Evidence summary
Evidence summary
Report AFM roughness only after checking tip and feedback artifacts, documenting leveling and line correction, defining the region and scale, and comparing multiple representative scans. Sa or Sq without preprocessing and scan-size context is not a reproducible surface claim.
Key takeaways
- 01Preserve the raw height map before flattening or line correction.
- 02Show how plane removal changes Sa and Sq.
- 03Compare equivalent scan sizes, pixel densities and regions.
- 04Inspect tip convolution, streaks, drift and feedback artifacts before statistics.
Reject artifacts before calculating roughness
Repeated features, streaks, abrupt line offsets and feedback oscillations can dominate surface statistics. Review trace and retrace where available and inspect line profiles in both scan directions.
- Look for tip duplication.
- Check edge streaks and missing lines.
- Review feedback and scan-speed context.
Document leveling as part of the result
Plane and line corrections are model choices. Compare the raw, plane-leveled and line-corrected surface, and exclude features from a fit only with a stated scientific reason.
- Record polynomial order.
- Show excluded masks.
- Run a preprocessing sensitivity check.
Match the parameter to the scale
Sa and Sq summarize height deviations but do not describe spatial organization. Report scan size, pixel spacing and, where relevant, autocorrelation length or power spectral density.
- Use equivalent scan dimensions across samples.
- Acquire more than one region.
- Keep units and parameter definitions explicit.
Make the comparison auditable
Export the processed map, raw map, correction settings, region mask and per-scan statistics. Group averages should retain between-region variability.
Methodology and scope
The guide aligns areal surface parameters with peer-reviewed evidence on leveling bias. It emphasizes a sensitivity check across reasonable preprocessing choices rather than a universal flattening recipe.
Limitations
- Finite tip radius limits lateral feature dimensions.
- A small scan may not represent a heterogeneous sample.
- Removing tilt or bow can also remove real long-wavelength topography.
- Roughness values from different scan sizes are not automatically comparable.
References
- [1]
How levelling and scan line corrections ruin roughness measurement and how to prevent it
Nečas, Valtr and Klapetek. Scientific Reports (2020).
doi:10.1038/s41598-020-72171-8 ↗ - [2]
Geometrical product specifications — Surface texture: Areal — Part 2
International Organization for Standardization. ISO 25178-2:2021 (2021).
Open source ↗
Suggested citation
Suggested citation
SciPhys Research Team. “AFM Surface Roughness.” SciPhys, August 5, 2026. https://www.sciphys.com/blog/afm-surface-roughness
Apply the workflow
Inspect roughness on your AFM height map.
Upload a surface file and keep leveling, profiles, regions and roughness metrics visible together.