Microstructural overview

 Thermodynamic simulations

To understand the possible phases present in the investigated materials, thermodynamic simulations were performed prior to the microstructural examination and characterisation. Figure 4.1 shows the predicted weight fraction of phases present in a typical P92 steel (composition 94566) as a thermodynamically stable at temperatures greater than 710^oC.

 Matrix microstructure

All the P92 samples involved in this study are tempered martensitic steels.

The matrix exhibits a typical martensitic lath structure which is subject to coarsening and deformation after long term creep testing. The typical matrix lath structure is shown in Figure 4.2.

 Major second phase particles

The major second phase particles, M23C6 and MX carbonitride precipitates, are produced following a normalising and tempering heat treatment at temperatures in the region of 1050 – 1071^oC and 770 – 780^oC respectively.

Figure 4.3 shows the fine dispersion of the chromium-rich M23C6 precipitates in the head and gauge sections of the creep sample using the ion beam imaging method. The chemical composition of M23C6 was determined using a TEM-EDS technique (shown in Table 4.1). MX carbonitrides are rich in V and/or Nb and finer than M23C6 precipitates in size. The STEM-HAADF images show the typical appearance of MX particles in P92 steels (Figure 4.4). The detailed composition of MX particles in the head section was

measured, and the Cr/Nb/V ratios are plotted in the ternary graph in Figure 4.5. This shows that the MX particles have a variable V and Nb content after long term aging.

Laves phase precipitates can form during high temperature creep exposure.

They are W and Mo rich intermetallic particles which appear bright in the SEM backscattered electron mode, which allows them to be easily discriminated and quantified. Laves phase was not found in the non-creep tested or unaged samples. Figure 4.6 shows the typical distribution of Laves phase particles in the head and gauge sections after high temperature creep exposure, where bright Laves phase particles can be seen to preferentially decorate the lath, subgrain and prior austenite grain boundaries.

Table 4.1 The chemical compositions of the M23C6 particles measured using STEM-EDS on the carbon extraction replica samples of three suppliers (in wt.%). 20 random particles were analysed in each sample.

Particles from Cr W Mo Fe Mn V Nb Ni

Base Metal A 54.6 14.8 2.7 24.9 0.9 1.5 0.4 0.3

±0.6 ±0.7 ±0.1 ±0.5 ±0.1 ±0.6 ±0.4 ±0.0 Base Metal B 54.6 13.9 2.5 25.3 1.1 2.1 0.5 0.2

±0.6 ±0.6 ±0.1 ±0.7 ±0.1 ±0.8 ±0.3 ±0.0 Base Metal C 53.2 14.1 2.9 25.8 0.8 2.8 0.4 0.1

±0.7 ±0.5 ±0.1 ±0.8 ±0.1 ±1.0 ±0.2 ±0.1 P92 steel as received,

after Panait (2010)

54.2 12.0 3.6 26.4 3.5 0.3 - -

±0.3 ±0.2 ±0.1 ±0.2 ±0.2 ±0.0 - -

Figure 4.3 Ion beam induced SE images with IEE (insulator enhanced etch) XeF2 flow showing M23C6 particles in the (a) head section and (b) gauge length.

Figure 4.4 STEM-HAADF images at different magnifications (a) ×57000 and (b) ×115000 showing the typical size and appearance of MX particles in carbon extraction replica samples.

(a) Head section (b) Gauge section

(a) ×57000 (b) ×115000

M

C

MX

0 20 40 60 80 100 phase particles in the head section after long term creep exposure at 600^oC.

Figure 4.6 Examples of the BSE images used for quantitative analysis of Laves phase particles in (a) head section and (b) gauge length.

 Inclusion particles

There are three major types of inclusion particles found in the steels studied in this work - BN, MnS and Al2O3. According to the thermodynamic simulation, BN does not form in the temperature range 550 – 650^oC; MnS is stable at the

(a) Head section (b) Gauge section

creep temperature, whilst Al2O3 was not able to be predicted due to the reason that O was not included in the calculations in this case due to lack of appropriate thermodynamic data in the database designed primarily for bulk steel microstructure determination. The finding of the three inclusions in the non-creep virgin samples is evidence for the fact that they formed in the previous steelmaking process of the steels studied. The detailed phase identification and characterisation of the inclusions will be discussed later in this chapter.

 Reversible ferrite

Ferrite regions, different from martensitic matrix, were observed in some G92 steels provided by supplier A. EBSD analysis showed that there was no sub-structure within the non-martensitic ferrite region (Figure 4.7). By comparing the chemical constituents of the ferrite region and the martensitic matrix using EDS, as shown in Figure 4.8, the concentrations of W, Mo and V in the ferrite region are higher than the martensitic matrix. From detailed STEM-EDS analysis (Figure 4.9), a high density of W-rich particles were observed in the ferrite region. EBSD spot analysis shows that the diffraction of the W-rich particles in the ferrite region is the same as the Laves phase particles in the matrix though the ones in the ferrite are finer and denser (Figure 4.10). In addition, needle-like V-rich particles which form skeleton-like structures associated with W-rich particles and a few Nb-rich particles are also present in the ferrite region. Although the crystallographic information is difficult to obtain due to the very small particle size, chemical analysis showed that the needle-like particles are likely to be VN precipitates, whilst the Nb-rich particles are likely to be NbC precipitates according to their particle size and shape. In some cases, inclusion particles were observed associated with the ferrite region, which will be discussed in detail in Chapter 6.

The ferrite region appeared as either an individual grain or a string of grains.

The size of the typical ferrite grain is less than 10 µm in feret diameter. In some rare cases large grains can be up to several tens of micrometres. A precipitation free zone (PFZ) can sometimes be observed in the inner side near boundaries of ferrite regions, as indicted and arrowed in Figure 4.8 (a).

The characteristics of the ferrite regions found in this study conform to the description of the ‘speckled’ reversible ferrite in Grade 92 steels. The formation of this kind of ferrite, which is different from the stable irreversible ferrite (delta ferrite) formed at high temperature, is probably due to the relatively slow cooling rate leading to the diffusional phase transformation from austenite to ferrite and carbides. This speckled ferrite can be removed by means of a re-heat treatment. A detailed description of the two types of ferrite in Grade 92 steels can be found in the paper of Knezevic et al. (2014).

Figure 4.7 (a) the ferrite region found in sample SA8 gauge close to the head section during SEM examination; (b) IQ (image quality) + GB (grain boundary) and (c) IPF (inverse pole figure) EBSD mapping showing no sub-structure within the ferrite region.

=20 µm; BC+GB; Step=0.1451 µm; Grid688x684 =20 µm; Map5; Step=0.1451 µm; Grid688x684

(a)

(b) (c)

Figure 4.8 (a) SEM image showing the typical ferrite region in the P92 samples with PFZ arrowed; (b) overlaid EDS spectrum showing the difference of the two regions (ferrite region in yellow and matrix region in red); the EDS chemical constituents of the two regions are shown in Table 4.2.

Table 4.2 Comparison of chemical compositions of reversible ferrite and martensitic matrix (in wt.%).

V Cr Mn Fe Mo W

Ferrite region 0.4 9.6 0.4 84.8 0.7 4.1

Martensitic matrix 0.2 9.6 0.4 87.2 0.4 2.2

(a) (b)

PFZ

Figure 4.9 STEM-EDS maps on the thin foil lift-out of the ferrite region in sample A5 head section showing the chemical distribution of W, Nb, V, and N elements.

Figure 4.10 EBSD Kikuchi patterns of (a) the Laves phase particle in the matrix and (b) indexed pattern of (a) with a CI = 0.241; and (c) the W-rich particle in the ferrite and (d) indexed pattern of (c) with a CI = 0.255.