REVIEW

Type IV cracking in ferritic power plant steels

J. A. Francis*1, W. Mazur2 and H. K. D. H. Bhadeshia3
There have been concerted world wide efforts to develop steels suitable for use in efficient fossil
fired power plants. Ferritic alloys containing between 9 and 12 wt-% chromium are seen as the
most promising materials in this respect, especially for thick walled components such as headers
and the main steam pipe in boilers. However, the performance of the improved steels has often
not been realised in service, because premature failures occur in the heat affected zone of
welded joints in a phenomenon referred to as type IV cracking. This review assesses the
relationship between the composition and microstructure of 9–12 Cr steels, the welding and
fabrication procedures and how these factors translate into a propensity for type IV failures.
Keywords: Creep, Steel, Heat affected zone, Power plant, Type IV, Weld

Introduction ferritic steels, at least as far as the goal of a 650uC steam

temperature is concerned.6
The prospect of global warming has stimulated the This goal has as yet proved impossible to achieve.
search for strategies which lead to a reduction in CO2 Alloys such as NF616 (9Cr–0.5Mo–1.8WVNb) and
emissions. The energy sector is a major CO2 producer HCM12A (11Cr–0.4Mo–2W–CuVNb) are currently
and hence has been the subject of particular scrutiny. In being tested for service at 625uC.8 But there are
the European Union, for example, y50% of electricity problems of premature failure at welded joints, parti-
production in the year 2000 was derived from the cularly in the heat affected zone (HAZ) of the welds.5,6
burning of fossil fuels.1 The cracking of welded joints is usually classified
Fossil fired power plants rely on steam turbines to according to the position of the crack. Type I and type
generate electricity. Their thermodynamic efficiency can II modes occur within the weld metal, the former
be increased if the temperature and pressure of the steam confined to the weld metal whereas the latter may grow
entering the turbines is increased.2 This is the reason for out of the weld into the plate. Type III cracking occurs
the intense efforts throughout the world to develop in the coarse grained region of the HAZ. Type IV is a
steels capable of sustaining the harsher conditions pernicious form of cracking where there is an enhanced
necessary for efficient power generation.1,3,4 The rate of creep void formation in the fine grained and
National Institute for Materials Science in Japan, for intercritically annealed HAZ of the weld, leading to
example, has a long term objective of developing steels early failure when compared with creep tests on the
for ultra supercritical power plants operating with steam unwelded steel.
conditions of 650uC at a pressure of 35 MPa,5 and there The purpose of this article is to review progress in
are similar efforts encouraged by the European Union. understanding and mitigating type IV failures in ferritic
Ferritic creep resistant steels containing up to power plant steels, including the relationship between
2.25 wt-% chromium and 1 wt-% molybdenum have the composition and microstructure of susceptible steels,
been available since the 1940s.6 They are capable of the welding and fabrication procedures and how these
supporting 40 MPa of stress at 560uC.7 In contrast, the factors translate into a propensity for type IV failures.
more expensive 304 austenitic stainless steel can cope Methods of weld repair for existing creep damage have
with 40 MPa at the substantially higher service tem- been reviewed elsewhere.9 The focus here is on the newer
perature of 650uC. However, austenitic steels suffer from steels containing between 8.5 and 12 wt-% chromium,
a much higher thermal expansivity and lower thermal i.e., candidate materials for future power plant applica-
conductivity, making them susceptible to thermal tions.4 We begin with a description of the steel
fatigue, particularly in thick sections and in operating metallurgy, followed by that of the welded structure,
conditions involving frequent shut-downs and variable in the context of type IV cracking.
power demands.1,7 The focus of research is therefore
Steels
The development of 9Cr steels with higher temperature
1
Research Fellow, School of Materials, University of Manchester, capabilities than the classical 2.25Cr–1Mo grades has
Grosvenor Street, Manchester, M1 7HS, UK
2
been described elsewhere6,7,10,11 but a summary is
CSIRO Manufacturing and Infrastructure Technology, Woodville, SA
5011, Australia relevant here to set the scene for type IV creep failures.
3
Department of Materials Science and Metallurgy, University of The first high chromium ferritic steels appeared in
Cambridge, Pembroke Street, Cambridge CB2 3QZ, UK Europe in the mid 1960s. A 9Cr–2Mo steel was developed
*Corresponding author, email [email protected] in France primarily for tubing and subsequently named

ß 2006 Institute of Materials, Minerals and Mining

Published by Maney on behalf of the Institute
Received 29 March 2006; accepted 12 May 2006
DOI 10.1179/174328406X148778 Materials Science and Technology 2006 VOL 22 NO 12 1387
Francis et al. Type IV cracking in ferritic power plant steels

1 Evolution of 9–12 Cr ferritic heat resistant steels17

EM12. This had a duplex microstructure containing d- retards the coarsening of M23C6 carbides (which
ferrite, giving poor impact toughness.6 At about the stabilise the martensite lath structure).14 However, at
same time a 12Cr–1Mo steel was developed in Germany concentrations in excess of 2 wt-%, the formation of
with the designation X20CrMoV12–1 and applied coarse laves phase (Fe2W) can lead to a deterioration of
throughout the world for tubes and pipes. While this creep properties.15 Tungsten also promotes the forma-
steel benefited from a fully martensitic microstructure, it tion of d-ferrite, so its use has to be balanced, either by
exhibited inferior creep strength to EM12 at tempera- reductions in the concentrations of other ferrite promot-
tures .520uC and was difficult to weld, primarily owing ing solutes such as molybdenum, or by adding austenite
to a high carbon content.6 stabilisers such as cobalt.16 The evolution of the 9–12 Cr
In the 1970s the Oak Ridge National Laboratory ferritic steels is illustrated in Fig. 1.17 Table 1 lists the
(USA) developed a modified 9Cr–1Mo steel,12 leading chemical compositions for various steels.6 It can be seen
ultimately to T91 for pressure tube applications and that many of the modern steels contain tungsten in the
alloy P91 for piping and headers; this superseded EM12 range 1–2 wt-%.
and X20CrMoV12–1. The alloy relies on a tempered Many of the new alloys which exhibited improved
martensitic microstructure stabilised by M23C6 carbides, creep rupture strength in short term tests have dis-
with further strengthening owing to molybdenum in appointed when 100 000 h data became available. As a
solid solution and a fine distribution of vanadium/ result, the focus of research has shifted to understanding
niobium rich carbonitride (MX) precipitates.10 the factors that affect the long term stability of M23C6
Reliable further improvements in creep strength have and MX precipitates. In creep tested steels a ‘modified
been achieved with the advent of steels such as NF616 Z-phase’, Cr(V,Nb)N (Ref. 18) seems to precipitate at
and HCM12A, where tungsten enhances the long term the expense of M23C6 and vitally, the MX precipi-
creep strength through solid solution hardening13 and tates.16,20 Recent studies have examined the effect of

Table 1 Typical chemical compositions for various steels.6

Typical chemical composition, wt-%

Steels C N Si Mn Cr Mo V Nb W Co Cu

ASME P/T22 0.12 – 0.3 0.45 2.25 1.0 – – – – –

ASME T9 0.12 – 0.6 0.45 9.0 1.0 – – – – –
HCM9M 0.07 – 0.3 0.45 9.0 2.0 – – – – –
EM12 0.10 – 0.4 0.10 9.0 2.0 0.30 0.40 – – –
X20CrMoV-12-1 0.20 – 0.4 0.60 12.0 1.0 0.25 – – – –
ASME P/T91 0.10 0.05 0.4 0.45 9.0 1.0 0.20 0.08 – – –
HCM12 0.10 0.03 0.3 0.55 12.0 1.0 0.25 0.05 1.0 – –
GX12CrMoWVNbN-10-1-1 0.13 0.05 0.3 0.60 10.5 1.0 0.23 0.08 1.0 – –
NF616 (ASME P/T92) 0.07 0.06 0.1 0.45 9.0 0.5 0.20 0.05 1.8 – –
HCM12A (ASME P/T122) 0.11 0.06 0.1 0.60 12.0 0.4 0.20 0.05 2.0 – 1.0
SAVE12 0.10 0.04 0.3 0.20 11.0 – 0.20 0.07 3.0 3.0 –

1388 Materials Science and Technology 2006 VOL 22 NO 12

 Francis et al. Type IV cracking in ferritic power plant steels

3 Schematic representation of welded joint in 9–12 Cr

PWHT: post-weld heat treatment; RT: room ferritic steel10
temperature10
2 Typical temperature schedules for welded joints in P91 thought to achieve higher toughness than those for
steel MMAW and FCAW, possibly owing to lower oxygen
contents in the weld metal.25
carbon concentration and controlled additions of boron In the context of type IV phenomena, there is little
on the M23C6 and MX precipitation behaviour.21,22 information on the effects of heat input and preheat
temperature on the subsequent creep performance of
welded joints. The choice of preheat temperature is
Welding procedures aimed solely at the avoidance of cracking on cooling
Ferritic 9–12 Cr power plant steels are generally after welding.10,11 It is also useful to deposit multiple
supplied in a normalised and tempered condition. beads to improve the toughness of welds. It has been
Tempering is recommended following welding, to reported that a narrow HAZ is conducive to enhanced
reproduce as far as is possible, a tempered martensitic creep performance and that reducing the groove angle in
microstructure and to relieve some of the stresses the joint preparation (Fig. 3) can improve creep life.23,26
induced by welding. The workpiece is frequently These conclusions hold for creep durations ,10 000 h,
preheated to avoid cold cracking, followed by natural but further testing at lower stresses is required to
cooling to ambient temperature which is slightly below establish whether the benefits are maintained over more
the martensite finish temperature.10,11 A typical tem- realistic time periods.
perature schedule for a welded joint in P91 steel is Albert et al.27,28 investigated the effect of the duration
presented in Fig. 2.10 of post weld heat treatment on the creep performance of
Figure 3 represents a cross-section through a welded welded joints in an 11Cr–0.5Mo–2WCuVNb steel. Their
joint in a thick walled 9–12 Cr ferritic steel pipe. Given samples were extracted from a gas tungsten arc weld in a
its ability to make high quality welds, gas tungsten arc 27 mm thick plate and subsequently tested at 70 MPa
welding (GTAW) is often used to complete the root pass and 650uC. All specimens were post-weld heat treated at
which, because it is the first physical joint between the 740uC, for between 15 and 260 min. No significant effect
component plates, is particularly vulnerable to contrac- of treatment time was noticed over this range, although
tion strains. The same process can be used to complete those samples treated for longer than 60 min performed
the joint but manual metal arc welding (MMAW) is slightly better.
often used in complex joints, or high productivity
processes such as flux cored arc welding (FCAW) and
submerged arc welding for deep welds. The choice of
Detail within HAZ
process may also depend on whether the pipe is to be Type IV failures occur because of the gradients of
welded in a workshop or in situ. Alternatives to arc microstructure in the HAZ of welds. The microstruc-
welding such as electron beam welding can also be tural regions that arise are illustrated in Fig. 4 and have
employed.23 been categorised by Mannan and Laha29 as follows:
Filler metals for welding the 9–12 Cr steels are (i) coarse grain region (CGHAZ): Material near
required to match the creep strength of the parent the fusion boundary that reaches a temperature
material in service. Ideally they would also have well above Ac3 during welding. Any carbides,
matching toughness at ambient temperatures, since which constitute the main obstacle to growth of
welded joints are exposed to transient stresses during the austenite grains, dissolve resulting in coarse
shut down periods. However, in early development work grains of austenite. In the 9–12 Cr steels, this
it became apparent that it would not be possible for weld austenite transforms into martensite on cooling
metals to simultaneously meet both requirements.10,11 (ii) fine grain region (FGHAZ): Away from the
As such, commercial filler metals tend to have a similar fusion boundary where the peak temperature TP
composition to the parent material, with matching creep is lower, but still above Ac3. Austenite grain
strength but lower toughness. There is a tendency to growth is limited by the incomplete dissolution
include higher levels of manganese (0.6–0.7 wt-%) and of carbides. Fine grained austenite is produced,
nickel (0.4–0.6 wt-%) in order to improve weld metal which subsequently transforms into martensite
toughness.24 Gas tungsten arc welding consumables are in the 9–12 Cr steels

Materials Science and Technology 2006 VOL 22 NO 12 1389

Francis et al. Type IV cracking in ferritic power plant steels

open points: furnace heat treatment; closed points: weld

simulator
5 Rupture times for HAZ simulation coupons in ASME-
P122 steel30

Tabuchi, Abe and co-workers also studied the P122

4 Schematic representation of microstructures developed HAZ using furnace simulated samples (Fig. 6).3,31 At all
in HAZ as approximate function of peak temperature stress levels, the minimum creep life was observed for TP
during welding29 Type IV failures are localised to the at or just above Ac3 (920uC (Ref. 30)). The minimum
FGHAZ region, adjacent to the ICHAZ hardness before creep testing occurred at TP5850uC
(ICHAZ in P122), as shown in Fig. 6. There was
(iii) intercritical region (ICHAZ): Here Ac1,TP, evidence that the minimum creep life and minimum
Ac3, resulting in partial reversion to austenite hardness occurred in the same region of the HAZ at high
on heating. The new austenite nucleates at the stresses, although the rupture location changed to a
prior austenite grain boundaries and martensite greater TP when the stress was reduced towards service
lath boundaries, whereas the remainder of the levels. In the latter case, the location corresponding to
microstructure is simply tempered. The auste- minimum hardness clearly is not useful in identifying the
nite (in 9–12 Cr steels) transforms into untem- location of creep failure.
pered martensite on cooling
(iv) over tempered region: With TP below Ac1 the Type IV failures
original microstructure of the plate material
undergoes further tempering. A limited number of uniaxial, cross-weld creep tests have
Post-weld heat treatment tempers any virgin martensite been openly reported for 9–12 Cr steels3,4,8,26,27,32–34
introduced by the welding thermal cycles, but gradients together with a few longitudinal tests on seam welded
in the microstructure and mechanical properties, extend- pipe sections subjected to internal pressure.4 In the latter
ing typically over a few millimetres from the fusion case the highest principal stress is transverse to the
surface, persist even after this treatment. The mechanical welding direction and the test is not uniaxial.
properties of individual zones can in principle be studied In a cross-weld test the sample fails in either the weld
by subjecting bigger steel samples to the expected metal, HAZ or parent material. For 9–12 Cr steels, the
thermal cycle, in order to produce a homogeneous creep strength of the weld metal usually at least matches
microstructure typical of one of the subzones described that of the parent material, so failures generally occur
above. either in the parent material or HAZ.10,11 Recently, an
Creep tests on HAZ simulated specimens for an 11Cr– analysis was carried out of published data relating to
0.4Mo–2WVNbCu (P122) steel with Ac1 and Ac3 at 820 cross-weld creep tests on several different 9–12 Cr
and 920uC respectively, were carried out by Albert steels.35 Based on the results of 53 tests, it was noted that
et al.30 Simulated HAZ zones representative of GTAW
were created using two techniques: furnace treatment
and a Gleeble simulator. The simulated samples were
‘post-weld’ treated at 740uC before creep testing. The
results are illustrated in Fig. 5. The lowest rupture time
at a given stress corresponds samples subjected to TP at
or just above Ac3 (equivalent to FGHAZ). The two
simulation methods produced comparable results except
for the highest values of TP. Unlike a furnace, a weld
simulator heats the sample locally with a uniform
temperature zone of y15 mm over a sample length of
120 mm. Furthermore, samples heated to temperatures
higher than Ac3 tended to rupture outside the uniform
temperature zone, in a location with a microstructure
corresponding to TP just above Ac3. Furnace heat
treatments therefore give more reliable simulated 6 Rupture times for furnace heat treated HAZ simulation
samples. coupons in ASME-P122 steel3

1390 Materials Science and Technology 2006 VOL 22 NO 12

 Francis et al. Type IV cracking in ferritic power plant steels

7 Type IV rupture data for welded joints in ASME-P122

steel at 650uC (Ref. 3)

type IV failures predominate when the applied stress is

less than 100 MPa. The shift in fracture location from
the base material to the HAZ is both stress and
temperature dependent, with stress being the more
important influence.10,11 At least one tube and pipe
8 Perceived significance of type IV rupture stress to a
manufacturer refers to a critical stress, y120 MPa,
variety of parameters36
below which weldments are expected to be limited by
type IV failure.10,11
The relationship between stress and the tendency for well behaved, it gives confidence in the interpretation of
type IV failure is nicely illustrated in data presented by the importance of the other parameters. There are clear
Abe and Tabuchi,3 who compared weld metal, base indications that an increase in the preheat temperature
metal and the simulated FGHAZ against cross-weld could be important in enhancing the rupture stress. The
rupture data for joints in P122 steel. Welds were made heat input is perceived to be insignificant, which is useful
with electron beam welding, as well as with the gas when high productivity welding is required. These
tungsten arc process, using two different joint prepara- predictions have yet to be verified systematically.
tions. It is evident in Fig. 7 that the creep strength of the
weld metal exceeds that of the base metal and that the Microstructural evolution
worst performance is of the simulated FGHAZ speci-
The 9–12 Cr alloys are supplied in a tempered
mens. All of the joints failed in the type IV region, i.e.
martensitic condition; the finely dispersed lath bound-
the fine grained HAZ region adjacent to the intercritical
aries, pinned typically by M23C6 carbides,10,11 are good
HAZ. The rupture times for the joints were between
for retarding creep. Further strengthening relies on solid
those of the base metal and simulated FGHAZ. At solution strengthening from solutes such as molybde-
stresses just above 100 MPa, the creep performance of num and tungsten and fine distributions of V/Nb
welds approaches that of the base metal. However, as carbonitrides (MX) which are seminal in ensuring very
the test stress is reduced the difference between the long term creep properties.38 The microstructure evolves
rupture lives of the welds and the base metal increases, during elevated temperature service, with general
until at low stresses the rupture lives for welds approach coarsening and the further precipitation of more stable
those for the simulated FGHAZ. and rather coarse phases such as laves and Z-phase.18
The data in Fig. 7 show some systematic differences in This depletes solid solution strengthening solutes and at
the creep lives of welded joints, which indicate that the the same time causes the dissolution of MX particles,
welding process and joint configuration must influence leading to a dramatic deterioration in creep properties.38
life, especially for high stresses. The influence of welding Coarse particles also stimulate the nucleation of voids
parameters on type IV failure has been investigated during the latter stages of creep.30,39
using a Bayesian neural network analysis of data The coarsening has been characterised quantita-
obtained from 53 well characterised cross-weld experi- tively; Hofer et al.40 studied a cast martensitic steel
ments.36 The input variables consisted of the chemical (GX12CrMoWVNbN-10-1-1), after aging for periods of
composition (including the 9–12 Cr alloys), details of the up to 33 410 h at 600uC (Fig. 9). Only M23C6 carbides
initial heat treatment of the steel, welding parameters, and vanadium nitride were detected in the as received
post weld heat treatment, test temperature and rupture condition (Fig. 9a). Laves phase and niobium carboni-
time. The output was the rupture stress. trides appeared after aging for 976 h (Fig. 9b). Once
The abilities of a variety of variables in explaining the present, laves phase was observed to coarsen rapidly
observed changes in type IV rupture stress (the (Fig. 9c and d) and Z-phase appeared (at the expense of
significance of each input) are shown in Fig. 8.36 A MX) after aging for 33 410 h. The results suggest that
large value of the significance means that the particular laves phase, because it coarsens rapidly, is likely to
variable is important in determining the rupture stress in promote the onset of tertiary creep over the long term.41
the context of the data examined. The time to rupture In contrast, the stability of M23C6 and MX is critical to
and test temperature obviously greatly influence the the long term performance of these alloys.
stress but some well known effects of normalising The effects of welding thermal cycles on the carbide
temperature,37 tempering and the strong effect of dispersions and creep strength in a 10Cr–3W–3CoVNb
tungsten6 are clearly evident. Given that the model is steel (SAVE12 in Table 1) have been investigated by

Materials Science and Technology 2006 VOL 22 NO 12 1391

Francis et al. Type IV cracking in ferritic power plant steels

9 Distributions of precipitates in X12CrMoWVNbN-10-1-1 steel in terms of an equivalent diameter De in a as received

condition, b 600uC/976 h, c 600uC/5 014 h and d 600uC/33 410 h (Ref. 40)

Hirata and Ogawa.42,43 They used induction heated this unusual HAZ microstructure have not been fully
HAZ simulated specimens with TP in the range 830 to clarified, but it is suggested22 that the low nitrogen
1200uC, subjecting them to creep at 650uC with a stress content reduces the content of MX, thus allowing the
of 98 MPa. The precipitates in the over tempered austenite grains to coarsen, eliminating the FGHAZ
regions did not dissolve as a consequence of the weld which is weak in creep. In any event, none of the welds
thermal cycle with TP5830uC, just above Ac1. In the studied exhibited type IV cracking, with fracture
coarse grained HAZ with TP51200uC, the precipitates occurring at the weld interface with rupture times
almost completely dissolved but were reprecipitated comparable to that of the parent metal. This is a system
during the post-weld heat treatment. In the fine grained which deserves further attention.
HAZ with TP51000uC (just above Ac3), carbides The mechanism of the boron effect is not clear; could
containing chromium or vanadium underwent partial it be that it influences austenite formation even at
dissolution with reprecipitation and relatively rapid such minute concentrations? Much may be attributable
coarsening during the post-weld heat treatment and to the absence of a fine grained HAZ, but it is also
subsequent creep, far more so than in the other zones. known that additions of boron have led to improved
After 5014 h there were indications of Z-phase in the creep performance for the parent material, through the
fine grained HAZ, a phase associated with reduced creep delayed coarsening of M23C6 carbides and a correspond-
properties.16,20,40 ing delay in the onset of tertiary creep.22
These observations are important because the partial
dissolution of M23C6 in the fine grained HAZ promotes Stress evolution
Z-phase which in turn destabilises MX. The resulting Watanabe et al.45 studied thermally aged 2.25Cr–1Mo,
deterioration of creep thus becomes localised to the type with and without an applied stress and found that
IV region. coarsening was accelerated in the presence of stress. It is
Some fascinating results have recently been reported expected therefore that residual stresses resulting from
on the influence of boron on type IV failure in 9–12 Cr welding are detrimental to creep properties, especially
steels.22,44 Albert et al.22 found that such failures were since coarse particles enhance the formation of
suppressed in a 9Cr–3W–3CoVNb steel containing voids.30,39
boron in the range 90–130 ppm, with nitrogen kept Although there are no specific data on residual
,0.002 wt-%. At stresses .100 MPa, fracture occurred stresses in 9–12 Cr steel weldments, the existence of
in the parent material, while for lower stresses the failure such stresses is well established in general.46–48
occurred at the fusion surface rather than in the usual Yaghi et al.49 recently made numerical predictions for
type IV region, which is the HAZ heated to near Ac3. the residual stresses that arise in P91 circumferential
There were reportedly no creep voids in the HAZ of pipe welds. Their results suggest that the location of the
fractured specimens. Curiously, the austenite grains peak tensile residual stresses will vary with the wall
remained large at a distance 1–2 mm from the fusion thickness. Nevertheless, as the wall thickness becomes
surface, the region where fine grained austenite is usually large, they consistently predict peak tensile residual
observed in the HAZ of welded joints. The reasons for stresses between 400 and 500 MPa near the outer surface

1392 Materials Science and Technology 2006 VOL 22 NO 12

 Francis et al. Type IV cracking in ferritic power plant steels

of the pipe; a result that becomes insensitive to further HAZ simulated specimens.53 Finally, they noted that
increases in wall thickness. There is a pressing need to once sliding was accommodated and constraint relaxed,
validate such numerical predictions with experimental the failure time for both cross-weld specimens and
data. Furthermore, studies specific to the power plant simulated type IV specimens was similar.
steels would be useful in assessing the importance of
parameters such as the preheat and post-weld heat Suggestions for future work
treatment in determining residual stress distributions
and the tendency for type IV cracking. As a result of this assessment of literature, we believe
Because of the gradients of microstructure in the that the following areas need particular attention in
HAZ, there are corresponding gradients in the creep future research:
properties. If the creep resistance reaches a minimum at (i) Controlled additions of boron: Evidence clearly
some location then complex constraint effects are suggests type IV cracking can be suppressed
expected during cross-weld loading, which may intensify using controlled additions of boron to 9–12 Cr
local damage. Using a two-dimensional finite element steels. Concentrations between 90 and 130 ppm
technique, Li et al.50 simulated creep in a P122 weldment result in the elimination of the conventional fine
by compiling a model consisting of four regions with grained HAZ; they stabilise M23C6 carbides,
different creep properties (weld metal, CGHAZ, apparently through the partial substitution of
FGHAZ and base metal). During creep, each zone carbon by boron. However, it is not clear which
within the weldment was assumed to obey Norton’s of these factors is responsible for the suppres-
creep law. The stress and strain distributions following sion of type IV failures, nor why it might be
1000 h of creep at 90 MPa and 650uC were thus counterproductive to exceed a boron level of
estimated. It was found that a high tensile first principal 130 ppm.
stress and high tensile hydrostatic stress were generated (ii) The stability of MX precipitates: MX precipi-
in the FGHAZ. It was argued that the strain focuses in tates are vital in determining the long term creep
the weaker FGHAZ during the early stages of creep, strength for 9–12 Cr steels. However, there are
leading to the nucleation of creep voids, which are then indications that the thermal cycles that are
experienced in the fine grained HAZ promote a
encouraged to grow by the strain mismatch and triaxial
chromium-rich ‘modified Z-phase’ of the form
stresses generated as creep progresses.
Cr(V,Nb)N during creep at the expense of MX
The triaxial stress state in the FGHAZ has been
precipitates. Further work is required to estab-
previously recognised to accelerate the growth of creep
lish how the fine grained HAZ is particularly
voids,28,50 but this may not necessarily lead to a
susceptible to this deleterious phase and
reduction in creep life. To illustrate this point, it is
whether its formation can be avoided.
worth considering the data in Fig. 7, for P122 steel.
(iii) Significance of welding parameters: There is a
Above 100 MPa, the creep performance of weldments is
dearth of studies relating welding parameters,
close to that for the parent material. Furthermore, as
particularly the preheat temperature and heat
the applied stress is reduced, the creep performance of
input, to the tendency for type IV cracking.
the weldments approaches (but still exceeds) that of the
(iv) Role of stress: The role of residual stresses
weakest zone within the weldment, i.e. the FGHAZ. If it arising from welding, on the creep performance
is assumed that the triaxial stress states that arise in the of 9–12 Cr steels, needs to be investigated with
FGHAZ accelerate the formation of creep voids, then it respect to their effect on the evolution of
is intriguing that the creep lives of the welded joints, microstructure in the HAZ. Further work is
which failed owing to type IV cracking, were still longer also necessary to clarify the extent to which
than those for the simulated FGHAZ specimens (these, grain-boundary sliding is responsible for the
after all, are mechanically homogeneous and will not formation of creep voids in longer-term creep
contain triaxial stresses). The data of Fig. 7 suggest that tests. In general, greater emphasis needs to be
the mechanical constraint resulting from the property placed on tests carried out at stresses below
gradients in the HAZ actually delays failure in the 100 MPa, as it is at these lower stress levels that
FGHAZ. type IV failures predominate.
An alternative approach to modelling type IV damage
is due to Kimmins and Smith,51 who used finite element
representations of Cr–Mo weldments, allowing for the Mechanism for type IV cracking
relaxation of constraint by the sliding of adjacent The focus on type IV cracking has in this review been on
elements during creep. It is known that models based the 9–12Cr steels which are leading the search for steels
on transverse strain compatibility give rise to significant useful in making ever more efficient power plants. For
constraint of creep deformation in the weak zone.52 these steels, it appears that the microstructure which is
However, it was pointed out that at lower service weakest in creep is that associated with the FGHAZ of
stresses grain boundaries are less resistant to sliding.53 a weld. This is the zone which reaches temperatures
Thus, in contrast to continuum models where damage just inside the austenite phase field, but not for long
accumulation is enhanced by multiaxial stresses, enough to allow carbide precipitates to completely
Kimmins and Smith suggested that constraint is relaxed dissolve. As a result, the austenite grains that form
by grain boundary sliding.51 They later presented remain relatively fine and transform to martensite on
evidence to suggest that grain boundary sliding in cooling. On post-weld heat treatment, the undissolved
cross-weld specimens gives rise to greater numbers of carbides coarsen with limited further precipitation.
cavities, consistent with their observation that greater The FGHAZ does not correspond to the heat affected
numbers of cavities are observed in welded joints than in region with the lowest hardness, which is the zone that is

Materials Science and Technology 2006 VOL 22 NO 12 1393

Francis et al. Type IV cracking in ferritic power plant steels

only partially austenitised (ICHAZ) and hence contains 10. K. Haarmann, J. C. Vaillant, B. Vandenberghe, W. Bendick and
A. Arbab: ‘The T91/P91 book’; 2002, Boulogne, Vallourec &
over tempered martensite.
Mannesmann Tubes.
At high stresses, using simulated samples with homo- 11. D. Richardot, J. C. Vaillant, A. Arbab and W. Bendick: ‘The T92/
geneous microstructures, the minimum creep strength P92 book’; 2000, Boulogne, Vallourec & Mannesmann Tubes.
microstructure coincides with the minimum hardness 12. V. K. Sikka and P. Patriarca: ‘Analysis of weldment mechanical
microstructure (ICHAZ). However, at the low stresses properties of modified 9Cr–1Mo steel’, Technical report, Metals
and Ceramics Division, Oak Ridge National Laboratory, Oak
typical of service conditions, the creep life is a minimum Ridge, TN, USA, 1984.
in the FGHAZ, which does not have the minimum 13. K. Iwanaga, T. Tsuchiyama and S. Takaki: Key Eng. Mater., 2000,
hardness. This means that hardness cannot be used as an 171–174, 477–482.
indicator of type IV susceptibility or type IV location. 14. F. Abe: Key Eng. Mater., 2000, 171–174, 395–402.
15. Y. Hasegawa, T. Muraki, M. Ohgami and H. Mimura: Key Eng.
The tests done on simulated samples containing Mater., 2000, 171–174, 427–436.
homogeneous microstructures are revealing, but type 16. C. Berger, A. Scholz, Y. Wang and K. H. Mayer: Z. Metallkd,
IV cracking occurs in the FGHAZ of cross-weld 2005, 96, (6), 668–674.
specimens where the microstructure is far from uniform. 17. H. Cerjak and E. Letofsky: Sci. Technol. Weld. Join., 1996, 1, (1),
36–42.
Finite element simulations which treat the cross-weld
18. A. Strang and V. Vodarek: Mater. Sci. Technol., 1996, 12, (7), 552–
test as a longitudinal series of different microstructures 556.
with different constitutive behaviours, predict triaxial 19. I. L. Papst, P. Warbichler, F. Hofer, E. Letofsky and H. Cerjak:
stresses in the FGHAZ, which should accelerate the Z. Metallkd, 2004, 95, (1), 18–21.
formation of creep voids. Taken at face value, this is odd 20. E. Letofsky and H. Cerjak: Sci. Technol. Weld. Join., 2004, 9, (1),
31–36.
because experiments with cross-weld samples show a 21. M. Taneike, K. Sawada and F. Abe: Metall. Mater. Trans. A, 2004,
greater resistance to creep than simulated FGHAZ 35A, (4), 1255–1262.
samples. This is in spite of the fact that triaxial stresses 22. S. K. Albert, M. Kondo, M. Tabuchi, F. Yin, K. Sawada and
will only be present in the FGHAZ of cross-weld F. Abe: Metall. Mater. Trans. A, 2005, 36A, (2), 333–343.
23. M. Tabuchi, M. Matsui, T. Watanabe, H. Hongo, K. Kubo and
samples.
F. Abe: Mater. Sci. Res. Int., 2003, 9, (1), 23–28.
The most reasonable explanation comes from the 24. A. W. Marshall, Z. Zhang and G. B. Holloway: Proc. 5th Int.
work of Kimmins and Smith, who allowed grain Conf. on ‘Welding and repair technology for power plants’, Point
boundary sliding in their finite element models of the Clear, AL, USA, June 2002, Electric Power Research Institute.
cross-weld scenario. They found that the constraint 25. E. L. Bergquist: Svetsaren, 1999, 54, (1–2), 22–25.
26. S. K. Albert, M. Tabuchi, H. Hongo, T. Watanabe, K. Kubo and
originating from the heterogeneous microstructure is M. Matsui: Sci. Technol. Weld. Join., 2005, 10, (2), 149–157.
relaxed by grain boundary sliding, making failure easier 27. S. K. Albert, M. Matsui, T. Watanabe, H. Hongo, K. Kubo and
in the FGHAZ. This is consistent with the poor creep M. Tabuchi: ISIJ Int., 2002, 42, (12), 1497–1504.
properties of a simulated, homogeneous FGHAZ. 28. S. K. Albert, M. Matsui, T. Watanabe, H. Hongo, K. Kubo and
M. Tabuchi: Int. J. Press. Vess. Pip., 2003, 80, 405–413.
These results lead to a simple conclusion, that 29. S. L. Mannan and K. Laha: Trans. Ind. Inst. Met., 1996, 49, (4),
anything carried out to strengthen grain boundaries, 303–320.
for example the addition of boron, may allow constraint 30. S. K. Albert, M. Matsui, H. Hongo, T. Watanabe, K. Kubo and
to appear in a cross-weld test, thereby increasing the life M. Tabuchi: Int. J. Press. Vess. Pip., 2004, 81, 221–234.
31. M. Tabuchi, T. Watanabe, K. Kubo, J. Kinugawa and F. Abe: Key
for type IV fracture.
Eng. Mater., 2000, 171–174, 521–528.
32. T. Kojima, K. Hayashi and Y. Kajita: ISIJ Int., 1995, 35, (10),
Acknowledgements 1284–1290.
33. K. Shinozaki, D. Li, H. Kuroki, H. Harada and K. Ohishi: ISIJ
This work was undertaken as part of a project that is Int., 2002, 42, (12), 1578–1584.
proudly supported by the International Science 34. R. Wu, R. Sandstrom and F. Seitisleam: J. Eng. Mater. Technol.,
2004, 126, (1), 87–94.
Linkages programme established under the Australian
35. J. A. Francis, W. Mazur and H. K. D. H. Bhadeshia: ISIJ Int.,
Government’s innovation statement Backing Australia’s 2004, 44, (11), 1966–1968.
Ability. 36. J. A. Francis, W. Mazur and H. K. D. H. Bhadeshia: Proc. 7th Int.
Conf. on ‘Trends in welding research’, Pine Mountain, GA, USA,
May 2005, ASM International.
References 37. F. Brun, T. Yoshida, J. D. Robson, V. Narayan, H. K. D. H.
Bhadeshia and D. J. C. Mackay: Mater. Sci. Technol., 1999, 15, (5),
1. H. Cerjak: Proc. Int. Conf. on ‘Achievements and perspectives in 547–554.
producing welded construction for urban environments’, 56th IIW 38. D. Li, K. Shinozaki, H. Harada and K. Ohishi: Metall. Mater.
Annual Assembly, Bucharest, Romania, July 2003, ASR, 15–24. Trans. A, 2005, 36A, (1), 107–115.
2. H. K. D. H. Bhadeshia: ISIJ Int., 2001, 41, (6), 626–640. 39. D. Li and K. Shinozaki: Sci. Technol. Weld. Join., 2005, 10, (5),
3. F. Abe and M. Tabuchi: Sci. Technol. Weld. Join., 2004, 9, (1), 22– 544–549.
30. 40. P. Hofer, H. Cerjak and P. Warbichler: Mater. Sci. Technol., 2000,
4. K. Shinozaki, D. Li, H. Kuroki, H. Harada, K. Ohishi and T. Sato: 16, (10), 1221–1225.
Sci. Technol. Weld. Join., 2003, 8, (4), 289–295. 41. F. Abe: Metall. Mater. Trans. A, 2005, 36A, (2), 321–332.
5. M. Tabuchi, T. Watanabe, K. Kubo, M. Matsui, J. Kinugawa and 42. H. Hirata and K. Ogawa: Weld. Int., 2005, 19, (2), 109–117.
F. Abe: Int. J. Press. Vess. Pip., 2001, 78, 779–784. 43. H. Hirata and K. Ogawa: Weld. Int., 2005, 19, (2), 118–124.
6. C. D. Lundin, P. Liu and Y. Cui: ‘A literature review on 44. Y. Otoguro, M. Matsubara, I. Itoh and T. Nakazawa: Nucl. Eng.
characteristics of high temperature ferritic Cr–Mo steels and Design, 2000, 196, (1), 51–61.
weldments’, WRC bulletin 454, Welding Research Council, Inc., 45. T. Watanabe, M. Yamazaki, H. Hongo, M. Tabuchi and
New York, NY, USA, 2000. T. Tanabe: Int. J. Press. Vess. Pip., 2004, 81, 279–284.
7. ‘Metals handbook’, 10th edn, Vol. 1, ‘Elevated temperature 46. P. Dong: Sci. Technol. Weld. Join., 2005, 10, (4), 389–398.
properties of ferritic steels’, 617–652; 1990, Materials Park, OH, 47. P. Dong: J. Press. Vess. Technol. Trans. ASME, 2001, 123, (2),
ASM International. 207–213.
8. M. Matsui, M. Tabuchi, T. Watanabe, K. Kubo, J. Kinugawa and 48. D. J. Smith, P. J. Bouchard and D. George: J. Strain Anal., 2000,
F. Abe: ISIJ Int., 2001, 41, S126–S130. 35, (4), 287–305.
9. S. Issler, A. Klenk, A. A. Shibli and J. A. Williams: Int. Mater. 49. A. H. Yaghi, T. H. Hyde, A. A. Becker, J. A. Williams and W. Sun:
Rev., 2004, 49, (5), 299–324. J. Mater. Process. Technol., 2005, 167, 480–487.

1394 Materials Science and Technology 2006 VOL 22 NO 12

 Francis et al. Type IV cracking in ferritic power plant steels

50. D. Li, K. Shinozaki and H. Kuroki: Mater. Sci. Technol., 2003, 19, 52. I. J. Perrin and D. R. Hayhurst: Int. J. Press. Vess. Pip., 1999, 76,
(9), 1253–1260. 599–617.
51. S. T. Kimmins and D. J. Smith: J. Strain Anal., 1998, 33, (3), 195– 53. D. J. Smith, N. S. Walker and S. T. Kimmins: Int. J. Press. Vess.
206. Pip., 2003, 80, 617–627.

Materials Science and Technology 2006 VOL 22 NO 12 1395