. All of these approaches currently have severe deficiencies that limit theirability to stabilize global climate. We conclude that a broad range of intensiveresearch and development is urgently needed to produce technological options thatcan allow both climate stabilization and economic development.

More than a century ago, Arrheniusput forth the idea that CO2 from fos-sil fuel burning could raise the infra-

red opacity of the atmosphere enough towarm Earth (1). In the 20th century, thehuman population quadrupled and primarypower consumption increased 16-fold (2).The fossil fuel greenhouse theory has becomemore credible as observations accumulateand as we better understand the links between

fossil fuel burning, climate change, and en-vironmental impacts (3). Atmospheric CO2

has increased from !275 to !370 parts permillion (ppm). Unchecked, it will pass 550ppm this century. Climate models and paleo-climate data indicate that 550 ppm, if sus-tained, could eventually produce globalwarming comparable in magnitude but oppo-site in sign to the global cooling of the last IceAge (4).

The United Nations Framework Conven-tion on Climate Change aims to stabilizegreenhouse gas concentrations at levels thatavoid “dangerous anthropogenic interferencewith the climate system (5).” AtmosphericCO2 stabilization targets as low as 450 ppmcould be needed to forestall coral reef bleach-ing, thermohaline circulation shutdown, andsea level rise from disintegration of the WestAntarctic Ice Sheet (6). Wigley and col-leagues developed emission scenarios to sta-bilize atmospheric CO2 at 350, 450, 550, 650,or 750 ppm (7). They minimized early emis-sion controls by initially following a busi-ness-as-usual scenario that combines eco-nomic growth of 2 to 3% year"1 with asustained decline of 1% year"1 in energyintensity (energy use per gross domesticproduct). Much larger cuts than those calledfor in the Kyoto Protocol are needed later,because the levels at which CO2 stabilizedepend approximately on total emissions.Targets of cutting to 450 ppm, and certain-ly 350 ppm, could require Herculean

effort. Even holding at 550 ppm is a majorchallenge.

Primary power consumption today is !12TW, of which 85% is fossil-fueled. Stabiliza-tion at 550, 450, and 350 ppm CO2 by Wigleyet al. scenarios require emission-free powerby mid-century of 15, 25, and #30 TW,respectively (8). Attaining this goal is noteasy. CO2 is a combustion product vital tohow civilization is powered; it cannot beregulated away. CO2 stabilization could pre-vent developing nations from basing theirenergy supply on fossil fuels (9). Hansen etal. call for reductions in methane and blacksoot, which also cause warming (10). Suchreductions are desirable but do not addressfossil fuel greenhouse warming. The KyotoProtocol calls for greenhouse gas emissionreductions by developed nations that are 5%below 1990 levels by 2008 to 2012. Paradox-ically, Kyoto is too weak and too strong: Toostrong because its initial cuts are perceived asan economic burden by some (the UnitedStates withdrew for this stated reason); tooweak because much greater emission reduc-tions will be needed, and we lack the tech-nology to make them.

Arguably, the most effective way to re-duce CO2 emissions with economic growthand equity is to develop revolutionary chang-es in the technology of energy production,distribution, storage, and conversion (8). Theneed to intensify research on such technolo-gies now is by no means universally appre-ciated. Present U.S. policy emphasizes do-mestic oil production, not energy technologyresearch (11). Misperceptions of technologi-cal readiness also appear in the latest “Sum-mary for Policymakers” by the “Mitigation”Working Group of the IntergovernmentalPanel on Climate Change (IPCC): “. . .known technological options could achieve abroad range of atmospheric CO2 stabilizationlevels, such as 550 ppm, 450 ppm or belowover the next 100 years or more. . . . Knowntechnological options refer to technologiesthat exist in operation or pilot plant stagetoday. It does not include any new technolo-gies that will require drastic technologicalbreakthroughs. . . .” (12)

This statement does not recognize theCO2 emission–free power requirements im-plied by the IPCC’s own reports (3, 8) and isnot supported by our assessment. Energy

1Department of Physics, 2Department of Biology,New York University, New York, NY 10003, USA.3Lawrence Livermore National Laboratory, Livermore,CA 94550, USA. 4Department of Physics and Astron-omy, University of California, Irvine, CA 92697, USA.5Institute of Space Systems Operations, University ofHouston, Houston, TX 77204, USA. 6Department ofEconomics, McGill University, Montreal, Quebec H3A2T7, Canada. 7MIT Laboratory for Energy and theEnvironment, Cambridge, MA 02139, USA. 8Depart-ment of Atmospheric Sciences, University of Illinois atUrbana-Champaign, Urbana, IL 61801, USA. 9Exxon-Mobil Research and Engineering Company, Annan-dale, NJ 08801, USA. 10Department of Earth andEnvironmental Engineering, 11Department of AppliedPhysics and Applied Mathematics, Columbia Univer-sity, New York, NY 10027, USA. 12Lunar and PlanetaryLaboratory, University of Arizona, Tucson, AZ 85721,USA. 13Centre for Climate and Global Change Re-search, McGill University, Montreal, Quebec H3A 2K6,Canada. 14Plasma Physics Division, Naval ResearchLaboratory, Washington, DC 20375, USA. 15NASAHeadquarters, Washington, DC 20546, USA. 16Na-tional Center for Atmospheric Research, Boulder, CO80307, USA.

*To whom correspondence should be addressed. E-mail: marty.hoffert@nyu.edu

S C I E N C E ’ S C O M P A S S ! R E V I E W

www.sciencemag.org SCIENCE VOL 298 1 NOVEMBER 2002 981

sources that can produce 100 to 300% ofpresent world power consumption withoutgreenhouse emissions do not exist operation-ally or as pilot plants.

Can we produce enough emission-freepower in time? Here we assess the potentialof a broad range of technologies aimed atmeeting this goal.

Improving EfficiencyEfficiency is the ratio of usable energy outputto energy input. Primary energy in metastablechemical and nuclear bonds includes fossilfuels, fission fuels, and fusion fuels. “Renew-ables” are primary energy in natural fluxes(solar photons, wind, water, and heat flows).Energy conversion always involves dissipa-tive losses, losses that in many cases engi-neers have already expended considerable ef-

fort to reduce. Opportunities still exist toimprove efficiency in power generation andend-use sectors: transportation, manufactur-ing, electricity, and (indoor) climate condi-tioning (13).

The efficiencies of mature technologiesare well characterized (14, 15). Most efficientare large electric generators (98 to 99% effi-cient) and motors (90 to 97%). These arefollowed by rotating heat engines that arelimited by the second law of thermodynam-ics: gas and steam turbines (35 to 50%) anddiesel (30 to 35%) and internal combustion(15 to 25%) engines. Electrolyte and elec-trode materials and catalysts limit electro-chemical fuel cells (50 to 55% now; 70%eventually). Fuel cells may replace heat en-gines but will likely run on hydrogen. Aseamless transition would use H2 extracted

from gasoline or methanol in reformers (75 to80%). Renewable energy converters includephotovoltaic (PV) cells (commercial arrays,about 15 to 20%; theoretical peak for singlebandgap crystalline cells, $24%; higher formultiband cells, lower for more cost-effectiveamorphous thin films) and wind turbines(commercial units, about 30 to 40%; theoret-ical “Betz limit,” $59%). High-pressure so-dium vapor (15 to 20%), fluorescent (10 to12%), and incandescent (2 to 5%) illumina-tion generate more heat than light. Photosyn-thesis has a very low sunlight-to-chemicalenergy efficiency, limited by chlorophyll ab-sorption bands (most productive ecosystemsare about 1 to 2% efficient; theoretical peakindependent of cell or ecosystem is $8%).

How much can energy efficiency im-prove? In a given technology class, efficiencynormally starts low, grows for decades tocenturies, and levels off at some fraction ofits theoretical peak (16). It took 300 years todevelop fuel cells from 1%-efficient steamengines. The earliest gas turbines could bare-ly turn their compressors. The developmentof fusion could be similar: The best experi-ments are close to balancing power to ignitethe plasma; power is carried off by fusion-generated neutrons, but no net power outputhas occurred yet. Fossil and nuclear fuels aremuch closer to their limits (Figs. 1A and 4A).Steam-cycle efficiencies (39 to 50%, includ-ing combined cycles and cogeneration) andoverall primary energy–to-electricity effi-ciency (30 to 36%, including transmissionlosses) yield the nominal thermal-to-electricpower conversion: 3 kW (thermal) $ 1 kWe

(electrical). Impressive reductions in wasteheat have been accomplished with compactfluorescents, low emissivity windows, andcogeneration (17). More efficient automotivepower conversion is possible (18, 19). Emis-sions depend on vehicle mass, drivingpatterns, and aerodynamic drag, as well aswell-to-wheels efficiency [(torque % angularvelocity at wheels)/(fossil fuel power in)].Power trains are typically 18 to 23% efficientfor internal combustion (IC), 21 to 27% forbattery-electric (35 to 40%, central powerplant; 80 to 85%, charge-discharge cycles; 80to 85%, motor), 30 to 35% for IC-electrichybrid (higher efficiency from electric powerrecovery of otherwise lost mechanical ener-gy), and 30 to 37% for fuel cell–electric (75to 80%, reformer; 50 to 55%, fuel cell; 80 to85%, motor).

Lifestyles also affect emissions. Ultra fu-el-efficient cars are available today that cantravel up to 29 km liter"1 [68 miles pergallon (mpg) U.S. Environmental ProtectionAgency highway driving cycle (EPA hwy)].But consumer demand for sport utility vehi-cles (SUVs) has driven the fuel economy ofthe U.S. car and light truck fleet to a 21-yearlow of 8.5 km liter"1 (20 mpg EPA hwy)

steam

SteamTurbine

watercondensate

Condensor

Generatorelectricity

out

Pump

High-pressure Boiler

fossil fuel in

air (O2)

CO2 up the stack

or sequesteredCO2 out

CO2to cooling

tower or cold river water

and/orcogeneration

cooling waterreturn flow

A

B

fossilfuel

sequestered CO2

CO2farm residue

or marinebiomass sunkto deep ocean

ocean releaseof CO2/CaCO3

mixture orliquid CO2

CO2CO2

electricity& H2 magnesium carbonate bricks

centralpower plants

biomassfuel

Coalmines

Gas & Oilfields

Deep coal beds,subterranean

aquifers

Carbon sequestration rates to produce10 TW CO2-emission-free from fossil fuels

Fossilfuel

Gas

Oil

Coal

Energy content[TW-yr]

Carbon content[GTC]

(Efuel/C)[TW-yr/GtC]

(E/C)[TW-yr/GtC]

Sequestrationrate [GtC/yr]

12001200

4800

570750

3690

2.11.6

1.3

1.9 - 1.61.4 - 1.2

1.2 - 1.0

5 - 67 - 8

9 - 10

Fig. 1. (A) Fossil fuel electricity from steam turbine cycles. (B) Collecting CO2 from central plantsand air capture, followed by subterranean, ocean, and/or solid carbonate sequestration, could fosteremission-free electricity and hydrogen production, but huge processing and sequestration rates areneeded (5 to 10 GtC year"1 to produce 10 TW emission-free assuming energy penalties of 10 to25%).

S C I E N C E ’ S C O M P A S S

1 NOVEMBER 2002 VOL 298 SCIENCE www.sciencemag.org982

(19). Even with SUVs, doubling (or more)efficiency is quite feasible. Unfortunately, theeffects of such efficiency could be over-whelmed if China and India follow the U.S.path from bicycles and mass transit to cars.(Asia already accounts for #80% of petro-leum consumption growth.) As a result, car-bon-neutral fuels or CO2 “air capture” maybe the best alternatives to develop.

Decarbonization and SequestrationReducing the amount of carbon emitted per unitof primary energy is called decarbonization.The long-term trend has been from coal to oil togas, with each fuel emitting progressively lessCO2 per joule of heat (20). Continuation of thetrend would lead to use of H2, a carbon-neutralfuel, but H2 does not exist in geological reser-voirs. Processes requiring energy are needed forits synthesis. The energy can come from fossilfuel feedstocks. H2 is produced today by steam-reforming natural gas (2H2O & CH43 4H2 &CO2). Energy can be transferred to H2 with anefficiency of about 72% from gas, 76% fromoil, and 55 to 60% from coal (21). Per unit ofheat generated, more CO2 is produced by mak-ing H2 from fossil fuel than by burning thefossil fuel directly. Emission-free H2 manufac-tured by water electrolysis that is powered byrenewable or nuclear sources is not yet costeffective.

Thus, the decarbonization of fuels alonewill not mitigate global warming. The under-lying problem is providing 10 to 30 TWemission-free in 50 years. Continuing thetrend to lower carbon fuels requires disposingof excess carbon because the trend opposesthe relative abundance of fossil resources—high-carbon coal being most abundant, fol-lowed by oil and gas (22, 23). One vision of“clean” coal incorporates CO2 capture andsequestration: Coal and/or biomass and wastematerials are gasified in an oxygen-blowngasifier, and the product is cleaned of sulfurand reacted with steam to form H2 and CO2.After heat extraction, the CO2 is sequesteredand the H2 used for transportation or electric-ity generation (24). Decarbonization is thusintimately linked to sequestration (25). Se-questration reservoirs include oceans, trees,soils, depleted natural gas and oil fields, deepsaline aquifers, coal seams, and solid mineralcarbonates (Fig. 1B). The main advantage ofsequestration is its compatibility with exist-ing fossil fuel infrastructures, including CO2

injections for enhanced recovery from exist-ing oil and gas fields and capture of CO2

from power plant flue gases.Recovery of fossil fuel CO2 emitted from

decentralized sources (like cars) may beneeded. The simplest air capture is foresta-tion. Tree and soil sequestration does notrequire combustion product separation ormore fuel, but the capacity to absorb CO2 islimited. Uptake occurs during growth of or-

ganic matter (CH2O), when the net photosyn-thesis-respiration reaction is to the right: h'& CO2 & H2O 3 CH2O & O2. HistoricalCO2 data and models imply a temperate for-est carbon sink today of 1 to 3 billion tons ofcarbon (GtC) year"1 (3), but some modelsshow forests reversing from sinks to sourceslater this century as global warming increasessoil respiration (26). The exchange time ofCO2 with trees is !7 years. Turnover of ironfertilization–enhanced plankton uptake (27)can be similarly fast if organic detritus oxi-dizes near the surface. Biological sequestra-tion approaches to longer term storage in-clude sealing undecayed trees underground(28) and sinking agricultural residues to thedeep ocean (29). Air capture by aqueous cal-cium hydroxide [Ca(OH)2] in shallow pools,with CO2 recovery by heating CaCO3 in aretort to produce CaO and CO2, has also beenproposed (30). This reaction (calcination) is akey step in making cement from limestone,but breaking the Ca–CO2 bond requires sub-stantial energy.

Also being explored is longer term CO2

sequestration in the deep sea (31). For a givenemission scenario, ocean injections can sub-

stantially decrease peak atmospheric CO2

levels, although all cases eventually diffusesome CO2 back to the atmosphere (32). Back-diffusion and pH impacts of ocean CO2 dis-posal could be diminished by acceleratingcarbonate mineral weathering that would oth-erwise slowly neutralize the oceanic acidityproduced by fossil fuel CO2 (33, 34). Afar-reaching removal scheme is reacting CO2

with the mineral serpentine to sequester car-bon as a solid in magnesium carbonate“bricks” by vastly accelerating silicate rockweathering reactions, which remove atmo-spheric CO2 over geologic time scales (35).Thus, carbon sequestration could be a valu-able bridge to renewable and/or nuclearenergy. However, if other emission-free pri-mary power sources of 10 to 30 TW areunavailable by mid-century, then enormoussequestration rates could be needed to stabi-lize atmospheric CO2 (Fig. 1B). Substantialresearch investments are needed now to makethis technology available in time.

RenewablesRenewable energy technologies include bio-mass, solar thermal and photovoltaic, wind,

N2 (liquid)

N2 (gas)

to superconductingelectric power grid

electricpower

conditioning

liquid nitrogenchiller

to hydrogenpipelines and

storage

H2

H2O O2electrolyzerphotovoltaic

arrays

wind farms

Renewableelectricity and

hydrogenA

B

Icosahedron: Anequilateral-triangle-faced solid thatreduces map projectionerrors

Antarctica

Europe

Asia

Africa

SouthAmerica

NorthAmerica

Oceania

Buckminster Fuller's Global Electrical Grid

Fig. 2. (A) Mass-produced widely distributed PV arrays and wind turbines making electrolytic H2 orelectricity may eventually generate 10 to 30 TW emission-free. (B) The global grid proposed by R.Buckminster Fuller with modern computerized load management and high-temperature supercon-ducting (HTS) cables could transmit electricity from day to night locations and foster low-lossdistribution from remote, episodic, or dangerous power sources. (The resistivity of HTS wiresvanishes below the 77 K boiling point of nitrogen available from air.)

S C I E N C E ’ S C O M P A S S

www.sciencemag.org SCIENCE VOL 298 1 NOVEMBER 2002 983

hydropower, ocean thermal, geothermal, andtidal (36). With the exception of firewoodand hydroelectricity (close to saturation),these are collectively (1% of global power.All renewables suffer from low areal powerdensities. Biomass plantations can producecarbon-neutral fuels for power plants ortransportation, but photosynthesis has toolow a power density (!0.6 W m"2) for bio-fuels to contribute significantly to climatestabilization (14, 37). (10 TW from biomassrequires #10% of Earth’s land surface, com-parable to all of human agriculture.) PV andwind energy (!15 We m"2) need less land,but other materials can be limiting. For solarenergy, U.S. energy consumption may re-quire a PV array covering a square !160 km

on each side (26,000 km2) (38). The electri-cal equivalent of 10 TW (3.3 TWe) requires asurface array !470 km on a side (220,000km2). However, all the PV cells shipped from1982 to 1998 would only cover !3 km2 (39).A massive (but not insurmountable) scale-upis required to get 10 to 30 TW equivalent.

More cost-effective PV panels and windturbines are expected as mass production driveseconomies of scale. But renewables are inter-mittent dispersed sources unsuited to baseloadwithout transmission, storage, and power con-ditioning. Wind power is often available onlyfrom remote or offshore locations. Meeting lo-cal demand with PV arrays today requirespumped-storage or battery-electric backup sys-tems of comparable or greater capacity (40).

“Balance-of-system” infrastructures couldevolve from natural gas fuel cells if reformer H2

is replaced by H2 from PV or wind electrolysis(Fig. 2A). Reversible electrolyzer and fuel cellsoffer higher current (and power) per electrodearea than batteries, !20 kWe m"2 for protonexchange membrane (PEM) cells (21). PEMcells need platinum catalysts, # 5 % 10"3 kg Ptm"2 (41) (a 10-TW hydrogen flow rate couldrequire 30 times as much as today’s annualworld platinum production). Advanced electri-cal grids would also foster renewables. Even ifPV and wind turbine manufacturing rates in-creased as required, existing grids could notmanage the loads. Present hub-and-spoke net-works were designed for central power plants,ones that are close to users. Such networks needto be reengineered. Spanning the world electri-cally evokes Buckminster Fuller’s global grid(Fig. 2B). Even before the discovery of high-temperature superconductivity (42), Fuller en-visioned electricity wheeled between day andnight hemispheres and pole-to-pole (43).Worldwide deregulation and the free trade ofelectricity could have buyers and sellers estab-lishing a supply-demand equilibrium to yield aworldwide market price for grid-providedelectricity.

Space solar power (SSP) (Fig. 3, A and B)exploits the unique attributes of space topower Earth (44, 45). Solar flux is !8 timeshigher in space than the long-term surfaceaverage on spinning, cloudy Earth. If theoret-ical microwave transmission efficiencies (50to 60%) can be realized, 75 to 100 We couldbe available at Earth’s surface per m2 of PVarray in space, !1/4 the area of surface PVarrays of comparable power. In the 1970s, theNational Aeronautics and Space Administra-tion (NASA) and the U.S. Department ofEnergy (DOE) studied an SSP design with aPV array the size of Manhattan in geostation-ary orbit [(GEO) 35,800 km above the equa-tor] that beamed power to a 10-km by 13-kmsurface rectenna with 5 GWe output. [10 TWequivalent (3.3 TWe) requires 660 SSP units.]Other architectures, smaller satellites, andnewer technologies were explored in theNASA “Fresh Look Study” (46). Alternativelocations are 200- to 10,000-km altitude sat-ellite constellations (47), the Moon (48, 49),and the Earth-Sun L2 Lagrange exterior point[one of five libration points corotating withthe Earth-Sun system (Fig. 3C)] (50). Poten-tially important for CO2 emission reduction isa demonstration proposed by Japan’s Instituteof Space and Aeronautical Science to beamsolar energy to developing nations a fewdegrees from the equator from a satellite inlow equatorial orbit (51). Papua New Guinea,Indonesia, Ecuador, and Colombia on thePacific Rim, and Malaysia, Brazil, Tanzania,and the Maldives have agreed to participatein such experiments (52). A major challengeis reducing or externalizing high launch

Solar powersatellite

phased-arraytransmittingantenna

Dt

Dr

sun-trackingPV arrays inorbit

to grid and localpower consumers

powerregulation

microwavepower beam

DC out

pilotbeam

rectenna(rectifyingantenna)

B

Generation of power on the Moon,transmission to Earth by microwave beams

•Assembled mainly from lunar materials

Generation of power in-orbit,transmission to Earth by microwave

Transmission of power fromone point to another

•Assembled entirely from terrestrial materialsor

•Assembled partly or mainly from lunar materials

powersource

orbiting solararrays withmicrowavetransmitters

solararrays

on Moon

powerconsumer

solar reflectorsin lunar orbit

Earth-orbitingmicrowave

reflectors

orbiting microwavereflector

Sun

Sun

A

x

x

x

L3

L1

L2

L5L4

~ 0.01 AU

2% of sunlightdiffracted to

bypass Earth

10 µm thick Fresnellens near L1

aerosolscatters inthe atmosphere

EARTH

SUN

parasol screenor lens

1 AU !150 x 106 km

Earth--SunLagrange

points

C

Fig. 3. Capturing and controlling sun power in space. (A) The power relay satellite, solar powersatellite (SPS), and lunar power system all exploit unique attributes of space (high solar flux, linesof sight, lunar materials, shallow gravitational potential well of the Moon). (B) An SPS in a low Earthorbit can be smaller and cheaper than one in geostationary orbit because it does not spread itsbeam as much; but it does not appear fixed in the sky and has a shorter duty cycle (the fractionof time power is received at a given surface site). (C) Space-based geoengineering. The Lagrangeinterior point L1 provides an opportunity for radiative forcing to oppose global warming. A2000-km-diameter parasol near L1 could deflect 2% of incident sunlight, as could aerosols withengineered optical properties injected in the stratosphere.

S C I E N C E ’ S C O M P A S S

1 NOVEMBER 2002 VOL 298 SCIENCE www.sciencemag.org984

costs. With adequate research investments,SSP could perhaps be demonstrated in 15 to20 years and deliver electricity to global mar-kets by the latter half of the century (53, 54).

Fission and FusionNuclear electricity today is fueled by 235U.Bombarding natural U with neutrons of a feweV splits the nucleus, releasing a few hundredmillion eV (235U & n 3 fission products &2.43n & 202 MeV) (55). The 235U isotope,0.72% of natural U, is often enriched to 2 to3% to make reactor fuel rods. The existing!500 nuclear power plants are variants of235U thermal reactors (56, 57): the light waterreactor [(LWR) in both pressurized and boil-ing versions]; heavy water (CANDU) reactor;graphite-moderated, water-cooled (RBMK)reactors, like Chernobyl; and gas-cooledgraphite reactors. LWRs (85% of today’s re-actors) are based largely on Hyman Rick-over’s water-cooled submarine reactor (58).Loss-of-coolant accidents [Three Mile Island(TMI) and Chernobyl] may be avoidable inthe future with “passively safe” reactors (Fig.4A). Available reactor technology can pro-vide CO2 emission–free electric power,though it poses well-known problems ofwaste disposal and weapons proliferation.

The main problem with fission for climatestabilization is fuel. Sailor et al. (58) proposea scenario with 235U reactors producing !10TW by 2050. How long before such reactorsrun out of fuel? Current estimates of U inproven reserves and (ultimately recoverable)resources are 3.4 and 17 million metric tons,respectively (22) [Ores with 500 to 2000parts per million by weight (ppmw) U areconsidered recoverable (59)]. This represents60 to 300 TW-year of primary energy (60).At 10 TW, this would only last 6 to 30years—hardly a basis for energy policy. Re-coverable U may be underestimated. Still,with 30- to 40-year reactor lifetimes, it wouldbe imprudent (at best) to initiate fission scale-up without knowing whether there is enoughfuel. What about the seas? Japanese research-ers have harvested dissolved U with organicadsorbents from flowing seawater (61).Oceans have 3.2 % 10"6 kg dissolved U m"3

(62)— a 235U energy density of 1.8 MJ m"3.Multiplying by the oceans’ huge volume(1.37 % 1018 m3) gives 4.4 billion metric tonsU and 80,000 TW-year in 235U. Runoff andoutflow to the sea from all the world’s riversis 1.2 % 106 m3 s"1 (63). Even with 100%235U extraction, the flow rate needed to makereactor fuel at the 10 TW rate is five times asmuch as this outflow (64). Getting 10 TWprimary power from 235U in crustal ores orseawater extraction may not be impossible,but it would be a big stretch.

Despite enormous hurdles, the mostpromising long-term nuclear power source isstill fusion (65). Steady progress has been

made toward “breakeven” with tokamak (atoroidal near-vacuum chamber) magneticconfinement [Q § (neutron or charged parti-cle energy out)/(energy input to heat plas-ma) ) 1] (Fig. 4B). The focus has been onthe deuterium-tritium (D-T) reaction (34He & n & 17.7 MeV). Breakeven requiresthat the “plasma triple product” satisfy theLawson criteria: n % * % kT $ 1 % 1021 m"3

s keV for the D-T reaction, where n is numberdensity; *, confinement time; T, temperature;and k, Boltzmann’s constant (66, 67). Bestresults from Princeton (Tokamak Fusion TestReactor) and Europe ( Joint European Torus)are within a factor of two (68). Higher Qs areneeded for power reactors: Neutrons pene-trating the “first wall” would be absorbed by

molten lithium, and excess heat would betransferred to turbogenerators. Tritium (12.3-year half-life) would also be bred in thelithium blanket (n & 6Li 3 4He & T & 4.8MeV). D in the sea is virtually unlimitedwhether utilized in the D-T reaction or theharder-to-ignite D-D reactions (3 3He &n & 3.2 MeV and 3 T & p & 4.0 MeV). IfD-T reactors were operational, lithium bredto T could generate 16,000 TW-year (69),twice the thermal energy in fossil fuels. TheD-3He reaction (3 4He & p & 18.3 MeV) isof interest because it yields charged particlesdirectly convertible to electricity (70). Stud-ies of D-3He and D-D burning in inertialconfinement fusion targets suggest that cen-tral D-T ignitors can spark these reactions.

steamgenerator

Fig. 4. (A) The conventional LWR employs water as both coolant and working fluid (left). Thehelium-cooled, graphite-moderated, pebble-bed, modular nuclear fission reactor is theoreticallyimmune to loss-of-coolant meltdowns like TMI and Chernobyl (right). (B) The most successful pathto fusion has been confining a D-T plasma (in purple) with complex magnetic fields in a tokamak.Breakeven occurs when the plasma triple product (number density % confinement time %temperature) attains a critical value. Recent tokamak performance improvements were capped bynear-breakeven [data sources in (68)]. Experimental work on advanced fusion fuel cycles andsimpler magnetic confinement schemes like the levitated dipole experiment (LDX ) shown arerecommended.

S C I E N C E ’ S C O M P A S S

www.sciencemag.org SCIENCE VOL 298 1 NOVEMBER 2002 985

Ignition of D-T–fueled inertial targets andassociated energy gains of Q " 10 may berealized in the National Ignition Facilitywithin the next decade. Experiments are un-der way to test dipole confinement by a su-perconducting magnet levitated in a vacuumchamber (71), a possible D-3He reactor pro-totype. Rare on Earth, 3He may someday becost-effective to mine from the Moon (72). Itis even more abundant in gas-giant planetaryatmospheres (73). Seawater D and outer plan-et 3He could power civilization longer thanany source other than the Sun.

How close, really, are we to using fusion?Devices with a larger size or a larger mag-netic field strength are required for net powergeneration. Until recently, the fusion commu-nity was promoting the International Thermo-nuclear Experimental Reactor (ITER) to testengineering feasibility. Enthusiasm for ITERwaned because of the uncertainty in raisingthe nearly $10 billion needed for construc-tion. The U.S. halted ITER sponsorship in1998, but there is renewed interest amongU.S. fusion scientists to build a smaller-sized,higher-field, nonsuperconducting experimentor to rejoin participation in a half-sized, re-designed ITER physics experiment. A “burn-ing plasma experiment” could produce netfusion power at an affordable scale and couldallow detailed observation of confined plas-ma during self-heating by hot alpha particles.The Fusion Energy Sciences Act of 2001calls on DOE to “develop a plan for UnitedStates construction of a magnetic fusionburning plasma experiment for the purpose ofaccelerating scientific understanding of fu-sion plasmas (74).” This experiment is acritical step to the realization of practicalfusion energy. Demonstrating net electricpower production from a self-sustaining fu-sion reactor would be a breakthrough of over-whelming importance but cannot be relied onto aid CO2 stabilization by mid-century.

The conclusion from our 235U fuel analy-sis is that breeder reactors are needed forfission to significantly displace CO2 emis-sions by 2050. Innovative breeder technolo-gies include fusion-fission and accelerator-fission hybrids. Fissionable 239Pu and/or233U can be made from 238U and 232Th (75).Commercial breeding is illegal today in theUnited States because of concerns over wasteand proliferation (France, Germany, and Ja-pan have also abandoned their breeding pro-grams). Breeding could be more acceptablewith safer fuel cycles and transmutation ofhigh-level wastes to benign products (76). This the more desirable feedstock: It is threetimes more abundant than U and 233U isharder to separate and divert to weapons thanplutonium. One idea to speed up breeding of233U is to use tokamak-derived fusion-fissionhybrids (68, 77). D-T fusion yields a 3.4-MeV alpha particle and a 14-MeV neutron.

The neutron would be used to breed 233Ufrom Th in the fusion blanket. Each fusionneutron would breed about one 233U and oneT. Like 235U, 233U generates about 200 MeVwhen it fissions. Fission is energy rich andneutron poor, whereas fusion is energy poorand neutron rich. A single fusion breedercould support perhaps 10 satellite burners,whereas a fission breeder supports perhapsone. A related concept is the particle accel-erator-fission hybrid breeder (56): Thirty3-MeV neutrons result from each 1000-MeVproton accelerated into molten lead; uponinjection to a subcritical reactor, these couldincrease reactivity enough to breed 233U fromTh, provide electricity, and power the accel-erator efficiently (!10% of the output). Theradiotoxicity of hybrid breeder reactors overtime is expected to be substantially belowLWRs.

These ideas appear important enough topursue experimentally, but both fission andfusion are unlikely to play significant roles inclimate stabilization without aggressive re-search and, in the case of fission, without theresolution of outstanding issues of high-levelwaste disposal and weapons proliferation.

GeoengineeringNo discussion of global warming mitigationis complete without mentioning “geoengi-neering” (78, 79), also called climate engi-neering or planetary engineering on Earth andterraforming on other planets (80). Geoengi-neering in the climate change context refersmainly to altering the planetary radiation bal-ance to affect climate and uses technologiesto compensate for the inadvertent globalwarming produced by fossil fuel CO2 andother greenhouse gases. An early idea was toput layers of reflective sulfate aerosol in theupper atmosphere to counteract greenhousewarming (81). Variations on the sunblockingtheme include injecting sub-micrometer dustto the stratosphere in shells fired by navalguns, increasing cloud cover by seeding, andshadowing Earth by objects in space (82).Perhaps most ambitious is a proposed 2000-km-diameter mirror of 10-+m glass fabricat-ed from lunar materials at the L1 (83) La-grange point of the Sun-Earth system (84)(Fig. 3C). The mirror’s surface would looklike a permanent sunspot, would deflect 2%of solar flux, and would roughly compensatefor the radiative forcing of a CO2 doubling.Climate model runs indicate that the spatialpattern of climate would resemble that with-out fossil fuel CO2 (84). Engineering theoptical properties of aerosols injected to thestratosphere to produce a variety of climaticeffects has also been proposed (85). Our as-sessment reveals major challenges to stabiliz-ing the fossil fuel greenhouse with energytechnology transformations. It is only prudentto pursue geoengineering research as an in-

surance policy should global warming im-pacts prove worse than anticipated and othermeasures fail or prove too costly. Of course,large-scale geophysical interventions are in-herently risky and need to be approachedwith caution.

Concluding RemarksEven as evidence for global warming accu-mulates, the dependence of civilization on theoxidation of coal, oil, and gas for energymakes an appropriate response difficult. Thedisparity between what is needed and whatcan be done without great compromise maybecome more acute as the global economygrows and as larger reductions in CO2-emit-ting energy relative to growing total energydemand are required. Energy is critical toglobal prosperity and equity.

If Earth continues to warm, people mayturn to advanced technologies for solutions.Combating global warming by radical re-structuring of the global energy system couldbe the technology challenge of the century.We have identified a portfolio of promisingtechnologies here—some radical departuresfrom our present fossil fuel system. Manyconcepts will fail, and staying the course willrequire leadership. Stabilizing climate is noteasy. At the very least, it requires politicalwill, targeted research and development, andinternational cooperation. Most of all, it re-quires the recognition that, although regula-tion can play a role, the fossil fuel greenhouseeffect is an energy problem that cannot besimply regulated away.

References and Notes1. S. Arrhenius, Phila. Mag. 41, 237 (1896).2. J. R. McNeill, Something New Under the Sun: AnEnvironmental History of the Twentieth Century (Nor-ton, New York, 2000).

3. J. T. Houghton et al., Eds., Climate Change 2001:Scientific Basis (Cambridge Univ. Press, New York,2001).

4. M. I. Hoffert, C. Covey, Nature 360, 573 (1992).5. United Nations Framework Convention on ClimateChange, Text of the Convention, May 1992, Rio deJaneiro, Brazil (United Nations Environment Pro-gram–World Meteorological Organization, ClimateChange Secretariat, Geneva, 1992).

6. B. C. O’Neill, M. Oppenheimer, Science 296, 1971(2002).

7. T. M. L. Wigley, R. Richels, J. A. Edmonds. Nature 379,240 (1996).

8. M. I. Hoffert et al., Nature 395, 881 (1998).9. B. Bolin, H. S. Kheshgi, Proc. Natl. Acad. Sci. U.S.A. 98,4850 (2001).

10. J. Hansen, M. Sato, R. Ruedy, A, Lacis, V. Oinas, Proc.Natl. Acad. Sci. U.S.A. 97, 9875 (2000).

11. R. B. Cheney et al., National Energy Policy: Report ofthe National Energy Policy Development Group ( TheWhite House, Washington, DC, May 2001); availableat www.whitehouse.gov/energy/National-Energy-Policy.pdf.

12. B. Metz et al., Eds., Climate Change 2001: Mitigation(Cambridge Univ. Press, New York, 2001), p. 8.

13. National Laboratory Directors, Technology Opportu-nities to Reduce U.S. Greenhouse Gas Emissions (DOE,Washington, DC, 1997); available at www.ornl.gov/climate_change/climate.htm.

14. M. I. Hoffert, S. E. Potter, in Engineering Response to

S C I E N C E ’ S C O M P A S S

1 NOVEMBER 2002 VOL 298 SCIENCE www.sciencemag.org986

Global Climate Change, R. G. Watts, Ed. (Lewis, BocaRaton, FL, 1997), pp. 205–259.

15. V. Smil, Energies (MIT Press, Cambridge, MA, 1999).16. J. H. Ausubel, C. Marchetti, Daedalus 125, 139 (1996).17. A. H. Rosenfield, T. M. Kaarsberg, J. Romm, Phys.Today, 53, 29 (November 2000).

18. M. A. Weiss, J. B. Heywood, E. M. Drake, A. Schafer,F. F. AuYeung, On the Road in 2020: A Life-CycleAnalysis of New Automobile Technologies, EL 00-003(MIT Energy Laboratory, Cambridge, MA, 2000); avail-able at web.mit.edu/energylab/www/pubs/el00-003.pdf.

19. K. H. Hellman, R. M. Heavenrich, Light-Duty Automo-tive Technology and Fuel Economy Trends: 1975Through 2001, EPA420-R-01-008 (U.S. Environmen-tal Protection Agency, Washington, DC, 2001); avail-able at www.epa.gov/OMS/cert/mpg/fetrends/r01008.pdf.

20. C. Marchetti, Nucl. Sci. Eng. 90, 521 (1985).21. C.-J. Winter, J. Nitsch, Eds., Hydrogen as an Energy-Carrier (Springer-Verlag, New York, 1988).

22. N. Nakicenovic, A. Grubler, A. McDonald, Global En-ergy Perspectives (Cambridge Univ. Press, New York,1998), p. 52.

23. Energy in recoverable fossil fuels (“reserves” plus“resources” from conventional and unconventionalsources) is !4800 TW-year for coal and !1200TW-year each for oil and gas [converted from valuesin (22); 1 billion metric ton oil equivalent) 44.8 EJ)1.42 TW-year; natural gas includes liquids, excludesmethane hydrates]. Total energy in recoverable fossilfuels ranges from 6400 to 8000 TW-year, two-thirdsof which is coal. Coal resources could roughly doubleif energy in shales could be extracted cost-effectively.

24. J. P. Holdren et al., Federal Energy Research andDevelopment for the Challenges of the Twenty-FirstCentury (President’s Committee of Advisors on Sci-ence and Technology [PCAST], Washington, DC,1997); available at www.ostp.gov/Energy/index.html.

25. D. Reichle et al., Carbon Sequestration Research andDevelopment (DOE, Washington, DC, 1999); availableat www.ornl.gov/carbon_sequestration/.

26. P. M. Cox, R. A. Betts, C. D. Jones, S. A. Spall, I. J.Totterdell, Nature 408, 184 (2000).

27. J. E. Martin, S. E. Fitzwater, R. M. Gordon, Glob.Biogeochem. Cycles 4, 5 (1990).

28. F. J. Dyson, G. Marland, Workshop on the GlobalEffects of Carbon Dioxide from Fossil Fuels, ReportCONF-770385 (DOE, Washington, DC, 1977), pp.111–118.

29. R. A. Metzger, G. Benford, Clim. Change 49, 11(2001).

30. K. S. Lackner, P. Grimes, H.-J. Ziock, Carbon DioxideExtraction from Air: Is It an Option? Technical ReportLA-UR-99-583 (Los Alamos National Laboratory, LosAlamos, NM, 1999); available at www.lanl.gov/energy/ecology/carbon/docs/AirExtraction.pdf.

31. H. Herzog, B. Eliasson, O. Kaarstad, Sci. Am. 282, 72(2000).

32. H. S. Kheshgi, B. P. Flannery, M. I. Hoffert, A. G.Lapenis, Energy 19, 967 (1994).

33. K. Caldeira, G. H. Rau, Geophys. Res. Lett. 27, 225(2000).

34. Carbon dioxide dissolves in water and slowly cor-rodes carbonate sediment, resulting in the storage offossil fuel carbon primarily as bicarbonate ions dis-solved in the ocean. Left to nature, the net reactionCO2 & H2O & CaCO3 3 Ca

2& & HCO3– would

occur on the time scale of !6000 years.35. K. S. Lackner, C. H. Wendt, D. P. Butt, E. L. Joyce Jr.,D. H. Sharp, Energy 20, 1153 (1995).

36. T. B. Johansson, H. Kelly, A. K. N. Reddy, R. H. Wil-liams, Eds., Renewable Energy (Island Press, Washing-ton, DC, 1993).

37. H. S. Kheshgi, R. C. Prince, G. Marland, Annu. Rev.Energy Environ. 25, 199 (2000).

38. J. A. Turner, Science 285, 687 (1999).39. H. C. Hayden, The Solar Fraud: Why Solar EnergyWon’t Run the World ( Vales Lake Publishing, PuebloWest, CO, 2001), p. 161.

40. J. K. Strickland Jr., Sol. Energy 56, 23 (1996).41. B. C. H. Steele, A. Heinzel, Nature 414, 345 (2001).42. J. G. Bednorz, K. A. Muller, Z. Phys. B 64, 189 (1986).43. R. B. Fuller, Critical Path (St. Martins, New York,1981).

44. P. E. Glaser, Science 162, 857 (1968).45. """" , F. P. Davidson, K. I. Csigi, Eds., Solar PowerSatellites (Wiley-Praxis, New York, 1997).

46. J. Mankins, Aerosp. Am. 35, 30 (1997).47. M. I. Hoffert, S. D. Potter, in Solar Power Satellites,P. E. Glaser, F. P. Davidson, K. I. Csigi, Eds. (Wiley-Praxis, New York, 1997), pp. 231–254.

48. D. R. Criswell, in Innovative Solutions to CO2 Stabili-zation, R. Watts, Ed. (Cambridge Univ. Press, NewYork, 2002), pp. 345–410.

49. """" , Ind. Physicist 8, 12 (April/May 2002).50. G. A. Landis, in Proceedings of Solar Power Satellite’97 Conference, 24 to 28 August 1997, Montreal,Canada (Canadian Aeronautics and Space Institute,Ottawa, Canada, 1997), pp. 327–328.

51. M. Nagatomo, Sol. Energy 56, 111 (1996).52. P. Collins, Y. Purwanto, X. C. Ji, Space Energy Transp.4, 47 (1999).

53. Committee for the Assessment of NASA’s Space So-lar Power Investment Strategy, Aeronautics andSpace Engineering Board, National Research Council,Laying the Foundation for Space Solar Power: AnAssessment of NASA’s Space Solar Power InvestmentStrategy (National Academy Press, Washington, DC,2001).

54. S. J. Smith, N. Mahasenan, J. F. Clarke, J. Edmonds,Space Solar Electric Power: Future Market Share Un-der Competition (Battelle Memorial Institute, Wash-ington, DC, 2001).

55. H. G. Graetzer, Am. J. Phys. 32, 10 (1964) [translationfrom O. Hahn, F. Strassmann, Naturwissenschaften27, 11 (1939)].

56. A. M. Weinberg, in Technologies for a Greenhouse-Constrained Society, M. A. Kuliasha, A. Zuker, K. J.Ballew, Eds. (Lewis, Boca Raton, FL, 1992), pp. 227–237.

57. R. L. Garwin, G. Charpak, Megawatts and Megatons: ATurning Point in the Nuclear Age (Knopf, New York,2001).

58. W. C. Sailor, D. Bodansky, C. Braun, S. Fetter, B. vander Zwaan, Science 288, 1177 (2000).

59. M. A. Adelman, J. C. Houghton, G. Kaufman, M. B.Zimmerman, Energy Resources in an Uncertain Future(Ballinger, Cambridge, MA, 1983), p. 334.

60. Fission energy per ton ( J t"1) 235U is

202MeV

235 amu% 1.661% 10–27 kg/amu%

1.602% 10–13 J

1 MeV%103 kg

1 ton) 8.3% 1016

J

t235U

The 235U energy per ton natural U is therefore0.0072% 8.3% 1016 J t"1$ 6% 1014 J t"1, and theenergy in proven U reserves and ultimately recover-able resources is !61 and 300 TW-year, respectively(1 TW-year ) 31.5 EJ).

61. S. Nakamura, K. Oguma, Y. Ito, J. Chem. Eng. Jpn. 28,660 (1995).

62. P. Henderson, Inorganic Geochemistry (Pergamon,New York, 1982), p. 282.

63. J. P. Peixoto, A. H. Oort, Physics of Climate (AmericanInstitute of Physics, New York, 1992), p. 271.

64. The flow rate through a hypothetical seawater ex-traction plant to yield 235U fuel at 10 TW is

10 TW%1.37% 1018 m3

8% 104 TW-year%

1 year

3.15% 107 s) 5.4% 106 m3s–1.

65. T. K. Fowler, The Fusion Quest ( Johns Hopkins, Balti-more, MD, 1997).

66. J. D. Lawson, Proc. Phys. Soc. London B 70, 6 (1957).67. G. H. Miley, Fusion Energy Conversion (American Nu-clear Society, LaGrange Park, IL, 1976).

68. W. Manheimer, Fusion Technol. 36, 1 (1999).69. L. M. Lidsky, in Energy Technology Handbook, D. M.Considine, Ed. (McGraw-Hill, New York, 1977), pp.5–134.

70. G. L. Kulcinski, J. F. Santarius, Nature 396, 724 (1998).71. J. Kesner, L. Bromberg, M. Mauel, D. Garnier, J. M.Dawson, The Dipole Fusion Confinement Concept: AWhite Paper for the Fusion Community (MIT PlasmaScience and Fusion Center, Cambridge, MA, 1998);available at www.psfc.mit.edu/ldx/pubs/dipole_wp.pdf.

72. L. J. Wittenberg, J. F. Santarius, G. L. Kulcinski, FusionTechnol. 10, 167 (1987).

73. J. S. Lewis, Space Power 10, 363 (1991).74. Fusion Energy Sciences Act of 2001, H.R. 1781 (107thCongress).

75. 239Pu and 233U don’t exist in nature. They must bebred from their feedstocks, 238U (!99% of U) and232Th (!100% of thorium), with the use of suitableneutron sources (238U & n 3 239Pu; 232Th & n 3233U). Fission energy per unit mass is comparable to235U (239Pu & n 3 fission products & 2.9n & 210MeV; 233U & n 3 fission products & 2.5n & 198MeV ). We estimate breeding could increase fissionenergy resources by factors of 60 and 180 for 239Puand 233U, respectively. Breeding rates depend onfeedstock mining rates and availability of neutrons.

76. Committee on Separations Technology and Transmu-tation Systems, Board on Radioactive Waste Manage-ment, Commission on Geosciences, Environment, andResources, National Research Council, NuclearWastes: Technologies for Separations and Transmuta-tions (National Academy Press, Washington, DC,1996).

77. H. Bethe, Phys. Today 32, 45 (May 1979).78. S. H. Schneider, Nature 409, 417 (2001).79. D. W. Keith, Nature 409, 420 (2001).80. M. J. Fogg, Terraforming: Engineering Planetary Envi-ronments (Society of Automotive Engineers Interna-tional, Warrendale, PA, 1995).

81. M. I. Budyko, Climate Changes, R. Zolina, Transl., L.Levin, Ed. (American Geophysical Union, WashingtonDC, 1977).

82. Panel on Policy Implications of Greenhouse Warming,Committee on Science, Engineering, and Public Poli-cy, National Academy of Sciences, National Academyof Engineering, Institute of Medicine, Policy Implica-tions of Greenhouse Warming: Mitigation, Adaptationand the Science Base (National Academy Press,Washington, DC, 1992), pp. 433–464.

83. J. T. Early, J. Br. Interplanet. Soc. 42, 567 (1989).84. B. Govindasamy, K. Caldeira, Geophys. Res. Lett. 27,2141 (2000).

85. E. Teller, L. Wood, R. Hyde, Global Warming and IceAges: Prospects for Physics-Based Modulation ofGlobal Change, Report UCRL-231636/UCRL JC128715 (Lawrence Livermore National Laboratory,Livermore, CA, 1997).

86. We thank D. Harvey, J. Katzenberger, M. MacCracken,G. Marland, J. Palfreman, V. Smil, and P. Werbos forvaluable discussions. K.C, H.H., and A.K.J. thank theDOE-Office of Biological and Environmental Researchfor support in this work.

S C I E N C E ’ S C O M P A S S

www.sciencemag.org SCIENCE VOL 298 1 NOVEMBER 2002 987

Planning for FutureEnergy Resources

WE AGREE WITH M. I. HOFFERT ET AL.(“Advanced technology paths to globalclimate stability: energy for a greenhouseplanet,” Review, 1 Nov., p. 981) that stabi-lizing atmospheric CO2 concentrations at 550parts per million (ppm) or below will requireinvestment in energy research and develop-ment well in excess of current levels.However, their conclusion—that known tech-nological options are not up to the task—suffers from two shortcomings related to howmuch decarbonization is required and howsoon we need it. First, they do not consideruncertainty in future energy demand, basingtheir analysis on a single reference scenario(1). In contrast, the most recent Intergov-ernmental Panel on Climate Change (IPCC)report on emissions scenarios (2) foresees awide range of plausibledevelopment paths leadingto global primary powerdemand of anywhere from20 to 50 TW by 2050. Rel-ative to these scenarios, asquantified by six differentintegrated assessmentmodeling teams, stabilizingat 550 ppm may not requireany additional energy fromcarbon-free technologiesover the next 50 yearsbeyond that produced byknown technologies forreasons unrelated to climate change. Or itcould require that additional zero-carbongenerating capacity deliver nearly 600 TW-years of energy over that same period. Policyresponses to climate change should be robustacross this wide range of uncertainty.

Second, we doubt whether the develop-ment and implementation of the radically newtechnologies such as fusion or solar powersatellites advocated in the article are feasiblewithin the time horizon necessary for CO2stabilization. The process from invention, todemonstration projects, to significant marketshares typically takes between five and sevendecades (3). Fundamentally new technologies

that have not been demonstrated to be feasibleeven on a laboratory scale today would there-fore likely come much too late to contribute tothe emissions reductions necessary by 2050,particularly for stabilization at 450 ppmv orbelow (4). We believe that the appropriate mixof investments must include an initial focuson technologies with proven feasibility if weare to embark on a path to stabilization. At thesame time, we should begin to explore newenergy sources that might then be available inthe long term to finish the job.

BRIAN O’NEILL, ARNULF GRÜBLER, NEBOJSA

NAKICENOVIC, MICHAEL OBERSTEINER, KEYWAN

RIAHI, LEO SCHRATTENHOLZER, FERENC TOTH

International Institute for Applied SystemsAnalysis, Schlossplatz 1,A-2361 Laxenburg,Austria.

References1. W. Pepper et al., Emissions Scenarios for the IPCC

(Cambridge Univ. Press, Cambridge, 1992).2. N. Nakicenovic et al., Special Report on Emissions

Scenarios (Cambridge Univ. Press, Cambridge, 2000).3. A. Grübler, Technology and Global Change (Cambridge

Univ. Press, Cambridge, 1998).4. B. C. O’Neill, M. Oppenheimer, Science 296, 1971 (2002).

THE REVIEW BY M. I. HOFFERT ET AL.(“Advanced technology paths to globalclimate stability: energy for a greenhouseplanet,” 1 Nov., p. 981) discusses a wide rangeof advanced technology solutions to achievingglobal climate stability. Their treatment

of nuclear energy, however,is completely inadequate.Nuclear electric power and,with a small extension,nuclear process heat are theonly alternatives amongthose considered that havebeen tested at a commercialscale. Because noncarbonalternatives to nuclearenergy are not yet provenon a commercial scale, awide range of options forsustainably applyingnuclear technology must

receive increasing attention. In the short term, there is no fuel resource

problem. Even a trebling of capacity to meetthe Kyoto accords is possible with uraniumfuel at reasonable cost for 50 years. Beyondthis, W. C. Sailor et al. (1) estimated that one-third of a postulated (high) 900 EJ/yearprimary energy demand in a 2050 worldcould be met by nuclear fission. To meet thislevel of demand, either cheaper fuel must befound, an increased cost must be accepted, orfuel must be bred from 238U or 232Th.

Breeding plutonium from 238U wouldextend the uranium resource base by afactor of about 70; higher-cost uranium

resources would then become feasible,extending that resource for 1000 years.

Although Hoffert et al. state that “breederreactors are needed for fission to significantlydisplace CO2 emissions by 2050,” the need fora breeder reactor is less immediate than wasperceived in the 1970s. The decrease in theprice of raw uranium presently makesbreeding uncompetitive and reduces the needfor a rapid expansion, so that even more safeand economic reactor designs with a lowerbreeding ratio can now be considered.Moreover, reprocessing and recycling of spentfuel can dramatically reduce the heat load andradiotoxicity of the long-lived actinides sent toany waste repository. “Waste form modifica-tion,” therefore, is being reconsidered forimproved repository performance independ-ently of perceived uranium resource issues.

Contrary to what Hoffert et al. state,breeding as well as reprocessing has notbeen illegal since the Reagan administration.

Hoffert et al. raise concerns about nuclearenergy but do not describe how theseconcerns are being addressed. Indeed, majoraccidents have occurred at the Windscale,Chernobyl, and Three-Mile Island nuclearpower plants. Much has been learned andapplied from these events. Analyses of thesefew serious accidents have improved opera-tional safety, which was already high.

Nuclear fission technology is indeeddeeply rooted in the bomb-making military.Materials generated as a byproduct ofcommercial nuclear power might lead toundesirable proliferation of nuclear weapons.Proliferation-resistant commercial fuel cyclesare being explored, although no nuclearweapons proliferation has been attributabledirectly to a commercial power plant or theattendant fuel cycle. Inefficiencies and publicconcerns led to cost increases between 1973and 1990; however, since 1990, theeconomics of nuclear power have improvedsignificantly. Several avenues should now bedeveloped simultaneously: (i) further devel-oping low-cost uranium and (ii) improvingthe economic and environmental characteris-tics of various breeder technologies. Fossil-coal and fissile-uranium share one commonfeature—they do not have a resource problemon the time horizon of 500 years. It is theenvironmental issues, in their broadest sense,that are likely to determine the choice.

RICHARD D. WILSON1 AND ROBERT KRAKOWSKI2

1Department of Physics, Harvard University,Cambridge, MA 02138, USA. 2Los Alamos NationalLaboratory, Los Alamos, NM 87545, USA.

Reference1. W. C. Sailor, D. Bodansky, C. Braun, S. Fetter, B. van der

Zwaan, Science 288, 1177 (2000).

Letters to the EditorLetters (~300 words) discuss material publishedin Science in the previous 6 months or issuesof general interest. They can be submitted bye-mail (science_letters@aaas.org), the Web(www.letter2science.org), or regular mail(1200 New York Ave., NW, Washington, DC20005, USA). Letters are not acknowledgedupon receipt, nor are authors generallyconsulted before publication. Whetherpublished in full or in part, letters are subjectto editing for clarity and space.

LETTERS

www.sciencemag.org SCIENCE VOL 300 25 APRIL 2003 581

[I]t would befoolhardy not to

assess a broad spectrum of advanced

energy sources,converters, and

enabling technologies.”

“–HOFFERT ET AL.

L E T T E R S

25 APRIL 2003 VOL 300 SCIENCE www.sciencemag.org582

WE DISAGREE WITH M. I. HOFFERT ET AL.’S(“Advanced technology paths to global climatestability: energy for a greenhouse planet,”Reviews, 1 Nov., p. 981) characterization ofthe IPCC Third Assessment Report’s conclu-sion that “known technological options couldachieve a broad range of atmospheric stabi-lization levels, such as 550 ppm, 450 ppm orbelow over the next hundred years or more” (1,2, p. 8), as “a misperception of technologicalreadiness.” First, Hoffert et al. analyze (anddismiss) individual technologies in isolationand do not consider their full combined poten-tial. Absent detailed argumentation at theenergy system level, background reports (3, 4)suggest that their critique rests on pessimisticassessments of the availability and efficiencyof renewable energy. The IPCC evaluated abroad array of demand and supply studies, notjust individual supply-side technologies (5).Most of these studies are much less pessimisticthan Hoffert et al. about biomass, solar energy,efficiency, and fossil fuel decarbonization.Second, the authors imply that technologiesnot technically feasible today (nuclear fusionand space solar power) are needed to stabilizeconcentrations. But their development anddiffusion may require more than 50 years, toolong for timely carbon stabilization at accept-able levels. None of the studies assessed by theIPCC assumed penetration rates of new tech-nologies higher than historical experience.Third, Hoffert et al. ignore the IPCC conclu-sion that no simple technological fix exists andthat a portfolio of available technologies mustbe evaluated “in combinationwith associated socio-economic and institutionalchanges” (5). Fourth, theyignore possible carbon emis-sions reductions unrelated toenergy services, such asoptions in the area of land-usechanges.

We agree that carbon stabilization at lowlevels will be difficult and not cost-free. Weagree that enhanced R&D and investment inconventional and new technologies is neces-sary. But we stand by the IPCC conclusion thattoday’s technically feasible technologiesincluding energy efficiency improvementscould stabilize carbon concentrations if furtherdeveloped and deployed, and if complementedby necessary nonenergy initiatives and associ-ated socio-economic and institutional changes.

ROB SWART,1* JOSE ROBERTO MOREIRA,2

TSUNEYUKI MORITA,3 NEBOJSA NAKICENOVIC,4

HUGH PITCHER,5 HANS-HOLGER ROGNER6

1National Institute of Public Health and Environment(RIVM), Post Office Box 1, 3720 BA Bilthoven,Netherlands. 2Biomass Users Network (BUN), RuaFrancisco Dias Velho 814, 04581-001 CEP, Sao Paulo,Brazil. 3National Institute for Environmental Studies,16-2 Onogawa, Tsukuba, Ibaraki 305-8506, Japan.4Transitions to New Technologies Project, IIASA,

Schlossplatz 1, Laxenburg A-2361, Austria. 5JointGlobal Change Research Institute, Pacific Northwest,National Laboratory, 8400 Baltimore Avenue, CollegePark, MD 20740, USA. 6Department of Nuclear Energy,International Atomic Energy Agency (IAEA), PostOffice Box 100,A-1400,Vienna,Austria.*To whom correspondence should be addressed. E-mail: rob.swart@rivm.nl

References and Notes1. “Known” refers to “technologies that exist in opera-

tion or pilot plant stage today. It does not include anynew technologies that will require drastic technolog-ical breakthroughs” (2, p. 8).

2. B. Metz, O. Davidson, R. Swart, J. Pan, Eds., Climate Change2001 (Cambridge Univ. Press, Cambridge, 2001).

3. H. D. Lightfoot, C. Green, Report No. 2002-5 (McGillUniversity, Montreal, Canada, 2002).

4. H. D. Lightfoot, C. Green, Report No. 2002-9 (McGillUniversity, Montreal, Canada, 2002).

5. T. Morita et al., in Climate Change 2001, B. Metz, O.Davidson, R. Swart, J. Pan, Eds. (Cambridge Univ. Press,Cambridge, 2001).

Response EXISTING TECHNOLOGIES CAN CONTRIBUTEto global warming mitigation. However,projected levels of emission-free powerneeded later this century to stabilize climatechange appear to be so unprecedented (1, 2)that it would be foolhardy not to assess abroad spectrum of advanced energy sources,converters, and enabling technologies.

The IPCC Special Report on EmissionScenarios (SRES) projects 40 energyscenarios (3). Unfortunately, no reliabletheory exists to assess their probabilities. Our33 TW primary power in 2050 is close to themidcentury mean of the SRES range. Unlike

SRES, we specify a range ofconcentration targets andcompute CO2 emission–freepower required as a function oftime. We recently extended ouranalysis to global warmingtargets, including climate sensi-tivity uncertainty effects (4).

For example, a 2°C warming target (whichcan still produce adverse climate impacts)requires non–CO2-emitting primary power inthe 10 to 30 TW range by 2050.

The crux of our disagreement with theIPCC Mitigation Panel is whether “knowntechnologies”—which they define as alreadyexisting “in operation or as pilot plants”—cangenerate such massive emission-free power.Remarkably, their definition excludes fossil-fueled zero emission plants (ZEPs), with CO2sequestered. DOE just announced plans tobuild the first ZEP pilot plant by 2010–15 (5).

O’Neill et al. say that fusion and solarpower satellites are not feasible because theprocess “from invention, to demonstrationprojects, to significant market shares typicallytakes between five and seven decades.”Fusion power reactors may be unlikely beforethe latter half of the 21st century, but a fissionpath employing fusion-fission hybrid breedersbased on paid-for tokamak technology (advo-

Discussion of the Hoffert et al. Reviewcontinues online in Science's dEbates.

dEbates!

cated by Andrei Sakharov) could come onlineearlier (2, 6). Contrary to O’Neill et al. andSwart et al., both the NASA “Fresh LookStudy” and recent U.S. National ResearchCouncil assessments find space solar powerfeasible on decadal time scales (7). Leisurelymarket penetration times may apply to classicfuel substitutions, but not, historically, totechnologies accelerated by governmentresearch: Gas turbines, commercial aircraft,spaceflight, radar, lasers, integrated circuits,satellite telecommunications, personalcomputers, fiber optics, cell phones, and theInternet all developed faster (8).

What about demand? Our 10 to 30 TWemission-free requirement by 2050assumes ~2%/year growth in primarypower demand: ~3%/year GDP growth(for some measure of equity for devel-oping nations) less ~1%/year fromdeclining E/GDP (energy per unit ofGDP). The latter is where efficiencyimprovements come in (9, 10). We realizethere are many efficiency improvementspossible. The question is whether they addup to >1%/year (11).

We agree with Krakowski and Wilsonthat fission can contribute fundamentallyto global climate stability. Today, anxietiesover waste disposal and diversion toweapons are evident in Nevada’s opposi-tion to a spent nuclear fuel repository inYucca Mountain and the Pentagon’sdeployment of long-range bomberscapable of destroying North Korea’sYongbyon reactor complex. These issuesmay indeed be amenable to technical solu-tions (12). But, as indicated above, holdingglobal warming to <2°C requires 10 to 30TW emission-free power in 50 years forplausible economic growth, regardless ofpower sources. W. C. Sailor and colleaguesindependently recognized this by putting~10 TW from fission by 2050 in theirnuclear scenario (13).

Although it is no longer technically illegalin the United States, commercial breeding offissile fuels is not being done anywhere todayto our knowledge (the United States, France,Japan, and Germany have suspended theircommercial breeder reactor programs).Continued 235U burning at 10 TW rates willrequire finding major new high-gradeuranium deposits to prevent rapid exhaustion(2). Low-grade ores face serious environ-mental and cost issues. Our finding ofmassive flow rates needed for seawaterextraction of 235U surprised us. And we arenowhere near able to breed on the scaleneeded to realize theoretical factors of 60(238U → plutonium) or 180 (Th → 233U)increase in fissionable fuels. The issue forglobal warming is not breeding, as such, butour ability to breed fast enough. This willrequire drastic shifts in technology and

substantial research and development.We are astonished at continued confident

forecasts by Swart et al. that “existing” tech-nology can accomplish the mitigation jobahead, while they discount or ignore tech-nologies they deem too advanced. Expertpredictions of technological readiness arenotoriously unreliable (14). The near-termmaturity of highly desired technologies iscommonly overestimated (ballistic missiledefense, cancer cures, controlled fusion), evenas promising innovations perceived as toofuturistic are often underestimated (8, 15–17).

Market penetration rates of new technolo-gies are not physical constants. They can bestrongly impacted by targeted research anddevelopment, by ideology, and by economicincentives. Apollo 11 landed on the Moonless than a decade after the program started.We are confident that the world’s engineersand scientists can rise to the even greaterchallenge of stabilizing global warming. Butit does not advance the mitigation cause togloss over technical hurdles or to say that thetechnology problem is already solved.

MARTIN I. HOFFERT,1* KEN CALDEIRA,3 GREGORY

BENFORD,4 TYLER VOLK,3 DAVID R. CRISWELL,5

CHRISTOPHER GREEN,6 HOWARD HERZOG,7 ATUL K.JAIN,8 HAROON S. KHESHGI,9 KLAUS S. LACKNER,10

JOHN S. LEWIS,12 H. DOUGLAS LIGHTFOOT,13

WALLACE MANHEIMER,14 JOHN C. MANKINS,15

MICHAEL E. MAUEL,11 L. JOHN PERKINS,3 MICHAEL E.SCHLESINGER,8 TYLER VOLK,2 TOM M. L. WIGLEY16

1Department of Physics, 2Department of Biology,New York University, New York, NY 10003, USA.3Lawrence Livermore National Laboratory,Livermore, CA 94550, USA. 4Department of Physicsand Astronomy, University of California, Irvine, CA92697, USA. 5Institute of Space Systems Operations,University of Houston, Houston, TX 77204, USA.6Department of Economics, McGill University,Montreal, Quebec H3A 2T7, Canada. 7MITLaboratory for Energy and the Environment,Cambridge, MA 02139, USA. 8Department ofAtmospheric Sciences, University of Illinois atUrbana-Champaign, Urbana, IL 61801, USA. 9Exxon-Mobil Research and Engineering Company,Annandale, NJ 08801, USA. 10Department of Earthand Environmental Engineering, 11Department ofApplied Physics and Applied Mathematics, ColumbiaUniversity, New York, NY 10027, USA. 12Lunar andPlanetary Laboratory, University of Arizona, Tucson,AZ 85721, USA. 13Centre for Climate and GlobalChange Research, McGill University, Montreal,Quebec H3A 2K6, Canada. 14Plasma Physics Division,Naval Research Laboratory, Washington, DC 20375,USA. 15NASA Headquarters, Washington, DC 20375,USA. 16National Center for Atmospheric Research,Boulder, CO 80307, USA.*To whom correspondence should be addressed.E-mail: marty.hoffert@nyu.edu

References and Notes1. M. I. Hoffert et al., Nature 395, 881 (1998).2. M. I. Hoffert et al., Science 298, 981 (2002).3. N. Nakicenovic et al., Eds., Special Report on Emissions

L E T T E R S

www.sciencemag.org SCIENCE VOL 300 25 APRIL 2003

L E T T E R S

25 APRIL 2003 VOL 300 SCIENCE www.sciencemag.org

Scenarios (Cambridge Univ. Press, New York, 2000).4. K.Caldeira,A.K. Jain,M. I.Hoffert,Science 299,2052 (2003).5. Department of Energy, Vision 21: Power Plant of the

Future (available at www.fe.doe.gov/coal_power/vision21/index.shtml).

6. T. K. Fowler, The Fusion Quest (Johns Hopkins,Baltimore, MD, 1997), p. 195.

7. Committee for the Assessment of NASA’s Space SolarPower Investment Strategy, Aeronautics and SpaceEngineering Board, National Research Council, Layingthe Foundation for Space Solar Power: An Assessmentof NASA’s Space Solar Power Investment Strategy(National Academy Press, Washington, DC, 2001).

8. M. I. Hoffert, S. D. Potter, in R. G. Watts, Ed.,Engineering Response to Global Climate Change,(Lewis Publishers, Boca Raton, FL, 1997), pp. 205–259.

9. A. B. Lovins et al., Least-Cost Energy: Solving the CO2Problem (Rocky Mountain Institute, Snowmass, CO, 1989).

10. H. D. Lightfoot, C. Green, Energy Intensity DeclineImplications for Stabilization of Atmospheric CO2Content (Report 2001-7, McGill Centre for Climate andGlobal Change Research, Montreal, Canada, 2001).

11. Large carbon emission reductions over the 21stcentury from “efficiency” improvements reflected inE/GDP declines as high as 2%/year (a factor of 7.2 by2100) have been proposed (3, 9). However, theaverage global energy intensity decline over the 20thcentury was <1%/year, and some analyses of thecombined potential from sectoral change and engi-neering efficiency improvements suggest that1%/year may be an upper limit (2, 10).

12. R. Krakowski, R. Wilson, in Innovative Energy Strategiesfor CO2 Stabilization, R. G.Watts, Ed. (Cambridge Univ.Press, Cambridge, 2002), pp. 211–323.

13. W. C. Sailor, D. Bodansky, C. Braun, S. Fetter, B. van derZwaan, Science 288, 1177 (2000).

14. A. C. Clarke, Profiles of the Future:An Inquiry Into the Limitsof the Possible (Holt,Rinehart & Winston,New York,1982).

15. D. Walter, Today Then: America’s Best Minds Look 100Years into the Future on the Occasion of the 1893World’s Columbian Exposition (America & WorldGeographic Pub., St. Helena, MT, 1992).

16. J.Verne, Paris in the Twentieth Century, translated by R.Howard (Random House, New York,1996).

17. This “lost novel” by Jules Verne (16) written in the 1860swas rejected for publication in its time because it pictureda future too strange to be credible. In this work, Verneimagined a future in the 1960s where people traveled bysubway and in gas-driven cars, where they communi-cated by fax and telephone, where they used computers,and where “electric concerts” provided entertainment. Inthis world, everyone could read, but no one read books. Itwas a society dominated by money where destitutehomeless people roamed the streets. Strange indeed.

CORRECTIONS AND CLARIFICATIONS

Technical Comments: Response to a Comment on“No major schizophrenia locus detected on chromo-some 1q in a large multicenter sample” by D. F.Levinson (20 Dec., www.sciencemag.org/cgi/content/full/298/5602/2277a). In further discussionafter publication, the authors of the TechnicalComment (A. S. Bassett et al.) and the Response(Levinson et al.) have concluded that there was anerror in the Response. The empirical P valuesreported by L. M. Brzustowicz et al. [Science 288, 678(2000)] were incorrectly interpreted in the Responseas pointwise (uncorrected) values, but they wereactually corrected for multiple testing, as describedby F. Bonnet-Brilhault et al. [Eur. J. Hum. Genet. 7, 247(1999)] and C. R. Cloninger et al. [Am. J. Med. Genet.81, 275 (1998)]. The genome-wide P value forlinkage to schizophrenia on proximal 1q in theCanadian sample was 0.0002 to 0.00002, a highlysignificant result. The Response also noted thatsignificant linkage had not been reported in thelargest family in the Brzustowicz et al. sample. As apoint of clarification, the Zmax in this family atD1S1679 was 2.98 under a recessive model of inher-itance, considering individuals with schizophrenia orschizoaffective disorder as affected. Single-family lodscores were not presented in the original publicationbecause of space limitations.

TECHINICAL COMMENT ABSTRACTSCOMMENT ON “Arsenic Mobility and Groundwater Extraction inBangladesh” (I)Pradeep K. Aggarwal, Ashish R. Basu, Kshitij M. Kulkarni

Harvey et al. (Reports, 22 November 2002, p. 1602) concluded that irrigation pumping caused an influx of labile,carbon-laden recharge water in Bangladesh aquifers. In contrast, we present groundwater tritium data indicatingsimilar vertical flow times in pre- and postirrigation pumping periods, and long-term water level records showingconsistent seasonal fluctuations over a 30-year period.Full text at www.sciencemag.org/cgi/content/full/300/5619/584b

COMMENT ON “Arsenic Mobility and Groundwater Extraction in Bangladesh” (II)Alexander van Geen, Yan Zheng, Martin Stute, Kazi Matin Ahmed

Harvey et al. (Reports, 22 November 2002, p. 1602) claimed that elevated groundwater arsenic levels inBangladesh are linked to water pumping for irrigation. This does not appear to be supported by their data andother data indicating high arsenic concentrations in groundwater recharged well before the onset of massive irri-gation in the region.Full text at www.sciencemag.org/cgi/content/full/300/5619/584c

RESPONSE TO COMMENTS ON “Arsenic Mobility and Groundwater Extractionin Bangladesh”C. F. Harvey, C. Swartz, A. B. M. Badruzzaman, N. Keon-Blute, W. Yu, M. A. Ali, J. Jay, R. Beckie,V. Niedan, D. Brabander, P. Oates, K. Ashfaque, S. Islam, H. Hemond, M. F. Ahmed

Hydraulic and geochemical data indicate that groundwater flow, and hence pumping, influence arsenic concen-trations at our site.These data contradict the van Geen et al. claim that arsenic is mobilized in stagnant water andthe Aggarwal et al. claim that flow is unaffected by pumping. Their contrasting regional generalizations, craftedwith selected data from the same set, contain serious inconsistencies and ignore basic hydrologic processes.Full text at www.sciencemag.org/cgi/content/full/300/5619/584d