Engineering Failure Analysis 40 (2014) 141152

Contents lists available at ScienceDirect

Engineering Failure Analysis

journal homepage: www.elsevier.com/locate/engfailanal

Correlation between microstructure and creep performance

of martensitic/austenitic transition weldment in dependence
of its post-weld heat treatment
Ladislav Falat a,, Lucia Ciripov a, Jn Kepic a, Jir Burk b, Ivana Podstransk b
a
b

Institute of Materials Research, Slovak Academy of Sciences, Watsonova 47, SK 040 01 Koice, Slovak Republic
Institute of Physics of Materials, Academy of Sciences of the Czech Republic, Zizkova 22, CZ 616 62 Brno, Czech Republic

a r t i c l e

i n f o

Article history:
Received 13 December 2013
Received in revised form 7 February 2014
Accepted 25 February 2014
Available online 3 March 2014
Keywords:
Transition weldment: T92/TP316H
Post-weld heat treatment
Microstructure
Creep failure mode

a b s t r a c t
This paper deals with the inuence of post-weld heat treatment (PWHT) of T92/TP316H
martensitic/austenitic transition weldment on the resulting microstructure and creep
characteristics. Experimental weldments were fabricated by gas tungsten arc welding
using a nickel-based weld metal (Ni WM). After the welding, two individual series of
produced weldments were heat-treated according to two different PWHT procedures.
The rst conventional PWHT was carried out via subcritical tempering (i.e. bellow Ac1
temperature of T92 steel), whereas the other one, the so-called full PWHT consisted of
a complete reaustenitization of the weldments followed by water-quenching and nal
tempering. The use of conventional PWHT preserved microstructural gradient of T92
steel heat-affected zone (HAZ), consisting of its typical coarse-grained and ne-grained
subregions with tempered martensitic and recrystallized ferriticcarbidic microstructures
respectively. In contrast, the full PWHT led to the complete elimination of the original
HAZ via transformation processes involved, i.e. the reaustenitization and back on-cooling
martensite formation. The observed microstructural changes depending on the initial
PWHT conditions were further manifested by corresponding differences in the weldments
creep performance and their failure mode. The weldments in conventional PWHT state
ruptured after long-term creep tests by premature type IV failure within their recrystallized intercritical HAZs. On the contrary, the long-term creep behavior of the weldments
processed by full PWHT was characterized by their remarkable creep life extension but
also by the occurrence of unfavorable decohesion failure along T92/Ni WM interface.
2014 Elsevier Ltd. All rights reserved.

1. Introduction
The 9%Cr martensitic steels are used in power generation industry for thick-walled boiler components such as steam
headers and main steam piping because of the relatively low coefcient of thermal expansion and favorable cost compared
to the high-alloyed austenitic steels [1]. However, the austenitic steels with their excellent corrosion and creep resistance are
frequently used for construction of superheaters [2]. This indicates that the joining of martensitic and austenitic steels is
rather necessary in supercritical boiler constructions.

Corresponding author. Tel.: +421 55 7922447; fax: +421 55 7922408.

E-mail address: lfalat@imr.saske.sk (L. Falat).
http://dx.doi.org/10.1016/j.engfailanal.2014.02.018
1350-6307/ 2014 Elsevier Ltd. All rights reserved.

142

L. Falat et al. / Engineering Failure Analysis 40 (2014) 141152

During the rst implementation of transition weldments between ferritic (e.g. tempered bainitic or tempered martensitic) steels and austenitic steels, the use of austenitic steel type welding consumables was typical [3]. However, further experience with these welds revealed a serious problem related to the carbon diffusion processes across the ferritic/austenitic
weld interface, resulting in the formation of soft carbon-depleted zones, the so-called white bands at the ferritic side of
the welded joints with deteriorated creep strength. To prevent this problem, alternative nickel-based welding ller materials
have been invented [4]. The main benet of Ni-based weld metal (Ni WM) comes from its low carbon solubility, so it acts as a
carbon diffusion barrier. Moreover, thermal properties such as thermal conductivity and expansivity of Ni-based alloys lie in
a medium range between the corresponding properties of ferritic and austenitic steels. Thus the use of Ni WM is very suitable in transition weldments for the lowering and/or moderation of the gradient of residual welding stresses [5,6].
Many previous investigations, e.g. [79] were focused on the Ni-based transition joints between 2.25Cr1Mo ferritic
bainitic steel and AISI 300 series austenitic steels. Recently published studies [2,1015] include the results obtained for
the martensitic/austenitic weldments involving 9%Cr martensitic base materials. Nevertheless, literature concerning the
effects of the variation of post-weld heat treatment (PWHT) conditions on microstructure and creep behavior of martensitic/austenitic weldments is very limited. In general, the weldments of martensitic steels require an application of PWHT
in order to relieve residual stresses and stabilize the microstructure with strengthening precipitates [1618]. Conventional
way of PWHT of these weldments consists of subcritical tempering with respect to the steels Ac1 temperature, typically in
the range from 720 to 760 C [16]. It is also well-known that the conventional PWHT improves the weldments toughness
but the remaining problem is their premature type IV failure in the heat-affected zone (HAZ) during operation in creep
conditions [1,19,20]. Albert et al. [19] concluded that this failure mode cannot be suppressed by any variation of subcritical
PWHT conditions. Recently, Abe et al. [21] found out that the welded joints of newly developed 9Cr3W3Co0.2V0.05Nb
steel with 160 ppm boron and 85 ppm nitrogen exhibited no type IV failure in relatively short to medium-term creep
conditions as a result of specic modication of HAZ microstructure. However, long-term creep tests of this new promising
martensitic steel with modied B and N contents (from there MARBN steel) are still in progress [21]. On the other hand,
Tezuka and Sakurai [22] and Kimura et al. [23] suggested that a possible way to avoid type IV failure of the weldments is
their full renormalization. In a specic case of the weldments between martensitic and austenitic steels, PWHT conditions
are commonly specied according to the conventional (subcritical) procedure with regard to the martensitic base material
of welded joint [24]. However, available information about the application of full PWHT for martensitic/austenitic
weldments is rather scarce [2527].
The present study represents an extended and continuing research work of the former study by Falat et al. [25]. It summarizes the results regarding the effects of conventional as well as full PWHT on microstructure and creep performance
of T92/TP316H transition weldments.

2. Experimental procedure
Dissimilar steels T92 and TP316H in the form of tubes with outer diameter of 38 mm and wall thickness of 5.6 mm were
welded by gas tungsten arc welding (GTAW) using Ni-based ller alloy Nirod 600. The electrode diameter was 2.4 mm and
the applied welding parameters were: welding current 70110 A, voltage 1217 V and heat input 912 kJ/cm. Chemical
compositions of the individual materials used for fabrication of T92/TP316H weldments are listed in Table 1.
After the GTAW process, two different PWHT procedures were applied to the rst and second series of the produced weldments respectively. The conventional PWHT consisted of subcritical tempering at 760 C for 1 h, followed by air cooling in
furnace to room temperature. The second series of the weldments was subjected to the so-called full PWHT including a
complete reaustenitization at 1060 C for 15 min with subsequent water-quenching and subcritical tempering at 760 C
for 1 h and air cooling down in furnace. Detailed PWHT diagrams of both the used procedures are shown in Fig. 1.
All experimental work was performed using cross-weld (c-w) samples. Prepared tubular weldments were cut into the c-w
blocks, as schematically shown in Fig. 2. The creep tests were performed using cylindrical tensile samples with a gauge
length of 40 mm, body diameter of 4 mm, and M6 head thread. Metallographic analyses involved light microscopy (LM),
scanning electron microscopy (SEM) with energy dispersive X-ray (EDX) spectroscopy, and transmission electron microscopy (TEM). Etched metallographic samples were used for LM and SEM analyses. The used etching solutions were specied
in [25]. Thin foils for TEM observations were prepared using focused ion beam (FIB) technique, applied perpendicularly onto
the longitudinally sectioned tensile creep specimens at the locations immediately beneath their creep fractures. Thermodynamic calculations of phase equilibria were performed using the software Thermo-Calc [28] and thermodynamic database
STEEL16 formulated by Kroupa et al. [29].

Table 1
Chemical composition (wt.%) of individual materials of the dissimilar weldment.

T92
Nirod 600
TP316H

Si

Mn

Cr

Mo

Ni

Al

Nb

Fe

0.11
0.05
0.052

0.056

0.38
0.3
0.51

0.49
3.0
1.77

0.019
0.03
0.031

0.002
0.015
0.006

9.08
20.0
16.76

0.31

2.05

1.57

0.0023

0.33
Balance
11.13

0.014

0.2

0.069
2.0

Balance
2.0
Balance

L. Falat et al. / Engineering Failure Analysis 40 (2014) 141152

143

Fig. 1. Schematic timetemperature diagrams clarifying the conditions of individual PWHT procedures: conventional PWHT (a) and full PWHT (b).

Fig. 2. Schematic sampling of cross-weld (c-w) blocks from the tubular weldments.

3. Results and discussion

3.1. The effect of PWHT conditions on microstructure and hardness
Figs. 36 show the microstructures of individual regions of the studied T92/TP316H weldments after the application of
different PWHT procedures. In Fig. 3a and b the microstructures of martensitic (T92) region close to the Ni WM are shown
in conditions after the conventional and full PWHT respectively. The heat-affected zone (HAZ) of the T92 base material
(T92 BM) consists of coarse-grained HAZ (CGHAZ), ne-grained HAZ (FGHAZ), intercritical HAZ (ICHAZ) and subcritical HAZ
(SCHAZ) subregions as a result of phase transformations taking place in T92 BM in dependence of a local temperature
reached during the welding (see Fig. 3a). The subregions of ICHAZ and FGHAZ are both essentially ne-grained because of
a very short duration of a ? c matrix transformation processes hindering the growth of austenite grains [16]. On the
contrary, microstructure in HAZ of T92 BM became coarse-grained and homogeneous after the full PWHT (see Fig. 3b)
as a result of transformation processes involved during the complete reaustenitization and back on-cooling fresh martensite
formation. This retransformed martensitic microstructure shows signicant coarsening with respect to its prior austenitic
grains, even in comparison to the CGHAZ microstructure after the conventional PWHT (see the details in Fig. 4). Although
microstructure of T92 steel after different PWHT procedures differs signicantly in the prior austenite grain size (see Fig. 4),
its phase composition in both PWHT states is basically the same, consisting of tempered martensite, i.e. ferriticcarbidic

144

L. Falat et al. / Engineering Failure Analysis 40 (2014) 141152

Fig. 3. Microstructures of martensitic (T92) region of transition T92/TP316H weldment in conditions after the conventional PWHT (a) and full PWHT
(b).

Fig. 4. Detailed microstructures of the martensitic region of the T92/TP316H weldment in conditions after the conventional PWHT (a) and full PWHT
(b).

Fig. 5. Microstructures of Ni WM (Nirod 600) near the interface with T92 steel in conditions after the conventional PWHT (a) and full PWHT (b). After
the selective etching of Ni WM, marginal T92 regions appear dark due to their over-etching.

L. Falat et al. / Engineering Failure Analysis 40 (2014) 141152

145

Fig. 6. Microstructures of austenitic (TP316H) region of the T92/TP316H weldment in conditions after the conventional PWHT (a) and full PWHT (b).

mixture. In accordance with the literature [16,30], the microstructures of Grade 92 steels typically contain two major
types of precipitates, namely the intergranular M23C6 (M = Cr, Fe) carbides and intragranular MX (M = V, Nb; X = C, N)
carbonitrides.
The microstructures of Ni WM (Nirod 600) in the vicinity of their interface with T92 steel are shown in both PWHT states
in Fig. 5. The microstructures of Nirod 600 weld metal are markedly heterogeneous with respect to the size, morphology and
distribution of grain boundaries and precipitates. In the state after the conventional PWHT (Fig. 5a) the Ni WM microstructure revealed well recognizable solidication grain boundaries and dendritic cell sub-boundaries. On the contrary, after the
full PWHT (Fig. 5b) the microstructure of Nirod 600 contains elongated grains with clearly visible straightened boundaries
as a result of their diffusional migration during the weldment austenitization and carbide precipitation. The size of precipitates in Ni WM after the full PWHT (Fig. 5b) is markedly smaller in comparison to that after the conventional PWHT
(Fig. 5a). This renement of the particles is likely to originate from the dissolution of precipitates during reaustenitization
and reprecipitation during subsequent subcritical tempering. The microstructures in Fig. 5 contain intragranular as well
as intergranular precipitates with different sizes and morphological appearance. According to the previous investigation
[25], both intragranular and grain-boundary precipitates were found to be rich in Nb by means of EDX analyses. This result,
with respect to the overall Ni WM composition, indicated these precipitates to be the NbC carbides. The nest precipitate
agglomerations were analyzed using EDX and electron diffraction on carbon extraction replicas and they were also found
to be the Nb-rich MC type precipitates with face-centered cubic crystal structure [25].
In Fig. 6 the microstructures of austenitic steel (TP316H) region adjacent to the Ni WM are compared for the conventional and full PWHT state. With respect to the grain size, the austenitic microstructures (Fig. 6) were generally less
affected by PWHT conditions compared to the martensitic ones (Fig. 4). Etching revealed a narrow band of coarse grains
fragmented by dendritic cell sub-boundaries along the fusion line in conditions after conventional PWHT (see Fig. 6a). After
the full PWHT, the dendritic band within the fusion zone disappeared as a result of homogenization and the rest of austenitic microstructure became slightly coarser (Fig. 6b). Both austenitic microstructures contain recrystallized polygonal grains
with mostly intergranular M23C6 (M = Cr, Fe) carbides which are known to be the major type of precipitates in non-stabilized
austenitic steels [31].
In order to evaluate the inuence of different PWHT conditions on the local mechanical properties of the studied weldment, cross-weld hardness measurements were carried out on polished metallographic samples (Fig. 7). The hardness prole

Fig. 7. Cross-weld hardness proles of the T92/TP316H weldment after different PWHT procedures.

146

L. Falat et al. / Engineering Failure Analysis 40 (2014) 141152

Fig. 8. Schematic formation of individual T92 HAZ subregions in correlation to the equilibrium phase diagram calculated using software Thermo-Calc [28]
and thermodynamic database STEEL16 [29]. The T92 steel composition is indicated by the corresponding isopleth at 0.11 wt.% C (vertical dash-line in the
phase diagram).

of T92/TP316H weldment in the state after the conventional PWHT exhibits a steep hardness gradient within the region of
T92 HAZ which reasonably reects the local microstructural characteristics of the individual T92 HAZ subregions (Fig. 3a).
The highest hardness was measured close to the fusion zone (FZ) of the T92 steel and Ni WM. This hardness peak can be
related to the transformation hardening, i.e. fresh martensite formation in the FZ during cooling after the welding [16]
as well as to the additional alloying originated from a mixture of partially melted T92 steel and Nirod 600 weld metal. Transformation hardening effects were also strongly manifested within the region of T92 CGHAZ heated up to the temperatures
below the range of solidication temperatures but still well above the Ac3 transformation temperature (see Fig. 8). Further
decreasing of the reached peak temperature of T92 HAZ down to the Ac3 at increasing distance from the weld FZ resulted in a
gradual decrease in hardness. In this region a short heating period allowed the austenite formation without its saturation
with alloying elements from the secondary carbide precipitates. Moreover, the undissolved carbides caused efcient hindering of grain growth within the T92 FGHAZ. The lowest hardness values were measured within the region of T92 ICHAZ
heated up to the temperatures between Ac1 and Ac3 as a result of severe microstructure degradation during the welding
thermal cycle. Primarily, this deterioration was caused by the grain-renement originating (as in the case of T92 FGHAZ)
from the non-saturated austenite formation during short-term heating and subsequent on-cooling formation of ne polygonal ferrite instead of lath martensite. The second contribution to the microstructure degradation of T92 ICHAZ was related
to the additional tempering of the remaining untransformed (originally tempered) martensite, i.e. the so-called doubletempering effect [16]. The hardness values of martensitic (T92) part of the studied weldment in the state after the full
PWHT are mostly lower, compared to those of the weldment in conventional PWHT state. In addition, the original hardness gradient of the former T92 HAZ was completely suppressed after the full PWHT and the hardness within the whole
T92 region was invariable (see Fig. 7). This behavior can apparently be related to the elimination of the former T92 HAZ during the reaustenitization period of the full PWHT (Fig. 3b). On the other hand, the increased hardness values of Ni WM and
TP316H regions after the full PWHT can likely be associated with redistribution of alloying elements and subsequent reprecipitation of ne particles during the tempering period of this heat treatment (Fig. 5b and Fig. 6b).

3.2. The effect of PWHT conditions on creep life and failure mode
Fig. 9 shows the dependencies of creep-rupture life (time to fracture) of the studied weldment on the applied stress and
used PWHT conditions for two creep testing temperatures. With increasing stress and/or temperature the fracture time of
the weldments in both the PWHT states decreases. In addition, Fig. 9 indicates that for all the applied stresses at a given
creep temperature, the fracture times of the weldments in the state after the full PWHT are always higher than those
of the weldments after the conventional PWHT. However, with decreasing applied stress the difference in creep rupture
life between the weldments in different PWHT states decreases. This effect is more signicant at a higher creep temperature.
Fig. 10 claries the effects of creep conditions (applied stress and creep temperature) and the used PWHT procedure on
the weldments failure mode [25]. Metallographic samples of cross-weld sections of ruptured specimens indicating typical
failures for conventional and full PWHT conditions are shown in Figs. 11 and 12 respectively. In accordance with Fig. 10
it can be stated that the long-term creep failure mode of the weldments processed by conventional PWHT was the type IV
failure within ne-grained T92 ICHAZ (see also Fig. 11b). On the other hand, the long-term creep failure mode of the
weldments processed by full PWHT was characterized by the decohesion failure along T92/Ni WM interface (see Figs. 10
and 12b). Signicant differences in local plasticity close to the fracture were observed for different specimens. Therefore

L. Falat et al. / Engineering Failure Analysis 40 (2014) 141152

147

Fig. 9. Creep fracture times of the T92/TP316H weldments in dependence of the applied stress, initial PWHT procedure and creep testing temperature:
625 C (a) and 650 C (b). Creep-rupture data according to [25].

Fig. 10. Occurrence of different failure mode types in T92/TP316H weldments in dependence of the used creep conditions and initial PWHT procedure:
conventional PWHT (a) and full PWHT (b).

Fig. 11. The creep failure types of the T92/TP316H weldment processed by the conventional PWHT: abrupt over-load failure at T92/Ni WM interface (a)
and type IV failure in T92 ICHAZ (b).

148

L. Falat et al. / Engineering Failure Analysis 40 (2014) 141152

Fig. 12. The creep failure types of the T92/TP316H weldment processed by the full PWHT: ductile failure in T92 BM (a) and brittle decohesion failure
at T92/Ni WM interface in long-term creep conditions (b).

Fig. 13. Reduction area (RA) of the T92/TP316H weldment depending on the time to fracture, PWHT procedure and creep temperature: 625 C (a) and
650 C (b).

reduction area (RA) was evaluated in dependence of time to rupture in order to assess the susceptibility to local embrittlement of the weldments (Fig. 13). For both the PWHT procedures, the measured RA values of T92/TP316H weldments show a
decreasing tendency with increasing creep fracture time. The only deviation from this rule was observed for the creep test
with the shortest time to rupture of weldment after conventional PWHT (Fig. 13a), which revealed over-load failure at
T92/Ni WM interface (Fig. 11a). However, a general comparison of RA values corresponding to the weldments in different
PWHT states indicates that the application of full PWHT results in much greater local embrittlement than that of the weldments processed by conventional PWHT. In accordance with Fig. 11b it is obvious that the type IV failure occurring in
T92 ICHAZ region was still characterized by considerable plasticity at fracture, whereas the decohesion failure at T92/Ni
WM interface (Fig. 12b) was completely brittle. This nding discredits the benets of creep life extension via the full PWHT
performed under the conditions of the present study.
3.3. Correlation between microstructure degradation and creep failure
Fig. 14 shows detailed SEM-images of T92 ICHAZ microstructure near the fracture surface after the type IV failure. It is
generally known [32,33] that the microstructure degradation in this region is caused by the welding process and subsequent

L. Falat et al. / Engineering Failure Analysis 40 (2014) 141152

149

Fig. 14. SEM-images of typical T92 ICHAZ microstructure in the region of type IV creep failure of T92/TP316H weldment, initially processed by the
conventional PWHT: secondary electrons (a) and back-scattered electrons (b) micrographs.

creep exposure. Detrimental effects during the creep are enhanced by gradual decrease in dislocation density and coarsening
of secondary phase precipitates [34,35]. Microstructural damage accompanied by nucleation and growth of creep voids is
pronounced by a multiaxial stress state [20,32] originating from the mechanical constraint in T92 HAZ region with lower
creep strength surrounded by the regions with higher creep strength (i.e. base material and weld metal). The microstructure
in Fig. 14 clearly reveals a crucial inuence of coarsened grain-boundary precipitates on the intergranular fracture characteristics of this failure type. The individual precipitates were differentiated using an atomic-number contrast of backscattered electrons (see Fig. 14b).
Fig. 15 shows detailed SEM-images of the T92/Ni WM interface after the nal decohesion failure. It is obvious that the
observed interfacial decohesion is closely related to a progressive growth and interlinking of individual precipitates along
the T92/Ni WM interface. With respect to their chemical composition, these interfacial precipitates are mostly M23C6
carbides and occasionally also Fe2(W,Mo) based Laves phase. An occurrence of interfacial precipitates in several types of
transition weldments has also been reported in many other studies, e.g. [6,36,37] and it has been ascribed to the alloying
elements redistribution across the weld metal interfaces [38,39].
With respect to the observed differences in microstructures and creep failure types of the studied T92/TP316H weldments, cross-weld hardness measurements were carried out to estimate their local mechanical properties in dependence
of initial PWHT procedure and subsequent creep exposure. Typical hardness proles of the weldments after the individual
creep failures are shown in Fig. 16. The general hardness proles tendencies of creep-exposed weldments (Fig. 16) are
mostly very similar to those of the weldments in their corresponding PWHT states without creep exposure (Fig. 7). Thus,
clear differences between the hardness proles corresponding to the individual PWHT states remain preserved after the
creep expositions. Irrespective of applied creep conditions, the hardness proles of the weldments processed by conventional PWHT show a steep hardness gradient within their T92 HAZ regions (Fig. 16a). In comparison to the corresponding
initial PWHT state (Fig. 7), the hardness gradients within T92 HAZs are even more pronounced after the creep testing
(Fig. 16a) as a result of continuing microstructure degradation involving the effects of precipitate coarsening as well as
gradual decreasing in dislocation density of tempered martensite due to the dynamic recovery processes [15,16].
The hardness proles of creep-exposed weldments initially processed by the full PWHT (Fig. 16b) differ vastly from
those of the weldments initially processed by the conventional PWHT (Fig. 16a). The hardness values within T92 region
after the full PWHT do not exhibit any strong variations among the individual creep-exposed states. In the case of

Fig. 15. Detailed SEM-images of typical T92/Ni WM interfacial microstructure in the region of decohesion failure of T92/TP316H weldment, initially
processed by the full PWHT: secondary electrons (a) and back-scattered electrons (b) micrographs.

150

L. Falat et al. / Engineering Failure Analysis 40 (2014) 141152

Fig. 16. Cross-weld hardness proles of the T92/TP316H weldments ruptured by individual creep failure modes in dependence of creep conditions and
initial PWHT procedure: conventional PWHT (a) and full PWHT (b). Interruptions in the hardness proles indicate the locations of individual creep
failures.

Fig. 17. Substructural characteristics of: T92 ICHAZ of the T92/TP316H weldment ruptured by type IV creep failure after 4060 h at 625 C, 100 MPa (a) and
T92 region near the T92/Ni WM interface of the weldment ruptured by decohesion failure at T92/Ni WM interface after the creep exposure at 625 C,
120 MPa for 5143 h (b).

long-term creep tests, the creep failure locations correspond reasonably with their most signicant microstructural damage
at the T92/Ni WM interface (Fig. 15).
Since the long-term creep failures of the investigated weldments always occurred in their martensitic parts, i.e. either in
T92 ICHAZ or at T92/Ni WM interface, these regions were subsequently subjected to the detailed substructural characterization (see Fig. 17). The TEM observations were performed on creep-exposed samples with comparable creep-rupture lives for
the individual PWHT states. Sampling procedure for TEM was carried out using FIB technique in the locations immediately
beneath individual creep fractures. In the case of creep-exposed weldment, initially processed by the conventional PWHT,
the T92 ICHAZ region was investigated (see Fig. 17a). On the other hand, in the case of creep-exposed weldment, initially
processed by the full PWHT, the TEM analysis was performed within T92 steel region next to the T92/Ni WM interface
(Fig. 17b). The substructure of T92 ICHAZ region after the creep exposure for 4060 h at 625 C, 100 MPa shows completely
recrystallized polyhedric grains with signicantly coarsened precipitates (Fig. 17a). On the contrary, the substructure of T92
interfacial region after the creep exposure for 5143 h at 625 C, 120 MPa exhibits lath-like tempered martensite with much
ner precipitates (Fig. 17b) in comparison to the previous case (Fig. 17a). Thus, the observed differences between individual
substructures (Fig. 17) of different T92 regions of the studied weldments correspond well with the differences in their creep
performance and failure mode characteristics in dependence of the used PWHT procedure. Despite the observed benecial
effects of the used full PWHT on the creep life increase of the studied T92/TP316H weldments via type IV failure
suppression, it is necessary to search for possibilities of creep failure mechanism modication from the completely brittle
interfacial decohesion towards less brittle failures.

4. Summary and conclusions

Variations in microstructures and creep performance of T92/TP316H martensitic/austenitic weldment were investigated
depending on the applied conditions of post-weld heat treatment (PWHT). The obtained results are summarized in the
following conclusions:

L. Falat et al. / Engineering Failure Analysis 40 (2014) 141152

151

 After the conventional PWHT the microstructure of the T92 base material adjacent to the interface with Ni-based weld
metal (Ni WM) shows a typical microstructural gradient. On the contrary, after the full PWHT the part of weldment
made of T92 steel does not contain such a structured HAZ as a result of complete reaustenitization. In addition, the microstructure is signicantly coarser with respect to prior austenite grain size in comparison to the original HAZ.
 Solidication grains with intergranular and intragranular NbC precipitates occur in microstructure of Ni WM (Nirod 600)
in conditions after the conventional PWHT. After the full PWHT the microstructure of Ni WM is signicantly affected
by diffusive migration of grain boundaries and the processes of NbC carbides dissolution and their back reprecipitation
during the individual stages of performed heat treatment. This resulted in a signicant increase in density of precipitates.
 The austenitic microstructures of TP316H steel base material are generally less affected by PWHT conditions in comparison to the martensitic microstructures of T92 base material. However, after the full PWHT the microstructure is
homogenized and slightly coarsened in grain size.
 In long-term creep conditions, the weldments with the initial application of conventional PWHT procedure ruptured
prematurely by the type IV failure mode within their T92 ICHAZ regions. On the contrary, the weldments initially processed by the full PWHT ruptured after their long-term creep tests by the decohesion failure within the fusion zone
along T92/Ni WM interface.
 Regardless of the applied creep conditions, the hardness proles of the weldments initially processed by the conventional PWHT always showed steep hardness gradients within their T92 HAZ regions. Moreover, these gradients of
creep-exposed weldments were even more pronounced, compared to the hardness prole corresponding to the initial
PWHT state. In contrast, the hardness values within T92 regions of the weldments initially processed by the full PWHT
did not exhibit any strong variations among the individual creep-exposed states.
 In the case of long-term creep tests, the creep failure locations of the weldments initially processed by the conventional
PWHT were associated with the most remarkable microstructural deterioration of T92 ICHAZ regions with the lowest
measured hardness values within the whole T92 regions of the investigated weldments. On the other hand, the creep failures of the long-term creep-exposed weldments initially processed by the full PWHT were related to their signicant
microstructural damage at the T92/Ni WM interfaces, but their eventual correlations with performed cross-weld hardness
measurements were not revealed by the used experimental approach.
 Comparison of substructural characteristics from the individual creep failure locations indicated their good correlations
with creep life of the weldments processed by different initial PWHT procedures. Specically, the substructure of T92
ICHAZ region after the long-term creep exposure exhibits fully recrystallized polygonal grains with markedly coarsened
precipitates. In contrast, after the comparable long-term creep exposure, the substructure of T92 interfacial region with
the Ni WM still contains highly tempered martensitic laths with relatively ner precipitates.
 The creep life of weldment processed by the full PWHT was signicantly longer than that of the weldment processed by
the conventional PWHT procedure. The observed creep life extension of the weldments was directly related to their
type IV failure elimination, due to the complete suppression of their former T92 HAZ microstructural gradient during
the full PWHT. However, the occurrence of unfavorable decohesion failure along T92/Ni WM interface in long-term
creep conditions apparently discredited the achievements in creep performance of the studied martensitic/austenitic
weldments obtained by the full PWHT.

Acknowledgement
Financial support by the Slovak Scientic Grant Agency (VEGA) under the Grant No. 2/0116/13 is gratefully acknowledged.
References
[1] Shibli A, Starr F. Some aspects of plant and research experience in the use of new high strength martensitic steel P91. Int J Press Ves Pip 2007;84(1
2):11422.
[2] Cao J, Gong Y, Zhu K, Yang Z-G, Luo X-M, Gu F-M. Microstructure and mechanical properties of dissimilar materials joints between T92 martensitic and
S304H austenitic steels. Mater Des 2011;32(5):276370.
[3] Celik A, Alsaran A. Mechanical and structural properties of similar and dissimilar steel joints. Mater Charact 1999;43(5):3118.
[4] Parker JD, Stratford GC. Characterisation of microstructures in nickel based transition joints. J Mater Sci 2000;35(16):4099107.
[5] Joseph A, Rai SK, Jayakumar T, Murugan N. Evaluation of residual stresses in dissimilar weld joints. Int J Press Vessels Pip 2005;82(9):7005.
[6] Samal MK, Seidenfuss M, Roos E, Balani K. Investigation of failure behavior of ferriticaustenitic type of dissimilar steel welded joints. Eng Fail Anal
2011;18(3):9991008.
[7] Laha K, Chandravathi KS, Rao KBS, Mannan SL, Sastry DH. An assessment of creep deformation and fracture behaviour of 2.25Cr1Mo similar and
dissimilar weld joints. Metall Mater Trans 2001;A32:11524.
[8] Parker JD, Stratford GC. The high-temperature performance of nickel-based transition joints I. Deformation behaviour. Mater Sci Eng 2001;A299(1
2):16473.
[9] Parker JD, Stratford GC. The high-temperature performance of nickel-based transition joints II. Fracture behaviour. Mater Sci Eng 2001;A299(1
2):17484.
[10] Lee H-Y, Lee S-H, Kim J-B, Lee J-H. Creep-fatigue damage for a structure with dissimilar metal welds of modied 9Cr1Mo steel and 316L stainless
steel. Int J Fatigue 2007;29:186879.
[11] Cao J, Gong Y, Yang Z-G, Luo X-M, Gu F-M, Hu Z-F. Creep fracture behavior of dissimilar weld joints between T92 martensitic and HR3C austenitic
steels. Int J Press Vessels Pip 2011;88(23):948.
[12] Chen G, Song Y, Wang J, Liu J, Yu X, Hua J, et al. High-temperature short-term tensile test and creep rupture strength prediction of the T92/TP347H
dissimilar steel weld joints. Eng Fail Anal 2012;26:2209.

152

L. Falat et al. / Engineering Failure Analysis 40 (2014) 141152

[13] Cao J, Gong Y, Yang Z-G. Microstructural analysis on creep properties of dissimilar materials joints between T92 martensitic and HR3C austenitic steels.
Mater Sci Eng 2011;A528(1920):610311.
[14] Chen G, Zhang Q, Liu J, Wang J, Yu X, Hua J, et al. Microstructures and mechanical properties of T92/Super304H dissimilar steel weld joints after hightemperature ageing. Mater Des 2013;44:46975.
[15] Falat L, Svoboda M, Vyrostkov A, Petryshynets I, Sopko M. Microstructure and creep characteristics of dissimilar T91/TP316H martensitic/austenitic
welded joint with Ni-based weld metal. Mater Charact 2012;72:1523.
[16] Cerjak H, Mayr P. Creep strength of welded joints of ferritic steels. In: Abe F, Kern T-U, Viswanathan R, editors. Creep-resistant
steels. Cambridge: Woodhead Publishing Ltd.; 2008. p. 47298.
[17] Paradowska AM, Price JWH, Kerezsi B, Dayawansa P, Zhao X-L. Stress relieving and its effect on life of welded tubular joints. Eng Fail Anal
2010;17:3207.
[18] Milovic Lj, Vuherer T, Blacic I, Vrhovac M, Stankovic M. Microstructures and mechanical properties of creep resistant steel for application at elevated
temperatures. Mater Des 2013;46:6607.
[19] Albert SK, Matsui M, Watanabe T, Hongo H, Kubo K, Tabuchi M. Variation in the Type IV cracking behaviour of a high Cr steel weld with post weld heat
treatment. Int J Press Vessels Pip 2003;80:40513.
[20] Francis JA, Mazur W, Bhadeshia HKDH. Type IV cracking in ferritic power plant steels. Mater Sci Technol 2006;22:138795.
[21] Abe F, Tabuchi M, Tsukamoto S, Shirane T. Microstructure evolution in HAZ and suppression of Type IV fracture in advanced ferritic power plant steels.
Int J Press Vessels Pip 2010;87(11):598604.
[22] Tezuka H, Sakurai T. A trigger of Type IV damage and a new heat treatment procedure to suppress it. Microstructural investigations of long-term exservice CrMo steel pipe elbows. Int J Press Vessels Pip 2005;82:16574.
[23] Kimura K, Watanabe T, Hongo H, Yamazaki M, Kinugawa J, Irie H. Effects of full annealing heat treatment on long-term creep strength of 2.25Cr1Mo
steel welded joint. Quarterly J Japan Weld Soc 2003;21:195203.
[24] Msch H, Meyer H, Remmert H, Schlter N. Erprobung von Austenit-Ferrit Verbindungen der Werkstoffe X3CrNiMoN 1713 (1.4910) und X10CrMoVNb
9 1 (1.4903) fr den Einsatz in modernen Hochtemperaturkraftwerken. VGB Kraftwerkstechnik 1997;77:85866.
[25] Falat L, Vyrostkov A, Svoboda M, Milkovic O. The inuence of PWHT regime on microstructure and creep rupture behaviour of dissimilar T92/TP316H
ferritic/austenitic welded joints with Ni-based ller metal. Kovove Mater 2011;49(6):41726.
[26] Rieth M, Rey J. Specic welds for test blanket modules. J Nucl Mater 2009;386388:4714.
[27] Widak V, Dafferner B, Heger S, Rieth M. Investigations of dissimilar welds of the high temperature steels P91 and PM2000. Fusion Eng Des 2013;88(9
10):253942.
[28] www.thermocalc.se.
[29] Kroupa A, Havrnkov J, Coufalov M, Svoboda M, Vretl J. Phase diagram in the iron-rich corner of the FeCrMoVC system below 1000 K. J Phase
Equilibr 2001;22(3):31223.
[30] Vyrostkov A, Homolov V, Pecha J, Svoboda M. Phase evolution in P92 and E911 weld metals during ageing. Mater Sci Eng 2008;A480(12):28998.
aplovic L, Gogola P, et al. Evolution of secondary phases in austenitic stainless steels during long-term
[31] Vach M, Kunkov T, Domnkov M, evc P, C
exposures at 600, 650 and 800 C. Mater Charact 2008;59(12):17928.
[32] Sawada K, Hongo H, Watanabe T, Tabuchi M. Analysis of the microstructure near the crack tip of ASME Gr.92 steel after creep crack growth. Mater
Charact 2010;61(11):1097102.
[33] Zhao L, Jing H, Xu L, Han Y, Xiu J. Experimental study on creep damage evolution process of Type IV cracking in 9Cr0.5Mo1.8WVNb steel welded
joint. Eng Fail Anal 2012;19:2231.
[34] Jandov D, Kasl J. Microstructural changes in creep exposed P91 steel weld joint. Mater High Temp 2011;28(2):13746.
[35] Wang Xue, Qian-gang Pan, Yao-yao Ren, Wei Shang, Hui-qiang Zeng, Hong Liu. Microstructure and type IV cracking behavior of HAZ in P92 steel
weldment. Mater Sci Eng 2012;A552:493501.
[36] Hosseini HS, Shamanian M, Kermanpur A. Characterization of microstructures and mechanical properties of Inconel 617/310 stainless steel dissimilar
welds. Mater Charact 2011;62(4):42531.
[37] Buchkremer HP, Ennis PJ, Stver D. Manufacture and stress rupture properties of hipped austeniticferritic transition joints. J Mater Process Technol
1999;9293:36870.
[38] Sopouek J, Foret R. More sophisticated thermodynamic designs of welds between dissimilar steels. Sci Technol Weld Joining 2008;13(1):1724.
[39] Hajiannia I, Shamanian M, Kasiri M. Microstructure and mechanical properties of AISI 347 stainless steel/A335 low alloy steel dissimilar joint produced
by gas tungsten arc welding. Mater Des 2013;50:56673.