EBSD Sample Preparation

By Rachael Beane, Bowdoin College

Sample preparation


Electron and optical microscopic investigations typically begin with a thin section of rock. Here we detail the method that we have successfully used for preparing such a thin sections for EBSD analyses. Thin sections used for EBSD analyses require additional preparation steps, because grit-based polishing of samples introduces crystal damage in the uppermost surface that most be removed to obtain quality patterns. Materials are specified because this information is frequently requested, and should not be viewed as product endorsements - equivalents and alternatives are available. For some alternative methods of preparing thin sections and other materials, please see information on TSL's website and Oxford/HKL's website.


  1. Make (or have commercially made) a microprobe quality polished thin section.
  2. Mount the thin section on a brass polishing holder (suggestion of M. Cheadle, University of Wyoming). The brass polishing holder can be prepared in a machine shop by cutting a brass rod in half, and trimming the halved-rod to fit a standard thin section (see photo). The thin sectionneeds to be adhered to the holder. One method is to heat the brass polishing weight on a hot plate, cut a thin section sized piece of sheetadhesive (e.g. WaferMount 562 sheet), lay it on the holder until the adhesive softens, then place the thin section (rock side up!) on the adhesive and remove holder+ thin section from the hot plate. Allow to cool prior to polishing.
  3. Prepare the vibratory polisher (e.g. Buehler Vibromet 2) by adhering a polishing pad (e.g. 12" Buehler Microcloth, PSA-backed) to the polisher.Slightly dampen the pad with water (e.g. squeeze DI water until the pad is 50-75% covered dampened). Add enough colloidal silica (e.g. Buehler Mastermet 2 non-crystallizing colloidal silica polishing suspension) to barely cover the pad completely).
  4. Place the thin section+holder (rock side down) in the colloidal silica on the vibratory polisher, and polish (SYTON method of Fynn and Powell, 1979). How long to polish? Long enough to obtain good quality patterns. The time required will vary depending on the thin section and on the minerals in the thin section. We typically polish thin sections, two at a time, for six hours. The polisher runs automatically with no user intervention required unless the thin section slips off from the holder (in which case it needs to be re-adhered and restarted on the polisher).
  5. Remove thin section + holder from polisher and rinse thoroughly with water. A thorough rinsing is needed to avoid having the colloidal silica crystallize on the thin section (note that in the photo of the mount the collidal silica – white - has crystallized slightly on the holder). Reheat the holder just until the thin section can be pushed off. Optional: rinse the thin section with ethanol. We find that little or no adhesive remains on the glass bottom of the thin section. Remains of the adhesive sheet on the brass holder can be removed with a razor blade (our recommendation) or dissolved chemically (per instructions with adhesive). At this point, new thin sections can be readied for polishing. If no further polishing is expected in the next day, then the polishing pad should be thoroughly rinsed to avoid crystallizing the colloidal silica on the pad. With thorough rinsing, the pad can be re-used for many polishes.

Reference cited:

Fynn, G. W. & Powell, W. J. A., 1979. The Cutting and Polishing of Electro-optic Materials. Adam Hilger, London. 216 pp.

Operating parameters

Here are a few parameters and questions to consider when conducting EBSD analyses and when writing the methods section of papers. Specific operating parameters will be instrument, user, and material dependent.

  1. How is electron charging on the thin section minimized? One possibility is to apply as thin as possible carbon coat over the thin section, or paint carbon around the edges and underneath a thin section. Another possibility is to reduce the vacuum pressure (e.g. in a variable pressure SEM). For our low-vacuum SEM, we find that a reduced chamber pressure of approximately 15 Pa, combined with the 70° tilt successfully minimizes charging.
  2. What electron microscope parameters are used? For example, we collect EBSD patterns with an accelerating voltage of 20kV, working distance of 25 mm, and probe current of 2.2nA. These parameters might vary, depending on the microscope and material, but likely should be specified.
  3. What EBSD system settings are used when collecting patterns? For example typical values in our lab are Hough resolution = 75, 7 bands, 80 reflectors, 2x2 or 4x4 binning, and high gain. These values will vary depending on EBSD system, instrument, and material analyzed.
  4. How are maps collected (if applicable)? Is the stage moved automatically under a stationary beam? Is the beam scanned over the entire sample? (And, if so, is a correction factor used to minimize distortion away from the center?) Or is the beam scanned over small areas with stage moved between areas?
  5. What is the data source for the lattice parameters used to create the reflector file (aka "match unit")? Consider citing the original article that contained the lattice parameters used to make the reflector file. The American Mineralogist Crystal Structure Database is a helpful source of crystallographic data.



Analysis methods

Similar to the section on operating parameters, this section offers some questions to consider when analyzing (or post-processing) data collected with EBSD methods. Not all questions will apply to all projects.

  1. What are the mean angular deviations between the detected Kikuchi bands and the simulations? Consider both any cutoffs that you set (e.g. reject all data with deviations greater than 1 degree or 1.5 degrees) as well as the average mean angular deviation for the data collected.
  2. Did you choose to remove isolated data points (aka wild spikes)?
  3. Did you observe systematic misindexing? If so, was it removed or kept?
  4. Did you extrapolate data from points where data was successfully collected into points that had no solutions? If so how? and what error – if any – might this introduce into the interpretation of the data?
  5. Did you create a subset of the collected data for final analysis?