MECHANISMS OF CREEP DEFORMATION IN AN ULTRAFINE- GRAINED ALUMINIUM PROCESSED BY ECAP METHOD

MECHANISMS OF CREEP DEFORMATION IN AN ULTRAFINE- GRAINED ALUMINIUM PROCESSED BY ECAP METHOD Jirí Dvorák a, Václav Sklenicka a, Petr Král a a Ústav fyziky materiálu Akademie ved Ceské republiky, Žižkova 22, 616 62 Brno, CR, E-mail: dvorak @ipm.cz Abstract Extremely coarse-grained aluminium (99,99%) was subjected to severe plastic deformation (ECAP) at room temperature using the die that had a 90 angle between the die channels. Using multiple passes (up to 12) and the application of different processing routes (A, Bc and C) an ultrafine-grained microstructure with the grain size ~ 1 m was attained in the whole volume of material. Creep tensile and tests were conducted on such material at an applied stress range of 15 25 MPa and at temperatures 423 523 K. For comparison reasons, parallel creep tests were also conducted on coarse-grained aluminium. The results indicate an increased creep resistant of ultrafine-grained aluminium compared to its coarsegrained state. For description of controlling creep mechanisms an activation energy of creep and stress exponent creep rate were determined. It was proposed, that creep of ultrafinegrained aluminium is probably control by the same creep mechanisms as coarse-grained one. However, the role of grain boundary sliding in creep considerably differs in both ECAP and coarse grained materials. 1. INTRODUCTION The mechanisms controlling the creep properties of pure metals have usually been identified from the dependence of the minimum and/or steady-state creep rate m on stress, temperature T and grain size d, using power-law expressions of the form n p A (1/ d) exp( Q / RT), (1) m c where A, n, p and R are constants and Q c is the activation energy for creep. With this approach, the fact that n, p and Q c are themselves functions of stress, temperature and grain size is conventionally explained by assuming that different mechanisms, each associated with different values of n, p and Q c control the creep characteristics exhibited in different stress/temperature regimes. In turn, the dominant mechanisms under specific test conditions is then generally decided by comparing experimentally determined values of n, p and Q c with the values predicted theoretically for different creep processes. Despite the widespread adoption of power law approaches over many decades, little agreement has been reached on the detailed mechanisms governing creep behaviour at elevated and/or high temperatures. Thus, pure metals are frequently considered to exhibit regimes with n 1 at low stresses and n 4 at high stresses, with n increasing rapidly with increasing stress in the power-law break down range. When n 4, creep is known to occur by diffusion-controlled generation and movement of lattice dislocations, but the precise mechanisms involved remain a matter of discussion. Controversy also continues over whether creep in the n 1 regime takes place by diffusional mechanisms which do not require dislocation movement (i.e. Nabarro-Herring or Coble creep 1,2) or by dislocation processes (often referred to as Harper-Dorn creep 2). One reason suggested to account for the uncertainly over creep mechanisms stems from the traditional assumption that the parameter of greatest significance is the secondary or steady- 1

state creep rate, ignoring the primary and tertiary stages. However, with most metals and alloys displaying standard creep curves, a genuine steady-state condition is rarely achieved. Instead, a minimum rate is commonly observed when the decaying primary creep rate is offset by the acceleration in rate associated with the damage processes which cause the tertiary stage. Thus, in most cases the secondary creep rate merely seems to be constant over a period which depends on the precision of the measured creep strain/time record. Many investigations concerned with creep mechanism identification have been undertaken using coarse grained pure aluminium 3. This choice is easily justified because aluminium is readily available, has a convenient melting point (T m = 933K) and possesses excellent oxidation resistance. However, even when compared with other face-centred-cubic metals, the creep properties of pure aluminium could be considered untypical. In particular, at 0.5T m (i) aluminium exhibits very large creep strains without the occurrence of necking and, (ii) grain boundary cavities do not form to cause intergranular creep fracture. However, it is logical to expect that mechanism of hardening/softening observed in ultrafine grained aluminium may be fundamentally different from that observed in the coarse grained one. Consequently it can not be excluded that creep in the ultrafine grained aluminium is (are) controlled by different creep mechanism(s) than the coarse grained material. Previously 4,5 we studied the creep behaviour of ultrafine-grained (UFG) aluminium prepared by equal-channel angular pressing (ECAP) 6 using the different processing routes. The present paper is devoted to the investigation of the acting creep deformation mechanisms to obtain a more definitive understanding of the creep processes in UFG materials. 2. EXPERIMENTAL MATERIAL AND PROCEDURES The material used in this investigation was an extremely coarse-grained (grain size 5 mm) high purity (99.99%) aluminium supplied in the form of rods. The rods were cut into short billets having a length of 60 mm and a cross-section 10 x 10 mm 2. ECAP was conducted at room temperature with a die that had an internal angle of 90 between the two parts of the channel and an outer arc of curvature of 20, where these two parts intersect. It can be shown from first principles that these angles lead to an imposed strain of 1 in each passage of the sample. The ECAP die was earlier described in detail by Dvorák et al 4. The pressing speed was 10 mm min -1. The subsequent extrusion steps were performed up to twelve passes to achieve high strains. Three different routes were applied in performing these steps to vary the overall shearing conditions by rotation of the sample between subsequent steps. In route A the specimen was removed from the die and the pressing was repeated without any sample rotation. In route B (B B C 4), the rotation was always 90 in the same sense. Finally, in route C the rotation was always 180. Creep tests were performed at a temperature interval from 423 to 523K both in tension and in under constant applied stress ranging between 10 and 25 MPa. The flat creep specimens were cut out of ECAP billets. While almost all the tensile specimens were run to final fracture creep tests were run up to a true strain of about 0.35 only. For comparison purposes, the same creep tests were performed also on coarse-grained aluminium. True strain-time readings were continuously recorded by the PC-based data acquisition system. Following ECAP and creep testing, samples were prepared for examination by means of scanning and transmission electron microscopy (Philips CM 12 TEM/STEM). All grain size measurements were made by the linear intercept method using a transmission electron microscope. The grain size is the mean intercept of the boundaries irrespective of their nature.the amount of grain boundary sliding (GBS) was determined by measuring the surface offsets produced at the intersections of grain boundaries with marker lines transversal to the stress axis. Displacement of the marker lines, u, due to GBS, together with the fraction of boundaries, s, with observable GBS, were measured 7. 2

3. RESULTS 3.1 Creep behaviour Representative creep curves are shown in Figs. 1 and 2. All of these plots were obtained at an absolute temperature of 473 K ( 0.5 T m ) and at an applied tensile or stress of 15 MPa. The creep tests in tension were running up to final fracture of creep specimens, whereas the creep tests in were interrupted at a true strain of about 0.35. The creep testing was conducted on billets after different number of ECAP passes (up to 12 passes) followed route B(BB c ) and, for comparison purposes, on (coarse grained) material. CREEP STRAIN RATE d/dt [s -1 ] 1 10-1 10-2 0.8 10-3 0.6 0.4 0.2 Al 99.99 Al 99.99 473 K, 473 15 MPa K, 15 MPa state tension tension state 0 0 50 1000 1050 1100 10-8 TIME t [h] 10-1 10-2 10-3 10-8 Al 99.99 473 K, 15 MPa tension 0 0.2 0.4 0.6 0.8 1 STRAIN state Fig. 1. Creep curves and creep rate versus time or strain for state and various number of ECAP passes via route B (tension creep up to fracture). 3

STRAIN 0.3 0.2 473 K, 15 MPa state 0.1 10-8 10-9 10-8 10-9 0 0 50 400 800 1200 TIME state t [h] A1 Al 99.99 Al 99.99 473 K, 15 MPa 473 K, 15 MPa 0 50 400 800 1200 TIME t [h] state 0 0.1 0.2 0.3 0.4 STRAIN Figs. 1a and 2a show standard vs t curves for aluminium and for the ECAP material for creep tests both in tension or. These standard vs t creep curves can be easily replotted in the form of the instantaneous strain rate d/dt versus time (as shown in Fig. 1b and 2b) and/or in the form of the instantaneous strain rate d/dt versus strain (Fig. 1c and 2c). As demonstrated by figures, significant differences were found in the creep behaviour of the ECAP material when compared to its coarse-grained counterpart. First, the ECAP materials exhibits markedly longer creep life (Figs 1a and 1b) or markedly longer 4 Fig. 2. Creep curves and creep rate versus time or strain for state and various number of ECAP passes via route B ( creep up to strain 0.35). duration of creep exposure to obtain a strain of 0.35 (Fig. 2a and 2b) than coarse grained aluminium. Second, the minimum creep rate for the ECAP material is about one to two orders of magnitude less than that of coarse-grained one. Third, the shapes of creep curves for the

ECAP material after a different number of pressing differ considerably. However, this difference in the shapes of the ECAP creep curves is more clearly illustrated for the tests conducted at small number of the ECAP passes. 3.2 Grain boundary sliding (GBS) Grain boundary sliding (GBS) may be an important mechanism of creep deformation in ultrafine grained materials at elevated temperatures. The process of GBS refers to the relative displacement of two adjacent grains in a polycrystalline matrix with the displacement taking place at, or in a zone immediately adjacent to, their common boundary. At present, however, little information exists regarding the effect of GBS on mechanical and/or creep properties of ultrafine grained materials. GBS were measured on the surfaces of the crept specimens; the creep tests in tension were terminated at similar predetermined strains ( 0.15) to permit quantitative measurements of GBS. SEM made it possible to detect GBS characterized by u 0.1 m. However, GBS does not place on all grain boundaries; that is why the relative frequency of sliding boundaries s (the fraction of boundaries with observable GBS) was determined. This fraction s equals N s /N, where N s in the number of boundaries on which detectable GBS take place and N is the total number of boundaries examined. Then the strain component gb due to GBS is expressed 7: gb = (1 + ) u. s / L, (2) where the mean grain size L is determined by the linear intercept method and the contribution of GBS to the total strain = gb /. The results of GBS measurements are summarized in Table I. Table 1. Summary of GBS measurements Specimen (Route-No of Passes).10 2 s.10 2 u[ m] L m gb.10 2 gb/.10 2 A1 = 16.70 0.782 0.393 5.8 6.184 37.03 12.86 0.744 0.387 9.0 3.610 28.07 15.05 0.957 0.788 8.9 9.786 63.14 19.22 0.982 0.603 5.7 12.385 64.44 15.60 0.945 0.622 6.5 10.454 67.01 It is evident that the fraction of boundaries s with observable GBS and the mean displacement u due to GBS increase as a number of ECAP passes increases. These results strongly support the idea that GBS is closely connected with microstructural changes of grain boundaries [6]. 3.1.3 Stress dependences of creep rate and lifetime 5

The difference in the minimum creep rate for the ECAP material and state consistently decreases with increasing number of ECAP passes (Figs. 1 and 2). Additional 10 difference 10 7-4 is illustrated by Fig. 3 showing the variation of the minimum creep rate with applied stress for ECAP specimens Al Al 99.99 99.99 processed by route B after 8 passes. The results demonstrate that at high stresses 473 473 the K Kminimum creep rate of ECAP material may be up to one 10 order 6 of magnitude lower than that of the material, although this difference decreases with decreasing applied stress so that, at 10 MPa is negligible. Furthermore, for both and ECAP material, the double logarithmic plots of the creep exposure time t, as 10 a function 5 of applied stress state are shown state - in Fig. 4. Again, the difference in the exposure time - increases with increasing applied stress. In fact, inspection of Fig. 4 reveals that at stress of 10 B MPa 10the - 6 ECAP route: B creep exposure 8 time 8 passes of the ECAP material seem to be essentially equal to that of the 10 4 state. tension Fig. 5 shows the dependence of the creep exposure time t (and/or the time to fracture t f for tensile specimens) on the creep rate d/dt. Experimentally determined values of t and/or t f correlate 10 10 3-7 well with the creep rate d/dt through the Monkman-Grant relationship, (d/dt) a 10 10 100 100.t const 7, where the exponent a is found to be about 0.84 (Fig.5). Thus creep exposure time t CREEP EXPOSURE TIME t [s] STRESS [MPa] Fig. 3. Stress dependences of creeprate for state ( creep only) and ECAP material after 8 passes via route B (both tension and creep). Fig. 4. Stress dependences of creep exposure time (the time to fracture for tension creep and the time since to achieve a strain 0.35 for creep) for state and ECAP material after 8 passes via route B. 6

CREEP EXPOSURE TIME t [s] METAL 2004 10 6 10 5 10 4 state - ECAP route: B 8 passes tension Al 99.99 473 K 10 3 Fig. 5. Dependences of creep exposure time on creep rate for state and ECAP material after 8 passes via route B. is approximately inversely proportional to d/dt. A possible explanation for this result may lie in the same creep mechanisms controlling both creep information and damage processes. 3.1.4 Temperature dependence of creep rate The minimum creep rate was measured at various temperatures in the interval from 423 to 523 K and at two constant tensile applied stresses 15 MPa and 20 MPa, respectively. The temperature dependences of minimum creep rate are shown in Fig. 6. TEMPERATURE T [K] 10-3 523 498 473 448 Al 99.99 ECAP route: B 8 passes 423 Q C =129,716 kj/mol 15 MPa 20 MPa Q C =110,98,7 kj/mol 19 20 21 22 23 24 10 4 /T [K -1 ] 7

Fig. 6. Temperature dependences of creep rate for different stresses and ECAP samples (8 passes via route B). Tension creep. The activation energy of creep Q c is defined as Q c ln min ( 1/ kt). (3) Thus, the activation energy Q c can be obtained as a k multiple of the slope of log m vs 1/T plots shown in Fig. 6. By the least square method a value of the apparent activation energy Q c was determined. The Q c is stress dependent and equal to 129,7 16 and 110,9 9 kj/mol for stress 20 MPa and 15 MPa, respectively. 4. DISCUSSION According to eq. (1) the stress dependence of the minimum creep rate (Fig. 3) at constant temperature and at constant grain size can be described by the power-law relationship of the n form: A, where A is independent of, while n = ( ln / ln ) T is the stress exponent of the creep rate. The observed values of n are 5.5 and 6.5 for ECAP samples and material, respectively. Similar values of n was reported by Tobolová and Cadek 8 for creep in polycrystalline aluminium with a standard grain size. As it was mentioned earlier, for n 4 creep is known to occur by diffusion-controlled generation and movement by dislocations within grains and by grain boundary sliding. The rate of grain boundary sliding * n* can be described by a similar way: gb A, where the stress exponent of grain boundary sliding n * 0.8n 7. Thus, the lower values of n for ECAP material may reflect the effect of more intensive grain boundary sliding in creep of ultrafine grained material (see Tab.1) in comparison with its coarse-grained counterpart. A mechanisms which most probably plays the dominating role in recovery in creep of aluminium is dislocation climb controlled by lattice self-diffussion. Therefore the activation energy for creep Q c should be the same as the value of the activation enthalpy of selfdiffusion H SD in aluminium ( H SD 127 143 kj/mol 8). The slightly higher experimental value of the activation energy for creep Q c 151 kj/mol was found by Tobolová and Cadek [ 8 ] for polycrystalline aluminium of 99.99 % purity with a mean grain size ~ 400 m. The values obtained for Q c (Fig. 6) in this work are somewhat lower than that of H SD. Supposing that grain boundary sliding is controlled by grain boundary diffusion (the activation enthalpy of grain boundary diffusion H gb H SD ) we have a further support to interprete the creep mechanism in ultrafine grained aluminium as a synergistic action of intragranular strain and grain boundary sliding. Further, the amount of grain boundary sliding to the total strain increases at the same temperature with a decrease in the applied stress 7. This can explains the lower value of Q c determined for = 15 MPa in comparison with the Q c ( = 20 MPa). 5. CONCLUSIONS The creep resistance of ultrafine grained (UFG) high purity aluminium is shown to be considerably improved compared to an material. The activation analysis of the 8

creep data has led to the conclusion, that creep in UFG aluminium is probably controlled by the same mechanisms as coarse-grained one. However, lower values of the stress exponent of the creep rate n and the apparent activation energy of creep Q c found for UFG aluminium in comparison to coarse-grained counterpart indicate an important role of grain boundary sliding in creep of UFG materials. ACKNOWLEDGEMENTS Financial support for this work was provided by the Grant Agency of the Academy of Sciences of the Czech Republic under Grant A204 1301. REFERENCES [1] CADEK, J. Creep in metallic materials. Amsterdam: Elsevier Science Publishers. [2] KLOC, L., FIALA, J., CADEK, J. Diffusional creep and Harper-Dorn creep at intermediate temperatures. In Creep behaviour of advanced materials for the 21 st Century. Warrendale: TMS, 1999, pp. 471-480. [3] BRIANT, C.L., DAVIDSON, P.L. Creep in polycrystalline aluminium. Materials Sci. Technology. 2000, 16, pp. 1102-1111. [4] DVORÁK, J., SKLENICKA, V., SVOBODA, M. Creepové chování ultrajemnozrnného hliníku. In METAL 2003: 12 Int. Conference: 20.-22.5.2003, Cervený zámek, Hradec nad Opavicí, Czech Republic CD-ROM. Ostrava: Tanger: ISBN 80-85988-82-8 In Czech. [5] SKLENICKA, V., DVORÁK, J., SVOBODA, M. Influence of processing route on creep of ultrafine grained aluminium prepared by ECAP. In: Ultrafine Grained Materials III, Warrendale, USA: TMS, 2004, in print. [6] VALIEV, R.Z., ISLAMGALIEV, R.K., ALEXANDROV, I.V. Bulk nanostructured materials from severe plastic deformation. Properties in Mater. Science, 2000, 45, pp. 103-189. [7] SKLENICKA, V., SAXL I., CADEK, J. Mezikrystalový lom pri vysokoteplotním creepu kovu a slitin. Praha: Academia, Studie CSAV c. 8, 1977 (In Czech). [8] TOBOLOVÁ, Z., CADEK, J. An interpretation of steady state creep. Phil. Magazine, 1972, 26, pp. 1419-1428. 9