Defect imaging of structural objects using positron annihilation spectroscopy

Nuclear Instruments and Methods in Physics Research B 241 (2005) 262–266 www.elsevier.com/locate/nimb

Defect imaging of structural objects using positron annihilation spectroscopy A.W. Hunt a,b,*, R. Spaulding a, J. Urban-Klaehn a, J.F. Harmon a,b, D.P. Wells a,b a

Idaho Accelerator Center, Idaho State University, Pocatello, ID 83209-8263, USA Department of Physics, Idaho State University, Pocatello, ID 83209-8106, USA

b

Available online 26 August 2005

Abstract Two techniques have recently been developed to quickly and easily apply positron annihilation spectroscopies to large structural components found in civil engineering, aviation, etc. In this paper, the authors discuss how to extend imaging capabilities to these new techniques, which will enable defect imaging similar to that obtained with positron micro-beams but at much larger sample sizes. Preliminary two-dimensional defect imaging results are presented from a highly damaged 30.5 · 30.5 cm copper plate.  2005 Elsevier B.V. All rights reserved. PACS: 78.70.Bj; 61.72.Ji; 61.72.Hh Keywords: Positrons; Positron annihilation; Doppler broadening spectroscopy

1. Introduction The discovery by DeBenedetti et al. that the c-rays emitted from positron–electron annihilations in a solid are sensitive to the electronsÕ momenta ushered in the use of positrons as a probe of condensed matter systems [1]. The utility * Corresponding author. Address: Idaho Accelerator Center, Idaho State University, Pocatello, ID 83209-8263, USA. Tel.: +1 208 282 5966; fax: +1 208 282 5878. E-mail address: [email protected] (A.W. Hunt).

of this probe was further enhanced by the realization that positrons have a propensity to trap in sub-nanoscale defects (mono-vacancies, di-vacancies, dislocations, etc.) [2–4]. Hence, methods involving positron annihilation spectroscopy have become well-established for detecting and studying sub-nanoscale defects at sizes and concentrations too small for other methods [5–11]. These subnanoscale defects are often important to the formation or prevention of macroscopic damage and/or material failure. Consequently, positron annihilation spectroscopies could be an excellent

0168-583X/$ - see front matter  2005 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2005.07.186

A.W. Hunt et al. / Nucl. Instr. and Meth. in Phys. Res. B 241 (2005) 262–266

nondestructive testing technique for detecting damage in structural materials, well before physical damage signs are present. Most contemporary materials studies using positrons today, implant the positrons into the material under study from mono-energetic positron beams. Research conducted with these beams have been extremely successful because they permit the creation of defect depth profiles and lateral defect images when scanning positron micro-beams are employed [6,10,12–16]. However, only the top tens of micrometers can be probed and the overall sample size has been limited to a few cm3 because of the vacuum system in which the material under study must be placed [6,17]. This makes positron annihilation defect studies in large structural components difficult and its use has not been extensively adopted for nondestructive testing. Two techniques, both of which use electron accelerators to create bremsstrahlung photon beams, have been developed that allow positron annihilation spectroscopies to be easily and quickly applied to large samples. In the first technique, which has been successfully commercialized, the material under study is activated by a high-energy bremsstrahlung beam (endpoint energy of 20 MeV), causing AX(c, n)XA 1 reactions in its constituent nuclei [18]. Since the resulting radioisotopes are neutron deficient, they have a tendency to be positron emitters. Then standard positron annihilation techniques such as Doppler broadening spectroscopy or positron lifetime spectroscopy can be utilized. In the second technique, positron–electron pairs are directly created in the sample by a bremsstrahlung beam below the neutron emission threshold (10 MeV) but above the pair creation threshold (1.022 MeV) thereby avoiding material activation [19–21]. During irradiation a well shielded HPGe detector records the annihilation c-rays for standard Doppler broadening analysis. Assuming at least two sides of the sample are accessible, these methods can interrogate specimens that are 40 g/cm2 thick due to highly penetrating nature of the bremsstrahlung and annihilation photons. While neither of the photon induced positron annihilation methods have been used to create

263

lateral defect images, both have this capability when combined with a scanning system. In this paper the authors report on preliminary defect imaging results obtained by photo-activating a highly damaged 30.5 · 30.5 cm copper plate. In addition, three-dimensional defect maps can be created by using computed tomography techniques. These approaches extend the microscopic defect imaging capabilities of positron annihilation spectroscopies to large scale structural components.

2. Defect imaging by photo-activation A 6.4 mm thick 30.5 · 30.5 cm OFHC copper plate was bent around a 3.2 mm radius to a 90 angle. The plate was then bent back to its original flat shape, creating a linear highly damaged region along the centerline of the plate. To ensure uniform irradiation, the plate was then placed 2 m from a 2.2 mm tungsten converter at the end of an electron LINAC. A 28 MeV, 16 lA electron beam impinged on the converter creating the bremsstrahlung photons that were incident on the copper plate. The dominate nuclear reactions were 65Cu(c, n)64Cu and 63Cu(c, n)62Cu. Both resulting radioisotopes are positron emitters with half-lives of 12.7 h and 9.74 min respectively. Neither isotope emits any intense c-rays and consequently there was no interference with the positron annihilation signal. The plate was irradiated for a total of 2.5 h, after which it had a dose rate of 600 mR/h on contact. In initial experiments the short half-life of 62 Cu was found to be problematic because it caused the count rate in the HPGe detector to change rapidly relative to the scan speed, which in turn affects the detectorÕs energy resolution. Thus, the plate was allowed to cool for 2 h, permitting the 62Cu to decay through 12.3 half-lives at which point the dose rate on contact was 20 mR/h. The plate was then mounted on a two axis translation stage in front of a collimated 28% relative efficiency HPGe detector. The translation stage had an accuracy and reproducibility better than 0.5 mm. The HPGe detector had a 9.5 mm diameter lead collimator, which was 5.1 cm thick, and was shielded by 10.2 cm of lead

A.W. Hunt et al. / Nucl. Instr. and Meth. in Phys. Res. B 241 (2005) 262–266

0

10

133

10-1

Ba

1346 keV 64 Cu 1481 keV Unknown

101

1185 keV 61 Cu

Yield (keV-1s-1)

102

X-rays

103

511 keV 662 keV 137Cs

on the remaining sides, allowing spectra from 3.1 cm2 regions to be acquired. Two preamplifier outputs from the HPGe detector were transmitted to separate channels in the data acquisition system. The first channel recorded c-rays from 40 to 700 keV and was used for Doppler broadening analysis. The second channel recorded c-rays from 10 to 2300 keV and was used to monitor all the decay c-rays. Two calibration sources, 133Ba and 137Cs, were placed near the detector to give a count rate of 100 s 1 in the 356 and 661 keV lines respectively. The data was acquired in an event-by-event mode so that any gain drifts in the detector could be monitored and corrected using the c-ray lines from the check sources. Fig. 1 shows a full c-ray spectrum acquired for 40 min, 4 h after irradiation. The count rate in the 511 keV annihilation peak was 50.7 s 1. The radioisotopes 64Cu and 61Cu were identified by their 1345 and 1185 keV c-ray lines respectively. There was no evidence of 62Cu, which was identified in spectra taken 2 h after irradiation by its 1173 keV c-ray. The 61Cu was created during the irradiation by (c, 2n) reactions. Four spectra with high counting statistics were first collected at different positions running along

1.04

10-2 10-3 0

500

1000 1500 Energy (keV)

a line perpendicular to the linear damaged region of the plate. Each spectrum had over 105 counts in the 511 keV annihilation line on which standard Doppler broadening analysis was performed. Near the damaged region spectra were acquired 1 cm apart followed by a spectra taken 4 cm away, presumably in an undamaged area. The relative S-parameter is shown in Fig. 2. The S-parameter in the damaged region is 3 standard deviations above the undamaged area due to the high concentration of defects. To create a lateral defect image, the plate was divided into a 1 cm · 1 cm grid and at each point a c-ray spectrum was collected. A total of 33 spectra were acquired, covering 11 cm perpendicular and 3 cm parallel to the linear damaged region. Due to the 12.7 h half-life of 64Cu, each spectra was only acquired for 30 min, leading to an average of 4 · 104 counts in the 511 keV annihilation line. The entire scan took over 16 h to complete. Upon performing standard Doppler broadening analysis, large statistical errors and hence large scatter in the S-parameter were found. Any images created with this S-parameter data was indiscernible. In part, this was caused by the S-parameter relying on the number of counts in the central region of the 511 keV annihilation line, which significantly increases the relative statistical error. For low statistics experiments, it was found that the FWHM was a better measure of the Doppler broadening because it exploits all the available

2000

Fig. 1. Full c-ray spectrum acquired from the Cu plate for 40 min, 4 h after the end of irradiation by the 28 MeV bremsstrahlung beam. The radioisotopes 64Cu, 61Cu, 137Cs and 133Ba are identified with the latter two from the check sources. All unlabeled peaks are from either 61Cu at energies above 400 keV. The tight group of peaks labeled X-rays is comprised of Pb X-rays and the 81 keV c-ray from 133Ba.

Relative S-Parameter

264

1.03 1.02 1.01 1.00 0.99 -8

-6

-4

-2

0

2

4

6

Position (cm) Fig. 2. The relative S-parameter measured along a line running perpendicular to the highly damaged region of the Cu plate.

A.W. Hunt et al. / Nucl. Instr. and Meth. in Phys. Res. B 241 (2005) 262–266 3.0 2.5

FWHM 37.29

2.0

Y (cm)

38.03

1.5

38.76

1.0

39.50

0.5 0.0 -5 -4 -3 -2 -1 0

1

2

3

4

5

X (cm) Fig. 3. Two-dimensional color scale image of the 511 keV FWHM from the Cu plate. A total of 33 c-ray spectra were collected in 1 cm steps. The color scale goes from red representing the narrowest peak to blue representing the broadest peak. The FWHM is measured in ADC channels with 84.7 eV per channel.

data. Fig. 3 shows a two-dimensional color scale image of the FWHM of the 511 keV annihilation line. The spatial resolution was 1 cm. In the highly damaged region of the plate where the positrons predominately trap in defects, the measured FWHM narrows over 3.5 standard deviations. The damaged region appears to extend out as far as 2 cm but with the poor statistics and limited resolutions it is difficult to draw any definitive conclusions.

265

(a common structural material) from photo-activation is 53Fe. Unfortunately, 53Fe only has a half-life of 8.5 min, making it difficult to image a large region. Finally, any additional radioisotopes produced in the material cannot emit c-rays that interfere with the annihilation line directly or indirectly by producing count rates too high in the detector. The positron–electron pair creation technique overcomes the reliance on photo-nuclear reaction and the resulting radioisotopes by using pair creation to generate the positrons in the material. While this method benefits from using high Z materials, it has been successfully applied to aluminum, which has Z = 13. Imaging capabilities can be added to this technique in a manner similar to the photo-activation method. A schematic of an experimental arrangement is presented in Fig. 4. An electron accelerator generates a bremsstrahlung beam that is hardened by low Z absorbers and is then directed through a collimator in the wall. The highly collimated photon beam impinges on the specimen creating positron–electron pairs. A well shielded HPGe detector records the annihilation c-rays from a voxel inside the object. By

HPGe Detector

D

et

ec

3. Conclusion These preliminary results demonstrate how positron annihilation spectroscopy can image damaged regions in large structural components. The photo-activation technique, however, has some shortcomings. First, after irradiation any object will undoubtedly be quite radioactive and is a potential radiation safety concern. Second, the constituent nuclei of the object to be imaged must be suitable for defect imaging by photo-activation. This includes having a component from which a positron emitting radioisotope can be created in (c, n) reactions with suitable activity and half-life. For example, the only positron emitting radioisotope that can be made in large quantities in iron

to

rS

hi

el

di

ng Object

Photon Beam Voxel Collimators Fig. 4. Schematic drawing of a three-dimensional defect imaging systems. A DC electron accelerator (not shown) to the left of the wall produces the 10 MeV bremsstrahlung beam which is incident on the specimen. The object is on a translation and rotation stage allowing three-dimensional defect maps to be created.

266

A.W. Hunt et al. / Nucl. Instr. and Meth. in Phys. Res. B 241 (2005) 262–266

mounting the specimen on a two axis translation stage lateral defect images can be created and by adding a rotation stage three-dimensional defect maps can be produced. The use of pulsed electron LINACs with 4 MeV electron energies has limited the 511 keV count rate to 2 s 1, which is too low for defect imaging applications. Since the typical duty factor of a pulsed accelerator is 0.1%, the 511 keV count rate can be increased by two orders of magnitude by using a DC electron accelerator with a modest current of 25 lA. An additional factor of 7 in this count rate can also be obtained by increasing the electron energy to 10 MeV, bringing the total 511 keV count rate to 1.5 · 103 s 1. With this rate, 105 511 keV counts can be collected in 70 s and hence data from 411 voxels in the sample can be accumulated in an 8 h period, making threedimensional defect maps possible.

References [1] S. DeBenedetti, C.E. Cowan, W.R. Konneker, Phys. Rev. 76 (1949) 440. [2] S. Berko, J.C. Erskine, Phys. Rev. Lett. 19 (1967) 307. [3] I.K. MacKenzie, T.L. Khoo, A.B. McDonald, B.T.A. McKee, Phys. Rev. Lett. 19 (1967) 946. [4] B.T.A. McKee, W. Triftsha¨user, A.T. Stewart, Phys. Rev. Lett. 28 (1972) 358.

[5] K.G. Lynn, J.E. Dickman, W.L. Brown, M.F. Robbins, E. Bonderup, Phys. Rev. B 20 (1979) 3566. [6] P.J. Schultz, K.G. Lynn, Rev. Mod. Phys. 60 (1988) 701. [7] R. Wu¨rschum, P. Farber, P.S.R. Dittmar, W. Frank, H.E. Schaefer, Phys. Rev. Lett. 79 (1997) 4918. [8] T. Wider, S. Hansen, U. Holzwarth, K. Mair, Phys. Rev. B 57 (1998) 5126. [9] P. Asoka-Kumar, J. Hartley, R. Howell, P.A. Sterne, T.G. Nieh, Appl. Phys. Lett. 77 (2000) 1973. [10] G. Brauer, E.M.N.W. Anward, J. Kuriplach, I. Procha´zka, F. Becˇva´rˇ, A. Osipowicz, P.G. Coleman, Phys. Rev. B 62 (2000) 5199. [11] V. Ranki, A. Pelli, K. Saarinen, Phys. Rev. B 69 (2004) 115205. [12] G. Ko¨gel, W. Triftsha¨user, Phys. Rev. Lett. 48 (1982) 1741. [13] P. Asoka-Kumar, K.G. Lynn, D.O. Welch, J. Appl. Phys. 76 (1994) 4935. [14] M. Haaks, K. Bennewitz, H. Bihr, U. Ma¨nnig, C. Zamponi, K. Maier, Appl. Surf. Sci. 149 (1999) 207. [15] A. David, G. Ko¨gel, P. Sperr, W. Triftsha¨user, Phys. Rev. Lett. 87 (2001) 067402. [16] C. Zamponi, U. Ma¨nnig, T.E.M. Staab, K. Maier, S. Eichler, R. Hammer, Appl. Phys. Lett. 83 (2003) 4128. [17] A.P. Mills, R.J. Wilson, Phys. Rev. A 26 (1982) 490. [18] D.W. Akers, A.B. Denison, F. Harmon, Appl. Surf. Sci. 194 (2002) 245. [19] F.A. Selim, D.P. Wells, J.F. Harmon, W. Scates, J. Kwofie, R. Spaulding, S.P. Duttagupta, J.L. Jones, T. White, T. Roney, Nucl. Instr. and Meth. B 192 (2002) 197. [20] F.A. Selim, D.P. Wells, J.F. Harmon, J. Kofie, R. Spaulding, G. Erikson, T. Roney, Nucl. Instr. and Meth. A 495 (2002) 154. [21] F.A. Selim, D.P. Wells, J.F. Harmon, J. Kwofie, G. Erikson, T. Roney, Radiat. Phys. Chem. 68 (2003) 427.