Processing through the application of severe plastic deformation (SPD), as in equal-channel angular pressing (ECAP), provides an opportunity for achieving exceptional grain refinement to the submicrometer or even the nanometer level. After SPD processing, these materials may be conveniently used for evaluating the effect of a reduced grain size on the creep behaviour at elevated temperatures under testing conditions of constant load or constant stress. This report provides an overview of the creep properties of ECAP-processed metals with an emphasis on the microstructural characteristics developed by SPD, on their thermal stability and especially on the creep mechanisms that control their flow behaviour. For convenience, these properties are generally compared with the creep behaviour of coarse-grained (CG) samples of the same materials tested under identical conditions.

Fig. 8 Influence of number of ECAP passes on (a) creep resistance,⁴³⁾ (b) strain to fracture.^37,45)

It is generally accepted that methods of severe plastic deformation (SPD) can refine the coarse-grained structure of polycrystalline metallic materials to the ultrafine-grained (UFG, <1 µm) or even the nanostructure (<100 nm) scale.¹⁾ It was found that severe plastic deformation (SPD) leads to significant strengthening of the material at ambient temperatures and to the exceptionally high tensile ductilities with elongations up to >1000% at elevated temperatures.¹⁾ The high temperature behaviour of materials exhibiting such a large ductility occurs through the process of grain boundary sliding (GBS) in which the individual grains are displaced with respect to each other leading to the formation of a randomly oriented microstructure.^2,3) This deformation mechanism was usually observed in alloys at strain rates of about 10⁻³–10⁻² s⁻¹ where GBS is characterized by a stress exponent of n = 2, and inverse grain size dependence with an exponent of 2 and a value for the creep activation energy, Q_c, close to the activation energy for grain boundary diffusion, Q_gb.⁴⁾ However, this is not necessarily the situation for ECAP-processed pure metals and many alloys when testing at medium and lower stresses.^5–7) Further, it was found that ECAP-processed materials exhibit significantly faster creep rates and higher ductility by comparison with their coarse-grained (CG) states and the creep characteristics, n and Q_c, are reasonably consistent with intragranular processes involving the glide and climb of free dislocations rather than with GBS. Although it seems acceptable to associate faster creep rate with smaller grain sizes after ECAP due to the more intensive GBS, there are probably additional creep mechanisms that strongly influence the creep behaviour of these ECAP-processed materials.^8–11) Thus, it is suggested that the various deformation mechanisms in these materials operate independently so that the slowest process is rate-controlling under any given creep conditions. The recent development of SPD-processed materials enables the possibility of studying creep behaviour when the grain size is comparable with a stationary subgrain size so that creep mechanisms related to the grain boundaries become more active.

Accordingly, the aim of this work is to provide a detailed overview of the main experimental results available to date related to the creep behaviour of ECAP-processed materials.

At the present time, much research involves a characterization of the microstructure after the application of severe plastic deformation.^1,6) Frequently, such microstructural characterization is focused only on the mean grain size after high numbers of passes by equal-channel angular pressing (ECAP) where typically this is 8 passes. Nevertheless, it is important to examine the influence of the numbers of ECAP passes on the microstructural heterogeneity and on the subsequent thermal stability during static annealing and creep.

The selected microstructural characteristics for pure ECAP-processed Al are summarized in Fig. 1 for (a) the mean grain size and (b) the frequency of high-angle grain boundaries (HAGBs) as a function of the number of ECAP passes. The results demonstrate^12–17) that increasing numbers of ECAP passes lead to a significant increase in the fraction of HAGBs and to a reduction in the grain size.

Fig. 1

Changes of microstructure characteristics in ECAP-processed Al: (a) reduction of grain size, (b) increase of HAGBs. Data were published in literature.^12–17)

Observations by transmission electron microscopy (TEM) revealed that the largest reduction in the (sub)grain size occurred after 1 ECAP pass and the subsequent ECAP passes had only a negligible effect on the (sub)grain size.^13,15) The increasing numbers of ECAP passes influenced particularly the homogeneity of the microstructure which can be characterized by the coefficients of the profile area variation, CVa:^15,18) thus, the higher are their values so the more pronounced is the inhomogeneity.

After the first pass, the boundaries are predominantly low-angle grain boundaries (LAGBs) in character. However, the fraction of HAGBs increases considerably in materials during the first four ECAP passes. Furthermore, the microstructure processed by low numbers of ECAP passes is very inhomogeneous and the microstructural characteristics may be significantly different in different areas within each specimen (Fig. 2). Studies investigating the evolution of the microstructure in single crystals during ECAP revealed that the macroscopic heterogeneity and dislocation structures can be related to the degree of concentration of the active slip systems with respect to the shear plane.^19,20) Thus, in microstructures processed by low numbers of ECAP passes the initial coarse grains contain predominantly LAGBs and also grains with new HAGBs formed during SPD (Fig. 2). It is readily apparent from Fig. 2 that it is very difficult to determine a mean grain size in the extremely coarse-grained materials processed by SPD.

Fig. 2

Heterogeneity of microstructure in Al after 2 ECAP passes. HAGBs are denoted by black lines and LAGBs are denoted by white lines.

The microstructures processed by 8 and 12 ECAP passes are also usually not fully homogeneous but they can contain large grains exceeding sizes of ∼1 µm and the fraction of LAGBs is generally about 20–30% (Fig. 1). However, the scatter in the microstructural characteristics decreases. A similar trend of microstructural characteristics was also observed in other metals processed by ECAP at room temperature.

It is generally accepted that the microstructure formed by severe plastic deformation techniques, especially of pure metals, is not fully stable even at room temperature²¹⁾ with results demonstrating significant coarsening of the UFG microstructures after self-annealing at room temperature for periods of up to 3²²⁾ and 7²³⁾ years. Grain growth also occurs during heating to the testing temperature and continues during creep testing by dynamic coarsening. The thermal stability of ECAP-processed microstructures is shown in Fig. 3 for (a) UFG Al and (b) an Al–Mg–Sc alloy.^24–31)

Fig. 3

Comparison of thermal stability of microstructure after static annealing and after creep testing (a) in UFG Al and (b) Al–Mg–Sc alloy. Comparison of experimental data from Refs. 24–31).

The results demonstrate that statically-coarsened grains just prior to creep testing are much larger than the size of the ultrafine grains before heating to the testing temperature. The annealed microstructure is quite stable against further static coarsening in short term but the growth of grains occurs during long-term annealing.^26,31) It was observed that the annealing leads to a homogenization of the grain structure but the microstructures with low numbers of ECAP passes remain heterogeneous and contain regions with coarse and fine grains.³²⁾ Furthermore, the annealed microstructure coarsens significantly during subsequent deformation.

The coarsening of ECAP-processed copper during tensile testing at elevated temperatures was investigated in situ within a scanning electron microscope.^33,34) It was found that the microstructure of Cu after 8 ECAP passes tested at 373 K coarsens locally even before the flow stress is reached. This local coarsening, caused probably by dynamic recrystallization, leads to the formation of a banded microstructure with a mix of fine-grained and dynamically-recrystallized regions. The occurrence of dynamically-coarsened regions, with a low defect density immediately after their formation, leads to a localization of plastic deformation into these regions. These results show that local dynamic recrystallization gives an improvement of ductility but does not cause a decrease in the strength.³⁴⁾

At higher creep temperatures, coarsening by static and dynamic recrystallization occurs in a large volume of the ECAP-processed microstructure.³³⁾ The microstructure changes by dynamic recrystallization were so intensive that they caused oscillations in the flow stress during tensile testing of Cu at 573 K. It was also observed that the first wave of dynamic recrystallization occurred at a lower relative maximum of the flow stress than the second wave. By contrast, the opposite results were found for CG Cu.³⁵⁾ This behavior of UFG Cu was explained by the recovery of dislocation at HAGBs where the influence of this recovery process depends on the relation between the grain size, d, and the mean free path of free dislocations, w. It was shown⁹⁾ that this relation is influenced by temperature and applied stress according to d = 2w = 2k_wbG/σ where G is the shear modulus, b is the Burgers vector length, σ is the applied stress and k_w is a numerical factor.

This means in practice that before the first wave the strength of the material is fully controlled by dislocation recovery at the HAGBs. The decrease in the HAGBs fraction and the formation of substructure in the interior of coarse dynamically-recrystallized grains leads to an increase of the strength closer to the CG state. Thus, the second wave for dynamic recrystallization is shifted to a higher stress.

There is a higher thermal stability of UFG microstructures in precipitation-strengthened alloys^28,29) because the grain coarsening in these alloys is restricted by the presence of nanometer-sized precipitates which block the movement of dislocations and stabilize the fine-grained microstructure.

Published research investigating the creep behaviour of alloys was mainly focused on the grain size stability or size of the precipitates.^11,30) However, it was suggested that an important component of the creep behaviour is the formation of dislocation substructure within the grains. In coarse-grained pure metals, the dislocations are usually distributed heterogeneously in the form of subgrains which limit their free path but in SPD-processed materials tested under the same creep conditions the grain sizes are generally smaller than the stationary subgrain sizes and in this case the free paths of dislocations are restricted by the grain boundaries. Some alloys may be strengthened predominantly by solid solution and it is generally accepted that solid solution alloys can exhibit, under some creep conditions, relatively random distributions of dislocations in the grain interiors without any evidence for subgrain formation. Although the dislocation density in solid solution UFG materials has not yet been systematically studied after creep testing, nevertheless the preliminary microstructural results, as shown in Fig. 4 for a Ti–6Al–4V alloy, suggest that grain boundaries in their neighbourhood may reduce the dislocation density.¹¹⁾

Fig. 4

Dislocation structure in grain interior of UFG Ti–6Al–4V alloy after compression creep at 873 K under applied stress 25 MPa.¹¹⁾

Representative creep curves are shown in Fig. 5 for ECAP-processed Al showing (a) creep rate and (b) creep strain plotted as a function of the testing time.^8,36) These plots were obtained at a temperature of 473 K under an applied uniaxial tensile or compressive constant stress or load of 20 MPa. The results demonstrate that ECAP-processed Al exhibits a slightly lower minimum creep rate and a slightly higher ductility than the CG aluminium.^8,36)

Fig. 5

Comparison of creep behaviour of CG and UFG Al: (a) creep rate vs. time, tensile and compression at const. σ and (b) creep strain vs. time, tensile tests under const. σ and const. load (F). Comparison of experimental data from Refs. 8, 36).

Differences in the creep curves were found also in ECAP-processed Cu compared to the CG copper^33,37) and ECAP Cu tested under constant load exhibits markedly longer creep life than the CG material as shown in Fig. 6. However, the minimum creep rate for the ECAP Cu tested at higher stress is slightly lower by comparison with the minimum creep rates of CG Cu but the opposite behaviour is found at low stresses as in Fig. 6. The shapes of the creep curves for the ECAP-processed Cu may exhibit waves of softening and hardening which indicate the occurrence of microstructural changes related to discontinuous dynamic recrystallization.^33,38)

Fig. 6

Creep behaviour of pure CG and UFG copper tested at low and high stresses. The creep curves exhibit waves of softening and hardening which can be related to discontinuous dynamic recrystallization.^33,37)

The coarse-grained and ECAP-processed pure Al in Fig. 5 exhibit a normal primary stage where the creep rate decreases with time into an extended stationary stage. The dislocation creep of pure metals is predominantly controlled by climb of dislocations because dislocation glide occurs very rapidly. But the creep curves of solid-solution alloys may be different because the creep behaviour may be influenced by the dragging of solute atom atmospheres when dislocation glide is dominant.^39,40)

Figure 7(a) shows creep curves for CG and a UFG Al–Mg alloy. The results demonstrate that the CG state exhibits a very short primary stage and it is apparent that the creep rate gradually increases from primary into the stationary stage in the CG state at 30 MPa. This creep behaviour is similar to the inverted primary creep where viscous glide is the rate-controlling process.⁴¹⁾ At the higher stress of 40 MPa the creep rate continues from the primary stage into the stationary region in a linear manner. Such creep behaviour can occur when the rates of glide and climb are approximately the same. The creep curve of the CG alloy measured at 50 MPa exhibits a normal primary stage which shows that dislocation climb is the rate-controlling process. The results demonstrate that all creep curves for the UFG Al–Mg alloy exhibit a normal character.

Fig. 7

Comparison of creep curves for CG and UFG: (a) Al–3Mg alloy,^{57,this work)} (b) Al–3Mg–0.2Sc alloy.³⁰⁾

By contrast, Fig. 7(b) shows the creep curves for CG and UFG Al–Mg–Sc alloy. The creep curves for the UFG state indicate faster creep rates and better ductility by comparison with the CG material and the results demonstrate that a decrease of applied stress below 50 MPa leads to a significant increase in the creep resistance of the CG ternary alloy. However, the significant increase in the creep resistance is also visible in the creep curves for the UFG ternary alloy³⁰⁾ measured at low stresses of 18 and 16 MPa. It was found that a decrease in the applied stresses by only 2 MPa caused a decrease in the stationary creep rate of about one order of magnitude.

In earlier works^5,36,42) it was found that pure metals after 1 ECAP pass tested in tension can exhibit significantly higher creep resistance by comparison with their CG states (Fig. 8(a)). This suggests that the highest creep resistance is exhibited by the specimens with an inhomogeneous microstructure containing a high fraction of LAGBs. The effect of LAGB strengthening decreases with a reduction of the grain size and with an increase in the fraction of HAGBs in the microstructure. Nevertheless, precipitate-strengthened alloys usually show a deterioration of creep resistance by comparison with their as-received states even after one ECAP pass.⁴³⁾

Fig. 8

Influence of number of ECAP passes on (a) creep resistance,⁴³⁾ (b) strain to fracture.^37,45)

It was found that a further reduction of grain size and increasing homogeneity of microstructure leads to a decrease in the creep resistance and to an improvement of the ductility in ECAP-processed materials as in Fig. 8.^37,43)

The dependence of creep rate on strain,³⁷⁾ measured in tension under constant load, demonstrates that an increasing number of ECAP passes, p, leads to an increase in the strain as shown in Fig. 8(b). It is readily apparent that the highest changes in creep resistance and ductility occur during the first four ECAP passes in Fig. 8(a), (b). A further increase of ECAP passes, leading to a finer grain size and better microstructural homogeneity, causes an additional slight decrease of the creep resistance and an improvement in the creep ductility. However some CG Cu specimens exhibit similar or even slightly larger fracture ductility in comparison to Cu processed by 1 ECAP pass. One can see that the largest part of ductility in some CG Cu specimens occurred predominantly in the tertiary stage due to multiple peaks of hardening and softening. The multiple peaks in creep curves for CG Cu can be caused by discontinuous dynamic recrystallization.³⁷⁾ The occurrence of dynamic recrystallization in the tertiary stage of CG Cu may explain the relatively long fracture period with the regions of softening and hardening.

It was found that the creep resistance of the CG and ECAP states is influenced by the type of creep testing.⁴⁴⁾ The results demonstrate that UFG Cu tested in tension under constant load exhibits a longer creep live and approximately a similar minimum creep rate as in CG Cu as shown in Fig. 9. Nevertheless, the opposite results were found when Cu was tested in compression under a constant stress.

Fig. 9

Creep behaviour of CG and UFG copper under tension and compression: (a) creep rate vs. time, (b) creep rate vs. strain.

The tensile creep tests at constant load are influenced during creep exposure by the increase of stress due to the decrease in the initial specimen cross-sectional area given for constant stress in tension. For this reason, the secondary stage may be understood as the transition region where the decrease of creep rate in the primary stage is compensated by the increase of creep rate in the tertiary stage.⁴⁵⁾ The tertiary stage of tensile creep is commonly discussed in terms of a creep acceleration by ongoing creep degradation processes and fracture.

However, the creep rate during tensile testing at constant load increases not only due to the damage processes and fracture because, even in the absence of fracture, there is a continuous increase of stress because the stationary creep rate is a strong function of stress. In order to determine whether the fracture or stress increases are the dominant cause of tertiary creep, it is necessary to compare the dependence of creep rate on stress in the tertiary stage and in the stationary stage. For the situation where fracture is absent the curves of creep rate vs. stress should merge with a stress dependence measured at the same creep conditions. This was found with a certain approximation for the creep of UFG Cu⁴⁶⁾ but not for CG Cu. Therefore, it was concluded that the tertiary stages of creep, as well as the values of the minimum creep rates, are probably influenced by fracture and should not be interpreted as stationary values resulting from uniform deformation.

For the situation where the influence of fracture was eliminated using compression tests at constant stress, the minimum creep rate of CG Cu was significantly displaced down to the slower creep rates. A compression test at 473 K and 80 MPa ended in the work hardening stage of the so-called primary creep because the overall creep had become too slow.

The dependences of the minimum creep rate on the applied stress for the ECAP-processed Al are shown in Fig. 10(a). It is observed that the minimum creep rate of the ECAP-processed Al measured at high stresses may be up to one order of magnitude lower than for the CG state. The results demonstrate that the difference in minimum creep rates of CG and ECAP-processed Al decreases significantly with decreasing applied stress. The values of the stress exponent $n = (\frac{d\,ln\,\dot{\varepsilon }}{d\,ln\,\sigma })$ were about 6.6 for the CG Al and about 4.8 for the ECAP-processed Al.³⁶⁾ Similar results were found also in another investigation⁸⁾ where it was observed that the stress exponent of the ECAP-processed Al is equal to ∼5 and the minimum creep rate is slightly slower by comparison with the CG state.

Fig. 10

Stress dependences of minimum creep rate for pure CG and UFG states: (a) Al tested at 473 K under constant stress,^8,36) (b) Cu tested at 295–573 K in tension,^47–49) stress dependence of CG Cu at 295 K is the result of the model.^48,49)

The stress dependences of the minimum creep rates for UFG and coarse-grained copper are shown in Fig. 10(b) where the differences in the minimum creep rates of CG and UFG Cu tested at 295–573 K depend on the applied stress and testing temperature.^46,47) The results demonstrate that the minimum creep rates of CG copper tested at 473 and 573 K at low stresses are slower by comparison with the UFG material. The values of the stress exponent, n, determined for CG Cu are higher than the values of n for UFG Cu. For this reason the stress dependences of CG and UFG Cu intersect and thus the opposite result is found at higher stress.

Creep tests carried out on CG and UFG Cu at temperatures below 473 K demonstrate that the strength of UFG Cu is significantly higher by comparison with the CG state. However, the differences in the creep strength of CG and UFG Cu decrease with decreasing value of the initial applied stress as shown in Fig. 10(b).

Figure 11 demonstrates the stress dependence for Al alloys tested at 473 K in compression under constant load.^30,43) Thus, a reduction in grain size leads to the faster minimum creep rate in both solid solution and precipitation-strengthened alloys by comparison with their CG counterparts. Inspection of Fig. 11 shows that, although the minimum creep rate for the UFG Al–3%Mg alloy is more than one order of magnitude faster than the CG state, the values of the stress exponents, n, are very similar. This implies that creep may be controlled by the same creep mechanism but the shapes of the creep curves suggest that the deformation mechanisms in UFG and CG Al–3%Mg alloy are probably different.

Fig. 11

Stress dependences of minimum creep rate for selected CG and UFG Al alloys.^30,43,57)

It was found that the UFG Al–3Mg–0.2Sc alloy strengthened by a combination of solid solution and precipitates exhibited a decrease of the stress exponent n at higher stress by comparison with the CG state. However, the slope of the stress dependence for the UFG Al–Mg–Sc alloy increases significantly at stresses lower than 20 MPa and the high value of the stress exponent indicates the effect of a threshold stress. Nevertheless, typical sigmoidal curves indicating a threshold stress are not observed in the CG state and this may be explained by the relatively low temperature used for the creep testing and the coarse grain size of ∼200 µm which leads to a change of the stress dependence curve from sigmoidal towards a flatter form. The occurrence of a threshold may be assumed in the CG ternary alloy at low stresses because of the appearance of the creep curves in Fig. 11 which show that the creep strain at 16 MPa is very low so that creep is almost unmeasurable. Although the ternary alloy is a fully strengthened Al alloy, the minimum creep rates are faster by comparison with the Al–Mg alloy. This different creep behaviour is explained by the higher grain size stability in the ternary Al alloy which leads to a higher activity of boundary mediated processes.

The largest differences in the minimum creep rates between the CG and UFG states were exhibited by a precipitation-strengthened Al–Sc alloy where the results demonstrated that Al–Sc showed no evidence for the occurrence of a threshold stress.⁵⁰⁾ It was reported that the threshold stress in Al–Mg–Sc and Al–Sc binary alloys occurs in a similar stress range when Al alloys are aged and treated in the same way to give similar precipitate radii. However, there is evidence for Al alloys^30,43) that processing by different heat treatment leads to different precipitate sizes and thus the threshold stress in the Al–Sc alloy is probably displaced to a lower stress.

Creep tests performed at different temperatures and the same constant stress permit a determination of the activation energy of creep, Q_c. Thus, the activation energy may be derived from the slope of a semi-logarithmic plot of dε/dt versus 1/T for a given temperature interval. The activation energy for creep, Q_c, is then defined as:   

Previously published data^5,6) showed that the activation energies of creep for the CG state are higher under the same testing conditions than those for UFG materials as shown in Fig. 12. It was also found that the value of Q_c for UFG states increases with an increasing value of the applied stress. Opposite results are found for CG Cu where the values of Q_c decrease with increasing values of the applied stress and the results demonstrate that the value of Q_c is associated with fraction of HAGBs in the microstructure which tend to shift the value of Q_c down to the activation energy for boundary diffusion. However, the value of the activation energy for grain boundary diffusion Q_gb is similar to the value of the activation energy for pipe diffusion Q_p.⁵²⁾ It can be suggested that the influence of the pipe diffusion on creep behavior increases with increasing density of dislocations in the grain interior which increases with the increasing value of the applied stress. On the basis of this assumption, the value of the activation creep energy should be low also at high stresses or even decrease with increasing value of applied stress down to the activation energy for pipe diffusion as a consequence of the increase of dislocation density. However it was not observed for pure aluminum.

Fig. 12

Activation creep energy Q_c for CG and UFG states. Experimental data were taken from Refs. 5, 6, 37, 42, 51).

But low values of the activation energies for creep close to Q_gb or Q_p were found in UFG Al alloy at low and also high stresses¹⁰⁾ and it was suggested that intragranular dislocation process can be influenced by dislocation pipe diffusion.

It is generally accepted that during intragranular dislocation creep processes there may be contributions from both lattice and pipe diffusion. This can be expressed by the following equation:⁴⁰⁾   

The creep mechanism(s) are usually determined from the value of the stress exponent n and activation energy for creep Q_c.⁴⁰⁾ For convenience, the main creep characteristics of different SPD-processed materials are listed in Table 1.

Table 1 Creep data for ECAP processed materials.

The proposed deformation mechanisms influencing creep behaviour of SPD processed metals may be divided into three main groups: a) dislocation creep with a different contribution from GBS,⁵⁾ b) enhanced recovery of dislocations at HAGBs⁹⁾ and c) a dislocation glide process.¹⁰⁾ The current microstructure results show that the fine grains formed during SPD are not stable but instead they grow during heating to the testing temperature and subsequent creep testing as in Fig. 3. The thermal instability of the initial UFG structure in pure metals probably leads to values of the stress exponent that are significantly higher by comparison with the values of n typical for grain boundary mediated creep processes such as diffusion creep or GBS. The values of the stress exponent n for pure UFG metals show that power-law creep associated with glide and climb of intragranular dislocations plays a dominant role, as demonstrated by the data summarized in Table 1. This suggestion is supported by an insufficient thermal stability of the UFG microstructure when the mean grain size corresponds with fine-grained or even with coarse-grained regions.

The experimental creep results in Figs. 8 and 10 show that ECAP-processed pure materials can exhibit faster minimum creep rates and higher ductilities by comparison with their CG counterparts. For the case where dislocation or power-law creep is the dominant process, the activation energy for creep Q_c should be close to the value for the activation enthalpy of lattice self-diffusion in pure metals.

The values of the activation creep energies determined in SPD-processed materials are usually lower^5,10,53,54) and decrease with decreasing applied stress. This observation suggests that creep processes in SPD-processed pure metals may be influenced by the large numbers of grain boundaries that are a direct consequence of the SPD processing.^9,36) Although the values of the stress exponent, n, indicate that power law creep is the dominant mechanism, nevertheless some experiments demonstrated the occurrence of GBS from surface observations on tensile specimens.^5,27,38)

There is also a report in which tensile flat specimens were tested only up to a predetermined strain of ∼0.15 in order to eliminate the influence of any damage processes.⁵⁾ In these experiments there were reports of the occurrence of GBS at some grain boundaries from surface observations and it was estimated that the contribution of GBS to the total creep strain was of the order of about 30% for UFG Al.

In other experiments it was suggested that the creep behaviour of UFG materials may be influenced by an enhanced storage and recovery of dislocations at boundaries for grains that are finer than the equilibrium subgrain size in the CG material.⁹⁾ It was shown^26,46) that this mechanism governs the creep behaviour when the grain size is reduced below a critical size given by the stationary subgrain size and thus the microstructure contains more than about 50% of HAGBs. However the influence of HAGBs on the creep strength also depends on the creep conditions.^26,33) The creep results (Fig. 10(b)) indicate that the enhancement or the deterioration of creep properties of UFG materials in comparison with their CG states is influenced by testing temperature and applied stress. One can see that the creep strength of UFG material becomes better than the creep strength of their CG state with decreasing testing temperature and increasing applied stress. This creep behavior was explained by Blum et al.^9,33) It was suggested that creep behavior is influenced by HAGBs which have two opposing effects. HAGBs form strong obstacles to dislocation glide and thus they tend to harden the material (well known under the term “Hall–Petch”). At the same time, they tend to soften the material through various hab-mediated processes. On the basis of this suggestion, all UFG materials can exhibit, depending on the creep testing conditions, an improvement or a deterioration of creep properties in comparison with their CG states. When creep tests are performed at high strain rates and/or low temperatures the influence of hab-mediated processes, such as GBS and enhanced recovery of dislocation at HAGBs, is suppressed and thus the HAGBs become strong obstacles to dislocation glide and accordingly they harden the material. However the effect of hab-mediated processes on creep of UFG materials is higher at higher temperature and lower applied stress where HAGBs have the tendency to soften UFG material.^9,33)

Experiments investigating the creep behaviour of aluminium alloys^10,30,43,60) found that the stress exponent n can be about 3 and this value is consistent with a dislocation glide process. The deformation process is then characteristic for solid solution alloys tested at intermediate stresses when the stress exponent is reduced from a value of ∼5 for control by dislocation climb to a value of ∼3 when viscous glide becomes the rate-controlling process. However, the creep behaviour of UFG alloys fails to show an inverse primary creep but rather it shows a normal primary creep which is typical of pure metals. The influence of grain size on jerky glide as in the Portevin-Le Chatelier effect was investigated in a UFG Al–3Mg alloy and it was found that jerky glide, which is controlled by a repetitive interaction and break-away of dislocations from clouds of solute atoms, can be suppressed in UFG alloys.⁶¹⁾ This suppression of jerky glide is influenced by the increasing effect of the HAGB-mediated processes in UFG alloys such as an enhanced recovery of dislocations at boundaries or the occurrence of GBS along the HAGBs.

Finally, it should be noted that a detailed analysis of the creep behaviour of Al–Sc and Al–Mg–Sc alloys showed that processing by ECAP at room temperature initially produced submicrometer grain sizes. In subsequent testing these materials gave rates that were too fast to be explained by Nabarro-Herring diffusional creep but instead they were in very good agreement with a model developed for flow by GBS in conventional materials where the grain size is smaller than the equilibrium subgrain size.^62,63) This latter result has now been supported by a more comprehensive analysis of a very large number of Al and Mg alloys processed by SPD techniques.⁶⁴⁾

An investigation of the surfaces of ECAP-processed specimens taken from the gauge lengths after creep testing revealed the appearance of mesoscopic shear bands lying near to the shear plane of the last ECAP pass as shown in Fig. 13(a). It was observed that the frequency of mesoscopic shear bands observed near the fracture increases rapidly with increasing numbers of ECAP passes.^43,65)

Fig. 13

Fracture behaviour of ECAP-processed materials: (a) formation of mesoscopic shear bands on surface of UFG Al–Sc alloy during creep,⁴³⁾ (b) intergranular creep cavitation in UFG Cu,⁶⁵⁾ (c) intergranular fracture in CG Cu–Zr alloy and (d) ductile intragranular fracture in UFG Cu–Zr alloy.

The formation of these mesoscopic shear bands is related to the inhomogeneity of the microstructure in ECAP-processed specimens after creep.³⁴⁾ It is apparent that CG Cu and Cu alloys tested in tension under constant load exhibit intergranular fracture as in Fig. 13(c). The intergranular cavity formation in Fig. 13(b) was also found in the surface of UFG pure Cu⁶³⁾ and its alloys after creep but creep fracture of ECAP-processed materials, tested under the same conditions, always exhibited a ductile intragranular fracture mode with numerous dimples on the relief as is evident in Fig. 13(d).

The production of SPD-processed materials permits an investigation of the effect of grain size and numbers of HAGBs on the creep behaviour of these materials. This overview focusses particularly on experimental results relating to creep microstructures and the creep mechanisms associated with SPD-processed materials.

This research was supported by the Ministry of Education, Youth and Sports of the Czech Republic through the project IPMinfra No. LM2015069. One of the authors (TGL) was supported by the European Research Council under ERC Grant Agreement No. 267464-SPDMETALS. Long-term interactions and stimulating discussions with Profs. W. Blum of University of Erlangen – Nürnberg, Erlangen, Germany and Z. Horita of Kyushu University, Fukuoka, Japan have been of great value.

• 1)   R.Z.  Valiev,  R.K.  Islamgaliev and  I.V.  Alexandrov: Prog. Mater. Sci. 45 (2000) 103–189.
• 2)   P.  Kumar,  M.  Kawasaki and  T.G.  Langdon: J. Mater. Sci. 51 (2016) 7–18.
• 3)  O.A. Kaibyshev: Superplasticity of Alloys, Intermetallides and Ceramics, (Springer-Verlag, Berlin, 1992).
• 4)   T.G.  Langdon: Acta Metall. Mater. 42 (1994) 2437–2443.
• 5)   V.  Sklenicka,  J.  Dvorak,  P.  Kral,  Z.  Stonawska and  M.  Svoboda: Mater. Sci. Eng. A 410–411 (2005) 408–412.
• 6)  W. Blum, P. Eisenlohr and V. Sklenička: Bulk Nanostructured Materials, ed. by M. Zehetbauer and Y.T. Zhu, 1st edition, (Wiley-VCH, Weinheim, 2009) Chapter 24, pp. 519–537.
• 7)   Y.J.  Li,  X.H.  Zeng and  W.  Blum: Acta Mater. 52 (2004) 5009–5018.
• 8)   M.  Kawasaki,  I.J.  Beyerlein,  S.C.  Vogel and  T.G.  Langdon: Acta Mater. 56 (2008) 2307–2317.
• 9)   W.  Blum and  X.H.  Zeng: Acta Mater. 57 (2009) 1966–1974.
• 10)   H.-K.  Kim: J. Mech. Sci. Technol. 24 (2010) 2075–2081.
• 11)   P.  Kral,  J.  Dvorak,  W.  Blum,  E.  Kudryavtsev,  S.  Zherebtsov,  G.  Salishchev,  M.  Kvapilova and  V.  Sklenicka: Mater. Charact. 116 (2016) 84–90.
• 12)   M.  Kawasaki,  Z.  Horita and  T.G.  Langdon: Mater. Sci. Eng. A 524 (2009) 143–150.
• 13)   M.  Cabibbo,  W.  Blum,  E.  Evangelista,  M.E.  Kassner and  M.A.  Meyers: Metall. Mater. Trans. A 39 (2008) 181–189.
• 14)   R.  Kocich,  L.  Kunčická,  P.  Král and  A.  Macháčková: Mater. Charact. 119 (2016) 75–83.
• 15)   I.  Saxl,  A.  Kalousová,  L.  Ilucová and  V.  Sklenička: Mater. Charact. 60 (2009) 1163–1167.
• 16)   L.  Kunčická,  R.  Kocich,  P.  Král,  M.  Pohludka and  M.  Marek: Mater. Lett. 180 (2016) 280–283.
• 17)   A.P.  Zhilyaev,  D.L.  Swisher,  K.  Oh-ishi,  T.G.  Langdon and  T.R.  McNelley: Mater. Sci. Eng. A 429 (2006) 137–148.
• 18)   L.  Ilucova,  I.  Saxl,  M.  Svoboda,  V.  Sklenicka and  P.  Kral: Image Anal. Stereol. 26 (2007) 37–43.
• 19)   H.  Miyamoto,  J.  Fushimi,  T.  Mimaki,  A.  Vinogradov and  S.  Hashimoto: Mater. Sci. Eng. A 405 (2005) 221–232.
• 20)   Y.  Fukuda,  K.  Oh-ishi,  M.  Furukawa,  Z.  Horita and  T.G.  Langdon: Acta Mater. 52 (2004) 1387–1395.
• 21)   Y.  Huang,  S.  Sabbaghianrad,  A.I.  Almazrouee,  K.J.  Al-Fadhalah,  S.N.  Alhajeri and  T.G.  Langdon: Mater. Sci. Eng. A 656 (2016) 55–66.
• 22)   P.  Král,  J.  Dvořák,  M.  Kvapilová,  W.  Blum and  V.  Sklenička: Mater. Lett. 188 (2017) 235–238.
• 23)   A.I.  Almazrouee,  K.J.  Al-Fadhalah,  S.N.  Alhajeri,  Y.  Huang and  T.G.  Langdon: Adv. Eng. Mater. 21 (2019) 1801300 (1–9).
• 24)   H.  Hasegawa,  S.  Komura,  A.  Utsunomiya,  Z.  Horita,  M.  Furukawa,  M.  Nemoto and  T.G.  Langdon: Mater. Sci. Eng. A 265 (1999) 188–196.
• 25)   W.Q.  Cao,  A.  Godfrey,  W.  Liu and  Q.  Liu: Mater. Sci. Eng. A 360 (2003) 420–425.
• 26)   W.  Blum,  J.  Dvorak,  P.  Kral,  P.  Eisenlohr and  V.  Sklenicka: J. Mater. Sci. Technol. 32 (2016) 1309–1320.
• 27)   P.  Kral,  J.  Dvořák,  A.  Jäger,  M.  Kvapilová,  Z.  Horita and  V.  Sklenička: Kovove Mater. 54 (2016) 441–451.
• 28)   P.H.R.  Pereira,  Y.  Huang and  T.G.  Langdon: Mater. Res. 20 (2017) 39–45.
• 29)   P.B.  Berbon,  S.  Komura,  A.  Utsunomiya,  Z.  Horita,  M.  Furukawa,  M.  Nemoto and  T.G.  Langdon: Mater. Trans. JIM 40 (1999) 772–778.
• 30)  V. Sklenicka, J. Dvorak, M. Svoboda, P. Král, M. Kvapilova and Z. Horita: Ultrafine Grained Materials IV, ed. by Y. Zhu, T.G. Langdon, Z. Horita, M.J. Zehetbauer, S.L. Sermiatin and T.C. Lowe, (TMS, Warrendale) pp. 450–464.
• 31)   P.  Kral,  J.  Dvorak,  M.  Kvapilova,  Z.  Horita and  V.  Sklenicka: Letters Mater. 5 (2015) 306–312.
• 32)   I.  Saxl,  V.  Sklenicka,  L.  Ilucova,  M.  Svoboda,  P.  Kral and  J.  Dvorak: Rev. Adv. Mater. Sci. 25 (2010) 233–240.
• 33)   W.  Blum,  J.  Dvořák,  P.  Král,  M.  Petrenec,  P.  Eisenlohr and  V.  Sklenička: Philos. Mag. 95 (2015) 3696–3711.
• 34)   P.  Král,  W.  Blum,  J.  Dvořák,  P.  Eisenlohr,  M.  Petrenec and  V.  Sklenička: Mater. Charact. 134 (2017) 329–334.
• 35)   L.  Blaz,  T.  Sakai and  J.J.  Jonas: Met. Sci. 17 (1983) 609–616.
• 36)   V.  Sklenicka,  J.  Dvorak and  M.  Svoboda: Mater. Sci. Eng. A 387–389 (2004) 696–701.
• 37)   W.  Blum,  J.  Dvořák,  P.  Král and  V.  Sklenička: Mater. Sci. Eng. A 731 (2018) 520–529.
• 38)   K.V.  Ivanov and  E.V.  Naydenkin: Mater. Sci. Eng. A 608 (2014) 123–129.
• 39)   F.A.  Mohamed and  T.G.  Langdon: Acta Metall. 22 (1974) 779–788.
• 40)  J. Cadek: Creep in Metallic Materials, (Elsevier, Amsterdam, The Netherlands, 1988).
• 41)  B. Wilshire and R.W. Evans (Eds): Creep Behaviour of Crystalline Solids, (Pineridge Press, Swansea, U.K., 1985).
• 42)   J.  Dvorak,  V.  Sklenicka,  P.  Kral,  M.  Svoboda and  I.  Saxl: Rev. Adv. Mater. Sci. 25 (2010) 233–240.
• 43)   P.  Král,  J.  Dvořák and  V.  Sklenička: Mater. Sci. Forum 584–586 (2008) 846–851.
• 44)   W.  Blum,  J.  Dvorak,  P.  Kral,  P.  Eisenlohr and  V.  Sklenicka: J. Mater. Sci. 49 (2014) 2987–2997.
• 45)   W.  Blum,  J.  Dvorak,  P.  Kral,  P.  Eisenloh and  V.  Sklenicka: J. Mater. Sci. Technol. 31 (2015) 1065–1068.
• 46)   W.  Blum,  J.  Dvorak,  P.  Kral,  P.  Eisenlohr and  V.  Sklenicka: Mater. Sci. Eng. A 590 (2014) 423–432.
• 47)  P. Kral, J. Dvorak, M. Kvapilova, M. Svoboda and V. Sklenicka: Conference on Creep and Fracture of Engineering Materials and Structures (JIMIS 11), (The Japan Institute of Metals, Sendai, 2012) No. C39 [CD-ROM].
• 48)   W.  Blum,  Y.J.  Li,  Y.  Zhang and  J.T.  Wang: Mater. Sci. Eng. A 528 (2011) 8621–8627.
• 49)   R.  Sandström and  H.C.  Andersson: J. Nucl. Mater. 372 (2008) 76–88.
• 50)   E.A.  Marquis,  D.N.  Seidman and  D.C.  Dunand: Acta Mater. 51 (2003) 4751–4760.
• 51)   Z.  Tobolova and  J.  Cadek: Philos. Mag. 26 (1972) 1419–1428.
• 52)  M. Ashby and H. Frost: Deformation-Mechanism Maps, (Pergamon Press, Oxford, U.K., 1982).
• 53)   Yu.R.  Kolobov,  G.P.  Grabovetskaya,  I.V.  Patochka and  K.V.  Ivanov: Russ. Phys. J. 41(3) (1998) 260–264.
• 54)   Y.J.  Li,  X.H.  Zeng and  W.  Blum: Acta Mater. 52 (2004) 5009–5018.
• 55)   W.  Blum,  Y.J.  Li and  K.  Durst: Acta Mater. 57 (2009) 5207–5217.
• 56)   C.  Xu and  T.G.  Langdon: Mater. Sci. Eng. A 410–411 (2005) 398–401.
• 57)  J. Dvorak, P. Král, M. Kvapilová, V. Sklenicka and M. Svoboda: 16th International Metallurgical & Material Conference (Metal 2007), Conference Proceedings, (Hradec nad Moravicí, 2007) No. 119 [CD-ROM] (in Czech).
• 58)   P.  Kral,  M.  Svoboda,  J.  Dvorak,  M.  Kvapilova and  V.  Sklenicka: Acta Phys. Pol. A 122 (2012) 457–460.
• 59)   J.  Dvorak,  P.  Kral,  M.  Svoboda,  M.  Kvapilova and  V.  Sklenicka: IOP Conf. Ser. Mater. Sci. Eng. 63 (2014) 012141.
• 60)   W.  Blum,  P.  Eisenlohr and  F.  Breutinger: Metall. Mater. Trans. A 33 (2002) 291–303.
• 61)   T.A.  Lebedkina,  M.A.  Lebyodkin,  T.T.  Lamark,  M.  Janeček and  Y.  Estrin: Mater. Sci. Eng. A 615 (2014) 7–13.
• 62)   M.  Kawasaki,  V.  Sklenička and  T.G.  Langdon: J. Mater. Sci. 45 (2010) 271–274.
• 63)   M.  Kavasaki,  V.  Sklenicka and  T.G.  Langdon: Kovove Mater. 49 (2011) 75–83.
• 64)   M.  Kawasaki and  T.G.  Langdon: J. Mater. Sci. 51 (2016) 19–32.
• 65)   V.I.  Betekhtin,  A.G.  Kadomtsev,  P.  Král,  J.  Dvorak,  M.  Svoboda,  I.  Saxl and  V.  Sklenička: Mater. Sci. Forum 567–568 (2008) 93–96.