Development of Combinatorial Defect Analysis with an Intense Positron Microprobe

Nagayasu Oshima1,∗, Yasuhiro Kamada2, Hideo Watanabe3, Atsushi Kinomura1,†, and Ryoichi Suzuki1 1Research Institute of Instrumentation Frontier (RIIF), National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Ibaraki 305–8568, Japan 2NDE and Science Research Center, Faculty of Engineering, Iwate University, 4–3–5 Ueda, Morioka 020–8551, Japan 3Research Institute for Applied Mechanics, Kyushu University, Kasuga 816–8580, Japan †Present address: Research Reactor Institute, Kyoto University, Kumatori-cho, Osaka 590–0494, Japan


Introduction
Positron annihilation lifetime spectroscopy (PALS) using a slow positron beam is a powerful and unique analytical method for evaluating atomic-scale defects near the sample surface [1,2].One can evaluate defect/pore properties at an arbitrary depth with a range from the surface to several micrometers by adjusting the positron injection energy over the range from zero to several tens of keV.In general, slow positron beam analysis requires samples with a size of around 10 mm because the typical positron beam diameter is more than several millimeters.
Recently, a positron lifetime measurement system with a high-intensity positron microbeam was developed in our facility at AIST, the positron probe microanalyzer (PPMA) [3].An intense slow positron beam is generated using a linear electron accelerator (LINAC) [4,5] and its brightness is enhanced by using combination of a focusing lens and a transmission moderator [6].The injection energy of the brightness enhanced beam into the sample is adjustable from 1 keV to 30 keV.The beam spot size at the sample is a few tens of micrometers [7].By using the PPMA, PALS can be applied at an arbitrary local spot (0.1 mm × 0.1 mm) within the scanable area (15 mm × 15 mm) on the sample.Positron lifetime spectra can be measured with a relatively high count rate of ∼10 3 s −1 and with sufficient time resolution for PALS, i.e. 200 ps-300 ps [8,9].
So far we have performed defect/pore analysis of various samples taking advantage of the capabilities of the PPMA i.e., PALS performed with good spatial resolution and high count rates.Analytical methods with the PPMA (i.e., how the PPMA was used) so far performed can be categorized as follows; (i) defect/pore analysis of small samples [10][11][12][13][14][15][16] (ii) defect/pore analysis for many local spots in large samples [3,[17][18][19] (iii) in-situ (in-air) defect/pore analysis of thin films [20,21] In the present study a new analytical method which we call the combinatorial technique is described.This combinatorial defect/pore analytical method can be categorized as follows; (iv) systematic defect/pore analysis of a large number of samples The combinatorial technique is a powerful method for evaluating large numbers of samples, prepared under different conditions, systematically.In this paper, the concept of defect/pore analysis using the combinatorial technique is proposed.A test performance of this combinatorial method is also presented by applying it to the analysis of ion beam irradiated Fe film samples.

Concept of combinatorial defect/pore analysis
Combinatorial defect/pore analysis with PPMA is a method to analyze the defect/pore properties of a large number of samples (up to a maximum of several thousand in principle) systematically within practically reasonable times i.e., within the limited machine-time of a LINAC-based or other high intensity facility, typically a few days in case of AIST.The conceptual difference between the traditional analytical method using a conventional slow positron beam (diameter ϕ ∼ 10 mm) and the combinatorial method with the AIST-positron microbeam (ϕ ∼ 0.1 mm) are schematically shown in Fig. 1.In the conventional method, samples are prepared with different fabrication/processing conditions with a typical size of 10 mm.We reasonably assume that only several samples can be mounted on a sample holder and positron lifetime of these samples are then measured automatically by moving the holder using a linear translation system without human intervention.One can continue PALS measurements by breaking the vacuum condition of the measurement chamber and replacing a sample holder, however, it is obvious that machine time cannot be effectively used.Therefore, we can estimate that this conventional method can measure only a limited number of samples within a typical machine time run.On the contrary, in the combinatorial method, large numbers of samples with relatively small sizes are prepared on the same substrate.As mentioned above, the AIST-PPMA can analyze samples with a lateral spatial resolution of less than 0.1 mm in the scanning area of 15 mm × 15 mm, and hence, up to ∼10 4 different samples can be mounted on one sample holder in principle.Therefore, we can prepare samples with many different fabrication/processing conditions and measure how these conditions affect the defect/pore properties systematically in a single run (within a typical LINAC machine time) by using the combinatorial technique.
In PALS, normally 1 × 10 6 -3 × 10 6 positron annihilation events are recorded to obtain a lifetime spectrum.These spectra are then usually analyzed by the least squared fitting method using a model several-thousand samples on a holder several samples on a holder function consisting of a sum of several decaying exponentials as follows [2]; Here, t i and I i are mean lifetime of the i-th component and its intensity.Since the counting rate of positron annihilation events using the PPMA is around 1 × 10 3 s −1 -3 × 10 3 s −1 , the required time for one sample analysis is of the order of ∼10 3 s.This means that the experiment would take more than 10 days if sample number is more than 10 3 .This long machine time is unpractical in normal circumstances.On the other hand, averaged positron lifetimes can be obtained to an accuracy of several ps by recording a relatively low number of annihilation events, about 10 4 counts [2,17].Here, the average positron lifetime is given as follows; Average positron lifetime can be a good index for characterizing defect/pores properties of samples [2], therefore, in the combinatorial method, it is suitable to analyze a large number samples by the following procedure.Process (a): averaged positron lifetime of samples are measured rapidly by recording only ∼10 4 positron annihilation events.Then, about 10 3 -10 4 samples can be analyzed/evaluated within practically reasonable times like a few hours-a few days.Techniques for rapid average lifetime measurement have already been developed [17].Process (b): secondly, complete positron lifetime spectra are measured for selected samples by recording ∼10 6 positron annihilation events.In this process about 10-10 2 samples can be analyzed within practically reasonable times.We should select 10-10 2 samples for PALS according to the results of process (a).

Application of the combinatorial defect analysis with PPMA
We applied this combinatorial technique using the AIST-PPMA for a Fe-based alloy to understand the role of open volume defects in the embrittlement mechanism of reactor materials.Matrix defects including vacancies and vacancy-clusters have been studied as one of main factors enhancing embrittlement and therefore positron annihilation techniques are often used for studying this field [15,[22][23][24][25].
Various parameters can affect the embrittlement of Fe-based alloys including chemical composition, heat environment (heat treatment), and radiation environment (neutron/ion irradiation flux and dose).Therefore, we expect that the combinatorial method is suitable for studying the dependence of defect characteristics on those various parameters systematically and effectively.Here, we prepared a test sample-piece consisting of nine Fe samples with different irradiation conditions (damage and irradiation-temperature) on the same substrate for a demonstration of combinatorial defect analysis.We describe the experimental results in the following sub-sections.

Experimental procedure
Pure-Fe thin film was prepared by an electron beam evaporation method in ultra-high vacuum.The top-surface-layer of Fe(001) (330 nm) and the following-buffer-layer of MgO (6 nm) were deposited on a MgO(001) substrate with a size of 10 mm × 10 mm × 0.5 mm.This piece (Fe (330 nm) / MgO(6 nm) / MgO) was annealed at a temperature of 500 • C for 10 min in vacuum.Nine different areas of this Fe film were assigned different sample numbers with a size of 2.0 mm × 2.0 mm and labelled as #1-#9 as shown in Fig. 2. Each sample was irradiated by a Cu 2+ ion beam where the irradiation time and sample temperature during irradiation was changed.The ion beam energy was  during irradiation was controlled using a movable beam shutter and a heating stage respectively (see Fig. 2).The status of the beam shutter and stage temperature during sample preparation are shown schematically in Fig. 3.The sample denoted as #1 was not irradiated.The samples denoted as #2, #3, #4, and #5 were irradiated at room temperature for 7.5 min, 15 min, 30 min, and 60 min, respectively.The samples denoted as #6, #7, #8, and #9 were irradiated at 290 Positron lifetime was measured in vacuum with the AIST-PPMA in which a pulsed and brightness enhanced positron beam with a diameter of 0.1 mm was injected into the samples.Positron injection energies were set to E = 3.0 keV and 7.0 keV in this experiment and the time resolution for measuring positron lifetime was ∼250 ps FWHM.In this test experiment 10 different samples were measured including one reference sample.The time required to obtain a complete lifetime spectra of all 10 samples is only a few hours, and hence, we skipped the rapid average-lifetime measurement described above (process (a)) and measured 1 × 10 6 events for all samples in this experimental demonstration.

Stopping profiles of ions and positrons
Figure 4(a) and (b) shows the respective stopping profiles of Cu 2+ ions and positrons in a Fe target calculated with our experimental parameters.The simulation code SRIM [26] was used for ions and the Makhovian function [1,2] is used for positrons.Most of the implanted ions pass through the Fe layer while positrons are stopped (i.e.their kinetic energy is reduced down to the thermal level) in the Fe layer.The mean implantation depths z m of positrons for E = 3.0 keV and 7.0 keV were approximately 30 nm and 110 nm, respectively.It should be noticed that the actual positron annihilation depth profile might be broadened from the calculated profile because of thermal diffusion of positrons.In Fig. 4(a), the damage profile due to ion beam irradiation for 1 hour is also plotted in units of displacement per atom (dpa) and is approximately given by D (dpa • h −1 ) ∼ 0.0034 × z (nm) + 0.8.Radiation doses around the positron mean implantation depths z m = 30 nm (E = 3.0 keV) and 110 nm (7.0 keV) are estimated to be ∼0.9 dpa • h −1 and ∼1.2 dpa • h −1 , respectively.

Unirradiated sample
Figure 5 shows positron lifetime spectra of sample #1 measured at E = 3.0 keV and 7.0 keV.A long lived positron lifetime of more than ∼1 ns was clearly seen only in the spectrum measured at E = 3.0 keV and not seen at E = 7.0 keV.In general the positron lifetime in bulk metals is much shorter than 1 ns.The origin of the positron lifetime longer than 1 ns can be explained as positrons which annihilate around the surface.Some fraction of the injected positrons can return to near the surface due to thermal back diffusion and be re-emitted into vacuum as a free positron or as the positron-electron bound state positronium and finally annihilate somewhere outside the sample.
■■■ 011306-5 JJAP Conf.Proc.(2014) 011306 These annihilation processes are responsible for the long lived component of more than ∼1 ns in the annihilation spectrum.In general, the intensity of such annihilation process becomes higher with lower injection energy.Therefore, the long lived component which only appeared in the spectrum for E = 3.0 keV can be reasonably explained by positron annihilation around the surface due to the back diffusion of thermalized positrons.The lifetime spectra of sample #1 measured at E = 7.0 keV was well characterized by a two component fit.The shorter lifetime τ 1 (longer lifetime τ 2 ) and its intensity I 1 (I 2 ) are τ 1 ∼ 0.25 ns (τ 2 ∼ 0.39 ns) and I 1 ∼ 55 % (I 2 ∼ 45 %), respectively.By comparing with calculated values of positron lifetime, the shorter lifetime τ 1 ∼ 0.25 ns can be attributed to positron annihilation at vacancy clusters with an average size of ∼V 5 while the longer lifetime can be attributed to positron annihilation at vacancy clusters with an average size of more than ∼V 15 and/or annihilation at the surface.We cannot derive a lifetime component of less than ∼100 ps which would be attributed to annihilation in a defect free region.This implies that this sample has a high density of positron trapping sites i.e., defects without ion beam irradiation.Therefore, we cannot evaluate defect densities by using the positron annihilation rate equation, the so-called "trapping-model", because this model cannot be applied in the present case when all positrons are annihilated in defects.
In general, defects in pure-Fe of bulk-samples observed by PALS disappear after annealing at 500 • C. The existence of vacancy clusters in our annealed sample may because we have used a thin film sample.The orientational relationship is Fe [110] (001) // MgO [100] (001) [27].The lattice constants for bulk of MgO and Fe are 0.4213 nm and 0.2866 nm respectively and hence the misfit of the periodic structure of MgO and Fe atoms along to the same direction is −3.8 %.We guess that lattice defects exist in Fe even after the 500 • C annealing due to this misfit although further studies are needed to investigate their origin.

Samples irradiated at RT
Examples of positron lifetime spectra of unirradiated samples (#1) and irradiated by ions at RT (#2 and #3) were shown in Fig. 6 where all spectra were measured at E = 3.0 keV.The intensity of the long lived component (longer than 1 ns) attributed to surface annihilation decreased after 7.5 min irradiation and disappeared after 15 min irradiation.For the E = 7.0 keV measurement, the intensity of the long lived component disappeared after only 7.5 min irradiation.This can be reasonably explained by an increase in defect density due to ion irradiation resulting in a reduction in the positron back diffusion probability.Therefore we can conclude that the positron diffusion length in RT-irradiated samples was much smaller than the mean positron implantation depth (30 nm for E = 3.0 keV and 110 nm for E = 7.0 keV), and hence, we can reasonably assume that the broadening of the positron stopping profile due to thermal diffusion is negligibly small.Lifetimes and intensities, extracted from the spectra at E = 7.0 keV for the RT irradiation samples (#2, #3, #4, and #5), are plotted with fitting error bars in Fig. 7.These spectra were well characterized using only a one component fit and the deduced lifetimes well agreed with each other, namely, τ 1 = 0.37 ns.This value for the lifetime indicates that main positron annihilation site after ion beam irradiation at RT is vacancy clusters ∼V 15 and its size weakly depends on the irradiation dose in our experimental range (0.1 dpa-1.2dpa).Our result is consistent with results from Doppler broadening of positron annihilation radiation (DBAR), reported by Iwai et al., in which the Doppler parameter (S parameter) of samples irradiated at RT did not depend on the irradiation dose for > 0.1 dpa [23].The results for the RT irradiation samples at E = 3.0 keV were very similar to those for E = 7.0 keV, indicating that there was no obvious depth dependence on defect information in our sample.

Samples irradiated at 290 • C
A lifetime component longer than ∼1 ns was not observed for samples irradiated at 290 • C indicating that the positron diffusion length is much smaller than mean positron implantation depths z m .In contrast to RT-irradiation, the positron lifetime spectra for the samples irradiated at 290 • C varied depending on the irradiation time (dose) in our experimental range (> 0.1 dpa).The fitting results for E = 7.0 keV are shown in Fig. 8.Only the positron lifetime spectrum of sample #9 was fitted by one component while the other spectra (#6, #7, and #8) were well fitted by a two component analysis.It should be noticed that the fitting result of #6 (0.15 dpa) is very similar to that of #1 (unirradiated) and fitting result of #9 (1.2 dpa) is the same as that of RT-irradiation case (> 0.1 dpa).The intensity of the relatively shorter lifetime ∼0.25 ns component gradually decreased with increasing irradiation time and become finally zero at 60 min irradiation (#9).So, heating of samples to 290 • C during ion irradiation seems to delay the defect evolution rate (or vacancy accumulation rate) by around at least 10 times compared with the RT irradiation case.The temperature dependence of defect evolution rates can be attributed to the balance between defect clustering and defect annihilation at sinks (e.g., dislocations and interfaces).Decrease of defect accumulation in Fe bulk samples by increasing irradiation temperature was reported for neutron irradiation experiments [22].On the other hand, it should be noticed that each sample has a different post irradiation annealing history in our experiment.We need further study to understand the difference of the heating effect between "during irradiation" and "after irradiation" by measuring many samples systematically with the combinatorial technique.The fitting results for the samples irradiated at 290 • C with E = 3.0 keV were similar to those for E = 7.0 keV indicating that there was no obvious depth dependence on defect information in our sample.

Summary
The concept of combinatorial defect/pore analysis using an intense positron microprobe is proposed.This combinatorial method is suitable to analyze defects/pores for a large number of samples systematically and in a single measurement run.A test of this method was performed by applying it to the analysis of an ion beam irradiated Fe film in which nine separate samples (2.0 mm × 2.0 mm) with different irradiation conditions were prepared on one 10 mm × 10 mm sized substrate.In principle, by preparing rather small sized samples on the same substrate one can analyze defect/pores of up to several thousand samples in a single measurement using an intense positron microprobe.

Fig. 1 .
Fig. 1.(a) Comparison of the measurement set-up with a conventional slow positron beam and (b) the combinatorial technique at the AIST-PPMA.

Fig. 2 .
Fig. 2. (a) One piece of Fe (330 nm) film with size of 10 mm × 10 mm was set in an ion beam irradiation chamber.(b) Nine different samples denoted as #1-#9 were prepared, each with a size of 2 mm × 2 mm where the ion beam irradiation time and temperature during irradiation were changed by controlling the beam shutter and heating stage.

Fig. 4 .
Fig. 4. Stopping profiles of 2.4 MeV Cu 2+ ions (a) and positrons (b) in a Fe target calculated for our experimental parameters.The damage profile due to irradiation of 2.4 MeV Cu 2+ for 1 hour is also plotted in units of displacement per atom (dpa) in (a).

Fig. 6 .
Fig. 6.Example of positron lifetime spectra of samples (#2 and #3) irradiated by ions at RT. Spectrum of unirradiated sample (#1) was also plotted.All spectra were measured with positron injection energy E = 3.0 keV.

Fig. 7 .
Fig. 7. Lifetimes and intensities obtained from fitting the measured positron annihilation lifetime spectra for the samples (#2, #3, #4, and #5) irradiated at RT.All irradiated spectra were fitted well by one component.PALS was performed with a positron energy E = 7.0 keV.The two-component-fit result for the unirradiated sample (#1) is also plotted.

Fig. 8 .
Fig. 8. Fitting results of samples ion-irradiated at 290 • C.Only the positron lifetime spectrum of sample #9 was fitted by one component while the other spectra (#6, #7, and #8) were well fitted by a two component analysis.The result for the unirradiated sample (#1) is also plotted.