Electrochimica Acta 52 (2006) 973–978
Nano silicon for lithium-ion batteries Michael Holzapfel a,∗ , Hilmi Buqa a,1 , Laurence J. Hardwick a , Matthias Hahn a , Andreas W¨ursig a,2 , Werner Scheifele a , Petr Nov´ak a , R¨udiger K¨otz a , Claudia Veit b , Frank-Martin Petrat b a
Paul Scherrer Institut, Electrochemistry Laboratory, CH-5232 Villigen PSI, Switzerland b Degussa AG, Paul-Baumann-Str. 1, Building 1420/18, D-45764 Marl, Germany
Received 6 February 2006; received in revised form 3 May 2006; accepted 28 June 2006 Available online 8 August 2006
Abstract New results for two types of nano-size silicon, prepared via thermal vapour deposition either with or without a graphite substrate are presented. Their superior reversible charge capacity and cycle life as negative electrode material for lithium-ion batteries have already been shown in previous work. Here the lithiation reaction of the materials is investigated more closely via different electrochemical in situ techniques: Raman spectroscopy, dilatometry and differential electrochemical mass spectrometry (DEMS). The Si/graphite compound material shows relatively high kinetics upon discharge. The moderate relative volume change and low gas evolution of the nano silicon based electrode, both being important points for a possible future use in real batteries, are discussed with respect to a standard graphite electrode. © 2006 Elsevier Ltd. All rights reserved. Keywords: Nano silicon; Graphite; Lithium-ion batteries; Composite electrodes; Rechargeable
1. Introduction Lithium-ion batteries are now the cell-of-choice to power portable electronic applications; more than a billion cells were sold in 2004. Due to their high energy density (more than twice the one of NiMH batteries) and very high efficiency (up to 95% overall), there is more and more discussion for the utilisation of lithium-ion batteries in electric vehicles (EVs) and hybridelectric vehicles (HEVs). For the negative electrode, in general, graphitic carbon is used because it shows relative safety upon cycling when compared to lithium metal. Due to graphite’s relatively low electrochemical charge capacity (theoretical value: 372 mAh g−1 ), however, the search for alternative negative electrode materials has been intensified, above all in the field of lithium–metal alloys. Wellknown examples are aluminium [1], tin [2,3], antimony [3], etc. ∗
Corresponding author. Tel.: +41 56 310 2116; fax: +41 56 310 4415. E-mail address:
[email protected] (M. Holzapfel). 1 Present address: High power Lithium S.A., PSE B, EPFL, CH-1015 Lausanne, Switzerland. 2 Present address: Fraunhofer Institute for Silicon Technology (ISIT), Fraunhoferstr.1, D-25524 Itzehoe, Germany. 0013-4686/$ – see front matter © 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2006.06.034
for binary alloys or copper–tin [4], copper–antimony [5], etc. for ternary alloys. In these systems, the metal reversibly forms alloys with lithium, which have very high capacities. A general disadvantage of alloy electrodes, however, is the huge volume change which occurs upon the insertion/deinsertion of the lithium. It can attain values of more than 200–300% [6] and leads to mechanical fatigue upon prolonged cycling. Much research has been conducted on silicon, as it reversibly forms, alike tin, electrochemically active binary alloys with lithium [7–9]. They can show a very high lithium insertion capacity of approx. 4200 mAh g−1 (for a theoretical composition of Li4.2 Si). This very high lithium content is accompanied by a huge volume change (of more than 300%), which leads to strong mechanical stress on the crystallites and, thus, to breaking and amorphisation of the particles and loss of the electrical contact [10–12]. A rapid loss of the reversible capacity upon prolonged cycling (fading) generally results. Besides the deposition of the silicon as a sub-micrometer thin film on a strongly adhering support by chemical vapour deposition [12–16] there are several other ways that have been explored. Intensive ball-milling to decrease the particle size and enhance the electric contact [17,18], carbon coating starting from the gas phase [19], and pyrolysis of an intimate mixture
974
M. Holzapfel et al. / Electrochimica Acta 52 (2006) 973–978
of precursors [20] have been adopted. Indeed, a reduction of the irreversible losses and, hence, an increase in the reversible charge capacity and cycle life could have been obtained, but in any case the cycle life was below 100 cycles. Other techniques for an eventual stabilisation of silicon negative electrodes could be the use of an additive which counterbalances irreversible losses during the first cycles (e.g. Li2.6 CoN, cf. [21]), or the use of an inert, buffering, metal matrix for the electrochemically active silicon [22,23]. A reduction of the particle size into the nanometre range can reduce the mechanical stress. An early study [19], showed the favourable behaviour of nano silicon/carbon composites. Recent literature [15,24,25] shows that with nano-scale materials capacities up to 1700 mAh g−1 , together with reduced fading can be reached. However, such materials are not yet comparable to common graphite electrodes as they suffer from low cycle life and high fading. Wang et al. [26] recently presented composite electrodes based on nano silicon inclusions in carbon aerogel. These electrodes show a stable charge capacity of 1450 mAh g−1 . The same group prepared also a promising high-capacity composite electrode by ball-milling, but these electrodes still suffer from a relatively high fading [27]. In this study we present nano-scale silicon materials, prepared by thermal vapour deposition, showing excellent electrochemical properties, above all high reversible specific charge capacity and low capacity fading during cycling. The relative volume changes of the composite electrode upon lithium uptake and release are shown not to be higher and the gas evolution to being less than for standard graphite. Two types of materials are presented: the first one is a composite electrode based on an intimate mixture of the nano silicon with a small particle graphite (TIMREX KS6; 20 and 80 wt.% Si). For the 80 wt.% silicon electrode also 2% of TIMREX Super P carbon black have been added. The second material is based on direct deposition of the nano silicon on the surface of the fine particle graphite. In the following, this material is called “compound material”. Preliminary results concerning both materials have already been communicated [28,29]. Both materials show impressive results with respect to their high reversible charge capacity, low irreversible capacity upon prolonged cycling and both, long cycle life and low capacity fading. The first cycles are particularly characterised by in situ Raman microscopy, slow scan cyclic voltammetry, dilatometry measurements and measurements on gas evolution upon cycling. 2. Experimental The nano silicon material was produced by a pyrolysis process of mono silane (SiH4 ), either without substrate to form the nano silicon, or with TIMREX KS6 (TIMCAL SA, Bodio, Switzerland) fine particle graphite as substrate, to form the “compound” material. The composite electrodes were prepared as follows: the silicon material (20 wt.%) is mixed with 70 wt.% of TIMREX® KS6 and 10% of SOLEF® PVdF 1015 binder (Solvay SA, Belgium) in a N-methylpyrrolidone solution, mixed thoroughly and cast on a pre-treated (with a polymer based primer from Contitech,
Nordhausen, Germany) copper foil which serves as current collector. The primer enhances the adherence of the active material on the copper. The typical mass load is of 2–3 mg/cm2 . The “compound” electrode is made in the same manner by mixing the active material with 10% SOLEF® PVdF 1015 binder. Lithium metal is used as counter electrode in both cases. The electrolyte used is battery grade ethylene carbonate and dimethyl carbonate (1:1), with 1 M LiPF6 to which 2% of vinylene carbonate were added and was obtained from Ferro Corp. (USA). The electrochemical measurements were conducted in combined galvanostatic-potentiostatic protocol. First, classical galvanostatic (constant current) cycling with a specific current of 74 mA g−1 (10 mA g−1 in the first cycle) was performed until a lower voltage limit of 5 mV versus Li/Li+ and an upper voltage limit of 1.0 V versus Li/Li+ , respectively, for the charge and discharge. At the end of each charge and discharge step a potentiostatic step followed with a reduction of the current, at the fixed upper or lower potential limit, respectively, down to a value of 5 mA g−1 , to complete the charge/discharge (the charge capacity which passes at this potentiostatic step generally represents less than 5% of the total specific charge capacity). The cells were cycled galvanostatically by means of a computer-controlled cell capture system CCCC (Astrol Electronics AG, Oberrohrdorf, Switzerland). A confocal Raman microscope (Labram series, Jobin Yvon SA, ex DILOR SA), using a He–Ne laser at 632.8 nm with ca. 1 mW power as the excitation source, was used to acquire the in situ Raman spectra. Raman band positions were calibrated against the spectrum of a neon lamp (Penray, Oriel) with a resolution of 4 cm−1 with an 80× objective. Each spectrum required 180 s measurement-time, and because of the acceptable signal to noise ratio, with only one accumulation with a confocal resolution of 2–3 m3 . For the thickness variation measurements we used a purpose made dilatometer [30]. This consists of a cell stack, the active electrode and a LiCoO2 counter electrode separated by a fixed glass diaphragm. This assembly is soaked with electrolyte. A lithium wire serves as reference. The working electrode is free to move against a constant load (20 N) applied by means of a spring. The lower (and much bigger) electrode is fixed and serves as the counter electrode. The overall height change of the cell is monitored by an inductive displacement transducer mounted on top of the plunger which contacts the working electrode. As the glass diaphragm cannot move, the measured expansion can be solely attributed to the working electrode. The gas evolution experiments have been performed in an all titanium electrochemical cell with in situ head space gas analysis under flowing argon using a quadrupole mass spectrometer (cf. [31]). 3. Results and discussion The composite electrode is a homogeneous dispersion of the silicon particles within the graphite matrix. The silicon particles form long aggregates, with diameters of some 100–200 nm, intimately connected to the larger graphite particles. For the “compound” material the silicon forms small (20 nm) spher-
M. Holzapfel et al. / Electrochimica Acta 52 (2006) 973–978
975
ically shaped particles on the surface of the graphite. Both materials have been described in more detail in previous communications [28,29]. The materials show impressive specific charge (up to 1000 mAh g−1 for the composite material) and cycling stability upon electrochemical cycling [28,29]. The good behaviour of both types of electrodes has been attributed to several important points. First, a much smaller mechanical degradation upon the intercalation/deintercalation process which is probably due to the very small particle size and the favourable distribution within and good electrical contact to the supporting graphite/carbon black matrix (cf. also Liu et al. [32]). Second, to a favourable electrode manufacture, i.e. the larger amount of graphite accommodating the volume change of the silicon. Third, in the case of the compound electrode, the fact that the silicon particles are chemically bound on the graphite surface. This prevents contact loss during the huge volume change upon the lithiation–delithiation process via the vapour deposition process. In addition one has to note that the homogeneous distribution of the nano silicon particles on the graphite reduces the influence upon the neighbouring Si particles and hence the mechanical stress upon cycling. Fig. 1 shows slow scan cyclic voltammograms of the first lithiation/delithiation of the 20% composite and the “compound” electrode compared to a KS6 graphite electrode, respectively, below 0.6 V versus Li/Li+ , where the reversible lithium insertion takes place. For the graphite electrode, the lithium intercalation/deintercalation potentials are practically unchanged for the first and the second cycle. However, the silicon containing electrodes behave differently. Both show a peak
shift to lower potentials upon first lithiation, in relation to both, their second cycles and to graphite. This overpotential can be attributed on the one hand to the semiconducting behaviour of silicon which is a poor electronic conductor in its pristine state and should become conductive after partial filling of the conduction band with electrons during lithium insertion. On the other hand it can be attributed in part also to the breaking of the crystal structure upon the first lithiation. The kinetics of the electrochemical reaction are thus greatly enhanced (and the overpotentials for charge transfer are reduced) once the Si particles have been lithiated. This causes that the first and second delithiation do not show major differences but that the potentials of the peaks in the second lithiation are shifted to the right, compared to the first cycles, for the silicon containing materials. The typical features of graphite can be acknowledged in all three materials. The signatures coming from the silicon are mostly visible, for the lithiation reaction, as unspecified features at voltages above 200 mV versus Li/Li+ and increasing the intensity of the graphite signal. Upon delithiation, both an increasing of the intensity of the graphite pattern and a distinct peak between 400 and 500 mV versus Li/Li+ , reminiscent of the delithiation of crystalline lithiated silicon phases formed [9], can be acknowledged. The first lithiation reaction was followed using in situ Raman microscopy (Fig. 2) at several points on the surface, two of which are shown. Both points (A) and (B) (with different intensity) display a prominent silicon signal with the most notable Raman feature seen at ca. 518 cm−1 , with a full width at half maximum (FWHM) of 6 cm−1 . This can be assigned to the scattering of the first order optical phonon mode (TO). Two peaks at ca. 300 and 950 cm−1 can also be observed and these are assigned to the scattering of the second-order transverse acoustic phonon mode (2TA) and the second-order optical phonon mode (2TO), respectively [33–35]. The main bands of graphite
Fig. 1. Slow scan cyclic voltammetry (5 V/s) on both silicon-based electrodes, compared to graphite. The specific current is given in mA g−1 of active material.
Fig. 2. In situ Raman spectroscopy of the lithium intercalation into a 20% nano silicon composite electrode.
976
M. Holzapfel et al. / Electrochimica Acta 52 (2006) 973–978
are also observed, with the D band seen at 1330 cm−1 , the G band at 1585 cm−1 ([36]). During the first lithium insertion into this material from the open circuit potential (ca. 3 V versus Li/Li+ ) to 170 mV versus Li/Li+ the silicon lines are seen to rapidly diminish and then disappear completely at most of the points of the surface (represented by point A). This is because of the inserted lithium breaking down the sp3 symmetry of the structure. The diamond-like structure of silicon becomes amorphous through lithium insertion. This observation is in agreement with previous Raman measurements upon nano silicon, where a decrease of the 520 cm−1 band was also detected [37]. The decrease in intensity of the TO mode of silicon, for all points, can be observed. The intercalation spectral characteristics of graphite are also seen, especially at 170 mV versus Li/Li+ where the stages 3 and 4 GIC G-Band doublet appears [38–40] (represented by point B). The return to a single band seen at 1590 cm−1 below 155 mV indicates a stage 2 GIC and is detected at lower potentials (not shown). Some additional points measured still displayed a typical open circuit potential spectrum for the silicon. These measurement points could have been isolated particles. For the first delithiation no reappearance of a silicon line, at 520 cm−1 , was detected for any of the points measured during the cycling of potential from 5 mV to 1.5 V versus Li/Li+ . With the early decrease of the silicon line during the first lithium insertion it is unlikely a sufficient amount of lithium leaves the silicon for the initial crystalline structure to return for the TO mode of silicon to be detectable. The reappearance of graphite doublet G-band returning to singlet G band (ca. 1580 cm−1 ) is observed. This indicates that lithium is deintercalating from the graphite present in the electrode (not shown). Fig. 3 shows the electrochemical cycling behaviour of the “compound” material at different specific current rates. The charge reaction has always been done at C/7.4 rate in order to have comparable conditions for each cycle (50 mA g−1 ; the C-rate is given here with respect to the theoretical specific capacity of graphite, i.e. 372 mAh g−1 ). The galvanostatic discharge rate was varied between rates of C/7.4 and 10C (50 mA g−1 to 3.72 A g−1 ). This high current regime was followed by a poten-
tiostatic regime down to a specific current of 10 mA g−1 , in order to compare the galvanostatic (high regime) capacity to the total electrochemical discharge capacity. It can be seen that, even if the fading is somewhat higher than for normal cycling, the total capacity is not affected much by high rates. The galvanostatic discharge capacity of the “compound” material decreases steadily with the discharge rate, but the value always remains higher than the total capacity of graphite and reaches this value only for 10C (3.72 A g−1 ). The high rate capability of different types of graphite already shown in a recent study [41] is hence affected only little for the silicon coated “compound” material. As discussed in the introduction, the main problem of silicon and comparable materials (tin, antimony, aluminium) is the important volume change upon the electrochemical reaction with lithium. However, as mentioned above, if the material is present as sufficiently small particles, the mechanical stress is reduced. Secondly, if the electrical contact is well maintained by intimate mixing in an adapted conductive matrix, the loss of contact leading to capacity fading can be minimised. Besides the problem of the volume change of the particles themselves, also the volume change of the entire electrode is an important question in a real lithium-ion battery. Fig. 4 shows the evolution of the relative change of the electrode thickness of a 20% nano silicon composite electrode. The specific charge capacities are (in mAh g−1 ) for this experiment: first charge 940, first discharge 780, second charge 750, second discharge 720. The lower capacity, when compared to the results presented in Ref. [28], comes from the fact that the dilatometry cell used is not optimally designed for proper electrochemistry. A practically linear thickness change can be observed along with galvanostatic cycling, i.e. that a given amount of lithium inserted or deinserted accounts for the same volume change, irrespective of whether silicon or graphite is the host material. Only at the beginning of the delithiation reaction a lower slope can be acknowledged, which could be related to the stages I–II transformation of the graphite. A comparison to dilatometry measurements performed on graphite [42] reveals that, indeed, the relative thickness change per specific capacity is comparable for graphite and silicon/graphite
Fig. 3. Rate capability behaviour of the nano silicon “compound” electrode. The C rate is given with respect to graphite, i.e. C/1 corresponds to 372 mA g−1 .
Fig. 4. Thickness variation of a 20% nano silicon composite electrode during electrochemical cycling.
M. Holzapfel et al. / Electrochimica Acta 52 (2006) 973–978
977
pound” material shows a relatively high rate capability for the discharge reaction. The overall relative volume change of a nano silicon based electrode has been shown to be about the same as for classical graphite electrodes. Finally, the gas evolution is reduced when compared to graphite electrodes. Acknowledgement The authors express their gratitude to Dr. Frank Krumeich (ETH Z¨urich, Switzerland) for his valuable help and fruitful discussions. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] Fig. 5. Gas evolution for a nano silicon graphite electrode compared to a graphite electrode upon the first electrochemical cycle.
electrodes. This means that the use of a material with inherent high volume change does not necessarily lead to a higher volume change in an electrode made out of it. The use of silicon-based material, with increased specific capacity, in classic lithium-ion batteries would therefore not need a new battery design. Fig. 5 shows the comparative evolution of hydrogen and ethylene for both nano silicon and the graphite electrodes in the standard electrolyte ethylene carbonate: dimethyl carbonate (1:1), 1 M LiPF6 + 2% vinylene carbonate (VC), for approximately the same total charge capacity of the electrode. The amounts of active material were ca. 40 mg of graphite and ca. 15 mg of the nano silicon/graphite composite, which accounts for approximately the same amount of electrochemically stored lithium. It can be seen that the gas evolution is 4–5 times less for the composite electrode which means that the nano silicon does not account for the reduction of the electrolyte under gas evolution. The pressure build-up within a lithium-ion battery could thus be reduced, which increases the safety of the battery. A comparable result has already been described by Wagner et al. for tin-based electrodes [43]. 4. Conclusions
[11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29]
Thermal vapour deposition from silicon-containing precursors has lead to free nano-scale silicon and nano-scale silicon particles deposited on a fine particle graphite (the “compound” material). Upon the first lithiation the rapid formation of lithiated phases causes the early disappearance of the Raman signal and increases the kinetics for the lithiation reaction. The “com-
[30] [31] [32]
M.J. Lindsay, G.X. Wang, H.K. Liu, J. Power Sources 119 (2003) 84. M. Winter, J.O. Besenhard, Electrochim. Acta 45 (1999) 31. J.L. Tirado, Mater. Sci. Eng. R-Rep. 40 (2003) 103. K.D. Kepler, J.T. Vaughey, M.M. Thackeray, Electrochem. Solid-State Lett. 2 (1999) 307. J. Yang, M. Wachtler, M. Winter, J.O. Besenhard, Electrochem. Solid-State Lett. 2 (1999) 161. J.O. Besenhard, J. Yang, M. Winter, J. Power Sources 68 (1997) 87. W.J. Weydanz, M. Wohlfahrt-Mehrens, R.A. Huggins, J. Power Sources 82 (1999) 237. R.N. Seefurth, R.A. Sharma, J. Electrochem. Soc. 124 (1977) 1207. T.D. Hatchard, J.R. Dahn, J. Electrochem. Soc. 151 (2004) A838. M. Winter, J.O. Besenhard, M.E. Spahr, P. Nov´ak, Adv. Mater. 10 (1998) 725. J. Yang, M. Winter, J.O. Besenhard, Solid State Ionics 90 (1996) 281. S. Bourderau, T. Brousse, D.M. Schleich, J. Power Sources 82 (1999) 233. S. Ohara, J. Suzuki, K. Sekine, T. Takamura, J. Power Sources 119 (2003) 591. H.J. Jung, M. Park, S.H. Han, H. Lim, S.K. Joo, Solid State Commun. 125 (2003) 387. H.J. Jung, M. Park, Y.G. Yoon, G.B. Kim, S.K. Joo, J. Power Sources 115 (2003) 346. J. Graetz, C.C. Ahn, R. Yazami, B. Fultz, Electrochem. Solid-State Lett. 6 (2003) A194. N. Dimov, S. Kugino, M. Yoshio, Electrochim. Acta 48 (2003) 1579. J.J. Niu, J.Y. Lee, Electrochem. Solid-State Lett. 5 (2002) A107. A.M. Wilson, J.R. Dahn, J. Electrochem. Soc. 142 (1995) 326. D. Larcher, C. Mudalige, A.E. George, V. Porter, M. Gharghouri, J.R. Dahn, Solid State Ionics 122 (1999) 71. Y. Liu, K. Hanai, K. Horikawa, N. Imanishi, A. Hirano, Y. Takeda, Mater. Chem. Phys. 89 (2005) 80. M.D. Fleischauer, J.M. Topple, J.R. Dahn, Electrochem. Solid-State Lett. 8 (2005) A137. H.Y. Lee, Y.L. Kim, M.K. Hong, S.M. Lee, J. Power Sources 141 (2005) 159. H. Li, X.J. Huang, L.Q. Chen, Z.G. Wu, Y. Liang, Electrochem. Solid-State Lett. 2 (1999) 547. B. Gao, S. Sinha, L. Fleming, O. Zhou, Adv. Mater. 13 (2001) 816. G.X. Wang, J.H. Ahn, J. Yao, S. Bewlay, H.K. Liu, Electrochem. Commun. 6 (2004) 689. G.X. Wang, J. Yao, H.K. Liu, Electrochem. Solid-State Lett. 7 (2004) A250. M. Holzapfel, H. Buqa, W. Scheifele, P. Novak, F.-M. Petrat, Chem. Commun. (2005) 1566. M. Holzapfel, H. Buqa, F. Krumeich, F.-M. Petrat, C. Veit, Electrochem. Solid-State Lett. 8 (2005) A515. M. Hahn, O. Barbieri, F.P. Campana, R. K¨otz, R. Gallay, Appl. Phys. A 82 (2006) 633. P. Novak, D. Goers, L. Hardwick, M. Holzapfel, W. Scheifele, J. Ufheil, A. Wuersig, J. Power Sources 146 (2005) 15. W.R. Liu, Z.Z. Guo, W.S. Young, D.T. Shieh, H.C. Wu, M.H. Yang, N.L. Wu, J. Power Sources 140 (2005) 139.
978
M. Holzapfel et al. / Electrochimica Acta 52 (2006) 973–978
[33] H. Richter, Z.P. Wang, L. Ley, Solid State Commun. 39 (1981) 625. [34] B.B. Li, D.P. Yu, S.L. Zhang, Phys. Rev. B 59 (1999) 1645. [35] S.L. Zhang, X. Wang, K.J. Ho, J.J. Li, P. Diao, S.M. Cai, J. Appl. Phys. 76 (1994) 3016. [36] J.L.K.F. Tuinstra, J. Chem. Phys. 53 (1970) 1126. [37] H. Li, X.J. Huang, L.Q. Chen, G.W. Zhou, Z. Zhang, D.P. Yu, Y.J. Mo, N. Pei, Solid State Ionics 135 (2000) 181. [38] M. Inaba, H. Yoshida, Z. Ogumi, T. Abe, Y. Mizutani, M. Asano, J. Electrochem. Soc. 142 (1995) 20.
[39] W.W. Huang, P. Frech, J. Electrochem. Soc. 145 (1998) 765. [40] L. Hardwick, H. Buqa, P. Novak, submitted for publication. [41] H. Buqa, D. Goers, M. Holzapfel, M.E. Spahr, P. Novak, J. Electrochem. Soc. 152 (2005) A474. [42] M. Hahn, O. Barbieri, F.P. Campana, R. Gallay, R. K¨otz, Proceedings of the 14th International Seminar on Double Layer Capacitors and Hybrid Energy Storage Devices, vols. 6–8, Deerfield Beach, USA, December, 2004, p. 40. [43] M.R. Wagner, P.R. Raimann, A. Trifonova, K.C. Moeller, J.O. Besenhard, M. Winter, Electrochem. Solid-State Lett. 7 (2004) A201.
