Unusual precipitation of amorphous silicon nitride upon nitriding Fe–2at.%Si alloy

Abstract

Silicon-nitride precipitation in ferritic Fe–2at.%Si alloy was investigated upon nitriding in NH₃/H₂ gas mixtures, using light microscopy, hardness measurements, scanning and transmission electron microscopy, X-ray diffraction and electron-probe microanalysis for microstructural characterisation. Surprisingly, ideally weak nitriding behaviour occurred upon nitriding thick (1 mm) recrystallised Fe–2at.%Si alloy specimens. This phenomenon can be attributed to the onset of silicon-nitride precipitation only after a certain degree of nitrogen supersaturation has been established at all depths in the specimen. Silicon-nitride precipitates formed inside the ferrite grains and also along the ferrite grain boundaries. The precipitates were amorphous and had a stoichiometry of Si₃N₄. The amorphous nature of the tiny precipitates has a thermodynamic origin. Nitride precipitation occurred very slowly due to the very large volume misfit of the nitride with the matrix. An anomalous non-monotonous hardness change occurred with increasing nitriding time, which was ascribed to the initially fully elastic accommodation of precipitate/matrix misfit. The nitrogen uptake rate increased upon continued nitriding as a result of “self-catalysis”. The possible, favourable application of amorphous silicon-nitride precipitates as grain-growth inhibitors in the production of grain-oriented electrical steel is discussed.

Keywords:

• nitriding
• Fe–Si alloy
• precipitation kinetics
• amorphous silicon nitride
• electrical steel

Acknowledgements

We are grateful to P. Kress and the late J. Koehler for assistance with the nitriding experiments, S. Haug for assistance with the EPMA measurements, W.-D. Lang for TEM sample preparation, S. Kühnemann for SEM investigations (all with the Max Planck Institute for Intelligent Systems).

Notes

Notes

1. This calculation is based on the crystalline modification of Si₃N₄. For the amorphous modification of Si₃N₄, the volume misfit will be even somewhat larger.

2. Consideration of N₂ desorption increases the calculated time needed for nitrogen saturation of the specimens by about 10%. This increase in the nitrogen saturation time does not affect the conclusions drawn in this work.

3. The amount of nitrogen removed during the denitriding treatment of the 30-h nitrided specimen (0.15at.%) is smaller than the equilibrium N solubility of unstrained pure ferrite (0.27at.% at 580°C and at a r _N of 0.104 atm^−1/2 41), which is attributed to lower nitrogen solubility of ferrite in the presence of dissolved Si 46,47.

4. Although several theories have been proposed for the mechanisms which can lead to the preferential, abnormal growth of “Goss” grains in electrical steel, none of them can explain completely the underlying mechanisms [51–53].