Preventing damage and redeposition during focused ion beam milling: The “umbrella” method

Ultramicroscopy 186 (2018) 35–41

Contents lists available at ScienceDirect

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

Preventing damage and redeposition during focused ion beam milling: The “umbrella” method T. Vermeij a,b, E. Plancher a,∗, C.C. Tasan a,∗ a b

Massachusetts Institute of Technology, Department of Materials Science and Engineering, 77 Massachusetts Avenue, Cambridge, MA 02139, USA Eindhoven University of Technology, Department of Mechanical Engineering, Den Dolech 2, 5612 AZ Eindhoven, The Netherlands

a r t i c l e

i n f o

Article history: Received 20 September 2017 Revised 1 December 2017 Accepted 6 December 2017 Available online 7 December 2017

a b s t r a c t Focused ion beam (FIB) milling has enabled the development of key microstructure characterization techniques (e.g. 3D electron backscatter diffraction (EBSD), 3D scanning electron microscopy imaging, sitespecific sample preparation for transmission electron microscopy, site-specific atom probe tomography), and micro-mechanical testing techniques (e.g. micro-pillar compression, micro-beam bending, in-situ TEM nanoindentation). Yet, in most milling conditions, some degree of FIB damage is introduced via material redeposition, Ga+ ion implantation or another mechanism. The level of damage and its influence vary strongly with milling conditions and materials characteristics, and cannot always be minimized. Here, a masking technique is introduced, that employs standard FIB-SEM equipment to protect specific surfaces from redeposition and ion implantation. To investigate the efficiency of this technique, high angular resolution EBSD (HR-EBSD) has been used to monitor the quality of the top surface of several micro-pillars, as they were created by milling a ringcore hole in a stress-free silicon wafer, with or without protection due to an “umbrella”. HR-EBSD provides a high-sensitivity estimation of the amount of FIB damage on the surface. Without the umbrella, EBSD patterns are severely influenced, especially within 5 μm of the milled region. With an optimized umbrella, sharp diffraction patterns are obtained near the hole, as revealed by average cross correlation factors greater than 0.9 and equivalent phantom strains of the order 2 × 10−4 . Thus, the umbrella method is an efficient and versatile tool to support a variety of FIB based techniques. © 2017 Elsevier B.V. All rights reserved.

1. Introduction In the last couple of decades, focused ion beam (FIB) milling has been established as a key tool to manufacture, modify or polish specimens at the micrometer and nanometer scales [1]. A wide variety of tools to analyze the microstructure (e.g. transmission electron microscopy (TEM) [2], 3D tomographic imaging [3], 3D electron backscatter diffraction (EBSD) [4], 3D electron channeling contrast imaging (ECCI) [5] and atom probe tomography [6]) and micromechanical testing techniques (e.g. micro-pillar compression [7], micro-beam bending [8] and in situ tensile tests [9–11]) strongly rely on FIB to quantitatively characterize microstructural features or mechanical fields down to atomistic scale [12,13]. Yet, measurements in the vicinity of FIB-milled areas are often disturbed by undesired features stemming from ion implantation [14–18] or material redeposition [19].

∗

Corresponding authors. E-mail addresses: [email protected] [email protected] (C.C. Tasan). https://doi.org/10.1016/j.ultramic.2017.12.012 0304-3991/© 2017 Elsevier B.V. All rights reserved.

(E.

Plancher),

Ion implantation is caused by the capture of Ga+ ions within the crystal [18]. The implantation process leads to the creation of crystallographic defects such as vacancies and dislocation loops [17], and even leads to amorphization [20]. The TEM bright field image shown in Fig. 1(a) illustrates the difference in defect density observed in a copper film between the undamaged crystal and an area exposed to the Ga+ ion beam [16]. Similar observations have been made with ECCI [21]. Increasing defect density (i) blurs channeling and diffraction contrasts in ECCI images and Kikuchi patterns used for EBSD, creating significant challenges for several microstructure characterization techniques in the scanning electron microscope (SEM) [22]; and (ii) has significant influence on the mechanical testing of plasticity mechanisms [14]. Redeposition takes place since some of the atoms removed by the ion beam (which cannot all be evacuated by the vacuum system) reattach elsewhere on the surface [23]. Fig. 1(b) shows how redeposition builds up at the edges of a trench created in steel by FIB milling [19]. As seen in the atomic force microscopy (AFM) measurement, redeposition pile-ups are formed at the edges of the trench. The amount of redeposition gradually decreases with the distance to the trench. When the thickness of the amorphous

36

T. Vermeij et al. / Ultramicroscopy 186 (2018) 35–41

ever, fully shielding the surface against FIB-induced damage with such an approach is not possible without a thorough optimization of the umbrella material and shape. Additionally, we propose HREBSD as a powerful in situ tool to assess the amount of FIB damage. To evaluate the efficiency of the umbrella method, it is used here to monitor the difference in quality of the top surface of several micro-pillars, as they are being created by milling a ringcore hole in a stress-free silicon wafer, with or without the protection provided by the umbrella. 2. Methods Fig. 1. Ion implantation and redeposition induced by FIB milling. (a) Defects introduced in a copper film partially exposed to FIB. The TEM bright-field image shows the difference in defect density between the undamaged crystal and a damaged area (reproduced with permission from [16]). (b) Surface topography changes due to redeposition close to a trench. The top part shows a secondary electron image of the FIB-milled trench. The bottom part reports the AFM height profile measured over the above line, showing redeposition outside the trench (reproduced with permission from [19]).

redeposition layer approaches the size of the backscatter electron interaction volume, channeling and diffraction signals are diffused, leading to a diminution of the contrast in ECCI images and Kikuchi patterns. In conventional ringcore experiments, redeposition also tends to cover the patterns used for digital image correlation of secondary electron images [12]. The contrast distribution is disrupted in the SE images as the milling proceeds, potentially leading to erroneous measurements of the initial strain state. To overcome these drastic effects, various strategies have been developed to prevent FIB damage [10,24–27]. Most common approaches are based on optimizing FIB parameters or the milling geometry. Examples include using multiple FIB currents in a single machining process, milling the surface to be analyzed from the backside [10], avoiding the use of FIB images by more careful SEM imaging [24]. Such approaches successfully reduce ion implantation or redeposition, however, it is difficult to ensure for all cases that a fully damage-free microstructure is obtained. The success thereof often depends on how focused the beam is at low accelerating voltages, a parameter difficult to assess experimentally. A preliminary ECCI investigation carried out by the authors (not reported here) has shown that even at low accelerating voltages (∼1 kV) and under grazing incidence conditions, FIB damage could not be avoided in a dual-phase steel. Other more intrusive approaches focus on processing the sample either prior to the milling step (e.g. by adding sacrificial layers [25,26]) or after it (e.g. by carrying out annealing treatments [27]). These methods are also successful but introduce other challenges: heat treatments have unwanted effects on the microstructure and sacrificial layers limit the access to the surface. It might be that other strategies to decrease FIB damage have been investigated and even employed, but the literature is currently limited on this topic despite a steady increase in the popularity of FIB. The task is especially challenging because assessing damage in situ (without removing the sample from the FIB instrument) is hardly possible. This brief overview reveals an urgent need for a generic method to efficiently protect the sample surface against all FIB damage. This method should: (i) enable the use of damage-sensitive techniques such as ECCI and high angular resolution EBSD [28] (HREBSD) after FIB milling; (ii) require only standard equipment and accessories found in dual-beam FIB-SEM instruments; and (iii) enable site-specific protection that can be applied to a wide range of specimen shapes and material types. In what follows, we present the “umbrella” method that involves temporarily masking a given surface area of interest with a soft polymer block to fulfil the three conditions set above. This is a simple idea at first glance, how-

A schematic overview of the umbrella method is shown in Fig. 2. The specific goal here is to protect the top surface of micropillars obtained by ringcore drilling experiments [29]. In this work, several approaches have been investigated. They are illustrated in Fig. 2(a). Without any protection (red path) extensive FIB damage, due to both ion implantation and redeposition, is observed on the sample surface. Some improvement is observed when an unoptimized umbrella is temporarily placed on the top of the pillar during the milling step, using a micro-manipulator (MM) (orange path). In this scenario, the umbrella shape and material are not optimized so the amount of damage observed is reduced but not minimized. The surface is mostly shielded from direct ion implantation but redeposition occurs under the umbrella, as atoms ejected from the hole penetrates in the gap between the surface and the umbrella. Most of our trials have fallen into this category. To achieve an optimal protection, the gap should be closed, by enabling an excellent contact between the umbrella and the surface of the material to be milled (green path). Additionally, the um-

Fig. 2. Schematic overview of the umbrella method. (a) Damage obtained in a FIB drilling experiment of a ringcore hole, without protection (red), with protection of an unoptimized umbrella (orange) and the optimized umbrella (green). Damage is shown as red glow on the surface. (b) Schematic representation of the umbrella manufacturing process. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

T. Vermeij et al. / Ultramicroscopy 186 (2018) 35–41

brella should not remain partially or fully adhered to the surface after the milling operation, nor be chemically reacting with it. The choice of the umbrella material and shape results from a systematic screening and testing process on materials of varying stiffness, beam sensitivity, adhesion and chemical interaction tendencies. We have considered or tested steel, silicon, aluminum, Polydimethylsiloxane (PDMS) and natural rubber. Blocks made up of aluminum or stiffer materials did not have good contact with the sample surface and allowed redeposition to penetrate under the block. In the other extreme, rubber blocks had the advantage of being compliant, but their micro-machining to the desired size or shape using FIB was problematic due to charging effects. Unwanted adhesion and chemical interactions were also observed. We have identified PDMS to be the best candidate since: (i) it can be shaped by micro-molding, which creates flexibility in size and shape; (ii) its stiffness can be controlled by varying the amount of curing agent and by tuning curing conditions; (iii) it can also be observed at low voltage in the SEM-FIB without too much charging and (iv) by applying a platinum coating at the bottom of the block, adhesion and chemical interaction tendencies with the sample can be minimized. The umbrella preparation steps are presented in Fig. 2(b). Umbrellas are cut and machined out from custom-designed PDMS pillars. While Fig. 2(b) gives a schematic overview of the fabrication process, Fig. 3 illustrates four steps in the umbrella fabrication process. Soft lithography was carried out to mold PDMS (Sylgard 184) onto a silicon wafer containing FIB-machined holes (5 μm of diameter, 20 μm deep). A 1:20 weight ratio of curing agent was used and curing conditions were set to 6 hours at 60 °C to create a soft material with controlled Young’s modulus [30]. After demolding, fresh micrometric pillars were obtained on a bulk PDMS layer, as seen in Fig. 3(a). To prepare the umbrella and carry out the experiments, we employed a FEI dual beam FIB-SEM (Helios 600), equipped with an Omniprobe micro-manipulator, a Pt gas injection

Fig. 3. Overview of the umbrella fabrication (a) PDMS micro pillar molded by soft lithography. (b) Roughly shaped umbrella resting on a Si substrate, with the micromanipulator in view. (c) Rotated micro-manipulator, showing the bottom of the umbrella, to be coated with Pt. The inset shows the configuration used to cut the bottom surface flat. (d) Final umbrella attached to the micro-manipulator. The scale bars represent 5 μm in each image.

37

system and an Oxford Nordlys EBSD detector. The electron beam was operated at 1 kV, 98pA while the ion beam voltage was kept at 30 kV. One PDMS pillar was cut off at 2.8nA and picked up with the micro-manipulator (Pt deposition was carried out at 45pA) (cf. Fig. 3(a)). The pillar, which is from this point referred to as the rough umbrella, was then placed near the edge of a silicon wafer, which was subsequently taken out of the microscope and put on a 45° pre-tilted stub (cf. Fig. 3(b)). The Si edge was sticking out from the topside of the stub. After reinsertion in the FIB-SEM, the holder was rotated 180° and tilted an extra 7° to align the Si surface with the FIB x axis. The rough umbrella was then picked up with the micro-manipulator, and its bottom was cut flat with consecutive FIB steps as shown in the inset of Fig. 3(c) (with decreasing intensities, from 1nA to 28pA). After putting the rough umbrella down on the Si wafer, the stage tilt was brought back to 0° and the stage was rotated 180° again. Next, the stage was tilted to 35° and the rough umbrella was picked up and cut into the required shape (a disk of diameter 4 μm). The micro-manipulator was rotated over its axis 160° counterclockwise and a 1μm-thick layer of Pt was deposited at 45pA, as illustrated in Fig. 3(c). It was then rotated back to obtain an umbrella ready for use, shown in Fig. 3(d) from two angles. The performance of the umbrella method was evaluated by monitoring the quality of several Kikuchi patterns acquired on a stress-free silicon single crystal during several drilling experiments of ringcore holes. FIB was operated at coarse (30 kV, 90pA) or fine (5 kV, 47pA) settings. The sample surface was tilted 80° from the horizontal and placed at a 4 mm working distance according to the electron beam. In this configuration SEM, FIB, EBSD and MM could be used at once. The ion beam was used to dig holes with a ringcore shape, creating a free-standing pillar with a diameter of 5 μm. A first experiment was conducted to evaluate the extend of FIB damage without using protective means. Two ringcore holes were machined with coarse and fine FIB parameters, far away from one another, without using the umbrella protection. To prevent additional Ga+ ion implantation in the area, FIB imaging was avoided beforehand to determine the position of the holes, instead, the electron and ion beams were aligned and the SEM image was used to position the milling pattern. After milling, two HR-EBSD line acquisitions were performed over 40 μm, with a step size of 1 μm, starting at the hole and heading towards the clean crystal along the tilt axis. The surface topography around the ringcore milled with rough FIB settings was quantified using AFM to study the redeposition layer thickness. A Digital Instruments Nanoscope V apparatus was used to acquire height profiles with sub-nanometer resolution along three 17μm-long scans (starting from the edge of the hole and heading towards the pristine silicon). A slight surface tilt was corrected by considering the last 7 μm of the scan as a baseline. In a second in situ experiment, the umbrella method was used to protect the top surface of the pillar created by drilling a similar ringcore hole using coarse FIB parameters. First, three Kikuchi patterns were taken at 9 positions (spaced out on a 3 × 3 grid with a step size of 750 nm) before creating the ringcore hole. Then, the umbrella was placed on that position using the MM. Special care was taken to press down the umbrella on the surface without using sliding motions, creating a good coverage of the area of interest, without causing any additional damage. The ringcore hole was milled using the FIB. Finally, a second EBSD acquisition was conducted after removing the umbrella, at exactly the same 9 spots as before and by taking three Kikuchi patterns per point. The HR-EBSD method [28,31] was used to compare the quality of the Kikuchi patterns acquired near the hole after milling with the one obtained from clean surfaces. This technique is extremely sensitive to the crystal condition in a 50–100 nm top layer.

38

T. Vermeij et al. / Ultramicroscopy 186 (2018) 35–41

It is used here to provide quantitative indexes of surface damage. Full-definition Kikuchi patterns (640 × 480 pixels) were taken with an exposure time of 850 ms, averaged over 12 frames. The electron beam was set to 20 kV and 2.7nA. The reference points for the line scans were taken 40 μm away from the holes. During the in situ measurements (drilling with protection), neither the EBSD detector nor the sample had moved during the full extent of the experiment. The reference points were taken at each spot before the milling step. A drift of less than 1 μm was observed in the SEM image but the patterns were taken from the same physical points on the surface. This drift correlates to a modification of the pattern center which is less than 3 μm and therefore does not strongly influence the results. The patterns were analyzed by using the StrainCorrelator software [31]. Normalized averaged cross correlation factors (XCF) and equivalent (von Mises) Lagrange elastic strains are reported here. As a stress-free state was present in the silicon wafer, these strains stem from FIB-induced damage. For clarity, they are referred to as “phantom” strains. In best HR-EBSD measurements, strain errors are expected to be below 2 × 10−4 [32,33]. As the reference pattern is always of optimal quality (taken from a clean, undamaged surface), the XCF acts as a marker of quality for the other patterns. A value close to 0 indicates a poor quality and a value close to 1 an optimal quality. 3. Results Fig. 4 shows the detrimental effect of redeposition on the HREBSD measurements taken around two ringcore holes milled without using protection. One of the two micro-pillars obtained after milling is shown in Fig. 4(a). The position of the HR-EBSD lines-

Fig. 4. HR-EBSD assessment of damage near a FIB hole. (a) Example of a ringcore hole produced by FIB milling in a stress-free silicon single crystal. The black line indicates the position used for the HR-EBSD linescans. (b) Average cross correlation factor (XCF) and equivalent phantom strain plotted against the distance to the hole for coarse and fine milling parameters. Kikuchi patterns taken 1 μm from the hole in the coarse and fine cases are shown along with the reference pattern taken on an undamaged surface (40 μm from the hole). (c) Height profiles obtained by AFM plotted against the distance to the ringcore hole drilled with coarse milling parameters.

can is indicated by the black line. The results of this measurement are reported in Fig. 4(b). The AFM height profiles measured near a ringcore hole are introduced in Fig. 4(c). Note that, with the exception of the milled area, the sample surface investigated was never intentionally exposed to direct Ga+ irradiation (FIB images and snapshots were entirely avoided). A visual inspection of the Kikuchi patterns taken 1 μm from the trench shows a significant decrease in pattern quality near the hole. The difference between coarse (30 kV, 90pA) and fine (5 kV, 47pA) FIB settings has a visible influence on the pattern quality: coarse parameters induce a lower sharpness of the Kikuchi bands, suggesting more damage such as thicker redeposition layers. After using coarse FIB settings, the HREBSD algorithm can barely conduct an analysis up to 5 μm from the hole, as shown by XCFs smaller than 0.7 and phantom strains over 10−3 . Between 5 μm and 25 μm from the hole, errors in the HREBSD measurement decrease, with XCFs approaching 1 and equivalent phantom strains decreasing to 2 × 10−4 . The AFM topography measurement in Fig. 4(c) highlights a 5 nm-thick pileup of redeposition at the edge of the ringcore milled with coarse settings. The amount of redeposition decreases rapidly to meet the baseline 5 μm from the hole. In the first 5 μm, the Kikuchi pattern quality clearly improves in conjunction with a diminution of the redeposition layer. Between 5 μm and 17 μm, no significant height variation is detected with AFM. In that range, the potential thickness gradient (<1 nm/μm) of the redeposition layer is identical to the natural waviness of the silicon wafer and the measurement appears inconclusive. From 25 μm to 40 μm, the patterns cannot be visually distinguished, yet, XCF and phantom strain data still show a slight improvement. In this range, the hypothesized effect of the redeposition is comparable to other sources of HR-EBSD errors such as sample misalignment and lens distortions effect [34]. These effects are closely related to the distance between the reference and the measurement point. Fabricating the hole with fine FIB settings brings significant benefits compare to the coarse setting case, up to 20 μm from the hole. For example, at 5 μm, the XCF reaches 0.9 for the fine settings, while in the coarse experiment it only approaches 0.8. Similar differences are observed in the phantom strain values, yet, a clear similarity in trends between coarse and fine settings is seen. Fig. 5 shows the results of the in situ experiment when the umbrella was used. The secondary electron image in Fig. 5(a) shows the position of the 9 spots where Kikuchi patterns were acquired. The contrast in the image comes from the contamination traces left on the surface. Fig. 5(b) shows the umbrella placed on the surface before using the ion beam, a good contact is obtained between the Si surface and the umbrella. The surface state after milling, with the umbrella still on the surface, is illustrated in Fig. 5(c). A full ringcore cannot be machined because the umbrella remains attached to the MM during milling, blocking the ion beam in a portion of the image. The protected surface is revealed in Fig. 5(d), where new HR-EBSD spot scans were acquired (at the same positions as the previous spots). Redeposition is seen as a brighter layer in the image. On the outer side of the hole this is observed as a halo effect, the central part shows sharp changes from white to dark, indicating pileups of redeposition against the umbrella. Pileups were also detected when the umbrella was removed by a strong adhesion with the surface. The HR-EBSD results used to assess the surface quality are given in Fig. 5(e). Kikuchi patterns after milling are similar to the reference pattern (see Fig. 4(b)), and show optimal contrast. In Fig. 5(e), XCF and equivalent phantom strains are plotted against the spot number defined in the central inset. For all spots, XCF are greater than 0.9 and phantom strain values fluctuate around 2 × 10−4 respectively, indicating few artefacts in the measurements. The surface under the umbrella has been adequately protected during the milling step.

T. Vermeij et al. / Ultramicroscopy 186 (2018) 35–41

Fig. 5. The umbrella method is used to protect the top surface of a micro-pillar created by drilling a ringcore hole in a stress-free silicon single crystal (a) Secondary electron image of the sample surface after 9 reference HR-EBSD spots have been taken. (b) An umbrella attached to the MM is placed on the surface. (c) A ringcore hole is milled by FIB with the umbrella providing cover. (d) Protected area where the new HR-EBSD spot scans have been performed. The inset shows the same ringcore observed at 0° tilt. (e) Results of the HR-EBSD spot analysis: XCF and equivalent phantom strains are shown for each spot. Post-milling Kikuchi patterns are shown for spot 2 and 8. The scale bars respresent 3 μm in each image.

4. Discussion In this study, HR-EBSD is used as a tool to quantify FIB damage. It is well suited for this purpose as the technique can be used in situ, is extremely sensitive to small changes in the Kikuchi patterns and the interaction volume investigated (50–100 nm [35]) is similar to that of other SEM-based observation techniques such as ECCI. A stress-free single crystal (similar to the one used to assess the accuracy of HR-EBSD [32]) was employed to make sure that FIB

39

damage is the main mechanism explaining Kikuchi pattern degradation. In particular, the experiments have been designed to ensure that no actual strains are present where the measurements are carried out (on the top surface, at least 1 μm away from the trench). At the very edge of the pillar, in the first 50 nm, actual residual stress up to 500 MPa can build up as a result of direct 30 kV Ga+ ion implantation [15]. However, this local residual field hardly disrupts the elastic field at the micron scale, enabling accurate measurements with the conventional DIC-based ringcore method [12]. As great care was taken not to expose the surface to ions before or after the milling step, redeposition is thought to be the driving damage mechanism, especially under the umbrella. The redeposition process involves low energy, heavy atoms being gradually reattached on the surface in small quantity (<10 nm) and therefore is unlikely to induce a significant stress state in the material underneath. Under this assumption, any strain measured is considered an erroneous representation of the state in the actual strain-free crystal and is referred to as “phantom strain”. The decrease in XCF, which provides a quantitative measure of the overall pattern quality (relatively to the perfect pattern from the pristine silicon crystal), may be explained by the amorphous nature of the redeposition layer which blurs the diffraction signal. In the literature, studies on the effect of redeposition are usually limited to its effect on the shape and quality of the trench or fabricated feature [36,37]. Little data is available regarding the surrounding surface. This last phenomenon of ‘surface redeposition’ is more challenging to predict and assess, as there is often no straight path for the sputtered atoms to travel from inside the trench onto the surface. Interestingly, AFM height measurements indicates a pileup of redeposition only in the direct vicinity of the hole (<5 μm). In that region, there is a strong correlation between the amount of redeposition and the quality of the Kikuchi patterns. Farther away, the decrease of the redeposition layer thickness still remains a strong candidate to explain the further improvement in XCF, even though that can hardly be proven here. Sputter atoms are known to travel long distances in SEM chambers and redeposit on detectors and chamber walls. Arguably, the baseline height seen in Fig. 4(c) doesn’t reflect the actual silicon wafer surface as a thin homogenous redeposition layer could spread over a large area. An alternative explanation would be direct irradiation (e.g. coming from a natural divergence in the ion beam) but the likelihood of such a phenomenon is low: direct ECCI observations have revealed sharp transition (<500 nm) at the edge of a FIB milled area in a stainless steel [21]. In any case, in situ HR-EBSD has been shown to be an adequate way to detect low intensity FIB damage, stemming most likely from surface redeposition. The phantom strains reported here are a good indicator of the magnitude of strain errors that would arise in a measurement performed on samples with actual residual strains. Phantom strains convey a different information than the XCF, as the later mainly provides a quantitative measure of surface quality but is hardly correlated with the accuracy of strain measurements. The precise origin of phantom strains is difficult to interpret. Phantom strains reflect small shifts measured in certain regions of the EBSP with respect to the reference state. A quick inspection reveals that the amorphous redeposition layer tends to blur the diffraction signal from high order crystal planes first, and this change in the patterns is reflected as a shift by the cross correlation algorithm. The new redeposition layer also changes the depth from which the EBSP signal comes from, altering the complex dynamical diffraction effect [35,38]. It might finally be reflecting actual stress coming from the creation of a local mismatch at the new interface. When the umbrella is used, the number of artefacts observed is drastically reduced, compared to the unprotected case: XCF values are greater than 0.9 and the equivalent phantom strains are below 4 × 10−4 . Only the equivalent elastic strain was reported

40

T. Vermeij et al. / Ultramicroscopy 186 (2018) 35–41

in the paper for clarity, but it should be noted that single elastic strain components stay below 2 × 10−4 . This shows that the umbrella adequately protects the surface from damage, as no significant HR-EBSD artefacts are created during the milling operation. Larger scans with different shapes (e.g. cross instead of square grids) could be used in the future to investigate the extent of the protected area near the pillar edges. In an additional experiment (not shown here) where the umbrella was left on the surface for a longer period of time, without any FIB milling in between, no noticeable phantom strain or decrease in pattern quality was observed. Some limitations in the protection provided by the umbrella can be observed in Fig. 5. By comparing the umbrella shape and position on the surface in Fig. 5(c) with the shape of the dark area in Fig. 5(d), it can be seen that the left side of the area was not fully protected (the white glow indicates some redeposition). By comparing Fig. 4(b) and 5(e), one can also note that the XCF and the phantom strains in Fig. 5(e) show random fluctuations slightly out of the optimal range of HR-EBSD errors. Thus, the umbrella method could be further optimized to ensure a more reliable protection. Decreasing the long HR-EBSD acquisition and FIB milling time would also reduce SEM artefacts such as drift [39]. With the current experimental configuration, the smallest area we expect to protect reliably is 1μm2 . Limitations to the ability to protect a smaller surface come from the difficulty (i) to mold a tall, thin and soft umbrella, (ii) to pick it up with a micro-manipulator of a similar size and (iii) to position it accurately on the surface. The largest umbrella we manufactured was 10 μm in length with an elliptical shape (5 μm in breadth). With an increased size, obtaining a flat bottom surface by milling and a smooth layer of Pt becomes more challenging. Adhesion effects with the surface are also higher, increasing the risk of losing the umbrella during the positioning and removal steps of the in situ experiment. None of these limitations appear intrinsically dictated, i.e., they may be overcome by improving the experimental configuration further, or by using more advanced milling or deposition technologies. Finally, the umbrella method is an encouraging step on the path to accurate absolute HR-EBSD measurements. Absolute HR-EBSD measurements have been attempted mostly by using simulationbased Kikuchi patterns for cross-correlation [40–44] or 3D Hough transforms [45]. Unfortunately, the strain errors reported with those methods are still very large (∼10−3 ) and stem from the difficulty to precisely determine the pattern center position [46]. Here, it has been shown in a stress-free crystal that one could perform HR-EBSD during a local relaxation experiment without strong artefacts (cf. the low phantom stain values in Fig. 5c). Such relaxations experiments provide a stress free state that can be used as a reference state for the HR-EBSD measurements. In other words, one could substitute the use of digital image correlation by HR-EBSD in the ringcore method [29], to get locally the full elastic strain tensor with an excellent sensitivity. The knowledge of local stress states, coupled with total strain measurements would provide direct insights into the local mechanical behavior of polycrystals [47]. 5. Conclusions A novel masking technique has been introduced to avoid FIB damage. It consists of protecting a sensitive area from redeposition and ion implantation by temporarily placing an optimized block of PDMS on the surface, during the milling step. The performance of the protective method was validated with ringcore hole drilling experiments where FIB-damage was monitored using HR-EBSD. Without protection, even with fine FIB milling settings, sensitive measurements like HR-EBSD are hardly possible near the hole. For convenient coarse FIB settings, FIB damage has dreadful effects on the surface quality in the first 5 μm, with XCFs below 0.7

and equivalent phantom strains over 10−3 . Both values are far outside the usual range of HR-EBSD errors. At 25 μm away from the hole, XCF and phantom strains reach acceptable values of ∼1 and 2 × 10−4 respectively By using the umbrella method, (i) sharp patterns suitable for HR-EBSD are obtained 2 μm from the hole, (ii) the surface quality after FIB milling approaches that of a pristine surface (XCFs are greater than 0.9, equivalent phantom strain values fluctuate around 2 × 10−4 ) and (iii) the in situ requirements are fulfilled using only standard FIB-SEM equipment. The umbrella fabrication method can be used to create umbrellas with a shape and size tailored to protect specific surfaces on a wide range of materials. The umbrella method would be especially useful to protect the top surface of micro-sized tensile specimens [48] and micro-cantilever beams [49,50], often employed in SEM in situ experiments to characterize the material behavior at the grain scale. The use of high-sensitivity characterization techniques such as ECCI and HR-EBSD on areas previously affected by FIB is expected to provide new and artifactfree insights into plasticity mechanisms and local stress states. Acknowledgments The authors are in debt to Dr. Shiahn J. Chen for his valuable advice and assistance during the experiments. We gratefully acknowledge Dr. Claire Maurice for her long-term support and for providing the StrainCorrelator software. Dr. ir. Johan Hoefnagels and Marc van Maris are acknowledged for their input and assistance in the finalization of the manuscript. This work made use of the shared experimental facilities supported in part by the MRSEC program of the National Science Foundation under award number DMR – 1419807. References [1] C. Kim, S. Ahn, D. Jang, Review: developments in micro/nanoscale fabrication by focused ion beams, Vacuum 86 (2012) 1014–1035. http://doi.org/10.1016/j. vacuum.2011.11.004. [2] D. Tomus, H.P. Ng, In situ lift-out dedicated techniques using FIB–SEM system for TEM specimen preparation, Micron 44 (2013) 115–119. http://dx.doi.org/10. 1016/j.micron.2012.05.006. [3] R.K. Bansal, A. Kubis, R. Hull, J.M. Fitz-Gerald, High-resolution threedimensional reconstruction: a combined scanning electron microscope and focused ion-beam approach, J. Vac. Sci. Technol. B 24 (2) (2006) 554–561. http://dx.doi.org/10.1116/1.2167987. [4] J. Guyon, N. Gey, D. Goran, S. Chalal, F. Pérez-Willard, Advancing FIB assisted 3D EBSD using a static sample setup, Ultramicroscopy 161 (2016) 161–167. http://dx.doi.org/10.1016/j.ultramic.2015.11.011. [5] J. Man, T. Vystave, A. Weidner, I. Kubena, M. Petrenec, T. Kruml, J. Polák, Study of cyclic strain localization and fatigue crack initiation using FIB technique, Int. J. Fatigue 39 (2012) 44–53. http://dx.doi.org/10.1016/j.ijfatigue.2011.05.002. [6] K. Babinsky, R. De Kloe, H. Clemens, S. Primig, A novel approach for sitespecific atom probe specimen preparation by focused ion beam and transmission electron backscatter diffraction, Ultramicroscopy 144 (2014) 9–18. http: //dx.doi.org/10.1016/j.ultramic.2014.04.003. [7] D. Kiener, C. Motz, G. Dehm, Micro-compression testing: a critical discussion of experimental constraints, Mater. Sci. Eng. A 505 (2009) 79–87. http://doi. org/10.1016/j.msea.2009.01.005. [8] C. Motz, T. Schoberl, R. Pippan, Mechanical properties of micro-sized copper bending beams machined by the focused ion beam technique, Acta Mater. 53 (2005) 4269–4279. http://dx.doi.org/10.1016/j.actamat.2005.05.036. [9] R.D. Field, P.A. Papin, Location specific in situ TEM straining specimens made using FIB, Ultramicroscopy 102 (2004) 23–26. http://dx.doi.org/10.1016/ j.ultramic.20 04.08.0 02. [10] C. Du, J.P.M. Hoefnagels, L.I.J.C. Bergers, M.G.D. Geers, A uni-axial nanodisplacement micro-tensile test of individual constituents from bulk material, Experimental 57 (2017) 1249–1263. http://dx.doi.org/10.1007/ s11340- 017- 0299- 6. [11] G. Dehm, M. Legros, B. Heiland, In-situ TEM straining experiments of Al films on polyimide using a novel FIB design for specimen preparation, J. Mat. Sci. 41 (2006) 4484–4489. http://dx.doi.org/10.1007/s10853- 006- 0087- 7. [12] A.J.G. Lunt, A.M. Korsunsky, A review of micro-scale focused ion beam milling and digital image correlation analysis for residual stress evaluation and error estimation, Surface Coat. Technol. 283 (2015) 373–388. http://dx.doi.org/ 10.1016/j.surfcoat.2015.10.049. [13] C. Mansilla, V. Ocelík, J.Th.M.De Hosson, Local residual stress measurements on nitride layers, Mater. Sci. Eng. A 636 (2015) 476–483. http://dx.doi.org/10.1016/ j.msea.2015.04.023.

T. Vermeij et al. / Ultramicroscopy 186 (2018) 35–41 [14] J. Hutsch, E.T. Lilleodden, The influence of focused-ion beam preparation technique on microcompression investigations: Lathe vs. annular milling, Scripta Mater 77 (2014) 49–51. http://dx.doi.org/10.1016/j.scriptamat.2014.01.016. [15] A.M. Korsunsky, J. Guénolé, E. Salvati, T. Sui, M. Mousavi, A. Prakash, Erik Bitzek, Quantifying eigenstrain distributions induced by focused ion beam damage in silicon, Mater. Lett. 185 (2016) 47–49. http://dx.doi.org/10.1016/j. matlet.2016.08.111. [16] D. Kiener, C. Motz, M. Rester, M. Jenko, G. Dehm, FIB damage of Cu and possible consequences for miniaturized mechanical tests, Mater. Sci. Eng. A 459 (2007) 262–272. http://dx.doi.org/doi:10.1016/j.msea.2007.01.046. [17] H. Idrissi, S. Turner, M. Mitsuhara, B. Wang, S. Hata, M. Coulombier, J. Raskin, T. Pardoen, G. Van Tendeloo, D. Schryvers, Point defect clusters and dislocations in FIB irradiated nanocrystalline aluminum films: an electron tomography and aberration-corrected high-resolution ADF-STEM study, Microsc. Microanal. 17 (2011) 983–990. https://doi.org/10.1017/S143192761101213X. [18] C.-M. Park, J.A. Bain, T.W. Clinton, P.A.A. van der Heijden, T.J. Klemmer, Measurement of Ga implantation profiles in the sidewall and bottom of focusedion-beam-etched structures, Appl. Phys. Lett. 84 (2004) 3331–3333. http://dx. doi.org/10.1063/1.1715142. [19] S.N. Bhavsar, S. Aravindan, P. Venkateswara Rao, Experimental investigation of redeposition during focused ion beam milling of high speed steel, Precis. Eng. 36 (2012) 408–413. http://dx.doi.org/10.1016/j.precisioneng.2011.12.005. [20] S. Rajsiri, B.W. Kempshall, S.M. Schwarz, L.A. Giannuzzi, FIB damage in silicon: amorphization or redeposition? Microsc. Microanal. 8 (Suppl. 2) (2002). [21] I. Gutierrez-Urrutia, Analysis of FIB-induced damage by electron channelling contrast imaging in the SEM, J. Microsc. 265 (1) (2017) 51–59. http://dx.doi. org/10.1111/jmi.12462. [22] N. Mateescu, M. Ferry, W. Xu, J.M. Cairney, Some factors affecting EBSD pattern quality of Ga+ ion-milled face centred cubic metal surfaces, Mater. Chem. Phys. 106 (2007) 142–148. http://dx.doi.org/10.1016/j.matchemphys.2007.05. 048. [23] D.A.M. de Winter, J.J.L. Mulders, Redeposition characteristics of focused ion beam milling for nanofabrication, J. Vac. Sci. Technol. B 25 (2007) 2215–2218. http://dx.doi.org/10.1116/1.2806973. [24] V. Samaeeaghmiyonia, H. Idrissia, J. Grotenb, R. Schwaigerb, D. Schryvers, Quantitative in-situ TEM nanotensile testing of single crystal Ni facilitated by a new sample preparation approach, Micron 94 (2017) 66–73. http://dx.doi.org/ 10.1016/j.micron.2016.12.005. [25] A. Kannegulla, L. Cheng, Metal assisted focused-ion beam nanopatterning, Nanotechnology 27 (2016) 36LT01. 6pphttp://dx.doi.org/10.1088/0957-4484/27/36/ 36LT01. [26] J.P. Liebig, M. Göken, G. Richter, M. Mačkovic´, T. Przybilla, E. Spiecker, O.N. Pierron, B. Merle, A flexible method for the preparation of thin film samples for in situ TEM characterization combining shadow-FIB milling and electron- beamassisted etching, Ultramicroscopy 171 (2016) 82–88. http://dx.doi.org/10.1016/ j.ultramic.2016.09.004. [27] S. Lee, J. Jeong, Y. Kim, S. Min Han, D. Kiener, S. Ho Oh, FIB-induced dislocations in Al submicron pillars: annihilation by thermal annealing and effects on deformation behaviour, Acta Mater. 110 (2016) 283–294. http://dx.doi.org/10. 1016/j.actamat.2016.03.017. [28] A.J. Wilkinson, G. Meaden, D.J. Dingley, High-resolution elastic strain measurement from electron backscatter diffraction patterns: new levels of sensitivity, Ultramicroscopy 106 (2006) 307–313. http://dx.doi.org/10.1016/j.ultramic.2005. 10.001. [29] A.M. Korsunsky, M. Sebastiani, E. Bemporad, Focused ion beam ring drilling for residual stress evaluation, Mater. Lett. 63 (2009) 1961–1963. http://dx.doi.org/ 10.1016/j.matlet.2009.06.020. [30] Z. Wang, A.A. Volinsky, N.D. Gallant, Crosslinking effect on polydimethylsiloxane elastic modulus measured by custom-built compression instrument, J. Appl. Polym. Sci (2014) 41050. 4pphttp://dx.doi.org/10.1002/APP.41050. [31] C. Maurice, J.H. Driver, R. Fortunier, On solving the orientation gradient dependency of high angular resolution EBSD, Ultramicroscopy 113 (2012) 171–181. http://dx.doi.org/10.1016/j.ultramic.2011.10.013. [32] S. Villert, C. Maurice, C. Wyon, R. Fortunier, Accuracy assessment of elastic strain measurement by EBSD, J. Microsc. 233 (2) (2009) 290–301. https: //doi.org/10.1111/j.1365-2818.2009.03120.x.

41

[33] E. Plancher, J. Petit, C. Maurice, V. Favier, L. Saintoyant, D. Loisnard, N. Rupin, J.-B. Marijon, O. Ulrich, M. Bornert, J.-S. Micha, O. Robach, O. Castelnau, On the accuracy of elastic strain field measurements by laue microdiffraction and high-resolution EBSD: a cross-validation experiment, Exp. Mech. 56 (2016) 483–492. http://doi.org/10.1007/s11340-015-0114-1. [34] K. Mingard, A. Day, C. Maurice, P. Quested, Towards high accuracy calibration of electron backscatter diffraction systems, Ultramicroscopy 111 (2011) 320– 329. http://doi.org/10.1016/j.ultramic.2011.01.012. [35] W. Wisniewski, S. Saager, A. Böbenroth, C. Rüssel, Experimental evidence concerning the significant information depth of electron backscatter diffraction (EBSD), Ultramicroscopy 173 (2017) 1–9. http://dx.doi.org/10.1016/j.ultramic. 2016.11.004. [36] B.I. Prenitzer, C.A. Urbanik-Shannon, L.A. Giannuzzi, S.R. Brown, R.B. Irwin, T.L. Shofner, F.A. Stevie, The correlation between ion beam/material interactions and practical FIB specimen preparation, Microsc. Microanal. 9 (3) (2003) 216–236. https://doi.org/10.1017/S1431927603030034. [37] S. Rubanov, P.R. Munroe, FIB-induced damage in silicon, J. Microsc. 214 (3) (2004) 213–221. https://doi.org/10.1111/j.0 022-2720.20 04.01327.x. [38] S. Zaefferer, On the formation mechanisms, spatial resolution and intensity of backscatter Kikuchi patterns, Ultramicroscopy 107 (2) (2007) 254–266. https: //doi.org/10.1016/j.ultramic.20 06.08.0 07. [39] M.A. Sutton, N. Li, D. Garcia, N. Cornille, J.J. Orteu, S.R. McNeill, H.W. Schreier, X. Li, Metrology in a scanning electron microscope: theoretical developments and experimental validation, Meas. Sci. Technol. 17 (10) (2006) 2613–2622. http://dx.doi.org/10.1088/0957-0233/17/10/012. [40] J. Kacher, C. Landon, B.L. Adams, D. Fullwood, Bragg’s law diffraction simulations for electron backscatter diffraction analysis, Ultramicroscopy 109 (2009) 1148–1156. https://doi.org/10.1016/j.ultramic.20 09.04.0 07. [41] C. Maurice, R. Fortunier, J. Driver, A. Day, K. Mingard, G. Meaden, in: J. Kacher, C. Landon, L.A. Brent, F. David (Eds.), Comments on the Paper “Bragg’s Law Diffraction Simulations for Electron Backscatter Diffraction Analysis, 110, 2010, pp. 758–759. http://dx.doi.org/doi:10.1016/j.ultramic.2010.02.003. [42] T.B. Britton, C. Maurice, R. Fortunier, J.H. Driver, A.P. Day, G. Meaden, A.J. Wilkinson, Factors affecting the accuracy of high resolution electron backscatter diffraction when using simulated patterns, Ultramicroscopy 110 (12) (2010) 1443–1453. https://doi.org/10.1016/j.ultramic.2010.08.001. [43] D. Fullwood, M. Vaudin, C. Daniels, T. Ruggles, S.I. Wright, Validation of kinematically simulated pattern HR-EBSD for measuring absolute strains and lattice tetragonality, Mater. Charact. 107 (2015) 270–277. https://doi.org/10.1016/ j.matchar.2015.07.017. [44] J. Alkorta, M. Marteleur, P.J. Jacques, Improved simulation based HR-EBSD procedure using image gradient based DIC techniques, Ultramicroscopy 182 (2017) 17–27. http://dx.doi.org/10.1016/j.ultramic.2017.06.015. [45] C. Maurice, R. Fortunier, A 3D Hough transform for indexing EBSD and Kossel patterns, J. Microsc. 230 (3) (2008) 520–529. http://dx.doi.org/10.1111/j. 1365-2818.2008.02045.x. [46] C. Maurice, K. Dzieciol, R.A. Fortunier, Method for accurate localisation of EBSD pattern centres, Ultramicroscopy 111 (2) (2011) 140–148. http://dx.doi.org/10. 1016/j.ultramic.2010.10.007. [47] E. Plancher, V. Favier, C. Maurice, E. Bosso, N. Rupin, J. Stodolna, D. Loisnard, J.-B. Marijon, J. Petit, J.-S. Micha, O. Robach, O. Castelnau, Direct measurement of local constitutive relations, at the micrometre scale, in bulk metallic alloys, J. Appl. Crystallogr. 50 (3) (2017) 940–948. https://doi.org/10.1107/ S160 05767170 06185. [48] Y. Kihara, T. Nagoshi, T.F.M. Chang, H. Hosoda, S. Tatsuo, M. Sone, Tensile behavior of micro-sized specimen made of single crystalline nickel, Mater. Lett. 153 (2015) 36–39. http://dx.doi.org/10.1016/j.matlet.2015.03.119. [49] G.P. Zhang, K. Takashima, Y. Higo, Fatigue strength of small-scale type 304 stainless steel thin films, Materials Science and Engineering 426 (1) (2006) 95–100. http://dx.doi.org/doi:10.1016/j.msea.2006.03.090. [50] J. Gong, A.J. Wilkinson, Anisotropy in the plastic flow properties of singlecrystal α titanium determined from micro-cantilever beams, Acta Mater. 57 (19) (2009) 5693–5705. http://dx.doi.org/doi:10.1016/j.actamat.2009.07.064.