In-situ investigation of austenite-ferrite phases in creep strength enhanced ferritic steels using neutron diffraction

Abstract

Creep-strength-enhanced ferritic steels, particularly 9%Cr steels such as P91 and P92, are widely used in high-temperature applications like power plants due to their superior mechanical properties, oxidation resistance and corrosion protection. However, their performance, especially in welded joints may be affected by the presence of δ-ferrite with a body-centred cubic (δ-BCC) crystal structure. The study focuses on understanding the behaviour of these steels in fully ferritic and high-temperature regions where austenite with a face-centered cubic (FCC) and δ-BCC phases coexist, with the main objective of understanding the mechanisms responsible for the retention of δ-BCC at room temperature, particularly in the weld metal.

The primary objective of this study was in-situ measurements of phase changes at high temperatures. The study employed neutron powder diffraction, neutron texture measurements, dilatometry, electron backscatter diffraction, optical microscopy and thermogravimetric analysis coupled with differential scanning calorimetry to analyse these transformations.

In-situ neutron powder diffraction was performed at temperatures ranging from room temperature to 1350°C allowing for direct observation of lattice parameters and phase fractions in real time. Dilatometry was used to measure length changes and phase transformation temperatures up to 1000°C, while thermogravimetric analysis coupled with differential scanning calorimetry provided insights into thermal events such as heat absorption and release up to 1000°C. Neutron texture measurements and electron backscatter diffraction were employed to examine the texture evolution and preferred crystallographic orientation in P91 tube at room temperature, and optical microscopy was used to investigate microstructural changes after the neutron powder diffraction thermal cycle.

For pure Fe, the α-ferrite (α-BCC) to FCC transition was detected at approximately 910°C using neutron powder diffraction and dilatometry. The phase transformation temperatures observed in pure Fe closely matched published values giving confidence in measurement of sample temperature.

The onset temperature of the α-BCC to FCC transformation, Ac1 and the completion, Ac3 were observed at 800–805°C and 885–890°C in P91 and 814–816°C and 928–930°C in P92, as revealed by neutron powder diffraction. The weighted-profile R-factor, Rwp values confirmed that P91 base metal showed more preferred crystallographic orientation effects than P92 base metal. Significant variations in lattice parameters were detected, particularly at temperatures exceeding 1200°C. The first formation of δ-BCC from FCC on heating was observed first in P92 base metal, at 1215–1220°C when compared to P91 base metal, at 1255–1260°C. Both P91 and P92 exhibited δ-BCC retention at room temperature, indicating incomplete phase transformation during cooling from δ-BCC to FCC at an average cooling rate of 300°C/h.

In the as-welded condition, both P91 weld metals contained ferrite, martensite and residual austenite at room temperature. The residual austenite in P91 weld metal 1 was retained on heating until fresh austenite formed at 765–770°C. In contrast, the residual austenite in P91 weld metal 2 completely disappeared at 655–660°C. It should be noted that P91 weld metal 2 contained approximately 8% retained FCC phase at room temperature, in contrast to about 1% in P91 weld metal 1. Ac1 and Ac3 were observed at 815–820°C and 920–922°C. There was no variation in the Ac4 temperature where δ-BCC started forming. The rate of change in the δ-BCC content with increasing temperature above the Ac4 temperature differed significantly between the different 9%Cr alloys, particularly among weld metals. This difference may constitute a significant aspect of the FCC–BCC transformation behaviour. On cooling, pure Fe completely transformed from austenitic to ferrite at 899–894°C and 877–872°C. The martensite start temperature, Ms of the 9%Cr steels ranged from 377 to 391°C and the martensite finish temperature, Mf ranged from 240 to 297°C.

Dilatometry results showed Ac1 and Ac3 transformation temperatures of 817°C and 880°C for P91 base metal, and 836°C and 923°C for P92 base metal. Differences in the rates of phase transformations and thermal expansion behaviour were also observed, with P91 exhibiting faster transformation than P92. The tangential method was used to calculate the phase fractions within the α-BCC + ɣ-FCC phase region. The dilatometry curve of the P91 weld metal 1 was complex, showing multiple transition regions which were difficult to analyse without the insights provided by the neutron powder diffraction results.

From the differential scanning calorimetry results, the phase transformation of α-ferrite to austenite was endothermic while for δ-BCC is exothermic. The differential scanning calorimetry curves showed Ac1 temperatures in P91 and P92 base metal close to that measured by neutron powder diffraction and dilatometry. Ac3 temperatures were subject to large variations.

Optical metallography highlighted notable microstructural differences between P91 and P92 base metals. As-received microstructure of P91 base metal consisted of martensite with coarse, prior-austenite grain boundaries. The as-received P92 base metal exhibited a finer microstructure of fully tempered martensite with sub-grains and a distribution of coarse and fine precipitates. Heating and subsequent cooling during neutron powder diffraction measurements significantly changed the microstructures of P91 and P92 base metals. The presence of δ-BCC in the microstructure of the 9%Cr steels could not be verified.

While neutron powder diffraction provided precise phase fraction data and lattice parameter measurements, it required complementary techniques to offer a comprehensive understanding of the phase transformations. Dilatometry effectively captured transformation temperatures but could not distinguish phases in dual-phase regions, especially in the weld metals. Thermogravimetric analysis coupled with differential scanning calorimetry was instrumental in identifying thermal events, though sensitivity limitations were observed at high temperatures. The findings of this study contribute to metallurgical knowledge by highlighting the significance of employing in-situ techniques to investigate the various sample parameters influenced during phase transformations.

The practical implications of these findings are significant for both alloy design and welding practices. The presence of residual austenite in the weld metal can influence the mechanical performance of the joint, particularly its toughness, hardness, and dimensional stability during and after post-weld heat treatment. The observation that residual austenite persists to relatively high temperatures in some welds underscores the need for precise control over weld composition and thermal cycles. Furthermore, the study shows that the formation and retention of δ-ferrite cannot be explained solely in terms of the Ac4 temperature, indicating that additional factors such as local compositional variations, segregation during solidification, and thermal gradients play a key role. These findings suggest that current predictive models based only on phase diagrams or Ac4 temperatures may be insufficient for weld metal behaviour and must be refined to account for kinetic and compositional effects during welding and post-weld heat treatment.