2019 Volume 60 Issue 3 Pages 391-399
Alloy of composition Ti–5Ta–2Nb (numbers in mass%) was synthesised directly from the corresponding mixed metal oxides via the FFC-Cambridge process. Compacted powder discs of TiO2–Ta2O5–Nb2O5 were cathodically polarised against a graphite anode in a molten CaCl2 electrolyte, and the effect of various process parameters on the synthesis of the alloy was investigated. The samples retained their disc-type shape throughout the oxide-to-metal conversion under appropriate processing conditions. XRD analysis showed that the electrochemically prepared alloy existed as single-phase α-Ti and changed into dual-phase (α+β)-Ti upon heat treatment. SEM/EDX analysis revealed a nodular porous microstructure and confirmed the intended chemical composition. Backscattered SEM analysis after heat treatment provided direct evidence of micron-sized β-phase precipitates at the boundaries of the α-phase particles. Microhardness of the heat-treated alloy compared well with literature data for the same material prepared by conventional metallurgical methods. Overall, the study has demonstrated the feasibility of the single-step electrochemical fabrication of Ti–5Ta–2Nb alloy bodies directly from oxides for structural application.
Schematic representation of synthesis of Ti–5Ta–2Nb alloy by electrochemical deoxidation of metal oxide mixtures in a molten salt for measurement of various properties.
Titanium and its alloys find extensive application worldwide in the nuclear,1) marine,2) aerospace3) and biomedical4,5) sectors as well as in the chemical, electrochemical and power plant industries.6) This is due to their excellent strength and ductility and their exceptional oxidation and corrosion resistances in various extreme environments, often rendering Ti or a Ti alloy the only option in safety-critical cases. The existing alloy types are hence continuously improved and new ones developed.7) Transition metal alloying elements, such as Nb, Ta, Mo and W, stabilise the β-phase of Ti and improve the oxidation and corrosion resistance.7) In practice however, the addition of β-stabilisers to Ti master alloys is often restricted, because they are costly, and because they have much higher melting temperatures than pure Ti, increasing the difference between liquidus and solidus temperatures8) and thus causing segregation problems during solidification.
Ti–5Ta–2Nb (TTN alloy) in particular is being employed in the reprocessing of spent nuclear fuel as a structural material in vessels for dissolvers, evaporators and waste storage tanks, where it is exposed to highly oxidising and corrosive conditions.1,9) The commercial production of TTN alloy is by a multistep process that comprises compacting highly pure Ti sponge and Ta and Nb sheets, arc melting thrice, and then casting. This is followed by hot extrusion into rods, cold swaging, annealing, and finally rolling into sheets.10) TTN alloy has been studied extensively with respect to its microstructure, mechanical properties and corrosion behaviour in the wrought and welded states.11–16)
The conventional production of TTN alloy requires high-purity metals of Ti, Ta and Nb, the production of which involves chlorides and fluorides at high temperatures and is laborious, energy intensive and environmentally unfriendly.17) Various economically viable and greener technologies to produce Ti and other reactive metals are currently being researched and developed.18) One of these is the FFC-Cambridge process,19) in which a metal oxide is made the cathode in a bath of molten CaCl2 electrolyte against a carbon anode, such that the O2− ions are expelled from the oxide into the melt without the electrolyte undergoing electrolysis. This electro-deoxidation process is explored in the world over for various metals including Ti, Zr, Hf, Nb, Ta, Si and U.20–22) The process is particularly favourable for the preparation of alloys that are difficult to achieve via classical routes, for instance, because the individual metals have high melting points, different liquid state densities, or large sensitivity to oxidation. Key examples are the synthesis of alloy powders of Ti/Nb,23) Ti/Ni,24) Ti/W,25) Ti/Mo,26) and Nb/Ta.27) The process also offers the opportunity of preparing three-dimensional metallic bodies, by using solid oxide bodies of defined geometry as the starting material and maintaining this geometry throughout the reduction. Examples are the near-net-shape fabrication of solid bodies of α-Zr and α-Zr/Nb,28) (α+β)-Ti/Zr,29) Ti–6Al–4V,30) and β-Ti/Nb.31)
Heat treatment of Ti and its alloys is key to establishing the ratio between the low-temperature α-phase and the high-temperature β-phase as well as the microstructure.32,33) Single-phase α-alloys are only used in applications that focus on corrosion resistance but are not particularly demanding in terms of strength. In contrast, dual-phase (α+β)-alloys offer better mechanical properties. They are therefore more appropriate for structural applications, due to their higher strength and toughness, and for biomedical applications, due to their lower elastic modulus, and they are also suitable at elevated temperatures up to about 773 K. TTN alloy is a near α-phase Ti alloy, in which typically about 8–15% of the β-phase, enriched with the Ta and Nb, is retained after heat treatment.1,16) It thus falls into the dual-phase category.
The goal of the present work has been to demonstrate the feasibility of TTN alloy preparation from the corresponding mixed metal oxide precursor via the FFC-Cambridge process, and to study the influence of various factors on the electrochemical reduction, such as precursor morphology, and voltage and duration of reduction. Also investigated has been the impact of a specific heat treatment on the alloy’s microhardness.
Anhydrous CaCl2 was prepared from commercially available CaCl2·2H2O (Sigma Aldrich) by first drying at 443 K in air for 48 h and subsequent drying at the same temperature in vacuum for 48 h. Powders of TiO2, Ta2O5 and Nb2O5 (TTN oxide) were mixed in the ratio required for the synthesis of Ti–5Nb–2Ta (TTN alloy, numbers in mass%). The properties of the oxides, including particle size and crystallographic phase, are listed in Table 1. The powder was made into a slurry in isopropanol with added 1 mass% of polyvinyl alcohol and 0.5 mass% of polyethylene glycol as binder and plasticiser. Dry TTN oxide powder was obtained by evaporating the solvent and pressed into discs of either 1 g in mass and 13 mm in diameter or 3 g and 25 mm. These were sintered in air at 1223, 1323 and 1373 K for 3 h.
For each experiment about 500 g of dried CaCl2 salt was filled into an alumina crucible and placed inside an Inconel retort of which the top flange had provisions for gastight insertion of electrodes. The assembly was heated to the operating temperature of 1173 K in an electrical tube furnace and continuously flushed with argon. The CaCl2 melt was pre-electrolysed to remove redox impurities, using a Ni coil cathode and a graphite rod anode of 10 mm diameter. This was done at an applied DC voltage of 2.8 V for typically 3–12 h. Electro-deoxidation experiments were then performed, using a TTN oxide disc, tied to a stainless steel current collector, as the cathode and the same graphite rod as the anode. The geometric surface area of the fully immersed oxide discs was between 2.2–2.7 cm3, and the surface area of the immersed section of the graphite rod was about 15 cm3. The experiments were done at applied DC voltages of 2.5 and 3.1 V for durations of 3, 6, 12, 18 and 22 h. A DC power supply (Aim-TTi PL154) and a data acquisition system (Agilent 34970A) were used to conduct the experiments. After completion of each experiment, the electrodes were lifted above the melt and the retort was allowed to cool. The processed samples were rinsed with distilled water and dried in vacuum.
The sintered TTN oxide precursor discs and the electrochemically prepared TTN alloy samples were both analysed with respect to their phase compositions by means of X-ray diffraction (XRD) (Philips PW 1830). Their morphologies were assessed through scanning electron microscopy (SEM) (Nova NanoSEM 450) of fracture and polished surfaces, and local chemical compositions were determined through energy-dispersive X-ray spectroscopy (EDX) (Bruker X Flash 6I100). Open porosities were determined by Archimedes’ method. The residual oxygen contents of the TTN alloy samples were determined with the hot extraction method (ELTRA ONH2000), using about 50 mg in three separate tests.
The reduced TTN alloy samples were subjected to a defined heat treatment. To that end, the reduced samples were ground with SiC paper, and fragments of them were wrapped with Ta foil and vacuum-sealed into quartz ampoules. The sealed samples were heat treated at 1473 K for 3 h using heating and cooling rates of 5 K/min. After that the samples were characterised again by XRD and SEM/EDX, now using secondary and backscattered electrons. Microhardness measurements were done before and after the heat treatment with a Vickers indenter (Laizhou Huayin 200HV-5) along polished cross-sections, by applying a load of 300 g for a dwell of 15 s and averaging at least ten measurements. Light optical microscopy (Olympus BHM) was used to visualise the indentations.
Figures 1(a)–(c) display the XRD patterns of the TTN oxide precursors sintered in air at 1223, 1323 and 1373 K for 3 h. It was confirmed that each sample contained TiO2 in the rutile phase and Ta2O5. It is known that TiO2 and Nb2O5 react during sintering to form TiNb2O7,34) but this phase was not seen probably owing to its relatively small proportion. A small degree of shrinkage was found to occur with increasing sintering temperature. The insets of the figure are photographs of the sintered oxide discs.
XRD patterns of TiO2–Ta2O5–Nb2O5 oxide precursors sintered in air for 3 h at (a) 1223 K, (b) 1323 K, and (c) 1373 K. SEM images of the same oxide precursors sintered at (d) 1223 K, (e) 1323 K, and (f) 1373 K. Insets of (a), (b) and (c) show the sintered oxide discs, with diameters and thicknesses of, respectively, 12.7 and 3.5 mm, 12.2 and 3.3 mm, and 11.7 and 3.2 mm.
Figures 1(d)–(f) show SEM images of the cross-sections of the same TTN oxide precursors. The sample sintered in air at 1223 K, was composed of mainly spherical particles with size of around 150 nm (Fig. 1(d)). When the temperature was increased to 1323 K, particles grew to around 400 nm and neck formation commenced between some of the particles (Fig. 1(e)). At 1373 K, particles grew further, with some remaining more spherical with size of around 600 nm and others showing enhanced necking with size of more than 1000 nm (Fig. 1(f)). With increasing temperature, the open porosity of the precursors was measured as 43%, 36% and 20%, and this was accompanied by noticeable shrinkage of the disc diameter.
3.2 Electro-deoxidation of mixed oxide precursor 3.2.1 Variation of applied potential and temperatureFigure 2(a) presents the current versus time curves recorded during the electro-deoxidation of 1 g TTN oxide precursors sintered at 1373 K under three different sets of conditions, viz. 2.5 V and 1173 K, 3.1 V and 1098 K, and 3.1 V and 1173 K, all for a duration of 18 h. In each case, the current peaked initially, then dropped off, and finally approached a constant background value, as it had likewise been observed for the reduction of pure TiO2.35,36) As expected, the magnitude of the current increased with increase in applied potential and electro-deoxidation temperature. The insets of the figure are photographs of the processed samples and show that these were in the form of powder, fragments, or a complete disc.
(a) Current vs time curves during electro-deoxidation of TiO2–Ta2O5–Nb2O5 precursor discs of 1 g (sintered in air at 1373 K for 3 h) in CaCl2 melt at different applied voltages and temperatures for 18 h. Insets show the reduced samples. XRD patterns of reduced samples at (b) 2.5 V and 1173 K, (c) 3.1 V and 1098 K, and (d) 3.1 V and 1173 K. SEM images of reduced samples at (e) 2.5 V and 1173 K, (f) 3.1 V and 1098 K, and (g) 3.1 V and 1173 K. (h) High-magnification SEM image of reduced sample shown in (e), and (i) low-magnification SEM image of reduced sample shown in (g) with EDX analysis. (j) EDX area maps of reduced sample shown in (g).
Figures 2(b)–(d) display the XRD patterns of processed TTN samples, recorded from either powder samples or polished surfaces. For 2.5 V and 1173 K, the main phases observed were α-Ti, CaTiO3 and CaCO3. The CaTiO3 indicated that the TTN oxide precursor was only partially reduced, while the CaCO3 pointed to the presence of CaO in the sample. For 3.1 V and both 1098 and 1173 K, the only phase seen was α-Ti. This proved that complete reduction had occurred. The peak positions in the measured and the reference XRD patterns were identical, indicating that the quantities of Ta and Nb dissolved in the Ti had no influence on the lattice parameters.
Figures 2(e)–(i) show SEM images of processed TTN samples. For 2.5 V and 1173 K, the material was composed of comparatively small particles on the order of 0.5 µm with nearly spherical and nearly cubic shapes (Figs. 2(e), (h)). In line with earlier studies, and as will be discussed further below, the more spherical particles were suboxidic and metallic particles, while the more cubic and facetted particles were Ca-containing intermediates, such as CaTiO3 in case of Ti.37) For 3.1 V and 1098 K, the material exhibited a nodular microstructure in which the individual particles had coalesced into slightly larger ones (Fig. 2(f)). For 3.1 V and 1173 K, the material was in the form of a continuous nodular sponge with large particles on the order of 20 µm. The microstructure indicated complete reduction, and the coarsening pointed to significant in-situ sintering (Figs. 2(g), (i)). Figure 2(j) shows EDX area maps for Ti, Ta and Nb that confirm the homogeneous distribution of these elements. Quantitative EDX analysis verified that the alloy composition was close to the target value.
Similar current versus time curves, phase compositions, microstructures and chemical compositions were also observed for the electro-deoxidation of TTN oxide precursors sintered at 1223 and 1323 K (not shown).
The porosity of the reduced TTN samples was dependent on the sintering temperature of the oxide precursor and the electro-deoxidation time. Quantitatively, porosities ranged within 27–44% for oxide sintering temperatures within 1223–1373 K, and a higher porosity was typically accompanied by a slightly smaller average particle size. The residual oxygen contents of all samples processed at 3.1 V varied within 2050–4440 ppm.
3.2.2 Variation of duration and mechanistic considerationsFigure 3(a) presents the current versus time curves from the electro-deoxidation of 1 g TTN oxide precursors sintered at 1373 K for three different durations, viz. 3, 6 and 12 h, all at 3.1 V and 1173 K. The curves were similar to the ones described earlier. The insets are photographs of the processed samples and show that these were either fragmented or complete discs. Visual examination of the sample recovered after 3 h revealed that there was a clear distinction between an outer layer and an inner core. These could readily be separated manually and then investigated individually. The sample obtained after 6 h was similar, with a comparatively thicker outer layer and a thinner core. The sample after 12 h exhibited no such separation.
(a) Current vs time curves during electro-deoxidation of TiO2–Ta2O5–Nb2O5 precursor discs of 1 g (sintered in air at 1373 K for 3 h) in CaCl2 melt for different durations at 3.1 V and 1173 K. Insets show the reduced samples. XRD patterns of (b) outer region of sample polarised for 3 h, (c) inner region of sample polarised for 3 h, (d) sample polarised for 6 h, and (e) sample polarised for 12 h.
Figures 3(b)–(e) display the XRD patterns of the three processed samples. The 3 h sample contained the α-Ti and CaTiO3 phases in the outer region (Fig. 3(b)), and the TiO, Ti2O3 and CaTiO3 phases in the inner region (Fig. 3(c)). The 6 h sample consisted of the α-Ti and CaTiO3 phases (Fig. 3(d)), while no suboxide phases could be resolved. The 12 h sample had the α-Ti phase alone (Fig. 3(e)).
Figure 4(a) shows a low-magnification SEM image of the cross-section of the 3 h sample, which demonstrates that the inner and outer regions were clearly discernible. Figure 4(b) is the corresponding EDX line scan, which revealed that the inner region had Ti and Ca as the main constituents together with minor amounts of Ta and Nb, and that the outer region had Ti alone as the main constituent together with some Ta and Nb and little Ca. These findings are in agreement with the XRD results and indicate that reduction advanced faster in the outer region while Ca metallates were retained in the inner region. The average thickness of the outer layer increased with time, and the inner core disappeared within 6 to 12 h when no CaTiO3 was found anymore.
(a) Low-magnification SEM image of cross-section of sample polarised at 3.1 V and 1173 K for 3 h, with (b) corresponding EDX line scan. SEM images of (c) inner region of sample polarised at 3.1 V and 1173 K for 3 h, and (d) inner region of sample polarised under the same conditions for 6 h, with EDX analysis of (e) the area indicated in (c), and (f) the area indicated in (d). SEM images of (g) outer region of sample polarised at 3.1 V and 1173 K for 3 h, (h) outer region of sample polarised under the same conditions for 6 h, and (i) sample polarised under the same conditions for 12 h.
Figures 4(c), (d) show SEM images of the inner regions of the 3 and 6 h samples, which were characterised by particles of different shapes, some of them facetted. As reduction advanced, the size of these particles decreased. Figures 4(e), (f) are EDX results that suggest that the inner regions contained CaTiO3 and metal suboxides. The individual particles could be identified based on the results of an in-depth study into the electro-deoxidation of TiO2, which had shown that Ca metallate particles are indeed distinctly facetted while the suboxide particles became more spherical as their oxygen content lowered.37) Figures 4(g)–(i) show SEM images of the outer regions of the 3 and 6 h samples and of the fully reduced 12 h sample, which were characterised by nodular particles. This is expected for Ti sponge prepared by electro-deoxidation.35,36,38) The average particle size in the samples grew from slightly below 1 µm (Fig. 4(g)), via around 2–3 µm (Fig. 4(h)), to 3–5 µm (Fig. 4(i)) and was always found to be larger closer to the surface. This is indicative of in-situ sintering via particle coalescence. Similar observations had been made in the reduction of UO239) and Nb2O5.40) Figure 5 shows a high-magnification SEM image of a fully reduced sample. It reveals that the large-angle boundaries of the particles remained clearly visible after coalescence and in-situ sintering. This is again similar to the findings for pure Ti.35,36,38)
High-magnification SEM image of fully reduced Ti–5Ta–2Nb alloy, prepared by electro-deoxidation of TiO2–Ta2O5–Nb2O5 precursor disc of 3 g (sintered in air at 1223 K for 3 h) in CaCl2 melt at 3.1 V and 1173 K for 22 h, showing grain boundaries between individual particles after coalescence and in-situ sintering.
The electro-deoxidation of almost all metal oxides proceeds in a series of steps, often involving the transient formation and decomposition of Ca metallates. In-depth studies have established the following kinetic pathways for TiO2, Ta2O5 and Nb2O5.
Figure 6(a) displays the XRD pattern of a reduced TTN alloy sample after heat treatment at 1473 K in vacuum for 3 h. While before heating only the peaks of the α-Ti phase could be resolved, now several peaks pertaining to the β-Ti phase were seen too. This implies that dual-phase (α+β)-Ti with a small but significant proportion of the β-phase had been formed. This is because the applied temperature was about 300 K above the β-transus11) and because slow cooling is known to retain a proportion of the high-temperature phase without giving rise to martensite formation.11) The inset of the figure is the section of the XRD trace with the strongest β-phase peak.
(a) XRD pattern of Ti–5Ta–2Nb alloy after heat treatment at 1473 K in vacuum for 3 h. Inset shows section with characteristic β-phase peak. SEM images of fracture cross-sections of Ti–5Ta–2Nb alloy in (b) as-reduced state, and (c) heat-treated state. Backscattered SEM images of polished cross-sections of Ti–5Ta–2Nb alloy in (d) as-reduced state and (e) heat-treated state. Insets show light optical microscopic images of Vickers indentations. EDX area maps of the alloy in (f) as-reduced state, and (g) heat-treated state. EDX analysis of (h) the area shown in (f), and (i) the area shown in (g).
Figures 6(b), (c) show SEM images of a fracture surface of a reduced TTN alloy sample before and after heat treatment. The effect of the heat treatment on the microstructure is most obvious in that there were bigger particles and less porosity afterwards. Porosity of the heat-treated sample was measured to be about 15%, which was significantly below the initial porosity of the as-reduced sample of 27%. This is because softening of metals and associated microstructural changes typically progress rapidly at temperatures higher than 50–60% of the melting temperature. Given that the melting point of TTN alloy will be similar to that of pure Ti, this is at about 970–1165 K and thus lower than that of the heat treatment.
Figures 6(d), (e) show backscattered SEM images of a polished surface of a reduced TTN alloy sample before and after heat treatment. In the former case the microstructure had a uniform appearance, while in the latter case it clearly comprised two phases. With a view to the XRD results, it is evident that the main phase was α-Ti, and that the secondary phase was β-Ti that had precipitated in the form of micron-sized particles along the boundaries of the much bigger α-Ti particles during the heat treatment and was then retained upon cooling. Figures 6(f), (g) show EDX area maps for Ti, Ta and Nb before and after heat treatment. In the former case the distribution of these elements was uniform, while in the latter case the β-phase was enriched with Ta and Nb that are both β-stabilisers. Quantitative EDX data are given in Figs. 6(h), (i).
The microhardness of the as-reduced TTN alloy sample was measured as 115 HV0.3, while the microhardness of the same sample after heat treatment was significantly higher at 176 HV0.3. This increase of hardness is known to result from the generation of the (α+β)-type microstructure.10,11) A further small contribution will have originated from the decrease in porosity because the relationship between hardness and porosity is of an inverse nature. The insets of Figs. 6(d), (e) are light optical images of Vickers indentations on the sample surfaces, which show that these gave clean impressions without cracks or irregular deformation at the edges and corners.
The measured hardness for the present TTN alloy of 176 HV0.3 is in accordance with the hardness of conventionally prepared TTN alloys of the same composition of 210 HV0.1, when considering that the latter were nearly dense.15) With the common approximation, that hardness expressed in MPa and ultimate tensile strength are in a ratio of 1 to 3, measured data for both quantities can be compared directly. This shows that the hardness of the present TTN alloy agrees well with that derived from strength tests on conventionally prepared alloy of 205 HV.10) Hardness of a Ti alloy is sensitive to the oxygen content, but the effect is difficult to quantify for TTN alloy due to scarcity of data. It is evident however that the hardness of the present TTN alloy is also consistent with the tensile strengths of cp Ti grades 3 and 4, with allowed oxygen contents between 2500 and 4000 ppm.43) All quantitative information is compiled in Table 2.
Ti–5Ta–2Nb alloy was synthesised from sintered oxide mixtures of TiO2–Ta2O5–Nb2O5 by electrochemical deoxidation in a CaCl2 melt, following the concept of the FFC-Cambridge process. The preferred conditions were an applied voltage of 3.1 V and an operating temperature of 1173 K. Complete deoxidation was achieved within 12 h under these conditions, and the processed samples retained their disc-type shape. The alloy was of target composition with a uniform distribution of the three elements, and the residual oxygen content was within 2050–4440 ppm which is an acceptable level for Ti and its alloys. Partially reduced samples contained significant quantities of CaTiO3 and Ti suboxides, indicating that the reduction pathway was in several respects similar to that of pure TiO2.
Heat treatment in vacuum transformed the as-reduced single-phase α-Ti into dual-phase (α+β)-Ti and increased microhardness by more than 50%. The values obtained were close to those reported in the literature for samples prepared via conventional routes. Considering that the volume fraction of the β-phase depends critically on the type and quantity of the alloying elements and on the conditions during the heat treatment, there is significant scope for further optimisation of the hardness and associated mechanical properties of this and similar Ti alloys. Overall, the present study suggests a promising route for the affordable production and subsequent optimisation of porous bodies of α-phase and (α+β)-phase Ti alloys for a range of structural and refractory applications.
This study was funded through the Research Chair Grant Program of The Research Council of the Sultanate of Oman.
