Microstructural characterization of U-7.5Nb-2.5Zr alloy after ageing and constrained fatigue

U-7.5Nb-2.5Zr cladded in Zircaloy-4 is one of the most studied fuel prototypes. The Nb and Zr are added to the U to stabilize the body-centered cubic (BCC) gamma phase and grant mechanical and swelling resistance. The U-7.5Nb-2.5Zr undergoes the gamma to α′′ phase transformation, generating compressive stresses due to the volume reduction. The α′′ phase also can transform to a combination of α + γ 2 phase (equilibrium phases), which are known to be hard and brittle. This work had the objective to test the effect of ageing the gamma to α′′ phase in thermal cycling of a U-7.5Nb-2.5Zr cladded in Zircaloy-4 part. A co-laminated specimen was aged and thermally cycled in a dilatometry experiment. The samples were characterized through X-ray diffraction and metallography. The results show that the gamma to α′′ phase transformations occur at the beginning of the initial ageing, followed by stress relief. During the cycling, the sample demonstrated non-equal thermal strains and presented fractures along with the U matrix.


Introduction
Metallic uranium has high density, which allows to increase the amount of U 235 inside the reactor core without the necessity of increase the enrichment which contributes to the non-proliferation policies 1 .Metallic U also presents better thermal conductivity compared to traditional ceramic fuels like UO 2 or the ones like UN 2 .
Pure metallic uranium presents three allotropic forms.At the room temperature, it presents itself in the base-centered orthorhombic structure (U α ).The BCO structure presents high plasticity mainly deforming by twinning 3 , but this structure also presents thermal expansion anisotropy, for example, 'a' and 'c' crystalline axis expands while 'b' axis shrinks.This behavior causes severe dimensional instability 4 .At 661 • C the pure U transforms in to a tetragonal structure (U β ), undergoing a 1% contraction.The tetragonal phase is hard to stabilize and the literature on this topic is scarce.Finally, at 769 • C, metallic U transforms into a body-centered cubic (BCC) structure (U γ ) undergoing a 0.6% volume contraction 5 .
In order to solve the dimensional instability problem, alloying elements like Nb and Zr are added to U 6 .The addition of circa of 9 mass% of Nb 7 or 20 mass% of Zr 8 individually are capable of stabilize the BCC (U γ ).Both elements also enhances other properties, like Nb enhances the plasticity of the (U γ ) and the Zr increases the swelling resistance, making combinations of Nb and Zr a good solution to avoid high alloying, which reduces the alloy density, and combining the best of their properties 6 .
The addition of alloying elements also brings the possibility new phase transformations to metallic U. Low Nb additions (≤ 3%) inhibits the formation of the (U β ) and causes the formation of the U α ′ phase, a martensitic distortion on the (U α ) phase.Nb additions about 5.8% can lead to the precipitation of another martensitic phases, the U α ′′ , a monoclinic distortion of the orthorhom-bic (U α ) phase and the U γ 0 phase, a tetragonal distortion of the (U γ ) BCC phase 9 .Zr additions below 20 mass% also causes the precipitation of martensitic U α ′ phase 8 .In addition to the martensitic phases, the equilibrium U γ 2 and U γ 3 can precipitate after diffusional decomposition of the (U γ ).Table 1 contains a description of each U phase and system were it occurs.
The U-7.5Nb-2.5Zr,also known as Mulberry alloy, is one the U-Nb-Zr alloys studied as can-didate to fuel research reactors.This alloy presents mechanical properties that can withstand the fabrication process of the alloy and of a monolithic fuel element 11 .
According to the time-temperature-transformation published by Lopes 12 , the Mulberry alloy can suffer the phase transformations γ → α′′ and γ → α′ + γ 3 .Both phase transformations should present volume shrinkage 12 .The first transformation is expected to occur between 390 • C and 400 • C after 10 minutes, the second transformation should occur after 100 minutes.Materials Research

Monolithic fuels
The monolithic fuels consist in a foil of U alloy cladded in Zr or Al alloys.The U alloy works as fuel and the cladding protects it from the fluid on the primary circuit of a nuclear reactor.The fabrication of the monolithic fuels is performed by hot isostatic pressing (HIP) 13 or by hot rolling the fuel alloy and the cladding alloy in order to promote diffusion welding 14 .Since the U and Zr have complete solubility in the (U γ ) phase, the hot rolling is performed above the (U γ ) formation temperature, but below the transus temperature of Zr to avoid the Zr β → Zr α reaction, which causes volume shrinkage and could lead to crack formation at the interface between the fuel and the cladding.
Once both alloys are bind, they must withstand the mechanical and thermal stresses together, but they have different linear thermal expansion coefficients, which can cause thermal fatigue.The open literature on thermal fatigue on monolithic fuels is scarse.Lopes performed thermal cycling of a U-10Mo foil cladded in Zircaloy-4 and observed that the alloy fracture due the strain hardening caused by the constraining 15 .The (U γ ) U-10Mo is stable at the cycled temperature range, so no phase transformation was observed by Lopes.Since there is little information on thermal fatigue of monolithic prototypes, this work presents a result of thermal cycling after ageing a of Mulberry/Zircaloy-4 monolithic fuel prototype.

Sample preparation
One U-7.5Nb-2.5Zralloy slug was Vacuum Induction melted in a carbon crucible using as raw materials natural metallic U (U 235 < 0.7%, with major impurities: 147 ppm Mn , 95 ppm Ni, 47 ppm Cu, 19 ppm Fe and 15 ppm O), sponge Zr (major impurities: 0.166 ppm Fe and 0.022 ppm Hf, 2 ppm O) and high purity Nb (major impurities: 0.14 ppm Ta, 25 ppm O).The melting procedure was similar to the one reported in the previous works of the group [16][17][18][19][20] .The molten alloy was then cast into a copper mold.The chemical composition of the obtained alloy was determined as U-7.4Nb-2.3Zrwas measured through inductively coupled plasma -optical emission spectrometry (ICP-OES).After the casting, the samples were homogenized at 1000 • C for 5h in high purity Argon atmosphere, followed by water quenching in order to stabilize the γ phase.
The Zircaloy-4 frame cladding was machined by wire electro-erosion.The covers and the frame were polished to 1 µm diamond paste.The Mulberry alloy was also polished to 1 µm diamond paste and stored in analytic ethyl alcohol.The set was assembled and welded using auto-genous gas tungsten arc welding (GTAW).The assembly was cladded in a colloidal carbon painted steel frame.The sample was hot rolled together with the Zircaloy-4 sheet using the procedure re-ported by Lopes 21 , in order to fabricate a prototype of co-rolled plate.

Thermal cycling
After the hot rolling, the plate was cut in to two 5 mm x 5 mm x 2 mm pieces.The sample con-sists a 'sandwich' of U-7.5Nb-2.5Zr in Zircaloy-4, this design alloys the sample to freely expands and shrink in the normal direction, since there is no constraining in that direction.One piece was used as control sample, the other sample was placed on the thermomechanical analyzer (TMA) instrument Setaram Multi HTC (Setaram 20 -1600 • C) aged for 24 h at 400 • C, then thermally cycled.
The cycles were carried in the interval 200 • to 400 • under high purity Ar atmosphere in a 5 K/min heating/cooling rate.The actuator of the TMA equipment was placed in contact with the Zircaloy-4 in the normal direction.The heating and cooling rate chosen in the thermal cycling allows to compare the obtained results with other dilatometric results in the literature, such the presented by Lopes 11 .

Sample characterization
The treated samples were cold mounted in polymethyl methacrylate (PMMA) resin for surface preparation.Microstructural characterization was performed using optical microscopy.Samples were ground with SiC paper and mechanically polished with 1 µm diamond paste.Electropolish-ing was then carried out using an electrolyte containing 10% oxalic acid solution in water for 5 s using 2 V potential and using stainless steel as the cathode material.The same electrolyte was used for etching using 5 V potential at room temperature.
X-ray diffraction (XRD) was performed using a diffractometer with linear detector.The measurements were performed at room temperature with filtered Cu Kα radiation (λ = 0.15418 nm) in the θ /2θ mode, with 2θ range between 30º to 55º, with a 0.03º step.The U phases were identified U-Nb, U-Nb-Zr after diffusional decomposition of the Uγ using models suggested by Lopes, and Rietveld refined to find a match in the XRD results 22 , the Zr phases were identified using the crystallographic information file (CIF) proposed by Wyckoff 23 .Both phase identification and the peak indexation were performed using the GSAS software based on data available in literature 24 .

Microstructural characterization of the initial condition
Figure 1a shows the initial microstructure on the interface of the of the control sample and the Figure 1b shows the initial microstructure of U matrix.The equiaxial Uγ grains shows uniform bounding with the Zircaloy-4 plate.No cracks are perceptible neither on the U matrix, interface or on the Zircaloy-4.Some inclusions are perceptible on the U matrix.The Figure 2 shows the XRD pattern for the initial condition.
The diffractogram shows the presence of the BCC U γ phase and the HCP Zr α phase.
The initial microstructure is typical compared to hot rolled U γ U-Nb-Zr alloys, as it can be seen in the literature 17,18 .No secondary phase can be seen in the uranium matrix.The X-ray diffraction confirms that there is only Zr α and U γ phases before the thermal cycling.The Table 2 shows the lattice parameters of the identified phases of the initial condition.

Thermal ageing fatigue
Figure 3a shows the relative displacement of the sample during the ageing.At the beginning of the ageing, there is a strong shrinkage due the γ → α′′ phase transformation.This phase transformation is expected and this behavior was previously described on the literature 11 .After 17 h, the sample presents a slight expansion (∆L/L 0 < 0.001).In the time-temperature-transformation diagram proposed by Lopes 12 , the U-7.5Nb-2.5Zralloy should undergo the γ → α′ + γ 3 phase transformation at 400 • C which would cause another shrinkage so, this expansion may be related to a stress relieving on the assembly.
Figure 3b shows the relative displacement of thermal fatigue.The red lines indicate the furnace temperature and the black lines indicates the relative displacement.It is important to remember that the beginning of the cycle is a cooling step, going from 400 • C to 200 • C, which will cause the sample to shrink.
Since there is no constraining in the normal direction, the fatigue process occurs on the transversal and in the normal direction (shearing).Zircaloy-4, in the hexagonal close packed (HCP) structure, has a linear thermal expansion coefficient (α T ) of about 6×10 −6 K −1 25 .The linear thermal expansion coefficient of the Mulberry alloy is not reported on the literature, for instance of comparison, the linear thermal coefficient to the U γ of the U-10 mass % Mo alloy Table 2. Lattice parameters of the identified phases on x-ray diffraction of the initial condition sample.

Phase
Lattice parameter "a" Lattice parameter "c" Space group  In both cases (U γ or (U α ′′ ), the thermal cy-cling would cause one material expand more than the other, creating the shearing stress to nucleate and propagate cracks.
In order to better visualize the difference between cycles, Figure 3c shows the amplitude of the cooling, heating and the residual strain after each complete cycle.Only cycles 5 and 6 presents zero residual strain.From the first to the last cycle, there is the tendency of negative residual strain, which indicated that the sample expanded a little more at each cycle, which could be indicative of the crack propagation.After the final cycle, the sample conserves a -0.0085 mm shrinkage.

Microstructural characterization of the final condition
Figure 4a presents the microstructure of the U matrix and Figure 4b presents the microstructure on the interface after the cycling.The U matrix presents a long crack perpendicular to the normal direction and a fully transformed U α ′′ matrix.The interface shows no visible cracks in the Zircaloy-4 or in the interdiffusion layer, but show some cracks on the U matrix.
The microstructures of U α ′′ can present itself in the accicular 18 , banded and discontinuous morphology 12 depending the alloy composition and the process that generates it.The presented microstructure corresponds to the accicular form, but presenting distorted deformation bands, sheared on the horizontal direction in reference of the image, indicating that the structure was submitted to shearing and/or tearing stresses during the thermal cycling.
Based on the microstructure and in the dilatometric results, the stresses during the ageing and thermal cycling can be exemplified in the Figure 5.The initial phase transformation introduces a strong compressing stress in the normal direction.This compressing stress hinders the crack open-ing in the normal direction, the stress release generates the expansion before the cycling.Since there is no constraining in the normal direction, there is no stresses applied on the normal direction during the thermal cycling.The thermal expansion and thermal shrinkage applies only shearing and tearing stresses on the  sample.Since the Zircaloy-4 has greater linear thermal expansion than the U α ′′ in the matrix, it might generate cumulative strain hardening, as proposed by Lopes 21 to happen in the U γ phase of the U-10Mo alloy in similar conditions.Once the matrix reaches the limit of strain hardening, the stresses causes the crack nucleation and propagation.
Figure 6 shows the superposition of the X-Ray diffraction pattern of the control sample and of the cycled sample.The Zr α peaks shows displacement, but does not present any of the Zr β peaks.The displacement could be related to the residual strain due the cycling.The results shows strong U α ′′ peaks, but shows no peaks for U α ′ or any U γ variation, indicating that the cycling process occurred in the U α ′′ phase.
The Table 3 shows the lattice parameters calculated for Uα′′ phase.As expected, it varies from the literature reference, probably due the stresses on the structure generated by the thermal cycling.

Conclusions
This work performed the ageing of a Mulberry alloy cladded in Zircaloy-4 at 400 • C for 24 h followed by 10 thermal cycles in the 400 • C to 200 • C temperature range.The objective of this work was evaluate how the assembly would perform under the stress provided by difference of the linear thermal expansion coefficients.The following conclusions can be drawn: • The ageing at 400 • C for 24 h caused a complete γ → α′′ phase transformation, but was not enough to cause the α′′ → α′ + γ 3 reaction; • In the presented conditions ageing cycling the cracks tended to initiate perpendicular to normal direction on the U matrix, but no crack was visible on the Zircaloy-4 or in the inter-diffusion layer; • The amplitude variation can be an indicative of crack propagation.

Figure 3 .
Figure 3. Results from1t0he cycling experiments.(a) Relative displacement (∆L/L0) for the ageing heat treatment; (b) Relative displacement (∆L/L0) for the thermal cycling; (c) Amplitude of each heating and cooling cycle.The blue dots represent the residual strain after each cycle.

Figure 5 .
Figure 5. Stresses during the ageing and thermal cycling on the sample.

Figure 6 .
Figure 6.X-Ray diffraction pattern after the cycling.

Table 3 .
Lattice parameters of the U α ′′ after the thermal cycling.