Károly Havancsák, Nóra Nagy and András Szigethy
Technoorg Linda Ltd, Ipari Park utca 10, 1044 Budapest, Hungary
Gábor Varga and Zoltán Dankházi
Eötvös Loránd University, Faculty of Science Research and Instrument Core Facility
Hungary, 1117 Budapest, Pázmány Péter sétány 1/a.
Sample preparation plays a crucial role in any investigation and it is particularly true for electron backscatter diffraction (EBSD) measurements. In this technique the information depth is as shallow as some tens of nanometer and this is the reason why this method requires damage- and oxide-free sample surfaces. There are a number of techniques to prepare specimens for EBSD, among others mechanical grinding and polishing, chemical etching, electropolishing and different ion polishing treatments. Unfortunately, there is no single technique and treatment that works with all materials and this is why it is essential to have a suitable protocol to achieve the desired result. In the past decades the ion milling and polishing techniques have been used increasingly. These techniques use either low-energy (0.1-2 keV) and relatively high-energy (2-20 keV) ions of inert gases (mostly Ar or Xe) in near parallel beams, or ions of metallic origin (in most cases Ga) as in the focused ion beam (FIB) systems. The main benefit of the ion beam milling and polishing methods is that they are less sensitive to the chemical composition of the sample material. Nevertheless, the ion beam sample preparation also needs different approaches to the conductive and non-conductive, hard and soft or homogeneous and inhomogeneous materials. In this presentation we show some examples how to achieve good EBSD results in case of different and even hardly polishable materials applying high- and low-energy Ar+ ion sample preparation method.
Scanning electron microscopy (SEM) is a highly informative and more than ever popular experimental tool in materials science. At the same time sample preparation is an absolute prerequisite for every microscopy. Every specimen that goes into SEM needs some form of sample preparation.
Recently, the SEM apparatuses to an increasing number are equipped with a detector for electron backscattered diffraction (EBSD). EBSD can be used to determine both grain size and orientation, and even to phase identification in a polycrystalline material. This is why EBSD is so useful in materials investigations. Since, in EBSD measurement the diffraction information comes from a shallow surface layer (a few tens of nanometers) the most critical issue of this measurement is the surface quality. The surface should be perfectly clean, free of deformed or amorphous surface layer and moreover it should be flat because of the shadowing effect. Lack of these factors can result either no or faded diffraction pattern. The traditional sample preparation techniques, such as mechanical or electro polishing methods can hardly satisfy all demands, and they are usually time consuming treatments. The mechanical surface treatment has another drawback; after grinding and polishing a heavily deformed or amorphous layer of 1-100 nm thickness remains on the surface. Decrease or removal of this layer is difficult or needs lengthy and careful mechanical treatments.
These are the reasons that in the last decades a new promising technique has been developed. This new method is based upon the ion milling, the physical basis of which is the sputtering of atoms from the surface. Ion milling is the erosion of solid surfaces due to ion bombardment. During ion milling the incident ions transfer energy to the atoms of the target through collisions. If the target atoms gain sufficient energy then they escape from the surface. The rate of this erosion depends mainly on the material of the target (via surface binding energy) and the energy density given to the surface target atoms. This energy density depends on the mass, the energy, and the direction of the incident ions.
Different ion beam methods have been used in the last decades for TEM lamella thinning or to prepare flat sample surface and cross sectional specimens for SEM measurements. These methods use either low energy (0.1-2 keV) or relatively high energy (2-20 keV) ions of inert gases (e.g. Ar+) or some ions of metallic origin, (mostly Ga+) in the focused ion beam (FIB) technique. The main benefit of the ion milling techniques is that they are less sensitive to the microstructure and chemical composition of the samples. The focused ion beam and the near parallel broad beam techniques are used for different purposes. The advantages of the inert gas parallel beam technique are the relatively large size of the processed area and the low costs of the treatment. While using FIB the size of the achievable area is in the order of 100 µm × 100 µm, with parallel noble gas beams one can easily produce a 1000-2000 µm wide and 100-300 µm deep cut/polished area, not to mention that not every scanning electron microscope has a built-in FIB system.
In the framework of the cooperation between Technoorg Linda Co. Ltd. and Eötvös Loránd University the SEM and EBSD measurements presented in this paper have been carried out at the University’s SEM Laboratory using a FEI “Quanta 3D FEG” SEM and its EDAX EBSD device .
The sample preparations were accomplished using Technoorg Linda’s SEMPrep-2 ion mill (SC-2000). The SC-2000 device developed by Technoorg Linda Ltd  was designed specifically for SEM sample preparation. It uses near parallel Ar+ ion beams and capable of both surface polishing and cross-sectional sample preparation (slope cutting). The device has two ion guns; a high energy gun with energy range of 2-16 keV and a low energy one with energy range of 0.1-2 keV. The sample stage can be tilted in the range of 0-30° towards the active ion source and the surface polishing sample holder can do oscillation or full in-plane rotation in order to have homogeneous surface treatment. In case of both guns the near parallel Ar+ ion beam treats an area of about 100 mm2, but the homogeneous central part is about 15 mm2 as it can be seen in Fig. 1. During the ion milling the sample is located in a vacuum chamber which holds ~10-3 Pa dynamic pressure.
Fig. 1. The quality of the Ar+ ion treated surface as a function of the distance from the beam center.
The slope cutting head units are offered with fixed 30°, 45° and 90° stage pretilt, but the most suitable for EBSD sample preparation appears to be the 30° version. For slope forming, a titanium (Ti) mask blocks the lower half of the incoming Ar+ ion beam. The shaded sample part remains untreated, but the rest of the sample surface is sputtered by the upper half of the beam, therefore parallel to the beam, a smooth EBSD-ready surface is created. During slope cutting, the sample can oscillate in a ±40° angular range. To obtain a good surface it is recommended first to use the high-energy ion gun for slope cutting, and afterwards the low-energy one to remove the backsputtered atoms from the freshly cut surface. Using the high energy gun with energy of 10 keV and with anode current of 4 mA, the milling rates at different materials are as follows: 2.5 μm/min at Cu, 1.7 μm/min at Si and 0.8 μm/min at WC‑Co. Fig. 2 shows a slope cut in Si single crystal. The cut was made at 30o, the milling time was 60 min and accordingly the maximum depth of the cut is ~ 100 μm.
Fig. 2. Slope cut in Si single crystal. The dark central part is the new surface.
The aim of the present paper is to demonstrate the advantages of Ar+ ion polishing and the unique capabilities of the SC-2000 apparatus. In this paper presentation we show some examples how to achieve good EBSD results in case of different and even hardly polishable materials applying high- and low-energy Ar+ ion sample preparation method.
III. RESULTS AND DISCUSSION
EBSD and Ar+ ion polishing
Elastically backscattered electrons from a crystalline sample produce a diffraction pattern (Kikuchi pattern) in a scanning electron microscope. This image is detected by a fluorescent screen, and the pattern is recorded by a high speed CCD camera. The Kikuchi pattern shows bands corresponding to Bragg reflection directions from the grain lattice planes where the focused electron beam is directed to. As usual, the diffraction pattern gives information about the type and direction of the crystal just hit by the beam. In such a way we get phase and directional information in every measured points of the scanned area and as a result we can draw an orientation map (OM) of it. The EBSD measurement conditions were the following: 70° sample tilt, 20 kV accelerating voltage and 4 nA sample current for the SEM electron beam.
The EBSD program can give information about the surface quality of the measured area. To define the quality of a Kikuchi pattern the system calculates an image quality (IQ) value as the intensity sum of the indexed Kikuchi bands. There is direct relation between the Kikuchi pattern and the IQ: the blurrier or vaguer the bands are, the smaller the IQ is. Since we get an image quality value from every point of the measured area, we can use this parameter either for having an image quality map of the area selected for investigation or averaging these values for the measured area, we can use the average value as a measure of the surface quality. The link between the IQ and the diffraction pattern is direct, but the relation of the IQ to the material itself is complex. It depends on many factors: the crystal orientation, the conditions of the measurement (contrast, brightness etc.), the strains in the material and naturally the quality of the surface are the most important. To be careful we kept all possible conditions constant during the measurements, so the image quality will give us information on the quality of the surface.
In order to prepare the samples for EBSD measurement, the first step always consisted of conventional mechanical grinding and polishing in order to remove the rough bumpiness, and ensure a rather clean, uniform surface to start the Ar+ ion milling. The steps of the mechanical treatment were as follows: grinding with abrasive paper of 600-, 1200-, 2500- and 4000-grit and polishing with alumina paste of 1 µm average particle size.
Following the mechanical treatment we continued sample preparation with Ar+ ion milling. For a successful Ar+ ion polishing we should know the angle and the time of the treatment. In a previous work  we have figured out that in case of high energy Ar+ ion beam the optimum angle lays between 4-7o. At smaller angles the time of the treatment is too long (>30 min), while at larger angles the surface bumpiness (i.e. crater formation etc.) can be increased due to the ion bombardment. This angle range proved to be usable for miscellaneous materials.
The milling time depends much more on the material of the sample treated than the milling angle. Therefore, when having a new material, first we always have to find the optimum treatment time, and this procedure needs additional samples. This preliminary operation is accomplished by step by step increasing the time and measuring the image quality (IQ) parameter on the same sample, as it can be seen in Fig. 3. It is a good practice to choose the time as optimum at that point where the IQ first reaches the saturation level. It can be seen in Fig. 3 that in case of copper the optimum parameters are 7o and 6 min. In case of an extremely hard material such as martensitic steel the optimum time can be as long as 26 min at angle of 7o as it is seen in Fig. 4 . At the same time this figure demonstrates that the over-treating by Ar+ ion is not a good choice.
Fig. 3. Time dependence of the IQ value in case of an annealed Cu sample. Inserted images show the results of EBSD measurement before and after Ar+ ion treatments made at the optimum parameters (7o, 6 min).
Fig. 4. Time dependence of the IQ value in case of martensitic steel. Inserted image shows the results of EBSD measurement at the optimum Ar+ ion treatment (7o, 26 min). The in-grain structure shows the structure of martensitic lath packets.
In case we found the right parameters (optimum angle and time) for Ar+ ion treatment, the resulting surface is flat and clean. In such a surface condition of the sample the grain structure of a polycrystalline material can be seen even without EBSD measurement owing to the channeling effect of the backscattered electrons using the ETD or BSED detectors of the SEM. This is demonstrated on an annealed Ni wire in Fig. 5 after Ar+ ion polishing of 6 min at 6o.
Fig. 5. Grain structure of an annealed Ni wire measured after Ar+ ion polishing.
The potentiality of Ar+ ion polishing is even more obvious in Fig. 6 a and b. where part of a LED is shown before and after Ar+ ion treatment. The LED firs was cut by diamond saw, then mechanically polished in the last step by alumina past of 1 μm. Fig. 6a shows the surface after mechanical polishing. Afterwards Ar+ ion polishing was applied using the high energy gun at 8 kV, 4o for 30 min. The resulting surface can be seen in Fig. 6b demonstrating that after such a treatment the surface is much cleaner and flatter and details of the sample can be seen much better.
Fig. 6. Details of the active GaN layer and the Au connection in LED after mechanical polishing (a) and Ar+ ion treatment (b).
EBSD and slope cut
There are materials which are hardly polishable. The inhomogeneous (soft and hard) and porous material are the best examples for such materials. It is extremely difficult to treat such material mechanically and their Ar+ ion polishing is challenging as well. In such case the slope cut can give a solution.
As an example here we show the slope cut of sintered WC-Co material at a degree of 30o. Nowadays this material is extremely popular due to its outstanding mechanical properties. However its mechanical preparation is problematic owing to the difference in hardness of WC and Co. The Ar+ ion cut process is less sensitive to the hardness differences, thus the cut surface is flat enough as it can be seen in Fig. 7. At the right side of the image the cut edge is seen.
Fig. 7. The new surface of a WC-Co sintered material cut at angle of 30o.
This surface is suitable for EBSD measurement as it is seen in Fig. 8a, but some backsputtered material blurs the grains. Using the low energy gun, at 5o for 20 min the surface is cleaned and the EBSD measurement gives an extremely good quality result.
Fig. 8. EBSD orientation map measured at 30o cut surface on WC-Co sintered material after high energy cut (a) and after subsequent low energy cleaning (b).
The limestone is an even more demanding material for EBSD, because it is an isolating, porous material which is extremely sensitive to backsputtering which can prevent a good quality EBSD image. In Fig. 9a we can see the new surface after the cut using the high energy gun with parameters: 10 keV, 60 min. The new surface is flat enough, but the EBSD image in blurred due to the backsputtered material (Fig. 9b). Using the low energy beam at 1 keV for 5 min the surface is cleaned and the EBSD measurement gives a good quality image (Fig. 9c). All SEM measurements on limestone sample were made after coating the surface by carbon with thickness of 3 nm.
Fig. 9. The new surface of limestone material cut at angle of 30o(a), EBSD orientation map after cut with the high energy gun (b), and after cleaning with the low energy gun (c).
The examples presented in this paper demonstrate the usefulness of Ar+ ion polishing and slope cut for SEM and EBSD measurements even in case of difficult materials.
Polishing with a high energy Ar+ ion beam can produce an EBSD-quality surface in relatively short time. The grain structure of such a high quality surface can be observed even without EBSD measurement.
The slope cut can be useful when the material is investigated in depth. However, the cut procedure is useful in case of hardly polishable materials as well such as inhomogeneous or porous ones.
 Z. Dankházi, Sz. Kalácska, A. Baris, G. Varga, Zs. Radi and K. Havancsák, EBSD sample preparation: high energy Ar ion milling, Materials Science Forum Vol. 812 (2015) pp 309-314
 T. Berecz, Sz. Kalácska, G. Varga, Z. Dankházi, K. Havancsák, Effect os high energy Ar-ion milling on surface of quenched low-carbon low-alloyed steel, Materials Science Forum Vol. 812 (2015) pp 285-290