Stereological pitfalls in prior austenite grain size analysis

H. K. D. H. Bhadeshia

Discussion on: I. Frantsuzov, Z. Dong, O. Korobov, E. S. Statnik, A. I. Salimon, A. M. Korsunsky ‘Comparative analysis of prior austenite grain size determination in high grade pipeline steels’, Fracture and Structural Integrity, 78 (2026) 474–488.

1. Context and problem

In high-strength pipeline steels, accurately determining the prior austenite grain size can be important because the parent boundary network governs transformation products, effective cleavage packet dimensions, and fracture propagation paths. Frantsuzov and co-workers benchmarked reconstructed electron backscatter diffraction (EBSD) datasets against traditional light optical metallography to evaluate prior austenite grain structures.

While parent-phase reconstruction using orientation relationships provides crystallographic resolution, concerns arise regarding how spatial dimensions are extracted from the resulting orientation maps. A widespread oversight in modern microscopy is confusing planar observations with true volumetric morphology, fundamentally neglecting the laws of stereology.

Concern: Reconstructed EBSD orientation maps identify two-dimensional sections (areal intercepts), not true three-dimensional grain entities. Treating each distinct colour-indexed polygon as an absolute measurement of grain size introduces systematic bias into both the reported distributions and subsequent cross-method comparisons.

2. Areal intercepts vs true grain size distributions

In the publication's statistical analysis (notably in Figure 8), the compiled frequency charts are presented as grain size distributions. From a rigorous metallographic perspective, these plots are not grain size distributions; they are collections of planar areal intercepts (A) or equivalent circle diameters (ECD = √(4A/π)) sliced through a spatial polyhedral network.

When a plane polishes through an assembly of equiaxed three-dimensional polyhedra:

A planar cut rarely passes through the true equatorial diameter of a grain; it intersects grains at arbitrary elevations, inherently generating an apparent population of small sections even in a monodisperse assembly.
A distribution of 2D section areas cannot be directly mapped one-to-one to a 3D volumetric grain size distribution without using formal stereological transformations (e.g. Saltykov or Scheil–Schwartz–Saltykov unfolded matrix corrections).
Without applying such stereological corrections, planar EBSD intercept data only yields an uncorrected profile distribution rather than an accurate 3D metric.

3. Flawed comparison: lineal intercepts versus 2D areal maps

The comparison between the mean lineal intercept (L̅) method and uncorrected reconstructed EBSD planar maps constitutes an error in quantitative microscopy:

Metric	Dimensionality	Physical meaning	Stereological relationship
Mean lineal intercept (L̅)	1D line through 3D volume	Direct measure of interfacial boundary area per unit volume (S_V).	Exact: S_V = 2 / L̅ (independent of grain shape assumptions).
Planar EBSD polygonal intercepts	2D area on planar section	Section profile area (A), not spatial volume (V).	Requires unfolding matrix conversion to infer 3D equivalent sphere/polyhedron diameter.

The classic mean lineal intercept method possesses direct thermodynamic and mechanical relevance because S_V = 2 / L̅ directly gives the total parent grain boundary surface per unit volume without requiring assumptions about polyhedral geometry. Consequently, L̅ serves as a sound parameter for Hall–Petch strengthening and transformation nucleation density.

Comparing raw lineal intercept measurements directly with uncorrected 2D EBSD profile diameters, under the presumption that the latter represents the ‘true’ grain size, inevitably leads to erroneous conclusions regarding technique accuracy. The discrepancy observed between the two methods does not simply reflect experimental error or etching failure; it is an intrinsic mathematical mismatch caused by contrasting dimensional projections.

Audio podcast Study guide

            Conclusion: To ensure sound pipeline alloy characterisation, reconstructed EBSD maps must be corrected using proper stereological procedures before being presented as true 3D spatial grain size distributions. Until such transformations are applied, comparisons between planar EBSD data and lineal intercept metrics remain fundamentally invalid.
        

References

I. Frantsuzov, Z. Dong, O. Korobov, E. S. Statnik, A. I. Salimon, A. M. Korsunsky: ‘Comparative analysis of prior austenite grain size determination in high grade pipeline steels’, Fracture and Structural Integrity, 78 (2026) 474–488.
E. E. Underwood: Quantitative Stereology, Addison-Wesley, Reading, MA, 1970.