Invention and fundamentals of the FFC Cambridge Process

CHAPTER 11

Invention and fundamentals of the FFC Cambridge Process George Z. Chen1, 2, Derek J. Fray3 1

Department of Chemical and Environmental Engineering, Faculty of Engineering, University of Nottingham, Nottingham, United Kingdom; 2Department of Chemical and Environmental Engineering, Faculty of Science and Engineering, University of Nottingham Ningbo China, Ningbo, P. R. China; 3 Department of Materials Science and Metallurgy, University of Cambridge, Cambridge, United Kingdom

1. Background: how the concept of electro-deoxidation came about The path to the discovery of the Fray-Farthing-Chen (FFC) Cambridge Process started in the late 1980s when Tom Farthing visited Derek Fray in Cambridge seeking a way to reduce the alpha case (a solid solution of oxygen in titanium) that forms on the surface of titanium whenever it is heated at an elevated temperature in an oxidising environment. At that time, and even today, the only way to remove this brittle layer was either to dissolve the coating in hydrofluoric acid (HF) or grind it away, both methods resulting in a slight change in the dimensions of the artifact and the creation of hazardous wastes. The ideal solution would be to remove the oxygen from the alpha case without removing the titanium and changing the dimensions. Vacuum treatment might be a possibility but a very lower pressure would be required to dissociate the solid solution to titanium and oxygen. An alternative method was thought to use electrochemistry as the electrochemical potential to reduce titanium oxides to titanium and oxygen is only a few volts against the potential for chlorine gas evolution. In order for this to be practical, it would need to be performed in a molten salt with a solubility for oxygen ions. Initially, it was thought that if the alpha-cased titanium sample was made the cathode in a molten salt, the reaction on the surface of the sample would be Ti[O] þ 2e ¼ Ti þ O2

(11.1)

where [O] represents oxygen atoms dissolved in the alpha case [1]. When considering also the oxide scale above the alpha case, if present, the following reactions should occur as well [2]. Extractive Metallurgy of Titanium ISBN 978-0-12-817200-1 https://doi.org/10.1016/B978-0-12-817200-1.00011-9

Copyright © 2020 Elsevier Inc. All rights reserved.

227

228

Extractive Metallurgy of Titanium

TiO2 þ 4e ¼ Ti þ 2O2 

TiO þ 2e ¼ Ti þ O

2

(11.2) (11.3)

In the above reactions, the produced oxygen ions should dissolve in, and transport through a suitable molten salt electrolyte to be discharged on the anode as illustrated in Fig. 11.1A. The next challenge was to find a very stable salt that would not decompose under the applied potential for dissociation of the oxide, and also dissolve the oxygen ions. The consideration was to avoid the decomposition of the molten salts as illustrated in Fig. 11.1B. Calcium chloride is such a salt which is slightly more stable than TiO, so should not decompose, and has a significant solubility for oxygen ions. It is also very cheap as it is used for deicing the roads in winter and non-toxic as it is used as a food additive. The concept was to place the sample, coated with the alpha case, in a bath of molten calcium chloride and apply a cathodic potential of about 3 V versus Cl2/Cle. Theoretical calculations showed that this was a possible approach but funds were needed to demonstrate its practicality. Approaches were made to many of the industrial producers and users of titanium but none showed any interest in the concept. About 4 years later, research funds became available from the EPSRC, a UK government agency and, in 1994, George Chen joined Fray in the University of Leeds and work commenced on the project. In 1996, Chen and Fray transferred to the University of

Figure 11.1 The original drawings for the illustration of possible electrode reactions in a molten chloride salt, MCl, at appropriate applied temperature and voltage that could enable (A) cathodic ionisation of dissolved oxygen in the alpha case on titanium, and anodic discharge of the dissolved oxygen anions, and (B) the competing reactions of cathodic deposition of the metal, M, and anodic discharge of chloride anion, leading to decomposition of the molten salt [1].

Invention and fundamentals of the FFC Cambridge Process

229

Cambridge where the work started in earnest and they quickly showed that the alpha case could be removed (deoxygenated and deoxidised) by applying a voltage of 3 V within an hour [3]. They were very excited by the results and revisited the companies which Fray had originally contacted but, again, no interest was generated. About this time, a very thick layer of oxide thermally grown on a titanium foil in air was reduced to porous titanium, see Fig. 11.2, and it was surprising how easily it was reduced which posed the question e could the method be used to reduce pellets of TiO2 which is effectively an insulator [2]? The first successful experiment was carried out in 1997 and it showed that die-pressed and sintered pellets of TiO2 powder could be reduced to pure titanium metal in a time from 5 to 24 h, depending on the size and numbers of pellets, as exemplified in Fig. 11.3. Such a result was however met with a considerable amount of skepticism by the titanium community, including Tom Farthing! Eventually, QinetiQ with support from James Hamilton, a British business person, agreed in 1999 to put funds into the project for pilot testing, whilst Hamilton formed the first company, British Titanium plc, to commercialise the FFC Cambridge Process for making titanium and its alloys. In 2001, the Cambridge University Challenge Fund decided to exploit the commercial feasibility of using the FFC Cambridge Process for making non-titanium metals and alloys, and helped the creation of FFC Ltd which was subsequently Metalysis Ltd. At a slightly later time, BHP Billiton invested in the development of the FFC Cambridge Process, whilst DARPA also provided an R&D grant. These developments were finally incorporated into the Metalysis portfolio. Needless to say, a very

(A)

(B)

Figure 11.2 Photographs of (A) a piece of Tie6Ale4V alloy foil with a thin surface oxide scale and the underneath alpha case before (dark sample) and after (light gray sample) 1 h electro-deoxidation, and (B) the cross section of a piece of Ti foil with a thick oxide scale (ca. 50 mm) formed upon heating in air before (left) and after (right) 4 h electro-deoxidation in molten CaCl2 at 3.0 V and 950 C [2,3].

230

(A)

Extractive Metallurgy of Titanium

(B)

(C)

Figure 11.3 Photographs (A, B) and SEM images of microstructures (C, D) [2] of diepressed and sintered pellets of TiO2 powder before (A, C) and after (B, D) electrodeoxidation in molten CaCl2 at 950 C and 3.0 V for 12 h.

significant amount of negotiation was necessary to bring all these parties together and coupled with investments of millions of dollars to enable a very simple idea to become a reality, as shown in the paper by Ian Mellor.

2. Understanding of electro-deoxidation: interactions of the oxide cathode with molten salts At the same time as the commercialisation, detailed studies showed that the reactions at the cathode were far more complicated than originally thought. Chen and Fray [4] first observed that before the cathodic reduction of a metal oxide, MxOy, reached at the final metallic product, various intermediate products formed in the cathode pellets. Here, if MxOy is TiO2, x ¼ 1 and y ¼ 2, whilst for Cr2O3, x ¼ 2 and y ¼ 3. These intermediates were mostly co-oxides of CaO and the precursor oxide, or their partially reduced forms which were found to be conducting to electrons and hence could be further reduced to the respective metals, as described in the following postulated reactions [4]. MxOy þ CaO ¼ CaMxOyþ1 (e.g., TiO2 þ CaO ¼ CaTiO3) (11.4) MxOy þ 2ee ¼ MxOy-1 þ O2 (e.g., nTiO2 þ 2ee ¼ TinO2n-1 þ O2, n ¼ 2, 1)

(11.5)

MxOy-1 þ CaO ¼ CaMxOy (e.g., Ti2O3 þ CaO ¼ CaTi2O4) (11.6) MxOy þ Ca2þ þ 2ee ¼ CaMxOy (e.g., nTiO2 þ Ca2þ þ 2ee ¼ CaTinO2n)

(11.7)

MxOy þ Ca2þ þ 2(mþ1)ee ¼ nM þ CaMxenOyem þ mO2 (11.8)

Invention and fundamentals of the FFC Cambridge Process

231

CaMxOyþ1 þ 2yee ¼ xM þ CaO þ yO2

(11.9)

CaMxOy þ 2(ye1)ee ¼ xM þ CaO þ (ye1)O2

(11.10)

CaMxenOyem þ 2(yeme1)ee ¼ (xen)M þ CaO þ (yeme1)O2

(11.11)

Note that some of these reactions involving CaO are chemical in nature, but the original molten CaCl2 contained only a minimal amount of CaO, likely resulting from hydrolysis of CaCl2 when the salt was thermally dried in air. Thus, the large amount of the co-oxides, e.g., CaTiO3 and CaTi2O4, observed in the intermediate products suggests two routes for calcium intake into the cathode. Firstly, when electro-deoxidation of the metal oxide produced a sufficiently high concentration of O2 inside the pores of the oxide cathode, the activity product of O2 and Ca2þ could exceed the solubility product constant of CaO in the molten salt contained in the pores of the cathode. The consequence may be precipitation of CaO in the pores of the cathode, but more likely reaction between the freshly formed CaO and the metal oxide and its partially reduced forms in the cathode to produce the various co-oxides. Alternatively, the co-oxides could be electrochemically formed via (11.7) and (11.8). In another effort, Schwandt and Fray [5] obtained different products from electro-deoxidation of TiO2 pellets (relative density: 65%e70%) at stepwiseincreased cell voltages from 2.5 to 2.9 V. Based on structural, microscopic and elemental analyses, such as those shown in Figs. 11.5 and 11.6, they were (A)

(B)

Figure 11.4 Intermediate products from electro-deoxidation of (A) TiO2 and (B) Cr2O3 in molten CaCl2, as observed under SEM. EDX analyses suggested these crystallites to be co-oxides with elemental compositions of significant amounts of O and Ca plus (A) Ti and (B) Cr [4].

232

Extractive Metallurgy of Titanium

Figure 11.5 (A) Current-time plot obtained from electro-deoxidation of TiO2 in molten CaCl2 at 2.5 V and 900 C. The inserted image shows TiO cubes observed in partially reduced samples. (B) The XRD pattern of a product from 4 h electrolysis [5].

Figure 11.6 (A) EDX spectrium, SEM image (insert), and (B) XRD pattern of CaTi2O4 found in the product from electrolysis at 2.5 V for 8 h and then 2.7 V for 4 h in the 900 C molten CaCl2 [5].

able to show that the reactions on a TiO2 pellet cathode during the electrolysis were as follows: 5TiO2 þ Ca2þ þ 2ee ¼ Ti4O7 þ CaTiO3

(11.12)

4Ti4O7 þ Ca2þ þ 2ee ¼ 5Ti3O5 þ CaTiO3

(11.13)

3Ti3O5 þ Ca2þ þ2ee ¼ 3Ti2O3 þ CaTiO3

(11.14)

2Ti2O3 þ Ca2þ þ 2ee ¼ 3TiO þ CaTiO3

(11.15)

CaTO3 þ TiO ¼ CaTi2O4

(11.16)

CaTi2O4 þ 2ee ¼ 2TiO þ Ca2þ þ 2O2

(11.17)

TiO þ 2(1ex)ee ¼ TiOx þ (1ex)O2

(11.18)

Invention and fundamentals of the FFC Cambridge Process

233

As can be seen, for reactions (11.12) to (11.16), Ca2þ ions are removed from the melt into the TiO2 cathode, but there is no change in the oxygen content. It is not until the penultimate reduction step when the Ca2þ and O2 ions are returned to the melt. Note that the current-time plot in Fig. 11.5A was recorded at 2.5 V to purposely promote formation of different intermediate products, instead of the final metal. The cubic crystallites in the SEM image inserted in Fig. 11.5A, and the XRD pattern in Fig. 11.5B are solid evidence for the formation of TiO at relatively short times. It is interesting to note that the cubic TiO crystallites disappeared during later electrolysis when the cell voltage was increased from 2.5 to 2.7 V as shown in Fig. 11.6A and B. This observation was considered as evidence for the chemical consumption of TiO via reaction (11.16) to produce CaTi2O4. It should be also pointed out that the fairly pure CaTi2O4 phase is due partly to the applied cell voltage (2.7 V) being insufficient to invoke further reduction, such as reaction (11.17), or to form the Ti metal, and also the fairly long time of polarisation at the applied cell voltage. Obviously, it is a very complicated multistep process as the CaO content increases in the TiO2 cathode through a series of microscopic changes until the formation of CaTi2O4 whose further reduction then returns CaO to the melt. This understanding was confirmed by the elegant work of Bhagat and others at Imperial College and University of Warwick who used in situ synchrotron diffraction to study the reduction of TiO2 in real time [6]. Their main findings are presented in Fig. 11.7, showing the phase compositions near the surface and at the center of a TiO2 pellet as a function of the time of electrolysis in molten CaCl2 at 3.1 V and 900 C. It can be seen that CaTi2O4 is absent at all times near the surface, but appears in a large proportion near the center of the pellet. This difference implies progressive reduction from surface to center, and the effect of diffusion. It is interesting to note in Fig. 11.7 that the Ca metal, which is likely to form due to direct reduction of the Ca2þ ion, was not detected in any region

Figure 11.7 Variations of phase composition as a function of electrolysis time (A) at the center and (B) near the surface of a TiO2 pellet in molten CaCl2 at 900 C and 3.1 V [6].

234

Extractive Metallurgy of Titanium

of the pellet at any time of the electrolysis. This fact confirms that the cathodic reduction of TiO2 is more likely electrochemical in nature. However, it should be mentioned that calciothermic or electro-calciothermic reduction was reported for either deoxygenation of titanium [7] or deoxidation of TiO2 [8], particularly when CaO was purposely added into molten CaCl2 to help generate the Ca metal on the cathode [9]. One of the consequences of Ca2þ intake without oxygen removal (i.e., in situ perovskitisation) in these intermediate steps is the volume expansion of the solid phase which can in turn block the pores in the oxide precursor (pellet) cathode for ion movement, and retard the reduction. To avoid such a negative impact, researchers at Wuhan and Nottingham Universities purposely mixed and pressed TiO2 with CaO or CaCO3 powders into pellets, followed by heating to 1300 C in air. This process led to the formation of porous CaTiO3 pellets whose microstructure is shown in Fig. 11.8A. These perovskite pellets were then subjected to electrodeoxidation in molten CaCl2 under similar experimental conditions for reduction of TiO2. Note that the reaction temperature (1300 C) used to produce CaTiO3 in a dry furnace was much higher than the molten salt temperature (ca. 900 C) for electro-deoxidation. This fact implies a much lower kinetic barrier for perovskitisation under the conditions of the latter. Their work can be represented by reactions (11.19) to (11.22) below. Compared with TiO2, CaTiO3 was electro-deoxidised twice faster with higher current and energy efficiencies [10].

Figure 11.8 Porous microstructures of (A) CaTiO3 as prepared by reaction of TiO2 with CaCO3 at 1300 C, and (B) Ti from electro-deoxidation of CaTiO3 in molten CaCl2 at 3.2 V and 850 C for 265 min. The EDX analysis covering the whole area of (B) produced the spectrum shown [10].

Invention and fundamentals of the FFC Cambridge Process

TiO2 þ CaCO3 ¼ CaTiO3 þ CO2 (ex-situ perovskitisation)

235

(11.19)

CaTiO3 þ 4ee ¼ Ti þ Ca2þ þ 3O2

(11.20)

CaTiO3 þ 2ee ¼ TiO þ Ca2þ þ 2O2

(11.21)

TiO þ 2ee ¼ Ti þ 2O2

(11.22)

3. Understanding of electro-deoxidation: metal/insulator/ electrolyte 3PI models It is known that TiO2 is effectively non-conducting to electrons, which naturally leads to a question on how electrolysis could be conducted on an insulator oxide. This was thought because the highly conducting Magnelli phases (TinO2n-1, n ¼ 4 to 10) would form easily when TiO2 is exposed to an oxygen-deficient or reducing environment. However, in a separate and successful attempt to electro-deoxidise ZrO2 and Cr2O3 which could not form the Maganeli phases, Chen and Fray considered the possibility that the reduction actually proceeded via the propagation of the metal/oxide/ electrolyte three phase boundary (or interline) [12]. It was experimentally observed that, such as the findings presented in Fig. 11.7, the reduction occurred always from the surface to the interior of the oxide pellet when it was suspended on a metal wire, which supports the thought that electrodeoxidation proceeds through the propagation of the three phase interline (3PI) joining the metal, oxide and electrolyte phases as schematically illustrated in Fig. 11.9 [4,11]. Basically, as illustrated in Fig. 11.9A, in the process, electrons were conducted to the 3PI (the reaction site) via the metal, leading to discharge of the O2 ion from the oxide at the 3PI into the neighboring molten salt, and then the conversion of the oxide to the metal. Then, the newly formed metal helps form and move a new 3PI into the oxide until completion of the oxide to metal conversion. The 3PI model was further investigated by the Wuhan/Nottingham team who designed appropriate electrodes to observe the movement of the 3PI of metal/insulator/electrolyte, and developed mathematical models for such systems. They studied firstly the electro-reduction of a thin layer of AgCl powder in aqueous KCl electrolyte, and demonstrated a thin layer model for electron and ion transfer at the Ag/AgCl/electrolyte 3PI. This thin layer model was also successfully applied to account for charge transfer processes at the Si/ SiO2/molten salt 3PI as shown in Fig. 11.10. In both cases, potentiostatic reduction and cyclic voltammetry were applied and produced successfully

236

Extractive Metallurgy of Titanium

Figure 11.9 (A) Charge (electron and O2) transfer at the metal (Mo wire)/oxide (TiO2 particle)/electrolyte (molten CaCl2) three phase interline (3PI). (B) Aggregation of metal atoms resulting from electro-deoxidation and oxide surface renewal processes. (C) Continuous and consecutive metallisation of oxide particles. (D) 3PI growth and surface metallization of a quarter oxide pellet [4,11].

further experimental verifications [13]. An important conclusion from this study is that electro-deoxidation can proceed on an insulator metal compound via the 3PI propagation mechanism. This thin layer model is of fundamental importance, and can also account for the surface metallisation process in the electro-deoxidation of metal oxide pellets. However, to quantify electrodeoxidation of bulk metal oxide, a more sophisticated model is needed and has been developed by the Wuhan-Nottingham team. This is exemplified in Fig. 11.11A for electro-reduction of a SiO2 (quartz) pellet. According to the model, Fig. 11.11B, the optimal pellet thickness and potential have been identified for the electro-deoxidation of a solid quartz pellet to pure silicon, as shown in Fig. 11.11C and D [14,15].

4. Understanding of electro-deoxidation: the metal-tooxide molar volume ratio An important factor to the commercialisation of the FFC Cambridge Process is its current efficiency and energy consumption. Although the invention of the process was started from titanium, it was found in later studies that TiO2 is actually more complicated to electro-deoxidise than other metal oxides, such as SiO2, Cr2O3 and even ZrO2. The current efficiency for electro-deoxidation of TiO2 was found to be highly dependent on the oxygen content to be achieved in the final metallic product. To

Invention and fundamentals of the FFC Cambridge Process

237

Figure 11.10 Illustrations of (A) the metal/compound/electrolyte three phase interline (3PI) where electron and anion (Ae) transfer occurs, and (B) a quartz sheathed W wire, i.e., the WeSiO2 electrode and the predicted SiO2 to Si conversion as a function of time during reduction [13]. SEM images of the WeSiO2 electrode end (C) before, (D) after potentiodynamic reduction (0.5 to 1.7 V vs. Pt) in molten CaCl2 at 850 C, and (E) after washing away the reduced thin layer in (D) (side view) [16].

achieve an oxygen level below 3000 ppm in the produced Ti metal, the current efficiency could be as low as around 15%. This was much lower than that for making other metals, such over 75% for Cr (<2000 ppm O) [11] and 45% for Zr (ca. 1800 ppm O) [17]. It is particularly puzzling considering that both Ti and Zr have many comparable properties, such as high solid solubility for oxygen. Apart from in situ perovskitisation as discussed above, the Wuhan/Nottingham team discovered another intrinsic barrier: the molar volume ratio of metal to its oxide. Table 11.1 lists data for some common metals and their oxides [18]. The metal-to-oxide molar volume ratio, Vm/Vo, has long been used as an indicator of the stability of a metal exposed to air at elevated temperatures. If the ratio is smaller than 1, the oxide layer formed on the metal is able to completely cover the metal surface and functions as a protection barrier against further oxygen access to the underneath metal. Otherwise, the metal is vulnerable to oxidation because the oxide layer would be

238

Extractive Metallurgy of Titanium

Figure 11.11 (A) A quartz pellet cathode. (B) The in-depth propagation 3PI model for electro-deoxidation of a solid insulator metal compound cylinder to porous metal from right to the left. (C) Current efficiency and (D) energy consumption in relation with the thickness and applied potential of the quartz pellet cathode in molten CaCl2. The solid lines show the influences of the applied potential for fixed reduction depths and the dash lines present the influences of the reaction depth at fixed potentials as indicated [14,15]. Table 11.1 Metal to oxide molar volume ratios. M/ MOx

Ta/ Ta2O5

Cr/ Cr2O3

Zr/ ZrO2

Al/ Al2O3

Mg/ MgO

Ti/ TiO2

Ti/ Ti2O3

Ti/ TiO

V m/ Vo

0.40

0.50

0.66

0.78

1.25

0.63

0.74

0.91

porous. Following the same thought but in reverse, if the ratio is smaller than 1, the metal formed on the oxide surface would be porous, allowing continuous ion transport through the porous metal layer and complete reduction. For metals with larger molar volumes, the metal formed would cover the surface of the oxide, preventing transport of ions and any further reduction of the underneath oxide. This principle is illustrated in Fig. 11.12A. According to Table 11.1, it would be challenging, if not impossible, to reduce MgO to Mg because Vm/Vo > 1. However, reduction of ZrO2 is feasible because Vm/Vo < 1. For electro-deoxidation of TiO2, the process is multistep with reaction (11.22), i.e., TiO þ 2e ¼ Ti þ O2, at the later stage according to reactions (11.12) to (11.22). Because the molar volume ratio of Ti/TiO is close to 1, the TiO to Ti conversion is expected to be slow for oxygen removal.

Invention and fundamentals of the FFC Cambridge Process

239

Figure 11.12 (A) Schematic illustration of the dependence of electro-deoxidation on the molar volume ratio of metal (Vm) to metal oxide (Vo). (B, C) Porous Ti from electrodeoxidation of a 1.0 g TiO2 pellet of 68% in porosity at 3.2 V and 850 C for 3 h, viewed at low (B) and high (C) magnifications [18].

Fortunately, Ti metal is able to passage O atoms, although the diffusion in solid would take a longer time. To address this challenge, the Wuhan/ Nottingham team made the TiO2 highly porous using a recyclable fugitive, NH4Cl, resulting in a similarly highly porous reduction product, as shown in Fig. 11.12. As a result, the full metallisation of the TiO2 pellet of 68% in porosity was achieved in 3 h as shown in Fig. 11.12B and C, almost three times faster than that of 44% in porosity. Nevertheless, it can be noticed that the nodular Ti particles in Fig. 11.12C are fairly small. Assuming an oxide layer of 2 nm on the surface of the metal particles, the calculated oxygen content agrees well with the measured value of 6800 ppm. To help growth of the particle size, the electrolysis was allowed to last for a further 3 h, but at a lower voltage of 2.6 V to firstly avoid re-oxidation of formed small Ti particles, but also to reduce the charge waste via electronic conduction in the molten salt. This measure led to highly satisfactory results of 1800 ppm in oxygen content, 32.3% in current efficiency, and 21.5 kWh/kg-Ti in energy consumption. As expected from consideration of electronic conduction, it was recorded that the electrolysis current was changed from 0.7 to 0.2 A when the cell voltage was dropped from 3.2 to 2.6 V. Electronic conduction in molten alkali and alkaline earth halide salts is well known [19,20], and it was suggested to follow an electron hopping mechanism in molten CaCl2 by the reactions below where the front subscripts A and B represents the relative locations of the species in electrolyte [21,22]. Ca(II) þ Ca(0) ¼ Ca(I) þ Ca(I)

(11.23)

ACa(I)

þ BCa(II) ¼ BCa(I) þ ACa(II)

(11.24)

ACa(0)

þ BCa(II) ¼ BCa(0) þ ACa(II)

(11.25)

240

Extractive Metallurgy of Titanium

Note that both Ca(0) and Ca(I) may be produced at the cathode when the cell voltage applied was about 3.0 V because the decomposition of CaO occurs at about 2.6 V. According to reactions (11.23) to (11.25), it is possible to eliminate or reduce the effect of electronic conduction by prevention of electron hopping from reaching the anode. To achieve this goal, an effective way is to use an ionic conducting but electronic insulating membrane between the cathode and anode. A good example of such a membrane is made from yttria stabilised zirconia (YSZ) and similar materials which are conducting to oxygen ions at elevated temperatures (800e1000 C) and widely used in solid oxide (membrane) fuel cells. It was used by Uday Pal and co-workers at Boston for electro-winning of magnesium from MgO dissolved in molten chloride salts [23]. The process was termed solid oxide membrane (SOM) technology. By combining the SOM with electro-deoxidation in molten CaCl2, a team in Shanghai University studied the difference between using the SOM based anode at 4.0 V and 1000 C and a graphite anode at 3.1 V and 900 C for electro-deoxidation of mixed TiO2, SiO2 and C to produce the Ti5Si3/TiC composites [24]. Their results included current efficiency and energy consumption of 61.9% and 15.2 kWh/kgTi5Si3/TiC with the SOM anode for 5 h of electrolysis, but 12.5% and 58.7 kWh/kg-Ti5Si3/TiC with the graphite anode for 12 h of electrolysis. It is worth noting that the SOM (particularly YSZ) electrolyte is often made into a tube with a closed end. Therefore, it can be connected to the molten salt from either the external or internal surface of the tube. For the external surface in contact with the molten salt, the advantage is that the tube can be arranged in the cell in the same way as a conventional graphite rod anode. In doing so, the Shanghai team filled the tube with carbon saturated liquid tin in which a molybdenum wire was inserted as the anode current collector [24]. Alternatively, it is possible to use the SOM tube as the container of the molten salt, whilst using oxidation resistant metals such as Ag or Pt to form the anode coated directly on the external surface of the SOM tube [25]. Fig. 11.13 illustrates schematically the two forms of connection between the YSZ tube and the molten salts. Unfortunately, currently known SOM membranes are unstable for long term and repeated uses (>10 h) [24e27], and require relatively high temperatures to be activated for conduction of the O2 ion at a sufficiently high speed. Nevertheless, the promising prospect of the SOM technology for the FFC Cambridge Process is that there are other O2 ion conducting membranes developed for solid oxide fuel cells, such as gadolinia or samaria

Invention and fundamentals of the FFC Cambridge Process

241

Figure 11.13 Schematic illustration of electro-deoxidation of metal oxides (MOx) with a tubular SOM membrane that is connected to molten CaCl2 at the (A) external surface and (B) internal surface. The drawings are based on refs. [24,25].

doped ceria that could work at fairly mild temperatures (500e700 C) [26,27]. Because these solid electrolytes have been mostly developed for, and tested under dry conditions, their performance in molten salts should be studied for the FFC Cambridge Process.

5. Development of an inert anode for electro-deoxidation in calcium chloride based melts The anodic reaction in the FFC Cambridge Process is that the oxygen ions, from the cathode, migrate to the anode where discharge takes place. For most experiments, carbon anodes have been used and these react with the evolved oxygen to form mainly carbon dioxide with a small amount of carbon monoxide which causes the carbon anode to be consumed. 2O2 þ C ¼ CO2 þ 4ee

(11.26)

Due to the creation of CO2 gas bubbles, these can also cause erosion of the carbon creating particles or debris that can collect on the electrolyte surface, leading to a short circuit between the anode and cathode, reducing the current efficiency and raising the energy consumption. Carbon dioxide also dissolves in the electrolyte, probably via reaction with the O2 ion to form the carbonate ion which can migrate to the cathode where the following reaction can take place. 2 o CO2 E (900 C) ¼ 1.626 V (vs. Ca2þ/Ca) (11.27) 3 þ 4e ¼ C þ 3O

The liberated oxygen ions can then diffuse back to the anode with the net result being the transfer of carbon from the anode to the cathode, reducing the current efficiency and increasing energy consumption. The

242

Extractive Metallurgy of Titanium

form of the carbon deposited or liberated close to the cathode could be a mixture of nanoparticles and even nanotubes [28]. Fig. 11.14A is a photograph of the top view of an electrolysis cell after removal of the electrodes, showing the carbon debris floating on the surface of molten CaCl2. The optical micrograph of the carbon debris observed on the surface of the solidified CaCl2 melt is shown in Fig. 11.14B. Apparently, both photographs were taken after a long term use of the melt for electrolysis of TiO2 [29]. Fig. 11.14C and D explain the physical and chemical consequences from the anodic formation of the CO2 gas during electrolysis. Replacing the carbon anode with an inert one has several benefits, including increasing the current efficiency, reducing the energy consumption and, overall, making the process far more attractive as the overall reaction would simply be the dissociation of the titanium oxides to titanium and oxygen. However, materials that can act as inert anodes are scarce due to the

Figure 11.14 (A) Photograph of top view of an electrolysis cell after removal of electrodes, showing carbon slug on the surface of the molten salt. (B) Optical micrograph of the carbon slug on the solidified melt of (A) [29]. (C) Cracks (e.g., grain boundaries), strained by growing CO2 bubbles, in the graphite anode propagate and connect, leading to the disintegration of structure (i.e., erosion) and formation of carbon debris in molten salt. (D) Carbon transport from anode to cathode via both the gas and liquid phases during electro-deoxidation of a metal oxide (MOx) in molten CaCl2 [30].

Invention and fundamentals of the FFC Cambridge Process

243

highly corrosive nature of molten salts. The aluminum industry has tried, unsuccessfully for over 100 years, to discover an inert anode which is capable of liberating oxygen in a highly corrosive environment. In the case of titanium extraction, many factors need to be taken into account and these include the physical stability of the anode at high temperatures, electrical conductivity, resistance to attack by the electrolyte, specific ions present in the electrolyte, and the anodic gases, e.g., chlorine gas and oxygen gas, and also the material’s robustness under thermal shock and ease of fabrication. Metals which are generally easy to fabricate, highly electronically conductive and possess very good thermal shock resistance, are mechanically robust, and can easily be fabricated. Most metals are reactive to oxygen at high temperatures, but some protective oxide layers can form preventing further oxidation. For a few metals, notably silver and iridium, the oxides become unstable at elevated temperatures. Iridium has been used to make the anode in molten iron oxide during the electrolytic production of molten iron [31]. According to HSC [32], all silver oxides are unstable above 573 K but, as far as is known, silver has never been investigated as an inert anode in molten salts, likely because of the spontaneous oxide-halide exchange reaction (11.28) below. Ag2O þ CaCl2 ¼ CaO þ 2AgCl, DG (900 C) ¼ 82.678 kJ/mol

(11.28)

The properties of ceramics are well known as having very high melting points, good resistance to oxidation and being inert in extreme environments. However, the vast majority are very poor electrical conductors although some demonstrate semi-conducting properties which allow conduction at high temperature. When a tin oxide anode was used in molten calcium chloride, an insulating layer of calcium stannate (CaSnO3) formed. There is no data on the electrical conductivity of calcium stannate which however gradually raised the resistance of the cell as it formed [33]. Similarly, when conducting titanium carbides were used on the anode, an insulating layer of calcium titanate formed when the oxide to be reduced was titanium dioxide [34]. Cermets are a well-known group of materials which are composed of both ceramics and metals and, in theory, combine the advantages of metals, such as high conductivity with the inert behavior of ceramics. In practice, these materials have not performed well [35]. Apparently, past research findings have demonstrated that ceramics and cermets remain promising, but need further improvement in electronic conductivity and chemical stability in the melt for uses as the anode materials.

244

Extractive Metallurgy of Titanium

Calcium titanate is a possible candidate as an inert anode but has a very low electronic conductivity and doping with ions of a different valence to titanium have been tried but without a great deal of success [36,37]. The maximum conductivity was observed by Iwahara by doping with Fe3þ which substituted for Ti4þ in the structure according to the following overall reaction. CaO þ xTiO2þ x/2 Fe2O3 ¼ CaTi(1-x)FexO(3-x/2) þ þ x/2V2 o þ xh

(11.29)

Unfortunately, the conductivity increase was modest, only up to 0.1 U/ cm at 1000 C [37], which is insufficient if the material is to act as an electrode and pass a meaningful current. In order to overcome this problem, a completely different approach was needed and this was to search for stable compounds that were highly electronically conducting and had a similar structure to calcium titanate. Calcium ruthenate is such a compound with a conductivity of 900 U/cm at 900 C and it also forms a continuous range of solid solutions with calcium titanate. This is important as ruthenium is an expensive and relatively rare element so the amount used needs to be minimised. Ways to reduce the amount needed include using a very thin layer of calcium ruthenate on a conducting substrate or diluting the calcium ruthenate with calcium titanate without dramatically decreasing the conductivity. As mentioned above, both calcium and strontium ruthenates form a complete series of solid solutions with their respective titanates but the electrical properties change. It has been found that metallic conduction exists down to x ¼ 0.50 in CaTi(1ex)RuxO3 or SrTi(1ex)RuxO3 with a gradual decrease in the conductivity but at lower values of x, the conductivity decreases faster with decrease in x to give a value of about 103 U/cm at room temperature when x ¼ 0.1. Fortunately, one of the properties of a semi-conductor is that the conductivity increases with temperature so at the temperatures for electro-deoxidation, about 800 C, conductivity can rise to values of about 10 U/cm for x ¼ 0.1. Experiments showed that, using a CaRuO3 anode in CaCl2 e CaO melts, the rate of oxygen released on the anode tracked the current and, after these experiments which lasted up to 150 h, there was no apparent attack on the anode. In order to reduce the amount of ruthenium in the anode, two approaches were tested with (1) the solid solutions of calcium ruthenate and calcium titanate, and (2) the solid solution coated

Invention and fundamentals of the FFC Cambridge Process

(A)

(B)

(C)

245

(D)

Figure 11.15 Photographs of a CaRuO3 anode (pellet, 25 mm in diameter and 3 mm in thickness) (A) before and (B) after electrolysis. In (C) and (D) are shown, respectively, a CaTiO0.81Ru0.19O3 (solid solution) anode (C) and a CaTiO3 anode coated with thin layer of CaTiO0.72Ru0.28O3 (D) after electrolysis. The electrolysis was carried out in molten CaCl2 at 900 C and 3 V with a cathode of TiO2 (or its mixture with NiO) that was replaced with a new one at about every 20 h [38].

calcium titanate. Similar results were obtained in all these tests [38]. Fig. 11.15 compares these anodes, showing negligible attack after use. Further work was performed by Jiao and his group in Beijing using sintered mixtures of TiO2$RuO2, with very similar results. It was also found that if there was any CaO in the melt, a solid solution of CaRuxTi(1ex)O3 eventually formed [39]. Overall, the anodic testing of calcium ruthenate shows that it may be suitable as an inert anode in calcium chloride containing melts as the experiments showed that when used as an anode it exhibited a low rate of corrosion in melts containing a small amount of calcium oxide, capable of producing oxygen on its surface, and not contaminating the melt. In order to reduce the amount of ruthenium in the anode, solid solutions of calcium ruthenate in calcium titanate were investigated. At low concentrations, the solid solution is a semiconductor with a relatively low conductivity at room temperature but at the temperature of operation, the material is an excellent electronic conductor. The other way of reducing the amount of ruthenium is to coat the solid solution onto a substrate. In this way, the substrate would give the mechanical strength whilst the coating would give the electrical conductivity and corrosion protection. However, longer term trials are necessary to observe whether the materials last for months or years, the time period required by industry.

246

Extractive Metallurgy of Titanium

6. Electro-deoxidation of other metal oxides Although the FFC Cambridge Process was first reported using the electrodeoxidation of TiO2 as an example, it is actually a generic technology for the extraction of almost every metal or semimetal listed in the periodic table from its solid compound, particularly oxide and sulfide. This was partly derived from the preliminary work of Chen and Fray on the cathodic refining of raw copper in molten salts [40]. It can be shown that electrochemical removal of oxygen, sulfur and selenium from raw copper is feasible according to reactions (11.30)e(11.33) and the relevant thermodynamic data. Ca2þ þ 2e ¼ Ca, E (vs. Ca2þ/Ca) ¼ 0 V

(11.30)

/2 O2 þ 2e ¼ O2 E (vs. Ca2þ/Ca) ¼ 2.60 V

(11.31)

/2 S2 þ 2e ¼ S2 E (vs. Ca2þ/Ca) ¼ 2.14 V

(11.32)

1

1

/2 Se2 þ 2e ¼ Se

1

2

E (vs. Ca /Ca) ¼ 1.63 V 2þ

(11.33)

Oxygen, sulfur and selenium are soluble in molten copper, but precipitate out as inclusions in the solidified metal as exemplified in Fig. 11.16A and B for Cu2O which appears in a unique blue color. In the (A)

(B)

(C)

(D)

Figure 11.16 Optical micrographs of a raw copper sample containing 5500 ppm oxygen (and other impurities, see Table 11.2) before (A, B) and after (C, D) cathodic refining in molten CaCl2 at 1180 C and 3.0 V. (B) and (D) are magnifications of the boxed portions in (A) and (C), respectively. The inset between (A) and (B) shows the blue colored Cu2O inclusions in the bronze colored copper.

Invention and fundamentals of the FFC Cambridge Process

247

Table 11.2 Elemental analysis of raw copper before and after cathodic refining at 1160 C for 2.0 h Electrolyzed Electrolyzed Electrolyzed Electrolyzed at 2.6 V in at 2.6 V in at 2.1 V in at 2.1 V in As CaCl2 BaCl2 CaCl2 Impurity received BaCl2

O (ppm) S (ppm) Se (ppm)

5500 10 25

30 Undetectable Undetectable

100 Undetectable Undetectable

50 Undetectable Undetectable

100 Undetectable Undetectable

very initial test, the raw copper sample with various impurities as listed in Table 11.2 was placed at the bottom of a graphite crucible for refining. The graphite crucible acted as both the molten salt container and the cathode (current collector) during electrolysis with a graphite rod anode at 1180 C and 3.0 V for 170 min. The refining result proved the removal of the Cu2O inclusions as shown in Fig. 11.16C and D. In a more detailed study [40], both CaCl2 and BaCl2 were used as the electrolyte, but the latter assisted better refining results as shown in Table 11.2. This difference was attributed to the different thermodynamic tendency of the following oxide-halide exchange reactions (11.34) and (11.35). The more feasible reaction (11.34) means dissolution of CuCl, and hence a lower refining efficiency. Cu2O þ CaCl2 ¼ CaO þ 2CuCl, DG (1160 C) ¼ 6.056 (38.780) kJ/mol

(11.34)

Cu2O þ BaCl2 ¼ BaO þ 2CuCl, DGo (1160 C) ¼ 121.675 (80.046) kJ/mol

(11.35)

Note that the DG values from Version 6.12 of HSC as given above differ from those (in brackets) from the 1994 Version as reported in Ref. [39], but both agree with that reaction (11.34) is more feasible. The successes in the cathodic removal of oxygen, sulfur and selenium from raw copper, and of oxygen from the alpha case on titanium, and the electro-deoxidation of TiO2 contributed to the original patent on the FFC Cambridge Process [41], and encouraged further efforts to electro-deoxidise many other metal oxides, particularly ZrO2 [17,42,43], SiO2 [16,44,45], Cr2O3 [11,46], Ta2O5 [47,48], Nb2O5 [49,50], Tb4O7 [51], Al2O3 [52,53] and Fe2O3 [54,55]. The initial test on deoxidation of ZrO2 was fairly straightforward [42], but further efforts discovered difficulties in relation with the thickness [17] and porosity [43] of the cylindrical ZrO2 pellet. It was found that when the porosity was not high enough, sintering of Zr on the outer layer of the

248

Extractive Metallurgy of Titanium

pellet would make it difficult for ion diffusion through the outer metal layer from the interior unreduced ZrO2 (and CaZrO3) [43]. With the porosity of the ZrO2 pellet being 40%e50%, it was found that the thickness of the porous Zr metal layer formed on top of the unreduced ZrO2/CaZrO3 core reached a maximum of 0.8 mm which means a total thickness of 1.6 mm of the fully metallised pellet [17]. This work then enabled complete metallisation with the obtained porous Zr metal having 1800 ppm oxygen at a satisfactory energy efficiency of 45%. Fig. 11.17A displays the cross section of a partially reduced ZrO2 pellet, showing clearly an unreduced core between the metal layers on both sides. The thickness of the porous metal layer increased as a function of square root of the electrolysis time before reaching a maximum of ca. 0.8 mm as shown in Fig. 11.17B. This is evidence of diffusion controlled growth of the metal layer which however became thinner after the maximum, likely due to sintering of the metal

Figure 11.17 (A) SEM image of the cross section of a partially reduced ZrO2 pellet showing an unreduced core of CaZrO3 in between the outer layers of porous Zr metal. The inset shows the nodular structure of the porous Zr. (B) Correlation between the thickness of the metallised layer in (A) and the electrolysis time (3.0 V, 850 C). (C) A tube of ZrO2 that was fully electro-deoxidised to (D) a consolidated Zr tube [17].

Invention and fundamentals of the FFC Cambridge Process

249

layer. Following these findings from electrolysis of ZrO2 pellets, the fabrication of a consolidated Zr tube from the ZrO2 tube precursor was achieved as shown in Fig. 11.17C and D [17]. It is surprising that the degree of sintering of the produced Zr was higher than that of Ti because the melting point of Zr is 1855 C, higher than that of Ti, 1668 C. This difference could have resulted from the fact that Ti2O3 and CaTi2O4 could both exist in partially reduced TiO2, but the same was not seen in partially reduced ZrO2. Without an intermediate Zr(III) state, reduction of ZrO2 (or CaZrO3) goes to the conducting ZrO phase directly, which means quicker metallisation and hence sintering. It is also worth noting that sintering at the mild electrolysis temperature should have also been assisted by the low oxygen content of the formed metals at the applied reducing potentials. Disregarding sintering, the products from electro-deoxidation of TiO2 and ZrO2 had similar nodular structures as can be seen in Fig. 11.8B and the inset in Fig. 11.17A. However, for electro-deoxidation of Cr2O3 and Fe2O3, it was found in the initial studies that the metals produced were cubic particles as shown in Fig. 11.18A and B, respectively. Also, the current efficiency for electrolysis of these two oxides was found much higher than that for TiO2 and ZrO2. For Cr2O3 reduction, it reached 75%, whilst for Fe2O3, the current efficiency could be over 90%. In terms of interaction with oxygen, both Ti and Zr can dissolve a large amount of oxygen in solid state, but solid Cr and Fe have almost no solubility for oxygen. This difference was considered to be responsible for the very different morphologies observed in these two types of metals, as illustrated in Fig. 11.18D to F according to the preferential crystal growth model [56]. Accordingly, it was expected that for metals with no or low oxygen solubility, the crystal growth should have also started from the nodular morphology, which was indeed observed for Cr [56] and Fe as shown in Fig. 11.18C [54]. Following the preferential growth model, formation of cubic Ti crystallites is also expected in longer electrolysis experiment, which was again experimentally confirmed [56]. In fact, if not considering the oxygen content, the current efficiency for electro-deoxidation of both TiO2 and ZrO2 to the metal phase could be high as well [18], but removal of the dissolved oxygen in the produced metal would take an extra electrolysis time. This step in turn decreased the current efficiency, mainly because of the electronic conductivity of molten CaCl2 as discussed above in relation to Reactions (11.23) to (11.25). Because electronic conduction is more significant at higher cell voltages,

250

Extractive Metallurgy of Titanium

Figure 11.18 SEM images of (A) cubic Cr, (B) cubic Fe and (C) nodular Fe particles produced by electro-deoxidation of the respective metal oxides in molten CaCl2 under conditions indicated [11,54]. Schematic illustration of (D) the two dimensional (2D) preferential crystal growth model, (E) normal metal crystal growth into regular crystallites, and (F) irregular crystal growth resulting from the interruption of oxygen atoms in metals having sufficient solubility for oxygen in solid state [56].

following Ohm’s law, maintaining a minimum cell voltage at later stages of electrolysis would be a cheaper and fairly effective approach to increasing the current efficiency [18,50]. Another example of fast electro-deoxidation with high current efficiency, as shown in Fig. 11.11C, is from SiO2 to Si, and silicon does not dissolve oxygen in solid state. The electro-reduction of solid SiO2 was first reported by the Ito group at Kyoto University based on a pinpoint contacting SiO2 electrode as shown in Fig. 11.19 [44]. In the work, the principle of charge transfer at the “metal/oxide/electrolyte” three phase boundary [12] was further demonstrated. Fig. 11.19A to C show clearly that the electro-reduction of SiO2 (quartz) progressed via the expansion of the “conductor (Mo or Si)/insulator (SiO2)/electrolyte (molten salt)” three phase boundary (same as the 3PI mentioned above, comparing Fig. 11.19 with Figs. 11.10 and 11.11). However, unlike refractory metals, such as Ti, Zr, Ta, and Nb, which are inert at cathodic potentials in molten CaCl2, unwanted reactions could occur between Si and Ca to form various calcium silicides (and other alkali and alkaline earth metal silicides if their ions are present in the molten salt)

Invention and fundamentals of the FFC Cambridge Process

(A)

(B)

(C)

251

(D)

(E)

Figure 11.19 Photographs of (A) a point-contact electrode (Mo wire tip on quartz glass) before and after potentiostatic electrolysis in molten CaCl2 at 0.7 V (vs. Ca/Ca2þ) and 850 C for (B) 1 min and (C) 15 min [44]. SEM images of Si nanowires produced in molten CaCl2 at 900 C by (D) potentiostatic electrolysis of porous pellets of SiO2 nanopowder at 1.2 V (vs. Pt) for 4 h [45], and (E) by constant voltage electrolysis of porous blocks of mixed SiO2/NiO (10:1, mol. ratio) at 1.5 V for 3 h [57,58].

[59,60]. Fortunately, the cathode potentials for the formation of these silicides were found more negative than that for SiO2 reduction to pure Si, but more positive than that for deposition of Ca (or other alkali and alkaline earth metals) as expected from the depolarisation effect of silicide formation. Thus, to avoid formation of silicides, the electrolysis requires a proper control of the cathode potential, for example, by using a reference electrode [14,45,59,60]. Alternatively, Yang and co-workers from the General Research Institute for Nonferrous in Beijing successfully prepared straight Si nanowires from electrolysis of SiO2 in molten CaCl2, using nickel as the catalyst (cf. Fig. 11.19E) [57,58]. Their success was also related with a very low constant cell voltage of 1.5 V for the electrolysis and, possibly more importantly, a large area anode (graphite crucible) that could have minimised the polarisation for discharging the O2 ion to CO2, and hence stabilised the anode potential. This means a stable cathode potential as well at the applied constant cell voltage [61]. The benefits of using a large area anode were first purposely demonstrated in electro-deoxidation of metal oxides, particularly Ta2O5 [61]. Ta2O5 has excellent dielectric properties and is a key electronic capacitor material in computers and mobile phones. It is insulating and hence not used directly, but formed on top of Ta nanoparticles by appropriate anodisation. Therefore, the industrial need is high quality Ta powder. The high melting point of Ta metal (3020 C) means that the Ta particles

252

Extractive Metallurgy of Titanium

formed by electro-deoxidation of Ta2O5 in molten CaCl2 at the mild temperatures of 850e950 C would not be able to sinter significantly. This was confirmed experimentally by several groups [33,47,48]. Fig. 11.20A and B present the SEM images of the Ta2O5 powder and its electrodeoxidation product, the Ta powder [48]. It can be seen that both powders were of nanoparticles, but the electrolytic Ta powder had smaller nodular particles, indicating no or very little sintering. By electro-deoxidation of porous thin pellets of Ta2O5 (1.3e1.5 mm thick, 45% in porosity) via potentiostatic electrolysis for 5 h, the current efficiency could reach 78% with about 2000 ppm oxygen in the produced Ta nanopowder [48]. However, when performing the electrolysis by constant cell voltage, 3.1 V was found to be necessary, indicating polarisations in the cell. Performing potentiostatic electrolysis with varying the area of the graphite anode confirmed that the polarisations were mostly on the graphite anode in the electro-deoxidation cell [61]. Fig. 11.20C and D demonstrate that changing the anode area (367.8 cm2 for the graphite crucible anode, and 35.6 cm2 for the graphite rod anode) did not affect the current-time profile significantly, which was determined by the cathode reactions. However, upon a ten-fold increase in the anode area, both the

Figure 11.20 SEM images of powders of (A) Ta2O5 and (B) Ta prepared by potentiostatic electrolysis of porous thin pellets of Ta2O5 in molten CaCl2 (850 C, 1.3 V vs. Ag/ AgCl, 5 h) [48]. Current-time plots recorded during potentiostatic electrolysis with a graphite (C) rod and (D) crucible anode, and the corresponding (E) cell voltage-time, and (F) anode potential-time plots [61].

Invention and fundamentals of the FFC Cambridge Process

253

cell voltage and anode potential decreased correlatively by about 1.0 V, as shown in Fig. 11.20E and F, corresponding to nearly 30% saving in the electric energy input for the process. The electrolytic Ta powder was tested directly in capacitors and offered capacitance values of industrial expectation, except that its carbon content of 620e1100 ppm exceeded the specifications of several tens of ppm [48]. This is an issue related to the use of a carbon anode as illustrated in Fig. 11.14, and it could be addressed by replacing the carbon anode with an inert anode [33], or using the SOM to isolate the carbon anode [62]. It is interesting to note that under comparable electrolysis conditions, electro-deoxidation of Ta2O5 was found to be faster and more efficient than that of Nb2O5 [48,49,61,63]. This was of course partly because of the use of porous thin pellets of Ta2O5 nano-powder [48], but the fact that Nb has a lower melting temperature than Ta (2469 vs. 3020 C) may also have contributed. In other words, the time needed to remove dissolved oxygen in the larger Nb particles should have been responsible for the lower electrolysis efficiency at constant cell voltages. Indeed, under potentiostatic control, it was found that the electrolysis efficiency could be much improved, but the simultaneously monitored cell voltage decreased in the manner as the decreasing current as shown in Fig. 11.21A and B [50]. Because the cathode potential was constant, the cell voltage variation was a good indication of the anodic polarisation. An interesting innovation from this work was the use of a computer to output the same cell voltage profile obtained from potentiostatic electrolysis, and apply the profile to control electrolysis in the two electrode cell. With this computer assisted control (CAC), the electrolysis produced much improved results. For example, under a constant cell voltage of 3.0 V, electrolysis for 12 h produced Nb with 15,000 ppm oxygen, but under the same conditions, the oxygen level was 3900 ppm in the produced Nb from the CAC electrolysis. The CAC electrolysis also consumed 37.4% less energy than the constant voltage electrolysis as shown in Fig. 11.21C, and produced the same nodular Nb metal as presented in Fig. 11.21D. It can be seen in Fig. 11.21A that the time for the reduction current to reach the background level was gradually reduced with negative shifting the cathode potential (except for 0.85 V at which only partial reduction occurred). The charge passed in this period was found approximately the same as, or below the theoretical value for the complete oxide-to-metal conversion. The inset in Fig. 11.21A shows approximately a linear correlation between the square of the reduction charge and the time of electrolysis, indicating diffusion control [13e15]. More detailed analyses of the

254

Extractive Metallurgy of Titanium

Figure 11.21 (A) Current-time, and (B) cell voltage-time plots recorded during potentiostatic electrolysis of Nb2O5 at the indicated cathode potential in molten CaCl2 at 850 C. (C) Energy consumption profiles of electrolyses under constant voltage control and computer assisted control (CAC). (D) SEM image of the Nb product from CAC electrolysis [50].

electrolysis data revealed that the reduction was more efficient at potentials between 0.65 and 0.15 V. At less positive potentials, e.g., 0.05 and 0.05 V versus Ca/Ca2þ, the initial reduction was too fast, leading to saturation and precipitation of CaO or formation of various calcium niobates according to Reaction (11.36) below. NbOx þ gO2 þ dCa2þ þ 2(d e g)ee # CadNbOy (2  x  2.5, g  d  1, y ¼ x þ g)

(11.36)

This finding is against the initial thought that the electrolysis could have benefited from any Ca metal deposited on the cathode to promote calciothermic reduction according to Reactions (11.30) and (11.37) [7e9,29], in parallel with electrochemical reduction (i.e., electro-deoxidation). Ca2þ þ 2ee # Ca 2Nb2O5 þ 5 Ca ¼ 4Nb þ 5CaO(11.37)

(11.37)

Invention and fundamentals of the FFC Cambridge Process

255

The role of cathodically formed Ca metal has been an interesting research topic [22]. The Ca metal is soluble in molten CaCl2 and can contribute to electronic conduction via Reactions (11.23) to (11.25) as discussed above. Dissolved Ca may also transport through the melt to react with the carbon anode to form CaC2, bringing more carbon debris into the melt as shown in Fig. 11.14A and B. As mentioned above, the FFC Cambridge Process has benefited from the fact that Ca does not react with refractory metals. However, this is not so for Si as mentioned above, and aluminum, Al. It is a common practice in molten salts electrochemistry to use alumina (Al2O3) tubes as sheaths on metal wire electrodes. When using the tube as the cathode sheath in electro-deoxidation experiments, metallic coatings were often observed on the surface of the Al2O3 tube, or even beads at the bottom of the crucible. An attempt to electrolyze a small section of Al2O3 tube at 900 C and 3.1 V for 66 h produced metallic beads in molten CaCl2eNaCl (32.2 mol%). SEM, EDX and ICP analyses of the beads revealed an Al-rich alloy with Ca (1.8 at%), but Ca was present in the product as CaAl4 [52]. In another attempt at 550 C in molten CaCl2e NaCl (equimol.), a porous pellet of Al2O3 powder was electrolyzed at 3.1 V for up to 12 h. It was found that the pellet turned to a gray color on surface. Washing away the solidified salt and melting the reduced pellet produced metallic beads whose XRD pattern agreed with that of Al metal [53]. It should be pointed out that reported studies on electro-deoxidation of Al2O3 are still rare in the literature, but the limited information still confirms that, like the formation calcium silicides in electro-deoxidation of SiO2, calcium aluminides could also form when electro-deoxidising Al2O3 which is challenging also because of the low melting point of Al (660 C) and that Al2O3 (3.95 g/cm3) has a higher density than Al (2.70 g/cm3). Electro-deoxidation at temperatures below the melting point may encounter difficult electron transfer kinetics, whilst the fact that liquid Al can wet Al2O3 and hence fill up the pores of a porous Al2O3 cathode means also ion diffusion challenges. On the other hand, Al is already industrially produced by electrolysis of dissolved Al2O3 in molten salts at high efficiency, which obviously discourages investments and research effort in electro-deoxidation of Al2O3 to make Al metal at large scales. Another Ca related challenge is to electro-deoxidise the oxides of rare earth metals, particularly those heavy ones, because their reduction potentials are close to, or even more negative than that for Ca deposition. For example, it can be derived from thermodynamic data that the reduction

256

Extractive Metallurgy of Titanium

Figure 11.22 (A) Calculated reduction potentials of Dy2O3, Tb2O3 and Y2O3. (B) Cyclic voltammograms of a bare metallic (Mo) cavity electrode (MCE, see the inserted SEM image), and the MCE loaded with Tb4O7 (which was converted to Tb2O3). (C) The cross section of a partially reduced Tb2O3 pellet. (D) Variation of the metal layer thickness of a partially reduced Tb2O3 pellet. The inset in (D) shows the cross section of a partially reduced Tb2O3 pellet with a thick layer of deposited Ca [51].

potentials of Dy2O3, Tb2O3 and Y2O3 are very close to, or more negative than that for Ca deposition as shown in Fig. 11.22A [32]. These predictions agree broadly with cyclic voltammetry of these oxides, as exemplified in Fig. 11.22B for a metallic cavity electrode (MCE) with or without Tb4O7 (actually Tb2O3 as explained below) [51]. It can be seen that current peaks c1 and a1 appear on both the CVs, which are evidence of Ca deposition and re-dissolution. On the CV of Tb2O3, peaks c2 and a2 appear, and must have resulted from the redox reactions of the oxide. The important conclusion from these two CVs in Fig. 11.22B is that the electro-deposited Ca contributed very little, if any, to the reduction of Tb2O3. This is because the Ca re-dissolution peak a1 is the same in current with or without Tb2O3 on the electrode. In other words, very little or no Ca that was deposited during the negative potential scan reacted with Tb2O3 on the electrode, but the reduction of Tb2O3 proceeded dominantly or completely electrochemically.

Invention and fundamentals of the FFC Cambridge Process

257

Obviously, the MCE could only load a very small amount of the oxide powder (a few tens of micrograms), which means the influences of electron and ion conduction through the oxide were absent or minimised. To study such influences, pellets of Tb4O7 were sintered in argon and decomposed to Tb2O3 (as confirmed by XRD analysis), and were electro-deoxidised at constant cell voltages from 3.1 to 4.3 V. Under visual inspection, partially deoxidised pellets all showed a metallised surface enclosing an unreduced core, as shown in Fig. 11.22C. Using the increase of the thickness of the metallised surface layer as an indication of the reduction rate, it was found that the reduction became faster with increasing the cell voltage. Fig. 11.22D plots the thickness of the surface metal layer as a function of the cell voltages. It can be seen that for the same electrolysis time, the metal layer thickness increased almost linearly with the cell voltage from 3.1 until about 3.8 V when the reduction rate plateaued. The linear increase of the reduction rate with the cell voltage cannot be explained by electrocalciothermic reduction as represented by Reactions (11.30) and (11.37), replacing Nb2O5 by Tb2O3. This is because once enough deposited Ca has formed in its own phase on the cathode, the Ca activity becomes constant. This in turn means that the reduction of Tb2O3 should have proceeded at the same rate, independent of the cell voltage, which is against the experimental observations. The plateau on each of the two plots in Fig. 11.22D was found to have resulted from a large amount of Ca deposited on, and covered up the surface of the oxide pellet as shown in the inset of Fig. 11.22D, impeding further increase of the reduction rate. These findings and analyses demonstrate that electro-deoxidation of Tb2O3 could only occur at potentials near or more negative than that for Ca deposition. However, the deposited Ca metal could have only minor contribution, if any, to the reduction of Tb2O3. This understanding helped designing the cell and process for electro-deoxidation of Dy2O3 and Y2O3. Particularly, the success in electro-deoxidation of Y2O3 to Y is strong evidence supporting electrochemical reduction, because Y2O3 is more stable than CaO.

7. Electro-desulfidation of metal sulfides As discussed above on cathodic refining of impure copper, the FFC Cambridge Process is also applicable to extract metals from their sulfides. There are many sulfide-based minerals of transition metals, such as NiS (Millerite), Cu2S (Chalcocite)/CuS (Covllite), MoS2 (Molybdenite), FeS2 (Pyrite) CuFeS2 (Chalcopyrite), whilst sulfidation is an important cause for

258

(A)

(D)

Extractive Metallurgy of Titanium

(B)

(C)

(E1)

(E2)

(E3)

(E4)

Figure 11.23 SEM images of powders of (A) Mo, (B) Cu (inset: as-prepared copper pellet from electrolysis of Cu2S), and (C) W (inset: EDX spectrum of the W powder in the SEM image) obtained from electrolysis of solid MoS2 in molten CaCl2 [64], CusS in molten equimolar NaCleCaCl2 [66] and WS2 in molten equimolar NaCleKCl [67]. (D) A typical XRD pattern of the solidified product collected from the anode gases during electrolysis of metal sulfides in molten salts [64]. (E) Photographs of bottom (E1, E3) and side views (E2, E4) of a graphite rod anode before (E1, E2) and after (E3, E4) repeated uses for electrolysis of WS2 (60 g in total) in molten NaCleKCl at 2.7 V and 700 C for 60 h [67].

steel corrosion. An interesting current industrial process is to convert MoS2 to MoO3 with a large amount of SO2 as the by-product, and reduce the oxide by H2 or Al to the Mo metal [64]. Therefore, an effective and affordable process for direct desulfidation of sulfide minerals to the respective metals is of both scientific curiosity and commercial desire [65]. It has been reported that MoS2 [64], CuS [66] and WS2 [67] could all be electro-desulfidised to the respective metals in molten salts as shown in Fig. 11.23A and B. Although both electro-deoxidation and electrodesulfidation follow the same principle of electron transfer induced reduction and ionisation process, the practical aspects were found quite different.

Invention and fundamentals of the FFC Cambridge Process

259

Firstly, most stable metal oxides are poor electronic conductors, but metal sulfides often have metallic lustres and are comparable to metals in many other aspects, such as high conductivity. Also, metal sulfides are mostly less stable then their oxide counterparts, whilst sulfur, unlike oxygen, does not dissolve in most, if not all, solid metals. These properties of sulfides and sulfur are beneficial to increasing the rate of sulfide electrolysis. Indeed, as indicated in Fig. 11.23A and C, the cathodic conversion of sulfide to metal would usually take only 2e3 h. Secondly, sulfur has a boiling temperature of 444.7 C, and is not as reactive as oxygen with carbon as indicated by Reactions (11.38) to (11.41) below [65]. Considering reaction kinetics, there should not be significant reactive interactions between sulfur and carbon. C þ S2 ¼ CS2 DG (900 C) ¼ 18.5 kJ/mol

(11.38)

C þ 1/2 S2 ¼ CS DG (900 C) ¼ 109 kJ/mol

(11.39)

C þ O2 ¼ CO2 DG (900 C) ¼ 396 kJ/mol

(11.40)

C þ 1/2 O2 ¼ CO DG (900 C) ¼ 215 kJ/mol

(11.41)

Indeed, it was observed that during electro-desulfidation in molten salts, the discharge of S2 ions at the graphite anode produced the S2 gas which then condensed and solidified in a collection vessel. Fig. 11.23D presents the XRD pattern of such collected anodic product, confirming it to be indeed sulfur. The absence of significant reactions between sulfur and carbon is further confirmed by the fact that the graphite anode functioned as a true inert anode. Fig. 11.23E1 e E4. compare the photographs of a graphite rod anode before and after 60 h of electro-desulfidation in molten NaCleKCl (1:1 M ratio), showing that no change had occurred to the very detailed surface features [67]. Thirdly, O2 ions can only dissolve in a few commonly known molten chloride salts, such as LiCl, CaCl2 and BaCl2. However, S2 ions are soluble in many molten chloride salts, such as NaCl and KCl in which electrodesulfidation proceeds easily, but deoxidation would not occur. This is again beneficial because it was found that in CaCl2 based molten salts, precipitation of CaS2 in the cathode could slow the desulfidation and was also difficult to remove by washing in water. However, using NaCl and/or KCl would not encounter such problems [67]. Also, electronic conduction through either molten NaCl or KCl, or their mixture is much less significant as in molten CaCl2, making electrolysis more efficient. For example, the Wuhan/Nottingham team reported that electrolysis of 1.7 g of WS2 at

260

Extractive Metallurgy of Titanium

2.7 V and 700 C with a graphite rod anode proceeded to producing 99 wt% pure W nanopowder in 35 min with a high current efficiency of 94%, whilst allowing the electrolysis to last for 2 h led to a product with undetectable S by EDX [67]. With all these advantages in electro-desulfidation, one cannot help to think extraction of Ti from its sulfides, e.g., TiS2, Ti2S3 and/or TiS. TiS2 has a metallic golden yellow color, a layered crystal structure, and metallic conductivity. It was one of the early tested positive electrode materials for lithium intercalation [68], and may therefore be an alternative candidate to TiO2 for making Ti by electro-desulfidation. Thermodynamically, electrolysis of TiS2 requires a lower decomposition voltage than that of TiO2 disregarding the anodic formation of CO and/or CO2. TiS2 ¼ Ti þ S2 DE (900 C) ¼ 0.864 V

(11.42)

TiO2 ¼ Ti þ O2 DE (900 C) ¼ 1.897 V

(11.43)

A recent report from Suzuki’s group at Hokkaido University described electro-calciothermic reduction of TiS2 in molten CaCl2 with added CaS2 (0.5 mol%) and a carbon anode. When the electrolysis was carried out at 3.0 V and 900 C, and lasted for just under 7 h, Ti was produced, containing 0.01 wt% S and 0.96 wt% O. The charge passed in the electrolysis was 400% of the theoretical value for complete reduction [69]. It is worth mentioning that the following sulfurisation, i.e., Reaction (11.44), was claimed to be successful for the preparation of TiS2 by reacting CS2 and TiO2 with added S in a pressurised reactor at 800 C [70], whilst Reaction (11.38) was proposed for re-generation of CS2 [69]. TiO2 þ CS2 ¼ TiS2 þ CO2 DG (800 C) ¼ 30.304 kJ/mol (11.44) TiS2 þ Ca ¼ Ti þ CaS2 (Ca2þ þ S2, in molten salt)

(11.45)

Ca2þ þ 2e ¼ Ca (at cathode)

(11.30)

S2 ¼ S2 þ 2e (at carbon anode)

(11.46)

S2 þ C ¼ CS2 (at carbon anode)

(11.38)

However, Reaction (11.44) is thermodynamically unfavourable at 800 C according to HSC [32], whilst to realise Reaction (11.38) would also be challenging on a graphite anode in molten salts, considering Fig. 11.23D and E. Of course, if carrying out Reaction (11.38) in a separate reactor with carbon in an appropriate powdery form, the conversion may succeed,

Invention and fundamentals of the FFC Cambridge Process

261

considering that the produced CS2 gas (CS2 boiling point: 46.3 C) can be easily removed from the reactor. Alternatively, the sum of Reactions (11.44) and (11.38) leads to a more direct route below [71]. TiO2 þ S2 þ C ¼ TiS2 þ CO2 DG (1200 C) ¼ 7.541 kJ/mol (11.47) TiO2 þ S2 þ 2C ¼ TiS2 þ 2CO DG (1200 C) ¼ 79.742 kJ/mol (11.48) Reaction (11.47) is also thermodynamically unfavourable at temperatures below 1690 C, but reaction (11.48) has a negative Gibbs free energy change at temperatures higher than 775 C [32]. However, the realisation of Reaction (11.48) is expected not to be straightforward because Reaction (11.38) may happen first with CS2 having a very low boiling point, and a pressurised reactor is also necessary. Nevertheless, the expected ease of electro-desulfidation of TiS2 is attractive to further studies. The Suzuki group also reported electro-calciothermic reduction of V3S4 in molten CaCl2 with added CaS (2.0 mol%). Upon supplying 4.3 times of the theoretical charge at 3.0 V, they obtained a V sample containing 3390 ppm O and 210 ppm S [72]. They also found that with increasing the CaS content in the melt, the electrolysis current increased, but the S and O contents in the product increased as well. In addition, they compared the electrolytic reduction of V2O3 and V2S3 under similar electrolysis conditions and, as expected, found that carbon contamination of the molten salt was obvious in the case of the former, but absent in the latter [72].

8. Electro-deoxidation of mixed metal oxides Removal of oxygen from the thin oxide scale on the Tie6Ale4V alloy plate as shown in Fig. 11.2A indicated the possibility to electro-deoxidise mixed metal oxides because the dark brown color of the scale indicated the presence of oxides of vanadium and low valence titanium. To confirm this expectation, the mixture of TiO2, Al2O3 and V2O3 powders giving the final atomic ratio of the Tie6Ale4V alloy was pressed and sintered into pellets, and electrolyzed under the same conditions as that for electrolysis of TiO2. Fig. 11.24 presents the photographs (A, B, C) and electron micrographs of pellets of pressed and sintered mixed TiO2eAl2O3eV2O3 powders before and after electro-deoxidation. It was noted that the slip-cast pellet of mixed TiO2eAl2O3eV2O3 was light green in color, but changed to dark brown, apparently resulting from the oxidation of V(III) to V(V) upon sintering at 950 C in air, see Fig. 11.24A. Also, the original powder particles were spherical for TiO2 and

262

(A)

Extractive Metallurgy of Titanium

(B)

(C)

(D)

(E)

(F)

(G)

Figure 11.24 (AeC) Photographs of (A) a slip-cast and sintered (950 C) pellet of mixed TiO2, V2O3 and Al2O3 powders, (B) an as-reduced and washed (to clean away the solidified salt), and (C) the reduced and surface polished version of TiO2eAl2O3eV2O3 pellets similar to that in (A). (D) SEM image of particles in the sintered slip-cast TiO2eAl2O3eV2O3 pellet similar to that in (A). (E) Low and (F) high magnification backscattered SEM images of the polished cross-section, and (F) the polished and etched cross-section of Tie6Ale4V pellets similar to that in (B). Electrolysis: molten CaCl2, 3.1 V, 950 C, 12 h.

Al2O3, and irregular for V2O3, but these obviously reacted and changed to short rods after sintering in air, as show in Fig. 11.24D. The electrolysis produced, somehow surprisingly, very well sintered and strong metallic pellets whose surface turned to metallic luster after slight polishing (see Fig. 11.24B and C). To investigate what had happened to the interior of the pellets, a hacksaw had to be used to cut the pellet into two halves, revealing completely metallised and highly densified cross sections. SEM inspections of the polished cross section of the pellet demonstrated again high densification (see Fig. 11.24E), although closed pores could be seen scattered in the surface of the polished cross section (see Fig. 11.24E to G). Most of these pores were a few mm in size, but rarely larger ones up to a couple of hundreds mm could also be found (see Fig. 11.24D). The Tie6Ale4V alloy

Invention and fundamentals of the FFC Cambridge Process

263

Table 11.3 EDX analyses of the Tie6Ale4V alloy from electro-deoxidation of porous pellets of mixed powders TiO2-A2O3eV2O3 in molten CaCl2 at 3.0 V and 950 C for 12 h. Data in brackets are measured from a commercial standard Tie6Ale4V alloy sample. Element Ti (wt%) Al (wt%) V (wt%)

Bright phase composition Dark phase composition

87.08 (86.14) 90.07 (91.85)

3.10 (3.74) 6.03 (5.59)

9.82 (10.12) 2.78 (2.19)

has both the alpha and beta phases, and this was featured by the bright and dark phases as shown in Fig. 11.24E and F. Table 11.3 presents EDX detected elemental compositions in a good agreement with those of a commercial sample. According to the findings from this work, several interesting points arise for discussion. Firstly, it appears that electro-deoxidation of mixed oxides to make the alloy is quicker and more efficient than that of TiO2 alone to make the pure Ti metal. This might have been benefited from two additional processes occurring when reducing the mixed oxides. Alloy formation is spontaneous with a negative Gibbs free energy change, which in turn helps shift negatively the overall Gibbs free energy change of the deoxidation process. In addition, as shown in Fig. 11.24D, co-oxides must have formed after sintering the mixed oxide precursor, which may help not only uniform mixing at atomic level, but also avoidance of in situ perovskitisation and hence acceleration of the process. Secondly, the Tie6Ale4V alloy is industrially produced by the melting-solidification process, i.e., arc melting the individual metals at high temperatures, and then cooling the melt down to the solid phase. However, in electro-deoxidation, the elements have never been molten. In terms of phase transition, in the melting-solidification process, the phase change goes from the b-phase at high temperatures to the aþb-phase at low temperatures, but it occurs from the a-phase to the aþb-phase in electrodeoxidation, as illustrated in Fig. 11.25. Such a difference in the order of phase formation could in principle alter the phase structure of each grain in, and properties of the alloy. The consequence is still unknown, and worth further investigation. Thirdly, the well maintained shape of the oxide precursor in the highly densified alloy product from electro-deoxidation suggests the possibility of near-net-shape manufacturing of engineering artifacts. It is obviously much more economical to manufacture an oxide precursor in various shape, and

264

Extractive Metallurgy of Titanium

Figure 11.25 Schematic illustration of the Tie6Ale4V alloy formation from the b phase to the aþb phase via the industrial melting-solidification process (blue dashed line), and from the a phase to the aþb phase via the electro-deoxidation process (red dashed line).

then metallise it. An early example of near-net-shape fabrication by electrodeoxidation is shown in Fig. 11.17C and D for making the Zr tube from the ZrO2 tube. Similarly, the recent effort from Hu and co-workers in Nottingham has shown a very promising prospect. By slip-casting the mixed oxides into various shapes, including hollow structures, they have successfully metallised the precursor as shown in Fig. 11.26. In all these cases, highly densified alloy samples were obtained [73]. It should be mentioned that in comparison with Fig. 11.24A to C, more significant shrinkage occurred to the hollow structured samples. In the case of the hollow sphere, the outer diameter was 15.6 mm for the oxide precursor, but 9.4 mm for the metallised product. Although there was insignificant (A)

(B)

(D)

(E)

(C)

(F)

(G)

Figure 11.26 Photographs of (A) slip-cast hollow sphere of mixed TiO2-3A2O3e2V2O3, (B) metallised version of (A), and cross section of (B), and (D) slip-cast miniature golf club head of TiO2-3A2O3e2V2O3, (E) metallised version of (D), and (F) cross section of (E) [73], and (G) a commercial Ti golf club head.

Invention and fundamentals of the FFC Cambridge Process

265

shape distortion as shown in both Figs. 11.24 and 11.26, further studies are needed to confirm if metallisation can still maintain other shapes, particularly those with high aspect ratios. Last, but not the least, Ti metal is light (density: 4.506 g/cm3) which means possible composition segregation when alloying with a heavy metal via melting and solidification. Without melting, electro-deoxidation has been proven very successful in alloying Ti uniformly with all tested heavy metals, such as Ti-Mo [74], TieW [75], Ti-Ni [76], Ti-Fe [77], Ti-Ta [78], and TieNb [17]. With elements of comparable density or lighter, such as Al, Si and Ge, intermetallic compounds are more likely to form [79e81]. In addition to Tie6Ale4V as discussed above, three or more element alloys were also prepared by electro-deoxidation, for example, Tie5Tae2Nb [82], Ti-Nb-Ta-Zr [78,83] and TieNbeTaeZr-Hf [83]. The latter two alloys are also known as the Gum metal [84], or high entropy alloy [85], having low elastic modulus, but high strength, yield strain and ductility that are desirable for many applications, such as vehicle components and life supporting implants. A special type of structural and functional materials comprising titanium belongs to the metal carbide or nitride based MAX phase (Mnþ1AXn, n ¼ 1 to 3, M ¼ early transition metal, A ¼ A-group element, X ¼ carbon or nitrogen) [86,87]. Making metal carbides via electro-deoxidation differs from all other cases mentioned above in that carbon oxides are all gases and hence it is solid carbon that is available for use. Using carbon directly could be, in principle, beneficial to electro-deoxidation as carbon is conducting, and there is no need for an extra amount of charge (energy) to reduce carbon oxides. However, in practice, solid carbon can react with oxygen in air, or most metal oxides in Ar or vacuum, and be lost as CO2 or CO during sintering of the metal oxide precursors at elevated temperatures. Therefore, pressed pellets of mixed carbon and metal oxide powders without sintering [86], or after annealing at a low temperature (150 C) [87] that should not cause sintering were electrolyzed directly in molten salts. It was found that electro-deoxidation of the TiO2eAl2O3eC mixture experienced in-situ formation of both CaTiO3 and various co-oxides and oxychlorides of Ca and Al, e.g., Ca12Al14O33 and Ca12Al14O32Cl2, but eventually led to the targeted product, Ti3AlC2 or Ti2AlC, both of which are layer structured as shown in Fig. 11.27A and B. A reasonably good correlation was confirmed between the compositions of the precursor and final product [86,87].

266

(A)

Extractive Metallurgy of Titanium

(B)

Figure 11.27 SEM images of (A) Ti3AlC2 and (B) Ti2AlC prepared by electrodeoxidation of TiO2eAl2O3eC with molar ratio of Ti:Al:C ¼ 3:1.8:1.8 and 2:1.5:0.9, respectively [87].

However, loss of Al could be derived from the compositions of the precursor (Ti:Al ¼ 3.0:1.8) and product (Ti:Al ¼ 3:1), which was attributed to the low melting point of Al [87]. An implied cause from the work is the formation of calcium aluminides (CaAl2 and CaAl4). On the other hand, it can be noted that the carbon percentage in the precursor was slightly lower than that in the final product, which was to avoid formation of TiC during or at the end of electrolysis [87]. However, the excess could serve to compensate for any loss of C due to its reaction with the metal oxides if sintering is necessary for making a workable precursor. Nevertheless, by comparing the outcomes from electrolysis of TiO2, TiO2eC, TiO2eAl2O3 and TiO2eAl2O3eC under comparable conditions (950 C, 3.1 V, 4 h, molten CaCl2), it was found that the oxygen contents in the products were in the order TiO2eC (15,000 ppm) > TiO2eAl2O3 (10,500 ppm) > TiO2 (8000 ppm) > TiO2eAl2O3eC (4000 ppm). Another interesting finding is that graphite powder seemed to be most effective to help formation of Ti3AlC2 with the lowest oxygen content, in comparison with activated carbon and carbon nanotubes [87]. The FFC Cambridge Process has been proven to be a very generic and versatile method for making many other alloys and intermetallic compounds. Preparation of Nb-based intermetallics by electro-deoxidation was reported for superconductor applications in 2002 [88], whilst the high strain (up to 5%) ferromagnetic shape memory alloy, Ni2MnGa, was also successfully prepared by electro-deoxidation in molten CaCl2 [89]. The electrolysis was carried out at 3.0 V and 900 C in molten CaCl2 for 26 h, and it was highly effective in metallisation of the oxide precursor, see Fig. 11.28C to E. Particularly, EDX detected no oxygen which is not surprising because oxygen does not dissolve in these metals in solid state.

Invention and fundamentals of the FFC Cambridge Process

267

Figure 11.28 Flowcharts of (A) current industrial and (B) the FFC processes for making Ni2MnGa. (C) SEM image of the interior of (D) the electro-deoxidised pellet (of mixed NiOeMnO2-Ga2O3) whose (E) powder XRD pattern and EDX analysis confirm the successful reduction in terms of crystal structure and elemental composition [89].

The produced porous metallic pellet could be easily crashed and ground into powder, and was found to be ferromagnetic. In terms of process engineering, Fig. 11.28A and B compares the current industrial and the FFC Cambridge processes for making the Ni2MnGa alloy. Further work on preparation of Nb-based superconductor materials from Nb2O5 mixing with SnO2 or TiO2 was successful [90]. In molten CaCl2eNaCl (eutectic mixture) at 950 C, electrolysis of the porous Nb2O5eTiO2 and Nb2O5eSnO2 pellets was carried out using either a graphite rod anode or a YSZ membrane anode (cf. Fig. 11.13). It was observed that at a cell voltage of 3.0 V or 3.1 V, the electro-deoxidation of mixed oxide pellets also proceeded via the 3PI propagation as shown in Fig. 11.29A. Removal of oxygen continued with electrolysis time. For Nb2O5eTiO2, oxygen removal achieved 99.7% (1020 ppm O) after electrolysis for 25 h, and 99.96% (140 ppm O) in 88.43 h. However, the

268

Extractive Metallurgy of Titanium

electrolysis of Nb2O5eSnO2 seemed to be slower as the oxygen content was still 8700 ppm after 25 h, and 4490 ppm after 78 h. The finally metallised pellets consisted of slightly sintered micro-nodules, see Fig. 11.29B, and performed satisfactorily in superconductivity analysis as shown Fig. 11.29C. A large group of functional alloys and intermetallics have been developed for hydrogen storage. The attempt to electro-deoxidise porous pellets of ground natural ilmenite, i.e., FeTiO3, produced successfully the ferrotitanium alloy, FeTi, see Fig. 11.30A and B, an AB type hydrogen storage alloys (HSAs). It was found that the deoxidation started by producing Fe and CaTiO3, and the latter was then reduced on the preformed Fe to produce the FeTi alloy. The energy consumption was as low as 14.4 kWh/ kg-FeTi, which compares favourably with the energy consumption for producing the individual metals in the industry, i.e., 45e55 kWh/Ti-kg and 4e6 kWh/kg-steel. It can be anticipated that by grinding the natural ilmenite into finer and more uniform powders, electro-deoxidation could be more efficient. The as-produced FeTi alloy powder had a similar nodular morphology as pure Ti from deoxidising TiO2, but it was not as good as desired for electrochemical hydrogen storage. A significant improvement, as shown in Fig. 11.30C, was achieved by simply mixing the natural ilmenite power with a small amount of NiO (4.46 wt%), and then electro-deoxidise the mixture. This Ni doped FeTi alloy exhibited an initial discharging capacity that was about three times as large as the FeTi alone. However, as can be

Figure 11.29 (A) Photograph of partially electro-deoxidised pellets of mixed Nb2O5 and SnO2. (B) SEM image of product from electro-deoxidation of mixed Nb2O5. (C) Dependence of AC susceptibility of indicated electro-deoxidation products on temperature [90].

Invention and fundamentals of the FFC Cambridge Process

269

Figure 11.30 SEM images of (A) interior of a pressed porous pellet of the ground powder of natural ilmenite, and (B) the nodular powder of FeTi alloy produced by electrolysis of the porous pellet of natural ilmenite in molten CaCl2 at 900 C and 3.0 V for 12 h. (C) Cycling capacity profiles of electrochemical charging-discharging of the FeTi powder of (B) and the FeTiNi powder prepared by electro-deoxidation of natural ilmenite in (A) with added 4.46 wt% NiO [91].

seen in Fig. 11.29C, the cycling stability of the Ni doped FeTi still needs further improvement. It can be anticipated that by further fine-tuning the composition of the oxide precursor, e.g., mixing ilmenite with FeO and NiO so that the produced FeTiNi alloy will have the composition satisfying Fe ¼ Ti þ Ni, better performance for hydrogen storage should be achievable. Such an effort is worth making because the abundant availability of ilmenite with the simple and clean electro-deoxidation process can make very cheap FeTi-based hydrogen storage alloys for large scale applications. There are of course other better HSAs, such as the well-known La-based AB5 [92] and Zr-based AB2 [93] alloys. The work on electrodeoxidation of mixed La2O3 and NiO with or without other added oxides revealed several challenges. Firstly, La2O3 is very hygroscopic and it was impossible to prepare the oxide precursor in air. Also, La2O3 can react with CaCl2 to produce LaOCl which would destabilise the oxide precursor cathode in the molten salt. Further, unlike those main group elements and transition metals, rare earth metals are highly reactive and the electrolytic product must be washed with appropriate organic solvents to remove the solidified salts. An effective solution was proposed and demonstrated to be highly successful. This is to pre-compound La2O3 with NiO to form cooxides at 900e1400 C, such as LaNiO3 (<1000 C), La4Ni3O10, La3Ni2O7 (1100e1250 C), and/or La2NiO4 (>1300 C). The relevant reactions all involved O2 as shown by Reaction (11.49) as an example. 3La2O3 þ 4NiO þ 0.5O2 / 2La3Ni2O7

(11.49)

270

Extractive Metallurgy of Titanium

It was found that particles in all the compounded oxides or co-oxides were more crystalline and much larger than the parent oxide particles as shown in Fig. 11.31A and B. The co-oxides were no longer hygroscopic, and remained stable in molten CaCl2. Following this strategy, electrodeoxidation of mixed La2O3 and NiO (La:Ni ¼ 1:5, atomic ratio), mixed La2O3, NiO and Co2O3 (La:Ni:Co ¼ 1:4:1), and a more complicated mixture of oxides of the rare earth mischmetal (Mm) and Co, Mn, and Al was all successful in producing the respective alloys whose micrographs were typical of that in Fig. 11.31C. More importantly, electrolysis of the co-oxide was more efficient than that of TiO2, whilst the energy consumption was less than 5 kWh/kg-LaNi5. Tests of the as-produced alloys in 6.0 mol/L KOH for electrochemical hydrogen storage led to an initial

Figure 11.31 SEM images of (A) as-pressed ball-milled mixture of NiO and La2O3 (La:Ni ¼ 1:5 at. ratio) and hydrated to La(OH)3 in air, (B) after sintering at 1200 C for 2 h in air, and (C) after electrolysis at 3.1 V in molten CaCl2 at 850 C for 6 h, and washing in dimethylsulfoxide. (D) Capacity profiles from electrochemical charging-discharging cycling (60 mA/g, 1.45 V/0.9 V) of the indicated alloys. (E) Illustration of current industrial and FFC Cambridge processes for making the LaNi5 powder [92].

Invention and fundamentals of the FFC Cambridge Process

271

discharging capacity being comparable to the theoretical expectation, and satisfactory cycling stability, see Fig. 11.31D. Based on these findings, a comparison is given in Fig. 11.30E between the current industrial and the FFC Cambridge processes for making the LaeNi based hydrogen storage alloys [92]. The success in making the La-based AB5 HSAs encouraged further efforts to electro-deoxidise mixed ZrO2 and Cr2O3 with or without added other oxides to make the Zr-based AB2 HSAs [93]. Electrolysis of porous pellets of mixed ZrO2eCr2O3 (1:1 mol ratio) at 3.1 V and 900 C in molten CaCl2 for 12 h produced the ZrCr2 in a great stoichiometric agreement. It was found that during electrolysis, Cr2O3 was reduced first to Cr along with the formation of CaaZrOb which was later reduced on the formed Cr to produce the ZrCr2. The experimental findings suggested cathodic reactions as follows, Cr2O3 þ 6ee ¼ 2Cr þ 3O2

(11.50)

ZrO2 þ aCa2þ þ gO2 þ 2(a e g)ee ¼ CaaZrOb

(11.51)

CaaZrOb þ 2(b e a)ee ¼ Zr þ bO2 þ aCa2þ

(11.52)

2Cr þ CaaZrOb þ 2(b e a)ee ¼ ZrCr2 þ aCa2þ þ bO2 (11.53) Zr þ 2Cr ¼ ZrCr2

(11.54)

where 0  g  a  1, b ¼ g þ 2. Note that Reaction (11.51) is a general form of perovskitisation, which is a chemical reaction for g ¼ a, but an electrochemical one for g < a [10]. However, it was found that in situ perovskitisation seemed to have a much smaller impact on the electrolysis of mixed ZrO2 and Cr2O3. This was possibly because the cathode was partly made of ZrO2, and the formed perovskites could only occupy part of the pores in the cathode, exerting a smaller effect on electrolysis [93]. ZrCr2 is known to be difficult to activate for hydrogen storage, and the solution is to add the third or more elements to help the activation. Using Ni to replace a portion of Cr seems to be often effective, whilst replacing parts of Zr by Ti and Cr by V and Ni were also studied [93]. The work proved electro-deoxidation to be highly effective in producing multi-element alloys as shown in Fig. 11.32 in which the EDX revealed compositions are compared with the targets. For energy consumption, it was 15.0 kWh/kgalloy to produce the five-element alloy, Zr0.5Ti0.5V0.5Cr0.2Ni1.3, with 2700 ppm O. Again, when tested for electrochemical hydrogen storage, these as-produced electrolytic Zr-based alloy powders performed to the expectation.

272

(A)

Extractive Metallurgy of Titanium

(B)

Figure 11.32 SEM images of (A) ZrCr2 and (B) Zr0.5Ti0.5V0.5Cr0.2Ni1.3 prepared by electro-deoxidation of the mixtures of the respective metal oxides. Data presented in each image are EDX determined elemental compositions (at.%) and those in brackets are targeted compositions [93].

It should be mentioned that in most studies on the FFC Cambridge Process, constant cell voltage electrolysis and potentiostatic electrolysis are the common methods used. This is mainly because the need to control the cathode reaction for electro-deoxidation without invoking unwanted side reactions, such as electro-deposition of Ca, or formation of Ca containing compound, such as calcium silicides and calcium aluminides as mentioned above. However, in terms of process engineering, the current response to these two control methods is typical of that shown in Figs. 11.5, 11.20 and 11.21. Obviously, such varying/decreasing current profiles are not convenient to use for assessing the rate of production. To control the production rate, the electrolysis industry uses the constant current control for large scale electrolysis to produce, for example, aluminum. In addition, changing the control from cell voltage or cathode potential to the cell current may bring about interesting information in relation to the progress of electro-deoxidation of the cathode. In a recent study, electro-deoxidation of the sintered mixture of cobalt and chromium oxides, Co3O4eCoCr2O4 or CoxCryO4 (1  x  3, 0  y  2), was carried out under constant current to produce a Co-Cr alloy of 30 wt.% Cr. The variations of the cathode and anode potentials and the cell voltage were recorded against the electrolysis time, as shown in Fig. 11.33A. It can be noted that the anode potential remained almost constant, suggesting the same reaction (of discharging of the O2 ion) as expected. The cathode potential went through four plateaus, indicating four different reactions occurring at the cathode. Terminating the electrolysis and removing the cathode for XRD analysis at points labeled by A,

Invention and fundamentals of the FFC Cambridge Process

273

Figure 11.33 (A) Plots of cathode and anode potentials, and cell voltage against the time of electrolysis of CoxCryO4 in molten CaCl2 at 850 C and a constant current of 0.5 A. (B) Cathode potential variation against electrolysis time at indicated constant currents. The electrode potential was measured against a graphite pseudo-reference electrode [94].

B, C and D in Fig. 11.33A, and also checking the thermodynamic data suggested the following cathodic reactions in association with each plateau as indicated in brackets. Co3O4 ¼ 3CoO þ 2O2 CoO þ 2ee ¼ Co þ O2 (Plateau A) CoCr2O4 þ Ca2þ þ 2ee ¼ CaCr2O4 þ Co (Plateau B) CaCr2O4 þ 6ee ¼ 2Cr þ Ca2þ þ 4O2 (Plateau C)

(11.55) (11.56) (11.57) (11.58)

Ca2þ þ 2ee ¼ Ca (Plateau D)

(11.59)

The four plateaus were more distinguishable at smaller currents as shown by the cathode potential profile of 0.1 A in Fig. 11.33B. By increasing the constant current from 0.1 to 2.0 A, the variation of the cathode potential became much faster. However, it can be derived from Fig. 11.33B that the total amount of charge passed remained fairly constant (4200 C) when the current increased from 0.1 to 1.0 A, but it dropped to 3600 C at 2.0 A, indicating incomplete reduction at high currents, likely due to the diffusion rate being unable to match that of the reduction. Fig. 11.33B also shows that the cathode potential profile shifted negatively (downwards) with increasing the current, which could be due to the increasing ohmic polarisation (or iR drop) with increasing the current.

9. Titanium based medical implant materials Titanium is physiologically inert, biochemically stable, and hence a good candidate material for making medical implants. The main issue of using

274

Extractive Metallurgy of Titanium

metals to make bone replacements is that a solid metal has a much greater mechanical strength than bone which is also often porous. As it can be seen in many examples discussed above, Ti metal and alloys prepared by electrodeoxidation of the respective oxides were all porous to different degrees. This is also beneficial to making medical implants because the porous structure, in principle, could allow body fluid to access, and hence promote bone growth into the interior of the implant. However, in these studies, formation of the porous structures occurred naturally because of the use of porous oxide precursor and the removal of oxygen. Therefore, efforts to prepare the FFC Ti metal with controlled porous structures were made using powders of graphite and polymers as pore-making fugitives which should burn to CO2 upon sintering the oxide precursors in air [95]. Fig. 11.34 compares the particulate morphology of the fugitives and their effects on the produced porous TiO2 precursors. The work showed that up to 80% porosity can be achieved with the addition of 45 wt% graphite or polyethylene in the TiO2 precursor, followed by sintering in air at 800 C. It should be pointed out that using polymer based fugitives has a negative effect because slow burning of polymers at medium (A)

(B)

(C)

(D)

Figure 11.34 SEM images of powders of (A) graphite, (B) polyethylene, and (C) the TiO2 precursor prepared using the (C) graphite and (D) polyethylene powders as the pore making fugitives [95].

Invention and fundamentals of the FFC Cambridge Process

275

temperatures, e.g., 800 C, may produce unhealthy or harmful gases. The later work, as discussed above, on using NH4Cl as a recyclable fugitive to prepare highly porous TiO2 precursors was equally effective, but much cleaner. Such prepared TiO2 precursors were successfully electrodeoxidised to highly porous Ti products (cf. Fig. 11.12B). It was also noticed that the rate for reducing these highly porous TiO2 precursors was significantly increased [18,95]. As shown in Figs. 11.12B, 12C, 34C and 34D, using fugitives with appropriate sizes and shapes, not only the oxide precursors, but also the produced Ti metal by electro-deoxidation can be hierarchically porous. In this line, another more complicated “gel casting” method was used to prepare uniform porous TiO2 precursors. This method involved mixing the TiO2 powder with a monomer, crosslinker, surfactants and water into a slurry, and then adding the initiator and catalyst to induce in situ polymerisation. It was then followed by mechanical foaming and setting (polymerisation), and then programmed heating up to 1450 C and sintering at the final temperature [96]. Electro-deoxidation of this uniform porous TiO2 led to similarly porous but much shrunken Ti product as shown in Fig. 11.35A and B. The X-ray diffraction based mCT method was used to determine the internal porous structures of the porous TiO2 precursor and the reduced porous Ti metal. The results are presented in Fig. 11.35C and D, in comparison with similarly determined structures of a commercial porous Ti metal, and a typical porous human bone, as shown in Fig. 11.35E and F, respectively [96].

Figure 11.35 Optical images of (A) porous TiO2 precursor and (B) the reduced Ti foam, and computer modeling of microstructures of (C) sintered porous TiO2; (D) porous FFC Ti foam; (E) commercial Ti foam (78% porosity); (F) typical spongy bone in human [96].

276

Extractive Metallurgy of Titanium

Similarly produced porous FFC Ti alloys were also analyzed for their mechanical properties. It was found that by controlling the density of the oxide precursor, the maximum compressive stress and modulus of the FFC Tie6Ale4V alloy were approximately 243 MPa and 14 GPa, respectively, matching with the requirements for medical implants [73]. Other tests on the porous FFC TiZr [97] and TiNbx (x ¼ 24, 35, 42) [98] alloys demonstrated elastic moduli between 20 and 40 GPa, and 8.0e11.2 GPa, respectively, comparing very well with that of natural bone (10e30 GPa). Testing the FFC TiZr and Tie13Zre13Nb alloys in the Ringer solution [97,99], and TiNbx alloys in the Hank solution [98] (simulated human body fluids) confirmed excellent stability and resistance to corrosion. With reference to commercial implant materials, the as-prepared porous FFC Ti and alloy samples also presented very similar or better biocompatibility for bone cell adaption and growth [100]. These preliminary experimental findings indicate a great potential of the porous FFC Ti metal and alloys in bone implant applications.

10. Cathodic protection of titanium Ti metal and its alloys are usually excellent anticorrosion materials, but they become more vulnerable to oxidation attack when the temperature increases, particularly when the metal is molten. In order to prevent oxidation of molten Ti, a flux is frequently applied but the fluxes often contain fluoroborates and boric acid which should be reduced by titanium, thereby introducing oxygen into the metal. An alternative method is to use an inert gas such as helium, which is expensive, or argon, which is much cheaper but no way as efficient. Cathodic protection is an alternative approach which is applied extensively at ambient temperatures to prevent corrosion of constructions made of steel such as ships, pipelines and other structures. This is achieved by applying a moderate cathodic potential to the metal component coupled to a suitable anode. Recently, this method has been examined as a means of prevention of oxidation of Ti at elevated temperatures [101]. Complete protection was achieved by applying a modest cathodic potential to a titanium crucibles that was half filled with molten calcium chloride, against an anode of either doped SnO2 or iridium. It was found that, in contrast to the unprotected area, there was no noticeable oxidation or corrosion after exposure to air for several hours. At more elevated temperatures, this technique was also successfully applied to protect a molten titanium alloy.

Invention and fundamentals of the FFC Cambridge Process

277

An interesting challenge is whether this technique can be used to cathodically protect titanium components whilst undergoing welding titanium in air. This would be desirable, instead of welding under a protective atmosphere that is provided by either a gas blanket or undertaken in a gas filled cubicle, as done in established welding technologies. This idea was proven by spot welding a Tie6Ale4V alloy plate with an ytterbium fiber laser under a calcium-fluoride based flux cover and polarising the titanium alloy against an iridium inert anode partially immersed in the salt flux. The analysis of the titanium alloy after the welding experiments showed that that the oxygen contents of the weld metal were very close to that of the starting material, whilst the nitrogen contents had increased very slightly [102]. Fig. 11.36 shows (A) the experimental set up, (B and C) oxygen and nitrogen contents of the samples as determined by the hot extraction method, and (D, E, and F) SEM images of the cross sections of the welds under different coverages, which demonstrates conclusively that this may be a method for welding titanium alloys in air without the use of inert gas.

Figure 11.36 Welding of Ti6Al4V under cathodic production. (A) Photograph of the experimental set up for laser beam welding, showing an Ir anode inserted into a layer of salt powder (CaF2eNaF) covering a Ti6Al4V plate. (B) Oxygen and (C) nitrogen contents detected before and after laser beam spot welding under the indicated conditions. (D-F) SEM images of the cross section of the spot welds prepared under coverage of (D) Ar, (E) air, and (F) air, molten salt and cathodic polarisation [102].

278

Extractive Metallurgy of Titanium

11. Outlook and Prospective Twenty nineteen marks the 20th anniversary of the publication of the original patent of the FFC Cambridge Process [103] and, as discussed above, there has been tremendous progress in both fundamental research and the development of the technology. Whilst commercialisation has also been progressing [104,105], better understanding of both the cathode and anode processes in the FFC Cambridge Process is still needed, particularly in terms of in situ analyses with spectroscopic tools and cell designs. Detailed mechanisms on the electronic conduction through the molten salts are still pending, whilst prevention or minimisation of the negative impact of electronic conduction requires both physical and chemical interventions, plus process engineering design. For example, it is possible to reduce the electronic conduction by using an oxygen ion conducting membrane better than that of yttria stabilised zirconia, and by the addition of an absorbent of the redox shuttling species, particularly dissolved Ca(0) and Ca(I) as indicated by reactions (11.23) to (11.25). The idea of computer assisted control (CAC) with reference to Fig. 11.21 is also worth further refining in connection with cathode potential monitoring. The search for suitable inert materials for fabrication of the anode shall continue, following the work reported in Ref. [38] in terms of optimisation of the material composition and manufacturing of the anode in desirable shapes and dimensions. The use of metal sulfides as the cathodic feed in the FFC Cambridge Process is promising in terms of process engineering and efficiency of electrolysis, whilst an efficient, clean and low cost process is needed to convert metal oxides to metal sulfides, particularly TiO2 to TiS2 [69e71]. Furthermore, the versatile and unprecedented capability of the FFC Cambridge Process to make various metals, alloys, intermetallics and nearnet-shape products is far from being fully explored. The success in laboratory preparation of multi-metal alloys, particularly the high entropy alloys of four or more refractory metal elements is strong evidence that electrodeoxidation can achieve what the conventional methods are incapable of, or difficult to achieve. Another opportunity is to apply the FFC Cambridge Process in the development of the new concept of regenerative fuels that can help capture and storage of renewable energy for not only transport, but also seasonal and regional usages [106]. There has been the hope that electrodeoxidation of lunar rock and soil (which are typically oxide-based minerals containing metallic elements such as Ca, Mg, Al, Si, Fe, and Ti) could become the sources of oxygen and metals supply on the moon [107,108].

Invention and fundamentals of the FFC Cambridge Process

279

Figure 11.37 A group of second grade students in Springfield, Illinois, USA working on a FIRST LEGO League Jr. project where the students are trying to build a model out of LEGO to show how to obtain and use oxygen on the moon by the FFC Cambridge process. (Use of this photograph here has permission from the project leader, the students and parents of the students. Photograph courtesy of Tesch Woods.)

Perhaps, what is even more exciting is that the European Space Agency is developing the FFC Cambridge Process, with Metalysis Ltd., to treat the moon’s regolith and exporting the technology out into the solar system [109]. Such an anticipation has stimulated citation of the FFC Cambridge Process for smelting purposes on the moon in scientific fiction [110], and inspired educational activities in school to build a LEGO model of the FFC Cambridge Process operating on the moon, as shown in Fig. 11.37. Obviously, the list of future opportunities is too long to be completed here for the FFC Cambridge Process which however has been proven, so far, via research and development in the past 2 decades, to be a highly useful tool for both scientific and technological adventures, including space exploration.

12. Conclusions In the past 20 years several new molten salt electrochemical methods for the cathodic reduction of titanium dioxide have been investigated as reviewed recently [111e113]. A few of these, the OS, FFC Cambridge, SOM and MOE processes, have managed to gain further funding to advance work to a larger scale and industrial grade quantities. This paper has concentrated on the FFC Cambridge Process, outlining its mechanism of

280

Extractive Metallurgy of Titanium

reduction both the simple oxides and mixtures of multioxides which can lead to interesting alloys, including high entropy alloys, which are difficult to make using standard metallurgical methods. Important parameters such as the ratio of the density of the oxide to the reduced metal are explored. Considerable progress has been made in reducing both oxides and sulfides. In the conventional FFC Cambridge Process the anode is usually carbon which can lead to carbon being transferred from the anode to cathode causing contamination and reducing the current efficiency. It is not easy to find a suitable material for an inert anode by alloying or finding a suitable oxide but, eventually, calcium ruthenate, a highly electronically material, was found to be inert in molten calcium chloride under the conditions for oxygen evolution. As well of the FFC Cambridge Process being used for reducing metal oxides it can also be used to prevent oxidation and to allow welding without the use of inert gases. Overall, considerable advances have been made in the past 20 years which are now finding industrial applications. We look forward to similar progresses being made in the next 20 years, particularly with the expected contributions from enthusiastic children like those in the photograph of Fig. 11.37 who will grow into next generation of scientists, engineers and managers.

References [1] D.J. Fray, Investigations of an electrolytic method for the removal of dissolved oxygen from group IV metals, Research Grant of the Engineering and Physical Science Research Council (1994). Grant No. GR/J57650/01. [2] G.Z. Chen, D.J. Fray, T.W. Farthing, Direct electrochemical reduction of titanium dioxide to titanium n molten calcium chloride, Nature 407 (2000) 361e364. [3] G.Z. Chen, D.J. Fray, T.W. Farthing, Cathodic deoxygenation of the alpha-case on titanium and alloys in molten calcium chloride, Metallurgical and Materials Transactions B 32 (2001) 1041e1502. [4] G.Z. Chen, D.J. Fray, Understanding the electro-reduction of metal oxides in molten salts, Light Metals 2004 (2004) 881e886. [5] C. Schawndt, D.J. Fray, Determination of the kinetic pathway in the electrochemical reduction of titanium dioxide in molten calcium chloride, Electrochimica Acta 51 (2005) 66e76. [6] R. Bhagat, D. Dye, S.L. Raghunathan, R.J. Talling, D. Inman, B.K. Jackson, K.K. Rao, R.J. Dashwood, In situ synchrotron diffraction of the electrochemical reduction pathway of TiO2, Acta Materialia 58 (2010) 5057e5062. [7] T.H. Okabe, M. Nakamura, T. Oishi, K. Ono, Electrochemical deoxidation of titanium, Metallurgical and Materials Transactions B 24 (1993) 449e455. [8] K. Ono, R.O. Suzuki, A new concept for producing Ti sponge: calciothermic reduction, Journal of Occupational Medicine 54 (2002) 59e61. [9] R.O. Suzuki, Calciothermic Reduction of TiO2 and in Situ Electrolysis of CaO in the Molten CaCl2, vol. 66, 2005, pp. 461e465.

Invention and fundamentals of the FFC Cambridge Process

281

[10] K. Jiang, X.H. Hu, M. Ma, D.H. Wang, G.H. Qiu, X.B. Jin, G.Z. Chen, Perovskitization assisted electrochemical reduction of solid TiO2 in molten CaCl2, Angewandte Chemie International Edition 45 (2006) 428e432. [11] G.Z. Chen, E. Gordo, D.J. Fray, Direct electrolytic preparation of chromium powder, Metallurgical and Materials Transactions B 35 (2004) 223e233. [12] G.Z. Chen, D.J. Fray, Novel cathodic processes in molten salts, in: N. Chen, Z. Qiao (Eds.), Proceedings of the 6th International Symposium on Molten Salt Chemistry and Technology, Shanghai University Press, Shanghai, China, 2001, pp. 79e85, 0813 Oct. 2001. [13] Y. Deng, D.H. Wang, W. Xiao, X.B. Jin, X.H. Hu, G.Z. Chen, Electrochemistry at conductor/insulator/electrolyte three-phase interlines: a thin layer model, Journal of Physical Chemistry B 109 (2005) 14043e14051. [14] W. Xiao, X.B. Jin, Y. Deng, D.H. Wang, X.H. Hu, G.Z. Chen, Electrochemically driven three-phase interlines into insulator compounds: electro-reduction of solid SiO2 in molten CaCl2, ChemPhysChem 7 (2006) 1750e1758. [15] W. Xiao, X.B. Jin, Y. Deng, D.H. Wang, G.Z. Chen, Three-phase interlines electrochemically driven into insulator compounds: a penetration model and its verification by electroreduction of solid AgCl, Chemistry e A European Journal 13 (2007) 604e612. [16] X.B. Jin, P. Gao, D.H. Wang, X.H. Hu, G.Z. Chen, Electrochemical preparation of silicon and its alloys from solid oxides in molten calcium chloride, Angewandte Chemie International Edition 43 (2004) 733e736. [17] J.J. Peng, K. Jiang, W. Xiao, D.H. Wang, X.B. Jin, G.Z. Chen, Electrochemical conversion of oxide precursors to consolidated Zr and Zr-2.5Nb tubes, Chemistry of Materials 20 (2008) 7274e7280. [18] W. Li, X.B. Jin, F.L. Huang, G.Z. Chen, Metal-to-oxide molar volume ratio: the overlooked barrier to solid-state electro-reduction and a green bypass through recyclable NH4HCO3, Angewandte Chemie International Edition 49 (2010) 3203e3206. [19] N.H. Nachtrieb, Conduction in fused salts and salt-metal solutions, Annual Review of Physical Chemistry 31 (1980) 131e156. [20] A. Selloni, P. Carnevali, P. Car, M. Parrinello, Localization, hopping, and diffusion of electrons in molten salts, Physical Review Letters 59 (1987) 823e826. [21] M. Ma, D.H. Wang, W.G. Wang, X.H. Hu, X.B. Jin, G.Z. Chen, Extraction of titanium from different titania precursors by the FFC Cambridge process, Journal of Alloys and Compounds 420 (2006) 37. [22] G.H. Qiu, K. Jiang, M. Ma, D.H. Wang, X.B. Jin, G.Z. Chen, Roles of cationic and elemental calcium in electro-reduction of solid metal oxides in molten calcium chloride, Zeitschrift für Naturforschung 62a (2007) 292e302. [23] U.B. Pal, D.E. Woolley, G.B. Kenney, Emerging SOM technology for the green synthesis of metals from oxides, Journal of Occupational Medicine 53 (2001) 32e35. [24] X.L. Zou, X.G. Lu, Z.F. Zhou, C.H. Li, Direct electrosynthesis of Ti5Si3/TiC composites from their oxides/C precursors in molten calcium chloride, Electrochemistry Communications 21 (2012) 9e13. [25] H.B. Hu, Y.M. Gao, Y.G. Lao, Q.W. Qin, G.Q. Li, G.Z. Chen, Yttria stabilized zirconia aided electrochemical investigation on ferric ions in mixed molten calcium and sodium chlorides, Metallurgical and Materials Transactions B 49 (2018) 2794e2808. [26] R.M. Ormerod, Solid oxide fuel cells, Chemical Society Reviews 32 (2003) 17e28. [27] B. Singh, S. Ghosh, S. Aich, B. Roy, Low temperature solid oxide electrolytes (LTSOE): a review, Journal of Power Sources 339 (2017) 103e135.

282

Extractive Metallurgy of Titanium

[28] H.V. Ijije, C.G. Sun, G.Z. Chen, Indirect electrochemical reduction of carbon dioxide to carbon nanopowders in molten alkali carbonates: process variables and product properties, Carbon 73 (2014) 163e174. [29] K. Dring, Direct Electrochemical Production of Titanium, 2007. http://www.okabe. iis.u-tokyo.ac.jp/core-to-core/rmw/RMW3/slide/RMW3_06_Dring_F.pdf. [30] G.Z. Chen, Electro-reduction of solid TiO2 in molten CaCl2: barriers and feasible solutions for the new making of titanium, in: 2nd International Round Table for Titanium Production in Molten Salts, Trondheim, 19-22 September 2010. [31] H. Kim, J. Paramore, A. Allanore, D.R. Sadoway, Electrolysis of molten iron oxide with an iridium anode: the role of electrolyte basicity, Journal of the Electrochemical Society 158 (2011) E101eE105. [32] HSC Chemistry 6.12 (2007). [33] R. Barnett, K.T. Kilby, D.J. Fray, Reduction of tantalum pentoxide using graphite and tin-oxide-based anodes via the FFC-Cambridge Process, Metallurgical and Materials Transactions B 40 (2009) 150e157. [34] V. Kaplan, E. Wachtel, I. Lubomirsky, Titanium carbide coating of titanium by cathodic deposition from a carbonate melt, Journal of the Electrochemical Society 159 (2012) E159eE161. [35] P. Meyer, M. Gibilaro, L. Massot, I. Pasquet, P. Tailhades, S. Bouvet, P. Chamelot, Comparative study on the chemical stability of Fe3O4 and NiFe2O4 in molten salts, Materials Science and Engineering B 228 (2018) 117e122. [36] A.A. Murashkina, A.N. Demina, Doping of calcium titanate with chromium and indium, Inorganic Materials 41 (2005) 402e405. [37] T. Esaka, T. Fujii, K. Suwa, H. Iwahara, Electrical conduction in CaTi1xFexO3d under low oxygen pressure and its application for hydrogen production, Solid State Ionics 40e41 (1990) 544e547. [38] S.Q. Jiao, D.J. Fray, Development of an Inert Anode for Electrowinning in calcium chlorideecalcium oxide melts, Metallurgical and Materials Transactions B 41 (2010) 74e79. [39] L.W. Hu, Y. Song, J.B. Ge, S.Q. Jiao, J. Cheng, J. Electrochemical metallurgy in CaCl2-CaO melts on the basis of TiO2$RuO2 inert anode, Journal of the Electrochemical Society 163 (2016) E33eE38. [40] G.Z. Chen, D.J. Fray, Cathodic refining in molten salts: removal of oxygen, sulfur and selenium from static and owing molten copper, Journal of Applied Electrochemistry 31 (2001) 155e164. [41] D.J. Fray, T.W. Farthing, Z. Chen, Removal of Oxygen from Metal Oxides and Solid Solutions by Electrolysis in a Fused Salt, 5 June 1998. Patent, WO9964638, UK filing date. [42] G.Z. Chen, D.J. Fray, Electro-deoxidation of metal oxide, Light Metals 2001 (2001) 1147e1151. [43] K.S. Mohandas, D.J. Fray, Electrochemical deoxidation of solid zirconium dioxide in molten calcium chloride, Metallurgical and Materials Transactions B 40 (2009) 685e699. [44] T. Nohira, K. Yasuda, Y. Ito, Pinpoint and bulk electrochemical reduction of insulating silicon dioxide to silicon, Nature Materials 2 (2003) 397e401. [45] J.Y. Yang, S.G. Lu, S.R. Kan, X.J. Zhang, J. Du, Electrochemical preparation of silicon nanowires from nanometre silica in molten calcium chloride, Chemical Communications (2009) 3273e3275. [46] E. Gordo, G.Z. Chen, D.J. Fray, Toward optimisation of electrolytic reduction of solid chromium oxide to chromium powder in molten chloride salts, Electrochimica Acta 49 (2004) 2195e2208.

Invention and fundamentals of the FFC Cambridge Process

283

[47] X.F. Hu, Q. Xu, Preparation of tantalum by electro-deoxidation in a CaCl2-NaCl melt, Acta Metallurgica Sinica 42 (2006) 285e289. [48] T. Wu, X.B. Jin, W. Xiao, X.H. Hu, D.H. Wang, G.Z. Chen, Thin pellets: fast electrochemical preparation of capacitor tantalum powders, Chemistry of Materials 19 (2007) 153e160. [49] X.Y. Yan, D.J. Fray, Production of niobium powder by direct electrochemical reduction of solid Nb2O5 in a eutectic CaCl2-NaCl melt, Metallurgical and Materials Transactions B 33 (2002) 685e693. [50] T. Wu, X.B. Jin, W. Xiao, C. Liu, D.H. Wang, G.Z. Chen, Computer-aided control of electrolysis of solid Nb2O5 in molten CaCl2, Physical Chemistry Chemical Physics 10 (2008) 1809e1818. [51] D.H. Wang, G.H. Qiu, X.B. Jin, X.H. Hu, G.Z. Chen, Electrochemical metallisation of solid terbium oxide, Angewandte Chemie International Edition 45 (15) (2006) 2384e2388. [52] X.Y. Yan, D.J. Fray, Direct electrolytic reduction of solid alumina using molten calcium chloride-alkali chloride electrolytes, Journal of Applied Electrochemistry 39 (2009) 1349e1360. [53] H.W. Xie, H. Zhang, Y.C. Zhai, J.X. Wang, C.D. Li, Al, Preparation from solid Al2O3 by direct electrochemical deoxidation in molten CaCl2-NaCl at 550 C, Journal of Materials Science and Technology 25 (2009) 459e461. [54] G.M. Li, D.H. Wang, Z. Chen, Direct reduction of solid Fe2O3 in molten CaCl2 by potentially green process, Journal of Materials Science and Technology 25 (2009) 767e771. [55] G.M. Haarberg, B.Y. Yuan, Direct electrochemical reduction of hematite pellets in alkaline solutions, ECS Trans 58 (2014) 19e28. [56] G.Z. Chen, D.J. Fray, A morphological study of the FFC chromium and titanium powders, Mineral Processing and Extractive Metallurgy Transactions of the Institutions of Mining and Metallurgy: Section C 115 (2006) 49e54. [57] S. Fang, H. Wang, J.Y. Yang, S.G. Lu, B. Yu, J.T. Wang, C.R. Zhao, Electrochemical preparation of silicon nanowires from porous NiO/SiO2 blocks in molten CaCl2, Materials Letters 160 (2015) 1e4. [58] S. Fang, H. Wang, J.Y. Yang, B. Yu, S.G. Lu, Formation of Si nanowires by the electrochemical reduction of SiO2 with Ni or NiO additives, Faraday Discussions 190 (2016) 433e449. [59] K. Yasuda, T. Nohira, Y.H. Ogat, Y. Ito, Direct electrolytic reduction of solid silicon dioxide in molten LiCleKCleCaCl2 at 773 K, Journal of the Electrochemical Society 152 (2005) D208eD212. [60] W. Xiao, X.B. Jin, G.Z. Chen, Up-scalable and controllable electrolytic production of photo-responsive nanostructured silicon, Journal of Materials Chemistry 1 (2013) 10243e10250. [61] H.L. Chen, X.B. Jin, L.P. Yu, G.Z. Chen, Influences of graphite anode area on electrolysis of solid metal oxides in molten salts, Journal of Solid State Electrochemistry 18 (12) (2014) 3317e3325. [62] C.Y. Chen, X.G. Lu, Q. Li, H.W. Cheng, Tantalum preparation by SOM process, Rare Metal Materials and Engineering 36 (2007) 1991e1995. [63] X.Y. Yan, D.J. Fray, Electrochemical studies on reduction of solid Nb2O5 in molten CaCl2-NaCl Eutectic – II. Cathodic processes in electrodeoxidation of solid Nb2O5, Journal of the Electrochemical Society 152 (2005) D12eD21. [64] G.M. Li, D.H. Wang, X.B. Jin, G.Z. Chen, Electrolysis of solid MoS2 in molten CaCl2 for Mo extraction without emission, Electrochemistry Communications 9 (2007) 1951e1957.

284

Extractive Metallurgy of Titanium

[65] R. Woods, Extracting metals from sulfide ores, in: Electrochemistry Encyclopedia, The Electrochemical Society, 2010. http://knowledge.electrochem.org/encycl/artm02-metals.htm. [66] X.L. Ge, X.D. Wang, S. Seetharaman, Copper extraction from copper ore by electroreduction in molten CaCl2eNaCl, Electrochimica Acta 54 (2009) 4397e4402. [67] T. Wang, H.P. Gao, X.B. Jin, H.L. Chen, J.J. Peng, G.Z. Chen, Electrolysis of solid metal sulfide to metal and sulfur in molten NaCl-KCl, Electrochem. Commum. 13 (2011) 1492e1495. [68] M.S. Whittingham, Lithium batteries and cathode materials, Chemical Reviews 104 (2004) 4271e4302. [69] N. Suzuki, M. Tanaka, H. Noguchi, S. Natsui, T. Kikuchi, R.O. Suzuki, Calcium reduction of TiS2 in CaCl2 melt, Materials Transactions 58 (2017) 360e370. [70] M. Ohta, S. Satoh, T. Kuzuya, S. Hirai, M. Kunii, A. Yamamoto, Thermoelectric properties of Ti1þxS2 prepared by CS2 sulfurization, Acta Materialia 60 (2012) 7232e7240. [71] C.S. Wu, M.S. Tan, G.Z. Ye, D.J. Fray, X.B. Jin, High-efficiency preparation of titanium through electrolysis of carbo-sulfurized titanium dioxide, ACS Sustainable Chemistry & Engineering 7 (2019) 8340e8346. [72] T. Matsuzaki, S. Natsui, T. Kikuchi, R.O. Suzuki, Electrolytic reduction of V3S4 in molten CaCl2, Materials Transactions 58 (2017) 371e376. [73] D. Hu, W. Xiao, G.Z. Chen, Near-net-shape production of hollow titanium alloy components via electrochemical reduction of metal oxide precursors in molten salts, Metallurgical and Materials Transactions B 44 (2013) 272e282. [74] R. Bhagat, M. Jackson, D. Inman, R. Dashwood, The production of Ti-Mo alloys from mixed oxide precursors via the FFC Cambridge Process, Journal of the Electrochemical Society 155 (2008) E63eE69. [75] R. Bhagat, M. Jackson, D. Inman, R. Dashwood, Production of Ti-W alloys from mixed oxide precursors via the FFC Cambridge process, Journal of the Electrochemical Society 156 (2009) E1eE7. [76] Y. Zhu, M. Ma, D.H. Wang, K. Jiang, X.H. Hu, X.B. Jin, G.Z. Chen, Electrolytic reduction of mixed solid oxides in molten salts for energy efficient production of the NiTi alloy, Chinese Science Bulletin 51 (2006) 2535e2540. [77] G.M. Li, X.B. Jin, D.H. Wang, G.Z. Chen, Affordable electrolytic ferrotitanium alloys with marine engineering potentials, Journal of Alloys and Compounds 428 (2009) 320e327. [78] T. Wu, Studies of Electrolysis of Solid Oxides for Preparation of VB Metals and Their Alloys in Molten Salts, Wuhan University, 2008. PhD Thesis. [79] X.L. Zou, X.G. Lu, Z.F. Zhou, C.H. Li, W.Z. Ding, Direct selective extraction of titanium silicide Ti5Si3 from multi-component Ti-bearing compounds in molten salt by an electrochemical process, Electrochimica Acta 56 (2011) 8430e8437. [80] S.S. Li, X.L. Zou, K. Zheng, X.G. Lu, Q. Xu, C.Y. Chen, Z.F. Zhou, Direct production of TiAl3 from Ti/Al-containing oxides precursors by solid oxide membrane (SOM) process, Journal of Alloys and Compounds 727 (2017) 1243e1252. [81] Y.-S. Wang, X.-L. Zou, X.-G. Lu, S.S. Li, K. Zheng, S.-J. Wang, Q. Xu, Z.F. Zhou, Electrolytic production of Ti-Ge intermetallics from oxides in molten CaCl2-NaCl, Transactions of Nonferrous Metals Society of China 28 (2018) 2351e2359. [82] J. Sure, D.S.M. Vishnu, R.V. Kumar, C. Schwandt, Molten salt electrochemical synthesis, heat treatment and microhardness of Ti-5Ta-2Nb alloy, Materials Transactions (2019), https://doi.org/10.2320/matertrans.MA201808. [83] J. Sure, D.S.M. Vishnu, C. Schwandt, Direct electrochemical synthesis of highentropy alloys from metal oxides, Applied Materials Today 9 (2017) 111e121.

Invention and fundamentals of the FFC Cambridge Process

285

[84] S. Kuramoto, T. Furuta, J.H. Hwang, K. Nishino, T. Saito, Metallurgical and Materials Transactions A 37 (2006) 657. [85] J.-W. Yeh, S.-K. Chen, S.-J. Lin, J.-Y. Gan, T.-S. Chin, T.-T. Shun, C.-H. Tsau, S.Y. Chang, Nanostructured high-entropy alloys with multiple principal elements: novel alloy design concepts and outcomes, Advanced Engineering Materials 6 (2004) 299e303. [86] S.S. Li, X.L. Zou, X.L. Xiong, K. Zheng, X.G. Lu, Z.F. Zhou, X.L. Xie, Q. Xu, Electrosynthesis of Ti3AlC2 from oxides/carbon precursor in molten calcium chloride, Journal of Alloys and Compounds 735 (2018) 1901e1907. [87] J.H. Fan, D.D. Tang, X.H. Mao, H. Zhu, D.H. Wang, An efficient electrolytic preparation of MAX-phased Ti-Al-C, Metallurgical and Materials Transactions B 49 (2018) 2770e2778. [88] B.A. Glowacki, X.Y. Yan, D.J. Fray, G. Chen, M. Majoros, Y. Shi, Niobium based intermetallics as a source of high-current/high magnetic field superconductors, Physica C 372e376 (2002) 1315e1320. [89] A.J. Muir Wood, R.C. Copcutt, G.Z. Chen, D.J. Fray, Electrochemical fabrication of nickel manganese gallium alloy powder, Advanced Engineering Materials 5 (2003) 650e653. [90] X.Y. Yan, D.J. Fray, Electrosynthesis of NbTi and Nb3Sn superconductors from oxide precursors in CaCl2-based melts, Advanced Functional Materials 15 (2005) 1757e1761. [91] M. Ma, D.H. Wang, X.H. Hu, X.B. Jin, G.Z. Chen, A direct electrochemical route from ilmenite to hydrogen storage ferrotitanium alloys, Chemistry e A European Journal 12 (2006) 5075e5081. [92] Y. Zhu, D.H. Wang, M. Ma, X.H. Hu, X.B. Jin, G.Z. Chen, More affordable electrolytic LaNi5-type hydrogen storage powders, Chem. Commum. (2007) 2515e2517. [93] J.J. Peng, Y. Zhu, D.H. Wang, X.B. Jin, G.Z. Chen, Direct and low energy electrolytic co-reduction of mixed oxides to zirconium-based multi-phase hydrogen storage alloys in molten salts, Journal of Materials Chemistry 19 (2009) 2803e2809. [94] D.J.S. Hyslop, A.M. Abdelkader, A. Cox, D.J. Fray, Utilization of constant current chronopotentiometry to synthesize a CoeCr alloy, Journal of the Electrochemical Society 157 (2010) E111eE115. [95] R.L. Centeno-Sanchez, D.J. Fray, G.Z. Chen, Study on the reduction of highly porous TiO2 precursors and thin TiO2 layers by the FFC-Cambridge process, Journal of Materials Science 42 (2007) 7494e7501. [96] R. Singh, P.D. Lee, J.R. Jones, G. Poologasundarampillai, T. Post, T.C. Lindley, R.J. Dashwood, Hierarchically structured titanium foams for tissue scaffold applications, Acta Biomaterialia 6 (2010) 4596e4604. [97] J.J. Peng, H.L. Chen, T. Wang, X.B. Jin, D.H. Wang, G.Z. Chen, Phase-tuneable fabrication of consolidated (aþb)-TiZr alloys through molten salt electrolysis of solid oxides, Chemistry of Materials 21 (2009) 5187e5195. [98] D.S.M. Vishnua, J. Surea, Y.J. Liu, R.V. Kumar, C. Schwandt, Electrochemical synthesis of porous Ti-Nb alloys for biomedical applications, Materials Science and Engineering: C 96 (2019) 466e478. [99] T. Yu, H.Y. Yin, Y. Zhou, Y.N. Wang, H. Zhu, D.H. Wang, Electrochemical preparation of porous Ti-13Zr-13Nb alloy and it’s corrosion behavior in Ringer’s solution, Materials Transactions 58 (2017) 326e330. [100] X. Yang, D.H. Wang, Y.D. Liang, H.Y. Yin, S. Zhang, T. Jiang, Y.N. Wang, Y. Zhou, A new implant with solid core and porous surface: the biocompatability with bone, Journal of Biomedical Materials Research Part A 102A (2014) 2395e2407.

286

Extractive Metallurgy of Titanium

[101] C. Schwandt, D.J. Fray, Use of molten salt fluxes and cathodic protection for preventing the oxidation of titanium at elevated temperatures, Metallurgical and Materials Transactions B 45 (2014) 2145e2152. [102] C. Allen, C. Schwandt, D.J. Fray, Laser welding studies on Ti-6Al-4V in air in conjunction with cathodic protection, Welding in the World 60 (2016) 689e696. [103] D.J. Fray, T.W. Farthing, Z. Chen, Removal of Oxygen from Metal Oxides and Solid Solutions by Electrolysis in a Fused Salt, 1999. Patent, WO9964638. [104] Metalysis, Technology, 2019. http://www.metalysis.com/technology/. [105] GLABAT, Development of Negative Electrode Materials, 2015. http://www.glabat. com/article/content/view?id¼23. [106] L. Xia, L.P. Yu, D. Hu, G.Z. Chen, Electrolytes for electrochemical energy storage, Materials Chemistry Frontiers 1 (2017) 584e618. [107] R.O. Colson, L.A. Haskin, Oxygen from the lunar soil by molten silicate electrolysis, in: F.A. McKay, D.S. McKay, M.B. Duke (Eds.), Space Resources, Materials, vol. 3, NASA, Lyndon B Johnson Space Center, Houston, TX, 1992, pp. 195e209. [108] C. Schwandt, J.A. Hamilton, D.J. Fray, I.A. Crawford, The production of oxygen and metal from lunar regolith, Planetary and Space Science 74 (2012) 49e56. [109] A. Williams, V500,000 available in Metalysis and ESA space exploration, https:// 3dprintingindustry.com/news/e500000-available-in-metalysis-and-esa-spaceexploration-competition-147483/(accessed 31.03.2019). [110] A. Weir, Artemis, Del Rey, London, 2017, p. 203 & p. 216. [111] W. Xiao, D.H. Wang, The electrochemical reduction processes of solid compounds in high temperature molten salts, Chemical Society Reviews 43 (2014) 3215e3228. [112] Y. Zhang, Z.Z. Fang, P. Sun, S.L. Zheng, Y. Xia, M. Free, A perspective on thermochemical and electrochemical processes for titanium metal production, Journal of Occupational Medicine 69 (2017) 1861e1868. [113] D. Hu, G.Z. Chen, Chapter 25 e advanced extractive electrometallurgy, in: C. Breitkopf, K. Swider-Lyons (Eds.), Springer Handbook of Electrochemical Energy, Springer, Berlin, 2017, pp. 801e834.