Nb-Al Binary System: Reevaluation of the Solubility Limits of the (Nb), Nb 3 Al, Nb 2 Al and NbAl 3 Phases at High Temperatures

In this work a re-investigation of the solubility limits of the (Nb), Nb 3 Al, Nb 2 Al and the Nb-rich side of the NbAl 3 phases on the Nb-Al system is presented. Alloys in the binary fields ((Nb)+Nb 3 Al, Nb 3 Al+Nb 2 Al and Nb 2 Al+NbAl 3 ) were arc-melted, and then equilibrated at 1000, 1200 and 1400 °C. The phases were confirmed via X-ray powder diffractometry, and their compositions were determined via EPMA measurements. The results showed agreement with the literature concerning the solubility limits of (Nb), Nb 3 Al and NbAl 3 phases, while important differences in the values were found for the Nb 2 Al phase. In addition, the lattice parameters of the Nb 2 Al phase were determined via Rietveld refinement. This new set of more accurate experimental information indicates that a thermodynamic reassessment is necessary to precisely describe this system.


Introduction
Accurate description of binaries and ternaries phase diagrams is of fundamental importance for the development of thermodynamic databases, useful to predict phase relations, and to define processing conditions for multicomponent alloys. Investigations carried out in our group have contributed to better description of phase diagrams for several binaries systems [1][2][3][4][5][6] . Based on inconsistencies between the most recent assessments 7,8 as well as data from heat-treated ternary alloys [8][9][10][11] containing Nb and Al, new investigations on the solubility limits of the intermetallic phases of the Nb-Al system are necessary. Thus, in this work, the solubility limits of the (Nb), Nb 3 Al, Nb 2 Al and the Nb-rich side of the NbAl 3 phase were reevaluated via Electron Probe Microanalysis -Wavelength dispersive spectrometry (EPMA -WDS) from equilibrated alloys. Table 1 summarizes the experimental information available for the Nb-Al system concerning solidus/liquidus temperatures, phase solubility range, activity data and enthalpy of formation.

Phase equilibria data
Due to its importance for the development of superconductors and high-temperature materials, many authors have investigated the Nb-Al phase diagram. The early studies of this system [12][13][14][15][16][17][18][19] are all in relatively good agreement in terms of phase stability. Besides the terminal compounds, the Nb 3 Al (A15), Nb 2 Al (σ) and NbAl 3 (D0 22 ) phases are reported as stable (see Table 2 for crystallographic structural information), however, Richards 15 indicates the presence of 2 extra high temperature phases (Nb 7 Al 3 and Nb 17 Al 3 ) which have not been reported by other authors. The congruent formation of NbAl 3 is well established, however, there have been some discrepancies in terms of the nature of formation of the other phases. For example, Nb 3 Al is reported to be formed either peritectoidically 15,17 or peritectically 14,16,19 , while Nb 2 Al is reported to be formed either congruently 15,19 or peritectically 14,16,17 . The Al-rich side of this system is characterized by a degenerated equilibrium in which the Liquid, NbAl 3 and (Al) phases are involved. This invariant reaction has been reported either as eutectic 14,15 or peritectic 12,13,16 .
The most complete experimental work on the Nb-Al system was carried out by Jorda et al. 20 . In this paper, the authors determined the phases' solubilities ranges via metallography, XRD and EPMA analysis of samples heat-treated from 24 hours up to 1 month according to the temperature of heat treatment. The authors also used levitation thermal analysis (LTA) and differential thermal analysis (DTA) to determine the temperature of the invariant reactions, solidus and liquidus, and the peritectic nature for the Nb 3 Al and Nb 2 Al formation

Enthalpy of Formation of Intermetallic Phases
Colinet et al. 27 LMTO-FP De Boer et al. 28 Miedema Model Meschel et al. 29 DRC Shilo et al. 23 Knudsen Effusion George et al. 31 EMF Mahdouk et al. 30 DRC  24 performed Differential Scanning Calorimetry (DSC) measurements with different scanning rates from heat treated samples in order to determine the nature of the Al-rich equilibrium involving Liquid, NbAl 3 and (Al). The suggested temperature was 661.44 °C, leading to a peritectic type reaction because it is higher than the melting point of pure Al (660.3 °C). Witusiewicz et al. 7 performed new experiments (DTA and Pirani-Alterthum method) aiming at the determination of the high temperature solidus and liquidus lines. In general, their results are in good agreement with previous information 20 . Witusiewicz et al. 7 also measured the temperature of the degenerated Al-rich reaction as 657 °C ± 5 (DTA), despite this, they modeled the reaction as peritectic. More recently, Stein et al. 25

Thermodynamic data and CALPHAD modeling
Several studies present estimated data for enthalpies of formation of the Nb-Al compounds based both in calculations as well as on experimental results. Gelashvili and Dzneladze 26 estimated the enthalpies of formation calculating the changes in the free energy of the process of reduction of Al and Nb oxides with CaH 2 . Colinet et al. 27 reported the enthalpies of formation of the intermetallic phases via first principle calculations (Full Potential Linear Muffin Tin Orbital, FP-LMTO) and de Boer et al. 28 31 and Shilo et al. 23 also have measured the activities of Al in the Nb-Al system.
The Nb-Al system was firstly described according to the CALPHAD methodology by Kaufman and Nesor 32 , considering all compounds as stoichiometric. Latter, it was reassessed by Kaufman 33 where the Nb 2 Al phase was modeled as a substitutional solid solution. Subsequent studies have modeled the Nb 3 Al phase either as (Nb) 3 (Al,Nb) 1 8,24,34,35 or (Al,Nb) 3 (Al,Nb) 1 7,36 , the Nb 2 Al phase mostly as (Al,Nb) 5 (Nb) 2 (Nb,Al) 8 and the NbAl 3 phase has been modeled as either stoichiometric 24,34 or (Al,Nb) 1 (Al,Nb) 3 7,8,35,36 . Joubert 37 investigated the Nb 2 Al site occupancy via Rietveld refinement of X-ray diffraction data and Mathieu et al. 38 investigated simplifications for the σ phase sublattice models, evaluating the best agreement with the experimental phase diagram. The model type (Al,Nb) 2 (Al,Nb) 8 (Al,Nb) 5 should be used in order to respect the crystal structure and the nature of the defects in this phase. Table 3 summarizes the sublattices models applied for the description of these intermetallic phases from assessments of different authors.

Experiments Procedure
Alloys with initial masses between 1 and 2 g were weighed on an analytical balance with accuracy of 0.1 mg from high purity raw materials, Al (min. 99.999 wt. %) and Nb (min. 99.8 wt. %)

Arc-melting
The alloys were arc-melted in a water-cooled copper crucible under argon atmosphere (min. 99.995%) and nonconsumable tungsten electrode. Five melting steps were carried out for each alloy to ensure chemical homogeneity, turning the ingots upside-down from one melting step to the next. Before each melting step a piece of pure Ti (getter) was melted to remove residual gas impurities from the furnace atmosphere. After melting, the ingots were weighed to evaluate possible mass losses during arc-melting.

Reference
Sublattice model  Table 3. Sublattice models used for the Nb-Al phases in literature CALPHAD assessments of the Nb-Al system.

Heat treatments
Aiming to reach thermodynamic equilibrium conditions, all alloys where heat treated at 1400 °C for 75 h in a resistive (Ta heating element) furnace under argon. The temperature was measured by an optical pyrometer calibrated against the melting point of pure elements. Subsequently, samples of each alloy were wrapped in thin Ta foil, encapsulated in quartz tubes (under argon) and heat treated at 1200 °C for 200 h or 1000 °C for 600 h using tubular resistive furnaces.

X-ray diffractometry (XRD)
For the X-ray diffraction experiments, the samples were analyzed in powder form, with powder size below 80 mesh (178 µm). The following conditions were adopted: Cu-Kα radiation, 40 kV voltage, 30 mA current, 0.02º angular step, 15s per step, and angle (2θ) ranging from 10 to 90°. The phases present in the samples were identified by comparison between experimental and simulated diffractograms, using PowderCell Software 39 with crystallographic information reported by Villars and Calvert 40 . The lattice parameters for the Nb 2 Al phase were obtained via Rietveld refinement using the software FullProf 41 .

Scanning electron microscopy and electron probe micro-analysis
The samples were prepared according to the following route: (1) hot mounting; (2) manual grinding with SiC sand paper, in the sequence: 220, 400, 600, 1200, 2400, 4000;  Table 4 shows the chemical composition of the prepared alloys, the mass losses associated with the melting steps, and the calculated composition interval for each alloy assuming that all mass losses were either from Nb or Al volatilization. Alloy Nb60Al presented an important mass variation, however, EDS analysis indicated that the global composition of the sample was kept. Thus, it should have occurred due to macroscopic pieces of the alloy that were thrown out of the crucible during cooling inside the arc-melter.  Figure 1 presents the X-ray diffractograms of the alloys equilibrated at 1000, 1200 and 1400 °C. All peaks were identified and only the expected phases are present in the alloys: (Nb) + Nb 3 Al (Nb12Al); Nb 3 Al + Nb 2 Al (Nb23Al); Nb 2 Al+NbAl 3 (Nb60Al). Figure 2 presents SEM micrographs of the heat-treated alloys, with significant microstructural differences from the as-cast condition, indicating that significant diffusion process occurred, and the equilibria was achieved. Alloy Nb12Al in all conditions presented grains of (Nb) and intergranular Nb 3 Al. In the interior of the (Nb) grains, precipitates of Nb 3 Al are observed in the samples equilibrated at 1000 and 1200 °C, but not in the sample equilibrated at 1400 °C. The formation of   Table 5 shows the composition of the alloys measured via EDS and the phases measured via EPMA along with the error which is calculated based on the standard deviation of the measured values. The presence of precipitates in the interior of the (Nb) phase equilibrated at 1000 °C did not allowed reliable EMPA measurements. Results of EPMA measurements are also plotted in Figure 3 along with selected experimental data available in the literature as well as the calculated phase diagrams with the parameters optimized by Witusiewicz et al. 7 and He et al. 8 . The measurements in the (Nb), Nb 3 Al and NbAl 3 phases are in agreement with the literature experimental data as well as the Nb-rich limit of the Nb 2 Al phase. The Al-rich limit of the Nb 2 Al phase exhibits important discrepancies with the most recent assessment.     Figure 5 shows the lattice parameters "a" and "c" for the Nb 2 Al phase in function of temperature, along with the results from Kokot et al. 21 and Joubert 37 . The symbols in black (◄and ) represent the parameters obtained for the Nb-rich side of the Nb 2 Al phase (results from alloys in Nb 3 Al+Nb 2 Al field), while the gray symbols (◄and■) the parameters obtained for the Al-rich side of the phase (results from alloys in Nb 2 Al+NbAl 3 field). It should be stated that Joubert 37 results (symbol •) is from a Nb 2 Al single phase alloy with composition 34.2 at.% of Al. In general, our results are in agreement with Kokot et al. 21 as well as results from Joubert 37 . The presence of excess Al in the structure of Nb 2 Al promoted the decreasing of the parameter "a" and increasing of the parameter "c" of the crystal structure. This can be noted both in a fixed temperature, comparing the Al-rich side with the Nb-rich side as well as the lattices parameters change when the solubility of Al increases with the temperature (Al-rich side).

Conclusions
Experiments aiming to determine the solubility limits of the intermetallic phases in the Nb-Al system confirmed the solubility range for the phases (Nb) and Nb 3 Al. The Nb-rich sides of the phases NbAl 3 , Nb 2 Al were also in agreement with the previous data. Different values from those reported in the literature were found for the Al-rich border of the Nb 2 Al phase. The present experimental results are consistent with the experimental data in the ternary alloy systems containing Nb and Al [8][9][10][11] , suggesting necessary changes in the currently accepted Nb-Al phase diagram and the thermodynamic description of this system, specially concerning the Nb 2 Al phase.