zirconium oxidation on the atomic scale
TRANSCRIPT
ARTICLE IN PRESS
Ultramicroscopy 109 (2009) 667–671
Contents lists available at ScienceDirect
Ultramicroscopy
0304-39
doi:10.1
� Corr
E-m
journal homepage: www.elsevier.com/locate/ultramic
Zirconium oxidation on the atomic scale
Daniel Hudson �, Alfred Cerezo, George D.W. Smith
Department of Materials, University of Oxford, Oxford OX1 3PH, UK
a r t i c l e i n f o
PACS:
07.78.+s
81.05.Bx
81.65.Mq
82.20.�w
Keywords:
3D atom probe analysis
Field ion microscopy
Zirconium
Sub-oxide
91/$ - see front matter & 2008 Elsevier B.V. A
016/j.ultramic.2008.10.020
esponding author. Tel.: +441865283658.
ail address: [email protected]
a b s t r a c t
Zirconium alloys are used in the nuclear industry as fuel rod cladding. They are chosen for this role
because of their good mechanical properties and low thermal neutron absorption. Oxidation of these
alloys by coolant is one of the chief limiting factors of the fuel burn-up efficiency. The aim of the present
study is to understand these oxidation mechanisms. As a first step, a fundamental study of the oxidation
of commercially pure zirconium has been conducted using the 3D atom probe (3DAP). The current
generation of 3DAPs allows both voltage and laser pulsing, providing data sets of many millions of ions.
According to the literature the only stable oxide of zirconium is ZrO2. However, the 3DAP shows that
an initial layer a few nanometres thick forms with a composition of ZrO1�x when subjected to light
oxidation. This result confirms and extends the work of Wadman et al. [Colloque de Physique 50 (1989)
C8 303; Journal de Physique, 11 (1988) C6 49] and Wadman and Andren [in: C.M. Euchen, A.M. Garde
(Eds.), Zirconium in the Nuclear Industry: Ninth Symposium, ASTM STP 1132, ASTM, USA, 1991, p. 461],
who used 1DAP techniques, obtaining reduced data sets. Segregation of hydrogen to the metal–oxide
interface and a distinct ZrH phase were observed in this study. A novel kinetics study of the room
temperature oxidation of zirconium showed the ZrO layer to be non-protective over the time period
investigated (up to 1 h).
& 2008 Elsevier B.V. All rights reserved.
1. Introduction
Atom probe experimentalists examining zirconium and itsalloys have long been hampered by the tendency of the specimenneedles to break due to the stresses of analysis [1,2]. Wadmanet al. [2–4] used a 1D atom probe technique to examine Zircalloy-4.They found that an amorphous layer of composition similar toZrO formed on their field ion microscopy (FIM) needles [2].Nano-crystalline oxide particles were observed in this layer usingTEM. These atom probe results suffered from very limitedcounting statistics, typically a few hundred atoms per sample.Sub-oxides of zirconium have been suggested by other authors[5], but the current consensus of opinion in the literature is thatthe only stable oxide is ZrO2 [6]. Several systems demonstrate thetendency for different oxides to grow in very different ratios ofthickness [5] and it is possible that the presence of nanometricZrO oxide could be masked from many analytical techniques suchas electron or X-ray diffraction by a much thicker layer of ZrO2
grown on the ZrO [7].Zirconium dioxide is an n-type semiconducting ceramic
material [8]. The mechanism of growth is believed to be aniondiffusion from the surface towards the metal–oxide interface [5].If in fact there were to be an intermediate layer of ZrO betweenZrO2 and the metal, then cation diffusion would be expected in the
ll rights reserved.
(D. Hudson).
inner layer and the formation of new oxide would be expected atthe oxide–oxide interface [5]. It is therefore of upmost importanceto fully characterise the nature of the metal–oxide interface inorder to allow a full mechanistic understanding of the complexoxide growth of zirconium-based metals.
A variety of zirconium alloys are used in the construction ofnuclear fuel rod cladding. Analysis of the fundamental oxidationof pure zirconium has been conducted to act as a base line forfuture investigations into these various alloys. The 3D atom probe(3DAP) [9] has been used in the analysis of needle-like specimenssubjected to mild oxidation in various conditions. A localelectrode atom probe (LEAPs)1 [10] was used in the analysis ofmore heavily oxidised specimens, and to verify the results foundusing the 3DAP.
2. Experimental method
Zirconium was supplied in the form of as-drawn wire of0.125 mm diameter and 99.2% nominal purity purchased fromGoodfellow of Cambridge [11]. The composition of the bulkmaterial from which the wire was drawn is shown in Table 1. Ashas previously been reported, the use of FIM/atom probetechniques has a tendency to cause zirconium-based materials
1 LEAPs is a registered trademark of Imago Scientific Instruments Corporation,
5500 Nobel Drive, Madison, WI 53711, USA.
ARTICLE IN PRESS
D. Hudson et al. / Ultramicroscopy 109 (2009) 667–671668
to fracture during analysis [1,2]. It was found that vacuumannealing the wire at 700 1C for 5 h vastly improved the lifespanof specimens [1].
Specimens were prepared from the vacuum-annealed wireusing a two-stage electro-polishing process. The first stageelectrolyte consisted of 25% perchloric acid (60%), 75% acetic acid(100%) with an electric potential of 20 V D.C. The second stage was2% perchloric acid (60%), 98% 2-butoxyethanol at 10 V DC. Thespecimens were polished at room temperature. These solutionshave been commonly used for the FIM preparation of manymetals, such as iron, nickel, manganese, aluminium, uranium andzirconium [12].
Zirconium was field evaporated in the atom probe at atemperature of 90 K. The pulse fraction used was 18% at afrequency of 20 kHz with �8 ns pulse duration. The evaporationrate (ions detected per pulse) was maintained at a level of 1%.When operated in this manner the 3DAP gives an adequate massto charge resolution, typically FWHM ¼ 1/500.
Field evaporation or 3DAP analysis was used to remove fromthe apex region any impurities introduced during electro-polish-
Table 1Nominal compositional information for 99.2% pure bulk zirconium.
Element ppm (wt.) ppm (at.) at%
C 250 1887.8 0.19
Hf 2500 1269.2 0.13
Fe 200 203.8 0.02
Cr 200 348.5 0.03
N 100 646.9 0.06
O 1000 5663.7 0.57
H 10 898.9 0.09
Zr 995,740 98,901.1 98.91
Total impurity: 0.426 wt%.
Fig. 1. Concentration-depth profile from the 3DAP analysis of a specimen oxidised
in 150 1C air for 15 min.
Fig. 2. Separated 3D atom maps of a specimen oxidised in boiling water
ing. These field-evaporated ‘clean’ specimens were then subject tocontrolled oxidation. Oxidation was carried out in air at roomtemperature, air in a furnace at 150 1C or boiling water, for avariety of time periods between a few seconds and several hours.A kinetics curve was plotted for a sample by repeated oxidationand analysis, the destructive analysis process effectively cleaningthe surface for the next data point.
The LEAPs is also a 3DAP with enhanced field of view andimproved mass/charge resolution [10]. Laser pulsing was usedwith the LEAP to verify the experimental result found in the 3DAPusing different equipments. The LEAP analysis was undertaken ata temperature of (7070.1) K using a pulse duration of aprox. 10 ps,at a frequency of 200 kHz and pulse energy of (0.670.02) nJ. TheLEAP was also used to analyse specimens that had undergonemore severe oxidation and formed ceramic oxides, which cannotbe analysed using the conventional voltage-pulsing atom probes,which require the sample to be conductive.
The time of flight and positional data collected by the atomprobe were reconstructed into a series of 3D atom maps using theIVASTM software package [13]. The data from each specimen arepresented here in a variety of forms such as atom maps,compositional profiles and ladder diagrams.
3. Results
3.1. The initial oxidation of zirconium
Zirconium was oxidised for short periods at elevated tempera-tures of up to 150 1C, or for longer periods at room temperature.The result of this oxidation was the formation of a surface filmwith the composition ZrO1�x, where x is small, typically less than0.1. This result can be seen in the concentration profile in Fig. 1, aspecimen oxidised in air at 150 1C for 15 min. A ZrO oxide layerwith thickness of the order of a few nanometres was also seen onneedles oxidised in boiling water and the oxide produced duringsample preparation had the same composition. Figs. 2 and 3 are3D atom maps that show examples of the distribution of ionsbetween ZrO regions, hydrogen-rich layers and the matrix withinthese oxidised samples. A more precise value for the compositionof the oxide film can be calculated from a ‘ladder diagram’ ofoxygen against zirconium-containing ions detected in the sample.The ladder diagrams in Fig. 4 give values of the ratio of O/Zr asmeasured from the gradient of the curve. This information issummarised in Table 2.
The 3DAP is equally sensitive to all species irrespective ofmolecular weight. This makes the technique suitable for the studyof hydrogen uptake by material and its segregation within thebulk. Segregation of H and ZrH ions to the metal–oxide interfacewas observed in several specimens oxidised in air, an example is
for 3 s. The key shows the ion type detected and associated colour.
ARTICLE IN PRESS
Fig. 3. 3D atom map of a specimen oxidised during sample preparation.
Fig. 4. O/Zr ladder diagrams for specimens oxidised (a) in boiling water for 3 s, (b)
by the sample preparation process and (c) in 150 1C air for 10 min.
Table 2Summary of data from Fig. 4.
Oxidation method Temp. (1C) O/Zr ratio in oxide O/Zr ratio in matrix
Water 100 0.90270.001 0.005470.0002
Polishing Ambient 0.90770.001 0.004770.0003
Air 150 0.92570.001 0.004670.0003
Fig. 5. A comparison between the distinctive Zr profile seen in the 2+ charge state
(above) and a Zr/ZrH profile seen in hydride regions (below).
D. Hudson et al. / Ultramicroscopy 109 (2009) 667–671 669
given in Fig. 1. ZrH species can be distinguished from Zr because ofa change in the distinctive zirconium mass/charge spectrum, asillustrated in Fig. 5. More significant hydrogen uptake was alsoobserved in some air-oxidised samples, to the extent where aseparate hydride phase is formed. A wedge-shaped hydrogen-richregion with uniform composition of approximately ZrH is shownin Fig. 6. The transition between the hydride and the matrix takesplace within the 2 nm red region. Near the interface the mainhydrogen-rich ion type changes from ZrH(2+) to H(+). Thecalculated compositional information for this data set is given inTable 3.
3.2. More severe oxidation of zirconium
In order to establish the validity of these findings it was alsonecessary to examine specimens demonstrating the expected
composition. Fig. 7 shows a concentration profile for a specimenoxidised in air at 250 1C for 30 min. The expected ZrO2 phase ispresent. This data was collected using the LEAP in laser-pulsingmode, where the other data sets in Figs. 1–6 were created usingthe conventional 3DAP using voltage pulsing.
In order to rule out the possibility of a significant systematicerror in the interpretation of the results it is necessary to showthat the ZrO result for short periods of oxidation is also obtainedusing the LEAP. Figs. 8 and 9 are the 3D atom map andconcentration profile, respectively, for a specimen oxidised duringthe electro-polishing process and room temperature air andanalysed using the LEAP. From the concentration profile Fig. 9we see that the composition of the oxide is ZrO and that hydrogensegregates to the interface, the same result seen in the conven-tional 3DAP in Fig. 1, although introduced by a different method.
3.3. Short-term oxidation kinetics
The initial oxidation kinetics of atomically clean zirconium inroom temperature air is shown in Fig. 10. Each data point iscreated by a separate 3DAP data set, where the number of oxygenatoms in the oxide area of the region of interest (ROI) is divided bythe surface area of the ROI. The error in each point is derived fromthe uncertainty in count, oxidation time period and ROI surfacearea. The zero offset suggests that a very thin oxide layer growsvery rapidly, on a timescale shorter than can be successfullyexamined using this technique. The curve is linear within thelimits of error, suggesting that the initial oxide is effectively non-protective over this small time interval.
4. Discussion
The purity of the commercially pure zirconium was found to besignificantly better than the manufacturers’ nominal rating givenin Table 1. All minor spectra peaks could be identified and therewere very few multiple detector hits (typically less than 0.2%),
ARTICLE IN PRESS
Fig. 9. A LEAP concentration-depth profile of a specimen oxidised during sample
preparation.
Fig. 6. Separated 3D atom maps of a hydride-containing sample created by exposure to room temperature air for 15 min. The composition of the wedge-shaped region is ZrH
and was detected as Zr, ZrH and H ions.
Table 3Calculated compositional information for a uniform hydride region of approxi-
mately 2 million ions.
Species Total ions 71 s.d. at% 71 s.d. (%)
Zr 1,014,996 1007.5 50.265 0.035
O 6883 83.0 0.341 0.004
H 997,421 998.7 49.394 0.035
H/Zr ratio 0.9827 0.0014
Data taken from within the ZrH wedge shape region in Fig. 5.
Fig. 7. A LEAP concentration-depth profile of a specimen oxidised in air at 250 1C
for 30 min.
Fig. 8. A LEAP atom map of a specimen oxidised during sample preparation.
D. Hudson et al. / Ultramicroscopy 109 (2009) 667–671670
which limits the accuracy of the technique. So there can beconfidence that the ionic species emitted from the tip weredetected in the correct ratio. The transition between the ZrO layerand the metal was sharp, as seen in Figs. 1 and 4. The level ofoxygen seen in solid solution in the zirconium metal in Table 2was similar to the total amount of oxygen expected to be in bulkfrom Table 1.
ZrO2 layers cannot be analysed using voltage pulsing in theatom probe as it is an insulating ceramic. Similarly the kineticscurve plotted using the 3DAP in voltage-pulsing mode cannot beextended to show a transition between different oxides. This plotwas merely intended to identify the protective properties of theZrO scale. A comparison between the adjusted R2 values of linearand parabolic regression allows the significance of the reduceddegree of freedom of the parabolic plot to be compared with its
improved fit. A linear fit is statistically favourable within thelimited region shown in Fig. 10. A linear gradient would suggestthat the scale is non-protective. An extension of the curve with
ARTICLE IN PRESS
Fig. 10. The initial oxidation kinetics of atomically cleaned zirconium in air.
Adjusted correlation coefficient R2¼ 0.972 for linear fit shown. Adjusted
R2¼ 0.965 for parabolic trend.
D. Hudson et al. / Ultramicroscopy 109 (2009) 667–671 671
new data is not easily obtained except by the use of the samespecimen or of a specimen with exactly the same orientation. Theoxidation rates of different textures would be expected to be non-uniform due to the anisotropy of HCP metals. So a comparisonbetween kinetics curves plotted from specimens of different grainorientations plotted over different time ranges would not beconsistent.
There are three significant routes for the ingress of hydrogeninto the material; one of these is introduction by moisture orliquid during the intentional oxidation of the specimen. Secondly,for samples that have undergone electrochemical polishing buthave not had surface impurities removed by atom probe analysisthere is the possibility that hydrogen was introduced by thefabrication process. Figs. 3 and 9 demonstrate such hydrogenpickup. This is not the case for ‘cleaned’ specimens, where thesurface has been removed and the material underneath has beenverified to consist of zirconium metal only. Finally, for both freshlyprepared and ‘cleaned’ specimens there is the possibility of low-level hydrogen pickup from the vacuum system. It is possible thatspecimens displaying hydrogen segregation to the interface arecompromised by hydrogen adsorption, where hydrogen adsorbedonto zirconium metal but not the ZrO layer. However, if this wereto be the case we may expect to see greater penetration ofhydrogen from the vacuum system into the metal. The g-ZrHhydrides in Figs. 2 and 6 are not typical of hydrogen adsorbedfrom the vacuum system because of their bulk scale and near-stoichiometric composition. As the surfaces of these specimenshave been previously evaporated it is suggested that theseshydrides are products of the corrosive environment to which theywere intentionally exposed after field evaporation.
The agreement between the composition of the ZrO layermeasured in the LEAP using laser pulsing and the 3DAP using
voltage pulsing is good, and confirms the results found using thelatter device.
At the transition between the oxide and the metal there canoften be a miniature fracture of the specimen, where the smoothoxide cap is removed to produce a clean but uneven end form ofthe bulk metal. This makes the analysis over this transition regionchallenging. The size of the region of analysis, a few tens orhundred of nanometres deep, also presents difficulties whenexamining the possibility of a layer of ZrO, which may be only1–2 nm thick under a layer of ceramic oxide, which can grow tomany microns in thickness.
5. Conclusions
�
A ZrO film of the order of a few nanometres grows onzirconium in air or aqueous solution after mild oxidation. � Using laser pulsing the 3DAP can analyse the ceramic ZrO2oxide produced by more severe oxidation.
� Hydrides and hydrogen segregation can be observed inzirconium using the atom probe.
� The atom probe technique can be used to study early oxidationreaction kinetics.
Acknowledgements
The authors would like to thank their collaborators from EDFEnergy, Westinghouse, and Open and Manchester Universities,which make up the MUZIC consortium. The authors would alsolike to acknowledge the contribution and improvement to thispaper by an anonymous reviewer. This research was funded by theEngineering and Physical Sciences Research Council (EPSRC) andUK MoD [DSTL] under Grant number EP/E036384/1.
References
[1] J.J. Carroll, A.J. Melmed, Surface Science 58 (1976) 601.[2] B. Wadman, H.-O. Andren, L.K.L. Falk, Colloque de Physique 50 (1989)
C8 303.[3] B. Wadman, H.-O. Andren, U. Rolander, Journal de Physique 11 (1988) C6 49.[4] B. Wadman, S. Andren, in: C.M. Euchen, A.M. Garde (Eds.), Zirconium in
the Nuclear Industry: Ninth Symposium, ASTM STP 1132, ASTM, USA, 1991,p. 461.
[5] O. Kubaschewski, B.E. Hopkins, Oxidation of Metals and Alloys, Butterworths,London, 1962.
[6] B. Cox, Journal of Nuclear Materials 336 (2005) 331.[7] R.T. Phelps, E.A. Gulbransen, J.W. Hickman, Industrial and Engineering
Chemistry 18 (1946) 391.[8] G. Sainfort, Memoires Scientifiques de la Revue de Metallurgie 56 (1959) 704.[9] P.A.J. Bagot, A. Cerezo, G.D.W. Smith, Surface Science 601 (2007) 2245.
[10] A. Cerezo, P.H. Clifton, M.J. Galtrey, C.J. Humphreys, T.F. Kelly, D.J. Larson,S. Lozano-Perez, E.A. Marquis, R.A. Oliver, G. Sha, K. Thompson,M. Zandbergen, R.L. Alvis, Materials Today 10 (12) (2007) 36.
[11] Goodfellow Cambridge Limited, Huntingdon PE29 6WR, England.[12] M.K. Miller, A. Cerezo, M.G. Hetherington, G.D.W. Smith, Atom Probe Field Ion
Microscopy, Clarendon Press, Oxford, 1996.[13] IVAS, version 3.0.6, Imago Scientific Instruments Corporation, 5500 Nobel
Drive, Madison, WI 53711, USA.