Journal of Nuclear Materials 436 (2013) 84–92

Contents lists available at SciVerse ScienceDirect

Journal of Nuclear Materials

journal homepage: www.elsevier .com/locate / jnucmat

Hydride precipitation and its influence on mechanical properties of notched andunnotched Zircaloy-4 plates

Zhiyang Wang a,b,⇑, Ulf Garbe b, Huijun Li a, Robert P. Harrison c, Karl Toppler c, Andrew J. Studer b,Tim Palmer c, Guillaume Planchenault d

a Faculty of Engineering, University of Wollongong, Northfields Avenue, Wollongong, NSW 2522, Australiab The Bragg Institute, Australian Nuclear Science and Technology Organisation, New Illawarra Road, Lucas Heights, NSW 2234, Australiac The Institute of Materials Engineering, Australian Nuclear Science and Technology Organisation, New Illawarra Road, Lucas Heights, NSW 2234, Australiad Electricite De France, 6 Avenue Montaigne, 93192 Noisy Le Grand Cédex, France

a r t i c l e i n f o

Article history:Received 23 November 2012Accepted 24 January 2013Available online 1 February 2013

0022-3115/$ - see front matter � 2013 Elsevier B.V. Ahttp://dx.doi.org/10.1016/j.jnucmat.2013.01.330

⇑ Corresponding author at: Faculty of EngineeringNorthfields Avenue, Wollongong, NSW 2522, Australi+61 2 42213238.

E-mail address: zw603@uowmail.edu.au (Z. Wang

a b s t r a c t

The hydride formation and its influence on the mechanical performance of hydrided Zircaloy-4 platescontaining different hydrogen contents were studied at room temperature. For the unnotched plate sam-ples with the hydrogen contents ranging from 25 to 850 wt. ppm, the hydrides exerted an insignificanteffect on the tensile strength, while the ductility was severely degraded with increasing hydrogen con-tent. The fracture mode and degree of embrittlement were strongly related to the hydrogen content.When the hydrogen content reached a level of 850 wt. ppm, the plate exhibited negligible ductility,resulting in almost completely brittle behavior. For the hydrided notched plate, the tensile stress concen-tration associated with the notch tip facilitated the hydride accumulation at the region near the notch tipand the premature crack propagation through the hydride fracture during hydriding. The final brittlethrough-thickness failure for this notched sample was mainly attributed to the formation of a continuoushydride network on the thickness section and the obtained very high hydrogen concentration (estimatedto be 1965 wt. ppm).

� 2013 Elsevier B.V. All rights reserved.

1. Introduction

Zirconium alloys are extensively used as nuclear fuel claddingand fuel assembly components because of their unique combina-tion of low neutron absorption cross-section, good corrosion resis-tance and attractive mechanical properties. One factor that needsto be considered for zirconium alloy components during serviceis the hydrogen absorption and the subsequent precipitation of hy-drides, which severely degrades the ductility and fracture tough-ness of the material. Moreover, some zirconium alloycomponents are susceptible to a crack initiation and propagationprocess called delayed hydride cracking (DHC), potentially reduc-ing their usable lifetime [1,2]. Due to its technological importance,the hydride-induced embrittlement in zirconium alloys has at-tracted considerable attention in recent decades [2–6]. Previous re-search efforts have established that the degree of theembrittlement depends mainly on the features of the hydridesformed in zirconium alloys, such as the hydride distribution, pre-cipitation location and amount [2,7,8]. It has been found that clo-

ll rights reserved.

, University of Wollongong,a. Tel.: +61 2 42214143; fax:

).

sely spaced hydride platelets are more detrimental to tensileductility than widely spaced hydrides [3,4]. Additionally, Qinet al. suggested that the intergranular hydrides may be more dele-terious to ductility than the intragranular ones [8]. A brittle frac-ture behavior tends to result when a continuous hydride networkforms in the microstructure [5,9]. More interestingly, zirconium al-loys may undergo an abrupt ductile-to-brittle transition at roomtemperature when the hydride or hydrogen content in the materialreaches a critical level [5,6,9–11]. The sensitivity to hydrideembrittlement in zirconium alloys is influenced by multiple fac-tors, including the material microstructure and testing conditions.For example, Bai et al. suggested that the use of fine microstruc-tures with long elongated grains in the loading direction may delaythe ductility reduction of hydrided Zircaloy-4 [5]. Huang and Yehfound that a premature brittle-to-ductile transition occurred in ahydrided Zircaloy-4 sample at room temperature when it wastested in a hydrogen atmosphere with a pressure of 1010 kPa [6].It is important to note that in previous studies the evaluation ofthe influence of hydride precipitation on the mechanical behaviorof zirconium alloys was normally conducted through the uniaxialtensile tests on hydrided samples with relatively uniform distribu-tion of hydrides [5,6,9–11]. However, as indicated by Bertolinoet al. [12], the homogeneous hydride distribution may not properlyrepresent the actual hydride distribution present in the

Z. Wang et al. / Journal of Nuclear Materials 436 (2013) 84–92 85

components under service conditions. In fact, the inhomogeneoushydride distribution in the in-service component has been recog-nized, with the representative example of the formation of macro-scopic hydride blisters in zirconium alloy pressure tubes [13]. It istherefore essential to evaluate the distribution features of hydridesformed under different hydriding conditions (e.g. in the presence/absence of stress gradients), and its potential impact on the failuremicromechanisms of hydrided zirconium alloys, in order to obtainan in-depth understanding on the hydride embrittlement phenom-enon in zirconium alloys.

In this work, a notched Zircaloy-4 plate with the inclusion of awedge sample in the notch was prepared and then subjected togaseous hydriding. The notched sample was designed to inducestresses at the notch tip and to enable the potential effect of thestress on the hydride precipitation and distribution features inthe sample to be studied. Some unnotched plates charged with var-ious hydrogen contents and relatively uniform hydride distributionwere also prepared and the mechanical behaviors of these unnot-ched samples after hydriding were evaluated by the uniaxial ten-sile tests. The main objective of the present study was toascertain the influences of hydride precipitation and its distribu-tion features on the mechanical properties and fracture micro-mechanisms of hydrided Zircaloy-4 plates.

2. Experimental

2.1. Material

Hot-rolled and annealed Zircaloy-4 plate material (ASTM GradeR60804, purchased from Wah Chang, USA) was used in this study.The chemical composition of the Zircaloy-4 plate was Zr-1.56Sn-0.22Fe-0.11Cr-0.14O (wt.%), and the plate exhibited an equiaxedgrain structure with an average grain size of �10 lm, as shownin Fig. 1a. The crystallographic texture of the Zircaloy-4 platewas measured in this study by neutron diffraction and the{0002}, {1010} and {1120} pole figures are demonstrated inFig. 1b. It may be noted that the dominant texture feature is the{0002} double pole maxima inclined at about ±30� from the platenormal direction (ND) and extended towards the transverse direc-

RD

TD

(b)

(a)

Fig. 1. (a) Typical microstructure of the Zircaloy-4 plate matrix; and (b) the texture of{1010} and {1120} pole figures. The rolling direction (RD), transverse direction (TD) an

tion (TD); moreover, there is a {1010} prismatic maximum alongthe rolling direction (RD).

2.2. Hydriding procedures

Prior to the gaseous hydrogen charging, the Zircaloy-4 sampleswere immersed in an acid solution consisting of 20 ml HF, 90 mlHNO3 and 20 ml distilled water for 2 min to remove the oxide layerpresent on the surface. Two sets of plate samples were prepared forhydriding. A notched sample with the geometry schematicallyshown in Fig. 2 was first produced, and a wedge-shaped 316 stain-less steel sample was pressed into the notch. The localized plasticdeformation occurred near the edges of the notch during theassembling process due to the larger size of the steel wedge thanthat of the notch. The as-prepared whole sample (referred to asnotched sample in the following text) was hydrided using highpurity hydrogen (99.9%) in a Sieverts device (Advanced MaterialsCorporation, USA) under a pressure of 60 atm. It was soaked at atemperature of 450 �C for 24 h and then cooled down to room tem-perature in the furnace with an estimated cooling rate of2.2 �C min�1. After hydriding, a short macroscopic crack extendingalong the pre-existed notch direction was observed on the notchedsample (Fig. 2) and, subsequently, the entire through-thicknessrupture of this sample was achieved by hand. Besides, a batch ofunnotched plate samples was hydrided in a self-developed furnaceat ANSTO with a chamber volume of 100 cm3. They were exposedto the high purity hydrogen gas (99.9%) at 1 atm and maintained at520 �C, followed by furnace cooling to room temperature with anapproximate cooling rate of 1 �C min�1. Depending on the expo-sure time, the obtained hydrogen contents for this batch of sam-ples varied from 25 to 850 wt. ppm, as determined using a LECOhydrogen analyzer with an error of �5 wt. ppm.

2.3. Mechanical property testing

To understand the premature rupture behavior of the notchedsample after hydriding, mini-tensile tests on the unnotched platesamples charged with different hydrogen contents were per-formed. Miniature tensile samples with the geometry depicted in

ND

the hot-rolled and annealed Zircaloy-4 plate represented by experimental {0002},d normal direction (ND) are indicated.

Zircaloy-4 plate

Wedge-shaped steel sample

Force

Crack

60atm H2 / 450°C

Furnace

8mm

13mm

~1mm

5mm

RD

TD ~0.8mm

Fig. 2. Schematic drawing showing the set-up for the hydrogen charging experiment with a notched Zircaloy-4 sample.

86 Z. Wang et al. / Journal of Nuclear Materials 436 (2013) 84–92

Fig. 3 were prepared and tested in accordance with the AS 1391-1991 standard. The length and width directions of the tensile sam-ples were parallel to the TD and RD of the plates, respectively. Thegauge length was 10 mm. The mini-tensile tests were carried out atroom temperature on an Instron 5967 universal testing machine atan extension rate of 0.50 mm min�1. The 0.2% offset yield strength(r0.2) and ultimate tensile strength (rUTS) were determined fromthe recorded load-extension curves. The elongation and reductionof area were calculated from the dimensional measurements ofthe original and ruptured tensile samples. For the rupturednotched sample, the reduction of area was also evaluated via thepre- and post-crack measurements. The Young’s modulus (E) ofthe virgin Zircaloy-4 plate and the hydrided sample containing850 wt. ppm of hydrogen were determined by the impulse excita-tion technique using a GrindoSonic instrument (LEMMENS Com-pany, Belgium), in order to monitor the possible changes in Einduced by the hydride precipitation. The measurements of E wereconducted at room temperature along the RD and TD directions ofthe plates.

2.4. Microstructural characterization

The microstructures of hydrided Zircaloy-4 samples wereexamined by mounting them in epoxy, polishing according tometallographic techniques and finishing on 1200 grit silicon car-bide paper. The polished samples were etched with a solutioncontaining HF, HNO3 and H2O in a volume ratio of 1:10:10 for30 s and observed using a JEOL JSM-6490LA scanning electronmicroscope (SEM). The fracture surfaces of the tensile samples

Fig. 3. The geometry of mini-tensile test samples. The width and length directionsof the tensile sample correspond to the RD and TD of the plate, respectively.

after testing or the notched sample after hydriding were exam-ined under SEM. Phase identification of the hydrided sampleswas performed by a GBC MMA X-ray diffractometer (XRD) withCu Ka radiation (k = 1.5418 Å) at 28.6 mA and 35 kV, using a con-tinuous-scan mode at 1 �C min�1. An additional diffraction mea-surement of the notched sample after hydriding was conductedusing neutrons at a wavelength of 2.4143 Å and a beam size of20 mm � 50 mm on Wombat, the high intensity powder diffrac-tometer at ANSTO [14]. By analyzing the acquired neutron diffrac-tion pattern using the Rietveld refinement techniqueimplemented in the GSAS code [15], the phase fraction of hy-drides formed in the notched sample was quantified. The Riteveldphase quantification method enables the accurate phase analysiswithout the requirement of standards or laborious experimentalcalibration procedures [16]. This approach is based on the propor-tional relationship between the weight fraction of a phase and thescale factor derived from the Rietveld analysis, which can be gi-ven by [16]:

Wp ¼ SpðZMVÞpXn

i

SiðZMVÞi

,

where Wp is the relative weight fraction of phase p in a multiphasesystem with n phases, and S, Z, M, and V are, respectively, the Riet-veld scale factor, the number of formula units per cell, the mass ofthe formula unit (in atomic mass units) and the unit cell volume (inÅ3).

3. Results and discussion

3.1. Mechanical properties

Fig. 4 shows the changes of strength and ductility of the hydrid-ed Zircaloy-4 plates with the charged hydrogen contents up to850 wt. ppm. There is an overall trend that the ultimate tensilestrength (rUTS) was slightly increased by increasing the hydrogencontent, while the 0.2% yield stress (r0.2) was almost independentof the hydrogen content (Fig. 4a). Additionally, Table 1 lists theYoung’s modulus (E) of the unhydrided Zircaloy-4 plate and hyd-rided sample with a hydrogen concentration of 850 wt. ppm. Itshows that E was almost unchanged before and after hydriding,with the typical E value maintained at a level of �94 GPa. Grangeet al. [11] proposed that this observation accounted for the factthat r0.2 remained nearly independent of the hydrogen content(Fig. 4a). Further, the stable elastic modulus before and after hyd-riding indicated that the elastic modulus of hydride and Zircaloy-4matrix were similar, both having a magnitude of �94 GPa. Thisconclusion is also supported by a recent study on the hydrided Zir-caloy-2 by Synchrotron X-ray diffraction, showing that the hydridemodulus was around 100 GPa [17]. Regarding the observed slight

Fig. 4. Effect of hydrogen content on the mechanical properties of the hydrided Zircaloy-4 samples: (a) ultimate tensile strength and yield strength; and (b) elongation andreduction of area.

Table 1The Young’s modulus (E) of the virgin Zircaloy-4 plate and hydrided plate containing850 wt. ppm of hydrogen. The measurements are preformed at room temperaturealong the RD and TD of plates.

Sample E measured alongRD (GPa)

E measured alongTD (GPa)

Virgin Zircaloy-4 plate 94.8 94.5Hydrided Zircaloy-4 plate with

850 wt. ppm H94.1 94.1

Fig. 5. Typical XRD patterns of hydrided samples containing various hydrogencontents.

Z. Wang et al. / Journal of Nuclear Materials 436 (2013) 84–92 87

elevation of tensile strength with an increase in the hydrideamount, two major explanations have been proposed in the litera-ture for this strengthening effect. One suggestion is that the hy-drides precipitated in the Zircaloy-4 matrix normally possessedhigher strength [18] than that of the surrounding matrix and henceacting as the reinforcing phase and resulting in the mechanicalstrengthening of the hydrided material. Another suggestion is thata large number of dislocations were generated in the a-Zr matrixduring the hydride formation, contributing to the strengtheningof hydrided Zircaloy-4 [19,20]. The generation of the dislocationsin the matrix is a consequence of the accommodation of the dilata-tional misfit strain associated with the hydride precipitation, andhas been evidenced by transmission electron microscope (TEM)investigations [19,20]. Although the strength of hydrided sampleswas insignificantly affected by the hydrogen content, the ductilityreflected by the elongation and reduction of area (RA) values wasclosely linked to the hydrogen concentration (Fig. 4b). Specifically,the tensile ductility of hydrided plates generally diminished withincreasing hydrogen concentration (i.e. hydride content). Note thatthe virgin Zircaloy-4 had excellent ductility, represented by a highRA of 53% and elongation of 31%. Minor reductions in these valueswere found for the hydrided plate with a low hydrogen content of25 wt. ppm. As the hydrogen content increased to 110 wt. ppm, theRA decreased by �9% whereas the elongations only reduced by�1% compared with the initial values, which suggested that the ef-fect of hydrogen on RA is more distinct than that on elongation [6].At a high hydrogen concentration of 850 wt. ppm, the elongationand RA drastically reduced to 5% and 10%, respectively, indicatingthe substantial loss of ductility. Furthermore, for the hydridednotched sample, the RA was close to zero based on the pre- andpost-crack measurements, revealing that a ductile-to-brittle tran-sition occurred in this case.

3.2. Hydride phase identification

As the type of hydrides precipitated in zirconium alloys de-pends on the cooling rate applied during hydriding and theachieved hydrogen content [2,21,22], it is necessary to identifythe hydride phase present in the hydrided samples. Fig. 5 depictsthe typical XRD patterns of hydrided Zircaloy-4 plates containingvarious hydrogen contents. The diffraction peaks correspondingto hexagonal closed-packed (hcp) a-Zr (JCPDS Card No. 05-0665)were generally identified in all hydrided samples. Besides the a-Zr diffraction peaks, relatively weak peaks corresponding to face-centered cubic (fcc) d-ZrH1.66 (JCPDS Card No. 34-0649) were alsodetected in the hydrided sample containing 850 wt. ppm H. Forthe samples with hydrogen contents in the range of 25–450 wt. ppm, the hydride phase was not detectable by the presentXRD measurements (the representative XRD pattern of the samplewith 450 wt. ppm H was shown here). This is related to the lowvolume fractions of hydrides for the samples with low levels ofhydrogen (6450 wt. ppm). Nevertheless, it is rational to assume

Fig. 6. The GSAS refinement of the neutron diffraction pattern measured from thenotched sample after hydrogen charging. The graph shows the measured datapoints (+ symbols), the refinement result (fitted line) and their difference curve. Thev2 parameter, a metric indicating the goodness of fit between the actual data andthe fitted pattern, was 2.1 in this work.

88 Z. Wang et al. / Journal of Nuclear Materials 436 (2013) 84–92

that d-hydrides were present at low hydrogen concentrations(6450 wt. ppm), as the slow cooling rate applied in the hydridingprocess (1 �C min�1) favored the formation of the equilibrium d-ZrH1.66 phase [2,23]. For the notched sample after hydriding, theevident diffraction peaks for the d-ZrH1.66 phase were detected,which was also manifested by the neutron diffraction measure-ment, as revealed in Fig. 6. Considering that only d-hydrides werepresent in the samples and the hydrogen solubility in the Zircaloy-4 matrix was very low (less that 10 wt. ppm at room temperature[22]), it is safe to conclude that a greater concentration of hydrideswere precipitated at higher hydrogen levels. Further, the Rietveldrefinement analysis, which allows the quantification of phase frac-tion, was performed on the recorded neutron diffraction (Fig. 6).The refinement result showed a weight fraction of 11% for d-ZrH1.66 phase, corresponding to a hydrogen content of�1965 wt. ppm. As the neutron diffraction data representing theinformation from the bulk sample were used in the Rietveld phase

Fig. 7. SEM micrographs showing typical hydride morphologies in the hydrided Zirc850 wt. ppm.

quantitative analysis, the high accuracy of the determined hydridecontent should be expected. Based on this quantitative analysis, itis reasonable to consider that, compared with other hydrided sam-ples in this study, a higher hydrogen concentration, i.e. a largeramount of hydrides was attained for the notched sample after hyd-riding under a hydrogen pressure as high as 60 atm.

3.3. Metallographic and fractographic observations

3.3.1. Hydrided plates containing 25–850 wt. ppm of hydrogenFig. 7 shows the typical microstructures of the Zircaloy-4 plates

with hydrogen contents ranging from 110 to 850 wt. ppm. Thecommon acicular or platelet-like hydrides were clearly observedand, as expected, the distribution density of hydrides steadily in-creased with increasing hydrogen concentration. In addition, themicroscopic features of the hydrides in terms of the precipitationsites and alignments exhibited some distinct variations withhydrogen contents. At a relatively low hydrogen content of110 wt. ppm, the randomly distributed grain boundary hydrideplatelets were predominantly observed (Fig. 7a). This result sup-ported the previous suggestion that grain boundaries (GBs) maybe preferred sites for hydride formation relative to the intra-grainregion [8,24]. As the hydrogen content was increased to 380 or450 wt. ppm, intergranular hydrides as well as the gradual devel-opment of intragranular hydride needles were observed (Fig. 7band c). According to the theoretical analysis by Qin et al. [8], thefavorable GBs for the precipitation of intergranular hydrides wereessentially those lying on the basal plane of the matrix (the usualhabit plane of hydrides [25]) and having a high energy. When ahigh level of hydrogen (P380 wt. ppm) was introduced into thematrix, the preferred GBs sites tended to exhaust for the formationof intergranular hydrides, and the excess hydrogen could promotethe development of intragranular hydrides (Fig. 7b–d). At an evenhigher hydrogen content of 850 wt. ppm, the intergranular andintragranular hydrides were both frequently observed. These hy-dride platelets became thicker and were preferentially alignedalong the RD of the plate (Fig. 7d). The microstructure of the platewith a very low hydrogen concentration of 25 wt. ppm was similarto that of the virgin plate (Fig. 1a), demonstrating little or no evi-

aloy-4 plates with various hydrogen contents: (a) 110; (b) 380; (c) 450 and (d)

Z. Wang et al. / Journal of Nuclear Materials 436 (2013) 84–92 89

dence of hydrides. Apart from the hydrides primarily concerned, itcan be noted that the initial equiaxed a-Zr grain structure wasmaintained after hydriding.

To investigate the fracture behaviors of hydrided Zircaloy-4 andunderstand the hydride-induced embrittlement, the fracture sur-faces of hydrided plates with different hydrogen contents(6850 wt. ppm) were examined, as illustrated in Fig. 8. At a lowhydrogen content of 25 wt. ppm, the hydrided plate clearly under-went a ductile fracture process, evidenced by the abundant dimplestructures with a mean size of �10 lm present in the fracture sur-face (Fig. 8a). When the hydrogen content rose to 110 wt. ppm,some secondary cracks appeared (Fig. 8b) while the fracture sur-face still dominantly contained dimpled regions. It is also foundthat some short secondary cracks (pointed in Fig. 8b) were coinci-dent with the hydride precipitation sites, which implied that thehydrides contributed to the development of secondary cracks. Ata hydrogen concentration of 450 wt. ppm, besides the dimplestructures and secondary cracks, some cleavage facets around thecracks were distinctly observed, as indicated by arrows in Fig. 8c.In contrast, the fracture surface at an even higher hydrogen levelof 850 wt. ppm exhibited the brittle features, characterized by con-siderable cleavage facets and longer secondary cracks along thecrack propagation direction (Fig. 8d). In this case, small-sized andshallow dimple structures were occasionally observed in the sur-face (Fig. 8e), suggesting the abrupt reduction of ductility.

The hydrides present in zirconium alloys are recognized as afracture initiator because of their intrinsic brittleness, as reflectedby the low fracture toughness of the order of 1 MPa m1/2 at room

Fig. 8. SEM images showing characteristic fracture surfaces of hydrided plates containinoccasionally observed dimple structures at 850 wt. ppm. Crack growth direction is from

temperature [26]. It has been reported that the presence of brittlehydride phase, which provides an easy crack path, played a keyrole in determining the fracture micromechanisms and ductilityof hydrided zirconium alloys [5,6,27]. The combination of metallo-graphic and fractographic investigations (Figs. 7 and 8) explicitlysuggested that the fracture behavior of hydrided Zircaloy-4 platesunder uniaxial tension loading was closely related to the hydrideconcentration (i.e. hydrogen content). With increasing hydrogencontent, the fracture mode experienced a successive change froma pure ductile fracture (25 wt. ppm) to a mixture of ductile andbrittle fracture (450 wt. ppm), and then to a near complete brittlefracture (850 wt. ppm). At the low hydrogen level of 25 wt. ppm,negligible hydrides were formed, thereby leading to less degrada-tion of ductility (Fig. 4b) and the resultant ductile fracture(Fig. 8a). At 450 wt. ppm H, in conjunction with the metallographicobservation (Fig. 6c), it could be confirmed that the crack path pri-marily followed the brittle hydrides, which ruptured by cleavage,whereas the ductile matrix between hydrides fractured by micro-void coalescence. As increasing the hydrogen content to850 wt. ppm, the brittle fracture of hydrides with a high concentra-tion in this range may lead to the prevalence of cleavage fractureand, moreover, as these hydrides were preferentially aligned alongthe RD of the plate (Fig. 6d), the long secondary crack could be pro-gressively generated through linking up the adjacent fractured hy-drides arranged in the crack advancing direction (Fig. 8d).Furthermore, the almost complete brittle behavior exhibited in thiscase coupled with the drastic reduction of ductility (Fig. 4b), indi-cated that critical hydrogen content beyond which the ductile-to-

g different hydrogen contents: (a) 25; (b) 110; (c) 450 and (d) 850 wt. ppm. (e) Thebottom to top.

90 Z. Wang et al. / Journal of Nuclear Materials 436 (2013) 84–92

brittle transition occurred at the uniaxial tensile condition wasapproaching 850 wt. ppm.

3.3.2. Notched plate after hydridingFig. 9b–e display the characteristic microstructures for the hyd-

rided notched sample at various regions. The hydride morphologyand distribution features at these selectively observed areas clearlyexhibit some differences (Fig. 9b–e). Nevertheless, a close compar-ison of Figs. 7 and 9b–e reveals that the average concentration ofhydrides in the notched plate was generally higher than that inother hydrided plates investigated here, consistent with the hy-dride quantification analysis via neutron diffraction (Fig. 6). Atthe region below the fracture surface (i.e. the RD–ND section), agreat amount of grain boundary hydrides were precipitated, form-ing a continuous network along grain boundaries and, meanwhile,many long intragranular hydrides going from side to side of indi-vidual a-Zr grains and near dense hydride regions were present(Fig. 9b). The brittle hydrides distributed in this manner signifi-cantly destroyed the microstructural integrity of the initial mate-rial. At the region far from the notch edge (indicated by ‘‘A1’’ inFig. 9a), the microstructure consisted of many long hydride plate-lets, which tended to align along the RD of the plate (Fig. 9c). Also,

Zircaloy-4 notched plate

Steel wedge

Developing Crack

A2

RD

TD

A1

Region with marked plastic deformation

Biaxial tensile stress concentration region

60atm H2 / 450°C

ND

A3

(a)

(

(e)

(c)

(

Fig. 9. (a) Schematic illustrating the region with marked plastic deformation in the notchthe biaxial tensile stress concentration area around the notch resulting from the thermalmorphologies in the hydrided notched plate at different regions: (b) below the fracture s‘‘A1’’ in (a); (d) near the notch edge, as marked by ‘‘A2’’ in (a); and (e) near the edge of

the hydride concentration in this area seems to be lower than thatin the region below the fracture surface (Fig. 9b). It was believedthat the concentration of biaxial tensile stress at the root of thenotch generated by the inserted steel wedge (schematically shownin Fig. 9a) enhanced the hydrogen diffusion during hydriding to-wards the notch root [28] and, accordingly, resulting in a consider-ably higher concentration of hydrides precipitated in the areabelow the fracture surface (Fig. 9b). Although the direct measure-ment of the stress state around the notch root was presently unre-alized in this work, it is reasonable to assume that, the pre-existednotch (Fig. 2) introduced a biaxial tensile stress distribution with astress concentration at the root of the notch during hydriding(Fig. 9a). On one hand, the constrained 316 stainless steel wedgein the designed notch possesses a high linear thermal expansioncoefficient (17.5 � 10�6 �C�1 [29]), nearly three times the valuefor Zircaloy-4 (6 � 10�6 �C�1 [30]). Consequently, the steel wedgeeffectively exerted a compressive force to the surrounding Zirca-loy-4 plate during hydriding at 450 �C and, accordingly, generatingthe biaxial tensile stress at the root of the notch. On the other hand,previous theoretical and experimental studies have confirmed theexistence of biaxial tensile stress concentration at the root of thenotch [31]. At the area near the notch edge (marked by ‘‘A2’’ in

d)

b)

ed Zircaloy-4 plate generated during the installation of large-sized steel wedge andexpansion of steel wedge during hydriding; (b–d) SEM images showing the hydride

urface and close to the root of the notch; (c) far from the notch edge, as indicated bythe developed crack, as indicated by ‘‘A3’’ in (a).

Fig. 10. (a) SEM image showing typical fracture surface of the ruptured notched plate after hydriding; and (b) is a local magnification of (a). Crack growth direction is frombottom to top.

Z. Wang et al. / Journal of Nuclear Materials 436 (2013) 84–92 91

Fig. 9a), a considerable number of long hydride platelets thataligned along the RD of the plate was present in the microstructureand, interestingly, a hydride content gradient was observed, show-ing a greater concentration of hydrides at the position approachingthe notch edge (Fig. 9d). As previously stated, the marked plasticdeformation regions around the notch edges were created(Fig. 9a) due to the forced fit of the large-sized 316 stainless steelwedge into the notch. The plastic deformation is normally accom-panied by the generation of microstructural defects (e.g. vacanciesand dislocations), which could act as hydrogen traps [32,33] andprovide preferred hydride nucleation sites [34,35], thereby facili-tating the formation of more hydrides at the position close to thenotch edge (Fig. 9d). At the region near the crack edge (indicatedby ‘‘A3’’ in Fig. 9a), a similar microstructure to that of Fig. 9d wasfound, demonstrating numerous hydride platelets with a pro-nounced orientation along the RD and a decreasing hydride con-tent along the direction from the crack to the extreme of thesample (Fig. 9e). Based on Fig. 9d and e, it is important to pointout that the plane stress state supposed in the thin notched samplecontributes to the marked orientation of hydrides formed near thenotch/crack edges, yielding a dominant orientation of hydridesperpendicular to the stress component direction.

Fig. 10a depicts the fracture surface for the hydrided notchedsample. Extensive flat cleavage facets appeared in the fracture sur-face while no microvoid coalescence was found, demonstrating acomplete brittle fracture behavior (Fig. 10a). A close examinationof Fig. 10a also reveals that many cleavage facets were surroundedby short secondary cracks. Combined with the metallographic find-ing that a continuous hydride network was formed in this case(Fig. 9b), it was considered that the fracture path for this notchedsample mainly followed the intergranular hydrides, leading to theemergence of extensive cleavage facets with sizes comparable tothat of the a-Zr grains (Fig. 10b). In addition, a high-magnificationSEM micrograph (Fig. 10b) shows that many secondary cracks al-most perpendicular to the crack propagation direction are presentin the brittle facet area, implying that this area is much more brittlethan the surrounding areas. These secondary cracks were identifiedto coincide with the locations of some intergranular hydrides(Fig. 9b), which indicates that the fracture of intragranular hydridesalso occurred during the rupture process of the notched sample.

For the notched plate, it was notable that the crack was prema-turely initiated and propagated along the notch direction duringhydriding at 450 �C and under a high hydrogen pressure of60 atm (Fig. 2). This may be attributed to a synergistic effect ofhydrogen gas and tensile stress concentration at the root of notchin the sample (Fig. 9a). The metallographic observation (Fig. 9b)showed that a great concentration of hydrides tended to accumu-late at the region near the root of the notch, resulting from the en-hanced hydrogen diffusion by the tensile stress concentration atthis region. According to Huang and Yeh’s study [6], it was thoughtthat these concentrated hydrides may be instantly cracked because

of the stress concentration at the notch root, i.e. the delayed hy-dride cracking (DHC) occurs. This hydride cracking was possibleat the hydriding temperature of 450 �C because the d-hydride(ZrH1.66) remains brittle in tension up to 500 �C [36] with the frac-ture toughness varying slightly from 1 MPa m1/2 at room tempera-ture to 3–4 MPa m1/2 at 300 �C [7]. Therefore, many microcrackscould be generated by the fracture of the accumulated hydrides(Fig. 10). During the hydriding process using a high hydrogen pres-sure, the occurrence of DHC associated with the microcracks actingas short circuits for the further hydrogen penetration into the re-gion ahead of the notch root [6] promotes the continuous hydrideprecipitation at the notch tip. Again, the cracking of these precipi-tated hydrides could be triggered by the concentrated stress. Dueto this repeated hydride formation and fracture process (DHC),the crack propagation could be progressively realized throughthe fracture of hydrides, which was promoted by the tensile stressconcentration. However, the DHC only occurs when hydride con-centration increases to a critical value under certain stress level.Further, as the stress concentration was only limited in the regionnear the notch root, the propagation of the crack may be arrestedwhen the crack tip reaches the regions with a low level of stress,so only a short flaw was developed during hydriding (Fig. 2). Afterhydriding, the ease in breaking this thin notched sample as well asthe measured value of RA approaching zero reveals that thenotched sample became totally brittle, in agreement with the frac-tographic analysis (Fig. 10). Although the actual hydrogen contentin the hydrided notched sample was not measured, it is reasonableto consider that this sample had a very high hydrogen content, atleast, exceeding 850 wt. ppm, based on the combined metallo-graphic and neutron diffraction analyses (Figs. 6 and 9b–d). Over-all, two main factors, i.e. the very high hydrogen concentrationachieved in this thin notched sample (thickness �0.8 mm) andthe formation of a continuous hydride network on the sectionalong the thickness direction, contributed to the emergence of a ra-pid ductile-to-brittle transition at room temperature and the resul-tant through-thickness brittle failure for the hydrided notchedsample.

4. Conclusions

In this work, the hydride precipitation and its influence on themechanical properties of the hydrided Zircaloy-4 plates with vari-ous hydrogen contents were investigated at room temperature.Special attention was devoted to the analysis of the hydride distri-bution features and premature cracking behavior in a designednotched plate. The most important conclusions are summarizedin the following:

(1) Hydrides exerted only minor influences on the ultimate ten-sile strength and yield stress of the hydrided Zircaloy-4plates containing hydrogen contents up to 850 wt. ppm.

92 Z. Wang et al. / Journal of Nuclear Materials 436 (2013) 84–92

(2) The uniaxial tensile ductility of the hydrided plates was sig-nificantly degraded by the hydride precipitation. The frac-ture mode and the degree of embrittlement were closelydependent on the hydrogen content.

(3) Under the uniaxial tensile condition, the hydrided plate with850 wt. ppm H exhibited a negligible ductility with the elon-gation of 5% and reduction of area of 10%, leading to almostcompletely brittle fracture behavior.

(4) For the hydrided notched plate, the tensile stress concentra-tion associated with the notch tip promoted the hydrideaccumulation at the region near the notch tip and the pre-mature crack propagation through the hydride fracture dur-ing hydriding. The eventual brittle through-thickness failurefor this notched sample was mainly ascribed to the forma-tion of a continuous hydride network on the thickness sec-tion and the obtained very high hydrogen concentration(estimated to be 1965 wt. ppm).

Acknowledgments

The authors thank Dr. Jianfeng Mao and Dr. Mohammad Ismailfor their technical assistance with the hydrogen charging experi-ments. One of the authors (Zhiyang Wang) is very grateful forthe financial supports from the China Scholarship Council (CSC)and the Australian Nuclear Science and Technology Organisation(ANSTO).

References

[1] K. Nuttall, D.P. McCooeye, A.J. Rogowski, F. Havelock, Scripta Metall. 10 (1976)979–982.

[2] D.O. Northwood, U. Kosasih, Int. Mater. Rev. 28 (1983) 92–121.

[3] C.E. Coleman, D. Hardie, J. Less-Common Met. 11 (1966) 168–185.[4] K.S. Chan, Acta Metall. Mater. 43 (1995) 4325–4335.[5] J.B. Bai, C. Prioul, D. François, Metall. Mater. Trans. A 25 (1994) 1185–1197.[6] J. Huang, M. Yeh, Metall. Mater. Trans. A 29 (1998) 1047–1056.[7] L.A. Simpson, C.D. Cann, J. Nucl. Mater. 87 (1979) 303–316.[8] W. Qin, N.A.P. Kiran Kumar, J.A. Szpunar, J. Kozinski, Acta Mater. 59 (2011)

7010–7021.[9] J.B. Bai, D. François, C. Prioul, S. Lansiart, Scripta Metall. 25 (1991) 2559–2563.

[10] F. Yunchang, D. Koss, Metall. Mater. Trans. A 16 (1985) 675–681.[11] M. Grange, J. Besson, E. Andrieu, Metall. Mater. Trans. A 31A (2000) 679–690.[12] G. Bertolino, J. Perez Ipiña, G. Meyer, J. Nucl. Mater. 348 (2006) 205–212.[13] M.A. Vicente Alvarez, J.R. Santisteban, G. Domizzi, J. Almer, Acta Mater. 59

(2011) 2210–2220.[14] A.J. Studer, M.E. Hagen, T.J. Noakes, Phys. B 385–386 (2006) 1013–1015.[15] A.C. Larson, R.B. Von Dreele, General Structure Analysis System (GSAS), Los

Alamos National Laboratory, Report LAUR 86-748, 2004.[16] R.J. Hill, C.J. Howard, J. Appl. Crystallogr. 20 (1987) 467–474.[17] M. Kerr, M.R. Daymond, R.A. Holt, J.D. Almer, J. Nucl. Mater. 380 (2008) 70–75.[18] S. Arsene, J.B. Bai, P. Bompard, Metall. Mater. Trans. A 34 (2003) 579–588.[19] G.J.C. Carpenter, J.F. Watters, R.W. Gilbert, J. Nucl. Mater. 48 (1973) 267–276.[20] J.B. Bai, N. Ji, D. Gilbon, C. Prioul, D. François, Metall. Mater. Trans. A 25 (1994)

1199–1208.[21] B. Nath, G.W. Lorimer, N. Ridley, J. Nucl. Mater. 58 (1975) 153–162.[22] R.S. Daum, Y.S. Chu, A.T. Motta, J. Nucl. Mater. 392 (2009) 453–463.[23] K.V. Mani Krishna, A. Sain, I. Samajdar, G.K. Dey, D. Srivastava, S. Neogy, R.

Tewari, S. Banerjee, Acta Mater. 54 (2006) 4665–4675.[24] J.E. Bailey, Acta Metall. 11 (1963) 267–280.[25] N.A.P. Kiran Kumar, J.A. Szpunar, Z. He, J. Nucl. Mater. 403 (2010) 101–107.[26] L. Simpson, M. Puls, Metall. Mater. Trans. A 10A (1979) 1093–1105.[27] S. Oh, C. Jang, J.H. Kim, Y.H. Jeong, Mater. Sci. Eng. A 527 (2010) 1306–1313.[28] M. Puls, Metall. Mater. Trans. A 21A (1990) 2905–2917.[29] J.R. Davis (Ed.), Properties and Selection: Irons, Steels, and High-Performance

Alloys, ASM International Materials Park, 1990.[30] D.L. Douglass, The Metallurgy of Zirconium, International Atomic Energy

Agency, Vienna, Austria, 1971.[31] M. Kerr, M.R. Daymond, R.A. Holt, J.D. Almer, Acta Mater. 58 (2010) 1578–

1588.[32] N. Gerard, J. Less-Common Met. 131 (1987) 13–23.[33] K. Une, S. Ishimoto, J. Nucl. Mater. 322 (2003) 66–72.[34] V. Perovic, G.C. Weatherly, J. Nucl. Mater. 126 (1984) 160–169.[35] C. Roy, Atomic Energy of Canada Limited, Report, AECL-2297, 1965.[36] K.G. Barraclough, C.J. Beevers, J. Mater. Sci. 4 (1969) 518–525.