BACKGROUND Various metals, metalloids, and alloys have been investigated for use as active anode compositions for lithium-ion batteries. These materials are attractive because they potentially have higher gravimetric and volumetric capacities than carbon and graphite, both of which are currently used as anodes in lithium-ion batteries. One problem with these materials, however, is that they experience large volume expansion during battery operation as a result of lithiation and delithiation. This volume expansion, in turn, causes the materials to deteriorate, thus limiting cycle life. In addition, the methods used to prepare these materials do not always lend themselves readily to large-scale manufacturing.

SUMMARY The invention provides electrode compositions suitable for use in lithium-ion batteries in which the electrode compositions exhibit high capacities and good cycle life.

In addition, the electrode compositions, and batteries incorporating them, are readily manufactured.

To achieve these objectives, the invention features, in a first aspect, an electrode composition that includes particles having an average particle size ranging from 1 um to 50 urn, in which the particles comprise an electrochemically active phase and an electrochemically inactive phase that share at least one common phase boundary. The electrochemically active phase comprises elemental silicon and the electrochemically inactive phase comprises at least two metal elements in the form of an intermetallic compound, a solid solution, or combination thereof. In some embodiments, the electrochemically inactive phase further comprises silicon. Each of the phases is free of crystallites that are greater than 1000 angstroms prior to cycling. Moreover, the

electrochemically active phase is amorphous after the electrode has been cycled through one full charge-discharge cycle in a lithium-ion battery. Preferably, the electrochemically active phase remains amorphous during additional charge-discharge cycles when the voltage is greater than 70mV vs. Li/Li+, more preferably greater than 50mV vs. Li/Li+.

An"electrochemically active"material is a material that reacts with lithium under conditions typically encountered during charging and discharging in a lithium-ion battery.

An"electrochemically inactive"material is a material that does not react with lithium under those conditions.

An"amorphous"material is a material that lacks the long range atomic order characteristic of crystalline material, as observed by x-ray diffraction or transmission electron microscopy.

The electrode composition may be prepared according to a process that includes (a) melting together elemental silicon and two or more additional metal elements in an inert atmosphere to form an ingot; (b) melting the ingot in an inert atmosphere to form a molten stream; (c) rapidly quenching the molten stream on the surface of a rotating wheel to form a ribbon; and (d) pulverizing the ribbon to form particles having an average particle size ranging from 1 um to 50 um.

The details of one or more embodiments of the invention are set forth in the accompanying drawings and the description below. Other features, objects, and advantages of the invention will be apparent from the description and drawings, and from the claims.

DESCRIPTION OF DRAWINGS FIG. 1 is an x-ray diffraction profile for the melt-spun silicon-aluminum-iron powder described in Example 1.

FIG. 2 illustrates the cycling performance, in terms of capacity vs. cycle number, for half-cells based upon the melt-spun and non-melt spun silicon-aluminum-iron powders described in Example 1.

FIG. 3 is a scanning electron microscope (SEM) photograph of the melt-spun silicon-aluminum-iron powder described in Example 1.

FIG. 4 illustrates the differential capacity vs. voltage for a half-cell based upon the melt-spun silicon-aluminum-iron powder described in Example 1.

FIG. 5 is an x-ray diffraction profile of the melt-spun silicon-aluminum-iron powder described in Example 1 prior to cycling and after 35 cycles.

FIG. 6 is a plot of capacity vs. cycle number for the powder described in Example 2.

Like reference symbols in the various drawings indicate like elements.

DETAILED DESCRIPTION Electrode compositions are described that are particularly useful as anodes for lithium-ion batteries. The electrode compositions feature an electrochemically active phase that includes elemental silicon and an electrochemically inactive phase that includes two or more metal elements and, preferably, silicon. Examples of suitable metal elements include iron, aluminum, nickel, manganese, cobalt, copper, silver, and chromium, with iron, copper, and aluminum being particularly preferred. The two phases have the microstructure described in the Summary, above.

The electrode compositions are preferably prepared by a chill block melt spinning process. Such processes are described generally, for example, in"Amorphous Metallic Alloys, "F. E. Luborsky, ed., Chapter 2, Butterworth & Co. , Ltd. (London), 1983.

According to this process, ingots containing silicon and two or more metal elements are melted in a radio frequency field and then ejected through a nozzle onto the surface of a <BR> <BR> rotating metal wheel (e. g. , a copper wheel). Because the surface temperature of the copper wheel is substantially lower than the temperature of the melt, contact with the surface of the wheel quenches the melt. Quenching prevents the formation of large crystallites that are detrimental to electrode performance.

The electrode compositions are particularly useful as anodes for lithium-ion batteries. To prepare a battery, the electrode is combined with an electrolyte and a cathode (the counterelectrode). The electrolyte may be in the form of a liquid, solid, or gel.

Examples of solid electrolytes include polymeric electrolytes such as polyethylene oxide, polytetrafluoroethylene, fluorine-containing copolymers, and combinations thereof.

Examples of liquid electrolytes include ethylene carbonate, diethyl carbonate, propylene carbonate, and combinations thereof. The electrolyte is provided with a lithium electrolyte salt. Examples of suitable salts include LiPF6, LiBF4, and LiC104. Examples of suitable cathode compositions include LiCo02, LiCoo. 2Nio. 802, and LiMn204.

EXAMPLES Example 1 6.34 g of aluminum shot, 12.10 g of silicon flakes, and 6.56 g of iron flakes (all 99.9% or better purity) were weighed in a weighing dish and then placed in an arc furnace.

The mixture was melted in an Ar atmosphere in the presence of a Ti pool oxygen getter to yield 25 g of an ingot having the composition Si55Al3oFel5, where all amounts are in atomic percent.

The ingot was broken into pieces less than 15 mm in diameter. 10 g of this material was placed inside a quartz tube ending in a 0.035 mil (0.89 pm) diameter nozzle.

A thin carbon sleeve was also inserted in the tube as a radio frequency coupler to initiate melting of the ingot. The tube was placed in the chamber of a melt spinner above a 200 mm diameter copper wheel such that the distance from the nozzle orifice to the wheel surface was 10 mm. The chamber was then evacuated to 80 mTorr and backfilled with He to 200 Torr. The ingot was then melted in a radio frequency field. When the melt had reached 1150°C, the molten liquid was ejected at 80 Torr He overpressure onto the copper wheel rotating at a surface speed of 35 m/sec to quench the melt and form ribbon fragments. Approximately 9 g of ribbon fragments were collected.

The ribbon fragments were pulverized by ball milling in an aqueous slurry in a planetary mill for 1 hour to form a powder. After air-drying at 80°C in an oven, the powder was classified by sieving through sieves having pore sizes of 53 microns, 32 microns, and 20 microns. The fraction between 32 and 53 microns was selected for further investigation. Its x-ray diffraction pattern was collected using a Siemens Model Kristalloflex 805 D500 diffractometer equipped with a copper target x-ray tube and a diffracted beam monochromator. The results are shown in Fig. 1. Analysis of the peak widths suggests a crystallite size of 494 angstroms for the elemental silicon phase and 415 angstroms for the iron and aluminum-containing phase.

Fig. 3 is a scanning electron microscopy (SEM) photograph of the classified powder. As shown in Fig. 3, the microstructure of the powder features discrete regions of elemental silicon that share a phase boundary with regions of the silicon-aluminum-iron ternary alloy.

The remaining ingot material, which had not been subjected to melt spinning, was similarly pulverized to form a powder and classified, and the x-ray diffraction pattern of the fraction between 32 and 53 microns measured. Analysis of peak widths suggests a crystallite size of 1243 angstroms for the elemental silicon and 732 angstroms for the remaining components. Melt-spinning, therefore, resulted in the formation of materials with significantly smaller crystallite sizes.

To prepare electrodes for electrochemical cycling, 0.8 g of each powder was suspended in 1 g of N-methyl-2-pyrrolidinone (NMP). Next, 3.6 g of a 6% solids suspension of super P carbon (available from MMM, Belgium) in NMP and polyvinylidene fluoride (Kynar 461, available from Elf Atochem), 1: 1 by weight, were added to the powder suspension. The resulting suspension was stirred at high shear for 5 minutes, and then coated onto a 12 mil (0.305 mm) copper foil with a notch bar to provide an 80% active, 10% polyvinylidene fluoride, 10% super P carbon coating. The coating was dried in vacuo at 150°C for 4 hours to form the electrode. The electrode was then used to construct 2325 coin cells by combining it with a metallic lithium anode, two layers of Cellgard 2400 as the separator, and 1 M LiPF6 in a 1: 2 mixture of ethylene carbonate and diethyl carbonate as the electrolyte.

The cells were cycled using a MACCOR cycler at a constant current of 0.125 mA between 0.9V and 0.025V for the first cycle, and at a constant current of 0. 5mA between 0.9V and either 0. 050V or 0. 005V for all additional cycles. The results are shown in Fig.

2. As shown in the figure, the performance of the melt-spun material (black triangles), with its smaller crystallites, was superior to the performance of the non-melt spun material (black diamonds). In addition, enhanced performance was observed for voltages above about 50mV. Specifically, the melt-spun material exhibited an average coulombic efficiency of 99.3% when cycled to 50mV (black triangles). However, that value dropped to 98.2% when the material was cycled to 5mV (open squares).

The differential capacity curve of Fig. 4 contains three curves. Curve (a) represents results obtained after one cycle. Curve (b) represents results obtained after two cycles. Curve (c) represents results obtained when lithiation was limited to 50mV. The results illustrate that the amorphous silicon phase of the melt-spun material remains amorphous when lithiation is limited to about 50mV. Values below 50mV, on the other hand, result in the formation of crystalline silicon.

Fig. 5 compares the x-ray diffraction pattern of the anode prior to the first cycle <BR> <BR> (trace (a) ) and after 35 cycles (trace (b) ). As shown in the figure, after 35 cycles the silicon phase was amorphous, but the crystallite size of the silicon-aluminum-iron phase remained substantially unchanged.

Example 2 Melt spun SissAl3oFels ribbon was prepared, pulverized, and classified as described in Example 1. The fraction between 32 and 20 microns was isolated. A portion of this fraction was coated with a porous layer of Ag according to the procedure described in Krause et al. , U. S. S. N. 09/883,865 filed June 18,2001 and entitled"Electrode Compositions Having Improved Cycling Behavior,"which is assigned to the same assignee as the present application. The weight uptake was 10%. The silver-coated particles were dispersed in methyl ethyl ketone and further reacted with 3- aminopropyltrimethyoxysilane (Aldrich Chemical) (60 mg silane per 1 g of powder) by shaking for 8 hours.

The treated powder was used to prepare electrodes as described in Example 1 except that the binder was a fluorochemical elastomer available from Dyneon LLC under the name FC-2179, the carbon was Super S carbon, and the final coating composition contained 80% active powder, 14% carbon, and 6% binder. The performance of half cells incorporating these electrodes, in terms of capacity vs. cycle number, is shown in Fig. 6.

The half cells were prepared as described in Example 1. As shown in Fig. 6, the cells exhibited good cycling performance.

Example 3 6.98 g of aluminum shot, 14.80 g of silicon flakes, and 8.22 g of copper shot (all 99.9% or better purity) were weighed in a weighing dish and then placed in an arc furnace.

The mixture was melted in an Ar atmosphere in the presence of a Ti pool oxygen getter to yield a 30 g ingot having the composition Si57AI28CU14, where all amounts are in atomic percent.

The ingot was broken into pieces less than 15 mm in diameter. 10 g of this material was placed inside a carbon tube ending in a 0. 030 mil (0. 76, um) diameter nozzle.

The tube was placed in the chamber of a melt spinner above a 200 mm diameter copper

wheel such that the distance from the nozzle orifice to the wheel surface was 10 mm. The chamber was then evacuated to 80 mTorr and backfilled with He to 200 Torr. The ingot was then melted in a radio frequency field. When the melt had reached 1200°C, the molten liquid was ejected at 80 Torr He overpressure onto the copper wheel rotating at a surface speed of 35 m/sec to quench the melt and form ribbon fragments. Approximately 9 g of ribbon fragments were collected.

The ribbon fragments were pulverized by grinding in a mortar and pestle. The powder was classified by sieving through sieves having pore sizes of 53 microns, 32 microns, and 20 microns. The fraction between 32 and 53 microns was selected for further investigation. Its x-ray diffraction pattern was collected using a Siemens Model Kristalloflex 805 D500 diffractometer equipped with a copper target x-ray tube and a diffracted beam monochromator. The XRD pattern showed the presence of only the phases Si and Al2Cu. Analysis of the peak widths suggests a crystallite size of 395 angstroms for the elemental silicon phase and 270 angstroms for the Al2Cu phase.

The powder sample was made into a coated electrode, incorporated into an electrochemical cell, and cycled as described for the powder sample in Example 1.

Cycling was done by constant current (0.25 mA) charge and discharge between 0.9 V and 0.05 V for the first cycle, and 0.9V and 0.070 V for all additional cycles. The cell had a first discharge capacity of 1680 mAh/g and had a differential capacity curve showing only the characteristics of fully amorphous silicon after the first cycle.

To confirm that the Al2Cu phase was electrochemically inactive, 9.18 g of aluminum and 10.82 g of copper (all 99.9% or better purity) were placed in an arc furnace.

The mixture was melted in an Ar atmosphere in the presence of a Ti pool oxygen getter to yield a 20 g ingot having the composition Al2Cu. The ingot was ground with a mortar and pestle, and classified by sieving through sieves having pore sizes of 53 microns, 32 microns, and 20 microns. The fraction between 32 and 53 microns was selected for further investigation. Its x-ray diffraction pattern, which was collected as described above, corresponded to that of the Al2Cu phase.

The powder sample was made into a coated electrode, incorporated into an electrochemical cell, and cycled as described above. Cycling was done by constant current (0.25 mA) charge and discharge between 0.9 V and 0.005 V. The cell showed no capacity from the Al2Cu phase, establishing that it was electrochemically inactive.

A number of embodiments of the invention have been described. Nevertheless, it will be understood that various modifications may be made without departing from the spirit and scope of the invention. Accordingly, other embodiments are within the scope of the following claims.