Challenges for Na-ion Negative Electrodes

Challenges for Na-ion Negative Electrodes

The MIT Faculty has made this article openly available. Please share how this access benefits you. Your story matters.

Challenges for Na-ion Negative Electrodes

V.L.Chevrier and G.Ceder*,z

Department of Materials Science and Engineering,Massachusetts Institute of Technology,Cambridge,Massachusetts

02139,USA

Na-ion batteries have been proposed as candidates for replacing Li-ion batteries.In this paper we examine the viability of Na-ion negative electrode materials based on Na alloys or hard carbons in terms of volumetric energy density.Due to the increased size of the Na atom compared to the Li atom,Na alloys would lead to negative electrode materials with roughly half the volumetric energy density of their Li analogs.Volumetric energy densities obtainable with sodiated hard carbons would also be signi?cantly less than those obtainable with lithiated graphite.These?ndings highlight the need of novel ideas for Na-ion negative electrodes. V C2011The Electrochemical Society.[DOI:10.1149/1.3607983]All rights reserved.

Manuscript submitted May10,2011;revised manuscript received June13,2011.Published July14,2011.

Li-ion batteries dominate the portable electronics market and are currently seen as the only viable option in all-electric vehicles.Li-ion batteries are even considered for use in load-leveling applications on power grids.Li-ion technology is used because its high speci?c capacity and high voltage combine to yield an unmatched volumetric energy density.

Sony?rst commercialized the Li-ion battery in the early1990’s. The original chemistries for the negative and positive electrodes were based on hard carbon and LiCoO2respectively,although graphite quickly replaced hard carbon.Over the last twenty years three other major chemistries have been developed for the positive electrode: LiFePO4,LiMn2O4,and transition metal substituted variants of LiCoO2.1For the negative electrode,alloy-based materials are pre-dicted to replace graphite,which remains the vastly dominant chemis-try in commercial cells.2–4Sony has offered a Sn-based negative electrode since2005,5and Panasonic has announced a Si-based nega-tive electrode for2013.6

When graphite is fully lithiated only1/6of a Li can be stored for every C atom.On the other hand in alloys up to approximately four Li atoms can be stored for every active metal(or semimetal).2–4The advantage of alloy-based negative electrodes therefore lies in signif-icantly increased volumetric and speci?c energy densities.

Na batteries were originally researched in tandem with Li batteries in the late1970’s and early1980’s.7–14As the successes of Li bat-teries outnumbered those of Na batteries,the focus of the scienti?c community shifted away from Na.Na-ion batteries have therefore not been nearly as extensively studied as Li-ion batteries.In the last few years,a variety of factors have caused a renewal of interest in Na-ion batteries.The availability of Li in terms of a natural resource has been put into question with some claiming an eminent shortage15if the automotive market embraces Li-ion batteries,while others claim either abundant,16or suf?cient reserves if large-scale recycling is implemented.17Another motivation for the study of Na-ion batteries is that even though the?eld of Li-ion batteries has become increas-ingly competitive with more research effort,future directions for increasing the energy density through new chemistries remain unclear. Indeed,many of the commercial Li-ion cells with the highest energy densities remain based on graphite and LiCoO2–derived cathodes.18 Our MIT group has used high-throughput computing to evaluate sev-eral tens of thousands of compounds.19,20While some interesting novel compounds have emerged21,22and are in the process of being developed,it is becoming clear that the options for much higher posi-tive electrode energy density are becoming increasingly limited.

Most of the renewed academic interest in Na-ion batteries has focused on cathodes as the possibility of new Na-ion applicable crys-tal structures proves enticing.23–25Recent computational studies of a wide variety of Na containing structures as positive electrodes for Na-ion batteries have shown them to operate on average approxi-mately0.4V lower than their Li-ion analogs.26This will present a challenge for Na-ion batteries as a lower positive electrode voltage leads to lower overall energy density.Assuming an excellent positive electrode is found for Na-ion batteries,a negative electrode will obvi-ously also be required for Na-ion batteries to be commercially viable.

Few studies of candidate negative electrodes exist in the current scienti?c literature.Graphite cannot be used as a negative electrode as Na atoms will not intercalate between the carbon sheets.14,27,28 Na metal appears to form dendrites much like Li metal but its safety is further put into question by its low melting point of97.7 C com-pared to180.5 C for Li.Hard carbons have been identi?ed as possi-ble negative electrode materials,where Na atoms adsorb onto the surfaces of nanoscopic pores throughout the hard carbon particles. Na alloys as negative electrode materials have yet to be extensively studied.If the evolution of Li-ion and Na-ion batteries follow a sim-ilar path,Na alloys are likely to become a?eld of growing interest for Na-ion batteries.

This paper will seek to answer the question of whether hard car-bons and Na alloys will lead to Na-ion batteries that are viable com-petitors to Li-ion batteries in terms of volumetric energy density. Much of the analysis uses insights developed for Li-ion batteries and is inspired in large part by Ref.29.

Theory

Volumetric energy density.—Several metrics are used to qualify electrode materials.The most common is the speci?c capacity,typi-cally with units of mAh/g,because it is the easiest to measure.How-ever it can be convincingly argued that volumetric energy density (Wh/cc)is the most relevant metric for most Li-ion battery applica-tions.If one considers Na-ion batteries as a candidate substitute, similar considerations should apply.

Firstly,the most widespread application of Li-ion batteries is portable electronics where the volume is the limiting factor.The weight of the battery could vary considerably without a signi?cant impact to the user experience,but the volume must typically be as small as possible.Even in electric vehicles the volume is arguably the limiting factor.In power grid applications cost may trump vol-ume considerations,however volumetric energy density is also an important driver for cost reduction as cells have a large?xed cost (electrolyte,separator,current collector,casing,etc...)and a higher volumetric energy density electrode implies that fewer cells have to be fabricated to achieve a total energy storage capability.

Secondly,it is the energy,not the charge,which is stored in the battery that will dictate the amount of work a battery can accom-plish.One therefore seeks to maximize the voltage window of the battery by maximizing and minimizing the positive and negative electrode potentials respectively.In order to compare negative elec-trodes one can compare them by using a hypothetical positive elec-trode of constant voltage.

One can therefore conclude that the volumetric energy density should be one of the?rst criteria considered when evaluating an

*Electrochemical Society Active Member. z E-mail:gceder@https://www.wendangku.net/doc/2e19032102.html, Journal of The Electrochemical Society,158(9)A1011-A1014(2011)

0013-4651/2011/158(9)/A1011/4/$28.00V C The Electrochemical Society

A1011

electrode and this paper will focus on the volumetric energy density (Wh/cc)obtainable with hard carbons and Na alloys.

The volumetric energy density (~U

)is de?ned as ~U ?

F v ex f Te

x f 0V etTex TàV eàTex Tàádx [1]where F is Faraday’s constant (26.802Ah/mol),V is the voltage of the positive (t)and negative (à)electrodes,x is the number of moles of Na (or Li)per mole of host alloy,and v is the density of the sodiated (or lithiated)active material in cc per mole of host alloy.

Binder limitations.—The large amount of Li that can be alloyed on a per atom basis in alloy negative electrode materials causes mas-sive changes in volume.4For example,the full lithiation of Sn and Si leads to volume expansions of 260and 280%respectively.Sev-eral studies have shown that this massive volume expansion can lead to poor cycle life.Capacity fade can be caused by pulverization of the active particles 30or degradation of the electrode coating.31The capacity fade of alloy based negative electrodes is very sensi-tive to the choice of binder.31,32A good binder must ideally main-tain adhesion of the electrode to the current collector,maintain ionic contact,and facilitate the formation of a stable interface with the electrolyte.

In general,the volume expansion of the negative electrode must be limited in order to yield long term stable cycling.Currently,the best way of limiting the volume expansion is by diluting the active material with inactive material,29leading to more stable cycling but also to lower speci?c and volumetric capacities.This is the approach taken in the commercialized Sn-based material where Co acts mainly as an inactive diluent.5

Universal expansion curve.—Experimental and theoretical stud-ies have shown that the volume occupied by Li atoms in Li-M alloy negative electrodes is essentially independent of M and of lithiation

level,and constant at 14.8A

?3(8.9ml/mol).Based on this fact,Obrovac et al .derived a “universal expansion curve”for the design

of alloy negative electrodes 29allowing the energy density (~U

)to be calculated based on the volume occupied by Li (k ),the average volt-age of the full cell (V avg )and ?nal volume expansion (n f )of the active/inactive negative electrode

~U ?F V avg n f f [2]where F is Faraday’s constant.This powerful equation allows one to

assess the merit of a negative electrode using readily available infor-mation.This approach can also be applied to the design of Na-based negative electrodes.All that is needed is the volume occupied by Na atoms and the average voltages of Na negative electrodes.Few ex-perimental studies of Na alloy negative electrodes exist in the litera-ture and signi?cant effort would be required to test several different Na alloys as negative electrodes.However,the ability to calculate accurate average lithiation voltages and lattice parameters using density functional theory has been established for years and can readily be applied to this problem.19,33

The voltages for Na alloys are therefore obtained computation-ally and a hypothetical positive electrode with a constant voltage of 3.75vs.Li or Na metal is used to calculate the full cell voltage.The hypothetical 3.75V positive electrode is used when considering both Li and Na alloys.Recent theoretical studies have shown that Na substitution into all the major Li-ion positive electrode structure classes leads to potentials that are on average 0.4V lower.26This would represent an additional drop in volumetric energy density at the full cell level,which is not considered in this paper.The possible shrinking of the Na-based positive electrode during desodiation is also not considered in this paper.Na positive electrodes shrink to a greater extent than their Li counterparts,which are often found to

vary very little in volume during delithiation.The total volume of a full cell is typically at its greatest in the charged state (lithiated/sodi-ated negative electrode)since the alkali atoms occupy greater vol-umes because they are not fully ionized.If one designs cell packag-ing to accommodate the volume of the full cell in its charged state,the volume of the desodiated or delithiated positive electrode should be considered.The difference in volumes between desodiated posi-tive electrodes and their delithiated counterparts are computation-ally found to be small.The possible greater shrinking of Na positive electrodes should therefore not affect the current conclusions.

Computational Methods

Density functional theory (DFT)calculations in the general gra-dient approximation (GGA)were performed.Projector augmented wave (PAW)(Ref.34)pseudopotentials included in the Vienna Ab-initio Simulation Package (VASP 5.2.2)were used with the Perdew-Burke-Ernzerhof (PBE)functional.35

Spin-polarized total energy calculations and structure relaxations were performed with VASP using a 500eV energy cut-off and appropriate k -point meshes to obtain a convergence of better than 10meV per formula unit.Structural relaxations were performed to a tolerance of 2?10à4eV/atom in the total energy.Voltage curves are calculated using standard methods.36,37

Results and Discussion

Figure 1shows the sodiation voltage curves obtained from the DFT total energies for Si,Ge,Sn,and Pb.Calculations were per-formed using known crystal structures obtained from the Inorganic Crystal Structure Database (ICSD).38Overall,the average sodiation voltages are on average 0.15V lower than calculated average lithia-tion voltages.A recent computational study of Na positive electro-des showed their voltages to be consistently lower by approximately 0.4V.26Experimental sodiation data in the literature were only found for Pb (Ref.12)and are included in Fig.1.Rather good agree-ment is found between the experimental and theoretical Pb sodiation voltages.The Na 5Pb 2phase was not included in the calculations as the Na sites in the crystal structure have not yet been experimentally de?ned.

The experimental volumes of the crystal structures used for the sodiation voltage curves were used to calculate the volume occupied by Na in Na-M alloys.Figure 2shows that structure volumes increase linearly with Na content con?rming Na occupies a constant

volume,which is calculated from the slope as being k Na ?30.3A

?3.Na atoms therefore occupy approximately twice the volume

of

Figure 1.(Color online)Na-M voltage curves calculated using DFT and known Na-M crystal structures.(a)M ?Si,(b)M ?Ge,(c)M ?Sn,and (d)M ?Pb.Experimental data for (d)are adapted from Ref.12.

Journal of The Electrochemical Society ,158(9)A1011-A1014(2011)

A1012

Li atoms (k Li ?14.8A

?3)when used in alloy negative electrodes.This is roughly in agreement with a comparison of the elemental

volumes of Li (21.6A

?3/atom)and Na (39.4A ?3/atom).Using the Na volume and the average sodiation voltages the vol-umetric energy density of Si,Ge,Sn,and Pb can now be plotted using Eq.2.Figure 3shows the volumetric energy density of Si,Ge,Sn,and Pb as a function of volume expansion.The volume expan-sion is in turn a linear function of sodium content.Figure 3shows that if a Si negative electrode is used with a binder able to sustain a 100%volume expansion,2.7Wh/cc can be achieved with Na.This should be contrasted with the 4.7Wh/cc obtainable with Li.29

Hard carbons have been reported as high capacity carbon-based negative electrodes for Na-ion batteries 39and their merit in terms of volumetric energy density should be evaluated.Hard carbons have very large surface areas and sodiation occurs through the adsorption of Na atoms onto the surfaces of nanoscopic pores throughout the hard carbon particles,which leads to low volume expansion.As a best-case scenario,we assume a zero volume expansion.The volu-metric energy density (1.7Wh/cc)can then be calculated from the experimental hard carbon density 39($1.5g/cc),capacity 27($340mAh/g)and average voltage (0.33V,obtained by digitizing and integrating the delithiation curve in Fig.6b of Ref.27).27,39,40The volumetric energy density obtained with hard carbons is included in

Figure 3for ease of comparison.As with Li,the carbon-based approach leads to a considerably lower volumetric energy density than the alloy-based approach.The volumetric energy density of hard carbons with Na (1.7Wh/cc)should be contrasted with the vol-umetric energy of graphite with Li (2.7Wh/cc).

Figure 4shows the universal expansion curves of Li and Na-based alloy negative electrodes.The volumetric energy densities of lithiated graphite and sodiated hard carbons for Na are also included for ease of comparison,although the volume expansion axis does not apply to them.These curves can be used to evaluate the volu-metric energy density obtained from an active/inactive composite where the active fraction is fully lithiated or sodiated.The reader is encouraged to read Ref.29for more details.

The large difference in volumetric energy density between Li and Na negative electrodes is a direct consequence of the greater volume occupied by the Na atom as can be seen from Eq.2.There-fore,if all parameters except k are kept identical,the volumetric energy density of a sodiated alloy negative electrode will be k Li /k Na ?49%of a lithiated one.

Figure 4also shows that even if a discovery allowed acceptable cycling of Na alloys at 300%volume expansion,these would still have a lower volumetric energy density than what is obtained with 50–100%volume expansion in Li alloys.An example of a Li alloy demonstrating good cycling is the Sn 30Co 30C 40attrited material of Ferguson et al .,which has a capacity of roughly 450mAh/g for 100cycles 41and a density of 6.7g/ml.42Assuming a constant Li volume of 8.9ml/mol,this represents a 100%volume expansion of the alloy.Acceptable cycling with 100%volume expansions in Li alloys is therefore already achievable with current technologies and it is unlikely that solutions to volume expansion challenges would be speci?c to Na-alloys and not applicable to Li-alloys.

In the current market,the cost of Li represents less than 3%of the production cost of a full commercial Li-ion battery.43Even if Na was considerably less expensive because of its widespread availabil-ity,it is worthwhile to question whether the savings in cost would be lost when taking into account the costs associated with having a negative electrode with a signi?cantly lower volumetric energy density.

Conclusion and Outlook

There is growing interest in Na-ion batteries as a candidate replacement for Li-ion batteries.A viable Na-ion battery will require a negative electrode yielding suf?cient volumetric energy density.In this paper we have shown that Na-ion negative electrodes based

on

Figure 2.(Color online)The volume change (A 3/M)compared to elemental M for known crystalline Na-M alloys (M ?Si,Ge,Sn,Pb)increases linearly as a function of Na content.A linear ?t shows each Na atom occupies 30.3A 3on average,independent of M,or Na

content.

Figure 3.(Color online)Volumetric energy density and volume expansion obtained from the corresponding sodiation voltage curves of Fig 1.and Eq.1.Ge and Pb curves overlap

signi?cantly.

Figure 4.(Color online)Universal expansion curves for Li and Na alloys.The volumetric energy densities of lithiated graphite and sodiated hard car-bon are included for ease of comparison but are not a function of the volume expansion axis.

Journal of The Electrochemical Society ,158(9)A1011-A1014(2011)A1013

current strategies and technologies are unlikely to be able to compete with their Li-ion analogs in terms of volumetric energy density. Hence,focus should be on novel ideas for the negative electrode side, or on positive electrode materials which have substantially higher energy densities than the current Li-ion positive electrodes in order to overcome the limitations on the negative electrode side.

These?ndings highlight the need for new strategies and approaches for Na-based energy storage technologies.Indeed,an attempt to mimic current Li-ion technologies is unlikely to lead to a viable competitor to Li-ion batteries.

Acknowledgments

This work was supported by the US Department of Energy under Contract No.DE-FG02-96ER45571and the BATT program under Contract No.DE-AC02-05CH11231.

References

1. B.L.Ellis,K.T.Lee,and L.F.Nazar,Chem.Mater.,22,691(2010).

2.J.O.Besenhard,J.Yang,and M.Winter,J.Power Sources,68,87(1997).

3. A.D.W.Todd,P.P.Ferguson,M.D.Fleischauer,and J.R.Dahn,Int.J.Energ.

Res.,34,535(2010).

4.W.-J.Zhang,J.Power Sources,196,13(2011).

5.H.Inoue,Abstract228,The International Meeting on Lithium Batteries,Biarritz,

France,June18–23,2006.

6.http://panasonic.co.jp/corp/news/of?cial.data/data.dir/jn091225-1/jn091225-1.html,

last accessed June28,2011.

7.M.Whittingham,Prog.Solid State Chem.,12,41(1978).

8. A.S.Nagelberg and W.L.Worrell,J.Solid State Chem.,29,345(1979).

9. C.Delmas,J.Braconnier,C.Fouassier,and P.Hagenmuller,Solid State Ionics,

3–4,165(1981).

10.J.Molenda,C.Delmas,and P.Hagenmuller,Solid State Ionics,9–10,431(1983).

11.J.-M.Tarascon and G.W.Hill,Solid State Ionics,22,85(1986).

12.T.R.Jow,L.W.Shacklette,M.Max?eld,and D.Vernick,J.Electrochem.Soc.,

134,1730(1987).

13.L.W.Shacklette,T.R.Jow,and L.Townsend,J.Electrochem.Soc.,135,2669(1988).

14.P.Ge and M.Fouletier,Solid State Ionics,28,1172(1988).

15.W.Tahil,“The Trouble with Lithium2,”William Tahil,Meridian International

Research,Paris,France,https://www.wendangku.net/doc/2e19032102.html,/Projects/Lithium_ Microscope.pdf,May29,2008,last accessed April23,2011.

16.R.K.Evans,“An Abundance of Lithium,Part Two,”July2008,www.worldlithium.

com,last accessed June28,2011.17.L.Gaines and P.Nelson,Lithium-Ion Batteries:Examining Material Demand And

Recycling Issues,The Minerals,Metals and Materials Society2010Annual Meet-ing,Seattle,WA.

18.J.R.Dahn and G.M.Ehrlich,in Handbook of Batteries,4th ed.,T.B.Reddy,Edi-

tor,p.26.47,McGraw-Hill,New York(2010).

19.G.Ceder,G.Hautier,A.Jain,and S.P.Ong,MRS Bull.,36,185(2011).

20. A.Jain,G.Hautier,C.J.Moore,S.P.Ong,C.C.Fischer,T.Mueller,K.A.Pers-

son,G.Ceder,Comput.Mater.Sci.(2011).

21.G.Hautier,A.Jain,H.Chen,C.J.Moore,S.P.Ong,and G.Ceder,J.Mater.

Chem.,accepted.

22.G.Hautier,A.Jain,S.P.Ong,B.Kang,C.J.Moore,R.Doe,and G.Ceder,Chem.

Mater.,accepted.

23.J.Barker,M.Saidi,and J.Swoyer,Electrochem.Solid-State Lett.,6,A1(2003).

24.P.Barpanda,J.-N.Chotard,N.Recham,C.Delacourt,M.Ati,L.Dupont,M.

Armand,and J.-M.Tarascon,Inorg.Chem.,49,7401(2010).

25.J.F.Whitacre,A.Tevar,and S.Sharma,https://www.wendangku.net/doc/2e19032102.html,mun.,12,463(2010).

26.S.P.Ong,V.L.Chevrier,G.F.Hautier,A.Jain,C.Moore,S.Kim,X.Ma,and G.

Ceder,Energy Environ.Sci.,accepted.

27. D.A.Stevens and J.R.Dahn,J.Electrochem.Soc.,148,A803(2001).

28.R.C.Asher,J.Inorg.Nucl.Chem.,10,238(1959).

29.M.N.Obrovac,L.Christensen,D.B.Le,and J.R.Dahn,J.Electrochem.Soc.,

154,A849(2007).

30.L.Y.Beaulieu,T.D.Hatchard,A.Bonakdarpour,M.D.Fleischauer,and J.R.

Dahn,J.Electrochem.Soc.,150,A1457(2003).

31.Z.Chen,V.Chevrier,L.Christensen,and J.R.Dahn,Electrochem.Solid-State

Lett.,7,A310(2004).

32.J.Li,D.-B.Le,P.P.Ferguson,and J.R.Dahn,Electrochim.Acta,55,2991

(2010).

33.V.L.Chevrier,S.P.Ong,R.Armiento,M.K.Y.Chan,and G.Ceder,Phys.Rev.

B,82,075122(2010).

34.P.E.Blo¨chl,Phys.Rev.B,50,17953(1994).

35.J.P.Perdew,K.Burke,and M.Ernzerhof,Phys.Rev.Lett.,77,3865(1996).

36.M.K.Aydinol,A.F.Kohan,G.Ceder,K.Cho,and J.Joannopoulos,Phys.Rev.B,

56,1354(1997).

37.I.A.Courtney,J.S.Tse,O.Mao,J.Hafner,and J.R.Dahn,Phys.Rev.B,58,

15583(1998).

38.ICSD,Inorganic Crystal Structure Database;Fachinformationszentrum Karlsruhe:

Karlsruhe,Germany,2006.

39. D.A.Stevens and J.R.Dahn,J.Electrochem.Soc.,147,1271(2000).

40. D.A.Stevens and J.R.Dahn,J.Electrochem.Soc.,147,4428(2000).

41.P.P.Ferguson,A.D.W.Todd,and J.R.Dahn,https://www.wendangku.net/doc/2e19032102.html,mun.,10,25

(2008).

42.P.P.Ferguson and J.R.Dahn,Can.J.Phys.,88,131(2010).

43.T.Hamilton,Technology Review,Cambridge(2009),www.technologyreview.

com/energy/23215,last accessed June28,2011.

Journal of The Electrochemical Society,158(9)A1011-A1014(2011) A1014