Preparation of Hydrogen Storage Ti-V-Cr Alloy from the Oxide Mixture in CaCl2

Yoshitatsu Kano1, Ryosuke O. Suzuki2
Dept. Energy Sci. & Techn., Kyoto University
Dept. Material Science, Hokkaido University

V-Ti-Cr based solid solution with body-centered cubic (BCC) structure has been developed to be one of the promising hydrogen storage alloys. In the industrial production of the elemental metals in this alloy, the Ti oxide is once converted to the chlorides and reduced by Mg, and Cr and V are extracted by the electrolysis or reduction by Al or Si. For alloying, these elemental metals are arc-melted several times, heated for homogenization, and pulverized. As shown in Fig. 1(a), this long process wastes thermal and chemical energy. It may become a barrier for mass production of this alloy. This work proposes a simplified synthesizing process to save energy consumption. Fig. 1(b) illustrates our proposal that the oxide mixture is simultaneously reduced and converted to the alloy powder. The co-reduction using calcium was reported as a successful method to synthesize Ti-Al, Nb-Al and Ti-V binary alloys or intermetallic compounds such as SmCo5, Nb3Sn and TiCr2. The purpose of this work is to apply the co-reduction technique to the formation of hydrogen storage alloys with the combination of CaO electrolysis in the molten CaCl2. Ti has a stronger affinity with oxygen than V and Cr, as evaluated from the oxygen potential of the binary systems of Ti-O, V-O and Cr-O. The Ca/CaO equilibrium can decrease the oxygen content in Ti to 0.05 mass% level. Namely, Ca can reduce all the V, Ti and Cr oxides to their metals with a low oxygen level. When Ca is reacted with the oxide mixture, we can expect alloying with these metals, i.e., reaching a lowest state of energy, in addition to the reduction.

Previous attempts of Ca co-reduction, however, showed higher residual oxygen content than inferred from the thermodynamic evaluation and the alloys were relatively heterogeneous. This is because the by-product CaO of this reduction disturbs the reduction and alloying, as illustrated in Fig. 2(b). Okabe et al. proved in the deoxidation of pure Ti that a lower oxygen content than that in the Ti/Ca/CaO equilibrium was attainable when molten CaCl2 was applied, because CaCl2 could dissolve, for example, about 20 mol% CaO at 1173 K. The dissolution of the by-product CaO in-situ into the molten CaCl2 was successfully applied to enhance the Ca reduction of TiO2.

For the alloy synthesis, we expect that the reduction, the deoxidation and the alloying would be accelerated by the addition of CaCl2, as shown in Fig. 2(c). The alloying between the metallic M1 and M2 will be enhanced by the CaO removal into the molten CaCl2, although Ca liquid flows above the molten CaCl2. It is noted that Ca can dissolve for a few mol% in the CaCl2, and that CaCl2 and its mixed salts can work as a strong reducing atmosphere. Additionally we plan to recycle the by-product CaO to the reductant Ca using the concept of CaO electrolysis in the molten salt, known as "OS process".

Figure 3 illustrates the final concept to synthesis the ternary Ti-V-Cr alloy powder directly from the oxide mixture. Ca electrochemically precipitates at the cathode from the dissolved CaO, and it dissolves near the cathode as the reductant for the oxide mixture that is injected near the cathode. The oxygen ion is removed as CO or CO2 gas by reacting with the carbon anode as,

Ca2+(in CaCl2) + 2e = Ca (in CaCl2) (1)
2 O2-(in CaCl2) + C = CO/CO2(gas) + 2 e- (2)
MO + Ca(in CaCl2) = M + Ca2+ + O2- (3)
M1 + M2 + M3 = (M1-M2-M3 alloy) (4)

For example, the theoretical decomposition voltages necessary for CaO and CaCl2 are 2.6 and 3.21 V, respectively, at 1173 K. Those for CO2 and CO gas evolution are lowered to 1.63 and 1.54V, respectively, if the carbon anode is used as eq.(2). The voltage increase due to the activity changes of Ca and CaO is evaluated < 0.1 V.

In case of this proposal, we may control the reduction rate using the supplied current, while the calciothermic reduction can not be well controlled because of its significant exothermic heat.

The carbon crucible (120mm ID., 320mm in height) was used as anode, and the Ti net (#100) was used as the holder (15mm ID, 60mm in height) of oxide powder (3g) and simultaneously as the cathode. They were heated at 1173 K in a purified Ar gas atmosphere. After the electrolysis at 1173 K, the sample in the cathode was cooled and rinsed with the distilled water to remove the solidified salt. Subsequently it was washed in acetic acid, water and ethanol for several times, centrifugally separated, and dried in vacuum. The hydrogen storage property was measured by the Sievelts method at 293 K.

During the electrolysis the gas bubbling at the anode was clearly observed after 1.2 ks, and the black or gray powders were obtained after 10.8-86.4 ks when 2.5-3.0V was applied. The average current density at the cathode and anode were 220-320 and about 10 mA/cm2, respectively. The recovered powders were mainly the BCC solid solution with a small amount of TiCr2 by XRD phase identification. The oxygen contents were in the level of 2000-4400 mass ppm. This means that the co-reduction worked well when the sufficient amount of electric charge was applied.

When the samples were cooled in the molten salt or in the furnace, a small amount of TiCr2 was found by XRD analysis. TiCr2 phase can coexist with the BCC solid solution at the higher temperatures and at the wide compositional region. Note that both the BCC and TiCr2 phases can absorb hydrogen, but the former was considered superior.

Although the phase diagram was not known below 973 K, TiCr2 phase might precipitate during our slow cooling from 1173 K with the solidified salt. Some samples were rapidly cooled by pulling up the cathode to the upper part of the furnace and by blowing the cool Ar gas. XRD measurement shows that the precipitation of TiCr2 phase was hardly found in these samples. Figure 4 show the SEM images and their elemental mappings in these samples. The primary particles in the obtained powder are a few mm in size, and they are slightly sintered and look like the coral with a wide surface area, suitable for hydrogen absorption. The elemental distribution is more homogeneous in the rapidly cooled sample.

Figure 5 shows the PCT curve at 293K of the rapidly cooled sample. Our sample absorbed about 1.9 mass% hydrogen at maximum after the normal activation procedure. The effective hydrogen storage is evaluated as 1.68mass%H, which is about 87.5% of the reported value at 298K for the arc-melted and well-annealed sample.


This work confirmed a proposal that the calciothermic co-reduction of the oxide mixture can form a fine alloy powder directly, and that the by-product CaO in the reduction can be recycled to the reductant Ca by the molten salt electrolysis, so called as "OS process". This new proposal will be applicable as the general production process for the alloy powder.

[Published in Proceedings of 11th World Conference on Titanium (Ti-2007), (Kyoto, Japan, June 3-7, 2007) "Ti-2007 Science and Techology", ed. by M.Niinomi, S.Akiyama, M.Ikeda, M.Hagiwara and K.Maruyama, The Japan Institute of Metals (JIM), Sendai, Japan, 2007, pp.107-110.]

Fig.1. Conventional synthesis of the alloy powder(a) and co-reduction from the oxide mixture combined with the OS process(b).

Fig. 2. Behaviour of CaO in CaCl2 melt.

Fig. 3. Proposal for alloy powder formation in CaCl2 with OS process.

Fig. 4. SEM image and element mapping for the rapidly cooled sample. Fig. 5.  PCT curve of the rapidly cooled sample.