- Email: [email protected]
Monitoring the oxidation of nuclear fuel cladding using Raman spectroscopy
Journal of Nuclear Materials 445 (2014) 7–11
Contents lists available at ScienceDirect
Journal of Nuclear Materials journal homepage: www.elsevier.com/locate/jnucmat
Monitoring the oxidation of nuclear fuel cladding using Raman spectroscopy Hongyi Mi a, Solomon Mikael a, Todd Allen b, Kumar Sridharan b, Darryl Butt c, James P. Blanchard b,⇑, Zhenqiang Ma a,b,⇑ a b c
Department of Electrical and Computer Engineering, University of Wisconsin-Madison, 1415 Engineering Drive, Madison, WI 53706, USA Department of Nuclear Engineering and Engineering Physics, University of Wisconsin-Madison, 1500 Engineering Drive, Madison, WI 53706, USA Department of Material Science and Engineering, Boise State University, 1910 University Drive, Boise, ID 83725, USA
a r t i c l e
i n f o
Article history: Received 8 April 2013 Accepted 26 October 2013 Available online 1 November 2013
a b s t r a c t In order to observe Zircaloy-4 (Zr-4) cladding oxidation within a spent fuel canister, cladding oxidized in air at 500 °C was investigated by micro-Raman spectroscopy to measure the oxide layer thickness. Systematic Raman scans were performed to study the relationship between typical Raman spectra and various oxide layer thicknesses. The thicknesses of the oxide layers developed for various exposure times were measured by cross-sectional Scanning Electron Microscopy (SEM). The results of this work reveal that each oxide layer thickness has a corresponding typical Raman spectrum. Detailed analysis suggests that the Raman scattering peaks around wave numbers of 180 cm 1 and 630 cm 1 are the best choices for accurately determining the oxide layer thickness. After Gaussian–Lorentzian deconvolution, these two peaks can be quantitatively represented by four peaks. The intensities of the deconvoluted peaks increase consistently as the oxide layer becomes thicker and sufficiently strong signals are produced, allowing one to distinguish the bare and oxidized cladding samples, as well as samples with different oxide layer thicknesses. Hence, a process that converts sample oxide layer thickness to optical signals can be achieved. Ó 2013 Elsevier B.V. All rights reserved.
1. Introduction Monitoring of spent fuel cladding can assist with the assurance of the integrity of the fuel within the storage canister, ensure the safety of the fuel within the cask during transportation, and assess fuel as it comes out of a spent fuel pool. The oxidation of Zircaloy-4 (Zr-4) cladding has been studied at different temperatures and the oxidation process has been simulated in various experimental settings [1–4]. In this paper we propose a system that can provide real-time, in situ monitoring of this oxidation. Raman spectroscopy is able to reveal vibrational mechanisms inside materials in different phases. By combining this technique with microscopy, it is possible to measure the properties of very small material changes, including the oxide layer thickness. One advantage of this approach is that it can be non-destructive and non-contact, which assures the integrity of the material being evaluated and permits long distance detection [5]. For this application, a high luminosity laser source and high sensitivity charge-coupled ⇑ Corresponding authors. Address: Department of Nuclear Engineering and Engineering Physics, University of Wisconsin-Madison, 1500 Engineering Drive, Madison, WI 53706, USA. Tel.: +1 608 261 1095 (Z. Ma), +1 608 263 0391 (J.P. Blanchard). E-mail addresses: [email protected] (J.P. Blanchard), [email protected]. edu (Z. Ma). 0022-3115/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jnucmat.2013.10.052
device (CCD), as used in a typical micro-Raman spectroscopy device, will permit high resolution (even submicron scale) thickness determination. Raman spectroscopy has been widely employed to study the properties of zirconium oxide. The crystal structure/phase of zirconium oxide can be distinguished from a typical Raman spectrum. It has been shown that the vibration status of the amorphous, tetragonal and monoclinic polymorphs can be reflected from the peak intensity and profile [6–9]. Based on the Raman spectra, the lattice mechanism can be used to identify zirconium oxide from other oxide materials [10]. Meanwhile, some research has focused on the Raman spectra to detect Zircaloy cladding oxidation. The Raman spectra could provide detail to observe the formation of a textured oxide layer constituted in the Zircaloy cladding corrosion process [11]. Very recently, results of experiments suggest that Raman investigation of Zircaloy oxidation kinetic processes is also possible. Both pre- and post-breakaway oxide scales formed on Zircaloy can be reflected in corresponding Raman scattering peaks [12]. This previous research has resulted in the foundation for further Raman scattering investigation on Zr-4 cladding health condition and characteristics. The purpose of this work is to investigate the Zr-4 cladding oxide layer thickness by micro-Raman spectroscopy. The relationship between the Raman spectrum and the corresponding oxide
H. Mi et al. / Journal of Nuclear Materials 445 (2014) 7–11
thickness on Zr-4 cladding will be discussed and a method to convert zirconium oxide thickness into scattering optical signals will be analyzed.
2. Experimental 2.1. Sample oxidation The tested Zr-4 cladding samples were cut by saw from 1 mm thick Zr-4 cladding into 1 cm 1 cm samples. The Zr-4 samples were ground and polished by a Struers TegraForce-5 polish machine on one side using Allied High Tech SiC papers from 120 grit to 1200 grit to remove the native surface oxide layer and provide a smooth sample surface. The morphology of the oxide may be affected by the cold-worked nature of the polished surface. Future work will have to address the comparison of our oxide thickness sensor for cold-worked and stress-relieved samples. The samples were loaded in a quartz tube with a 5 cm diameter and were oxidized in a Thermo Scientific F79335-70 furnace chamber at 500 °C in flowing air. Because the oxide layer thickness increases with the oxidation time [13], samples were oxidized for 5 h, 10 h, 20 h, 50 h and 100 h respectively in order to produce different oxide layer thicknesses for measurement with the Raman technique. These oxidation times were chosen such that the thickness of the grown oxide layer can be clearly distinguished in further study. After oxidation, the polished side of each oxidized sample was ready for further Raman laser scattering investigation.
2.2. Raman spectrum investigation The Raman spectra of the oxidized samples were measured with a Thermal Scientific DXR micro-Raman spectrometer. The excitation laser source is a DXR 532 nm green laser and the output power was 10 mW. The spectra were acquired by a 50 lm aperture. The objective of the microscope was chosen as 10 and the laser beam produced a spot size diameter of 2.1 lm. Raman spectra are collected from the entire 2.1 lm diameter spot. The scans implemented a 900 lines/mm grating to obtain full range spectra and find the interested spectrum range requiring further study. Systematic scans were carried out for several spots, randomly selected at different locations. Then, the representative spectra were examined to identify the peaks that would provide the best potential for identifying the oxide layer thickness. These representative Raman spectra can effectively detect the sample properties of interest.
2.3. Scanning Electron Microscopy (SEM) examination After the Raman spectrum measurement, the oxidized Zr-4 samples were prepared for cross-section measurement by SEM in order to provide an independent measure of the oxide layer thickness. Because of the different properties of phenolic (used for mounting the samples), zirconium oxide, and Zr-4, these three materials can be easily distinguished by different white-black color due to the contrast. Zirconium oxide is a high dielectric constant material, so compared with Zr-4, the zirconium oxide layer has a darker gray color in SEM images and the boundary between the oxide layer and the Zr-4 was obvious. Following this procedure, the zirconium oxide region could be identified in the cross-section SEM images. In the cross-section SEM images of each oxidized sample, the scale measurement function in the SmartSEM software was carried out to measure the zirconium oxide layer thickness accurately at different positions of cross-section.
3. Results 3.1. Raman spectra Raman scans were performed at different spots on bare Zr-4, as well as samples oxidized for 5, 10, 20, 50, and 100 h. The characteristic Raman scattering peaks appear between wave number of 160 cm 1 and 800 cm 1. Fig. 1 shows the selected representative Raman spectra in this wave number range. In this figure, a baseline has been subtracted from each spectrum, allowing all the spectra to be put at the same level for selection of the best peaks for measuring the oxide layer thickness. As the Raman spectra reveal, the characteristic Raman scattering peaks become more obvious as the sample’s oxidation time increases. In the spectrum from the nonoxidized sample, there is only one peak around 480 cm 1 with a very weak intensity. The spectrum from the 5 h sample exhibits the 480 cm 1 peak, which is similar to the bare sample. Meanwhile, two weak peaks around 180 cm 1 and 630 cm 1 appear with a broad peak width. In the spectrum from the 10 h sample, the peaks at 180 cm 1, 480 cm 1 and 630 cm 1 become stronger and wider. In addition, two peaks at 340 cm 1 and 380 cm 1 appear. The spectrum from the 20 h sample is similar to that of the 10 h sample, but the intensity of each peak increases slightly. The peaks at 180 cm 1 and 630 cm 1 begin to split into two subpeaks at 175 cm 1 and 187 cm 1 for the 180 cm 1 peak and 613 cm 1 and 638 cm 1 for the 630 cm 1 peak. In the spectrum from the 50 h sample, all these peaks still exist at the same positions, but with much stronger intensities. New peaks at 220 cm 1, 540 cm 1 and 580 cm 1 also appear within this spectrum. The spectrum from the 100 h sample contains all the peaks in the 50 h sample’s spectrum, but each of them displays an increase in intensity. Furthermore, there are two new peaks at 280 cm 1 and 300 cm 1 with weak intensities. As the oxidation time increases, thicker oxide layers are grown on top of Zr-4 and in turn the Raman spectrum of this sample can show stronger Raman scattering signals. Comparing all of these Raman spectra in detail, some characteristic peaks are helpful in identifying bare and oxidized samples. As discussed previously, there are 12 characteristic peaks, either appearing or growing, in the spectra of the tested samples. In the spectra of the bare and 5 h samples, the differences are only observed for peaks at
1200
1000
non-oxidized 5 hour 10 hour 20 hour 50 hour 100 hour
800
Intensity (a.u.)
8
600
400
200
0
Raman shift wavenumber (cm-1) Fig. 1. The selected representative Raman spectra of the non-oxidized, 5 h, 10 h, 20 h, 50 h and 100 h oxidized Zr-4 cladding samples. As the oxidation time increases, there are 12 characteristic peaks appearing in the in the spectra.
9
H. Mi et al. / Journal of Nuclear Materials 445 (2014) 7–11
(a)
(b) Phenolic Phenolic
Zirconium oxide
Zr-4 Sample cross-section
500nm
(d)
(c) Phenolic
Phenolic
Zirconium oxide Zirconium oxide Zr-4 Zr-4
2µm
1µm
(e)
(f)
Phenolic
Phenolic
Zirconium oxide Zirconium oxide Zr-4 Zr-4
10µm
5µm
Fig. 2. (a) Optical image of Zr-4 sample after mounted by Phenolic. (b–f) SEM images of polished cross-sections of 5 h, 10 h, 20 h, 50 h and 100 h oxidized samples.
3.2. SEM images To investigate the thicknesses of the oxide layers grown for different times, the cross-section of each sample was checked by Scanning Electron Microscopy as previously described. Fig. 2a shows samples after mounting with phenolic. The SEM images of the cross-sections are shown in Fig. 2b–f. In these SEM images, phenolic, zirconium oxide, and Zr-4 are clearly distinguished by white–black color contrast due to their differing material
15
Zircomium oxide layer thickness (µm)
180 cm 1 and 630 cm 1, which do not exist in bare sample spectrum. These two peaks exist in all the spectra of the oxidized samples (from 5 to 100 h) and their intensities increase consistently as the oxidation time increases, although there is splitting in these two peaks for long exposure times. These split peaks at 175 cm 1 and 187 cm 1, 613 cm 1 and 638 cm 1 correspond to the existence of the zirconium oxide tetragonal and monoclinic polymorphs in the oxide layer [7,14]. When an oxide layer grows thicker, there will be more zirconium oxide in these polymorphs. Oxide layers grown in environments other than air will likely have a different morphology than those studied here. Future work is necessary to establish whether the approach used successfully in this study will apply to oxides grown under other conditions.
Mean thickness of oxide layer
10
5
0 0
20
40 60 Oxidation time (hour)
Fig. 3. The calculated statistical average thickness and deviation of all oxidized samples as a function of oxidation time.
10
H. Mi et al. / Journal of Nuclear Materials 445 (2014) 7–11
100
100
(a)
(b) 80
Intensity (a.u.)
Intensity (a.u.)
80
60
40
60
40
20
20
0
0
150
160
170
180
190
200
210
560
220
580
1000
620
640
660
680
700
720
1000
(c)
(d)
800
800
Intensity (a.u.)
Intensity (a.u.)
600
Raman shift wavenumber (cm-1)
Raman shift wavenumber (cm-1)
600
400
200
400
200
0 150
600
0 160
170
180
190
200
210
220
Raman shift wavenumber (cm-1)
560
580
600
620
640
660
680
700
720
Raman shift wavenumber (cm-1)
Fig. 4. The Gaussian–Lorentzian deconvolution processes of characteristic peaks around 180 cm 1 and 630 cm 1 from the selected representative Raman spectra of the 10 h (a and b) and 100 h (c and d) oxidized samples. The peaks at 180 cm 1 and 630 cm 1 are deconvoluted into peaks at 175 cm 1 and 187 cm 1, and 613 cm 1 and 638 cm 1 positions respectively.
properties. The SEM images were taken in more than 20 different regions of each sample, so that the effects of spatial variations in the oxide layer could be minimized. Based on the thickness of the top oxide layer, as measured by the scale function using the Smart SEM software, the statistical average thickness and standard deviation of the top oxide layer are shown in Fig. 3. The thickness of the oxide layer increases roughly linearly with oxidation time. For the 5 h sample, there is a very thin uniform oxide layer on top of Zr-4. As the thickness increases, the top oxide layer becomes non-uniform and the interface between zirconium oxide and Zr-4 becomes rougher. The mean thickness of the oxide layer for the 5 h sample is about 0.55 lm with a deviation of about 0.066 lm. But for the 100 h sample, the mean thickness of the oxide layer is about 10.13 lm with a deviation of about 3.08 lm. This suggests that during oxidation, the growth rate of the oxide layer varies across the sample and the oxide layer will have an uneven profile.
peaks from the selected representative spectra of the 10 h and 100 h samples (same data as Fig. 1) as examples. In all cases, spectra are taken from multiple spots on each specimen to minimize the effects of spatial variability in surface morphology. After deconvolution, the two sub-peaks can be replaced by individual peaks at the 175 cm 1 and 187 cm 1, and 613 cm 1 and 638 cm 1 positions respectively. The intensities of these four peaks from the representative Raman spectra of all oxidized samples are shown in Fig. 5 as a function of the average thickness of the oxide layer. The intensities of all of these four deconvoluted peaks increase consistently as the oxide layer becomes thicker. For the 10 h sample, with an average oxide layer thickness of 1.30 lm, the intensities of the deconvoluted peaks are 44.1, 26.6, 21.3 and 39.5. For the 100 h sample with an average oxide layer thickness of 10.13 lm, the intensities have increased greatly to 645.5, 405.5, 264.9 and 356.5. 4. Discussion
3.3. Data analysis As discussed in the previous section, there are two characteristic peaks around 180 cm 1 and 630 cm 1 in the spectra and those can be candidate signals for detecting the Zr-4 cladding oxide layer thickness. In order to analyze these two peaks quantitatively, Gaussian–Lorentzian mode deconvolutions were carried out. Fig. 4 shows the deconvolution processes of these two characteristic
The deconvoluted characteristic peaks have been related to the oxide layer thickness, as shown in Fig. 5. The difference in intensities of the peaks at 175 cm 1, 187 cm 1, 613 cm 1 and 638 cm 1 are clear enough to identify each sample with the corresponding oxide layer thickness. Taking the peak at 175 cm 1 as an example, the peak intensity increases from 0 for a bare sample to 3.5, 44.1, 63.2, 186.6 and 645.5 for 0.55 lm, 1.30 lm, 2.36 lm, 5.27 lm
H. Mi et al. / Journal of Nuclear Materials 445 (2014) 7–11
700 175 cm-1 peak
613 cm-1 peak
187 cm-1 peak
638 cm-1 peak
600
Intensity (a.u.)
500
400
300
200
100
0 0
2
4
6
8
10
Oxide layer thickness (µm) Fig. 5. The intensities of the deconvoluted peaks at 175 cm 1, 187 cm 1, 613 cm 1 and 638 cm 1 from the selected representative Raman spectra as a function of the average thickness of the oxide layer. All of these peaks are enhanced when the oxide layer thickens.
and 10.13 lm oxide layer thicknesses. Because there is no peak at these positions for the bare sample, it is easy to distinguish it from the oxidized samples, even for the 5 h sample with a 0.55 lm thick oxide layer. Among the oxidized samples, the average thickness difference between 5 h and 10 h samples is only about 0.75 lm, but the signal intensity of the 10 h sample is almost 10 times higher than that of the 5 h sample. Another three peaks show similar trends. This result reflects that during the oxidation process at 500 °C, as oxidation time increases, the components of the tetragonal and monoclinic polymorph zirconium oxide become more prominent, and consequently, they show much stronger characteristic peaks in the Raman spectra [14]. The deconvoluted characteristic peaks at wave numbers of 175 cm 1, 187 cm 1, 613 cm 1 and 638 cm 1 have been collected and studied for all tested Raman spectra from different positions
11
of each oxidized sample. In Fig. 6, the maximum intensity deviation percentages of each characteristic peak for all oxidized samples are plotted as a function of the average thickness of the oxide layer. Here, in order to maintain consistency, the intensities of the deconvoluted characteristic peaks from the selected representative Raman spectra are taken as reference. It is observed that the maximum intensity deviations for these characteristic peaks are in the range from about 5% to a little lower than 25% for all tested positions on the surfaces for all oxidized samples. This intensity variation comes from the difference in the thickness of the oxide layer. This result means that the uneven condition of the oxide layer thickness for the sample can be roughly reflected in intensity variation of the characteristic peaks. After analysis of these characteristic peaks, they can be confidently considered as sufficient scattering signals to detect the oxide layer thickness on Zr-4 cladding, and further employed to finish the task of converting the oxide layer thickness to detectable optical signals. 5. Conclusions In order to validate the proposed approach for monitoring spent fuel cladding oxidation, Zr-4 cladding samples were oxidized in air to obtain different oxide layer thicknesses. These oxidized cladding samples were systematically investigated by micro-Raman spectroscopy and Scanning Electron Microscopy in order to observe the characteristic peaks and compare to the measured oxide layer thicknesses. Further analysis reveals that there are four characteristic peaks in the Raman spectra able to identify the bare and oxidized cladding samples, as well as samples with different oxide layer thicknesses. The results show this Raman scattering detection method displays high resolution in the determination of the oxide layer thickness. Meanwhile, the intensity variation of these four characteristic peaks can roughly present the changes in thickness of the oxide layer on Zr-4 sample. Acknowledgment This work was supported by DOE FAST IRP project under Grant No. 00121602with CFDA #81.049.
Maximum peak intensity deviation percentage (%)
References [1] [2] [3] [4] [5]
30
25
175 cm-1 peak deviation persentage 187 cm-1 peak deviation persentage 613 cm-1 peak deviation persentage 638 cm-1 peak deviation persentage
[6] [7]
20
[8] [9] [10]
15
[11]
10
[12] [13]
5 [14]
0 0
0.55
1.30
2.36
5.27
10.13
Oxide layer thickness (µm) Fig. 6. The maximum intensity deviation percentages of the deconvoluted characteristic peaks as a function of the average thickness of the oxide layer.
M. Suzuki, S. Kawasaki, J. Nucl. Mater. 140 (1986) 32–43. D.G. Boase, T.T. Vandergraaf, Nucl. Technol. 32 (1977) 60. J.S. Bryner, J. Nucl. Mater. 82 (1979) 84–101. B. Cox, J. Nucl. Mater. 336 (2005) 331–386. L. Lyon, C. Keating, A. Fox, B. Baker, L. He, S. Nicewarner, S. Mulvancy, M. Natan, Anal. Chem. 70 (1998) 341R–361R. V. keranidas, W. White, J. Am. Ceram. Soc. 57 (1) (1974) 22. R. Srinivasan, M. Harris, S. Simpson, R. DeAngelis, B. Davis, J. Mater. Res. 3 (4) (1988) 787. P. Barberis, T. Merle-Mejean, P. Quintard, J. Nucl. Mater. 246 (1997) 232–243. P. Barberis, G. Corolleur-Thomas, R. Guinebretiere, T. Merle-Mejean, A. Mirgorodsky, P. Quintard, J. Nucl. Mater. 288 (2001) 241–247. P. Quintard, P. Barberis, A.P. Mirgorodsky, T. Merle-Mejean, J. Am. Ceram. Soc. 85 (7) (2002) 1745. J. Moya, M. Diaz, J. Bartolome, E. Roman, J. Sacedon, J. Izquierdo, Acta. Mater. 48 (2000) 4749–4754. I. Idarraga, M. Mermoux, C. Duriez, A. Crisci, J. Mardon, J. Nucl. Mater. 421 (2012) 160–171. B. Cheng, P. Gilmore, H. Klepfer, Zirconium in the Nuclear Industry, in: 11th International Symposium, ASTM STP, vol. 1295, Garmisch-Partenkirchen, Germany, 1996. pp. 137. C. Phillippi, K. Mazdiyasni, J. Am. Ceram. Soc. 54 (5) (1971) 254.
Contents lists available at ScienceDirect
Journal of Nuclear Materials journal homepage: www.elsevier.com/locate/jnucmat
Monitoring the oxidation of nuclear fuel cladding using Raman spectroscopy Hongyi Mi a, Solomon Mikael a, Todd Allen b, Kumar Sridharan b, Darryl Butt c, James P. Blanchard b,⇑, Zhenqiang Ma a,b,⇑ a b c
Department of Electrical and Computer Engineering, University of Wisconsin-Madison, 1415 Engineering Drive, Madison, WI 53706, USA Department of Nuclear Engineering and Engineering Physics, University of Wisconsin-Madison, 1500 Engineering Drive, Madison, WI 53706, USA Department of Material Science and Engineering, Boise State University, 1910 University Drive, Boise, ID 83725, USA
a r t i c l e
i n f o
Article history: Received 8 April 2013 Accepted 26 October 2013 Available online 1 November 2013
a b s t r a c t In order to observe Zircaloy-4 (Zr-4) cladding oxidation within a spent fuel canister, cladding oxidized in air at 500 °C was investigated by micro-Raman spectroscopy to measure the oxide layer thickness. Systematic Raman scans were performed to study the relationship between typical Raman spectra and various oxide layer thicknesses. The thicknesses of the oxide layers developed for various exposure times were measured by cross-sectional Scanning Electron Microscopy (SEM). The results of this work reveal that each oxide layer thickness has a corresponding typical Raman spectrum. Detailed analysis suggests that the Raman scattering peaks around wave numbers of 180 cm 1 and 630 cm 1 are the best choices for accurately determining the oxide layer thickness. After Gaussian–Lorentzian deconvolution, these two peaks can be quantitatively represented by four peaks. The intensities of the deconvoluted peaks increase consistently as the oxide layer becomes thicker and sufficiently strong signals are produced, allowing one to distinguish the bare and oxidized cladding samples, as well as samples with different oxide layer thicknesses. Hence, a process that converts sample oxide layer thickness to optical signals can be achieved. Ó 2013 Elsevier B.V. All rights reserved.
1. Introduction Monitoring of spent fuel cladding can assist with the assurance of the integrity of the fuel within the storage canister, ensure the safety of the fuel within the cask during transportation, and assess fuel as it comes out of a spent fuel pool. The oxidation of Zircaloy-4 (Zr-4) cladding has been studied at different temperatures and the oxidation process has been simulated in various experimental settings [1–4]. In this paper we propose a system that can provide real-time, in situ monitoring of this oxidation. Raman spectroscopy is able to reveal vibrational mechanisms inside materials in different phases. By combining this technique with microscopy, it is possible to measure the properties of very small material changes, including the oxide layer thickness. One advantage of this approach is that it can be non-destructive and non-contact, which assures the integrity of the material being evaluated and permits long distance detection [5]. For this application, a high luminosity laser source and high sensitivity charge-coupled ⇑ Corresponding authors. Address: Department of Nuclear Engineering and Engineering Physics, University of Wisconsin-Madison, 1500 Engineering Drive, Madison, WI 53706, USA. Tel.: +1 608 261 1095 (Z. Ma), +1 608 263 0391 (J.P. Blanchard). E-mail addresses: [email protected] (J.P. Blanchard), [email protected]. edu (Z. Ma). 0022-3115/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jnucmat.2013.10.052
device (CCD), as used in a typical micro-Raman spectroscopy device, will permit high resolution (even submicron scale) thickness determination. Raman spectroscopy has been widely employed to study the properties of zirconium oxide. The crystal structure/phase of zirconium oxide can be distinguished from a typical Raman spectrum. It has been shown that the vibration status of the amorphous, tetragonal and monoclinic polymorphs can be reflected from the peak intensity and profile [6–9]. Based on the Raman spectra, the lattice mechanism can be used to identify zirconium oxide from other oxide materials [10]. Meanwhile, some research has focused on the Raman spectra to detect Zircaloy cladding oxidation. The Raman spectra could provide detail to observe the formation of a textured oxide layer constituted in the Zircaloy cladding corrosion process [11]. Very recently, results of experiments suggest that Raman investigation of Zircaloy oxidation kinetic processes is also possible. Both pre- and post-breakaway oxide scales formed on Zircaloy can be reflected in corresponding Raman scattering peaks [12]. This previous research has resulted in the foundation for further Raman scattering investigation on Zr-4 cladding health condition and characteristics. The purpose of this work is to investigate the Zr-4 cladding oxide layer thickness by micro-Raman spectroscopy. The relationship between the Raman spectrum and the corresponding oxide
H. Mi et al. / Journal of Nuclear Materials 445 (2014) 7–11
thickness on Zr-4 cladding will be discussed and a method to convert zirconium oxide thickness into scattering optical signals will be analyzed.
2. Experimental 2.1. Sample oxidation The tested Zr-4 cladding samples were cut by saw from 1 mm thick Zr-4 cladding into 1 cm 1 cm samples. The Zr-4 samples were ground and polished by a Struers TegraForce-5 polish machine on one side using Allied High Tech SiC papers from 120 grit to 1200 grit to remove the native surface oxide layer and provide a smooth sample surface. The morphology of the oxide may be affected by the cold-worked nature of the polished surface. Future work will have to address the comparison of our oxide thickness sensor for cold-worked and stress-relieved samples. The samples were loaded in a quartz tube with a 5 cm diameter and were oxidized in a Thermo Scientific F79335-70 furnace chamber at 500 °C in flowing air. Because the oxide layer thickness increases with the oxidation time [13], samples were oxidized for 5 h, 10 h, 20 h, 50 h and 100 h respectively in order to produce different oxide layer thicknesses for measurement with the Raman technique. These oxidation times were chosen such that the thickness of the grown oxide layer can be clearly distinguished in further study. After oxidation, the polished side of each oxidized sample was ready for further Raman laser scattering investigation.
2.2. Raman spectrum investigation The Raman spectra of the oxidized samples were measured with a Thermal Scientific DXR micro-Raman spectrometer. The excitation laser source is a DXR 532 nm green laser and the output power was 10 mW. The spectra were acquired by a 50 lm aperture. The objective of the microscope was chosen as 10 and the laser beam produced a spot size diameter of 2.1 lm. Raman spectra are collected from the entire 2.1 lm diameter spot. The scans implemented a 900 lines/mm grating to obtain full range spectra and find the interested spectrum range requiring further study. Systematic scans were carried out for several spots, randomly selected at different locations. Then, the representative spectra were examined to identify the peaks that would provide the best potential for identifying the oxide layer thickness. These representative Raman spectra can effectively detect the sample properties of interest.
2.3. Scanning Electron Microscopy (SEM) examination After the Raman spectrum measurement, the oxidized Zr-4 samples were prepared for cross-section measurement by SEM in order to provide an independent measure of the oxide layer thickness. Because of the different properties of phenolic (used for mounting the samples), zirconium oxide, and Zr-4, these three materials can be easily distinguished by different white-black color due to the contrast. Zirconium oxide is a high dielectric constant material, so compared with Zr-4, the zirconium oxide layer has a darker gray color in SEM images and the boundary between the oxide layer and the Zr-4 was obvious. Following this procedure, the zirconium oxide region could be identified in the cross-section SEM images. In the cross-section SEM images of each oxidized sample, the scale measurement function in the SmartSEM software was carried out to measure the zirconium oxide layer thickness accurately at different positions of cross-section.
3. Results 3.1. Raman spectra Raman scans were performed at different spots on bare Zr-4, as well as samples oxidized for 5, 10, 20, 50, and 100 h. The characteristic Raman scattering peaks appear between wave number of 160 cm 1 and 800 cm 1. Fig. 1 shows the selected representative Raman spectra in this wave number range. In this figure, a baseline has been subtracted from each spectrum, allowing all the spectra to be put at the same level for selection of the best peaks for measuring the oxide layer thickness. As the Raman spectra reveal, the characteristic Raman scattering peaks become more obvious as the sample’s oxidation time increases. In the spectrum from the nonoxidized sample, there is only one peak around 480 cm 1 with a very weak intensity. The spectrum from the 5 h sample exhibits the 480 cm 1 peak, which is similar to the bare sample. Meanwhile, two weak peaks around 180 cm 1 and 630 cm 1 appear with a broad peak width. In the spectrum from the 10 h sample, the peaks at 180 cm 1, 480 cm 1 and 630 cm 1 become stronger and wider. In addition, two peaks at 340 cm 1 and 380 cm 1 appear. The spectrum from the 20 h sample is similar to that of the 10 h sample, but the intensity of each peak increases slightly. The peaks at 180 cm 1 and 630 cm 1 begin to split into two subpeaks at 175 cm 1 and 187 cm 1 for the 180 cm 1 peak and 613 cm 1 and 638 cm 1 for the 630 cm 1 peak. In the spectrum from the 50 h sample, all these peaks still exist at the same positions, but with much stronger intensities. New peaks at 220 cm 1, 540 cm 1 and 580 cm 1 also appear within this spectrum. The spectrum from the 100 h sample contains all the peaks in the 50 h sample’s spectrum, but each of them displays an increase in intensity. Furthermore, there are two new peaks at 280 cm 1 and 300 cm 1 with weak intensities. As the oxidation time increases, thicker oxide layers are grown on top of Zr-4 and in turn the Raman spectrum of this sample can show stronger Raman scattering signals. Comparing all of these Raman spectra in detail, some characteristic peaks are helpful in identifying bare and oxidized samples. As discussed previously, there are 12 characteristic peaks, either appearing or growing, in the spectra of the tested samples. In the spectra of the bare and 5 h samples, the differences are only observed for peaks at
1200
1000
non-oxidized 5 hour 10 hour 20 hour 50 hour 100 hour
800
Intensity (a.u.)
8
600
400
200
0
Raman shift wavenumber (cm-1) Fig. 1. The selected representative Raman spectra of the non-oxidized, 5 h, 10 h, 20 h, 50 h and 100 h oxidized Zr-4 cladding samples. As the oxidation time increases, there are 12 characteristic peaks appearing in the in the spectra.
9
H. Mi et al. / Journal of Nuclear Materials 445 (2014) 7–11
(a)
(b) Phenolic Phenolic
Zirconium oxide
Zr-4 Sample cross-section
500nm
(d)
(c) Phenolic
Phenolic
Zirconium oxide Zirconium oxide Zr-4 Zr-4
2µm
1µm
(e)
(f)
Phenolic
Phenolic
Zirconium oxide Zirconium oxide Zr-4 Zr-4
10µm
5µm
Fig. 2. (a) Optical image of Zr-4 sample after mounted by Phenolic. (b–f) SEM images of polished cross-sections of 5 h, 10 h, 20 h, 50 h and 100 h oxidized samples.
3.2. SEM images To investigate the thicknesses of the oxide layers grown for different times, the cross-section of each sample was checked by Scanning Electron Microscopy as previously described. Fig. 2a shows samples after mounting with phenolic. The SEM images of the cross-sections are shown in Fig. 2b–f. In these SEM images, phenolic, zirconium oxide, and Zr-4 are clearly distinguished by white–black color contrast due to their differing material
15
Zircomium oxide layer thickness (µm)
180 cm 1 and 630 cm 1, which do not exist in bare sample spectrum. These two peaks exist in all the spectra of the oxidized samples (from 5 to 100 h) and their intensities increase consistently as the oxidation time increases, although there is splitting in these two peaks for long exposure times. These split peaks at 175 cm 1 and 187 cm 1, 613 cm 1 and 638 cm 1 correspond to the existence of the zirconium oxide tetragonal and monoclinic polymorphs in the oxide layer [7,14]. When an oxide layer grows thicker, there will be more zirconium oxide in these polymorphs. Oxide layers grown in environments other than air will likely have a different morphology than those studied here. Future work is necessary to establish whether the approach used successfully in this study will apply to oxides grown under other conditions.
Mean thickness of oxide layer
10
5
0 0
20
40 60 Oxidation time (hour)
Fig. 3. The calculated statistical average thickness and deviation of all oxidized samples as a function of oxidation time.
10
H. Mi et al. / Journal of Nuclear Materials 445 (2014) 7–11
100
100
(a)
(b) 80
Intensity (a.u.)
Intensity (a.u.)
80
60
40
60
40
20
20
0
0
150
160
170
180
190
200
210
560
220
580
1000
620
640
660
680
700
720
1000
(c)
(d)
800
800
Intensity (a.u.)
Intensity (a.u.)
600
Raman shift wavenumber (cm-1)
Raman shift wavenumber (cm-1)
600
400
200
400
200
0 150
600
0 160
170
180
190
200
210
220
Raman shift wavenumber (cm-1)
560
580
600
620
640
660
680
700
720
Raman shift wavenumber (cm-1)
Fig. 4. The Gaussian–Lorentzian deconvolution processes of characteristic peaks around 180 cm 1 and 630 cm 1 from the selected representative Raman spectra of the 10 h (a and b) and 100 h (c and d) oxidized samples. The peaks at 180 cm 1 and 630 cm 1 are deconvoluted into peaks at 175 cm 1 and 187 cm 1, and 613 cm 1 and 638 cm 1 positions respectively.
properties. The SEM images were taken in more than 20 different regions of each sample, so that the effects of spatial variations in the oxide layer could be minimized. Based on the thickness of the top oxide layer, as measured by the scale function using the Smart SEM software, the statistical average thickness and standard deviation of the top oxide layer are shown in Fig. 3. The thickness of the oxide layer increases roughly linearly with oxidation time. For the 5 h sample, there is a very thin uniform oxide layer on top of Zr-4. As the thickness increases, the top oxide layer becomes non-uniform and the interface between zirconium oxide and Zr-4 becomes rougher. The mean thickness of the oxide layer for the 5 h sample is about 0.55 lm with a deviation of about 0.066 lm. But for the 100 h sample, the mean thickness of the oxide layer is about 10.13 lm with a deviation of about 3.08 lm. This suggests that during oxidation, the growth rate of the oxide layer varies across the sample and the oxide layer will have an uneven profile.
peaks from the selected representative spectra of the 10 h and 100 h samples (same data as Fig. 1) as examples. In all cases, spectra are taken from multiple spots on each specimen to minimize the effects of spatial variability in surface morphology. After deconvolution, the two sub-peaks can be replaced by individual peaks at the 175 cm 1 and 187 cm 1, and 613 cm 1 and 638 cm 1 positions respectively. The intensities of these four peaks from the representative Raman spectra of all oxidized samples are shown in Fig. 5 as a function of the average thickness of the oxide layer. The intensities of all of these four deconvoluted peaks increase consistently as the oxide layer becomes thicker. For the 10 h sample, with an average oxide layer thickness of 1.30 lm, the intensities of the deconvoluted peaks are 44.1, 26.6, 21.3 and 39.5. For the 100 h sample with an average oxide layer thickness of 10.13 lm, the intensities have increased greatly to 645.5, 405.5, 264.9 and 356.5. 4. Discussion
3.3. Data analysis As discussed in the previous section, there are two characteristic peaks around 180 cm 1 and 630 cm 1 in the spectra and those can be candidate signals for detecting the Zr-4 cladding oxide layer thickness. In order to analyze these two peaks quantitatively, Gaussian–Lorentzian mode deconvolutions were carried out. Fig. 4 shows the deconvolution processes of these two characteristic
The deconvoluted characteristic peaks have been related to the oxide layer thickness, as shown in Fig. 5. The difference in intensities of the peaks at 175 cm 1, 187 cm 1, 613 cm 1 and 638 cm 1 are clear enough to identify each sample with the corresponding oxide layer thickness. Taking the peak at 175 cm 1 as an example, the peak intensity increases from 0 for a bare sample to 3.5, 44.1, 63.2, 186.6 and 645.5 for 0.55 lm, 1.30 lm, 2.36 lm, 5.27 lm
H. Mi et al. / Journal of Nuclear Materials 445 (2014) 7–11
700 175 cm-1 peak
613 cm-1 peak
187 cm-1 peak
638 cm-1 peak
600
Intensity (a.u.)
500
400
300
200
100
0 0
2
4
6
8
10
Oxide layer thickness (µm) Fig. 5. The intensities of the deconvoluted peaks at 175 cm 1, 187 cm 1, 613 cm 1 and 638 cm 1 from the selected representative Raman spectra as a function of the average thickness of the oxide layer. All of these peaks are enhanced when the oxide layer thickens.
and 10.13 lm oxide layer thicknesses. Because there is no peak at these positions for the bare sample, it is easy to distinguish it from the oxidized samples, even for the 5 h sample with a 0.55 lm thick oxide layer. Among the oxidized samples, the average thickness difference between 5 h and 10 h samples is only about 0.75 lm, but the signal intensity of the 10 h sample is almost 10 times higher than that of the 5 h sample. Another three peaks show similar trends. This result reflects that during the oxidation process at 500 °C, as oxidation time increases, the components of the tetragonal and monoclinic polymorph zirconium oxide become more prominent, and consequently, they show much stronger characteristic peaks in the Raman spectra [14]. The deconvoluted characteristic peaks at wave numbers of 175 cm 1, 187 cm 1, 613 cm 1 and 638 cm 1 have been collected and studied for all tested Raman spectra from different positions
11
of each oxidized sample. In Fig. 6, the maximum intensity deviation percentages of each characteristic peak for all oxidized samples are plotted as a function of the average thickness of the oxide layer. Here, in order to maintain consistency, the intensities of the deconvoluted characteristic peaks from the selected representative Raman spectra are taken as reference. It is observed that the maximum intensity deviations for these characteristic peaks are in the range from about 5% to a little lower than 25% for all tested positions on the surfaces for all oxidized samples. This intensity variation comes from the difference in the thickness of the oxide layer. This result means that the uneven condition of the oxide layer thickness for the sample can be roughly reflected in intensity variation of the characteristic peaks. After analysis of these characteristic peaks, they can be confidently considered as sufficient scattering signals to detect the oxide layer thickness on Zr-4 cladding, and further employed to finish the task of converting the oxide layer thickness to detectable optical signals. 5. Conclusions In order to validate the proposed approach for monitoring spent fuel cladding oxidation, Zr-4 cladding samples were oxidized in air to obtain different oxide layer thicknesses. These oxidized cladding samples were systematically investigated by micro-Raman spectroscopy and Scanning Electron Microscopy in order to observe the characteristic peaks and compare to the measured oxide layer thicknesses. Further analysis reveals that there are four characteristic peaks in the Raman spectra able to identify the bare and oxidized cladding samples, as well as samples with different oxide layer thicknesses. The results show this Raman scattering detection method displays high resolution in the determination of the oxide layer thickness. Meanwhile, the intensity variation of these four characteristic peaks can roughly present the changes in thickness of the oxide layer on Zr-4 sample. Acknowledgment This work was supported by DOE FAST IRP project under Grant No. 00121602with CFDA #81.049.
Maximum peak intensity deviation percentage (%)
References [1] [2] [3] [4] [5]
30
25
175 cm-1 peak deviation persentage 187 cm-1 peak deviation persentage 613 cm-1 peak deviation persentage 638 cm-1 peak deviation persentage
[6] [7]
20
[8] [9] [10]
15
[11]
10
[12] [13]
5 [14]
0 0
0.55
1.30
2.36
5.27
10.13
Oxide layer thickness (µm) Fig. 6. The maximum intensity deviation percentages of the deconvoluted characteristic peaks as a function of the average thickness of the oxide layer.
M. Suzuki, S. Kawasaki, J. Nucl. Mater. 140 (1986) 32–43. D.G. Boase, T.T. Vandergraaf, Nucl. Technol. 32 (1977) 60. J.S. Bryner, J. Nucl. Mater. 82 (1979) 84–101. B. Cox, J. Nucl. Mater. 336 (2005) 331–386. L. Lyon, C. Keating, A. Fox, B. Baker, L. He, S. Nicewarner, S. Mulvancy, M. Natan, Anal. Chem. 70 (1998) 341R–361R. V. keranidas, W. White, J. Am. Ceram. Soc. 57 (1) (1974) 22. R. Srinivasan, M. Harris, S. Simpson, R. DeAngelis, B. Davis, J. Mater. Res. 3 (4) (1988) 787. P. Barberis, T. Merle-Mejean, P. Quintard, J. Nucl. Mater. 246 (1997) 232–243. P. Barberis, G. Corolleur-Thomas, R. Guinebretiere, T. Merle-Mejean, A. Mirgorodsky, P. Quintard, J. Nucl. Mater. 288 (2001) 241–247. P. Quintard, P. Barberis, A.P. Mirgorodsky, T. Merle-Mejean, J. Am. Ceram. Soc. 85 (7) (2002) 1745. J. Moya, M. Diaz, J. Bartolome, E. Roman, J. Sacedon, J. Izquierdo, Acta. Mater. 48 (2000) 4749–4754. I. Idarraga, M. Mermoux, C. Duriez, A. Crisci, J. Mardon, J. Nucl. Mater. 421 (2012) 160–171. B. Cheng, P. Gilmore, H. Klepfer, Zirconium in the Nuclear Industry, in: 11th International Symposium, ASTM STP, vol. 1295, Garmisch-Partenkirchen, Germany, 1996. pp. 137. C. Phillippi, K. Mazdiyasni, J. Am. Ceram. Soc. 54 (5) (1971) 254.