Camera for transmission He+ ion microscopy

+ Camera for transmission He ion microscopy Karen L. Kavanagh, Christoph Herrmann, and John A. Notte Citation: Journal of Vacuum Science & Technology B, Nanotechnology and Microelectronics: Materials, Processing, Measurement, and Phenomena 35, 06G902 (2017); doi: 10.1116/1.4991898 View online: http://dx.doi.org/10.1116/1.4991898 View Table of Contents: http://avs.scitation.org/toc/jvb/35/6 Published by the American Vacuum Society Camera for transmission He1 ion microscopy Karen L. Kavanagha) and Christoph Herrmann Physics Department, Simon Fraser University, Burnaby, British Columbia V5A 1S6, Canada John A. Notte Carl Zeiss Microscopy, Peabody, Massachusetts 01690 (Received 23 June 2017; accepted 11 September 2017; published 26 September 2017) The intensity of transmitted Heþ ions and atoms from a focused Heþ beam was measured using a direct impact onto a camera located 20 cm below the sample. The camera consisted of a 256  256 array of Si p-i-n diodes (pixels) each 55  55 lm2 in area. Static, focused beam intensity profiles show a uniform distribution, as expected from the small de Broglie wavelength (80 fm) and coherent source. From the size of the beam spot, the half-angle beam convergence angle can be directly measured and compared with theoretical predictions based on column geometries. The detector count rate was consistent with an efficiency of 75%, when compared to the current measured by electrostatic beam blanking. The intensity profile of a neutral atom component was measured to have a 17% peak or 1.3% integrated intensity at a beam energy of 25 keV, compared to the ion beam. The intensity of transmitted particles through 50 nm amorphous carbon was less than that of graphite flakes, likely affected by thickness and channeling. Transmission images as a function of C 2017 American Vacuum Society. time to form 20 nm holes in graphite were obtained. V [http://dx.doi.org/10.1116/1.4991898] I. INTRODUCTION One of the first applications of helium (He) ions was to probe the nature of the atomic nucleus.1 The detection of backscattered 5 MeV Heþ2 ions confirmed the existence of a dense atomic nucleus and changed our understanding of the atom. The same approach was developed into a very powerful analysis technique generally known as Rutherford backscattering spectroscopy.2 Backscattered ion energy profiling, channeling, and blocking are used to study the atomic composition, crystallinity, atomic positions, and fraction of defects and interstitial atoms in materials. Subsequent experiments with lower energy Heþ2 ions and neutral atom beams have extended the beam solid interactions to include atomic surfaces and the detection of He atom diffraction.3,4 The development of field ion microscopy and liquid metal ion sources has enabled the construction of focused Heþ ion beams (20–40 kV) for milling, lithography, and scanning secondary electron (SE) imaging with a nanometer resolution.5,6 Most recently, time-of-flight detection of backscattered and secondary ions has been added to such techniques.7 The spatial coherence of these Heþ beams is thought to be significant due to the atomic size of the emission sources. Channeling is evident from the reduction in SE emission at the surface as a function of sample orientation;5,6,8 scanning transmission images with an atomic resolution are theoretically predicted,9 but such transmission images are yet to be obtained. We define three classes of transmitted Heþ detectors, depending on the information collected. The simplest detectors (class 1) detect all Heþ ions without any information on angular distribution. The detection of secondary electrons on a conversion plate, excited below thin samples, has been used as an endpoint or thickness detector,10 which is also a) Electronic mail: kavanagh@sfu.ca 06G902-1 J. Vac. Sci. Technol. B 35(6), Nov/Dec 2017 possible with a scintillator, a channeltron, or a microchannel plate. Class 2 detectors detect a specific range of transmitted angles via an aperture, such as annular dark field detectors in transmission electron microscopy.11,12 In this way, the influence of the sample thickness and density can be probed using a scanning focused beam. In this work, we detect the intensity and scattering angles of forward-scattered Heþ ions and atoms (class 3) using a camera, consisting of an array of Si diodes located below the sample stage of a Heþ ion microscope (HIM). We characterize transmitted neutral He atoms and Heþ ion beam intensity profiles as a function of microscope parameters and apply this to milling nanometer holes in graphite. II. EXPERIMENT The HIM (Zeiss Orion Nanofab) was operated at 25–35 keV with beam currents below 2 pA. Scanning SE images were obtained using an Everhart–Thornley electron detector. We added a camera below the sample stage, consisting of a commercially available square array of Si p-i-n diodes.13 Designed primarily for the detection of x-rays, it has 256  256 pixels each 55  55 lm2 square, for a total detector area of 1.4  1.4 cm2. The n-type Si side of the array (1 lm thick) was exposed directly to the Heþ beam without optical filters or scintillation layers, other than a Si native oxide. The camera was operated with a reverse bias voltage of 100 V, with a minimum signal discrimination of 4 kV. There were no counts detected (1 s exposure) if the valve between the ion source and the chamber is closed (preventing all He from reaching the detector). Thus, the dark current was zero save for infrequent cosmic rays (1/min) that were identified by their multiple pixel tracks. The camera is supported on a narrow Al plate (1 mm thick) on the floor of the HIM vacuum chamber, placed over an exit flange 2166-2746/2017/35(6)/06G902/5/$30.00 C 2017 American Vacuum Society V 06G902-1 06G902-2 Kavanagh, Herrmann, and Notte: Camera for transmission He1 ion microscopy 06G902-2 (turbopump below the chamber). It is powered via a universal serial bus cable to an external personal computer, through an added vacuum-compatible socket on the chamber left-hand side door. Images are collected using a software (Pixelite) interfaced to the camera controller. A diagram of the ion column is shown in Fig. 1, indicating the beam limiting aperture, the blanking plates, the objective lens, sample, and geometry of the camera at the bottom of the chamber.6 The distance between the surface of the camera (thickness 1.15 cm) and the bottom of the objective lens pole-piece is 21 cm. Like scanning electron microscopy, the beam current in a HIM can be controlled by the size of a beam-limiting aperture inserted before the final objective lens, and the spot size on that aperture is determined by the cross-over point of the upper lens. In addition, the current is also directly proportional to the He gas pressure in the gun, which determines the rate of adsorption and emission from the source tip. Beam currents are measured by deflecting the beam into a large fixed aperture below the blanking plates. Values as low as 10 fA were measureable, limited by the sensitivity of the current meter. The HIM has a standard sample holder that is moved onto the HIM stage through a vacuum load-lock. The stage provides the displacement of the sample holder in the (x, y) plane (625 mm) and out of the plane, z (8 mm), and rotation, R, about z (360 ). It is also attached to a tilting “cradle” that tilts the entire stage and the sample holder about the x-axis, T, (þ52 to 5 ). But there is no hole through the stage center. To allow beam transmission, it was necessary to add a slit (3.0  4.5 cm2) in the left side panel of the tilt cradle. With the cradle tilted þ45 and the stage displaced to an extreme corner (x, y) ¼ (25, 25) mm with respect to the centre, the beam avoids hitting the stage or the cradle. Samples were 3 mm diameter TEM-compatible grids or substrates that were mounted onto a special arm on the sample holder that extended (x, y) ¼ (–25, –25) mm over its edge, positioning thin samples under the beam. To provide normal incidence, this arm also had a –45 fixed tilt. III. RESULTS AND DISCUSSION Figure 2(a) shows a scanning SE, HIM image from a polycrystalline-Si film (90 nm thick) suspended in a Cu folding grid. The SE silicon intensity is the greatest where the sample is transparent, in between some of the grid openings. This might be due to charging but more likely due to the generation of SE from both sides of the sample. The projected range for Heþ ions (30 keV) into Si is 280 6 100 nm, and so, 99% of He transmitted the sample. Figure 2(b) shows the corresponding scanning transmission HIM image obtained after one focused beam scan (total exposure time 1 s). Detection of Heþ ions or atoms occurs over the entire area of the camera, which subtends a solid angle of 5 msr. Figure 2(c) shows a profile of the average intensity per pixel (counts/s) as a function of the position inside the rectangle in (b), showing the periodicity of the Cu grid openings. The maximum current detected was 7 cps/pixel through the J. Vac. Sci. Technol. B, Vol. 35, No. 6, Nov/Dec 2017 FIG. 1. (Color online) Schematic diagram of the Heþ ion column including the beam limiting aperture, beam blanker location, bottom objective lens, sample, and He camera added to the bottom of the chamber. [Figure 1.4 adapted with permission from Helium Ion Microscopy, edited by Hlawacek and G€ olzh€auser (Cham : Springer International Publishing : Imprint: Springer, 2016). Copyright 2016 by Springer Nature.) hexagonal Cu grid holes consistent with the fast focusedbeam scanning rate. No counts were detected within the region of poly-Si. The grid holes are visible as magnified circles spaced 35 pixels apart or 1.9 mm on the detector. Since their actual spacing is 150 lm, this means that a magnification of 13 occurs due to the location of the raster pivot point. The blurriness of the image in Fig. 2(b) is attributed to the divergence 06G902-3 Kavanagh, Herrmann, and Notte: Camera for transmission He1 ion microscopy FIG. 2. Poly-Si (90 nm) in between a Cu folding grid (3 mm diameter). (a) Scanning Heþ ion microscopy image via secondary electron detection (25 fA, 5 ls scan rate). (b) Corresponding He camera image (25 fA, total exposure 1 s). (c) Profile of average within the exposure time of the scan intensity (counts/s) collected from the pixels within the rectangle in (b). (d) Camera image with the Heþ ion beam blanked (500 s exposure). of the beam as it passes through the sample (a divergence half angle of a few mrad). Transmission is blocked by the sample holder on the periphery of the image. The pixels indicated as dead were previously exposed to a Heþ ion dose that exceeded a threshold for permanent failure. Dead pixels developed during our earliest investigations of the detector. The maximum number of counts for each pixel is 10 000, equivalent to only an ion current of 1.6 fA, if 1 count equals one He. Currents as high as 60 pA are readily available with the instrument, and so, it was easy to saturate a pixel. Too many ions into a biased pixel, apparently, permanently damage its sensitivity, but we have not investigated this in any detail. We subsequently avoided pixel currents above 1 pA and tried never to expose the direct beam to a biased detector for more than 10 ms. Figure 3 shows a camera image (0.1 s exposure) of the focused beam spot with the sample removed. In this case, the microscope column operated with an accelerating voltage of 25 kV, but with the same 20 lm limiting aperture and spot size of 4, with low He pressure such that the blanked beam current was 50 6 10 fA. The ion beam was static (spot mode) displaced from the center and focused at the sample, with a distance between the sample surface and the objective lens (working distance) of 10 mm. The larger spot is the ion beam, while the smaller one is due to ions that neutralized within the gun or elsewhere along the column. This neutral He fraction is not focused, and therefore, these accelerated atoms are detected predominantly directly below the gun position. The plot shows the average intensity (cps) through the beam spot center within the rectangular area. As expected, we see a flat distribution in the ion beam intensity 06G902-3 FIG. 3. (Color online) Camera image of the static beam (spot mode) with a profile plot of the average intensity per pixel (counts/s) as a function of the pixel position within the rectangle. The beam was focused at the sample height, and a limiting 20 lm aperture, an exposure of 0.1 s, and a current of 50 fA were used to obtain this image. The image has a grayscale contrast with white which is the highest range. to within a 10% statistical error. Furthermore, the focused ion beam has a diameter of 20 pixels for a total number of pixels in this spot of 314. Hence, given that the total beam current was 50 6 10 fA, the current per pixel was 0.16 6 03 fA or 1000 6 200 Heþ ions. Compared to the average number of counts detected (750 6 50/s), the efficiency of the detector is 75%. Thus, either 25% of the beam is lost between the blanking plates and the detector (unlikely) or the camera misses a statistical fraction of the He beam (some will not penetrate into Si to the sufficient depth to reach the depletion width of the reverse-biased diodes). When the ion beam is blanked, an accelerated, neutral He atom fraction continues to the detector. From the intensity profile in Fig. 3, the neutral-to-ion peak intensity ratio was 17%, while the integrated count from neutrals (2.5 kcps) compared to the ion beam (192 kcps) is 1.3 6 0.1%. The analysis of SE intensity ratios from images produced using higher beam currents (5 pA) found a smaller neutral fraction of 0.02%.5,6 The full width half maximum of the beam is 16 pixels, while the neutral was smaller, 3 pixels. The beam focused at the sample expanded to a circle on the camera of radius 440 lm (8 pixel radius  55 lm/pixel), giving a half-angle convergence, b, of 2.2 mrad (¼ 440 lm/20 cm). Measurements of the ion beam radius, as a function of the increasing spot size, and limiting aperture size, gave the expected reciprocal, parabolic behavior (larger spot size ¼ smaller beam area). Theoretical calculations of the beam convergence angle, b, give similar values.6 Most of the neutral beam intensity must originate close to the gun and the ion beam path since its footprint on the detector is a projection. Thus, its exact position and width are primarily determined by the mechanical position of the gun and the effect of the gun tilt JVST B - Nanotechnology and Microelectronics: Materials, Processing, Measurement, and Phenomena 06G902-4 Kavanagh, Herrmann, and Notte: Camera for transmission He1 ion microscopy and shift controls on the alignment condition. These change each time the trimer source is adjusted, and so, both the neutral atom spot and zero-deflection ion beam spot positions are alignment dependent. Figure 2(d) shows a camera image for a 500 s exposure with the ion beam blanked. The neutral atoms are most intense within the 3 mm diameter Cu grid area, but there is a small flux throughout the detector area from scattering or less frequent neutralization collisions. The connected pixel events are due to cosmic ray particles that impact our detector at a rate of 1/min. This means that there will always be a low level of unfocussed He atoms reaching the sample. However, their largest intensity is limited to a small number of pixels, and thus, they will have a negligible effect on SE images. Figure 4(a) shows a typical SE image obtained from a commercial graphite flake TEM support grid. It consisted of exfoliated graphite flakes supported by a lacey-carbon mesh on a Cu grid. The brightest regions are the amorphous laceycarbon (60 nm thick) which has a distinctive structure consisting of long narrow strands. The material in between is a film with abrupt steps in its SE intensity but continuous, covering 99% of the grid openings, often with faceted edges and filamentary white lines consistent with ripples common to the thinnest layers of graphene. STEM imaging and diffraction (not shown) concluded that the material is a multilayered, hexagonal matrix, but with random rotation about the perpendicular axis consistent with multiple graphite sheets. Electron energy loss spectroscopy finds the thickness to vary from 7 to 60 nm based on the ratio of the inelastic to zero-loss peak areas. Hole milling investigations confirmed that the brighter regions in the SE image of Fig. 4 were associated with the thickest graphite films. Figure 4(b) shows SE images after a hole was milled into a brighter region of the graphite. Next to it is a failed milling attempt where only a white donut shape is visible. Figure 4(c) shows the transmitted peak intensity as a function of time for a static, focused beam spot in darker (black squares) and brighter regions (red triangles) of the SE image. After a delay of seconds, the intensity begins to rapidly increase. The corresponding transmission images showed that a hole forms just when the intensity begins to rapidly increase, with the delay longer for the brighter regions compared to the darker graphite. The rate of increase is similar, but the maximum intensity for the brighter regions is greater than that for the darker since the hole size is larger. Both subsequently begin to decrease with time. The brighter regions took longer to mill compared to the darker regions, indicating that the thickness was indeed correlated with a brighter SE intensity. In many cases, a hole formed that was 20 nm in diameter, as visible via the SE image. In other cases, the same exposure resulted in a buildup of the material, not a hole, confirmed by a reduction in the camera signal. Carbon and associated organic residues are susceptible to beam-induced reactions by the He beam,14 resulting in surface diffusion and accumulation near the beam spot. Previous exposures at a given location made J. Vac. Sci. Technol. B, Vol. 35, No. 6, Nov/Dec 2017 06G902-4 FIG. 4. (Color online) SE images of a graphite on the lacey-carbon TEM support grid at a beam current of 1 pA: (a) typical area and (b) high magnification of the indicated square region in (a). The inset shows a higher magnification image of the milled holes. (c) Peak transmission intensity (counts/s) as a function of time for darker (black squares) and brighter (red triangles) SE regions. subsequent milling more difficult, probably due to the increased thickness due to the accumulation of carbonaceous deposits. Figure 5(a) shows a camera image from transmission through a darker graphite region in the SE image and (b) from the lacey carbon. The intensity profiles from these 06G902-5 Kavanagh, Herrmann, and Notte: Camera for transmission He1 ion microscopy 06G902-5 transmission Heþ ion microscopy. The camera counts single Heþ ions and neutral atoms (energy 20–40 keV), providing information on the beam position, shape, and angular intensity profile, as well as forward scattering patterns through thin samples. Based on the size of the beam at the detector, the half-angle divergence angle of the focused beam was 2 mrad for a limiting aperture of 20 lm and a beam energy of 30 keV. Forward-scattered, transmission images as a function of milling time of graphite were obtained. Nanometer scale holes (20 nm diameters) were obtained when surface carbon accumulation was avoided. An improved forward-scattering resolution would be feasible with a smaller camera pixel size or longer distance from the sample. Detection of channeling and diffraction from 30 keV He (de Broglie wavelength 80 fm; diffraction angle 0.2 mrad) should be feasible given suitably thin and uniformly thick membranes. ACKNOWLEDGMENTS The authors are grateful for funding from NSERC, CFI/ BCKDF, and facilities managed by 4DLabs. 1 E. Rutherford, Philos. Mag. 21, 669 (1911). L. C. Feldman, J. W. Mayer, and S. T. Picraux, Materials Analysis by Ion Channeling (Academic, New York, 1982). 3 Helium Atom Scattering from Surfaces, Springer Series in Surface Sciences Vol. 27, edited by E. Hulpke (Springer-Verlag, Berlin, 1992). 4 J. Seifert, J. Lienemann, A. Sch€ uller, and H. Winter, Nucl. Instrum. Methods Phys. Res., B 350, 99 (2015). 5 D. C. Joy, He Ion Microscopy Principles and Applications (Springer, New York, 2013). 6 olzh€auser Helium Ion Microscopy, edited by G. Hlawacek and A. G€ (Cham : Springer International Publishing : Imprint: Springer, 2016). 7 L. Pillatsch, N. Vanhove, D. Dowsett, S. Sijbrandijb, J. A. Notte, and T. Wirtz, Appl. Surf. Sci. 282, 908 (2013). 8 L. A. Giannuzzi and J. R. Michael, Micros. Microanal. 19, 344 (2013). 9 A. J. D’Alfonso, B. D. Forbes, and L. J. Allen, Ultramicoscopy 134, 18 (2013). 10 A. R. Hall, Microsc, Microanal. 19, 740 (2013). 11 J. A. Notte, R. Hill, S. M. McVey, R. Ramachandra, B. J. Griffin, and D. C. Joy, Microsc. Microanal. 16, 599 (2010). 12 T. Woehl, Microsc. Microanal. 22, 544 (2016). 13 “Modupix camera,” Institute of Experimental and Applied Physics, Czech Technical University, Prague, http://www.widepix.cz/products. 14 S. Tan, K. Klein, D. Shima, R. Livengood, E. Mutunga, and A. Vladar, J. Vac. Sci. Technol., B 32, 06FA01 (2014). 2 FIG. 5. (Color online) Camera images of the transmission of a static, focussed He beam through the graphite-coated TEM support grid of Fig. 4 (1 s exposure, 1 pA) located on (a) a thinner (darker) graphite area or (b) amorphous lacey carbon. (b) Intensity profiles from the rectangular area in (a) graphite (black squares) lacey carbon (red triangles). images show that lacey carbon had a much lower transmission intensity compared to the graphite region, primarily due to a difference in the thickness. The degree of channeling through the graphite films is under investigation. IV. CONCLUSIONS We have investigated the properties of a commercial Si p-i-n diode array (256  256 pixels) as a camera for JVST B - Nanotechnology and Microelectronics: Materials, Processing, Measurement, and Phenomena