Ultramicroscopy 144 (2014) 50–57

Contents lists available at ScienceDirect

Ultramicroscopy journal homepage: www.elsevier.com/locate/ultramic

Nano-dot markers for electron tomography formed by electron beam-induced deposition: Nanoparticle agglomerates application Misa Hayashida a,n, Marek Malac b,c, Michael Bergen b, Peng Li b a National Metrology Institute of Japan, National Institute of Advanced Industrial Science and Technology (AIST), 1-1-1, Higashi, Tsukuba, Ibaraki 305-8565, Japan b National Institute of Nanotechnology, 11421 Saskatchewan Drive, Edmonton, Canada c Department of Physics, University of Alberta, Edmonton, Canada T6G 2E1

art ic l e i nf o

a b s t r a c t

Article history: Received 23 October 2013 Received in revised form 4 February 2014 Accepted 18 April 2014 Available online 26 April 2014

A method allowing fabrication of nano-dot markers for electron tomography was developed using an electron beam-induced deposition in an ordinary dual beam instrument (FIB and SEM) or an SEM. The electron beam deposited nano-dot markers are suitable for automatic alignment of tomographic series. The accuracy of the alignment was evaluated and the method was demonstrated on agglomerated nanoparticle samples using a rod-shaped sample with no missing wedge effect. Simulations were used to assess the effect of marker size on alignment accuracy. & 2014 Published by Elsevier B.V.

Keywords: Electron tomography Nano-dot marker Fiducial marker Electron beam induced deposition FIB sample preparation Nanoparticle agglomerate

1. Introduction Electron tomography is a method employed in a transmission electron microscope (TEM) to reconstruct a three-dimensional (3D) volume from a series of images acquired at suitable tilt increments [1]. An accurate, preferably fully automated alignment of the individual images in the series is critical to obtain good quality 3D reconstruction of the sample. Fiducial markers, usually gold nanoparticles placed from a suspension at random positions within a sample, are often used for tomography of biological samples. For precise alignment, markers must be dispersed suitably over the region of the specimen that is to be observed while not interfering with the observed objects. However, it is difficult to obtain even dispersion because the colloidal gold nano-particles are usually dense materials [2,3]. If dense colloidal gold nano-particles are close to or within the area of interest of the specimen, they introduce artifacts into the reconstructed 3D images. In the case that no gold nano-particles actually exist, precise alignment cannot be carried out. An alternative method for alignment uses landmarks in the sample itself for cross correlation allowing to obtain lateral shift of

n Corresponding author at: Central 5, Higashi 1-1-1, Tsukuba, Ibaraki 305-8565, Japan. Tel.: +81 29 861 4171. E-mail address: [email protected] (M. Hayashida).

http://dx.doi.org/10.1016/j.ultramic.2014.04.005 0304-3991/& 2014 Published by Elsevier B.V.

images, but the results can be operator dependent. The presented e-beam fabricated fiducial marker method allows to place the markers at desired locations near the region of interest. The method makes it possible to obtain, in addition to lateral shift, image rotation and tilt and azimuth angles. The applications shown below make use of a rod-shaped specimens prepared by focused ion beam (FIB) instrument allow obtain tomograms without missing-wedge [4]. The rod-shaped specimens are typically used in materials science where high resolution, resulting in small field of view, is usually required making it difficult to disperse colloidal gold nano-particles near the observing area while not interfering with the objects of interest. Very accurate alignment is desirable for such samples. The e-beam fabricated marker method can be applied to radiation sensitive, polymer and biological samples, since the area of interest is not irradiated during the fiducial marker fabrication step. In a previous study, we used a helium-ion microscope (HIM) equipped with a tungsten carbonyl (W(CO)6) gas injection system (GIS) to form tungsten nanodots [5,6]. Moreover, we demonstrated for the first time, the use of nano-dot markers with  10 nm size on a  100 nm-diameter rod-shaped specimen for aligning a TEM tomographic tilt series. To make the method accessible to a broad research community the previously demonstrated nano-dot marker fabrication in a HIM must be transferred to instruments that are more wide spread than HIM. In this paper we report

M. Hayashida et al. / Ultramicroscopy 144 (2014) 50–57

51

fabrication of electron beam-fabricated tungsten nano-dots for automated alignment of electron tomographic data. We refer to the dots as “nano-dot markers”. The formation of electron beam induced deposition of tungsten nano-dot markers was achieved using a standard scanning electron microscope (SEM) equipped with a tungsten carbonyl (W(CO)6) gas injection system (GIS). We first describe the experimental set up, then discuss the fabrication of the nano-dot markers and demonstrate the application of the method to a sample of agglomerated TiO2 nanoparticles and a sample of a regular array of silver nanoparticles with sub-5 nm diameter. Both the annular dark field scanning TEM (ADF STEM) and the bright field TEM (BFTEM) operation modes for tilt series acquisition were tested for suitability of automated nano-dot markers detection and alignment. Moreover, we investigate the effect of markers' shape on alignment accuracy using simulated tilt series with varied marker sizes.

2. Instrumentation A Hitachi NB 5000 dual beam (FIB and (SEM)) instrument was used for sample preparation and for electron beam-induced deposition of nano-dot markers. For fabrication of the rodshaped specimen we used 40 keV Ga ion beam. For deposition of the nano-dot markers we used 5 keV electron beam. The angle between SEM column and FIB column is 581. Hitachi HF 3300, a 300 kV transmission electron microscope (TEM)/scanning TEM (STEM) equipped with a cold field emission electron gun (CFEG) was used for collecting tomographic tilt series. The data collection was assisted by the use of a Matlab™-based computer control system controlling the microscope and associated hardware [7]. The same micropillar holder was used for sample preparation in the NB 5000 dual beam instrument and HF 3300 TEM/STEM for data acquisition without the need for remounting a sample [8].

3. Fabrication and properties of e-beam deposited nano-dot markers 3.1. Experimental conditions Initially, the electron beam-induced deposition of tungsten from W(CO)6 was investigated using about 20 nm thick amorphous carbon substrate. The substrate was prepared by electronbeam evaporation of carbon onto a mica substrate. The carbon film was then floated on deionized water surface and picked onto a standard 200-mesh copper grid. For high reproducibility of the fabrication, W(CO)6 gas pressure and exposure time of electron beam was controlled. The timing diagram of the nano dot marker deposition is shown in Fig. 1. First, a W(CO)6 precursor gas source was opened for a 10 s period (Fig. 1a) and the fabrication of the first marker in each row commenced within delay time of 0.5 s after the W(CO)6 precursor gas supply was closed. While the gas source was opened the electron beam gun valve was closed in order to protect electron beam gun (Fig. 1b), it was opened after closing the gas source. Then, 10 markers were deposited subsequently (Fig. 1c). The W (CO)6 precursor partial pressure decreased as schematically shown in Fig. 1d. Bright field STEM images acquired in the NB 5000 of the nano dots deposited with zero delay time between depositions at individual positions are shown in Fig. 2a (the first to last nano dots are shown from left to right). Each row was fabricated with the same dwell time per nano dot (5 s, 3 s, 2 s, 10 s and 20 s) indicated near the first marker of each row and shown in Fig. 2a. The measured electron beam current was about 0.5 nA. The dwell time per individual nano dot was between 2 s and 20 s corresponding

Fig. 1. Timing diagram of procedure for deposition of nano-dot markers. The x-axis is time from the initial opening of the gas precursor source while the y axis represents the status of various experimental variables. (a) Delivery of the W(CO)6 gas precursor source. (b) Opening of the SEM column gun valve. (c) Exposure of the sample by the incident electron beam leading to deposition of nano dots. (d) W(CO)6 gas precursor partial pressure in the chamber.

to exposure dose between 1 nC and 10 nC per nano dot. Similar experiments were performed to explore the effect of the precursor delivery time where the dwell time was kept constant and the opening time of the gas source was changed from 10 s to 30 s. The thickness of the uniform carbon film used in this experiment was estimated to be about 20 nm. The image in Fig. 2b acquired at 71 tilt, i.e. 511 relative to the SEM column of the NB 5000, shows that WCX was deposited on both top and bottom surfaces of the amorphous carbon film. This is of importance when electron beam deposited markers are used on samples dispersed on an amorphous carbon support. Due to beam broadening in the amorphous carbon, the WCX deposit was broader on bottom (exit) surface of the carbon film. The results are further discussed in Section 3.2. 3.2. Results and discussion It is known that the electron beam-induced deposition is capable of fabricating extremely small dots. Sizes down to 2 nm were demonstrated [9–11] when using organometallic precursors and 200 kV ultra high vacuum TEM. Similar results were demonstrated using W(CO)6 precursor [12]. In our experiments we opted for the W(CO)6 precursor leading to deposition of mixture of tungsten and carbon, WCX [12]. The nano dot deposition is thought to be due to secondary electron induced decomposition of the precursor landing on the surface [12]. The smallest achievable size of the dots is thought to be limited by the secondary electrons emitted from the side of the growing nano dots leading to deposition of material on the side walls of the growing nano dots [13,14]. The size of the nano dots is also expected to be dependent on the amount of the precursor available for deposition and on the current and time of the electron beam exposure of the individual nano dots [8]. The partial pressure of the precursor can be

52

M. Hayashida et al. / Ultramicroscopy 144 (2014) 50–57

Fig. 2. Nano-dot markers fabricated in SEM with W(CO)6 gas injection system. Bright field STEM images obtained at 30 keV incident electron energy of the SEM column in the Hitachi NB 5000 dual beam instrument. (a) The sample was perpendicular to the incident electron beam of the SEM column. Each row was fabricated with the constant dwell time per nano dot (5 s, 3 s, 2 s, 10 s and 20 s) as indicated near the first marker of each row. The gas precursors supply was opened for 10 s prior to fabrication of each of the rows. (b) Higher magnification of section from (a) acquired at 391 tilt with respect to the SEM column. (c) Evaluation of mass thickness of the nano dots with dwell time. Lines are added in each figure to guide an eye.

conveniently controlled by the total time that the precursor was allowed in the microscope chamber and/or the time delay between when the precursor delivery was stopped and the nano dot marker writing started, as shown in Fig. 1. The first parameter affects the total amount of precursor allowed in the chamber, while the second one affects the amount of the precursor residual in the chamber (i.e. not pumped out) by the time the nano dot marker is being written. The nano dot marker fabrication therefore requires the presence of the precursor in gaseous form at suitable partial pressure. Since the ultimate vacuum of our NB 5000 dual beam instrument is in the order of 1  10  6 Torr and the desired partial pressure of the W(CO)6 precursor is of the same order of magnitude, it is more convenient to rely on the reproducibility of the procedure than to attempt to actively adjust the precursor partial pressure. In analogy with bright field TEM [14], the BF STEM image in Fig. 2 allows us to estimate the relative mass thickness of the individual nano dots when the background arising from uniformly thick amorphous carbon is accounted for. The decrease of intensity arising from presence of a nano dot marker was obtained by integrating an area about 80  80 nm2 of the BF STEM image in Fig. 2a. The natural logarithm of the intensity within the nano dot marker area is proportional to the local mass thickness [15,16]. Since the beam intensity in vacuum was not measured, only relative comparison of mass thickness among individual nano dots can be made using the relation: Log ðI carbon =I marker Þ ¼ ρt here Icarbon and Imarker are the image intensities obtained in the uniform carbon and nano dot marker areas respectively, ρt is mass thickness. Fig. 2c shows that there is about three-fold increase in mass thickness of the first two or three nano-dot markers in each row with increase of the dwell time per nano dot marker from 2 s to 20 s. It also shows that there is a decrease of nano dot marker mass thickness with time from the first nano dot marker deposition to the last one in each series. The decrease of the nano dot marker mass thickness from left to right in each row is an indication of

availability of the precursor as function of time, i.e. the rate at which the precursor is pumped out. Based on the mass thickness of the first three nano dots in each row (Fig. 2c), the increase in mass thickness is very large when the precursor exposure is raised from 10 s to 20 s, but much smaller when the precursor exposure was further increased from 20 s to 30 s. This indicates that 30 s precursor exposure time may be nearing the precursor partial pressure steady state near the sample [17]. The mass thickness of nano dots past the first three dots in each row (nano dots belonging to the same row are connected by a line in Fig. 2c) varies less predictably than the first three dots. This can arise from both the precision of our rudimentary mass thickness measurement method in an instrument that is not calibrated for mass thickness measurements [14] or from contribution of additional parameters such as availability of carboneous contamination on the surface of the amorphous carbon film. The decrease of the precursor availability with time from the closing of the precursor supply (Fig. 2c) indicates that the W(CO)6 precursor is available near the sample for several tens of seconds. However, the fact that we were able to obtain BF STEM images in the NB 5000 without significant contamination, such as shown in Fig. 2a, indicates that the precursor does not remain adsorbed on the sample surface, but its availability is determined by the amount of the precursor impinging on the substrate from gaseous phase. This is in agreement with observations reported in [12].

4. Application to agglomerated nanoparticle samples 4.1. Substrate preparation Two samples were prepared for electron tomographic experiments: a sample composed of TiO2 nanoparticles with lateral dimensions  50 nm and a sample of a few-layer thick ordered array of silver nanoparticles with sub-5 nm diameter. The procedure of the preparation of TiO2 nanoparticles is shown in Fig. 3. The TiO2 nanoparticle sample was prepared by sonicating the TiO2 nanoparticle powder in deionized water. A drop of the resulting

M. Hayashida et al. / Ultramicroscopy 144 (2014) 50–57

53

Fig. 3. Schematic procedure of TiO2 nanoparticle sample preparation. (a) A drop of deionized water with TiO2 powder is placed on a Si wafer. (b) Drying of (a). (c) E-beam induced deposition of 1 mm thick amorphous carbon layer in the Hitachi NB 5000 dual beam instrument. (d) Rod-shaped specimen is fabricated from (c). (e) Deposition of 6 nano-dot markers at 01 tilt. (f) Deposition of additional 6 nano-dot markers after 1351 tilt.

suspension was placed onto a Si wafer (Fig. 3a) and allowed to dry in laboratory air overnight (Fig. 3b). We adjusted the concentration of the TiO2 nanoparticles in the deionized water to obtain a uniform thickness layer of agglomerated particles on the Si wafer. A rod-shaped specimen can be then fabricated from any location across the sample. If the nanoparticle concentration is too low it is difficult to ensure that the sufficient number of TiO2 nanoparticles are contained in a 300 nm diameter rod-shaped specimen fabricated by a FIB. The sample of regular array of Ag nanoparticles was prepared in a similar manner. A suspension of Ag nanoparticles in deionized water was deposited onto a Si substrate that was previously oxygen-terminated by dipping the Si wafer in O3 solution after cleaning the surface of the substrate in an HF solution. The use of oxygen-terminated Si wafer is essential to obtain an ordered array of nanoparticles that is only a few layers thick [18].

the initial cube-shaped sample in the Hitachi NB 5000 dual beam instrument (Fig. 3d). The diameters of TiO2 and Ag nanoparticles rod-shaped specimens were about 300 nm and 220 nm, respectively. Nano-dot markers were then fabricated using electron beam induced-deposition of tungsten from W(CO)6 precursor (Fig. 3e). Each marker was fabricated as the following steps. First, gas precursor source was opened for a 10 s period. Immediately after the gas source was closed, electron beam gun was opened and position of electron beam was set on the specimen fabricating an individual nano-dot marker. The procedure was repeated with electron beam positioned at 30–40 nm intervals along the length of the carbon section of the micropillar, as seen in Fig. 4a. After depositing seven nano-dot markers without rotating the sample along the tilt axis, the sample was rotated 1351 and a second series of seven nano-dot markers was deposited (Fig. 3f) resulting in 14 nano-dot markers in total by using 30 keV electron beam. Typical time for the 14 nano dot marker fabrication was about 30 min.

4.2. Tomography sample preparation and nano-dot marker fabrication

4.3. Acquisition of tilt series

The prepared Si wafer substrates with TiO2 or Ag nanoparticles were transferred into a Hitachi NB 5000 dual beam instrument on an SEM stub. A 1 mm thick amorphous carbon layer was deposited over 1 um  1 um area of the each sample using electron beam induced carbon deposition from acetylacetone precursor (Fig. 3c). A rhombohedral-shaped specimen about 1 um  1 um  10 um (Width  Height  Length) with the amorphous carbon layer on top was trimmed from the bulk specimen by the FIB micro sampling technique [19] and was transferred onto a Hitachi 3D holder that allows for continuous rotation over the full 3601 range [8]. A rod-shaped specimen was fabricated by further FIB milling of

A densely-packed nanoparticle (NP) sample is difficult to image [20] thus providing a challenging sample for testing of new electron tomography methods. Small nanoparticle samples are also of practical interest in various application fields, such as catalysis. STEM was used for the TiO2 NPs and TEM was used for the Ag NPs to demonstrate suitability of the nano-dot markers for each method. The TiO2 tilt series was acquired in annular dark field (ADF) STEM mode with nominal probe size o1 nm and irradiation dose approximately 32 e  /Å2 per tilt (the pixel dwell time 10 ms and image 512  1024 pixels at 1.04 nm/pixel). The BFTEM images of Ag NPs were collected at 40 kx nominal magnification. A 3 s

54

M. Hayashida et al. / Ultramicroscopy 144 (2014) 50–57

acquisition per image corresponds irradiation dose approximately 48 e  /Å2 for the entire series. The image size was 1024  1024 pixels, with pixel size 0.45 nm/pixel corresponding to only 10 pixels per Ag nanoparticle diameter. No objective aperture was used resulting in collection semi-angle 108 mrad and scattering contrast with reasonably following linear increase in contrast with mass thickness [14]. The tilt step of both tilt series was 31 over 01 to 1801 tilt range resulting in 61 images.

4.4. Alignment of the images and tomographic reconstruction The alignment was performed using the nano-dot markers shown in Figs. 4a and 5a. First, the positions were automatically detected by using a cross-correlation with a template image [6]. After that, unknown parameters (translational X, Y values, rotational angle in X–Y plane, tilt angle and azimuth angle) were calculated by using fiducial-marker method [2]. In the fiducialmarker method, a marker position on the sample is estimated by minimizing the difference between markers' positions projected onto an image over the entire tilt range. The error in marker position at every tilt is then estimated as the difference between the expected projected position of the marker and the position observed in an experimental image. The total error in marker position over the entire tilt range is taken as the minimum rootmean-square (RMS) error between the expected and measured marker position in the projected images. The RMS error in the final positions of the markers from all images in tilt series of TiO2 nano-particles and Ag nano-particles, calculated using the fiducial-marker method, were 1.6 pixels (1.6 nm) and 1.9 pixels (0.86 nm) respectively. The reconstruction from the aligned images was performed using standard filtered back projection method (FBP). The alignment and reconstruction, was done on 512  1024 pixels and 1024  1024 pixel images, respectively.

4.5. Results and discussion The TiO2 NP sample, shown in Fig. 4a exhibits three distinct sections (starting from the top): amorphous carbon rod with nanodot markers, the TiO2 nanoparticle section and the supporting Si wafer. The diameter of the rod is about 300 nm near the Si/TiO2 NP interface. Three additional nano-dot markers were deposited on the Si wafer section of the sample to assist location of the sample during the data collection step. The reconstruction in Fig. 4 (b–f) shows a section through the reconstructed volume in a plane that includes the long axis (tilt axis) of the rod sample and a plane perpendicular to the long axis of the sample rod (the tilt axis) respectively. In both cases, the TiO2 nanoparticles are clearly resolved. TiO2 nanoparticles are shown in Fig. 4b, e and f. The cross-sectional image of the nano-dot markers is shown in Fig. 4c and d. While the TiO2 particles can be clearly distinguished in areas where they are not touching, the interface in areas where nanoparticles are touching is hard to detect and the NPs may be fused together. It is not clear that the NPs are merely stuck together during the drying process or during the FIB fabrication. Since the NPs fusing appears to be uniform throughout the cross section of the sample (Fig. 4b, e and f) the fusing is unlikely to be from the FIB fabrication process. If the FIB fabrication was responsible, the fusing of the NPs would likely be more pronounced at the perimeter of the 300 nm diameter rod-shaped sample where the FIB beam was incident. The electron beam induced damage [20] during the tilt series acquisition is also an unlikely cause since there is no observable difference between the first and last image of the tilt series. Fig. 4(c and d) shows reconstruction of the nano-dot markers (marked by a circle). It appears that the marker has internal structure: a high contrast (bright) shell-like area with a lower contrast core is clearly visible in Fig. 5c. We attribute this core– shell structure of the nano-dot markers to presence of carbouneous contamination on the rod-shaped sample during nano-dot

Fig. 4. Electron tomographic observation of TiO2 nano-particles on Si subsrate. (a) An experimental projected image. The regions are (starting from the top): amorphous carbon with nano-dot markers, TiO2 nanoparticles, Si wafer substrate. The images were acquired in ADF STEM mode and the scale bar is 200 nm. The cross sections for (c)– (f) are marked in the image. (b) A cross section from the reconstructed volume in a plane that includes the tilt axis. (c), (d) a cross section of reconstructed volume of nanodot markers in a plane perpendicular to the tilt axis. The plane of the cross section is marked in (a). (e), (f) a cross section of reconstructed volume of TiO2 nano-particles in a plane perpendicular to the tilt axis. The plane of the cross section is marked in (a). To be modified in Fig: is it worth marking the areas of cross section by letters corresponding to figure c–f rather numbers.

M. Hayashida et al. / Ultramicroscopy 144 (2014) 50–57

55

Fig. 5. Electron tomographic observation of Ag nano-particles on Si substrate. (a) Bright field TEM image of the entire specimen. The regions are (starting from the top): amorphous carbon with nano-dot markers, ordered array of Ag nanoparticles, Si wafer substrate. (b) A cross section of reconstructed volume of nano-dot markers in a plane perpendicular to the tilt axis. The plane of the cross section is marked in (a). (d), (e) a cross section of reconstructed volume of Ag nanoparticles in a plane perpendicular to the tilt axis. The plane of the cross section is marked in (a).

markers fabrication in the Hitachi NB 5000 instrument. It appears that during the initial period of the 10 s time of the individual nano dot fabrication the contamination was highly mobile and depositing within the electron beam irradiated area [14]. Later, the deposition of the W from the precursor was faster than the flow and cross linking of the carboneous contamination. This is consistent with the presence of carboneous contamination during the onset of nano dot marker deposition. Both the experimental projected images and the reconstructed volume show that WCX was deposited also on the exit side of the rod-shaped sample when the nano-dot markers were written. The deposits on the exit surface, marked by an ellipse in Fig. 4c and d, are fairly uniform and thin and they are not posing problems with either alignment or reconstruction. Fig. 5 shows the second application example of the nano-dot markers to a difficult test sample. As in the case of the TiO2 sample, the cross section of the sample shown in Fig. 5b–e exhibits sharp boundaries between the sample edge and vacuum: a witness to accurate alignment of the images composing the series being reconstructed. Not only the sample boundary appears sharp, but the individual sub-5 nm diameter Ag nanoparticles are clearly visible in a cross section of the reconstructed volume shown in Fig. 5d and e. The results show the alignment error of 0.86 nm (1.9 pixels) by using markers with about 15 nm diameter on rodshaped specimen with 220 nm diameter a visualization of sub5 nm nanoparticles (about 10 pixels) can be reliably achieved. The method can be applied to wide variety of samples, providing a suitable, low contrast area is available for marker deposition. In case of material science samples reported here an amorphous carbon layer is deposited in the dual beam instrument and the fiducial markers are deposited onto the amorphous carbon area. When samples are deposited onto thin flat carbon films from a suspension or typical biological, low atomic number samples, there is usually plenty of suitable area for marker deposition near an area of interest. The method can be also applied to radiation sensitive samples because the area of interest is not irradiated during the marker fabrication step. However, samples that need to be held at cryogenic temperature pose a challenge as they are likely to be contaminated by the W(CO)6 precursor.

5. The effect of marker shape on alignment accuracy Ideally the shape of a fiducial marker should be spherical, having the same projected appearance in all projections for easy detection using cross correlation with a suitable radially-symmetric template. As seen in Figs. 4 and 5 the nano-dot markers have a cross section close to a parabola. For automatic detection of positions of markers, we use a radially symmetric template that is shown in [6]. The nonsymmetric projected shape in the x axis direction deteriorates the accuracy of detection in the x-axis direction. Fig. 6a shows the error in the direction long the x-axis detection of the marker marked by a circle in Fig. 6a. It was calculated using the method discussed above in Section 4.4 and in [6]. The error, is typically below 2 pixels – but reaches as much as  5 pixels when the marker long axis is perpendicular to the incident beam (i.e. the marker is “on the side” of the rod-shaped sample). This is a likely consequence of the projection of marker that not radially symmetric but parabolic projected shape. The position detection along the long axis of the marker becomes inaccurate when the marker is projected with its long axis perpendicular to the incident electron beam. Simulations were performed to gain additional insight in the effect of marker shape. Seven tilt series with different size of parabola-shaped markers as shown in Fig. 6b were simulated. The simulated image size in each tilt series was 1024  1024 pixels, angular step in each tilt series was 31 over 1801 range to match our experimental data. Template identical to the one used for experimental data processing was used to detect the marker positions. To account for the worst case scenario, the error in detection was evaluated for markers with their height (b direction in Fig. 6 b) perpendicular to the projection direction. The projected marker shape for this situation is shown in Fig. 6c together with the template image used for detecting the marker. A circle with diameter of 36, 36, 36, 36, 18, 12 and 9 pixels were used as a template to detect the each marker for the respective tilt series 1–7. A selected image from the simulated tilt series is shown in Fig. 6d. The rod-shaped specimen had 520 pixels diameter and density about 1/10 of the markers. For each simulated tilt series 12 markers (marked by a circle in Fig. 6d) were detected and used for alignment. All 12 markers in each series had identical parameters

56

M. Hayashida et al. / Ultramicroscopy 144 (2014) 50–57

Fig. 6. Evaluation of accuracy of alignment using nano-dot markers with parabolic shape. (a) The error of the detected position along the x-axis in Fig. 6a of the marker marked by a circle in Fig. 6a. The error is about 2 pixels over most of the tilt range, but increase to about 5 pixels when the long, b, axis of the marker is near perpendicular to the projection (i.e. electron beam) direction. (b) Schematic drawing of cross-sectional shape of the nano-dot markers. (c) Table of marker size in pixels for seven simulated tilt series. (d) An example of simulated image. The size is 1024  1024 pixels. Both the markers and an area with simulated nanoparticles, aligned horizontally, is visible. The density of markers is Xx higher than that of the rod-shaped sample. (e) The RMS error in each of the simulated tilt series with various marker sizes.

a and b. Processing identical with the processing of the experimental data for marker detection and alignment was used. RMS error obtained for each tilt series is shown in Fig. 6e. The RMS error decreases with decreasing of height b of the marker. When both of diameter a and height b of markers decrease, the RMS error also decreases. However, the contrast of the marker visible in a projected image also decreases with decreasing marker size. Consequently, it becomes difficult to detect small markers especially when the marker is aligned with its long axis b parallel with the incident electron beam (i.e. with the direction of the projection). It was difficult to detect markers of series #7 (a ¼b¼ 9 pixels) resulting in very high RMS error for the series #7 (error was not displayed in Fig. 6e)). The Fig. 6e also shows that the detection accuracy, as measured by RMS error, is not sensitive to marker shape: the series #2 to #6 result in approximately same, acceptably low, RMS error of marker detection. Should further improvement of alignment accuracy be needed, the shape of the template image must better reflect the shape of a marker and the rotational angle around the tilt axis of specimen should be fit to the each marker in each image. It should be noted however that the current accuracy of automatic detection is sufficient for most practical purposes.

6. Conclusion We have demonstrated the feasibility of using standard SEM beam for fabricating WCX nano-dot markers for electron tomography. We have shown the application of the method on reconstruction and visualization of sample composed of agglomerated TiO2 nanoparticles. This method is expected to be applied to

analyze the 3D shape of primary nano-particles that are easily condensed. Moreover, a regular array of closely packed sub 5 nm Ag nanoparticles, a sample that is notoriously difficult to analyze, could be observed. The results showed ability of alignment using the nano-dot markers. The optimum shape of the markers was investigated by means of computer simulations. The simulations suggest that decreasing the marker size leads to increased accuracy (decreased RMS error) of alignment providing the markers provide sufficient contrast to be detected when projected together with a finite thickness rod-shaped sample.

Acknowledgments We are indebted to Dr. Takashi Nakamura (AIST) for Ag NP sample preparation and Martin Kupsta for support and for assistance on FIB.

References [1] J. Frank, Principles of electron tomography, in: J. Frank (Ed.), Electron Tomography: Three-dimensional Imaging with the Transmission Electron Microscope, Plenum Press, New York, London, 1992, pp. 1–13. [2] M.C. Lawrence, Least-squares methods of alignment using markers, in: J. Frank (Ed.), Electron Tomography, Plenum Press, New York, 1992, pp. 197–204. [3] D.N. Mastronarde, Fiducial marker and hybrid alignment methods for singleand double-axis tomography, in: J. Frank (Ed.), Electron Tomography: Methods for Three-Dimensional Visualization of Structures in the Cell, Springer, New York, 2006, pp. 163–185. [4] N. Kawase, M. Kato, H. Nishioka, H. Jinnai, Transmission electron microtomography without the missing wedge for quantitative structural analysis, Ultramicroscopy 107 (2007) 8–15.

M. Hayashida et al. / Ultramicroscopy 144 (2014) 50–57

[5] M. Hayashida, T. Iijima, T. Fujimoto, S. Ogawa, Position-controlled marker formation by helium ion microscope for aligning a TEM tomographic tilt series, Micron 43 (2012) 992–995. [6] M. Hayashida, T. Iijima, M. Tsukahara, S. Ogawa, High-precision alignment of electron tomography tilt series using markers formed in helium-ion microscope, Micron 50 (2013) 29–34. [7] M. Bergen, R.A. McLeod, M. Malac, D. Hoyle, Y. Taniguchi, T. Yaguchi, J. Chen, T. Yotsuji, Centralized instrument control for a TEM laboratory, Microsc. Microanal. 19 (Suppl. 2) (2013) S1394–S1395. [8] T. Yaguchi, M. Konno, T. Kamino, M. Watanabe, Observation of threedimensional elemental distributions of a Si device using a 360 degree-tilt FIB and the cold field-emission STEM system, Ultramicroscopy 108 (2008) 1603–1615. [9] M. Tanaka, F. Chu, M. Shimojo, M. Takeguchi, K. Mitsuishi, K. Furuya, Positionand size-controlled fabrication of iron silicide nanorods by electron-beaminduced deposition using an ultrahigh-vacuum transmission electron microscope, Appl. Phys. Lett. 86 (2005) 183104. [10] M. Tanaka, M. Shimojo, M. Han, K. Mitsuishi, K. Furuya, Ultimate sized nanodots formed by electron beam-induced deposition using an ultrahigh vacuum transmission electron microscope, Surf. Interf. Anal. 37 (2005) 261–264. [11] K. Furuya, K. Mitsuishi, M. Shimojo, M. Song, M. Tanaka, M. Takeguchi, Electron beam induced deposition of position and size controlled structures on the nanometre scale, Microsc. Semicond. Mater. Springer Proc. Phys. 107 (2005) 239–242.

57

[12] I. Sychugov, Y. Nakayama, K. Mitsuishi, Composition control of electron beam induced nanodeposits by surface pretreatment and beam focusing, J. Phys. Chem. C 113 (52) (2009) 21516–21519. [13] N. Silvis-Cividjian, C.W. Hagen, P. Kruit, Spatial resolution limits in electronbeam-induced deposition, J. Appl. Phys. 98 (2005) 084905. [14] M. Malac, R. Egerton, M. Freeman, J. Lau, Y. Zhu, L. Wu, Electron-beam patterning with sub-2 nm line edge roughness, J. Vac. Sci. Technol. B: Microelectron. Nanometer Struct. 23 (1) (2005) 271–273. [15] H.R. Zhang, R. Egerton, M. Malac, Local thickness measurement through scattering contrast and electron energy-loss spectroscopy, Micron 43 (1) (2012) 8–15. [16] L. Reimer, Transmission Electron Microscopy, 4th ed., Springer Verlag, Berlin, 1997. [17] L.I. Maisel, R. Glang, Handbook of Thin Film Technology, McGraw-Hill book Company, New York, 1970. [18] T. Yamamoto, Y. Wada, T. Sakata, H. Mori, M. Goto, S. Hibin, S. Yanagida, Microwave-assisted preparation of silver nanoparticles, Chem. Lett. 33 (2) (2004) 158–159. [19] T. Kamino, T. Yaguchi, T. Ohnishi, K. Umemura, S. Tomimats, Site specific TEM preparation using an FIB/STEM system, Microsc. Microanal. 6 (2000) 510–511. [20] R.F. Egerton, P. Li, M. Malac, Radiation damage in the TEM and SEM, Micron 35 (2004) 399–409.

Nano-dot markers for electron tomography formed by electron beam-induced deposition: nanoparticle agglomerates application.

A method allowing fabrication of nano-dot markers for electron tomography was developed using an electron beam-induced deposition in an ordinary dual ...
1MB Sizes 0 Downloads 3 Views