*3.1. Microstructures*

Since the qualitative microstructural characteristics of T92/T92 weldments are, in principle, symmetrically distributed with respect to the weld centerline, only one half part of the cross-weld microstructure was documented. Figure 4 shows the light-optical micrograph of T92/T92 weldment in conventionally tempered, i.e., PWHT-1 material state. It can be clearly seen that the weldment after the PWHT-1 exhibits a typical microstructural gradient consisting of individual microstructural sub-regions, i.e., BM, SC-HAZ, IC-HAZ, FG-HAZ, CG-HAZ, FZ, and WM. These microstructural sub-regions are generally formed of tempered martensitic-ferritic structures with different tempering grades of martensite, depending on the reached local peak temperatures (i.e., the temperature gradient) during the welding thermal cycle.

**Figure 4.** Light-optical micrograph of T92/T92 weldment after the tempering PWHT-1.

The microstructural transitions SC-HAZ/IC-HAZ, FG-HAZ/CG-HAZ, and FZ/WM are clear thanks to the observed differences in grain size and morphology. However, the microstructural transitions BM/SC-HAZ and IC-HAZ/FG-HAZ cannot be clearly differentiated by means of light optical microscopy and, thus, they are only roughly estimated in Figure 4. Figure 5 shows the light-optical micrograph of T92/T92 weldment in renormalized-and-tempered, i.e., PWHT-2 material state.

**Figure 5.** Light-optical micrograph of T92/T92 weldment after the renormalizing-and-tempering PWHT-2.

It can be seen that the weldment after the PWHT-2 shows quite homogenized microstructure as a consequence of the performed renormalization treatment. Within the renormalized-and-tempered weldment, only the regions of BM and WM can be clearly distinguished (Figure 5). This observation can be directly related to the homogenization effect of the applied PWHT-2 resulting in notable suppression of the original T92 HAZ microstructural gradient due to the performed renormalization. However, it should be noted that Figure 5 indicates also a certain microstructural refinement within the former CG-HAZ and WM regions compared to the microstructure of rest BM involving the renormalized-and-tempered regions of former SC-HAZ, IC-HAZ, and FG-HAZ. The detailed SEM-micrographs of individual microstructural zones of T92/T92 weldment after the PWHT-1 and PWHT-2 are shown in Figures 6 and 7, respectively.

**Figure 6.** SEM-micrographs of individual microstructural zones of T92/T92 weldment after the tempering PWHT-1: (**a**) BM; (**b**) SC-HAZ; (**c**) IC-HAZ; (**d**) FG-HAZ; (**e**) CG-HAZ; and (**f**) WM.

**Figure 7.** SEM-micrographs of individual microstructural zones of T92/T92 weldment after the renormalizing-and-tempering PWHT-2: (**a**) BM; (**b**) former SC-HAZ; (**c**) former IC-HAZ; (**d**) former FG-HAZ; (**e**) former CG-HAZ; and (**f**) WM.

The phase composition of normalized and tempered martensitic steels of T/P92 grade is generally known and consists of ferritic matrix and strengthening precipitates of intergranular M23C6 (M = Cr, Fe ... ) carbides and intragranular MX (M = V, Nb; X = C, N) carbo-nitrides [37–41]. The same phase composition is to be expected also in the currently studied T92/T92 weldments in both the conventionally tempered and renormalized-and-tempered PWHT conditions. Although predicted by the phase diagram in Figure 1, the precipitation of intermetallic Fe2(W,Mo) Laves phase is not to be expected in the currently studied material states (PWHT-1 and PWHT-2) due to insufficient time for its

creation related to slow di ffusion kinetics of the tungsten and molybdenum atoms in the ferrite solid solution. This assumption has already been evidenced in our several former studies [27–29] about the e ffect of PWHT conditions on microstructure and various properties of dissimilar T92/TP316H ferritic/austenitic weldments for high temperature applications.

By comparison of individual microstructures in Figures 6 and 7, it can be stated that the performed PWHT-2 did not induce full microstructural homogenization of studied weldment with respect to the grain size. Thus, from the observed microstructural characteristics in Figure 7, it cannot be explicitly judged about the appropriateness of the used PWHT-2. Although the original HAZ microstructural gradient has been considerably suppressed, some recognizable microstructural heterogeneity among former HAZ sub-regions is still to be observed. The originally fine-grained regions (IC-HAZ, FG-HAZ) related to PWHT-1 became notably coarse-grained after the PWHT-2. On the contrary, the originally coarse-grained regions (CG-HAZ and WM) became partly refined. The observed microstructural changes can be related to variant (non-uniform) microstructural evolution in individual sub-regions during the PWHT-2 due to pre-existing microstructural di fferences originated from the primary welding-induced microstructural changes. The IC-HAZ microstructure is formed in the region of BM heated up during the welding to inter-critical peak temperatures (i.e., the temperatures in Ac1-Ac3 range). Accordingly, the IC-HAZ is formed of fine-grained microstructure consisting of over-tempered (i.e., non-transformed) martensite (i.e., ferrite with coarsened precipitates of original undissolved carbides) and newly formed martensite created on cooling from fine-grained non-saturated austenite. The FG-HAZ microstructure is formed in the region of BM heated up during the welding to peak temperatures just above Ac3 up to about 1100 ◦C. After subsequent cooling, the resulting FG-HAZ consists of newly formed martensite created from fine-grained non-saturated austenite and coarsened precipitates of original undissolved carbides. As shown in Figure 7, after the PWHT-2 both the IC-HAZ and FG-HAZ microstructures exhibit pronounced grain growth which can be related to the lower pinning e ffect of the coarsened carbide precipitates on the grain boundaries. Thus, the evolution of pronounced grain coarsening in both originally fine-grained IC-HAZ and FG-HAZ microstructures during the renormalization seems to be associated with their low thermal stability in as-welded material condition, enhancing the thermodynamic driving force for the observed microstructural changes. In contrast to the IC-HAZ and FG-HAZ regions, the original CG-HAZ and WM regions are created as a result of on-cooling phase transformations from the highest peak temperatures reached during the welding thermal cycle. Thanks to carbide dissolution at considered peak temperatures, these regions consist of coarse grain structures with the highest level of transformation (i.e., martensitic) hardening. During renormalization at 1060 ◦C (Figure 1) of these primarily coarse-grained microstructures, thermodynamic conditions for further grain growth are rather unfavorable due to a relatively low renormalization temperature compared to the considered highest peak temperatures. Instead, thanks to the e ffect of carbon supersaturation in these martensitic microstructures, the creation of newly formed (small) austenite grains on heating to the renormalization temperature is thermodynamically favored. This transformation behavior during the weld renormalization followed by on-cooling phase transformations and final tempering is assumed to be the reason for the creation of partly refined microstructures of CG-HAZ and WM after the PWHT-2.

## *3.2. Mechanical Properties*

The e ffects of initial PWHT conditions in combination with subsequent electrolytic hydrogenation of the studied T92/T92 weldments on their room-temperature tensile properties are shown in Figure 8.

**Figure 8.** The effects of PWHT conditions and subsequent electrolytic hydrogenation on room-temperature tensile properties of investigated T92/T92 weldments: (**a**) strength properties and (**b**) deformation properties.

It can be seen that the renormalizing-and-tempering PWHT-2 of studied T92/T92 weldment resulted only in a small increase of the strength properties, i.e., the YS and UTS values, compared to those of the weldment after the tempering PWHT-1 (Figure 8a). Thus, it can be stated that the effect of various PWHT conditions on the resulting strength properties of studied weldment was rather insignificant. In addition, the differences in measured strength properties between the hydrogen-free and hydrogen-charged weldments in the both PWHT conditions were also quite negligible (Figure 8a). On the contrary, the renormalizing-and-tempering PWHT-2 of studied weldment resulted in a significant decrease of the deformation properties, i.e., the EL and RA values, compared to those of the weldment after the tempering PWHT-1 (Figure 8b). Additional hydrogen charging of the studied weldments in both PWHT material states led to further deterioration of their deformation properties (Figure 8b). However, the observed detrimental effect of the renormalizing-and-tempering PWHT-2 on the deformation properties of the weldments was much more pronounced in comparison with the effect of electrolytic hydrogenation. The measure of individual studied effects (i.e., the heat treatment procedure and electrolytic hydrogenation) on the plasticity deterioration can be quantitatively estimated using the so-called embrittlement index EI:

$$\text{EI} \ (0, \text{ x}) \ = \ \frac{RA\_0 - RA\_x}{RA\_0} \times 100\% \tag{1}$$

where *RA*0 and *RAx* are the values of reduction of area at fracture of two considered material states, and the subscripts "0" and "*x*" refer to the states selected as initial and final, respectively [30]. Thus, the calculated values of the embrittlement index using the average *RA* values (Figure 8b) are summarized in Table 2. From the calculated values of embrittlement index in Table 2, it is clear that the highest degree of embrittlement of studied weldments is caused by the application of renormalizing-and-tempering PWHT-2 (row 1). When comparing the effects of additional hydrogen charging, the higher measure of hydrogen embrittlement is indicated for the weldments processed by the PWHT-2 (row 3) compared to the weldments processed by the PWHT-1 (row 2).


**Table 2.** Embrittlement index for individual material states.

The obtained findings about the highly detrimental effect of the welds' homogenization treatment on their plastic properties can be explained by considering the microstructural changes induced by the renormalizing-and-tempering PWHT-2. The observed microstructural di fferences between the weldments in PWHT-1 and PWHT-2 material states (Figures 4–7) have crucial e ffects on the localization of plastic deformation during the tensile testing. In order to indicate local strain hardening behavior in the studied T92/T92 weldments during the tensile straining, c-w Vickers hardness measurements were carried out on the plain surfaces of longitudinal sections of fractured tensile samples after the tensile tests. Figures 9 and 10 show the c-w Vickers hardness profiles of the studied weldments initially heat treated using the PWHT-1 and PWHT-2, respectively. Visible interruptions within the c-w hardness profiles indicate the fracture locations of broken tensile test specimens after the room-temperature tensile tests. The c-w hardness profiles of the referential unstrained samples are also included for comparison. Significant di fferences between the c-w hardness profiles of T92/T92 weldments in various initial PWHT conditions are clearly visible when comparing Figures 9 and 10. By comparison of both unstrained samples (i.e., in PWHT-1 and PWHT-2 conditions), it can be concluded that the weldment after the tempering PWHT-1 shows a steep hardness gradient in its HAZ. The highest hardness values are measured in the WM and FZ regions which can be related to the highest measure of transformation (i.e., martensitic) hardening in these locations. On the other hand, the lowest hardness values are typically measured in the FG-HAZ, IC-HAZ, and partly SC-HAZ regions which are known to be locations with the greatest degradation of transformation hardening during the welding thermal cycle [10].

**Figure 9.** The effects of electrolytic hydrogenation and subsequent room-temperature tensile testing on cross-weld hardness profiles of T92/T92 weldments initially heat treated by conventional tempering PWHT-1: (**a**) overall hardness profiles and (**b**) detailed hardness profiles focused on the HAZ.

(**b**)

**Figure 10.** The effects of electrolytic hydrogenation and subsequent room-temperature tensile testing on cross-weld hardness profiles of T92/T92 weldments initially heat treated by renormalizing-and-tempering PWHT-2: (**a**) overall hardness profiles and (**b**) detailed hardness profiles focused on the former HAZ.

The hardness profiles of both the hydrogen-free and hydrogen-charged specimens show the strain-induced hardness peaks in the formerly soft (i.e., in the initial unstrained state) SC-HAZ in the location of tensile test fracture. Thus, the originally softer areas within the T92/T92 weldment in the PWHT-1 material state represent the locations of the localization of the highest plastic deformation during tensile testing.

The necking-related hardness peaks at fracture locations are the result of strain hardening due to the localization of plastic deformation after reaching plastic instability during tensile testing. This can be supported by the fact that the more ductile samples after the PWHT-1 (in both hydrogen-free and hydrogen-charged conditions, see Figure 9) show significant hardness peaks, compared to the less pronounced hardness peak related to the embrittled sample after the PWHT-2. Moreover, the hydrogenated sample after the PWHT-2 shows almost total suppression of its hardness peak in fracture location (see Figure 10). The localized plastic deformation can reasonably explain the better deformation properties of the highly heterogeneous T92/T92 weldment in the PWHT-1 material state, compared to the homogenized weldment in the PWHT-2 material state. The localized plastic deformation can be quantified by true fracture strain <sup>ε</sup>*f*, as it follows:

$$
\varepsilon\_f = \ln \frac{A\_0}{A\_f} \tag{2}
$$

where *A*0 and *Af* are the values of original cross-sectional area and final (i.e., minimal) cross-sectional area at the point of fracture of the tensile test specimen. The calculated ε*f* values indicating the level of accumulated plastic deformation at the location of final fracture are shown in Table 3:

**Table 3.** Calculated values of true fracture strain for studied weldment in individual material states.


In our case, the reason to cause localized plasticity is related to the tri-axial stress state (i.e., the stress triaxiality) induced by plastic instability at local microstructural heterogeneities (microstructural notches), namely the HAZ microstructural gradient and WM microstructural heterogeneity in the PWHT-1 and PWHT-2 material states, respectively. For the sake of completeness, it should be noted that for the weldments in PWHT-1 material state, the increased strain hardening effects were observed not only at the final fracture locations but also at the locations of the concurrent necking areas (Figure 9a). The reason for this behavior was to be expected since the c-w configuration of investigated tensile samples implied symmetrical occurrence of equivalent HAZ microstructural gradients at the both sides of welded base materials. The undeformed weldment after the renormalizing-and-tempering PWHT-2 shows equalized course of hardness values as a result of the weld microstructure homogenization (Figure 10). Moreover, the hardness profile of hydrogen-free weldment, initially subjected to the renormalizing-and-tempering PWHT-2, shows after the tensile test the localized strain hardening at the location of final failure in WM. The hydrogen-charged weldment shows the same failure location but without the occurrence of significant strain hardening at fracture location. The observed WM failures occurred likely due to the fact that the WMs represent the critical microstructural regions even after the performed homogenization treatment due to their original as-cast microstructure and thus higher impurity content and inhomogeneity compared to the BM. For comparison, Figure 11 shows four selected engineering stress-strain curves representing the overall tensile deformation behavior of the studied T92/T92 weldments in their individual material states with respect to the used PWHT conditions and hydrogen charging application. From the obtained results, it can be concluded that the tensile deformation and fracture behavior of studied T92/T92 weldments is more sensitive to the initial PWHT conditions than to the applied electrolytic hydrogenation. In spite of the occurrence of certain hydrogen embrittlement in the studied weldments, it has been shown that the thermal deterioration caused by the use of renormalizing-and-tempering PWHT-2 is much more significant. This statement is additionally supported by the fact that the final failure locations are primarily pre-determined by the used PWHT conditions, i.e., regardless of hydrogen charging application. The T92/T92 weldments after the tempering PWHT-1 were always broken in their SC-HAZs, whereas the weldments after the renormalizing-and-tempering PWHT-2 always fractured in their WMs.

**Figure 11.** The effects of PWHT conditions and subsequent electrolytic hydrogenation on tensile deformation behavior of investigated T92/T92 weldments.

Figure 12 shows representative photo-macrographs of longitudinal sections of the fractured tensile samples' counterparts showing two different failure locations of the studied T92/T92 weldments in their all investigated material states.

**Figure 12.** Photo-macrographs of fractured tensile samples' counterparts indicating various failure locations after room-temperature tensile tests of T92/T92 weldments in their individual material states: (**a**) "PWHT-1"; (**b**) "PWHT-1 + hydrogen"; (**c**) "PWHT-2"; (**d**) "PWHT-2 + hydrogen".

The detailed discussion on mutual correlation between the fractographic and microstructural characteristics of studied T92/T92 weldments will be provided within the following section.
