Rechargeable Lithium Batteries with Aqueous Electrolytes Author(s): Wu Li, J. R. Dahn, D. S. Wainwright Source: Science, New Series, Vol.

264, No. 5162 (May 20, 1994), pp. 1115-1118 Published by: American Association for the Advancement of Science Stable URL: http://www.jstor.org/stable/2885126 Accessed: 01/06/2010 13:37
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=aaas. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Association for the Advancement of Science is collaborating with JSTOR to digitize, preserve and extend access to Science.

http://www.jstor.org

mma relativeto the crater diameter is verytypical. Melt production increases with impactenergy, such that only largecraters areexpected to produce significantquantities of melt (31). This melt is deposited as thin veneers, flows,andponds(32), so thatmelt sufficient to affectthe spectralreflectivity is expected to remainnearthe surface only forrelatively large and young craters, postdatingwidespreadresurfacing fromimpactbasins. Smrekarand Pieters (33) showed that reflectance spectraof the anomalously red crater ringsof Tycho and Copernicusare consistent with the presence of iron-bearing glass. The melt is concentrated nearthe crater rim because it is producedfrom the deepest ejected targetmaterial.Therefore,we expect that all craters larger than about25 km in diameter and postdating the formation of the Imbrium basinwill have anomalously red near-rimcolors. This generalization is observedin the EM-2 data for the near-side cratersCopernicus,Zucchius, Pythagoras, Carpenter,Philolaus, Plato, Archimedes, Theophilus, Langrenus, Plinius, Geminus, Fabricus,and Hainzel, as well as for the far-side cratersOhm, Vavilov, and Hausen (5). In the case of Copernicanage craters, the spectral effectsof freshsoilspartlycounteract the reddishmelt, but they are still redder(and darker)than melt-freesoils of similarage and composition.The most recent of the large Copernicanage craters, such as Tycho, Jackson,and Kepler,have red and darkringsas compared relatively to craterinteriors and outerejecta and rays. The flybysof the moon by Galileo in 1990 and 1992 providedcalibratedmultispectralcoverageof about75%of the lunar surfacein the 400- to 1000-nm spectral region. The images cover well-characterized regions of the near side, including standard areasfor remotesensingand sample sites, that have enabledthe calibration of Galileodataforextrapolation to areason the northern, eastern, and western limbs and the far side that previously lacked quantitativemultispectral coverage.
REFERENCESAND NOTES
1. M.J. S. Beltonet al., Science 255, 570 (1992). 2. J. W. Head et al., J. Geophys. Res. 98, 17149 (1993). 3. C. M. Pieters etal., ibid., p. 17127. 4. R. Greeley et al., ibid., p. 17183. 5. A. S. McEwenet al., ibid., p. 17207. 6. T. B. McCord, M.P. Charette, T. V. Johnson, L.A. Lebofsky,C. M. Pieters,ibid. 77, 1349 (1972). 7. C. Pieters, Proc. LunarPlanet.Sci. Conf.9, 2825 (1978). 8. The SSIfilter set is described by M.J. S. Beltonet al. [Space Sci. Rev. 60, 413 (1992)] and its relation to lunar compositional studies is described in (1). 9. B. K. Lucchitta, U.S. Geol. Surv.Misc.Invest.Ser. Map 1-1062(1978). 10. D. E. Wilhelms, U.S. Geol. Surv.Prof Paper 1348 (1987), p. 1. 11. W. K. Hartmann and G. P. Kuiper,LunarPlanet. Lab. Comm.1 (no. 12), plate 12.51 (1962).
12. T. V. Johnson etal., Proc. Lunar Planet. Sci. Conf. 8, 1013 (1977). 13. K. A. Howard, D. E. Wilhelms, D. H. Scott, Rev. Geophys. Space Phys. 12, 309 (1974). 14. D. E. Wilhelms and J. F. McCauley, U.S. Geol. Surv. Misc. Invest. Ser. Map 1-703 (1971); N. J. Trask and J. F. McCauley, Earth Planet. Sci. Lett. 14, 201 (1972); W. R. Muehlberger et al., in "Apollo 16 Preliminary Science Report," NASA SP-315 (1972), p. 6-1; G. E. Ulrich et al., U.S. Geol. Surv. Prof. Paper 1048 (1981), p. 1; J. M. Boyce et al., Proc. Lunar Planet. Sci. Conf. 5, 11 (1974); V. R. Oberbeck etal., Moon 12, 19 (1975). 15. G. Neukum, Moon 17, 383 (1977); Habilitationsschriff, Ludwig Maximilianis Universitat, Munich (1983); G. Neukum et al., Moon 12, 201 (1975). 16. P. H. Schultz and P. D. Spudis, Proc. Lunar Planet. Sci. Conf 10, 2899 (1979). 17. J. W. Head and L. Wilson, Geochim. Cosmochim. Acta 56, 2155 (1992). 18. B. R. HawkeandJ. F. Bell, Proc. Lunar Planet. Sci. Conf. 12, 665 (1981); J. F. Bell and B. R. Hawke, J. Geophys. Res. 89, 6899 (1984). 19. B. R. Hawke et al., Lunar Planet. Inst. Tech. Report 92-09 (1992), p. 14. 20. J. W. Head et al., Lunar Planet. Sci. Conf 24, 629 (1993). 21. D. A. Williams, thesis, Arizona State University (1992); S. D. Kadel, thesis, Arizona State University (1993). 22. M. P. Charette, T. B. McCord, C. M. Pieters, J. B. Adams, J. Geophys. Res. 79, 1605 (1974); M. P. Charette, L. A. Sodenblom, J. B. Adams, M. J. Gaffey, T. B. McCord, Proc. Lunar Planet. Sci. Conf 7, 2579 (1976). 23. M. R. Robinson, B. R. Hawke, P. G. Lucey, G. A. Smith, J. Geophys. Res. 97, 18 (1992). 24. D. E. Wilhelms and F. El Baz, U.S. Geol. Surv.

REPORTS

Misc. Invest. Ser. Map 1-948 (1977). 25. J. W. Head et al., in Mare Crisium: A View from Luna 24, R. B. Merrill and J. J. Papike, Eds. (Pergamon, New York, 1978), pp. 43-74. 26. B. L. Barsukov et al., Proc. Lunar Planet. Sci. Conf. 8, 3319 (1977). 27. L. R. Gaddis, C. M. Pieters, B. R. Hawke, Icarus 61, 461 (1985). 28. L. Wilson and J. Head, J. Geophys. Res. 86, 2971 (1981). 29. G. Heiken et al., Geochim. Cosmochim. Acta 38, 1703 (1974). 30. B. R. Hawke et al., Proc. Lunar Planet. Sci. Conf 21, 377 (1991). 31.. R. A. F. Grieve et a/., in Impact and Explosion Cratering, D. J. Roddy, R. 0. Pepin, R. B. Merrill, Eds. (Pergamon, New York, 1977), pp. 791-814. 32. B. R. Hawke and J. W. Head, ibid., p. 815. 33. S. Smrekar and C. M. Pieters, Icarus 63, 442 (1985). 34. We thank the Galileo Project Office, the National Aeronautics and Space Administration (NASA), and numerous individuals who participated in the mission planning and initial data analysis, including C. Avis, T. Becker, L. Bolef, N. Bridges, C. Cunningham, E. DeJong, K. Edwards, E. Fischer, R. Garstang, A. Harch, S. Kadel, R. Kirk, J. Moersch, J. Plutchak, M. Robinson, R. Sullivan, W. Sullivan, J. Sunshine, S. Vail, L. Wainio, D. Williams, and J. Yoshimizu. The Galileo data set is available on CD-ROM through the NASA Planetary Data System. The National Optical Astronomy Observatories are operated by AURA, Inc., under a cooperative agreement with the National Science Foundation. 17 September 1993; accepted 24 March 1994

Rechargeable LithiumBatteries with Aqueous Electrolytes

Wu Li, J. R. Dahn,*D. S. Wainwright
Rechargeable lithium-ionbatteries that use an aqueous electrolyte have been developed. Cells with LiMn2O4and V02(B) as electrodes and 5 M LiNO3 in water as the electrolyte provide a fundamentally safe and cost-effective technology that can compete with nickelcadmium and lead-acid batteries on the basis of stored energy per unit of weight.

During the 1970sand 1980s, rechargeable lithium batterieswere touted by some researchers as providinga possiblelong-term solutionto the electricvehicle (EV)battery problem. The cells had about twice the energy density (measuredby watt-hours storedper kilogramof battery)of the best competing ambient temperature batteries. Many companiesmoved to commercialize the technology,beginningwith small cells for consumerapplications,in view of the largemarketsanticipated.These cells used lithium metal as the negative electrode, a transitionmetal oxide [suchas MnO2 (1)] or chalcogenide[suchas NbSe3 (2)] as the
W. Li and J. R. Dahn, Department of Physics, Simon Fraser University, Burnaby, British Columbia, Canada V5A 1S6. D. S. Wainwright, Moli Energy (1990) Limited, 20000 Stewart Crescent, Maple Ridge, British Columbia, Canada V2X 9E7. *To whom correspondence should be addressed.

positive electrode, and a nonaqueouselectrolytecontainingdissolvedLi ions. The operationof such cells is based on the abilityof the positiveelectrodematerial to reversibly "intercalate" Li. Intercalation is the insertion of a guest atom (Li, for
instance) into a host solid (such as MnO2),

accompanied by only slight, reversible structuralchanges in the host. Hosts for intercalationare commonly layered compoundssuch as graphite(3) or tunnel compoundssuch as MnO2or LiMn2O4 (4), in which the intercalatedLi can reside between the layersor in the tunnels. Intercalation of Li occurs because the chemical potentialof Li can be loweredwhen the Li atomis insertedinto the host, thus forming chemicalbonds. The bindingenergyof Li when intercalated into a varietyof hosts has been measuredwith respectto the bindingenergyof Li metal (Fig. 1). In each of the hostslisted,
1115

SCIENCE * VOL. 264 * 20 MAY 1994

the Li atomscan diffusereadilywithin the is near4 host, even whenthe bindingenergy eV. (LiFis includedin the figurefor reference, but it is not an intercalationcompound.) BecauseLi can be added to and it is common fromthesematerials, removed to designatethem as Li_(Host),with x designatingthe Li stoichiometry. The voltage and operationof cells with of Li metalarenow easy negativeelectrodes of the Duringthe discharge to understand. cell, Li at the negativeelectrodedissociates into ions and electrons that move to the positive electrode through the electrolyte and the externalcircuit, respectively.The of the ions andelectronsmeet at the surface positive electrode, where they intercalate within the host material.The open circuit voltage of the cell is given by the binding energyof Li in the host (in electronvolts), dividedby the electron charge. (Formally, the cell voltage is given by the difference betweenthe chemicalpotentialof Li in the host and in Li metal, divided intercalation by the chargeon the electron.)To recharge the cell, a chargeris used to extract elecions from the trons and the corresponding host, and Li metal is replatedon the negative electrode. cells with negative elecRechargeable trodesof Li metal can be made that have excellent performancecharacteristics(1, 2). However, it seems difficult to make them safe (5). At the heart of the safety is the reactivityof metallicLi with problem the nonaqueouselectrolytesused in these cells. Lithium reacts instantly with these electrolytesto form a passivatingfilm of on the Li metal insolublereactionproducts surface.Once the filmis about50 Athick, the reaction stops. Although the film is a Li-ionconductor,it is an electronicinsulaof Li atoms tor that preventsthe transport to the electrolyte,once the filmthicknessis greaterthan the electrontunnelinglength. However, as cells are repeatedlycharged the Li metal becomesvery and discharged, porous,and the areaof contactbetweenLi and electrolytebecomes very large. After severalhundredcycles, such cells become (suchas producing proneto safetyproblems intense smoke or even fire) if they reach aboveabout 120?Cbecauseof temperatures or electrical abuse.This problem mechanical led to a major recallin 1989involvproduct by Moli Energy ing cells manufactured (Bumaby,BritishColumbia,Canada) (5). planned introductions MostotherLi battery by otherfirmswerethen abandoned. To solve the safety problemassociated with Li metal, batteryscientistseliminated Li metalbut triedto retainthe high energy density attainable with the technology.
The so-called Li-ion cells use two different intercalation hosts as the positive and negative electrodes. For example, Sony Ener1116

6eV

--LiF

Air stable

5e e

Fig. 1. The binding energy of Li, intercalated within a variety of materials, measured relative to that of Li metal.

LiNiVO4
XNiO2, Li, -XCoO2 VLil - xMn2O4

4 eV VLi

3 eV 2 eV

t-Li MnO x 2

LXV(B
LiXV02(B)

A~~~~
U LixS2

LixMoS2

LixM0?2

1 eV ,LixWO3
LixCoke

metal 0 eV L Lithium

LixGraphite

the 3.5-V LiCoO2/carbon gytec introduced cell in 1992, and it has been verysuccessful (6). The Sony cell and others like it are the state-of-the-art powersources currently electronics.In sucha cell, the forconsumer electrodematerialsare chosen so that the chemical potentialof Li in each differsby severalelectron volts, with the result that high energydensitycan still be maintained. becausethe surface Safetyis also improved, area of the electrodes remains constant duringcycling, even thoughLi-intercalated carbonis almost as reactive as metallic Li (see Fig. 1). It is unclear,however,whether cells largeenough for EVs will be safe. is also expensive. The Li-iontechnology have ion conductivelectrolytes Nonaqueous lowerthan of magnitude itiesabouttwoorders so cell designs their aqueouscounterparts, (0.1 mm) must thin electrodes incorporating be used. Suitable Li salts, such as LiPF6, are also expensive, as are the microporous films used to separatethe electrodes and hold the electrolyte. Water must be rigorouslyexcludedduringsome manufacturing steps, which leads to additionalcosts.

Lithiumbatteries were historicallydesignedwith metallic Li as one component, which precludedthe considerationof water as an electrolytesolvent. The reaction productof Li and water, LiOH, is soluble in water, so that a passivatingfilm does not form and a violent reaction occurs. Similarly, in the current generation of carbon Li-ioncells that use Li-intercalated as the negative electrode, watercannot be usedas the electrolyte. However,once the Li in an intercalation host is sufficiently tightly bound (at about 3.2 ? 0.2 eV), it will not react with water to form LiOH
and H2 (7).

When a Li intercalationcompoundis put in water, one must considerwhether the reaction +xH20'= Host Lix(Host) (1) + xLiOH(aq) +x/2H2(g) will occur. We recently showed (8) that intercalationcompoundsthat are unstable in pure water can be made stable in concentrated LiOH solution. The chemical potentialof Li' in solution increaseswith

1.5 1.00.5
>

[H+]= 1 M 2H20 4H+ + 4e + 02 Lii-xMn2O4

[OH-] = 1 M

40H-O 2H20 + 02 + 4e 2H+ + 2e- -H2 Li1+xMn204 LiXV02(B) 2H20 + 2eH2 2 +20H-

w en -0.5
(0 Xi

0.0

Fig. 2. The potentialsof the indicated reactions versus the standard hydrogen electrode (SHE) in solutions where [Li+]= 1.0 M. Both acidic and basic electrolyteswith 1 M H+ or 1 MOH-, respectively,are considered.

-1.02

-1.5-

g -2.0-

-2.5-3.0-3.5 Li -Li++ e

SCIENCE

VOL. 264 * 20 MAY 1994

i REPORTS Tabletelectrodes compounds. intercalation 8 mm in diameterand with a total massof from each material. 0.10 g were prepared Cellcap Carbonblack (10%by weight) and ethylene propylenediene monomerbinder (3% Negativeelectrode by weight)were addedto the tabletmix to provide good electrical conductivity and good mechanicaltoughness.The mix was in a cylinin 0. 10-gallotments then pressed thickness(1.0 mm). dricaldie to the desired Separator Coin-type test cells were made with Positiveelectrode (the firsttwo numbersgive 1225 hardware the cell diameter in millimetersand the Cell can secondtwo give the cell height in tenths of Fig. 3. A schematic of the coin cell used to test a 1225 cell is 12 mm in diammillimeters; the aqueous Li-ion chemistry. eter and 2.5 mm thick). First, the electrodes were thoroughly wetted with an aqueoussolutionthat was 5.0 M in LiNO3 concentrationand tends to drive Eq. 1 to that Li-ioncells with and about 0.001 M in LiOH. A microthe left. This suggests aqueouselectrolytescan be consideredfor porous polypropyleneseparator (Celgard a wettingagent, chosen hosts. 3500), which incorporates appropriately We mustalso considerthe potentialsfor was used to separatethe electrodesas they in the cell case. Figure oxygen and hydrogenevolution that can weretightlycrimped occurin aqueouselectrolytes(Fig. 2). Fig- 3 shows a schematic of the finished cell. ure 2 also shows the standardelectrode Becausethe concentrationof OH is 0.001 M in this cell, we expect the potentialsfor potentialfor the reaction H2 and O2 evolution to shift up by about Li Li+ + e(2) with theirpositionsin 1 0.175 V, compared Figure1 lists the chemicalpotentialsof Li M OH-, as shown on the right of Fig. 2. with a current The cell wasthen charged intercalationcompoundsrelative to reaction 2, so we can locate electrodematerials of 1 mA. Duringthe charge,Li wasextractproducing on Fig. 2, as we have done for the reactions ed fromLiMn2O4, LilxMn2O4, into V02(B), producing and intercalated = LiMn2O4 Li,_.Mn204 + xLi+ + xeLiXVO2(B).Figure 4 shows the voltage (3) profileduringthis chargingand duringthe and cycles. The cell next few charge-discharge V02(B) + xLi+ + xe- ' LixVO2(B) shows excellent reversibility,an average (4) voltage near 1.5 V, and a capacityof 10 of mA hour. form a crystal ["B"designates particular chosen elecWe calculatedthe energydensityof the V02 (9).] In an appropriately trolyte, reactions3 and 4 shouldbe viable cell usingthe weight of the electrodes,the production cell voltage, and the cell capacity (we did without excessivesimultaneous of ?2 or H2. A cell basedon reactions3 and not include the electrolyte or cell-case weight). This cell's energy density is 75 4 will have a terminalvoltage near 1.5 V. watt hours/kg.Typically, the active elecWe synthesizedLiMn204 accordingto trode weights are about 50% of the total the methods described in (10) and V02 (B) in (11). to the methodsdescribed weight of practicalcells, if the Sony Li-ion according are framework-type product is used as an example. Thus, Both of these materials practical energy densities near 40 watt hours/kg can be expected for this .. .... 2.0 . . chemistrywhen used in largercells. Furthermore, the theoretical energy density for this cell is 112 wattehours/kg, assuming can be metal Li transition 0.5 cycledin per 1.5each electrode and that the average cell would voltage is 1.5 V. These assumptions 0 in a practical give about 55 watt-hours/kg with both Pb-acid cell, whichis competitive 0O.5 and Ni-Cd (about (about30 watt-hours/kg) 50 wattehours/kg) technologies. 0.01 In basic electrolytes, 02 evolution at 0 10 20 30 40 50 60 70 80 906100 Time (hours) Lil XMn2O4 may occur simultaneously with the removalof Li fromthe electrode, Fig. 4. The voltage plotted versus time as a
cell is repeatedly charged LiMn2O 4NO12(B) and discharged. The currents used were ?1 mA.The cell test was carriedout at 30?C. as indicated in Fig. 2. (This is analogous to the ?2 evolution that occurs in Ni-Cd cells during charging.) The ?2 should then difSCIENCE * VOL. 264 * 20 MAY 1994 Sealing gasket

fuse to the LiXVO2(B) electrode, where it will recombine with water and electrons to produce OH- again. Therefore, it should be possible to select electrolytes of appropriate pH that allow for efficient charging of the positive electrode and that also provide effective oxygen pressure control. The LiMn2O4/VO2 (B) couple by no means represents the optimum electrode pair for the aqueous Li-ion cell. Many types of Li manganese oxides are known to intercalate Li in the right chemical potential range for this type of cell (12). One approach to an extremely low-cost system would be to use LiMn2O4 for both electrodes. Because Li can be extracted from LiMn2O4 (to form Li1_xMn2O4) at 4 V versus Li and added to LiMn2O4 (to form Li1+XMn204)at about 3 V versus Li, a 1-V cell would result. Manganese and its oxides are cheap, plentiful, and less toxic than Ni and Co oxides, so a Li1+xMn2O4/ Li1l XMn2O4cell could have an enormous potential market. In a 1989 review of about 40 battery technologies for EV applications, Ratner et al. (13) concluded that there was no suitable technology currently available for electric vehicles. The review considered factors such as cost, safety, performance, and environmental friendliness. Today, the search for an acceptable EV battery continues. For example, the Ni-metal hydride battery developed by Ovonics (Warren, Michigan) has been shown to attain 80 wattehours/kg in sizes acceptable for EVs (14). However, some industry experts contend that the Ni-metal hydride technology is too expensive and is not environmentally acceptable for EV applications. It is our opinion that the aqueous Li-ion approach described here shows promise and needs to be included in the hunt for the EV battery. REFERENCESAND NOTES
1. K. Brandt, paper presented at the Fourth International Seminar on LithiumBattery Technology and Applications, Deerfield Beach, FL, 5 to 7 March 1989.

Annual 2. J. Broadhead,in Proceedingsof the Third and AdvancBatteryConferenceon Applications

es, California State University, Long Beach, CA, 12 to 18 January 1988; S. Basu and F. A. TrumSoc. 139, 3379 (1992). bore, J. Electrochem. M. S. Dresselhaus and M. Endo, in Graphite Intercalation Compounds 11,H. Zabel and S. A. Solin, Eds. (Springer-Verlag,Berlin,1992), pp. 347-407; J. R. Dahn et al., Electrochim. Acta 38,1179 (1993). M. M. Thackeray et al., J. Electrochem. Soc. 139, 363 (1992); M. M. Thackeray and R. J. Gummow, U.S. Patent 5,240,794 (1993). For example, see "Cellular phone recall may cause setback for Moli," Toronto Globe and Mail (Canada), 15 August 1989. T. Nagaura and K. Tozawa, Prog. Batteries Solar Cells 9, 209 (1990); G. Stix, Sci. Am. 269, 108 (October 1993). J. R. Dahn, U. von Sacken, M. W. Juzkow, H. Soc. 138, 2207 (1991). Al-Janaby, J. Electrochem. W. Li, W. R. McKinnon, J. R. Dahn, ibid., in press.

3.

4.

5.

6.

7. 8.

1117

ited by the slow diffusionof the molecules, because moves can be made to arbitrary positions in the zeolite. For chain molecules, however, this is not the case because of findinga positionwithout the probability overlap between hydrocarbonand zeolite decreasesexponentiallywith chain length. Recently, we have developed a method, Monte Carlo, to simuconfigurational-bias March 1994 9 February1994; accepted 30 late chain molecules (4, 5). We demoncan be usedto stratehere that this approach studythe behaviorof long chain hydrocarthe bons in zeolitesand allowsus to address much debated question of the preferential of the n-alkanesin the different adsorption of silicalite. channels Siepmannt BerendSmit*and J. Ilija Monte We used the configurational-bias of the adsorption of to study the adsorption technique simulate Carlo to applied is technique Carlo Monte The configurational-bias long chain alkanes in zeolites. This simulationtechnique is several orders of magnitude n-butane to n-dodecane in silicalite. In contrastto the conventional Monte Carlo more efficientthan conventionalmethods that can be used to simulatethe adsorptionof long chain alkanes. The calculated heats of adsorption are found to be in excellent technique, in the configurational-bias Monte Carlo technique a molecule is not chain lengthdependence data. The resultsshow a surprising agreementwithexperimental of the heats of adsorption.This dependence has a simple molecularexplanationin terms inserted at randombut is grown atom by atom such that overlap with the zeolite siting of the long chain alkanes. of preferential atoms is avoided (Fig. 1). This growing processintroducesa bias that is removedby adjusting the acceptance rules (4-7). A in cycles, and each simulations,usedwith molec- simulationis performed Computer Zeolites are crystallineinorganicpolymers ular dynamicsor Monte Carlo techniques, cycle consists of a number of randomly network of that form a three-dimensional are an attractivealternativeto experiments selected moves: displacementof particles, micropores.These pores are accessible to of a internal large because these methods can, in principle, rotationof particles,partialregrowing The molecules. variousguest of a molecule at a surface,the thermalstability,and the pres- provide informationfor conditions under molecule, and regrowing arenot feasible.Indeed, randomlyselected position. For the latter ence of acid sites make zeolites an impor- which experiments Monte two moves, the configurational-bias tant class of catalytic materialsfor petro- over the last few years there has been in simulatingadsorp- Carlo technique is used with a total simuprogress considerable chemicalapplications.Fora rationaluse of lation consistingof at least 106cycles. The zeolites, it is essential to have a detailed tion in zeolites [for a recent review, see with the modelof (8). alkanesaredescribed simucomputer In however, practice, knowledgeof the behaviorof the adsorbed (2)]. of description molecules inside the pores of the zeolites. lations have been limited to atomsor small This modelyieldsan accurate theirphasebehavior.FollowingKiselevand Unfortunately, such information is very molecularguestmoleculesand could not be difficult to obtain, particularlyfor long extended to molecules of catalytic rele- co-workers(9), we assumethat the zeolite vance. These limitations are discussedby lattice is rigid and that the zeolite-alkane molecules. chain hydrocarbon Catalyticconversioninsidethe poresof a June et al. (3); in this work, molecular interactions are dominated by dispersive interactions.Latticevibrationscan have a as a three- dynamicswas used to studythe behaviorof zeolitecan be seen schematically and diffusion butane and hexane in the zeolite silicalite. pronouncedeffect on the diffusionbecause step process:(i) the adsorption these vibrationsmay lower the diffusional of the reactants, (ii) the catalytic conver- June et al. observed that the diffusionof Becausethese barriers barriers. hardlyconof these alkanes is very slow and the rate of anddesorption sion, and (iii) the diffusion distribution,we fromthe zeolite.The overallactiv- diffusion decreases with increasing chain tribute to the equilibrium products length. Therefore,long simulationsare re- expect that the effect of lattice vibrations reactionis ity and selectivityof a particular is small. The on the equilibrium properties the resultof a delicatebalanceof these three quiredto obtain reliableresults. The Monte Carlo technique is not lim- details of the model are given in Table 1. and theoretiMuch experimental processes. cal effort is directed toward obtaining a of each of these steps Fig. 1. Schematic drawingof the growing of an alkane in a detailedunderstanding level (1). The high selectivity zeolite ina configurational-bias at a molecular MonteCarlomove. The black of zeolitesimplies that the behaviorof the circles represent the atoms of the zeolite, and the white It is moleculesis system-specific. adsorbed circles representthe atoms of the alkane. Seven atoms have thereforeessential to be able to study the been grown successfully, and an attempt is made to insert moleculesof inter- the eighth.The arrowsindicateseven trialpositionsforwhich behaviorof the adsorbed the energy u;is calculated. Outof these seven positionsone t k0' est underreactionconditions.
9. W. Theobald et al., J. Solid State Chem. 17, 431 etal., J. Electrochem.Soc. (1976); D. W. Murphy 128, 2053 (1981). J. Electrochem. T. Hirai, 10. T. Ohzuku,M. Kitagawa, Soc. 137, 769 (1990). 11. J. R. Dahn, T. van Buuren,U. von Sacken, U.S. Patent4,965,150 (1990). 12. For example, see T. Ohzuku,A. Ueda, T. Hirai, Chem. Express 7, 193 (1992); T. Nohma, T. H. Ikeda, J. Power Sources Saito, N. Furukawa, 26, 389 (1989); J. M. Tarascon and D. Guyomard, Electrochim.Acta 38, 1221 (1993).

P. C. Symons,W.Walsh,C. J. Warde, 13. E. Z. Ratner, technolo'Assessmentof battery G. L. Hendriksen, gies for electric vehicles," contract DE-AC07DC, Washington, 761D01 570 (U.S.Dept.of Energy, August1989). M.A. Fetcenko,J. Ross, Science 14. S. R. Ovshinsky, 260, 176 (1993). 15. The authors acknowledge funding fromthe NationalScience and EngineeringResearch Council of Canada's Operatinggrant program.

of Alkanesin Zeolites the Adsorption Simulating

0>

>

B. Smit,Shell Research B.V.Koninklijke/Shell-LaboraAmsterdam,Post Office Box 38000, 1030 BN torium, Amsterdam,Netherlands. of University of Chemistry, J. I.Siepmann,Department PA 19104-6323, USA. Philadelphia, Pennsylvania, *Towhom correspondence should be addressed. tPermanentaddress: Departmentof Chemistry,Universityof Minnesota,207 Pleasant Street SE, Minneapolis, MN55455, USA.

with is selected witha probability pi = exp(-ui/kBT)/wnew(f) andkB Wnew(f) = -iexp(-uj/kBT), where Tis thetemperature for the old configuration is Boltzmann'sconstant. Similarly, we calculate wOId(f) = ljexp(-uj/kBT). This is repeated until the entirechain of length m has been grown.Itcan be proven (5) that the bias of the growing is removed by the replacement of exp(-AU/kBT) by rIme1
Wnew(f)/I2e=1 wold(4)

withmolecular dynamicsshows that in the acceptance rule.Comparison

MonteCarlois two orders of magnitudemore efficientfor butane and up to 12 configurational-bias orders of maqnitudemore efficientfor dodecane. SCIENCE * VOL. 264 * 20 MAY 1994

1118