Abstract
Ballmilling driven Li reactions are shown to be of great value to either mimic electrochemical discharge reactions in secondary Li cells or synthesize new single-phase Li-based materials. The former is illustrated by the reduction of CoO and 
to Co metal plus 
or 
respectively. To demonstrate the second effect, room temperature synthesis of 
and 
a member of a new class of promising negative electrode materials, is presented. Using Li powder as the basic ingredient, such reactions are rapid and easy to perform. © 2003 The Electrochemical Society. All rights reserved.
Export citation and abstract BibTeX RIS
The current interest in the development of more efficient Li-ion technologies has prompted the study of several new families of either positive or negative electrode materials. On the positive side, the main focus is presently on phosphates and Cr-based layered oxides. On the negative side, in addition to the Sn-based and 3d metal oxides and intermetallics, an increasing interest is focused on metal nitrides, phosphides, or antimonides families as a possible alternative to carbon materials due to their ability to react reversibly with large amounts of Li per formula unit. For instance, 
-type phases 
1
2 were recently reported to react reversibly with about 9 lithium atoms per formula unit. Such phases are usually synthesized at high temperatures in sealed stainless steel (SS) tubes. However, the samples obtained at high temperatures always contained second phases, regardless of attempted synthesis conditions. Thus, the search for other alternatives continued, and due to a complex solution chemistry, the ballmilling synthesis approach was utilized. Chemical reactions induced in this manner have been recognized for quite some time, and have been widely applied to various research fields as a simple and environmentally friendly alternative to high temperature synthesis or solution chemistry.
Such a process involves repeated welding fracture and rewelding of powder particles in a dry, high energy ball mill, resulting in particle and grain size reduction, together with structural defects. Energetic lattice defects, combined with short diffusion distances, are the driving forces of a faster solid-state alloying and chemical reaction at low temperatures. Surprisingly, these advantages did not capture the field of Li energy storage until the pioneering work3 of Disma et al. on the profound effect of ballmilling on the electrochemical performance of carbonaceous materials toward Li. This technique has been broadly applied to either synthesize, amongst other things, intermetallic compounds able to alloy with Li,4
5 or to modify powders morphologies to obtain, for instance, electrochemically optimized 
powders operating at 3 V6 or, more recently, phosphates powders with enhanced electrochemical performances.7 Nevertheless, mechanical reactions between Li metal and various elements have until now attracted less attention, with the exception, as far as we know, of only one report dealing with the preparation of 
using metallic Li powders.8
We decided to pursue this route further, and we successfully used ballmilling as reported herein as: 
a way to mimic displacement reactions in primary cells9 such as 
which involves extensive bond breakage, atomic reorganization, and formation of new bonds, and 
as a means to easily synthesize Li-based phases at room temperature.
Experimental
Attempts to perform ballmilling experiments in the presence of metallic Li did not succeed, due to its high ductility that resulted in a sticky Li deposition on the grinding vial walls. To circumvent this limitation, metallic Li powders were used as the Li precursor, together with a solvent (selected from the hydrocarbon family, and more specifically, dodecane). The CoO powder was utilized as received from Union Minière. The Sb, Ti, and P precursor powders, purchased from Aldrich chemicals, were always stored and handled in an argon glove box. 
was synthesized by reacting stoichiometric mixtures of Co and Sb powders for 12 h at 750°C in sealed quartz ampules.
Ballmilling experiments (denoted hereafter as BM) were performed using a Spex 8000 mixer mill that generates normal mechanical strain. The grinding vials were loaded and sealed in an argon dry box, using a weight ratio of SS ball to powder of 8:1. Unless otherwise specified, 
of dodecane was added to 400 mg of precursor mixtures, and grinding times ranged from 2 to 48 h.
Powder purity and crystallinity were examined by X-ray diffraction (XRD), with a Scintag diffractometer operating in Bragg-Brentano geometry with a Cu Kα radiation or, unless otherwise specified, by a Siemens D8 diffractometer using Co Kα radiation, and equipped with a power spectral density detector. Because of the moisture sensitivity of the Li-based reduced samples, a hermetically closed sample holder was used. The powders morphology and composition were investigated by scanning electron microscopy with a Philips XL 30 field emission gun, coupled to an Oxford Link instrument for energy-dispersive X-ray spectroscopy (EDS). Transmission electron microscopy (TEM) was carried out using a Philips CM12 microscope equipped with an EDS analyzer.
The ballmilled samples were tested in Swagelok-type cells assembled in an argon-filled dry box, using Li metal as the negative electrode, and a Whatman glass filter, size D (GF/D) borosilicate glass fiber sheet, saturated with 1 M 
in ethylene carbonate (EC), dimethyl carbonate (DMC) (1:1 in weight) as the electrolyte. The positive electrodes consisted of a few milligrams of powder directly deposited on the swagelok plunger. The electrochemical lithium reactivity was monitored with a VMP potentiostat/galvanostat (Biologic SA, Claix, France), operating either in galvanostatic, galvanostatic intermittent titration technique (GITT) or potentiostatic intermittent titration technique (PITT) mode.
Results and Discussion
By ballmilling Li powders (5% in excess) with either CoO or 
a first attempt was made to simulate the electrochemical reduction of CoO and 
by Li, which was shown to lead to the formation of 
10 and 
11 composites, respectively. From a study of various grinding times coupled with XRD and SAED analyses, we concluded that the ballmilling-driven Li reduction reactions were complete after 10 h for the CoO system and about 12 h for the 
samples. The XRD pattern of the 10 h milled CoO sample was featureless with the exception of weak, broad peaks corresponding to Co, while its SAED pattern (Fig. 1a) showed well-defined rings corresponding to the 
and Co phases. There was no evidence for traces of residual CoO. Regarding the 12 h milled 
sample, its XRD pattern (Fig. 2) revealed the presence of a single face-centered cubic fcc phase 
with lattice parameter 
in agreement with the reported literature value of 6.57 Å (JCPDF 040791). Unlike the CoO sample, XRD showed no evidence of Co metal after ballmilling. The SAED patterns (not shown here) confirmed the formation of 
with no evidence of residual 
or Co metal, as reported for the electrochemical reduction of 
powders.12
Figure 1. The voltage/composition curve is shown for a composite electrode obtained by ballmilling a mixture of CoO and Li powders, together with its TEM micrograph (a) and SAED pattern (b). Note that the ballmilled composite consists of a mixture of 
and 
fully consistent with the 0.7 V starting cell potential.
Figure 2. X-ray powder pattern of a powder obtained after ballmilling of a 
composite, together with its electrochemical behavior in Li half cells. The 
-ray diagram is analogous to that of 
The two extra peaks around 30° and 33.5° in 2θ, denoted by asterisks, are due to the cell hardware.
The electrochemical properties of these samples were reported in Fig. 1 and 2. Note that 1.2 Li from the 
composite can be removed, while 2Li can be removed from the electrochemically formed sample.10 This difference is rooted in the nanoparticle size, which is about 100 Å (Fig. 1b) for the BM sample, compared to 20 Å for the electrochemically made sample,10 thus leading to the expectation of a very difficult 
conversion process. For the BM 
sample, the oxidation curve is similar to that observed with an electrochemically made 
composite. These support the use of mechanical milling in the presence of Li metal powders as a new way of obtaining Li-based reduced composites.
We further exploited the possibility of using room temperature milling of Li powders to synthesize phases, such as 
that usually require high temperature and a controlled environment. Highly divided 
powders were synthesized by mixing stoichiometric amounts of lithium and Sb powders together with 
of dodecane. The best synthesis results were obtained for grinding times ranging from 1 to 2 h. The XRD pattern of a sample ballmilled for 1 h exhibited Bragg peaks (Fig. 3b) that were indexed with 
corresponding to the 
fcc phase with 
This contrasted with the high temperature 
phase (Fig. 3a) made from Li foil and Sb powders in sealed SS tubes. Although the high temperature and ballmilled samples exhibited similar voltage profiles, as shown in the insets of Fig. 3a and b, respectively, the BM sample exhibited lower capacity. No simple explanation was identified for the latter feature.
Figure 3. X-ray powder pattern of 
made at (a) high temperature as compared to 
prepared by (b) ballmilling. Their corresponding electrochemical signature, referred to as a and b, are shown as inset. While the 
curves are similar, note that the ballmilled sample is devoid of non indexed extra peaks.
Recently, it was shown2
13
14 that 
-type phases hold great potential with respect to applications in Li-ion cells. However, the challenge consists in synthesizing such compounds as single phases. Their structure15 can be viewed as an fcc framework of Pnictogens (P, N, As, Sb,..) having 12 available cationic sites per formula unit, with Li and Ti statistically partially occupying the 8 tetrahedral and four octahedral sites. They are usually synthesized by 
introducing, within an argon dry box, a mixture of stoichiometric amounts of 
Ti, and P powders in an arc welded SS container, 
placing the container into an oven, whose temperature is raised to 950°C at 15°C per hour, and held at this temperature for 48 h, and 
quenching the tube in air. The resulting crystallized powders were always found (Fig. 4a) to be contaminated by second phases, primarily 
TiP, and 
whose cumulative amounts always exceeded 25%, despite efforts to modify the annealing protocol or lithium stoichiometry. Hence, the need to search for a different synthesis path to fully exploit the Li reactivity within these phases was clear. In continuing our demonstration of the positive attributes of Li powder ballmilling to synthesize Li-based electrodes, we decided to extend this method to the preparation of 
Figure 4. X-ray powder patterns of 
made by (a) high temperature synthesis, by (b) ballmilling, and by (c) ballmilling followed by 500°C post-treatment, together with their corresponding voltage composition curves. Note the single phase character of the ballmilled samples and the post treatment benefit on sample crystallinity.
The experiments were carried out by mixing, within an argon dry-box, raw powders of Ti, P, and Li elements in the right stoichiometry to reach the 
reported composition.14 The mixture was placed into an SS vial containing an SS ball whose weight was 10 times greater than the mass of the initial powder mixture. The optimum milling parameters were determined, and the best results were obtained for a grinding time of 12 consecutive hours at room temperature. The XRD powder pattern for the BM powder (Fig. 4b) showed Bragg peaks reminiscent of the 
phase with an 
axis of 6.02(2) Å only, as compared to 5.96 Å for the reported literature value (JCPDS 21-0529). Although such a lattice parameter value (6.02 Å) is not accurately determined due to the limited number of broad Bragg peaks used to extract it, we believe its difference with the reported one (5.96 Å) to be meaningful and indicative of the high density of lattice/structural defects introduced during the BM process. Surprisingly, no second phase could be detected, either by X-rays or by TEM measurements (not shown here), suggesting once again the profound advantages of the BM synthesis. However, caution has to be exercised in over-interpreting such a result, since the samples could be contaminated by amorphous phases.
Thus, to clarify this issue, the ballmilled samples were annealed at various temperatures and success was achieved by obtaining crystallized samples with 
devoid of any impurity phases (Fig. 4c) at temperatures as low as 300 and 500°C. The sample Li composition could not be ascertained accurately, since it was reported2
16 that changing the Li content from 
to 9 for the phosphide phase did not change the 
axis by more than 0.05%. Therefore, an 
value of 8.5 was determined by atomic absorption analysis. Annealing at temperatures greater than 900°C did result in the presence of impurity phases similar to those observed for the high temperature synthesis method, suggesting the metastability of 
and once again stressing the advantage of BM.
We performed electrochemical tests using 
half cells to further validate the mechanical milling-induced formation of pure 
phase. The voltage/composition curves for 
cells made of either the high temperature, room temperature ball-milled, or 500°C reannealed ballmilled sample, as the positive electrode, are shown in Fig. 4a, b, and c, respectively. Note that whatever the sample elaboration process, more than 6 lithium atoms can react reversibly with 
yielding capacities of at least 700 mAh/g, twice as much as the capacities obtained from carbonaceous materials. However, these samples exhibit differences in the voltage profile during oxidation. While the high temperature and the 500°C BM sample showed a well-defined plateau on oxidation, the room temperature ballmilled sample was devoid of this feature. This is mainly due to the nanosized character of the room temperature ballmilled produced particles, which generates many new indefinite interfaces with various interfacial energies, resulting in some potential distribution. In contrast, the origin of an oxidation plateau on charge as compared to a sloppy voltage discharge for the 500°C ballmilled and heat-treated sample is nested in amorphization/recrystallization matters, as discussed in a separate paper.14 Finally, though not plotted herein, the capacity retention of BM samples did not exhibit any improvement compared to the HT ones. We could only achieve 20 cycles at C/5 while maintaining 80% of the initial capacity. Therefore, efforts still have to be made toward the optimization of such recently identified electrode materials.
Conclusions
We showed that Li electrochemically driven displacement reactions, which involve extensive bond breakage, atomic reorganization, and formation of new bonds can be mimicked by Li-assisted mechanical milling. Similarly, we illustrate the benefits of ballmilling in the presence of Li powder to synthesize, under mild conditions, new metastable Li-based phases. Therefore, these BM-made phases suffer from poor capacity retention and further work is required on that issue. Overall, these ballmilling-driven Li reduction reactions are of great value in screening new materials, since they are free of the complexities associated with high temperature processes in sealed environments. Needless to say, such an approach is being widely implemented with a wide variety of materials.
Acknowledgments
The authors would like to give special thanks to L. Dupont for enabling TEM studies, and L. Aymard, A. Rougier and R. Janot for technical discussions.
Université de Picardie Jules Verne assisted in meeting the publication costs of this article.