Skip to content

Calibration and Standards

Paige Quarterman edited this page Sep 20, 2026 · 1 revision

Calibration & Standards

Quantitative microscopy is only as trustworthy as the calibration and reference measurements behind it. FermiViewer keeps those inputs explicit: spatial scale can differ by axis, instrument settings live in versioned profiles, and factors measured from known standards retain their source, region, conditions, and uncertainty.

Spatial calibration

The Calibration card in the Image inspector shows the active pixel scale. Use the familiar single-value mode for square pixels, or switch to per-axis mode to enter independent row and column extents. Files that already contain anisotropic calibration display both values automatically.

  • A horizontal calibration line updates the column extent.
  • A vertical calibration line updates the row extent.
  • A diagonal line is refused because one diagonal length cannot determine two independent scales without inventing an answer.
  • Changing Å / nm / µm converts the current values; it does not reinterpret the same number in a new unit.

Measurements, areas, FFT reciprocal axes, layer thicknesses, and rotated box profiles use the appropriate directional scale. Calibration changes are data edits and participate in undo/redo.

Calibration Center

Open the Calibration Center from the calibration controls to manage named profiles for a microscope, detector, camera, or acquisition setup. A profile can hold physical values and uncertainty, provenance, a validity window, and applicable beam-energy, magnification, or camera-length ranges.

Profiles are versioned. Editing creates a new version instead of rewriting the conditions behind an earlier result. Applying a profile snapshots its identity and values into the image; captured results copy that snapshot, so later edits or deletion cannot change what an existing measurement meant. Applicability warnings are findings, not silent substitutions.

For quantitative EDS, explicit request values take precedence, followed by the applied profile and then the documented route default. Responses say which source supplied beam energy, takeoff angle, probe current, and live time. This makes a fallback value visible beside the composition it affects.

Known-composition standards

A standard records:

  • a name and immutable version;
  • composition in an explicitly stated wt% or at% basis;
  • optional uncertainty and certified mass-thickness;
  • provenance and notes; and
  • named reference regions tied to source images.

The basis is not interchangeable. EDS Cliff–Lorimer and ζ derivations use weight fractions; EELS quantification uses atomic fractions. FermiViewer performs the required conversion and records the original basis.

Deriving factor sets

Measure either the full standard dataset or one of its named regions:

Technique Derived value Important conditions
EDS Cliff–Lorimer relative k factors reference element, beam energy, fitted peaks
EDS Watanabe absolute ζ factors certified mass-thickness, dose, detector/acquisition conditions
EELS partial cross-sections σ edge and background windows, beam energy, collection semi-angle, reference anchor

Region restriction changes the measurement itself, not just its label: only the selected pixels are summed, and ζ uses the corresponding share of the acquisition dose. Exact non-rectangular regions use their masks rather than their bounding boxes.

Derived sets are immutable measurements. Each entry carries its measured intensity, uncertainty, composition fraction, and provenance. EELS sets also retain the hydrogenic model value for comparison. A standard determines relative EELS sensitivities; the absolute scale still needs a reference cross-section, supplied explicitly or anchored to the model value for the chosen reference element.

Quality findings

Quantitative EDS responses include a QC block covering issues such as poor counting statistics, peak interference, structured residuals, strong self-absorption, factors used away from their derivation voltage, and parameters that fell back to defaults. These findings do not erase the result; they travel beside it so you can decide whether to acquire more counts, change the fit, or repeat the measurement under better conditions.

Next: Analysis Workshops or Scripting & Automation.

Clone this wiki locally