Reflections on creep: a review of the mst archive

Reflections on creep: a review of the mst archive

Amongst many themes embedded in Materials Science and Technology (mst), the phenomenon of creep has remained a significant component, seamlessly continuing from the journals through which mst was directly descended. The Journal of the Iron and Steel Institute (JISI), from its founding in 1869, recorded procedures to manufacture steels to withstand arduous conditions to improve the performance of furnaces and engines. The problem of creep was identified and its exploration included its occurrence in non-ferrous metals and alloys with results on these published in the Journal of the Institute of Metals (JIM). The sensitivity of creep to materials’ processing as well as to chemical composition was soon apparent. The brief commentary here outlines some of the main features in this area as reported by the successive line of journals, but it gives no more than a taste the wealth of information, ideas and proposals they contain.

Early studies were necessarily empirical but became more focused by progressively controlling parameters that led to variable results. The shape of the strain versus time curve at constant uni-axial stress and temperature was a prominent consideration with formulae proposed for its description.1 Such attention has continued with increasing sophistication2 aided by the data processing power of computers. Experimental results and the capability of their extrapolation are vital ingredients in attempting to make realistic prediction of long term behaviour to times greatly in excess of those feasible in laboratory testing. Within relatively narrow ranges of material composition and microstructure and of imposed conditions such efforts have made valuable progress but the high sensitivity of the creep process to both internal and external variables indicates that unique relationships to cover a wider range of conditions are unlikely to emerge in a simple, precise and accurate form.

It was not until the late 1940s, aided by improved understanding of the character and influence of crystal defects and of grain boundaries, that serious attempts could begin to interpret underlying mechanisms.3^–9 These efforts were urgently required to meet the universal demand for improved materials.10^–12 imposed by new engine designs, notably the gas turbine, and in the search for improved efficiency in fossil fuelled power plant and in the novel requirements emerging from nuclear power generation. Experimental investigations in these areas were aided by rapid innovations in laboratory instruments and techniques and were supported by greater knowledge of the presence of atomic scale defects in materials and of their mobility.

For the most demanding situations, requiring toughness, oxidation and corrosion resistance as well as withstanding high stresses at high temperatures, the excellent properties of nickel and nickel based alloys in high temperature environments were already recognised but much scope remained for further improvement. Additions of aluminium were noted to promote creep resistance when appropriate heat treatments were given. In 1956–57, JIM was first to publish convincing evidence of this13 through the identification of the importance of the gamma prime phase Ni₃Al. This foundation has expanded extensively over subsequent decades with transmission electron microscopy playing a predominant role. The consequences have been far reaching and of huge technological value through understanding particle characteristics, interfaces and dislocation–particle interaction. These studies have further aided improvement in high temperature mechanical properties of less expensive materials covering steels14,15 and some light alloys.

Many contributions have aimed at interpretation of creep behaviour.16^–19 These were aided by substantial support for basic scientific research in the post-war years that encouraged studies20^–24 on metals with well characterised microstructures and accurately known levels of impurity. Much of this work was nevertheless driven by practical and commercial needs, urgently seeking data and better understanding of the behaviour of materials under severe conditions and in entirely new environments such as those inherent in nuclear power generation. Papers concerned with these areas have necessarily invoked extensive understanding of broad areas of metallurgy. As well as enhancing surface reactions, elevated temperatures allow precipitates to coarsen, grain boundaries to migrate and shear, impurities to segregate, dislocations to climb as well as glide, vacancies to be created, diffuse, condense and be absorbed.

The discontinuation through merger of the much respected JISI and JIM led to the succession of new journals, Metal Science Journal (MSJ, subsequently Metal Science, founded in 1967) and Metals Technology (MT, which began publication in 1974). Indeed, the first paper25 in MSJ concerned the nucleation of grain boundary cavities in creep. This has proved to be a continually recurring theme and the Journal reflected the effort26 devoted to its understanding. Cavity growth has proved more revealing of its mode of operation,27 though not without an appreciation of its complexities and a recognition of the circumstances under which cavity growth may be constrained.28 Proposals have been extended29 for the condensation of vacancies to enlarge cavities on grain boundaries to a size where they eventually coalesce and cause fracture, even after small (a few per cent or even less) strain. Such processes can occur when the applied stress is low (a few megapascal) and the material relatively soft: a reminder that fracture can take place when thermodynamic conditions are satisfied and suitable mechanisms for atomic movement are available. The roles of directional vacancy flow and of dislocation glide were shown, however, not to be mutually exclusive when different regions, close to and away from cavities, were considered. In the first analysis of such a situation30 a convincing description was given of cavity growth by vacancy condensation in a material creeping mainly through dislocation movement. Such an approach has proved capable of further extensions. These include analyses of the influence of multi-axial stresses31,32 and of situations where cracks rather than cavities are responsible for grain boundary removal.33,34 Much has ensued from this, with the realisation that sudden fracture can occur in creep at low strains, often in materials that are more ductile at lower temperatures under higher stress.

Amongst the microstructural changes35 resulting from creep under low stresses, one of the more prominent has been the observation in some materials of precipitate denuded regions adjacent to those grain boundaries most closely perpendicular to the principal tensile stress. A satisfactory interpretation can be derived directly from creep occurring through the stress induced directional flow of vacancies.36 In polycrystals the occurrence of this phenomenon relies on the capability of grain boundaries to act as sources and sinks for vacancies and on the absence of intergranular dislocation movement. Where such conditions are met, soundly based analyses have been provided37 in the cases where the influence of grain or grain boundary diffusion predominates, to permit predictions of behaviour based on calculations of internal stress distribution. The rate of creep by this mechanism can be analysed in its entirety with formulae emerging that are free from empirical parameters. Thus, predictions can be made of the response to any system of stresses acting on components of any geometry in materials with grains of known size and shape.

Although results through conventional creep testing techniques have dominated in the majority of papers, novel methods have also been described, notably to increase sensitivity,38 to take into account specific geometrical features to throw light on the influence of different stress systems and to obtain creep data from miniature specimens.39^–42 Tests on helically coiled wires have shown how the accuracy of strain measurement can be increased by more than two orders of magnitude whilst minimising errors through temperature variation, though such tests invoke a non-uniform stress in testpieces and so errors are introduced in analysis unless creep rate is linearly related to stress. An analogous limitation arises when creep is studied in bend specimens, though if these have a bamboo grain structure there is opportunity for exploring circumstances where creep is related to grain boundary diffusion.43 Substantial practical benefit can be obtained through using miniature specimens, especially to obtain information of remaining life before failure, when these can be taken from materials with a previous life in service.

Numerous papers44^–48 have taken a broader look at the creep process, some renewing the search for strain–time relationship in creep, the onset of failure and the prediction of creep life.49 Studies of this kind have been made on materials similar or close to those envisaged for commercial application,17,50^–55 pointing the way towards useful further alloy development. These, more general approaches have been reported alongside more specific studies focusing, as far as possible, on the endeavour to identify individual relationships. Such studies have included the exploration of grain boundary sliding22,56^–58 and other themes, including dislocation rearrangements during creep59^–62 and the variation of density change with creep strain.63,64 Recovery processes have shed further light on internal restructuring.65^–68 Similarly, studies of the effects of complex stress systems,69 the combined effects of torsion and tension70 and of hydrostatic pressure71 have provided some understanding of cavity shrinkage for comparison with their rate of growth under tensile stresses. Parametric relationships have also received attention in attempts to extract overall quantitative evaluation of behaviour.19,72

In 1985, the first year of the life of mst, substantial attention was given to problems of creep and creep fracture of steels and nickel based alloys73,74 covering microstructural changes,75 crystal defect behaviour76^–78 and the influence of trace elements.79 Concepts of fracture mechanics,80,81 more familiar in the interpretation of brittleness at lower temperatures, were introduced to explain observations of multiple creep cracking. The influence of gas bubbles was also explored.82 However, the topic encompassing the largest number of papers83^–90 in the first volume of mst concerned the extent to which the ravages of creep damage could be rectified. Ingenious practical proposals of procedures for repair and rejuvenation were put forward, often with attempts to evaluate their success. In the following volume the alternative side to this problem was emphasised in a series of papers91^–114 aiming for reduction of the rate of degradation of superalloys in service through better schemes for their environmental protection. A wide range of coating methods and procedures are described for the surface protection of these alloys in gas turbine engines used in aerospace, marine and on land.

Several papers115^–120 have covered casting techniques of superalloys with regard to cost effective methods of production and the higher level of precision required in components of increasingly complex shape. This area has been explored in extensive detail, particular attention being paid to the resulting microstructures, with needs progressively extending from materials with random polycrystals, through directionally solidified anisotropic structures to single crystals with close and reliable control of their orientation.

In extending conventional investigations of creep, further parameters have been included. The effects of external stresses acting concurrently with internal stresses became of more significance when it was appreciated that internal stresses could be continuously generated.121 This realisation was enhanced through effects observed in materials in nuclear reactors, especially when metallic uranium was used as a fuel. The α phase of this element is highly anisotropic in its crystallographic properties. In polycrystals, internal stresses are set up between adjacent grains that maintain coherence, causing constraints in their shape change that neutron irradiation and its inducement of fission would produce.122 Although this has proved to be an extreme case, the same phenomenon has been established in other materials with anisotropic crystal structures. Moreover, radiation is not a prerequisite, since temperature fluctuations can produce similar effects. The scale of this phenomenon is greatest when the applied stress is small compared to the internal stress and, in some situations, the resultant creep rate can exceed the rate without internal stress by many times.

The extent of reversibility of small creep strains (∼10⁻⁴ essentially in the primary creep range) is a subject that links closely with anelastic behaviour. Sensitive measurements have shown123 that, in this range, the primary creep rate is enhanced by prior introduction of dislocations, so this form of creep has a faster rate in materials that are initially work hardened. However, strains of this magnitude are almost completely reversed, in a time dependent manner, when stress is removed.124 This area has links with the broader field of creep transients after a stress change.125 A much larger body of work is recorded at larger strains where reversibility of stress at high temperatures leads to a consideration of creep and fatigue interactions.126,127 Industrially, this is often of critical importance with creep processes generally dominating at low frequencies, with the combination of reversible and irreversible mechanisms of greater significance as the frequency is raised.

Increasingly, studies of creep have been concerned with the simultaneous exploration of microstructure and its progressive change. Collectively, these have led to the identification of ‘creep damage’, its character and development. Numerous papers128^–135 have examined these features and they have necessarily invoked extensive understanding of many areas of metallurgy. With the passage of time, investigations on creep have intertwined with other phenomena. The influence of fluctuating stresses, resistance to high temperature environmental attack, the effect of bubbles and, more particularly of trace elements79 have emphasised these aspects. Not surprisingly, these themes have rarely led to simple quantification but valuable data have nevertheless emerged.

The early 1990s saw renewed interest in the possibility of progressing beyond well tried superalloys into the field of intermetallics, metal matrix composites and even of monolithic ceramics. Much valuable information is reported136^–145 on these newer materials, often with a realisation of the need to ameliorate the low temperature brittleness that could continue to present overriding difficulties for widespread practical applications.

The need for adequate creep resistance at modest temperatures in light alloys is recognised and explored in many papers.146^–153 Such requirements have also brought in the effects of multi-axial stresses154 and stimulated the devising of novel testing techniques.155,156 The development of methods for extrapolation to predict behaviour over longer times has remained a prominent feature.157^–159 This has necessitated continuing efforts to describe mathematically the shape of creep curves. The underpinning science has not been overlooked. The identification of the character of crystal defects and their movement is central to this and ‘deformation mechanism maps’160 have provided a suitable means to represent the areas over which a particular deformation mechanism is expected to dominate. This approach, however, is restricted to a consideration of constant secondary creep rates. Mechanical properties of superalloys have received broader attention. Milestones have been reviewed161 in the progressive development of these alloys that have proved so outstandingly successful in practical applications in addition to their value in gas turbine engines.162 Studies have provided additional information163,164 on the influence of oxidising atmospheres on their mechanical behaviour. The complexities of the processes occurring and the many variables influencing them163^–173 has led to the increasing realisation that modelling offers the most promise as a means of representing and analysing data.173,174 Such an approach remains a stimulus for extensive experimental programmes.

Creep data steadily continue to accumulate, together with insight and understanding of underlying atomic processes.175^–182 Much of this progression is presented or reflected in a great variety of papers in mst: space, not content or quality, has prevented reference to many of them. This journey is not completed. The field remains wide, with realisation that simple answers are rarely forthcoming. Practical applications require the tailoring of materials’ properties and usually these include much more than an enhancement of resistance to creep. Consistency of behaviour must be assured and this requires careful reproducibility of fabrication procedures. Properties at low temperatures cannot be ignored. Notably, adequate fracture toughness is a pre-requisite for any material with enhanced creep resistance. The number of variables in alloy design and parameters governing operational use negate the expectancy of simple recipes. Modelling procedures have made much progress, especially in predictive capabilities. Microstructural characterisation and control have equally made large strides forward. Many of these aspects, relating to nickel based alloys, have been incorporated in papers183^–203 collected in a recent issue of mst that provides a comprehensive appraisal of the present situation. The elimination of grain boundaries and exploitation of microstructural control to increase creep strength in single crystals in critical components such as turbine blades have been hugely beneficial. This work has also emphasised the need to gather further data and to improve understanding of anisotropic features, inseparably linking microstructure and component design.

G. W. Greenwood

Department of Materials Science and Engineering, University of Sheffield

[email protected]

This is the latest in an occasional series of editorials reviewing the digitised archives of mst and Metal Science. Previous contributions appeared in the January and April 2010 and November 2011 issues.