1 Introduction

9 % chromium containing ferritic–martensitic steels are favoured structural materials for high temperature steam generator applications in thermal and nuclear power generating industries [1, 2]. The first in the series developed as plain 9Cr–1Mo steel is known as P9 or grade 9 steel [3]. Grade 9 steel offers a good combination of creep strength and ductility, weldability and microstructural stability over long exposures at elevated temperatures [4–6]. The advancement of new generation power plants such as ultra super critical (USC) power plants has increased the operating steam temperatures and pressures to achieve higher efficiency. This led to the development of modified versions with improved creep strength for steam generator applications [1]. 9Cr–1Mo steel modified by the addition of strong carbide/nitride forming elements such as vanadium and niobium along with controlled nitrogen, i.e., 9Cr–1Mo–V–Nb is designated as P91 or grade 91 steel [1–3]. The creep strength of grade 91 steel has been established to be significantly higher than grade 9 steel [1, 2]. The steel has been further modified by the addition of tungsten with reduced molybdenum as 9Cr–0.5Mo–1.8W–V–Nb and is known as P92 or grade 92 steel [1, 3]. Addition of W in grade 92 steel leads to further enhancement of creep strength over grade 91 steel [1].

The creep strength of 9 % Cr ferritic–martensitic steels is derived from the combination of (1) solid solution strengthening mainly due to the presence of Mo (grade 9 and 91 steels) and W (grade 92 steel), (2) high initial dislocation density arising from phase transformation, (3) boundary strengthening in terms of hierarchical microstructure comprising of prior austenite grains, packets, blocks, sub-blocks and laths and (4) precipitation hardening. In grade 9 steel, precipitation hardening results mainly from the presence of M23C6 and M2X precipitates [5, 6]. In grade 91 and 92 steels, the strength contribution from precipitation arise by the presence of M23C6 carbides at the hierarchical boundaries and fine distribution of MX-type Nb-rich carbonitride and V-rich carbonitride/nitride in the matrix regions [7–13]. The superior long-term creep strength of grade 91 and 92 steels results mainly from the presence of fine MX precipitates, which undergo insignificant coarsening during creep. Intermetallic Laves phase rich in Mo (Fe2Mo) forms during thermal aging and creep exposure in grade 9 and 91 steels [5, 6, 11–13]. In grade 92 steel, W with small amount of Mo along with Fe constitutes Laves phase as Fe2(W, Mo) [7, 9–11]. Both grade 91 and 92 steels suffers from the nucleation and growth of deleterious Z-phase Cr(V, Nb)N at the expense of fine MX precipitates [9, 11–13]. It has been reported that the kinetics of Z-phase formation and its growth in 9 % Cr steels is much slower than 12 % Cr steels, and large population of MX particles remain present in the long-term creep regime contributing towards higher creep strength in grade 91 and 92 steels [9, 11].

The paper deals with the comparative microstructural degradation and development of damage occurring during creep in grade 9 and 92 steels. The creep properties of both the steels have been presented in terms of the stress dependence of rupture life, variations of creep ductility with rupture life, creep rate-rupture life relationships, creep damage tolerance factor, fracture behaviour and microstructure.

2 Experimental

Constant load creep-rupture tests have been performed on grade 9 and 92 steels at 873 K. Grade 9 steel have been tested in quenched and tempered (Q + T) and simulated post weld heat treatment (SPWHT) conditions on specimens obtained from tube plate forging [5, 6]. The chemical composition (in wt%) was as follows: Fe–0.10C–0.75Si–0.63Mn–0.001S–0.02P–9.27Cr–1.05Mo. Tube plate was austenitized at 1223 K for 5 h (heating time: 8 h) followed by quenching in water. Tempering treatment involved soaking at 1023 K for 8 h (heating time: 8 h) followed by air cooling. SPWHT was given on Q + T specimen blanks as 998 K for 3 h with heating and cooling rates of 50 K h−1 above 673 K. Creep tests on grade 9 steel were conducted for stresses ranging from 50 to 175 MPa. Grade 92 steel was tested in normalised and tempered (N + T) condition at 873 K in the stress range 120–270 MPa. The chemical composition (in wt%) was as follows: Fe–0.124C–0.20Si–0.47Mn–0.006S–0.011P–9.07Cr–0.46Mo–1.78W–0.19V–0.063Nb–0.003B–0.043N. The steel was austenitized at 1343 K for 2 h followed air cooling and tempered at 1048 K for 2 h followed by air cooling. Optical metallographic examinations were performed on heat treated and tested specimens. Fractographic examinations were performed using scanning electron microscope (SEM). Transmission electron microscopic (TEM) examinations were conducted on the carbon replicas and thin foils from heat treated and tested specimens.

3 Results and Discussion

3.1 Microstructure Prior to Creep Tests

The microstructure of grade 9 steel tubeplate forging in Q + T and SPWHT conditions was composed of tempered lath martensite and a few stringers of pro-eutectoid ferrite (~2 %) at prior austenite grain boundaries [5, 6]. In both Q + T and SPWHT conditions, coarse precipitates at the boundaries of prior austenite grains, martensite laths and pro-eutectoid ferrite, and in the matrix regions were observed. TEM investigations revealed presence of M23C6 carbides (i.e., [CrFe]23C6) and chromium rich Cr2N precipitates. Microstructure of grade 92 steel was consisted of tempered lath martensite and precipitates along the grain and lath boundaries and in the matrix regions. TEM investigations suggested presence of chromium rich M23C6 carbides on the boundaries and within the intralath matrix regions. Fine MX type precipitates have been identified as spheroidal Nb-rich Nb(C, N) carbonitrides and plate like V-rich V(C, N)/VN carobnitides/nitrides in the matrix regions. Some MX precipitates have been observed as Nb(C, N)-VN complexes following nucleation and growth of VN on undissolved particles of Nb(C, N).

3.2 Creep-Rupture Properties

The relative creep-rupture strength of grade 9 and 92 steels is shown as the variations of rupture life with applied stress at 873 K in Fig. 1. The long-term creep-rupture data reported for grade 92 steel at 873 K [10] are superimposed. Stress dependence of rupture life in both the steels obeyed power law of the form

$$ \text{t}_{\rm r} = \text{A}^{\prime } \upsigma^{{ - {\rm n}^{\prime } }} , $$
(1)

where A′ is constant and n′ is the power law exponent. Grade 9 and 92 steels displayed two-slope behaviour in the stress dependence of rupture life with different values of stress exponents in the low and high stress regimes. In grade 9 steel, n′ = 5.5 and 9 have been obtained in the low and high stress regimes, respectively. Grade 92 steel exhibited n′ = 7.2 and 17.1 in the low and high stress regimes, respectively. Different values of stress exponents obtained in the low and high stress regimes in grade 9 and 92 steels are in agreement with the reported observations [10, 14, 15] and are in accordance with the stress dependence of minimum creep rate in both the steels [16, 17]. Grade 92 steel exhibited significantly higher creep-rupture strength than grade 9 steel (Fig. 1). This is also reflected in the higher values of respective stress exponents than grade 9 steel in the low and high stress regimes. A good agreement with the reported creep strength of grade 92 steel [10] in the long-term creep regime can be seen in Fig. 1.

Fig. 1
figure 1

Stress dependence of rupture life at 873 K for grade 9 and 92 steels. Rupture life values at low stresses reported for grade 92 steel are superimposed [10]

Full size image

The creep ductility of grade 9 and 92 steels is shown as the variations of reduction in area with rupture life at 873 K in Fig. 2. The reported ductility values for grade 92 steel at longer rupture lives [10] are also superimposed. A good agreement for ductility values can be seen. Grade 9 steel displayed significantly higher creep ductility at longer durations than grade 92 steel. On the contrary, grade 92 steel exhibited significant decrease in creep ductility at intermediate (Regime-II) and longer (Regime-III) rupture lives. In grade 9 steel, high creep ductility resulted in the occurrence of transgranular ductile fracture showing cup and cone fracture (Fig. 3a) and presence of ductile dimples as a consequence of micro-void coalescence (Fig. 3b). Grade 92 steel also exhibited transgranular fracture (Fig. 4). However, reduced ductile features and a few intergranular cracks appearing on the fracture surface (shown by arrows) can be seen in Fig. 4b. This has been confirmed by the presence of large number of secondary cracks (Fig. 5a) originating mainly from decohesion of Laves phase [Fe2(W, Mo)] at the grain and lath boundaries (Fig. 5b). The grey particles have been recognized as M23C6, whereas the Laves phase has been identified as white particles (marked 1 and 2 in Fig. 5b). Typical Edax profile confirming white particles as Fe2(W, Mo) is shown in Fig. 5c.

Fig. 2
figure 2

Variations of  % reduction in area with rupture life at 873 K for grade 9 and 92 steels. Ductility values at longer rupture lives reported for grade 92 steel are superimposed [10]

Full size image
Fig. 3
figure 3

SEM fractographs showing transgranular fracture in grade 9 steel specimen tested at 60 MPa for rupture life 12,575 h. Typical cup and cone fracture and presence of ductile dimples can be seen in a and b, respectively

Full size image
Fig. 4
figure 4

SEM fractographs showing transgranular fracture in grade 92 steel tested at 132 MPa for 57,421 h. Secondary cracks are shown by arrows in b

Full size image
Fig. 5
figure 5

SEM micrographs showing a large number of secondary cracks in the specimen tested at 132 MPa for rupture life 53,700 h and b formation of creep cavities (shown by arrows) due to decohesion at Fe2(W, Mo)-boundary interface. Representative Edax profile from Fe2(W, Mo) precipitates is shown in c

Full size image

3.3 Creep Rate-Rupture Life Relationships

The interrelation between creep deformation and rupture is described by Monkman and Grant [18] and modified Monkman–Grant [19] relationships between minimum or steady state creep rate (\( {\dot{\upvarepsilon }}_{\text{s}} \)) and rupture life (tr). MGR in the generalised form is expressed as

$$ {\dot{\upvarepsilon }}_{\text{s}}^{\text{m}} \cdot {\text{t}}_{\text{r}} = {\text{ C}}, $$
(2)

where m and C are the slope and intercept in the double logarithmic plot of tr versus \( {\dot{\upvarepsilon }}_{\text{s}} \), respectively. The generalised form of MMGR is expressed as

$$ {\dot{\upvarepsilon }}_{\text{s}}^{{{\text{m}}^{{\prime }} }} \cdot \frac{{{\text{t}}_{\text{r}} }}{{\upvarepsilon_{\text{f}} }} = {\text{C}}^{{\prime }} , $$
(3)

where εf is the strain to failure and m′ is the slope of the double logarithmic plot of tr/εf versus \( {\dot{\upvarepsilon }}_{\text{s}} \) and intercept C′ is a constant. The validity of MGR for grade 9 steel is observed as the slope of log tr versus log \( {\dot{\upvarepsilon }}_{\text{s}} \) equal to unity (Fig. 6) and accordingly Eq. (2) is expressed as

$$ {\dot{\upvarepsilon }}_{\text{s}} \cdot {\text{t}}_{\text{r}} = {\text{C}}_{\text{MG}} , $$
(4)

where CMG is the real Monkman–Grant constant. P9 steel exhibited two distinct values of CMG = 0.05 and 0.1 in the low and high stress regimes, respectively. Like MGR, the validity of MMGR as the slope of log(tr/εf) versus log \( {\dot{\upvarepsilon }}_{\text{s}} \) equal to unity is observed, and Eq. (2) is expressed as

$$ {\dot{\upvarepsilon }}_{\text{s}} \cdot \frac{{{\text{t}}_{\text{r}} }}{{\upvarepsilon_{\text{f}} }} = {\text{C}}_{\text{MMG}} . $$
(5)

In grade 9 steel, different values of CMMG as 0.1 and 0.2 have been obtained in the low and high stress regimes, respectively. The validity of both MGR (Eq. 4) and MMGR (Eq. 5) essentially indicate constant creep ductility in grade 9 steel. Further, the obtained low values of CMG and CMMG suggest that the contribution of secondary creep strain to the overall creep strain in grade 9 steel is small and most of the creep strain results from tertiary creep.

Fig. 6
figure 6

Rupture life versus steady state creep rate plot showing validity of Monkman–Grant relation with separate values of CMG = 0.05 and 0.1 for low and high stress regimes, respectively, in grade 9 steel

Full size image

Grade 92 steel followed generalized form of Monkman–Grant relation (Eq. 2) with slope m = 0.86 and C = 0.163 in the plot of log tr versus log \( {\dot{\upvarepsilon }}_{\text{s}} \) (Fig. 7). The slope m = 0.86 less than unity essentially indicate decrease in the Monkman–Grant strain, \( {\dot{\upvarepsilon }}_{\text{s}} \)·tr with increase in rupture life. Contrary to this, the steel obeyed Eq. (5) and the validity of MMGR is observed as the slope of log (tr/εf) versus log \( {\dot{\upvarepsilon }}_{\text{s}} \) equal to unity (Fig. 8). This is agreement with the recent observation reported on large body of creep data in another modified grade 91 steel [20]. A single value of CMMG = 0.154 has been observed for the stress range examined in grade 92 steel. From these observations, it can be inferred that the significant reduction in creep ductility with increasing rupture life (Fig. 3) results in grade 92 steel following generalized form of Monkman–Grant relation (Eq. 2). It can also be inferred that the reduction in creep ductility facilitate the observed validity of MMGR (Eq. 5).

Fig. 7
figure 7

Rupture life versus minimum creep rate plot showing applicability of generalised form of Monkman–Grant relation in grade 92 steel

Full size image
Fig. 8
figure 8

Rupture life versus minimum creep rate plot showing validity of modified Monkman–Grant relation in grade 92 steel

Full size image

3.4 Tolerance to Creep Damage and Microstructural Degradation

Creep deformation and damage have been treated appropriately in the framework of ‘Continuum Creep Damage Mechanics’ (CDM) approach [21, 22]. An important outcome of CDM approach is the creep damage tolerance factor λ defined as the ratio of strain to failure εf to Monkman–Grant strain \( {\dot{\upvarepsilon }}_{\text{s}} \).tr [23, 24] as

$$ \lambda = \frac{{\upvarepsilon_{\text{f}} }}{{{\dot{\upvarepsilon }}_{\text{s}} \cdot {\text{t}}_{\text{r}} }}. $$
(6)

λ is suggested to be a better measure of creep ductility [23] as it assess the susceptibility of a material to localised cracking [24]. Ashby and Dyson [24] demonstrated that each damage micromechanism, when acting alone, results in a characteristic value of λ and a characteristic shape of creep curve. Creep damage due to growth of cavities by coupled diffusion and power-law creep results in λ values in the range 1.5–2.5. When damage is dominated by thermal-coarsening of particles and dislocation substructure softening, λ can be as high as 5 or more. Equation (6) can be rearranged into a relation between ratio of strain to failure and rupture time (εf/tr), the average creep rate and minimum creep rate as

$$ \frac{{\upvarepsilon_{\text{f}} }}{{{\text{t}}_{\text{r}} }} = \lambda \cdot {\dot{\upvarepsilon }}_{\text{s}} . $$
(7)

Double logarithmic plot of εf/tr versus \( {\dot{\upvarepsilon }}{}_{\text{s}} \) following Eq. (7) gives the value of intercept as λ shown for grade 9 steel in Fig. 9. Grade 9 steel exhibited distinct but constant values of λ = 5 and 10 for high and low stress regimes, respectively. Grade 92 steel displayed a single value of λ = 6.5 in the stress range examined (Fig. 10).

Fig. 9
figure 9

Variations of average creep rate with steady state creep rate showing constancy of λ in grade 9 steel. Separate values of λ = 10 and 5 are obtained in the low and high stress regimes, respectively

Full size image
Fig. 10
figure 10

Variations of average creep rate with minimum creep rate showing constancy of λ in grade 92 steel

Full size image

The high values of λ essentially indicate microstructural degradation as the dominant creep damage mechanism in both the steels. High λ values in grade 9 steel are consistent with the observed high creep ductility, absence of intergranular cracking and extensive tertiary creep. Grade 9 steel also exhibited decrease in dislocation density, and coarsening of precipitates and dislocation substructure [5, 6]. Progressive coarsening of M23C6 carbides and nucleation and growth of Laves phase (Fe2Mo) with increase in rupture life are shown Fig. 11. Presence of M23C6 and Cr2N precipitates in grade 9 steel in the heat treated condition can be seen in Fig. 11a. Nucleation of Fe2Mo at Cr2N/matrix interface as multiple nodes and adjacent to M23C6 can be seen in Fig. 11b. Coarsening of Fe2Mo and M23C6 at the expense of finer precipitates leads to decreased number density of precipitates (Fig. 11c). Like grade 9 steel, grade 92 steel also exhibited decrease in dislocation density, coarsening of dislocation substructure and primary carbides/carbontrides, and nucleation and growth of Laves phase, Fe2(W, Mo) with increase in rupture life. Figure 12a shows dislocation substructure in terms of subgrains and dislocation networks along with fine M23C6 carbides and MX type carbides/carbonitrides in N + T condition. Recovery in terms of significant decrease in dislocation density, coarsening of dislocation substructure and precipitates can be seen in creep specimen tested for 32,909 h (Fig. 12b). The observed λ = 6.5 is consistent with the microstructural degradation in grade 92 steel. However, significant reduction in creep ductility at longer durations associated with large of number of secondary cracks occurring due to decohesion of Fe2(W, Mo) at the boundaries appears to be a matter of concern for grade 92 steel.

Fig. 11
figure 11

TEM micrographs showing a presence of M23C6 and Cr2N precipitates in Q + T condition, and coarsening of precipitates in specimens creep tested at b 90 MPa with rupture life 1050 h and c 60 MPa with rupture life 12,575 h. Formation of Laves phase (Fe2Mo) on Cr2N precipitates are shown by arrow in b. Dominance of precipitate coarsening can be seen in the low stress regime in c

Full size image
Fig. 12
figure 12

TEM micrographs showing microstructures in a normalised and tempered condition and b specimen creep tested at 145 MPa for 32,909 h. Increased recovery in terms of significant decrease in dislocation density along with coarsening of precipitates and subgrains can be seen in b

Full size image

4 Conclusions

Both grade 9 and 92 steels exhibited two-slope behaviour with separate values of stress exponents in the low and high stress regimes in the stress dependence of rupture life. Grade 92 steel displayed significantly higher creep-rupture strength than grade 9 steel. High creep ductility along with ductile transgranular fracture was observed in grade 9 steel. Grade 9 steel obeyed Monkman–Grant and modified Monkman–Grant relations. Significant loss of creep ductility due to secondary cracking originating from the decohesion of Laves phase was observed in grade 92 steel. The steel followed generalized form of Monkman–Grant relation with slope ‘m’ lower than unity. Both grade 9 and 92 steels displayed high values of creep damage tolerance factor associated with microstructural degradation in terms of decrease in dislocation density and coarsening of precipitates and dislocation substructure.