Research article Received: 16 February 2016 Revised: 24 April 2016 Accepted: 2 May 2016 Published online in Wiley Online Library (wileyonlinelibrary.com) DOI 10.1002/sia.6049 Nanoscale imaging of Li and B in nuclear waste glass, a comparison of ToF-SIMS, NanoSIMS, and APT Zhaoying Wang,a,b† Jia Liu,b† Yufan Zhou,b,c James J. Neeway,d Daniel K. Schreiber,d Jarrod V. Crum,d Joseph V. Ryan,d Xue-Lin Wang,c Fuyi Wanga* and Zihua Zhub* It has been very difficult to use popular elemental imaging techniques to image Li and B distribution in glass samples with nanoscale resolution. In this study, time-of-flight secondary ion mass spectrometry, nanoscale secondary ion mass spectrometry, and atom probe tomography (APT) were used to image the distribution of Li and B in two representative glass samples, and their performance was comprehensively compared. APT can provide three-dimensional Li and B imaging with very high spatial resolution (≤2 nm). In addition, absolute quantification of Li and B is possible, although there remains room for improving accuracy. However, the major drawbacks of APT include poor sample compatibility and limited field of view (normally ≤100 × 100 × 500 nm3). Comparatively, time-of-flight secondary ion mass spectrometry and nanoscale secondary ion mass spectrometry are samplefriendly with flexible field of view (up to 500 × 500 μm2 and image stitching is feasible); however, lateral resolution is limited to only about 100 nm. Therefore, secondary ion mass spectrometry and APT can be regarded as complementary techniques for nanoscale imaging of Li and B in glass and other novel materials. Copyright © 2016 John Wiley & Sons, Ltd. Additional supporting information may be found in the online version of this article at the publisher’s web site. Keywords: ToF-SIMS; NanoSIMS; APT; nanoscale imaging; lithium; boron; nuclear waste glass Introduction The current main-stream techniques of nuclear waste treatment are immobilizing nuclear wastes into glass or glass ceramic and then storing the nuclear waste glass or glass ceramic at pre-selected disposal facilities.[1,2] Although silicon oxide is the major matrix material in nuclear waste glasses and glass ceramics, boron oxide and lithium oxide are two important components, too. The presence of boron in glass as a network component reduces the thermal expansion coefficient, making it more resistant to thermal shock. In addition, boron as a modifier gives the glass a sharp dependence of viscosity on temperature, which is beneficial for vitrification. Lithium may present in some nuclear waste, but more lithium (with other alkali metal, such as Na, K, Ru, and Cs) is intentionally added into glass as modifier to facilitate nuclear glass fabrication and waste incorporation.[3] Therefore, nanoscale imaging of Li and B in nuclear waste glasses and glass ceramics is of great interest. Reasons for this are due, firstly, to glass ceramics showing separate phase structures, where the phase sizes can range from nanometers to tens of micrometers. Therefore, it is very important to confirm that Li and B stay in desirable phases so the stability of the whole system can be maximized. The second reason why Li and B imaging is of interest is related to understanding the corrosion behavior of glass and glass ceramic materials because radionuclides initially encapsulated in the waste forms will be released into the surrounding environment upon contact with ground water in the geological repository.[4–7] Therefore, understanding the behavior of different components (including Surf. Interface Anal. (2016) Li and B) during glass corrosion and developing a reasonable model to predict glass corrosion behavior in a geological repository are of great importance. It has been reported that the elemental profiles of corrosion frontlines may be irregular and many nanometer scale structures with different physical and chemical properties may exist.[5,8,9] Therefore, techniques that can provide two-dimensional * Correspondence to: Zihua Zhu, Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, Richland, WA 99352, USA. Fuyi Wang, Beijing National Laboratory for Molecular Sciences, CAS Key Laboratory of Analytical Chemistry for Living Biosystems, Beijing Centre for Mass Spectrometry, Institute of Chemistry, Chinese Academy of Sciences, Beijing 100190, China. E-mail: [email protected]; [email protected] † Equal contribution. a Beijing National Laboratory for Molecular Sciences, CAS Key Laboratory of Analytical Chemistry for Living Biosystems, Beijing Centre for Mass Spectrometry, Institute of Chemistry, Chinese Academy of Sciences, Beijing 100190, China b Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, Richland, WA, 99352, USA c School of Physics, State Key Laboratory of Crystal Materials & Key Laboratory of Particle Physics and Particle Irradiation (MOE), Shandong University, Jinan 250100, China d Energy and Environment Directorate, Pacific Northwest National Laboratory, Richland, WA, 99352, USA Copyright © 2016 John Wiley & Sons, Ltd. Z. WANG ET AL. and three-dimensional (3-D) elemental and isotopic distributions with nanometer scale spatial resolution would be a major benefit to the field. Currently, scanning electron microscope (SEM) and transmission electron microscope are the two most important techniques for nanoscale imaging of elements in scientific research. Combined with energy dispersive spectroscopy (EDX), SEM can perform elemental imaging with a lateral resolution of ≥1 μm. However, the major drawback of SEM/EDX is poor sensitivity for light elements, such as B and Li. In addition, 1-μm lateral resolution is not sufficient for the nanoscale imaging need for accurately locating Li and B in the pristine glass ceramic or altered glass structure. Scanning transmission electron microscope (STEM) can combine with EDX to provide better spatial resolution (down to 0.1 nm is possible), but it still suffers from EDX’s intrinsic low sensitivity for light elements. STEM can combine with electron energy loss spectroscopy (EELS), and this combination has been widely used for Li and B imaging with very high spatial resolution (down to 0.1 nm is possible). A few successful cases using STEM-EELS in glass corrosion research have been reported,[8,10] where B corrosion frontline can be clearly observed with ~30–50 nm lateral resolution, but no Li imaging data were reported. However, charging effects, relatively low concentrations of B and Li, and interference from other components make imaging of B and Li using STEM-EELS in glass a challenging task (especially, for Li imaging). X-ray photoelectron spectroscopy is also a very popular surface analysis technique, but it can only perform elemental imaging with a lateral resolution of ~10 μm, which is far away from nanoscale imaging necessary to identify the multiple phases present in glass ceramics. Additionally, the technique is only sensitive to the first few nanometers of the surface, and this limits its usefulness in terms of understanding the behavior and location of different elements during the glass corrosion process. Lastly, in principle, Auger electron spectroscopy may be used for B and Li imaging glass samples with sub-micrometer lateral resolution; however, severe charging effects make this task impractical. Time-of-flight secondary ion mass spectrometry (ToF-SIMS) has been used for elemental and isotopic depth profiling of leached glass samples for over 10 years.[4,10–19] It can also be used for elemental imaging (including B and Li) with ≥100 nm lateral resolution.[20,21] Recently, nanoscale secondary ion mass spectrometry (NanoSIMS) and atom probe tomography (APT) were used for elemental imaging (including B and Li) glass samples.[5,8,9,22] As far as we know, very few other laboratory tools can provide B and Li nanoscale imaging for glass samples. Therefore, it is reasonable to expect that ToF-SIMS, NanoSIMS, and APT will play a more and more important role in glass research in the future. However, so far, a comparison study of these three techniques in glass imaging is not available. This creates a challenge for scientists in the field to choose the right tools for their scientific research to answer the specific questions that they are asking. In addition, because Li and B play a very important role in numerous novel nanomaterials (e.g. novel electrode materials used in future Li ion batteries,[23,24] boron nitride nanotubes and nanosheets,[25] etc.), therefore, a comparison study of ToF-SIMS, NanoSIMS, and APT for Li and B imaging will benefit such fields, too. Time-of-flight secondary ion mass spectrometry, NanoSIMS, and APT are unique techniques, and a rare ensemble of instruments co-located at the Environmental Molecular Sciences Laboratory. Environmental Molecular Sciences Laboratory is a Department of Energy national scientific user facility at Pacific Northwest National Laboratory and might be the only place worldwide equipped with all three instruments (until December 2015). We have used these wileyonlinelibrary.com/journal/sia instruments extensively in glass research in the last several years[8,9,16,17,19,21,22] and are in a unique position to perform the comparison study of ToF-SIMS, NanoSIMS, and APT for Li and B imaging in glass samples. In this study, we report a complementary approach using these three techniques to image the distribution of Li and B in two representative samples (a glass-ceramic sample and a water-corroded glass sample). The strengths and limitations of each technique will be discussed. Experiment Sample preparation Glass-ceramic sample The glass ceramic was fabricated using traditional crucible-scale melting. Details of the preparation procedure are available elsewhere.[22] A piece of glass ceramic (~10 mm × 10 mm × 3 mm) was cut from the location of interest and immobilized into resin with one of the 10 mm × 10 mm sides exposed for polishing. After polishing, this sample was used for SEM, ToF-SIMS, and NanoSIMS analysis. Cutting, resin immobilization, and polishing details have been depicted in our previous paper.[22] Specimens for APT analysis were extracted from the polished glass-ceramic sample using standard focus ion beam (FIB) lift-out procedures for APT specimens with a dual-beam SEM/FIB (FEI Helios Nanolab 600, Hillsboro, OR, USA).[26] In brief, a Pt capping protection layer of 2 × 20 μm2 was deposited over a region of interest consisting of approximately 50 nm-thick electron beamdeposited Pt followed by 150 nm-thick ion beam-deposited Pt. Two rectangular trenches were made along the Pt cap at 60 ° with respect to the specimen surface to create a triangular lamellar wedge. The tungsten needle of a micromanipulator was then attached to one end of the wedge followed by cutting two attached ends releasing the wedge. Individual sections of the wedge were then attached to commercially available (CAMECA Instruments Inc., Madison, WI, USA) Si microposts. Specimens were sharpened into the conical geometry using annular mill patterns with a 30-keV Ga+ beam, and final specimen cleanup was performed with a 2-keV Ga+ beam to minimize ion-beam damage. The radius at the apex of final tip was ~50 nm. Corroded glass sample The glass used in this study was the Advanced Fuel Cycle Initiative (AFCI) glass, and the entire composition is described in a previous publication.[27] The mass percentage of B2O3 and Li2O in the glass was 9.65% and 4.50%, respectively. Briefly, the glass batch was prepared by first weighing the proper mass of each reagent-grade chemical in the form of metal oxides, carbonates, H3BO3, and salts. Batches were then homogenized by mixing in an agate mill and then melting in a platinum crucible for 1 h. The melts were then quenched by casting them onto clean stainless steel plates. The quenched glass was crushed in an agate mill, remelted for 1 h, and again quenched onto clean stainless steel plates. The corrosion procedure consisted of placing AFCI glass powder (32–75 μm) and six coupons (~10 mm × 10 mm × 1 mm each) in a polytetrafluoroethylene (PTFE) vessel enclosed in a steel pressure vessel and contacting the glass with high-purity water (18.2 MΩ/cm) for a glass-surface-to-solution-volume ratio of roughly 20 000 m 1. The pH was unconstrained. The samples were placed in a convection oven held at 90 ± 2 °C for 358 days Copyright © 2016 John Wiley & Sons, Ltd. Surf. Interface Anal. (2016) Nanoscale imaging of Li and B in nuclear waste glass at which point the glass coupons were removed, rinsed with water and then ethanol, and allowed to dry. For NanoSIMS imaging of the corrosion layer of the AFCI sample, a wedged crater was prepared by an FIB instrument. A brief introduction of the procedure is shown in the Supporting Information (Figure S1). The details of the preparation procedure of the trench can be seen in our previous publication.[9] As shown in Figure S1, the slope of the wedged crater was about 5 μm wide. One side of the slope was sample surface, and the depth of the other side of the slope was about 1 μm. Instrumentation and data analysis Time-of-flight secondary ion mass spectrometry imaging was performed using an IONTOF TOF.SIMS5 spectrometer. The glassceramic sample was coated with 10 nm Au to reduce charging effects. Before imaging analysis, the direct current (DC) beam from a 25-keV primary Bi ion source was used to scan an area of 12 × 12 μm2 for ~60 s to remove the Au coating. To collect image data, a 25-keV Bi+ analysis beam was used to scan an area of 10 × 10 μm2 at the center of the DC crater with a frame of 256 × 256 pixels. The 25-keV Bi+ beam was focused to a diameter of ~100 nm with a current of ~0.12 pA. It should be noted that only unit mass resolution spectra could be obtained because our ToFSIMS instrument cannot provide high mass resolution and high spatial resolution, simultaneously. The collection time for each frame was about 2 s, and a typical collection time for one measurement is about 3000 s. A low-energy (~10 eV) electron beam was used for charge compensation. Charge compensation tuning was tricky, and relevant parameters (e.g. reflector voltage) needed to be carefully adjusted before each measurement.[16] The data analysis software is SURFACELAB 6.4, which was provided by the instrument manufacturer. Nanoscale secondary ion mass spectrometry imaging was performed using a CAMECA NanoSIMS 50 L. The glass samples were coated with 10 nm Au to reduce charging. A 16.0-keV Cs+ beam (0.65 pA) was used as the primary beam, which was focused to a spot size around 70 nm. The Cs+ beam was used to scan an 8 × 8 μm2 area (for the glass-ceramic sample) or a 7 × 7 μm2 area (for the corroded AFCI sample) with 256 × 256 pixels per frame, and the data collection time was about 11 min. Low-energy (≤10 eV) electrons were introduced to the sputter interface to compensate surface charging so that reasonable signal intensity could be obtained. 7Li , 18O , 11B16O , and 27Al16O were collected simultaneously. The 16O signal was too strong, and it led to some signal saturation problems. All secondary ion images were acquired after a pre-sputtering process. The pre-sputtering was performed using the 16.0-keV Cs+ beam with an ~160 pA current over a 10 × 10 μm2 area. The pre-sputtering time was about 100 s. This process was carried out to remove the Au-coating layer, clean surface contamination, and also prepare a mature analysis area with enough Cs implantation for optimal intensities of negative ion signals. Image processing was carried out using the software IMAGEJ 1.46r (Wayne Rasband, National Institute of Health, USA, http://rsbweb.nih.gov.ij/index.html) equipped with the OpenMIMS plug-in (http://www.nrims.harvard.edu.software.php). The APT analyses were performed with a CAMECA LEAP 4000X HR in laser pulsing mode (λ = 355 nm).[28,29] The instrumental setting used in this research was similar with that used in our previous study of corroded SON68 glass samples.[8] In brief, the laser pulse repetition rate was set at 160 kHz, the laser pulse energy was 80 pJ per pulse, and the target evaporation rate was 3 detected ions per 1000 pulses. The APT specimen was held at 40 K during analysis Surf. Interface Anal. (2016) in ultrahigh vacuum (<1 × 10 8 Pa). A minimum of 5 million ions were collected during the analysis. Atom probe tomography reconstruction parameters were estimated based on the analysis volume measured by SEM imaging of each specimen prior to and after APT analysis. The data analysis software was integrated visualization and analysis software (version 3.6.8), which was provided by CAMECA. An iso-concentration surface (4 at.% Li) was used to define the topography of the droplet phase and matrix interface and thereby maximized the accuracy of the compositional and interfacial width measurements using the proximity histogram method.[29,30] Because of the complex nature of the glass-ceramic sample and limited mass resolution (M/ΔM ~800 for 16O+) in APT mass spectra, the major challenges in data reconstruction were peak identification and peak deconvolution. The most significant peak convolution in this study was 11B16Ox with 27Al16Ox 1. This convolution was partially recovered using the natural isotopic abundances of 10B (19.9%) and 11B (80.1%), but this recovery was insufficient for absolute compositional accuracy. More details on the APT analysis of the glass specimens have been discussed in a separate article.[8] Results and discussion Glass-ceramic sample In order to reveal the microstructure of the glass-ceramic sample after quenching, a cross section of the specimen was examined with SEM to observe the morphology and distribution of multiple phases. Figure 1 shows a representative microstructure of the amorphous region from glass ceramic. The phase with dark contrast and spherical shape was uniformly distributed in the lightcontrast background. The contrast difference presented in the two phases indicates a liquid phase separation during the vitrification process. In this study, the spherical-shape phase was called the droplet phase, while the background phase was called the matrix phase. It was difficult to use SEM/EDX to do elemental imaging for this sample because the size of the droplet phase (around 200–500 nm) was smaller than the best lateral resolution of SEM/EDX (normally ~1.0 μm). STEM/EDX can provide elemental distribution with higher spatial resolution. The previously obtained STEM/EDX images show that Al, Si, and Cs are enriched in the droplet phase and the alkaline earths and lanthanides are enriched in the matrix phase.[22] However, STEM/EDX cannot provide B and Li distributions because of the intrinsically weak signal of Li and B. 2 µm Figure 1. A scanning electron microscope image, showing the dropletmatrix phase separation in the glass-ceramic sample. Copyright © 2016 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/sia Z. WANG ET AL. STEM-EELS is sensitive to B and Li, and it has been used to image B distribution in a leached SON68 sample.[8,10] We tried STEM/EELS; however, our result (data not shown here) did not provide any useful signal of Li and B from this glass-ceramic sample because of high background noise, which can be attributed to charging effects, relatively low concentration of B and Li, and interference from other components. Atom probe tomography Normalized to total counts Atom probe tomography is a unique analysis technique that can provide 3-D chemical and structural information on a nanometer scale. The basic principle of this technique is detailed elsewhere.[31,32] Typical analyses may contain 1–10 s of millions of atoms reconstructed within volumes of approximately 100 × 100 × 500 nm3. From an APT analysis, the mass spectrum provides the chemical identity for each detected ion, which provides a means to quantify the composition based on the number of counts, while a correctly scaled 3-D atom reconstruction provides a means to quantify the spatial distribution of individual atoms and assess atomic-scale heterogeneity. Figure 2 shows the element-specific atom maps from a subsection of APT data reconstruction from the glassceramic specimen. Each dot in this image represents one detected ion. The curvature of surface contour indicates that the region on the left side of the interface represents the droplet phase and the region on the right side of the interface represents the matrix phase. The data clearly show Li and B were enriched in the matrix phase. It should be noted that the composition analysis of several representative heavy elements (data will be reported in another paper) shows Al and Cs were enriched in the droplet phase, which is consistent with our previous STEM/EDX results.[22] Atom probe tomography analysis shows some ‘matrix effect’ for the droplet-matrix sample, which suggested by the observation that the atomic density of all detected total ions in the matrix phase was higher than that in the droplet phase. This situation may be attributed to the different evaporation fields of materials with various compositions; therefore, normalization seems very necessary to obtain the absolute concentration of Li and B in the two phases. In addition, it should be noted that many complex ion peaks, e.g. BOx+ and AlOx+, were observed in the APT mass spectrum. During data analysis, the complex ions were decomposed to obtain atomic composition. Figure 3 shows the proximity histogram calculated across the iso-concentration interface to quantify the compositional differences.[33] The APT instrument we used can provide a depth resolution of 0.2 nm and a lateral resolution of 0.3 nm, when a desirable reference sample is tested. We therefore assigned distance uncertainty to 0.5 nm, which was double the typical resolution for this APT instrument. The concentrations calculated for Li, B, and O (including all atoms in both monatomic and complex ions) were normalized to the total number of counts with background subtracted and presented in Table 1. For convenience, all concentration reported in APT analysis was atomic percentage. It is seen that oxygen atoms showed similar concentration in the two phases (56% in the droplet phase and 53% in the matrix phase, as atomic percentage), consistent with our expectation. The Li atoms showed 1.8% in the droplet phase and 5.5% in the matrix phase; the B atoms showed 2.2% in the droplet phase and 3.8% in the matrix phase. In addition, despite the visual similarity of the B and Li atom images, the quantitative widths of these composition gradients were strikingly different. Using the ASTM International standard, the interfacial width can be regarded as the distance, over which (b) (a) 20 nm 0.70 Droplet 0.60 Matrix 0.50 Li O B 0.06 0.04 0.02 0.00 -10 -5 0 5 Distance (nm) 10 Figure 3. A proximity histogram calculated across the iso-concentration interface, showing the concentration change of Li, B, and O across the droplet-matrix interface. (c) Table 1. Concentrations (atomic percentage) and interfacial widths of Li, B, and O atoms obtained from the atom probe tomography analysis of the glass-ceramic sample (d) Figure 2. Element-specific maps of a 10-nm-thick subsection of the atom probe tomography data from a specimen lifted from the droplet-matrix phase separation region of the glass-ceramic sample. (a) Total ion; (b) 6 + 7 + 10 + 11 + 10 2+ 11 2+ 16 + Li + Li ; (c) B + B + B + B ; and (d) O . The scale bar is equal to 20 nm. wileyonlinelibrary.com/journal/sia O Li B Droplet phase Matrix phase Interfacial width (nm) 56% 1.8% 2.2% 53% 5.5% 3.8% — 2.4 4.8 Copyright © 2016 John Wiley & Sons, Ltd. Surf. Interface Anal. (2016) Nanoscale imaging of Li and B in nuclear waste glass there is a 16–84% change in signal intensity. If so, the interfacial widths were found to be 2.4 ± 0.5 nm for Li and 4.8 ± 0.5 nm for B. The data clearly show the advantage of APT on high resolution measurement of interfacial widths. In addition, it was demonstrated that each species may exhibit a significantly different interfacial width, which might only be observed using the APT method. Time-of-flight secondary ion mass spectrometry Figure 4 presents the ToF-SIMS images of Li+, B+, O+, and Al+, the overlay image of Li+ and Al+, and the overlay image of B+ and Al+. The results of this analysis show that Li+ was enriched in the matrix phase, whereas Al+ was enriched in the droplet phase, consistent with APT analysis. Such distributions could be highlighted by the overlay image of Li+ (red) and Al+ (green), where the contrast of red and green colors was vivid. However, the B+ image looked fuzzy because of relatively low counts, although careful examination showed it might share a similar distribution with Li+. Also, due to low counts of B+ signal, the overlay image of B+ (red) and Al+ (green) was dominated by Al+. In Fig. 4, signals showed some fall-off at the top, right, and bottom edges, which can be attributed to uneven charging in the analysis area. It should be noted that a low-current (tens to hundreds of pA) 1.0-keV O2+ beam, which was scanned on a 600 × 600 μm2 over Figure 4. Time-of-flight secondary ion mass spectrometry positive secondary ion images, showing the distribution of Li, B, O, and Al in the droplet-matrix 6 + 7 + 10 + 11 + 16 + 27 + phase separation region of the glass-ceramic sample. (a) Sum-counts image of Li and Li ; (b) sum-counts image of B and B ; (c) O ; (d) Al ; (e) overlay of image (a) and (d); and (f) overlay of image (b) and (d). The black spots are most probably the bubbles in the sample. Surf. Interface Anal. (2016) Copyright © 2016 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/sia Z. WANG ET AL. the imaging area during data acquisition, could be used to reduce such intensity fall-off. However, because the analysis area was small (10 × 10 μm2), such intensity fall-off was difficult to fully eliminate during measurement. In principle, semi-quantitative analysis is feasible for ToF-SIMS analysis. Normally, a reference signal is required, and it can be used to reduce ‘matrix effect’. Unfortunately, in this system, it was not easy to find a reference signal. Oxygen was expected to share similar concentrations in both matrix and droplet phases, but the O+ signal was too low to be used. All other elements showed some preference in the two phases, and they also could not be used. Therefore, only a very rough estimation of the Li and Al concentration ratios between the two phases was obtained based on Table 2. A comparison of lateral resolution and concentration ratios of Li, B, and Al in the matrix phase over the droplet phase in the glass-ceramic sample using APT, ToF-SIMS, and NanoSIMS Lateral resolution Limatrix/Lidroplet Bmatrix/Bdroplet Almatrix/Aldroplet APT ToF-SIMS NanoSIMS ≤2 nm ~ 110– 130 nm 1.3 ± 0.2 1.5 ± 0.3 0.29 ± 0.06 ~150 nm 3.0 ± 0.4 1.7 ± 0.2 0.34 ± 0.04 2.2 ± 0.3 3.6 ± 0.6 0.57 ± 0.07 APT, atom probe tomography; ToF-SIMS, time-of-flight secondary ion mass spectrometry; NanoSIMS, nanoscale secondary ion mass spectrometry. signal intensity ratios between the two phases across line-scan analysis (Figure S2). As shown in Table 2, Limatrix/Lidroplet ≈ 1.3 and Almatrix/Aldroplet ≈ 0.3 were observed. It should be noted that the B+ counts were so low that the line-scan analysis did not provide any meaningful result. Lateral resolution is an important parameter for SIMS imaging. Several line-scan profiles across the interface between the matrix phase and droplet phase indicate the lateral resolution of ToF-SIMS was about 130 nm if using Al+ image (Figure S2). If a strong signal was used, e.g. Na+, the lateral resolution could be as good as ~100 nm (Figure S3). This value is very good because it is close to the specification value of the best lateral resolution of this ToF-SIMS instrument (~100 nm). BO2 and BO were very strong signals in ToF-SIMS negative ion spectra. So the negative ion mode may provide better B images. Figure 5 shows the images of 16O signal, the sum-counts image of 10B16O , 11B16O , and 10B16O2 signals, the normalized image of the sum-counts of 10B16O , 11B16O , and 10B16O2 signals over 16 O signal, and the 7Li image. The O and B images showed some edge fall-off, too, because of uneven surface charging. However, the normalized B image did not show any noticeable edge fall-off. More importantly, after normalization, semi-quantitative analysis (i.e. determination of relative B concentration between the droplet and matrix phases) was expected to be more reasonable. Figure S4 shows a line scan in the normalized B image, and data show that the concentration of B in the matrix phase was about 1.5 times the B concentration in the droplet phase, qualitatively consistent with the APT data. Also, a very good lateral resolution (110–120 nm) could be obtained based on line-scan data, which was similar with the lateral resolution values observed in ToF-SIMS positive ion images. Figure 5. Time-of-flight secondary ion mass spectrometry negative secondary ion images, showing distribution of B, O, and Li in the droplet-matrix phase 10 16 11 16 10 16 16 B O , B O , and B O2 ; (b) O ; (c) the image of separation region of the glass-ceramic sample. (a) Sum-counts image of 10 16 11 16 10 16 16 7 B O + B O + B O2 normalized to O ; and (d) Li . wileyonlinelibrary.com/journal/sia Copyright © 2016 John Wiley & Sons, Ltd. Surf. Interface Anal. (2016) Nanoscale imaging of Li and B in nuclear waste glass Nanoscale secondary ion mass spectrometry Nanoscale secondary ion mass spectrometry is another new imaging technique that was introduced for glass imaging during the last 3–4 years. Similar with ToF-SIMS, either negative secondary ions or positive secondary ions can be collected using NanoSIMS. The specifications of our NanoSIMS indicate that the best resolution of negative ion mode can be as low as 50 nm, much better than the 200 nm for positive ion mode; therefore, only negative ion mode was tested in this study. It should be noted that a new O source has been developed for high lateral resolution (e.g. 40 nm) positive secondary ion imaging using NanoSIMS[34]; however, so far, only prototype instrument is available. Compared with ToF-SIMS, NanoSIMS can provide high spatial resolution and high mass resolution simultaneously, so it potentially can provide ion images with lower background and better quantitative performance than ToF-SIMS. Figure 6 shows the ion images of 7Li , 11B16O , 27Al16O , and 18O . It is very clear that Li and B were enriched in the matrix phase, and the Al was enriched in the droplet phase. To obtain a semi-quantitative result, normalization was necessary. The 18O signal could serve as a good reference signal because oxygen atoms were expected to share similar concentrations in the two phases. After normalized to the 18 O signal, the Li and B still showed enrichment in the matrix phase, and the Al enrichment in the droplet phase looked more vivid. To obtain the lateral resolution and relative concentration of Li, B, and Al in these two phases, a few representative locations were selected for line-scan analysis. Figure S5 shows a typical line-scan result. The lateral resolution was found to be around 150 nm. This value was considerably poorer than the Cs+ beam diameter measured on a Si grid reference sample (~70 nm). This degradation of lateral resolution may be attributed to charging effects. Charging effects led to low signals and degraded image quality, even when low-energy electron charge compensation was applied. In addition, significant image distortion and shifting were observed during imaging collection, which can be attributed to uneven charging at sputter interface. The line-scan data show that the concentrations of B, Li, and Al in the matrix phase were about 2.2, 3.6, and 0.57 times those in the droplet phase, respectively (Table 2). The results of B and Li enrichment in the matrix phase and Al enrichment in the droplet phase are consistent with the results from APT and ToF-SIMS data. Figure 6. Nanoscale secondary ion mass spectrometry negative secondary ion images, showing distribution of Li, B, O, and Al in the droplet-matrix phase 7 11 16 27 16 18 7 18 11 16 separation region of the glass-ceramic sample. (a) Li ; (b) B O ; (c) Al O ; (d) O ; (e) the image of Li normalized to O ; (f) the image of B O 18 27 16 18 11 16 7 normalized to O ; (g) the image of Al O normalized to O ; (h) overlay of normalized B O and normalized Li ; and (i) overlay of normalized 11 16 27 16 B O and normalized Al O . Surf. Interface Anal. (2016) Copyright © 2016 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/sia Z. WANG ET AL. Corroded Advanced Fuel Cycle Initiative glass sample The corroded AFCI glass is a very interesting sample, due to a porous alteration layer observed in SEM images (Fig. 7). In this study, APT was used for analysis of elemental distributions in these corrosion layers. However, the highly porous sample broke easily during APT measurement when a high electric field was applied. As a comparison, the electric field in SIMS measurement was relatively gentle, and the electric field-induced sample breakage was rarely observed. Sample preparation is a key issue for imaging glass corrosion layers. Generally, a corroded glass sample is immobilized in a resin matrix, the sample is vertically cut by a diamond saw, and then the Figure 7. A top-view scanning electron microscope image on the surface of the corroded Advanced Fuel Cycle Initiative glass sample, showing the morphology of the focus ion beam-prepared wedge, in which the nanoscale secondary ion mass spectrometry imaging measurement was 2 performed (the red square area, 7 × 7 μm ). cross section is polished for imaging analysis. However, the best lateral resolution of SIMS imaging is about 50–100 nm. If the corrosion layers are thin (e.g. ≤100 nm), it may be difficult for SIMS imaging to distinguish between the corrosion layers. Therefore, we developed a ‘wedged crater’ strategy and observed a fivefold increase in depth information using SIMS imaging.[9] The details of the preparation of a wedged crater have been introduced in our previous publication, where the corrosion layers of a SON68 glass sample were imaged by NanoSIMS.[9] In this study, we used the same strategy to analyze the corrosion layers of an AFCI glass sample using NanoSIMS. Figure 8 shows the NanoSIMS ion images of 6Li + 7Li , 16O , 10 16 10 16 B O2 , (6Li + 7Li )/16O , B O2 /16O , and 6 7 10 16 6 7 ( Li + Li )/ B O2 . The images of Li + Li , 16O , and 10B16O2 show some signal increasing at the glass surface and may be due to the Cr layer deposited on the sample surface.[9] In addition, the 16 O image shows some low-intensity locations at the shallow region (region I in Fig. 8b), possibly due to topographic issues (porous structure). Therefore, to obtain semi-quantitative information of Li and B, normalization was performed. The 16O signal could serve as a good reference because oxygen atoms were expected to share similar atomic densities in different corrosion layers. After normalization, the most characteristic feature was an irregular distribution of B. Generally speaking, more B depletion occurred in the shallow region (region I in Fig. 8b), and less depletion occurred at the deep region (region II in Fig. 8b). However, some strong depletion occurred at the deep region (e.g. 2# location in Fig. 8e), and some weak depletion occurred at the shallow region (e.g. 1# location in Fig. 8e). Apparently, the measured corrosion depth was deeper than 1 μm, but some pristine glass could exist at the shallow region (much less than 1 μm). This situation made it very difficult to define the corrosion front line. Another interesting observation came from the ratio image of Li over B (Fig. 8f). This image shows that the Li behavior was different Figure 8. Nanoscale secondary ion mass spectrometry negative secondary ion images, showing distribution of Li, B, and O in the focus ion beam-prepared 6 7 16 10 16 wedge at the surface of the corroded Advanced Fuel Cycle Initiative glass sample. (a) Sum-counts image of Li and Li ; (b) O ; (c) B O ; (d) the image of 6 7 16 10 16 16 6 7 10 16 Li + Li normalized to O ; (e) the image of B O normalized to O ; and (f) the image of Li + Li normalized to B O . wileyonlinelibrary.com/journal/sia Copyright © 2016 John Wiley & Sons, Ltd. Surf. Interface Anal. (2016) Nanoscale imaging of Li and B in nuclear waste glass from the B behavior in the corrosion layers. The Li/B ratio in the Bdepletion region was higher than the Li/B ratio in the pristine glass region, indicating that B is easier to be dissolved into solution than Li. A comparison of atom probe tomography, time-of-flight secondary ion mass spectrometry, and nanoscale secondary ion mass spectrometry Time-of-flight secondary ion mass spectrometry, NanoSIMS, and APT can perform nanoscale imaging of Li and B in glass samples; however, they have their own merits as well as limitations (summarized in Table 3). APT provides very high lateral resolution (≤2 nm in this research) with an intrinsic 3-D mode analysis. As a comparison, lateral resolutions of both ToF-SIMS and NanoSIMS are no better than 100 nm, which is far inferior to that of APT analysis. It should be noted that SIMS can also perform 3-D imaging analysis with very decent depth resolution (1–2 nm is possible) if combined with depth profiling; however, the poor lateral resolution makes 3-D SIMS analysis less appealing than APT analysis. Quantification performance of APT seems better than SIMS. For example, our APT analysis can provide absolute concentrations of Li and B in the droplet and matrix phases: 1.8%/5.5% for Li and 2.2%/3.8% for B (all atomic percentage) in the droplet/matrix phases, which cannot be obtained by SIMS analysis. The Li concentration values were in a very reasonable range, because the nominal concentration value of Li atoms in the entire glass-ceramic system was about 3%. However, the B concentration values were much lower than the nominal value (~12% atomic percentage) in the system (partially due to the overlap of 11B16Ox and 27Al16Ox 1 peaks). Therefore, the accuracy of quantification using APT needed to be improved for B. Quantification has long been a challenge in SIMS analysis.[35] Absolute quantification is normally impossible unless a reference sample is available. In this study, only concentration ratios between droplet and matrix phases were obtained. Table 2 shows that the concentration ratios of Li in the matrix phase over the droplet phase were about 1.3 ± 0.2 (ToF-SIMS) and 2.2 ± 0.3 (NanoSIMS), respectively. If compared with the value of 3.0 ± 0.4 from APT analysis, we could see considerable difference between these values; however, the good news is that they all showed some Li enrichment in the matrix phase. A similar situation was found for B and Al. This observation indicates that all the three techniques in this study may be affected by matrix effect in some degree. Because APT analysis can provide absolute atomic concentrations and SIMS cannot, the ratio data from APT analysis may be more reliable. On the other hand, although the ratio values from SIMS analysis may not be as accurate as APT data, SIMS can give a quick estimation of relative enrichment of Li and B in different phases. The biggest drawback of APT is sample compatibility. Our study shows that APT measurement might be very difficult for some porous samples, such as the corroded AFCI sample. As a comparison, SIMS imaging was easily applied for this sample. In addition, it should be noted that the wavelength of the APT laser plays an important role for glass sample analysis. Our experience shows that, for some glass samples, laser absorption was so weak that APT analysis did not work (data not shown here). Obviously, the laser absorption issue does not exist in SIMS analysis. Therefore, SIMS is generally more sample-friendly than APT. Another drawback of APT is limited field of view. Although APT can provide very nice 3-D spatial resolution, the field of view is normally no larger than a size of 100 × 100 × 500 nm3. As a comparison, the field of view of SIMS imaging can vary from sub-μm to 500 × 500 μm2. In addition, one more advantage for SIMS imaging is that image stitching is available. Therefore, it has been very convenient to use SIMS to image many interesting structures with a size range of several microns to tens of microns on the glassceramic sample,[22] but it is a big challenge for APT. Generally speaking, time cost is not a major consideration if some desirable information is very important. However, if a large amount of samples need to be tested, time cost may be an issue for APT analysis. For example, a common method of sample preparation for APT analysis is FIB cutting and polishing, which requires an FIB instrument and is very time-consuming. As a comparison, sample preparation for SIMS analysis is relatively simple, which is similar to traditional SEM sample preparation. Also, although FIB was used in this study to prepare a wedged crater on an AFCI sample for NanoSIMS analysis, the time cost for preparation of a wedged crater (2–3 h) was much less than the time cost for preparation of a set of APT samples from one interesting location (typically 5–10 h). It has been observed that ToF-SIMS and NanoSIMS provide similar lateral resolution for Li and B nanoscale imaging on the glass-ceramic sample. NanoSIMS can simultaneously image Li and B in negative secondary ion mode with high mass resolution, and Table 3. A detailed comparison of advantages and disadvantages of ToF-SIMS, NanoSIMS, and APT for nanoscale imaging of Li and B in nuclear water glass samples Best spatial resolution Mass resolution Quantification Field of view Sample preparation Typical time cost for sample preparation Sample compatibility ToF-SIMS NanoSIMS ~110 nm (2-D) Unit mass ~150 nm (2-D) ≥6000 Need a reference sample 2 Up to 500 × 500 μm 2 (ToF-SIMS) and 200 × 200 μm (NanoSIMS), image stitching feasible Easy 3–5 h per sample Regularly available APT ≤2 nm (three-dimensional) ~800 Feasible 3 ≤100 × 100 × 500 nm Time-consuming, focus ion beam is required. 5–10 h per interesting location Lack of experience, sample-to-sample case APT, atom probe tomography; ToF-SIMS, time-of-flight secondary ion mass spectrometry; NanoSIMS, nanoscale secondary ion mass spectrometry; 2-D, two-dimensional. Surf. Interface Anal. (2016) Copyright © 2016 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/sia Z. WANG ET AL. semi-quantitative analysis is easily available. As a comparison, ToF-SIMS needs to perform both positive and negative polarities to obtain reasonable quality Li (positive) and B (negative) images. It should be noted that quantification result from ToF-SIMS positive ion mode is less reliable than negative ion mode. Nonetheless, ToF-SIMS has some advantages, too. For example, charging effects cause less trouble in ToF-SIMS measurement than that in NanoSIMS measurement, because NanoSIMS uses line-by-line scan mode during data collection, while random imaging mode can be applied in ToF-SIMS measurement. However, a more important advantage of ToF-SIMS may be that ToF-SIMS instrumentation is much more available than NanoSIMS. Therefore, ToF-SIMS may be a more convenient tool for Li and B imaging than NanoSIMS, while NanoSIMS can provide better data quality. [7] [8] [9] [10] [11] [12] [13] [14] [15] Conclusions [16] We report a comparative study of ToF-SIMS, NanoSIMS, and APT in nanoscale imaging of Li and B in glass and glass-ceramic samples. Our findings show that SIMS is more sample-friendly than APT, while spatial resolution and quantification performance of APT are better than SIMS. In addition, although both SIMS and APT can perform nanoscale Li and B imaging, their field of view is different. Therefore, the relationship between SIMS and APT is complementary and not competitive. On the one hand, it is very important to choose the right techniques based on the scales of scientific questions. Our suggestion is APT should be the first choice if the desirable field of view is smaller than 1 × 1 × 1 μm3; otherwise, SIMS may be a better choice. On the other hand, to better understand the structure of a glass sample, chemical images from various scales are all necessary. For example, SIMS can be used as a screening tool to examine a large area to find interesting locations for APT analysis. It is very reasonable to expect that the combination of SIMS and APT will become more and more important in glass research. [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] Acknowledgements The research was performed at the Environmental Molecular Sciences Laboratory, a national scientific user facility located at Pacific Northwest National Laboratory (PNNL), and sponsored by the Department of Energy’s (DOE) Office of Biological and Environmental Research. PNNL is operated for DOE by Battelle Memorial Institute under contract number DE-AC06-76RLO-1830. F. Y. W. and Z. Y. W. thank the National Natural Science Foundation of China (NSFC grant nos. 21127901, 21135006, and 21321003) for support. We also appreciate Linda H. Burk’s English editing. [27] [28] [29] [30] [31] [32] References [33] [1] W. E. L. M. I. Ojovan, New Developments in Glassy Nuclear Wasteforms, Nova, Science Publishers, New York, 2007. [2] I. W. Donald, Waste Immobilization in Glass and Ceramic Based Hosts, Wiley, Chippenham, UK, 2010. [3] A. Quintas, D. Caurant, O. Majérus, T. Charpentier, J.-L. Dussossoy. In International Conference Atalante 2008. Nuclear Fuel Cycles for a Sustainable Future: Montpellier, France 2009, p P4_07. [4] C. Cailleteau, F. Angeli, F. Devreux, S. Gin, J. Jestin, P. Jollivet, O. Spalla, Nat. Mater. 2008, 7, 978–983. [5] S. Gin, C. Guittonneau, N. Godon, D. Neff, D. Rebiscoul, M. Cabie, S. Mostefaoui, J. Phys. Chem. C 2011, 115, 18696–18706. [6] S. Gin, A. Abdelouas, L. J. Criscenti, W. L. Ebert, K. Ferrand, T. Geisler, wileyonlinelibrary.com/journal/sia [34] [35] M. T. Harrison, Y. Inagaki, S. Mitsui, K. T. Mueller, J. C. Marra, C. G. Pantano, E. M. Pierce, J. V. Ryan, J. M. Schofield, C. I. Steefel, J. D. Vienna, Mater. Today 2013, 16, 243–248. S. Utsunomiya, A. B. Kersting, R. C. Ewing, Environ. Sci. Technol. 2009, 43, 1293–1298. S. Gin, J. V. Ryan, D. K. Schreiber, J. Neeway, M. Cabie, Chem. Geol. 2013, 349, 99–109. Y. C. Wang, D. K. Schreiber, J. J. Neeway, S. Thevuthasan, J. E. Evans, J. V. Ryan, Z. H. Zhu, W. D. Wei, Surf. Interface Anal. 2014, 46, 233–237. R. Hellmann, S. Cotte, E. Cadel, S. Malladi, L. S. Karlsson, S. Lozano-Perez, M. Cabie, A. Seyeux, Nat. Mater. 2015, 14, 307–311. S. Fearn, D. S. McPhail, B. Hagenhoff, E. Tallarek, Appl. Surf. Sci. 2006, 252, 7136–7139. T. Chave, P. Frugier, A. Ayral, S. Gin, J. Nucl. Mater. 2007, 362, 466–473. P. Jollivet, F. Angeli, C. Cailleteau, F. Devreux, P. Frugier, S. Gin, J. Non-Cryst. Solids 2008, 354, 4952–4958. M. Melcher, R. Wiesinger, M. Schreiner, Acc. Chem. Res. 2010, 43, 916–926. C. Cailleteau, F. Devreux, O. Spalla, F. Angeli, S. Gin, J. Phys. Chem. C 2011, 115, 5846–5855. Z. Y. Wang, K. Jin, Y. W. Zhang, F. Y. Wang, Z. H. Zhu, Surf. Interface Anal. 2014, 46, 257–260. J. J. Neeway, S. Kerisit, S. Gin, Z. Y. Wang, Z. H. Zhu, J. V. Ryan, J. Non-Cryst. Solids 2014, 405, 83–90. S. Gin, P. Jollivet, M. Fournier, F. Angeli, P. Frugier, T. Charpentier, Nat. Commun. 2015, 6, 6360. Z. Y. Wang, B. W. Liu, E. W. Zhao, K. Jin, Y. G. Du, J. J. Neeway, J. V. Ryan, D. H. Hu, K. H. L. Zhang, M. N. Hong, S. Le Guernic, S. Thevuthasan, F. Y. Wang, Z. H. Zhu, J. Am. Soc. Mass Spectrom. 2015, 26, 1283–1290. P. Collins, J. Coumbaros, G. Horsley, B. Lynch, K. P. Kirkbride, W. Skinner, J. Forensic Sci. 2003, 48, 538–553. S. Gin, P. Jollivet, M. Fournier, C. Berthon, Z. Y. Wang, A. Mitroshkov, Z. H. Zhu, J. V. Ryan, Geochim. Cosmochim. Acta 2015, 151, 68–85. J. Crum, V. Maio, J. McCloy, C. Scott, B. Riley, B. Benefiel, J. Vienna, K. Archibald, C. Rodriguez, V. Rutledge, Z. H. Zhu, J. Ryan, M. Olszta, J. Nucl. Mater. 2014, 444, 481–492. L. Q. Mai, X. C. Tian, X. Xu, L. Chang, L. Xu, Chem. Rev. 2014, 114, 11828–11862. J. M. Zheng, M. Gu, A. Genc, J. Xiao, P. H. Xu, X. L. Chen, Z. H. Zhu, W. B. Zhao, L. Pullan, C. M. Wang, J. G. Zhang, Nano Lett. 2014, 14, 2628–2635. D. Golberg, Y. Bando, Y. Huang, T. Terao, M. Mitome, C. C. Tang, C. Y. Zhi, ACS Nano 2010, 4, 2979–2993. K. Thompson, D. Lawrence, D. J. Larson, J. D. Olson, T. F. Kelly, B. Gorman, Ultramicroscopy 2007, 107, 131–139. J. V. Crum, A. L. Billings, J. Lang, J. C. Marra, C. Rodriguez, J. V. Ryan, J. D. Vienna, Baseline Glass Development for Combined Fission Products Waste Streams, Pacific Northwest National Laboratory, Richland, WA, 2009. T. F. Kelly, D. J. Larson, Annu. Rev. Mater. Res. 2012, 42, 1–31. S. V. N. T. Kuchibhatla, V. Shutthanandan, T. J. Prosa, P. Adusumilli, B. Arey, A. Buxbaum, Y. C. Wang, T. Tessner, R. Ulfig, C. M. Wang, S. Thevuthasan, Nanotechnology 2012, 23, 215704. O. C. Hellman, J. A. Vandenbroucke, J. Rusing, D. Isheim, D. N. Seidman, Microsc. Microanal. 2000, 6, 437–444. M. K. Miller, Atom Probe Tomography: Analysis at the Atomic Level, Kluwer Acedemic/Plenum Publishers, New York, 2000. B. Gault, M. P. Moody, J. M. Cairney, S. P. Ringer, Atom Probe Microscopy, 160, Springer, London, 2012. O. C. Hellman, J. A. Vandenbroucke, J. Rüsing, D. Isheim, D. N. Seidman, Microsc. Microanal. 2000, 6, 437–444. Oregon Physics, LLC, http://www.oregon-physics.com/hyperion2.php, access date: April 22, 2016 T. Wirtz, P. Philipp, J. N. Audinot, D. Dowsett, S. Eswara, Nanotechnology 2015, 26, 434001. Supporting information Additional supporting information may be found in the online version of this article at the publisher’s web site. Copyright © 2016 John Wiley & Sons, Ltd. Surf. Interface Anal. (2016)
© Copyright 2026 Paperzz