The influence of impurities on the ductility and toughness of a low-temperature-aged U-Nb alloy

Author’s Accepted Manuscript The influence of impurities on the ductility and toughness of a low-temperature-aged U-Nb alloy Dong Chen, Xinjian Zhang, Haoxi Wu, Dingmu Lang, Dawu Xiao, Zhenhong Wang, Bin Su, Daqiao Meng www.elsevier.com/locate/msea

PII: DOI: Reference:

S0921-5093(18)31311-X https://doi.org/10.1016/j.msea.2018.09.105 MSA36986

To appear in: Materials Science & Engineering A Received date: 23 June 2018 Revised date: 18 September 2018 Accepted date: 27 September 2018 Cite this article as: Dong Chen, Xinjian Zhang, Haoxi Wu, Dingmu Lang, Dawu Xiao, Zhenhong Wang, Bin Su and Daqiao Meng, The influence of impurities on the ductility and toughness of a low-temperature-aged U-Nb alloy, Materials Science & Engineering A, https://doi.org/10.1016/j.msea.2018.09.105 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

The influence of impurities on the ductility and toughness of a low-temperature-aged U-Nb alloy Dong Chen*, Xinjian Zhang, Haoxi Wu**, Dingmu Lang, Dawu, Xiao, Zhenhong Wang, Bin Su, Daqiao Meng*** China Academy of Engineering Physics, Mianyang 621900, PR China E-mail: [email protected] E-mail: [email protected] E-mail: [email protected]

*Corresponding author: Dong Chen; Tel: +86-10-3626826; Fax: +86-10-3625900; **Corresponding author: Haoxi Wu; Tel: +86-10-3626996; Fax: +86-10-3625900; ***Corresponding author: Daqiao Meng; Tel: +86-10-3626826; Fax: +86-10-3625900;

Abstract: The effect of impurity level and inclusions on the mechanical properties, especially the ductility and toughness, of two low-temperature-aged U-5.5Nb alloys were extensively investigated. Two U-5.5Nb alloys with different impurity levels, i.e., total C+N contents of 330 and 840 mass ppm, are discussed here. The methods used in the present study were scanning electron microscopy with energy-dispersive X-ray spectroscopy, optical microscopy, time-of-flight secondary ion mass spectrometry, tensile and impact mechanical property tests, and nanoindentation measurement. According to the results, both the low- and high-impurity-level samples contain two types of inclusions, namely, U(N,C) and Nb2C inclusions. The inclusion volume fraction, average inclusion radius, inclusion spacing, and number of inclusion particles per unit area all increase with increasing impurity level. In addition, the ductility, i.e., elongation, reduction in area, and impact toughness decrease with increasing impurity level. However, the impurity level has little effect on the yield strength. The inclusions have a detrimental effect on the ductility and toughness of the alloys because the inclusions are the preferential sites for void nucleation. Based on the experimental observations, an illustration of the failure mechanisms for the tensile and impact fracture behaviors for the low-temperature-aged U-5.5Nb alloy is presented.

Keywords: uranium; U-Nb alloy; impurity; inclusion; mechanical property

1 Introduction U-Nb alloys attract a significant amount of attention in the nuclear industry because of their excellent corrosion resistance, high strength, and high ductility [1-3]. Elemental additions of Nb are key for enhancing the corrosion resistance and improving the ductility of these alloys [4-5]. In the high-temperature γ-U phase, the Nb element is fully soluble, whereas Nb has minimal solubility in the low-temperature β-U and α-U phases [6]. However, a metastable martensitic phase with a homogeneous distribution of Nb can be obtained by quenching from the high-temperature γ phase as long as the cooling rate is greater than 20℃/s [7]. Different metastable phases (', " or ) can be obtained depending on the Nb content [8]. At a low niobium content, i.e., less than 3.5 wt.%, an ' metastable phase forms with an orthorhombic structure. As the niobium content is varied between 3.5 wt.% and 6.5 wt.%, a monoclinic " metastable phase is produced. Further increasing the niobium content to between 6.5 wt.% and 9 wt.% yields a tetragonal  phase. When the Nb content is greater than 9 wt.%, the high-temperature bcc  phase is retained to room temperature. Among the three metastable martensitic phases (', ", ), the " phase U-Nb alloys have been studied extensively [9-16], especially those with Nb contents near 6 wt.%, because the excellent ductility and superior corrosion resistance of this particular composition qualifies it for use in several applications. Furthermore, it is well known that " phase U-Nb alloys are age-hardenable by heat treatments below the 647℃ monotectoid isotherm [17-18]. The aging pathways in U-Nb alloys are complex and comprise at least five decomposition mechanisms, as discussed by Hackenberg et al. [19]. Two of the five decomposition mechanisms demand additional attention. One mechanism is cellular decomposition, which occurs at temperatures between 300℃ to 647℃ and produces a lamellar two-phase microstructure that includes a niobium-poor  phase and a niobium-rich  phase. The alloy after cellular decomposition generates a remarkable increase of yield strength,

from 150 to 1500 MPa, although the tensile ductility decreases from ~35% to nil and the corrosion resistance is almost entirely lost [19]. Another extensively studied decomposition mechanism of the " phase U-Nb alloys is the low-temperature-aging (<250℃) mechanism [9, 20-22], which significantly enhances the yield strength with almost no effects on corrosion resistance and only a slight loss in ductility. However, an experimental determination of the low-temperature-aging mechanism involved in the evolution of the fine-scale microstructure remains elusive. Despite this fact, the application of low-temperature-aged U-Nb alloys due to their excellent corrosion resistance has continued in practice. In addition to the typical fine-scale microstructure that consists of all microstructural features, both the impurity level and presence of inclusions significantly affect the mechanical properties of alloys, such as their ductility and toughness [23-25]. The inclusions that yield negative effects are typically nonmetallic compounds, for example, carbides, oxides, nitrides, or clusters of these compounds. Such inclusions can develop either by interacting with crucible materials or by reactions with impurity elements. In U-Nb alloys [26-27], C and N, derived from the graphite crucible and the poor vacuum environment, respectively, are the predominant impurity elements. The primary inclusions existing in U-Nb alloys include the U(N,C) inclusion with a rectangular shape, the Nb2C inclusion with an angular shape, and clusters derived from the aggregation of U(N,C) and Nb2C inclusions [27]. Carbon impurities account for a large fraction of the impurities in U-Nb alloys, and their content always varies from one hundred to several hundred ppm (in wt. parts per million). The lower Gibbs free energy of the Nb2C inclusion than that of the UC results in the reaction of Nb and C, leading to their preferential development in the U-Nb alloy [28]. Therefore, Nb2C inclusions will develop as long as the carbon content is sufficient. Chen et al. [26] experimentally proved that the area fraction of Nb2C inclusions increases with increasing additions of C, ranging from 0 ppm to 1000 ppm, in a U-5.5wt.%Nb alloy. The result indicated that the C content significantly affected the promotion of Nb2C inclusion formation in the alloy. More fine-scale determinations of the Nb2C inclusions in the U-6wt.%Nb alloy were carried out by

Kelly et al. [27]. According to these authors, the Nb2C inclusions were largely consistent with the AsNi structure of the -Nb2C phase (P63/mmc, a~0.31 nm, c~0.50 nm), as determined by transmission electron microscopy (TEM). However, the structural confirmation of U(N,C) inclusions in U-Nb alloys is not easy because the UN (NaCl structure, α~0.488 nm) and UC (NaCl structure, α~0.496 nm) have similar structures. A TEM micrograph of an FIB (focused ion beam) foil of the U(N,C) inclusion in a U-6wt.%Nb alloy showed a layered structure with the NaCl lattice structure (Fm 3 m, α~0.49 nm) [27]. It is worth mentioning that the mechanical properties of an alloy are affected not only by the inclusion type but also by the characteristics of the inclusion, including the inclusion volume fraction, inclusion spacing, and the resistance of the inclusion to void nucleation, which all significantly affect an alloy’s ductility and toughness. Garrison and Wojcieszynski [25] indicated that inclusion volume fraction and inclusion spacing were two independent variables; therefore, when discussing the effect of inclusion volume fraction alone on the toughness, inclusion spacing must be fixed at a constant. That is, to discuss the effect of inclusion spacing alone, one must compare the toughness as a function of inclusion spacing for a fixed inclusion volume fraction. The authors found that the toughness (  IC : crack tip opening displacement) scaled linearly with f 1/ 3 (f represents the inclusion volume fraction) in nine nickel steels at an inclusion spacing of 1.87 m, yielding the following trend:  IC increased with decreasing inclusion volume fraction [25]. Moreover, when the inclusion volume fraction was fixed at f=0.00027,  IC increased with increasing inclusion spacing. Furthermore, the resistance of inclusions to void nucleation is another characteristic that affects the mechanical properties of an alloy. Void nucleation in ductile fracture occurs either by particle cracking or by decohesion of the particle-matrix interface [29]. The particles can be classified into two types [30], primary particles and secondary particles, according to their resistance to void nucleation. The primary particles, namely, nonmetallic inclusions, which include sulfides, nitrides, and oxides

in steel, allow voids to generate more easily. The secondary particles, such as carbides and nitrides, which mostly precipitate during heat treatment, are more resistant to void nucleation than the primary particles and take on roles late during the fracture process [30]. In U-Nb alloys, U(N,C) inclusions can be treated as primary inclusions, which produce voids by cracking at low strains during deformation, whereas the Nb2C inclusions are likely secondary inclusions because they nucleate voids late in the fracture process by decohesion of the Nb2C-matrix interface [13]. However, investigations of the effects of inclusions on the mechanical properties of U-Nb alloys remain insufficient. Furthermore, it is somewhat surprising that very few studies have investigated the effects of C and N on the ductility and toughness of U-Nb alloys even though these two impurity elements are the two dominant impurities in the U and U-Nb alloys. As structural materials, the effects of impurities on the mechanical properties of U-Nb alloys simply cannot be ignored. Therefore, the aim of the present study is to investigate the effect of impurity level on the evolution of inclusion morphology in a low-temperature-aged U-5.5Nb alloy and its effect on the tensile and impact fracture behaviors of the alloy, in addition to determining the influence of impurity content on the mechanical properties.

2. Materials and methods Table 1 Chemical composition of the investigated U-5.5Nb alloy Impurity

Sample

C

N

B

S

Mg

Ti

Cr

Ca

Zn

Nb

L

210

120

≤20

≤20

≤20

≤20

≤20

≤20

≤20

5.47

330

H

650

190

≤20

≤20

≤20

≤20

≤20

≤20

≤20

5.43

840

level (IL)*

*IL=C+N Compositions are in wt. parts per million (ppm), except for Nb values, which are in wt.%.

Two experimental U-5.5Nb alloys (L and H) with different impurity levels were investigated. The alloys were prepared by vacuum-induced melting in a graphite crucible with a protective coating. The ingots were heat treated at 1100°C for 10 hours in a vacuum environment to homogeneously distribute the Nb, followed by furnace cooling. The alloys were then held at 850°C for two hours under vacuum, followed by water quenching. The as-quenched samples were aged at 200°C for two hours to

enhance the yield strength. The chemical composition of each alloy investigated is presented in Table 1. Different analysis methods were carried out to determine the impurity element content. Carbon was analyzed by combustion. To do so, the sample was heated by using a high-frequency induction furnace in an oxygen stream to convert the carbon in the sample to CO and CO2, which was then detected by infrared absorption spectroscopy. Nitrogen and sulfur were analyzed by inert gas fusion. All other reported elements were measured by inductively coupled plasma mass spectrometry (ICP-MS). Because carbon and nitrogen were the primary impurity elements, the impurity level (IL) was defined as the total content of C and N. Tensile samples were machined from the low-temperature-aged alloys to have a gauge length of 25 mm and diameter of 5 mm. Examples of original and after-fracture samples are shown in Fig. 1. The uniaxial tensile tests were performed at a cross-head speed of 1.5 mm/min on a servo-hydraulic MTS tester with a 100-kN load cell capacity. At least three samples were measured. Standard Charpy V-notch samples sized 10×10×55 mm3 were fabricated from the same low-temperature-aged alloys used for the tensile test. The V notch had a tip radius of 0.25 mm and was 2 mm deep with sides making an angle of 45°. The Charpy V-notch (CVN) impact toughness tests were carried out at room temperature on an Instron impact machine (Model CEAST9050) fitted with an Instron Dynatup Impulse data acquisition system, following the ASTM E-23 standard [31]. To investigate the effect of impurity level on alloy hardness, nanomechanical testing (Hysitron TI-950 US) was performed with Berkovich tip indenters to determine the nanohardness of the alloys. The system included an XYZ sample stage and a setup that comprised a piezo scanner with a transducer. Indentation on the two alloys (samples L and H) was carried out in a displacement-controlled mode with a loading rate of 1.6 mN/s that included holding for 2 seconds at a maximum load of 8 mN prior to unloading at the same rate. The load and displacement values were measured and recorded simultaneously during indentation. Standard calibration was performed at various depths on quartz before the tests were carried out on the experimental materials. The loading and unloading curves were used to calculate the hardness (H) and contact modulus (Er) by

employing the procedure suggested by Oliver and Pharr [32]. The elastic modulus (E) was calculated from the contact modulus (Er) and Poisson’s ratio () by using the following equation: 1 1  2 1  i   Er E Ei

(1)

where Ei and i are the Young’s modulus and Poisson’s ratio of the indenter, with values of 1140 GPa and 0.07, respectively. The Poisson’s ratio used for the low-temperature-aged U-5.5Nb alloy was 0.35 [33]. Each matrix was measured five times to obtain its average elastic modulus and hardness. Samples for metallography were cut from the two low-temperature-aged U-5.5Nb alloys. Such samples were used to determine the inclusion volume fractions, inclusion sizes and spacing, and the overall microstructure, including prior austenite grain size (high-temperature γ phase) by a mean linear intercept method. In all cases, the samples used for metallography were roughly ground using SiC waterproof abrasive paper of an increasingly fine grade to a P2000 grit, followed by polishing using a 1.5-m diamond abrasive attached to a fine-polishing cloth. Final preparation consisted of a brief and careful polish using a 0.5-m diamond abrasive on low-nap cloth with the purpose of completely maintaining the nonmetallic inclusions. The polished samples were etched by using two different etching solutions to clearly delineate the detailed microstructural information recorded by an Olympus DSX-500 optical microscope. In method 1, the samples were electroetched with a 4% oxalic water solution at 2 V for 4 seconds to reveal the inclusions; instead, throughout method 2, the samples were electroetched with a 10% oxalic water solution at 2 V for 10 seconds to reveal the grain boundaries. The phase composition of the alloys was determined via X-ray diffraction (XRD) (Empyrean, PANalytical, Holland) with Cu Kα radiation (λ=0.1540598 nm) at 40 kV. Scans were collected over a 2θ range of 30-70 with a step of 0.01. Although energy-dispersive X-ray spectroscopy (EDS) is traditionally used to determine the composition and types of inclusions in metal alloys, elements with an atomic number less than 10 cannot be detected accurately. As an advanced method for

the surface analysis of solid samples, time-of-flight secondary ion mass spectrometry (TOF-SIMS) has the remarkable characteristics of high spatial and mass resolution, and a high sensitivity in elemental and isotopic measurements over a wide mass range. In the present study, TOF-SIMS was used to analyze the types of inclusions and the distribution of impurity elements. TOF-SIMS includes a 25 keV liquid metal ion gun (LMIG) that can be used for both analysis and sputtering by providing a pulsed (analytical mode) or focused (sputtering mode) beam of primary ions, respectively. The beam energy can range from 10 kV to 25 kV, with a minimum spot size of 200 nm at 25 kV in pulsed (analytical) mode for

197

Au+. A background pressure of 110-9

mbar was maintained to prevent surface oxidation during the experimental process. The samples were mechanically grounded and polished according to the procedure described above. Before the measurement, the surface was sputtered in situ to depths between 1 nm and 20 nm using an Au+ sputter ion beam scanned over 1500 m× 2000 m to remove the oxide layer and obtain a raw surface. Images of the negative nonmetal ions N- and C- and the positive ions Nb+, U+, NbC+, UN+ and UC+ were acquired using 25 kV Au+ ions with currents between 1.5 and 2 pA. A metallographic analysis of damage accumulation was performed on fractured samples. For the tensile test samples, the longitudinal cross-sections were machined using wire electrical discharge machining to preserve the damage. The fractured impact test samples were machined through a thickness perpendicular to the notch in the longitudinal direction. The samples used for damage accumulation examination were prepared as described above, followed by etching. The damage was recorded by optical microscopy (OM) for both polished and electroetched conditions. In addition, the micro- and macroscale morphologies of the fracture surfaces of the failed samples, including those retrieved following the tensile test and the impact test, were observed using scanning electron microscopy (SEM) (model KYKY-EM3200).

Fig. 1. Tensile samples before and after test.

3 Results 3.1 Mechanical properties 3.1.1 Tensile test A typical engineering stress-strain curve of the two U-5.5Nb alloys is shown in Fig. 2. It can be seen that the alloys exhibit a typical “double yield” effect. This “double yield” phenomenon in as-quenched and aged U-Nb alloys has been widely investigated [5, 10, 20, 34-36]. Field et al. [10] suggested that the deformation process of the " phase U-Nb alloys can be divided into four zones. After the elastic stage, twinning deformation involving the motion of {172} twins dominates the deformation within the initial strain of 3%. At strains between 3% and 6%, the proposed dominant mechanism involves the cooperative migration of fine {130} twins and crossing {172} twin boundaries to eliminate fine twins. With further increases in strain, work hardening via a dislocation-slip mechanism controls the final deformation until fracture. The mechanical properties of the two U-5.5Nb alloys are shown in Table 2. The first yield strength (1y) and second yield strength (2y) of the two alloys are very similar. The average values of 1y in samples L and H are 3325 MPa and 3303 MPa, respectively. The average values of 2y in samples L and H are 7324 MPa and 7252 MPa, respectively. The ultimate tensile strength (UTS) of sample L is higher than that of sample H, with an average value of 8354 MPa (L). The tensile ductility is defined

by the total plastic elongation (εt) and reduction in area (Z). The average values of εt in samples L and H are 25.01.2% and 16.01.0%, respectively, and the average values of Z in samples L and H are 32.81.5% and 15.41.3%, respectively. The results indicate that the impurity level has a significant effect on ductility and a minimal effect on yield strength. High impurity levels result in a sharp decrease in ductility. The Charpy impact energies of the two samples are also different. The Charpy impact values are 18.6 J and 13.1 J for sample L and sample H, respectively, which is consistent with the observed trend of decreasing ductility with increasing impurity level.

Fig. 2. Engineering stress-strain curves of the U-5.5Nb alloys with different impurity levels.

3.1.2 Nanoindentation Nanoindentation was performed on both alloys with the purpose of identifying the effect of impurity level on elastic modulus and hardness. Fig. 3 shows the representative load-displacement curves obtained for the two samples. Oliver and Pharr’s method [32], described above, was used for nanoindentation data analysis. The measured modulus and hardness values are listed in Table 3. The values show a significant difference among the two samples. Sample L shows an average hardness of 2.000.14 GPa with a contact modulus of 92.862.70 GPa. For sample H, the

obtained hardness and modulus values are 2.520.25 GPa and 108.909.46 GPa, respectively. The elastic modulus was calculated from the measured contact modulus based on equation (1). The calculated elastic moduli for samples L and H yield average values of 88.172.78 GPa and 104.9410.06 GPa, respectively. Wheeler and Morris [33] reported that the tensile strength of as-cast uranium increased from 358 to 510 MPa as the carbon content was increased from 60 to 1250 ppm. The authors mentioned that when subjected to rapid cooling, carbon was retained in the solid solution alloy, which strengthened the matrix. Therefore, in the present study, it seems likely that the increased hardness observed in sample H is due to an increase of the carbon content. Although the effect of carbon content on elastic modulus is less clear, the potential reason is discussed in the following section. In addition to modulus and hardness, elasticity and plasticity can also be determined from the load-displacement curves obtained following nanoindentation [37]. The indexes defined as fractions of elastic recovery work (re) and energy dissipation (rd) can be used to characterize the relative elastic/plastic behavior of the material under external stress and strain, which can be expressed as: re 

rd 

(hm  h f ) hm hf hm

(2)

(3)

where hm is the penetration depth of the maximum load and hf is the residual depth measured when the load returns to a value of zero during unloading. The results are listed in Table 3. Sample L has a lower re of 0.22 and a higher rd of 0.78, which indicates that sample L exhibits relatively poor elasticity and good plasticity, whereas sample H presents a higher re of 0.24 and lower rd of 0.76, revealing relatively high elasticity during deformation.

Fig. 3. Load-depth curves of the matrices in the U-5.5Nb alloys Table 2 Uniaxial tensile test data of the U-5.5Nb alloy Sample

E/GPa

1y

2y

(MPa)

(MPa)

UTS (MPa)

εt (%)

Z (%)

CVN(J)

L H

592 3325 7324 8354 25.01.2 32.81.5 18.62 723 3303 7252 8023 16.01.0 15.41.3 13.11 Table 3 Mechanical properties of U-5.5Nb alloys determined by nanoindentation. Sample Er (GPa) E (GPa) H (GPa) re rd 0.22 92.862.70 88.172.78 2.000.14 0.78 L 0.24 108.909.46 104.9410.06 2.520.25 0.76 H

3.2 Microstructures and phase structure The microstructures of the two U-5.5Nb alloys are shown in Fig. 4. Figs. 4(a) and (b) reveal the distribution and morphologies of the inclusions. The total inclusion quantity in sample H is significantly higher than that in sample L. These inclusions, identified by our previous studies [13, 26], are mainly carbide inclusions, namely, Nb2C, and U(N,C). Based on the differences in contrast in the metallographic image, the white inclusions are Nb2C, and the rectangular and light gray inclusions are U(N,C). The U(N,C) inclusions remain rectangular in both alloys regardless of the impurity content. However, the Nb2C inclusions exhibit a globular shape in sample L, which has a lower carbon content; in contrast, the inclusions grow more quickly along their long axis to evolve into a strip-like shape when the carbon content is high and the availability of C is sufficient. The evolution of the shape of Nb2C inclusions

strongly depends on the carbon content of the alloy, as confirmed by our previous studies [26]. Furthermore, the clusters mainly form through the aggregation of Nb2C and U(N,C) inclusions. Figs. 4(c) and (d) reveal the prior  phase grain boundary. The average grain sizes of sample L and H are 25 µm and 33 µm, respectively. Nb2C inclusions form in the liquid and can act as the nucleus for heterogeneous nucleation during solidification. Therefore, the grain refinement in as-cast U-5.5Nb alloy is mainly attributed to the precipitation of Nb2C inclusions, as validated by our previous investigation [26]. Sunwoo et al. [38] also concluded that both silicon and carbon could promote the grain refinement in the α-U due to the formation of silicide and carbide. The U(N,C) inclusions precipitating during solidification can hinder grain growth in the subsequent heat treatment process. However, grain refinement by either Nb2C or U(N,C) inclusions must arise from the uniform distribution of the inclusions. Although the volume fraction of inclusions in sample H is higher than that in the sample L, the heavy aggregation of inclusions in sample H results in the inhomogeneous distribution of grain size. As a result, the grains close to the inclusions are very small, while those far from the inclusions are large. Because of the inhomogeneous grain size distribution, the average grain size in sample H is slightly larger than that in sample L, despite the high impurity level in sample H. XRD patterns of the two samples show the same " phase (Fig. 5), which is in accordance with the results reported for the U-5.5Nb alloy by other investigations [9, 15].

Fig. 4. Optical images of the U-5.5Nb alloys with different impurity levels: (a) and (c) sample L; (b) and (d) sample H, in which (a) and (b) show the morphologies of the inclusions, and (c) and (d) reveal the grain boundaries.

Fig. 5. XRD patterns of the two U-5.5Nb alloys.

3.3 Inclusion identification The inclusion characteristics can be defined by the inclusion volume fraction (f),

average inclusion radius (R0), inclusion spacing (X0) and number of inclusion particles per unit area (NA) [25]. The inclusion volume fraction is taken to be equal to the inclusion area fraction. The average inclusion radius is obtained following the relationship:

R0  ( / 4) H (d )

(4)

where H(d) is the harmonic mean of the apparent diameter of the inclusion. The inclusion spacing is taken to be the following [25]: X 0  (0.89R0 ) f 1 / 3

(5)

where f is the volume fraction, and R0 is the average radius of the inclusion. The inclusion characteristics of the two U-5.5Nb alloy are listed in Table 4. The total inclusion volume fraction of sample L is 0.00985, 0.00697 and 0.00288 of which are made up of the volume fractions of the Nb2C and U(N,C) inclusions, respectively, corresponding to total inclusion proportions of 71% and 29%. As the impurity content in sample H increases, the total inclusion volume fraction increases to 0.18475, 0.15665 and 0.0281 of which are made up of the Nb2C and U(N,C) inclusions, respectively, corresponding to total inclusion proportions of 85% and 15%. The increase in carbon content results in an increase in the Nb2C volume fraction, as confirmed by our previous study [26]. Although the volume fractions of these two inclusions are different between the two samples, there is no evidence showing that this difference can affect the ductility and toughness because both types of inclusions nucleate microvoids. Thus, the ductility and toughness mainly depend on the total inclusion volume. The average inclusion radius is 2.99 μm in sample L, while the value in sample H increases to 10.46 μm because of the aggregation of inclusions. Furthermore, the severe aggregation of inclusions in sample H leads to the inclusion spacing becoming slightly higher than that in sample L, even though the number of inclusion particles per unit area in sample H is far greater than that in sample L. Garrison and Wojcieszynski concluded that the inclusion volume fraction and inclusion spacing affected the toughness [25]. However, the formation of clusters in sample H leads to the value of inclusion spacing deviating from homogeneously

distributed inclusions. Therefore, the discussion of the difference in inclusion volume fraction between the two alloys will be more meaningful in the context of the effect of inclusions on the toughness. Speich and Spitzing experimentally proved that Charpy impact toughness increased as the inclusion volume fraction decreased [39]. Table 4 Inclusion characteristics of the two U-5.5Nb alloys Sample

f

R0 (μm)

NA (mm-2)

X0 (μm)

L

0.00985

2.99

785

12.39

H

0.18475

10.46

1396

16.34

To reveal the distribution of all concerned impurity elements and identify the types of the inclusions in the two U-5.5Nb alloys with different carbon contents, the samples were characterized by TOF-SIMS (Figs. 6 and 7). The analyzed region covered an area of 200×200 μm2. It is obvious that the Nb atoms are homogeneously distributed in the matrix, except at the site of Nb2C inclusions. Nb2C inclusions are the primary inclusion type formed by the interaction of Nb and C element in the U-5.5Nb alloy. Furthermore, UN inclusions are detected in both sample types. UN and UC have similar structures and can dissolve in each other. The TEM micrograph of an FIB (focused ion beam) foil of the U(N,C) inclusion in a U-6Nb alloy showed a layered structure that included both UN and UC [27]. Thus, UN and UC inclusions in a U-Nb alloy are always assigned as U(N,C). In the present study, although UC clusters are not observed in Figs. 6 and 7, the C element is homogeneously distributed in the matrix, including at the site of UN. Therefore, we infer that C dissolves in the UN and follow the habit of designating the mixed inclusion as U(N,C). The images of the negative nonmetal ions C- shown in Figs. 6 and 7 indicate that part of the carbon remains in the matrix. According to the U-C phase diagram, carbon has a certain solubility in liquid uranium [40]. Therefore, in the U-5.5Nb alloys studied in this work, no UC inclusions form in the liquid state. In contrast, Nb2C inclusions form in the liquid alloy and remain in the solid specimens during the subsequent heat treatment process. Thus, the reaction kinetics of Nb and C control the formation rate of Nb2C inclusions. Clearly, when the Nb content in the two U-5.5Nb alloys is sufficient, the carbon cannot be consumed completely. Thus, some of the

carbon precipitates to form UC inclusions during solidification, and another portion of the carbon remains in the matrix. It is hard to calculate the remaining carbon content because there is little knowledge of the reaction kinetics of Nb and C in the U-Nb alloy. In contrast, the higher the total carbon content in the matrix is, the greater the amount of carbon retained in the matrix, as shown in Figs. 6 and 7.

Fig. 6. Secondary ion images of sample L.

Fig. 7. Secondary ion images of sample H.

3.4 Fractograph 3.4.1 Tensile fracture surface Fig. 8 shows the morphologies of the tensile fracture surfaces of the two U-5.5Nb alloys. Fig. 8(a) shows a macro view of a fracture surface in sample L, which exhibits a typical “cup and cone” feature. However, the fracture surface in sample H has nearly lost the “cup and cone” feature and instead has flattened significantly (Fig. 8(b)). The high-magnification views in Fig. 8(c)-(h) indicate that the fracture exhibits ductile character with typical dimple features present in the fracture surface. However, the void formation mechanisms and fracture behaviors in the two alloys are different. The dimples in sample L are deeper with visible inclusions and are homogeneous and small in size. In contrast, the dimples are shallow in sample H, and some faceting can be observed in Fig. 8(d). Moreover, the dimples in sample H are significantly larger than those in sample L, due to the large

number of big clusters. The rapid coalescence of the voids in sample H leads to the rapid formation of cracks (Fig. 8(f)). As a result, the sample with the high impurity level exhibits a low ductility. It is worth mentioning that different inclusions have different void formation mechanisms, i.e., U(N,C) inclusions nucleate voids by cracking (Fig. 8(h)), whereas the Nb2C inclusions nucleate voids by decohesion of the Nb2C-matrix interface (Fig. 8(g)), which is validated by our previous investigations [13, 41]. Although both types of inclusions are involved in void nucleation, they exhibit different behaviors as a sample is strained during tensile deformation [13]. The voids first nucleate due to cracking of the U(N,C) inclusions at the end of the twinning deformation stage (a strain of 6.6%); then, as the strain is increased further to the middle deformation stage (a strain of 16%), the Nb2C inclusions generate voids by the decohesion of the Nb2C-interface [13]. The different void formation mechanisms at the various inclusions sites in the U-5.5Nb alloy are due to the differences in the mechanical properties of the inclusions [14]. Nanoindentation tests indicate that the U(N,C) inclusion has a low hardness and a high plasticity; as a result, the U(N,C) inclusion deforms and fractures at the early deformation stage with a low strain [13-14]. In contrast, the Nb2C inclusion shows a significantly higher hardness than that of the U(N,C) inclusion and matrix; the mismatch between the hardness and elasticity of the Nb2C inclusion and matrix causes the decohesion at the Nb2C-matrix interface at higher strains [14]. Therefore, the presence of U(N,C) inclusions has a significantly more detrimental effect on the ductility and toughness of these alloys due to the nucleation of voids at low strain values during deformation.

Fig. 8. Fracture morphologies of the U-5.5Nb alloys after tensile testing: (a), (c), (e) and (g) are the fracture surfaces of sample L at different magnifications; (b), (d), (f) and (h) are the fracture surfaces of sample H at different magnifications.

Cross-sectional micrographs of the microstructures of the failed tensile test samples are shown in Figs. 9 and 10, mainly to illustrate the patterns of the voids. Fig.

9(a) and Fig. 10 (a) show the distinct nature of the fracture behaviors in the two samples. In sample L, a typical “cup and cone” fracture can be observed, which indicates a change in the failure mechanism from ductile to shear failure that corresponds to increasing position relative to the center of the tensile sample (i.e., moving toward the sample edge). In sample H, the fracture surface grows flatter and tends to transform from a ductile fracture to a brittle fracture, although the fracture surface also exhibits dimples, as seen in Fig. 8(d). High-magnification views of the microstructures near the fracture surface are shown in Fig. 9(b-c) and Fig. 10(b-c). Voids nucleate at the sites of the U(N,C) inclusions and the Nb2C inclusions by cracking and debonding at the Nb2C-matrix interface, respectively. Large clusters dominate the void formation features shown in Fig. 10(b). The voids grow and coalesce with each other to form cracks after nucleation, as shown in Fig. 9(c).

Fig. 9. Optical micrographs of the longitudinal cross-sections of failed tensile sample L: (a) low magnification and (b) and (c) high magnification, in which the surface in (b) is etched and that in (c) is polished.

Fig. 10. Optical micrographs of the longitudinal cross-sections of failed tensile sample H: (a) low magnification and (b) and (c) high magnification, in which the surface in (b) is etched and that in (c) is polished.

3.4.2 Impact fracture surface Fractographs of the U-5.5Nb alloys tested by the Charpy impact test are shown in Fig. 11. No apparent cleavage facets can be observed in Fig. 11. The two samples also show a typical ductile fracture mode under impact loading. Two types of voids can be observed in the dimples. One is equiaxed voids with internal inclusions, which are similar to the dimples formed in the tensile test samples. The second type of void is larger and more elongated and forms due to shear during fracture. Inclusions also act as a source of void nucleation. U(N,C) inclusions not only crack to nucleate voids but also break into small fragments under the immense and instantaneously applied impact load (Fig. 11(e)). Most of the Nb2C inclusions do not break but hold their original shape due to their high strength and hardness (Fig. 11(e), (g) and (h)). Sectioned samples can be observed in the longitudinal plane near the fracture surface, as shown in Fig. 12 (c). Optical images of the failed samples are shown in Fig. 12(a) and (b). The fracture surface is flatter in sample H than that in sample L, which indicates that the cracks form and spread more quickly in sample H due to the large number of inclusions in the matrix.

Fig. 11. Fracture morphology of the U-5.5Nb alloys tested by Charpy impact test: (a), (c), (e) and (g) are the fracture surfaces of sample L at different magnifications; (b), (d), (f) and (h) are the fracture surfaces of sample H at different magnifications.

Fig. 12. Optical image of broken samples sectioned in the longitudinal-transverse plane: (a) sample L; (b) sample H; (c) schematic of the observed section used in OM.

4 Discussion 4.1 Effect of microstructure on the mechanical properties In the U-5.5Nb alloys studied in this work, the grain sizes (prior  grain size) of samples L and H are 25 µm and 33 µm, respectively. Moreover, the corresponding first yield strengths of samples L and H are 3325 MPa and 3303 MPa, and the second yield strengths are 7324 MPa and 7252 MPa, respectively. The results indicate that there is little difference in yield strength between the two alloys, even though the grain sizes of the two are different. The same phenomenon was reported by other investigators. Jackson [42] reported that the grain size of the U-5.3Nb and U-7.1Nb alloys had little effect on the hardness. Ren et al. [43] experimentally investigated the mechanical properties of two as-quenched U-5.5Nb alloys with a coarse grain of 70.8 m and a fine grain of 4.3 m. The coarse-grained alloy had a first yield point of 180 MPa, while the fine-grained alloy had a first yield point of 178 MPa. This result indicated that the substantial difference in grain size did not lead to an apparent variation in yield strength. However, as a general trend in structural

materials, the yield strength of a polycrystalline metal increases as the grain size decreases, which can be taken as proportional to

1 / d , where d is the average grain

diameter. This well-known theory of the Hall-Petch relationship is based on the presupposition of deformation caused by dislocation slip. The grain refinement increases the number of grain boundaries, which hinders the dislocation slip and increases the yield strength [43]. Moreover, the fine grain can easily accommodate the deformation to enhance the plasticity. In an " U-Nb alloy with a substructure, namely, transformed twins produced by quenching, twinning deformation dominates the initial strain of 6%, while the irreversible deformation process associated with dislocation slip occurs beyond the strain of 6% [10]. Therefore, the yield strength of the " U-Nb alloys depend on the mobility of the twin interface, not the impediment of dislocation. Furthermore, the coarse-grained sample has an elongation of 33.6%, while the grain size decreases by 94% to 4.3 m and the elongation increases by 30% to 43.7% [43]. However, in the present study, the grain size decreases from 33 m in sample H to 25 m in sample L, representing a 24% decrease, whereas the corresponding elongation values increase from 16% to 25%, representing a 56% increase. The large increase in elongation does not match the proportion of grain refinement. Thus, there must be another reason for the great variation in ductility. With regard to the fracture toughness, no systematic studies regarding the effect of grain size have been carried out in U-Nb alloys. However, Saxton [44] mentioned that the humidity, temperature and impurities had a significant effect on fracture toughness. Saxton emphasized that among these factors, impurities such as hydrogen and carbon sharply decreased the toughness and even caused the embrittlement of materials [44]. In the present study, no significant difference exists in the grain size between the two U-5.5Nb alloys, but the large difference in impurity levels probably causes the variation in impact toughness. More information about the effect of impurity level on toughness will be discussed in the following section. 4.2 Effect of impurity level on elastic modulus The elastic modulus is one of the most important mechanical properties of

structural materials. In general, the elastic property of single-phase structures of metallic alloys scales with the solute content in a nearly linear relationship [45]. Unlike steels [46], the elastic modulus of U-Nb alloys varies significantly with composition and heat treatment process [15, 42]. However, there has been no systematic study on the dependence of elastic modulus on the impurity level in U-Nb alloys. Fig. 13(a) shows that the elastic modulus in the present study increases with increasing impurity levels. However, not all the impurities in the present study that could affect the elastic modulus are directly related to solute content [45]. From the elemental distributions shown in Figs. 6 and 7, N is mainly present in UN, with little N dissolved in the matrix; in contrast, a large amount of C dissolves in the matrix. Therefore, it is more accurate to conclude that a relationship between C content and elastic modulus exists. The variation in elastic moduli, from both the literature and the present study, with carbon content is shown in Fig. 14. Accordingly, the elastic modulus has a linear relationship with carbon content, increasing with the carbon content in the U-5.5Nb alloy. Moreover, the plastic deformation of the " phase U-Nb alloy in the initial strain region of 6% is predominantly due to the motion of existing twin interfaces and not the motion of slip dislocations [42]. Therefore, the interstitial impurity pins the action of twin interfaces and changes the elasticity of the alloy [42].

Fig. 13. Elastic modulus versus impurity level (a), reduction of area versus impurity level (b), elongation versus impurity level (c) and impact energy versus impurity level (d).

Fig. 14. Elastic modulus versus carbon content.

4.3 Effect of impurity level on ductility In the present study, it was found that inclusions have a detrimental effect on alloy elongation and reduction in area (Fig. 13(b) and (c)). U(N,C) inclusions are the

most harmful inclusions in U-5.5Nb alloys because they crack in the early stage of deformation under low strain and consequently generate voids in the matrix, which coalesce to form cracks in the matrix. Moreover, Nb2C inclusions also have a detrimental effect on ductility, although such inclusions play a role in the middle strain range during deformation, mainly due to debonding at the Nb2C-matrix interface. Nevertheless, a few studies have discussed the effect of impurities on the ductility of U alloys. In U-0.1wt.%Cr alloy [47], the elongation decreased from 9.8% to 2% when the carbon content was increased from 470 ppm to 940 ppm. Fracture surface analysis showed that most of the carbides separated from the matrix and acted as a void generator. Furthermore, it is widely accepted that increasing ultimate tensile strength (UTS) tends to decrease ductility. Therefore, it is better to compare the product UTSεt when there is a difference in the UTS of the studied alloy with the purpose of compensating for this effect [24]. In the present study, the values of UTS

εt were 20,875 MPa% and 12,832 MPa% in samples L and H, respectively. This result indicates that the inclusions reduce the ductility. 4.4 Effect of impurity level on toughness 4.4.1 Impact fracture behavior The impact toughness of U-5.5Nb alloys is sensitive to the impurity level, and the necessary impact energy to cause fracture decreases with increasing impurity content (Fig. 13 (d) and Fig. 15(a)). Inclusions, as a source of cracks, play an important role in the fracture process under impact loading. Fig. 15(b) shows the detailed information of the energy needed for crack initiation and propagation in the two U-5.5Nb alloys. The load-displacement curve is reduced in height and grows narrower with increasing impurity content, and the values of displacement of the maximum impact force (Sm) and maximum impact force (Fm) decrease with increasing impurity content. The area encompassed by the load-displacement curve in Fig. 15(b) represents the total impact energy (Wt), which increases with decreasing impurity level. Furthermore, the crack initiation energy (Wi) and crack propagation energy (Wp) show the same trend as total impact energy when the impurity level decreases in the present

study. The low values of Wi and Fm in the high-impurity-level alloy (sample H) indicate that crack initiation becomes easier, due to the increased volume of inclusions, which act as the source of voids. The values of the proportion of crack initiation energy (Wi) to total impact energy (Wt) (Wi/Wt) in sample L and sample H are 0.43 and 0.36, respectively. The low value of Wi/Wt indicates that the toughness of sample H is significantly controlled by crack initiation. Moreover, the load-displacement curve of sample H decreases sharply after the maximum impact force shown in Fig. 15(a), which indicates that the alloy has a weak ability to hinder crack growth.

Fig. 15. Impact force and energy versus deflection (a) and a magnified view of the curve (b), in which Wi (Wi-L: crack initiation energy for sample L, Wi-H: crack initiation energy for sample H) is the crack initiation energy, Wp (Wp-L: crack propagation energy for sample L, Wp-H: crack propagation energy for sample H) is the crack propagation energy, Wt (Wt-L: total impact energy for sample L, Wt-H: total impact energy for sample H) is the total impact energy, Sm (Sm-L: displacement of maximum impact force for sample L, Sm-H: displacement of maximum impact force for sample H) is the displacement of maximum impact force and Fm (Fm-L: maximum impact force for sample L, Fm-H: maximum impact force for sample H) is the maximum impact force.

4.4.2 Dynamic mechanical properties Dynamic mechanical properties are also very important for structural materials, especially those must endure impact loads. Thus, dynamic tensile stress (yd) and dynamic fracture toughness (KId) are calculated from the Charpy test data using the following equations. The equation for dynamic tensile stress (dynamic UTS) can be described as follows [48]:

 yd 

2.99 PW B(W  a)2

(6)

where W is the width of the Charpy impact test sample, B is the height of the Charpy

impact test sample, a is the crack depth of the Charpy impact test sample, and P is the maximum load, which can obtained from the curves in Fig. 15. The equation for dynamic fracture toughness can be described as follows [49,50]: K Id 

3 S aY (a / W ) F (t ) 2 BW 2

(7)

where the parameters of W, B and a have the same meaning as those described above. S is the span length, F(t) is the applied load at any instant of time, and the maximum load in the load/time trace is chosen to represent the initiation of fracture in the present study. Y(/W) is a geometric correction factor and can be calculated from the following equation: Y (a / W )  1.090  1.735(

a a a a )  8.20( ) 2  14.18( ) 3  14.57( ) 4 W W W W

(8)

All the parameters used for calculation and the calculated results are listed in Table 5. The dynamic tensile stress and dynamic fracture toughness are 663 MPa and 66 MPam1/2 for sample L, respectively, and the corresponding values for sample H are 533 MPa and 53 MPam1/2. The results indicate that the impurity level has a significant effect on the dynamic mechanical properties of these alloys. With an increasing impurity level, the dynamic tensile strength and dynamic fracture toughness decrease. Therefore, to obtain a U-Nb alloy with excellent dynamic properties, the impurity content must be restricted to low values. Table 5 Dynamic mechanical properties determined from the Charpy impact test and the parameters used for calculation. Sample

W (mm)

B (mm)

L H

10 10

10 10

P (kN)

F(t) (kN)

yd

KId

(mm)

S (mm)

(MPa)

(MPam1/2)

2 2

40 40

14.2 11.4

14.2 11.4

663 533

66 53



4.5 The role of inclusions Microstructure and phase structure studies reveal no significant difference between the two U-5.5Nb alloys. Because the only detected significant difference between the two alloys is the impurity level, it can be concluded that the inclusions cause the variation observed in the mechanical properties. To interpret the failure mechanisms of the alloy after tensile and impact fracture testing, a schematic

illustration of the fracture process based on void nucleation, growth and coalescence mechanisms is shown in Fig. 16. As discussed above, the initial microstructure of the alloys contains two kinds of inclusions, i.e., Nb2C and U(N,C). A large number of clusters in this matrix formed by the aggregation of Nb2C and U(N,C) inclusions. During the tensile deformation process (Fig. 16(a)), the high stress concentration after twinning deformation first causes the cracking of U(N,C) inclusions due to their low hardness and strength, as discussed in our previous study [14]. Next, voids are generated in the matrix around the cracked U(N,C) inclusions. Furthermore, with increasing strain, these voids grow and coalesce before the cracks form in the matrix. Therefore, whether solitary or in clusters, U(N,C) inclusions yield the most deleterious effect on the mechanical properties of these alloys. Moreover, Nb2C inclusions also contribute to void formation by the decohesion of the interface between the Nb2C inclusions and matrix, which occurs when the strain reaches a moderate level [13]. At high impurity levels, more inclusions are involved in void nucleation, which leads to early fracture. Therefore, both inclusion types have a deleterious effect on alloy ductility. In contrast to the uniform deformation behavior followed by necking that was observed during tensile deformation, sample fracture under impact loading is primarily characterized by crack extension along the existing notch (Fig. 16(b)). Thus, the zone covering the effect of inclusions is concentrated within a small width near the fracture surface. Cracking of the U(N,C) inclusions and decohesion of the Nb2C-matrix interface promote the propagation of cracks in the matrix under impact loading. Following the same trend, increasing the impurity level decreases the impact toughness. Therefore, the most effective way to increase the ductility and toughness of the U-5.5Nb alloy is to remove the impurities. Some often-used purifying methods in industrial manufacturing are gravitational separation, electromagnetic separation, centrifugation, electrolysis and electroslag refining [51]. To our knowledge, few studies on the refinement of impurities in U-Nb alloys have been carried out, thus this work will be addressed in a future study. Finally, the inhomogeneous plastic deformation of the U-5.5Nb alloy is a worthy

issue for discussion, as the initiation of voids and cracks could be easily localized in some areas, such as inclusions and clusters. Inhomogeneous plastic deformation is a common phenomenon in alloys because of the strain localization during plastic deformation [52-55]. A well-known example of inhomogeneous plastic deformation is the formation of Lüders bands in mild steel [55]. The typical and most extensively investigated inhomogeneous plastic deformation in U-Nb alloy is the appearance of adiabatic shear bands (ASBs) when the materials endure high strain rate deformation [56,57]. Although no ASBs are found under low strain rate deformation, the clusters aggregating due to the Nb2C and U(N,C) inclusions in the matrix cause the inhomogeneity of the microstructure shown in Fig. 4. Thus, during the uniaxial tensile deformation in the U-5.5Nb alloy studied in this work, the inhomogeneous microstructure could induce strain localization in the matrix, and the sites where strain localization occurs will be the location of crack initiation. Therefore, further experimental investigation combined with numerical simulation based on the finite element method needs be carried out to illuminate the inhomogeneous plastic deformation in U-Nb alloys. All this work will be performed in our future studies.

Fig. 16 Schematic illustration of the ductility fracture during tensile and impact deformation in U-5.5Nb alloy: (a) tensile fracture reproduced from Ref. [58]; (b) impact fracture reproduced from Ref. [59].

5. Conclusions The effects of impurity level and inclusions in U-5.5Nb alloy on its mechanical properties, especially its ductility and impact toughness, were investigated. According to the results of this work, the following conclusions can be drawn. (1) The two U-5.5Nb alloys with different impurity levels show a typical single

" phase structure in the matrix. Two kinds of inclusions, i.e., U(N,C) and Nb2C, are observed in the studied alloys. As the impurity level is increased, the inclusion volume fraction, average inclusion radius, inclusion spacing and number of inclusion particles per unit area all increase. (2) The stress-strain curves of the U-5.5Nb alloys show a typical “double yield” effect. The first yield strength and second yield strength of the two alloys are very similar with corresponding values of 3325 and 7324 MPa in sample L and 3303 and 7252 MPa in sample H. The ultimate tensile strengths decrease slightly with increasing impurity level. However, the elongation and reduction in area of the two alloys show a significant difference. The average values of elongation in samples L and H are 25.01.2% and 16.01.0%, respectively, and the average values of reduction in area in samples L and H are 32.81.5% and 15.41.3%, respectively. In addition, the Charpy impact value decreases from 18.6 J in sample L to 13.1 J in sample H. (3) Inclusions have a significant effect on alloy ductility and toughness. Voids nucleate at sites of U(N,C) and Nb2C inclusions based on cracking and Nb2C-matrix decohesion phenomena, respectively. By increasing the impurity level, the quantity of inclusions increases, leading to a concurrent increased in the number of sites for void nucleation. The presence of more voids in the matrix leads to more rapid crack formation due to the coalescence of voids in the sample H. Finally, the spreading of cracks within the matrix results in alloy fracture.

Acknowledgments This research was financially supported by the Key Project of China Academy of Engineering Physics (CAEP) under grant no. TB120301. The authors gratefully acknowledge Lin Jin and Su Yang for performing sample preparation; Lifeng He, Ruiwen Li, Tian Yang and Fang Xu for performing mechanical property tests; and Qinguo Wang for performing XRD analysis.

References [1] H.F. Ji, P Shi, G.C. Hu, X.L. Chen, R.W. Li, J.R. Yang, X.L. Wang, The effects of microstr

ucture on the hydriding for 500℃/2h aged U-13at.%Nb alloy, J. Nucl. Mater. 488 (2017) 252-260.

[2] H.F. Ji, P Shi, R.W. Li, C.L. Jiang, J.R. Yang, G.C. Hu, The microstructure and hydriding characteristics of high temperature aged U-13 at.% Nb alloy, J. Nucl. Mater. 464 (2015) 43-47.

[3] A.J. Sunwoo, D.S. Hiromoto, Effects of natural aging on the tensile properties of water-quenched U-6%Nb alloy, J. Nucl. Mater. 327 (2004) 37-45.

[4] Y.P. Wu, Q.W. Wu, S.F. Zhu, Z. Pu, Y.Z. Zhang, Q.G. Wang, D.M. Lang, Y.P. Zhang, Effect of niobium element on the electrochemical corrosion behavior of depleted uranium, J. Nucl. Mater. 478 (2016) 7-12.

[5] H.M. Volz, R.E. Hackenberg, A.M. Kelly, W.L. Hults, A.C. Lawson, R.D. Field, D.F. Teter, D.J. Thoma, X-ray diffraction analyses of aged U-Nb alloys, J. Alloys Compd. 444-445 (2007) 217-225.

[6] X.J. Liu, Z.S. Li, J. Wang, C.P. Wang, Thermodynamic modeling of the U-Mn and U-Nb systems, J. Nucl. Mater.380 (2008) 99-104.

[7] K.H. Eckelmeyer, A.D. Romig, L.J. Weirick, The effect of quench rate on the microstructure, mechanical properties, and corrosion behavior of U-6 wt pct Nb, Metall. Trans. A 15A (1984) 1319-1330.

[8] M. Anagnostidis, M. Colombié, H. Monti, Phases metastables dans les alliages uranium-niobium, J. Nucl. Mater. 11 (1964) 67-76.

[9] A.J. Clarke, R.D. Field, R.E. Hackenberg, D.J. Thoma, D.W. Brown, D.F. Teter, M.K. Miller, K.F. Russell, D.V. Edmonds, G. Beverini, Low temperature age hardening in U-13 at.% Nb: An assessment of chemical redistribution mechanisms, J. Nucl. Mater. 393 (2009) 282-291.

[10] R.D. Field, D.J. Thoma, P.S. Dunn, D.W. Brown, C.M. Cady, Martensitic structures and deformation twinning in the U-Nb shape-memory alloys, Philos. Mag. A 81(7) (2001) 1691-1724.

[11] D.W. Brown, M.A.M. Bourke, R.D. Field, W.L. Hults, Neutron diffraction study of the deformation mechanisms of the uranium-7 wt.% niobium shape memory alloy, Mater. Sci. Eng. A. 421 (2006) 15-21.

[12] R.D. Field, D.W. Brown, D.J. Thoma, Texture development and deformation mechanisms during uniaxial straining of U-Nb shape-memory alloys, Philos. Mag. 85(13) 2005 1441-1457.

[13] D. Chen, X.J. Zhang, X.D. Meng, R. Ma, R.W. Li, Z.H. Wang, B. Su, D.M. Lang, T. Yang, D.Q. Meng, Mechanisms of void formation during uniaxial tensile testing in a low-temperature-aged U-Nb alloy, Mater. Sci. Eng. A. 723 (2018) 182-193.

[14] D. Chen, R.W. Li, D.M. Lang, Z.H. Wang, B. Su, X.J. Zhang, D.Q. Meng, Determination of the mechanical properties of inclusions and matrices in α-U and aged U-5.5Nb alloy by nanoindentation measurements, Mater. Res. Express 4 (2017) 116516.

[15] R.E. Hackenberg, U-Nb aging final report, Los Alamos National Laboratory, 2007, Report, LA-14327.

[16] R. Ma, Z.Y. Ren, Q.F. Tang, D. Chen, T.Y. Liu, B. Su, Z.H. Wang, C. Luo, Metallurgical characterization of melt-spun ribbons of U-5.4wt%Nb alloy, J. Nucl. Mater. 504 (2018) 281-288.

[17] R.E. Hackenberg, Property and lifetime prediction in aged U-Nb alloys-A statistical

assessment, Los Alamos National Laboratory, 2009, Report, LA-14389.

[18] R.J. Jackson, Isothermal transformations of uranium-13 atomic percent niobium, Dow Chemical Company, Rocky Flats Division, 1971, Report, RFP-1609.

[19] R.E. Hackenberg, The surprising occurrence of non-steady-state growth of divergent lamellar decomposition products in uranium-niobium alloys: A preliminary, Los Alamos National Laboratory, 2012, Report, LA-UR-12-25218.

[20] D.W. Brown, M.A.M. Bourke, A.J. Clarke, R.D. Field, R.E. Hackenberg, W.L. Hults, D.J. Thoma, The effect of low-temperature aging on the microstructure and deformation of uranium- 6 wt% niobium: An in-situ neutron diffraction study, J. Nucl. Mater. 481 (2016) 164-175.

[21] H.M. Volz, R.E. Hackenberg, A.M. Kelly, W.L. Hults, A.C. Lawson, R.D. Field, D.F. Teter, D.J. Thoma, X-ray diffraction analyses of aged U-Nb alloys, J. Alloys Compd. 444-445 (2007) 217-225.

[22] L. Hsiung, J. Zhou, Sinodal decomposition and order disorder transformation in a water quenched U-6wt%Nb alloy, Lawrence Livermore National Laboratory, 2006, Report, UCRL-TR-224432.

[23] A.S. Wu, D.W. Brown, B. Clausen, J.W. Elmer, The in fluence of impurities on the crystal structure and mechanical properties of additive manufactured U-14at.%Nb, Scripta Mater. 130 (2017) 59-63.

[24] H. Tervo, A. Kaijalainen, T. Pikkarainen, S. Mehtonen, D.Porter, Effect of impurity level and inclusions on the ductility and toughness of an ultra-high-strength steel, Mater. Sci. Eng. A. 697 (2017) 184-193.

[25] W.M. Garrison, A.L. Wojcieszynski, A discussion of the effect of inclusion volume fraction on the toughness of steel, Mater. Sci. Eng. A. 464 (2007) 321-329.

[26] D. Chen, R. Ma, X.L. Chen, Z.H. Wang, B. Su, X.J. zhang, D.Q. Meng, The influence of the carbon addition on the microstructure of a U-Nb alloy and the features of Nb2C inclusions, Metall. Res. Technol. 115 (2018) 309.

[27] A.M. Kelly, R.D. Field, D.J. Thoma, Metallographic preparation techniques for U-6wt.%Nb, J. Nucl. Mater. 429 (2012) 118-127.

[28] J.G. Avery, The effect of innoculation of uranium and uranium-6 weight percent niobium with 0.01 to 1.0 weight percent titanium on impurity levels, Dow Chemical Company, Rocky Flats Division, 1974, Report, RFP-2067.

[29] A. Pineau, A.A. Benzerga, T. Pardoen, Failure of metals I: Brittle and ductile fracture, Acta Mater. 107 (2016) 424-483.

[30] W.M. Garrison, A.L. Wojcieszynski, A discussion of the spacing of inclusions in the volume and of the spacing of inclusion nucleated voids on fracture surfaces of steels, Mater. Sci. Eng. A. 505 (2009) 52-61.

[31] ASTM E23, E23-09: Standard Test Method for Notched Bar Impact Testing of Metallic Materials, Annual Book ASTM Stand. ASTM, West Conshohocken, PA, 2009.

[32] W.C. Oliver, G.M. Pharr, An Improved Technique for Determining Hardness and Elastic Modulus Using Load and Displacement Sensing Indentation Experiments, J. Mater. Res. 7 (1992) 1564-1583.

[33] D.W. Wheeler, S.T. Morris, Micro-mechanical characterization of uranium, J. Nucl. Mater. 385 (2009) 122-125.

[34] A. Zubelewicz, F.L. Addessio, C.M. Cady, A constitutive model for a uranium-niobium alloy, J Appl. Phys. 100 (2006) 013523.

[35] C.N. Tupper, D.W. Brown, R.D. Field, T.A. Sisneros, B. Clausen, Large strain deformation in uranium 6 Wt Pct niobium, Metall. Mater. Trans. A. 43(2) (2012) 520-530.

[36] R.E. Hackenberg, Synthesis and characterization of nonbanded U-Nb Plate, Los Alamos National Laboratory, 20007, Report, LA-14316.

[37] Y.W. Bao, W. Wang, Y.C. Zhou, Investigation of the relationship between elastic modulus and hardness based on depth-sensing indentation measurements, Acta Mater. 52 (2004) 5397-5404.

[38] A. Sunwoo, D. Goto, Effects of processing on microstructure and properties of -uranium formed parts, Scripta Mater. 47 (2002) 261-266.

[39] G.R. Speich, W.A. Spitzig, Effect of volume fraction and shape of sulfide inclusions on through-thickness ductility and impact energy of high-strength 4340 plate steels, Metall. Trans. A 13A (1982) 2239-2258.

[40] P.Y. Chevalier, E. Fischer, Thermodynamic modelling of the C-U and B-U binary systems, J. Nucl. Mater. 288 (2001) 100-129.

[41] D. Chen, X.L. Chen, R. Ma, Q.F. Tang, B. Su, Z.H. Wang, X.J. zhang, D.Q. Meng, Effect of strain rate on fracture of U-5.5Nb alloy, Mater. Sci. Tech. 34(1) (2018) 12-19.

[42] R.J. Jackson: Elastic, plastic, and strength properties of U-Nb and U-Nb-Zr alloys, in Physical Metallurgy of uranium alloys, (ed. J. J. Burke et al), 611-656; 1974, Massachusetts, Brook Hill.

[43] Z.Y. Ren, R. Ma, G.C. Hu, J. Wu, Z.H. Wang, C. Luo, Grain refinement and mechanical response of U-5.5 wt%Nb alloy produced by cold rolling and heat treatment, J. Nucl. Mater. 494 (2017) 72-78.

[44] Saxton H J. Fracture of uranium alloys, in Physical Metallurgy of uranium alloys, (ed. J. J. Burke et al), 349-391; 1974, Massachusetts, Brook Hill.

[45] W.A. Counts, M. Friák, D. Raabe, J. Neugebauer, Using ab initio calculations in designing bcc Mg-Li alloys for ultra-lightweight applications, Acta Mater. 57 (2009) 69-76.

[46] S Reeh, D Music, T Gebhardt, M Kasprzak, T Jäpel, S. Zaefferer, D. Raabe, S. Richter, A. Schwedt, J. Mayer, B. Wietbrock, G. Hirt, J.M. Schneider, Elastic properties of face-centred cubic Fe-Mn-C studied by nanoindentation and ab initio calculations, Acta Mater. 60 (2012) 6025-6032.

[47] S. Zalkind, O. Sabag, I. Makover, S. Haroush, Influence of carbon on the tensile properties of U-0.1w%Cr, J. Mater. Sci. Lett. 21 (2002) 551-553.

[48] W.L. Server, General yielding of Charpy V-notch and precracked Charpy specimens, J Mater. Sci. Technol. 100(2) (1978) 183-188.

[49] K. Kishimoto, S. Aoki, M. Sakata, Simple formula for dynamic stress intensity factor of pre-cracked Charpy specimen, Eng. Fract. Mech. 13 (1980) 501-508.

[50] E. Martin, T. Macke, J.M. Quenisset, Fracture parameters determination at yield point and fracture initiation in instrumented impact testing, Eng. Fract. Mech. 24 (1986) 821-836.

[51] L.F. Zhang, J. Gao, L.N.W. Damoah, D.G. Robertson, Removal of iron from aluminum: A review, Min. Proc. Ext. Met. Rev. 33(2) 2012 99-157.

[52] A.A. Csontos, E.A. Starke, The effect of inhomogeneous plastic deformation on the ductility and fracture behavior of age hardenable aluminum alloys, Int. J. Plasticity 21 (2005) 1097-1118.

[53] J.D. Kang, D.S. Wilkinson, R.K. Mishra, W. Yuan, R.S. Mishra, Effect of inhomogeneous deformation on anisotropy of AZ31 magnesium sheet, Mater. Sci. Eng. A. 567 (2013) 101-109.

[54] X. Wu, J.H. Zhang, J.L. Liu, T. Jin, Y.B. Xu, Z.Q. Hu, Plastic deformation inhomogeneity in a single crystal nickel-base superalloy, Mater. Sci. Eng. A. 325 (2002) 478-483.

[55] P.B. Bowden, A criterion for inhomogeneous plastic deformation, Philos. Mag. 22 (1970) 455-462.

[56] J. Shi, X.Y. Wang, Y.W. Zhao, X.C. Wu, Effects of Heat treatment on microstructure and dynamic shear localization of U-Nb alloy, Rare Metal. Mat. Eng. 43 (2014) 2836-2840 (in Chinese edition).

[57] X.Y. Wang, D.M. Lang, D.P. Ren, W.L. Huang, G.F. Jiang, X.E. Tang, The fracture morphology and microstructure of U-Nb alloy at high strain rate, Rare Metal. Mat. Eng. 34 (2005) 1546-1549 (in Chinese edition).

[58] A. Das, J.K. Chakravartty, Fractographic correlations with mechanical properities in ferritic martensitic steels, Surf Topogr.: Metrol. Prop. 5 (2017) 045006.

[59] Y.Q. Chen, S.P. Pan, M.Z. Zhou, D.Q. Yi, D.Z. Xu, Y.F. Xu, Effects of inclusions, grain boundaries and grain orientations on the fatigue crack initiation and propagation behavior of 2524-T3 Al alloy, Mater. Sci. Eng. A. 580 (2013) 150-158.