MATERIALS SEIENEE ENGINEERING A ELSEVIER Materials Science and Engineering A260(1999)275-283 Microstructural evolution over a large strain range in aluminium deformed by cyclic-extrusion-compression M.Richert a,Q.Liu b.*,N.Hansenb Department of Structure and Mechanics of Solids,University of Mining and Metallurgy,PL-059,Krakow,Poland Materials Research Department,Riso National Laboratory.DK-4000.Roskilde,Denmark Received 20 April 1998:received in revised form 28 July 1998 Abstract Polycrystalline pure aluminium (99.99%)has been deformed at room temperature by the Cyclic-Extrusion-Compression (CEC)-method to strains in the range 0.9-60(1-67 cycles).At different strains,the microstructure and local crystallography have been characterised in particular by transmission electron microscopy.It has been found that the microstructure develops from a cell block structure into an almost equiaxed structure of cells and subgrains,that the spacing between the boundaries subdividing the structure is almost unaffected by the strain and that the misorientation across these boundaries increases with the strain over the whole strain range.At the largest strain,the average misorientation across the deformation induced boundaries is ~25.The flow stress in compression is measured after the cyclic deformation and it is found that the flow stress increases with strain towards a saturation level which is reached at a relatively low strain.The discussion comprises the effect of deformation mode and plastic strain over a large strain range on the microstructural evolution and mechanical behaviour of aluminium.1999 Elsevier Science S.A.All rights reserved. Keywords:Cyclic-extrusion-compression;Aluminium;Microstructure;Plasticity;Large-strain deformation 1.Introduction X-rays,the flow stress (0.2%offset)has been measured in compression and the microhardness has been mea- The microstructural and mechanical response to sured on the deformed specimens.Some of these results large strain deformation has been studied over many will be recapitulated in this paper together with new years [1-7].This refects that large strain behaviour is results mainly describing the effect of strain on mi- of both fundamental and technological importance [6]. crostructures and local crystallography,especially the Most of these studies have been carried out on speci- angle of misorientation across dislocation boundaries mens deformed in monotonic loading over a strain and grain boundaries.The principal experimental tech- range extending up to 6-7 [4].It is of interest to extend nique has been TEM including Kikuchi pattern analy- this strain range by introducing alternative deformation sis.The material investigated has been polycrystalline modes.This has been done in the present work concen- pure aluminium (99.99%).The specimens have been trating on specimens deformed by the Cyclic-Extru- deformed at room temperature over a very large strain sion-Compression(CEC)-method [5]. range of :=0.9-60,thereby allowing a comparison The CEC-method was invented to allow arbitrarily with previous experiments [5,8,9]. large strain deformation of a sample with the preserva- tion of the original sample shape and the deformed specimens to be examined extensively [5,8,9].For exam- 2.Experimental procedures ple,the microstructure evolution has been studied by optical metallography and Transmission Electron Mi- 2.1.Deformation method croscopy (TEM),the texture has been measured by *Corresponding author.Tel:+45-4677-5808;fax:+45-4677- The CEC-method is a combination of extrusion and 5758;e-mail:qing.liu@risoe.dk. compression which alternate (Fig.1).The sample is 0921-5093/99/S-see front matter 1999 Elsevier Science S.A.All rights reserved. P:S0921-5093(98)00988-5
Materials Science and Engineering A260 (1999) 275–283 Microstructural evolution over a large strain range in aluminium deformed by cyclic-extrusion–compression M. Richert a , Q. Liu b,*, N. Hansen b a Department of Structure and Mechanics of Solids, Uni6ersity of Mining and Metallurgy, PL-059, Krakow, Poland b Materials Research Department, Risø National Laboratory, DK-4000, Roskilde, Denmark Received 20 April 1998; received in revised form 28 July 1998 Abstract Polycrystalline pure aluminium (99.99%) has been deformed at room temperature by the Cyclic-Extrusion–Compression (CEC)-method to strains in the range 0.9–60 (1–67 cycles). At different strains, the microstructure and local crystallography have been characterised in particular by transmission electron microscopy. It has been found that the microstructure develops from a cell block structure into an almost equiaxed structure of cells and subgrains, that the spacing between the boundaries subdividing the structure is almost unaffected by the strain and that the misorientation across these boundaries increases with the strain over the whole strain range. At the largest strain, the average misorientation across the deformation induced boundaries is 25°. The flow stress in compression is measured after the cyclic deformation and it is found that the flow stress increases with strain towards a saturation level which is reached at a relatively low strain. The discussion comprises the effect of deformation mode and plastic strain over a large strain range on the microstructural evolution and mechanical behaviour of aluminium. © 1999 Elsevier Science S.A. All rights reserved. Keywords: Cyclic-extrusion–compression; Aluminium; Microstructure; Plasticity; Large-strain deformation 1. Introduction The microstructural and mechanical response to large strain deformation has been studied over many years [1–7]. This reflects that large strain behaviour is of both fundamental and technological importance [6]. Most of these studies have been carried out on specimens deformed in monotonic loading over a strain range extending up to 6–7 [4]. It is of interest to extend this strain range by introducing alternative deformation modes. This has been done in the present work concentrating on specimens deformed by the Cyclic-Extrusion–Compression (CEC)-method [5]. The CEC-method was invented to allow arbitrarily large strain deformation of a sample with the preservation of the original sample shape and the deformed specimens to be examined extensively [5,8,9]. For example, the microstructure evolution has been studied by optical metallography and Transmission Electron Microscopy (TEM), the texture has been measured by X-rays, the flow stress (0.2% offset) has been measured in compression and the microhardness has been measured on the deformed specimens. Some of these results will be recapitulated in this paper together with new results mainly describing the effect of strain on microstructures and local crystallography, especially the angle of misorientation across dislocation boundaries and grain boundaries. The principal experimental technique has been TEM including Kikuchi pattern analysis. The material investigated has been polycrystalline pure aluminium (99.99%). The specimens have been deformed at room temperature over a very large strain range of o=0.9–60, thereby allowing a comparison with previous experiments [5,8,9]. 2. Experimental procedures 2.1. Deformation method The CEC-method is a combination of extrusion and compression which alternate (Fig. 1). The sample is * Corresponding author. Tel.: +45-4677-5808; fax: +45-4677- 5758; e-mail: qing.liu@risoe.dk. 0921-5093/99/$ - see front matter © 1999 Elsevier Science S.A. All rights reserved. PII: S0921-5093(98)00988-5
276 M.Richert et al.Materials Science and Engineering A260 (1999)275-283 placed in a sectioned die marked(1)on Fig.I consist- =4n In- ing of an upper and lower chamber of equal diameters d..The chambers are connected by a channel of diame- ter d.The two halves of the sectioned die (1)are where,d,is the chamber diameter,dm the channel connected by the screws (4)and (5)and are fixed to the diameter and,n is the number of deformation cycles frame(6)by bolts (7)and (8).The frame (6)is con- [5].The correct conditions of deformation during the nected via the cantilever (9)to the stationary beam of cyclic deformation is controlled by continuous registra- the testing machine,whereas the inner frame (10)with tion of the tension and compression forces.In the the thrust screws (11)and (12)is connected via the present experiment d,and d are 10 and 8 mm,respec- cantilever (13)to the cross-head of the machine.The tively.This gives a strain of s=0.9 per cycle. deformation proceeds by a cyclic flow of the metal from one chamber to the other.For example in a particular 2.2.Sample preparation and experiment cycle,the sample is extruded from the upper to the lower chamber by the upper punch(2).Here,under the High purity (99.99%)samples (28 mm long and 10 action of the lower punch(3)compression occurs simul- mm in diameter)with a grain size of ~1.7 mm have taneously with the extrusion.so that the sample is been deformed in the strain range 0.9-60,correspond- restored to its initial shape.Before the beginning of the ing to a range of cycles from I to 67.The cross-head deformation process,the sample is compressed by the speed has been 0.17 mm s-.The temperature during thrust screws (11)and (12).In this way,the sample is deformation has been measured by a thermocouple stressed in hydrostatic compression permitting arbitrar- positioned close to the channel between the upper and ily high deformations without crack development.The the lower dies.The duration of the individual cycles magnitude of the cumulated true strain is approximat- (~100 s)was chosen so that the temperature did not ely: exceed room temperature by more than ~10C during the deformation process [8].On the deformed sample, the flow stress (0.2%offset)in compression was mea- 9 sured on cylindrical specimens with a diameter of 8 mm and a height of 8 mm.The cross-head speed used was 1 0.017 mm s-and the loading direction was the same 6 as the extrusion direction.Longitudinal sections for microscopy and microhardness (uHV)testing were ob- 2 tained by bisecting the deformed specimen by spark machining.For optical microscopy the specimens were 7 electrolytically polished in a 20%HCIO+C2HsOH reagent and then etched in the Keller reagent.Thin foils 1 for electron microscopy were prepared from the longi- tudinal section. 4 Local misorientations were measured by the Kikuchi 5 pattern techniques in a TEM [10].Texture measure- ments have previously been carried out by X-rays and 10 the results are reported and discussed in detail in [11]. Generally,it is observed that the texture intensity does 3 not appear to increase as the strain is increased in the large strain regime.However,the texture spread in- 8 creases with the strain,but this effect is not large [11]. 2 3 3.Results 3.1.Flow stress and hardness The flow stress (0.2%offset)is shown in Fig.2 as a function of the strain.After a strain of 0.9,the flow stress is ~65 MPa and increases to a saturation value of ~77 MPa which is reached in the strain range Fig.1.Equipment for deformation of metals by the Cyclic-Extru- e=5-8. sion-Compression (CEC)-method [5]. The microhardness as a function of the strain is
276 M. Richert et al. / Materials Science and Engineering A260 (1999) 275–283 placed in a sectioned die marked (1) on Fig. 1 consisting of an upper and lower chamber of equal diameters do. The chambers are connected by a channel of diameter dm. The two halves of the sectioned die (1) are connected by the screws (4) and (5) and are fixed to the frame (6) by bolts (7) and (8). The frame (6) is connected via the cantilever (9) to the stationary beam of the testing machine, whereas the inner frame (10) with the thrust screws (11) and (12) is connected via the cantilever (13) to the cross-head of the machine. The deformation proceeds by a cyclic flow of the metal from one chamber to the other. For example in a particular cycle, the sample is extruded from the upper to the lower chamber by the upper punch (2). Here, under the action of the lower punch (3) compression occurs simultaneously with the extrusion, so that the sample is restored to its initial shape. Before the beginning of the deformation process, the sample is compressed by the thrust screws (11) and (12). In this way, the sample is stressed in hydrostatic compression permitting arbitrarily high deformations without crack development. The magnitude of the cumulated true strain is approximately: o = 4n lndo dm , where, do, is the chamber diameter, dm, the channel diameter and, n is the number of deformation cycles [5].The correct conditions of deformation during the cyclic deformation is controlled by continuous registration of the tension and compression forces. In the present experiment do and dm are 10 and 8 mm, respectively. This gives a strain of o=0.9 per cycle. 2.2. Sample preparation and experiment High purity (99.99%) samples (28 mm long and 10 mm in diameter) with a grain size of 1.7 mm have been deformed in the strain range 0.9–60, corresponding to a range of cycles from 1 to 67. The cross-head speed has been 0.17 mm s−1 . The temperature during deformation has been measured by a thermocouple positioned close to the channel between the upper and the lower dies. The duration of the individual cycles (100 s) was chosen so that the temperature did not exceed room temperature by more than 10°C during the deformation process [8]. On the deformed sample, the flow stress (0.2% offset) in compression was measured on cylindrical specimens with a diameter of 8 mm and a height of 8 mm. The cross-head speed used was 0.017 mm s−1 and the loading direction was the same as the extrusion direction. Longitudinal sections for microscopy and microhardness (mHV) testing were obtained by bisecting the deformed specimen by spark machining. For optical microscopy the specimens were electrolytically polished in a 20% HClO4+C2H5OH reagent and then etched in the Keller reagent. Thin foils for electron microscopy were prepared from the longitudinal section. Local misorientations were measured by the Kikuchi pattern techniques in a TEM [10]. Texture measurements have previously been carried out by X-rays and the results are reported and discussed in detail in [11]. Generally, it is observed that the texture intensity does not appear to increase as the strain is increased in the large strain regime. However, the texture spread increases with the strain, but this effect is not large [11]. 3. Results 3.1. Flow stress and hardness The flow stress (0.2% offset) is shown in Fig. 2 as a function of the strain. After a strain of 0.9, the flow stress is 65 MPa and increases to a saturation value of 77 MPa which is reached in the strain range o=5–8. The microhardness as a function of the strain is Fig. 1. Equipment for deformation of metals by the Cyclic-Extrusion–Compression (CEC)-method [5].
M.Richert et al.Materials Science and Engineering A260 (1999)275-283 277 90 dent of strain,thus the hardness evolution with strain is 80 quite similar before and after storage.The hardness measurement indicates that some recovery has taken 70 place during storage.As this recovery appears to be independent of the strain it is believed that the mi- 60 crostructural examination of the stored specimens will give a reliable description of the microstructural MO changes caused by the deformation process itself.This 0 assumption has been examined by comparing early microscopic work from the group which produced the 30 specimens.The present work shows that the structure of the specimen stored for 9 years after deformation is 0 4 681012141618 very similar to the structure of the newly deformed TRUE STRAIN specimens observed in [8]with respect to both mor- phology and the measured dislocation boundary spac- Fig.2.Flow stress as a function of the accumulated strain.This curve also includes two data points obtained at strains lower than 0.9. ing These specimens have been deformed half a cycle and one cycle with a ratio dod=10/9 which gives strains 0.2 and 0.4,respectively. 3.2.Macrostructure Fig.4 shows the macrostructure observed in longitu- shown in Fig.3.The hardness changes correspond to dinal sections of samples deformed at strains in the the changes in flow stress.At large strain a slight range 3.6-30.3.It is apparent that the original equiaxed decrease in hardness appears to take place as the strain grain structure is partly maintained even at the largest increases in the range s=4-20.Above this strain,the strain examined(Fig.4(d)).The most typical feature is hardness is unaffected by the strain up to 60,which is a banded structure indicating localized deformation in the maximum strain examined.A comparison of Figs.2 slip bands and shear bands.Fig.4(a)shows short bands and 3 shows that the hardness is proportional to the on the grain scale which have a direction with respect flow stress with a proportional factor of 0.3. to the loading direction which varies from grain to The specimens for electron microscopy have been grain.At the larger strains(Fig.4(c))intersecting bands examined ~9 years after the deformation took place. appear with a length corresponding to the specimen The specimens have been stored at room temperature dimension.These bands have an angle of ~65 to the and the hardness measurements have been repeated to sample axis.Based on their length and their macro- detect if changes have occurred during the long storage. scopic orientation these bands have been identified as This check was necessary as equipment was not avail- macroscopic shear bands [5,8,9].A comparison of Fig. able for repetition of the deformation process.The 4(b)and (c)shows that the number of shear bands hardness data shows that the hardness has decreased by increases when the strain is increased from 4.5 to 22.5. 10-20 hardness units.This decrease is larger than However,a comparison of Fig.4(c)and(d)indicated can be explained by the uncertainty of the measure- that the tendency to form shear bands is reduced as the ments.The hardness decrease appears to be indepen- strain is further increased (in the present case to 30.6). 350 300 250 200 10 100 TRUE STRAIN Fig.3.Microhardness as a function of the accumulated strain
M. Richert et al. / Materials Science and Engineering A260 (1999) 275–283 277 Fig. 2. Flow stress as a function of the accumulated strain. This curve also includes two data points obtained at strains lower than 0.9. These specimens have been deformed half a cycle and one cycle with a ratio do/dm=10/9 which gives strains 0.2 and 0.4, respectively. dent of strain, thus the hardness evolution with strain is quite similar before and after storage. The hardness measurement indicates that some recovery has taken place during storage. As this recovery appears to be independent of the strain it is believed that the microstructural examination of the stored specimens will give a reliable description of the microstructural changes caused by the deformation process itself. This assumption has been examined by comparing early microscopic work from the group which produced the specimens. The present work shows that the structure of the specimen stored for 9 years after deformation is very similar to the structure of the newly deformed specimens observed in [8] with respect to both morphology and the measured dislocation boundary spacing. 3.2. Macrostructure Fig. 4 shows the macrostructure observed in longitudinal sections of samples deformed at strains in the range 3.6–30.3. It is apparent that the original equiaxed grain structure is partly maintained even at the largest strain examined (Fig. 4(d)). The most typical feature is a banded structure indicating localized deformation in slip bands and shear bands. Fig. 4(a) shows short bands on the grain scale which have a direction with respect to the loading direction which varies from grain to grain. At the larger strains (Fig. 4(c)) intersecting bands appear with a length corresponding to the specimen dimension. These bands have an angle of 65° to the sample axis. Based on their length and their macroscopic orientation these bands have been identified as macroscopic shear bands [5,8,9]. A comparison of Fig. 4(b) and (c) shows that the number of shear bands increases when the strain is increased from 4.5 to 22.5. However, a comparison of Fig. 4(c) and (d) indicated that the tendency to form shear bands is reduced as the strain is further increased (in the present case to 30.6). shown in Fig. 3. The hardness changes correspond to the changes in flow stress. At large strain a slight decrease in hardness appears to take place as the strain increases in the range o=4–20. Above this strain, the hardness is unaffected by the strain up to 60, which is the maximum strain examined. A comparison of Figs. 2 and 3 shows that the hardness is proportional to the flow stress with a proportional factor of 0.3. The specimens for electron microscopy have been examined 9 years after the deformation took place. The specimens have been stored at room temperature and the hardness measurements have been repeated to detect if changes have occurred during the long storage. This check was necessary as equipment was not available for repetition of the deformation process. The hardness data shows that the hardness has decreased by 10–20 hardness units. This decrease is larger than can be explained by the uncertainty of the measurements. The hardness decrease appears to be indepenFig. 3. Microhardness as a function of the accumulated strain.
278 M.Richert et al.Materials Science and Engineering A260 (1999)275-283 1mm ED Fig.4.Macrostructure in the longitudinal section at different accumulated strains:(a)3.6:(b)4.:5;(c)22.5:(d)30.6.The extrusion direction is marked ED [12]. 3.3.Microstructure (TEM) crobands (MBs)[12].The DDW/MBs bound cell blocks containing ordinary dislocation cells.Within The microstructure at low strain (s=0.9)shown in many of the cells dislocation tangles can be observed. Fig.5 is a characteristic deformation structure with The banded structure observed by TEM shows no extended Dense Dislocation Walls (DDWs)and Mi- relationship to the banded structure observed optically
278 M. Richert et al. / Materials Science and Engineering A260 (1999) 275–283 Fig. 4. Macrostructure in the longitudinal section at different accumulated strains: (a) 3.6; (b) 4.;5; (c) 22.5; (d) 30.6. The extrusion direction is marked ED [12]. 3.3. Microstructure (TEM) The microstructure at low strain (o=0.9) shown in Fig. 5 is a characteristic deformation structure with extended Dense Dislocation Walls (DDWs) and Microbands (MBs) [12]. The DDW/MBs bound cell blocks containing ordinary dislocation cells. Within many of the cells dislocation tangles can be observed. The banded structure observed by TEM shows no relationship to the banded structure observed optically
M.Richert et al.Materials Science and Engineering A260 (1999)275-283 279 indicating localized deformation in slip bands and shear mately 200 boundaries are analyzed at each strain bands.Localized deformation indicated by the presence After a strain of 0.9.all angles are below 10 with the of S-bands,which is composed of many sets of S- majority being below 2 as shown in Fig.9(a).With shaped dislocation boundaries,in the TEM-microstruc- increasing strains.the average misorientation angle in- ture has,however,been reported previously [5,9]in creases and the angular spread increases.Fig.9(b) CEC-deformed commercial purity aluminium (99.5%). shows an example of a specimen deformed to a strain At increasing strains,the structure gradually evolves of ~30.This figure shows that a substantial fraction of into an equiaxed cell/subgrain structure,where cell the boundaries can be characterized as high angle boundaries are less well formed,dislocation boundaries boundaries with an angle above 15.Finally,at a strain and subgrain boundaries are characterized by being of 60,Fig.9(c)shows that approximately two thirds of well formed,sharp boundaries.This is illustrated in the boundaries have developed into being high angle Fig.6 showing the structure of a sample after a strain boundaries.The presence of both low angle and high of 60(67 cycles).Also this structure contains disloca- angle boundaries in the microstructure is illustrated in tion tangles in the interior of the cells/subgrains.This Fig.10 showing the structure of a specimen deformed evolution from a banded to an equiaxed structure takes to a strain of 30.It is characteristic that low and high place gradually as a slightly banded structure is still angle boundaries are observed quite randomly dis- observed after strains of 22 and 30.An example of this tributed in the structure. transition structure is shown in Fig.7 for a specimen deformed to a strain of ~30(33 cycles).None of the specimens examined showed indications of dynamic or 4.Discussion static recrystallization.The spacing (D)between the boundaries in the deformed structure has been deter- 4.1.Deformation mode mined by counting the number of intersections between the boundaries and a random set of test lines in the Deformation by the CEC-method gives the possibil- longitudinal plane [13].The average spacing as a func- ity of deforming to very large cumulative plastic strains tion of the true strain is shown in Fig.8.This figure without sample shape changes.However,the method is shows that the spacing in the range 1.2-1.4 um is cyclic,i.e.the strain direction changes from one half almost independent of the strain over the whole strain cycle to the next.A comparison with,e.g.monotonic range taking into account that the standard deviation loading is therefore not straightforward.For that rea- of the spacing is ~10%. son,we shall first compare the present observations The misorientation angle (0)across the boundaries in with previous studies where a changing strain direction the deformed microstructure was determined by a has been applied,e.g.by cyclic loading in tension-com- Kikuchi pattern analysis in the TEM [10].Approxi- pression of pure aluminium [14]and multidirectional Fig.5.Thin foil from a longitudinal section of a specimen deformed to a strain of 0.9 (one cycle).An area shows narrow,long DDW/MBs forming cell blocks containing ordinary dislocation cells.The extrusion direction is marked ED
M. Richert et al. / Materials Science and Engineering A260 (1999) 275–283 279 indicating localized deformation in slip bands and shear bands. Localized deformation indicated by the presence of S-bands, which is composed of many sets of Sshaped dislocation boundaries, in the TEM-microstructure has, however, been reported previously [5,9] in CEC-deformed commercial purity aluminium (99.5%). At increasing strains, the structure gradually evolves into an equiaxed cell/subgrain structure, where cell boundaries are less well formed, dislocation boundaries and subgrain boundaries are characterized by being well formed, sharp boundaries. This is illustrated in Fig. 6 showing the structure of a sample after a strain of 60 (67 cycles). Also this structure contains dislocation tangles in the interior of the cells/subgrains. This evolution from a banded to an equiaxed structure takes place gradually as a slightly banded structure is still observed after strains of 22 and 30. An example of this transition structure is shown in Fig. 7 for a specimen deformed to a strain of 30 (33 cycles). None of the specimens examined showed indications of dynamic or static recrystallization. The spacing (Dr) between the boundaries in the deformed structure has been determined by counting the number of intersections between the boundaries and a random set of test lines in the longitudinal plane [13]. The average spacing as a function of the true strain is shown in Fig. 8. This figure shows that the spacing in the range 1.2–1.4 mm is almost independent of the strain over the whole strain range taking into account that the standard deviation of the spacing is 10%. The misorientation angle (u) across the boundaries in the deformed microstructure was determined by a Kikuchi pattern analysis in the TEM [10]. Approximately 200 boundaries are analyzed at each strain. After a strain of 0.9, all angles are below 10° with the majority being below 2° as shown in Fig. 9(a). With increasing strains, the average misorientation angle increases and the angular spread increases. Fig. 9(b) shows an example of a specimen deformed to a strain of 30. This figure shows that a substantial fraction of the boundaries can be characterized as high angle boundaries with an angle above 15°. Finally, at a strain of 60, Fig. 9(c) shows that approximately two thirds of the boundaries have developed into being high angle boundaries. The presence of both low angle and high angle boundaries in the microstructure is illustrated in Fig. 10 showing the structure of a specimen deformed to a strain of 30. It is characteristic that low and high angle boundaries are observed quite randomly distributed in the structure. 4. Discussion 4.1. Deformation mode Deformation by the CEC-method gives the possibility of deforming to very large cumulative plastic strains without sample shape changes. However, the method is cyclic, i.e. the strain direction changes from one half cycle to the next. A comparison with, e.g. monotonic loading is therefore not straightforward. For that reason, we shall first compare the present observations with previous studies where a changing strain direction has been applied, e.g. by cyclic loading in tension–compression of pure aluminium [14] and multidirectional Fig. 5. Thin foil from a longitudinal section of a specimen deformed to a strain of 0.9 (one cycle). An area shows narrow, long DDW/MBs forming cell blocks containing ordinary dislocation cells. The extrusion direction is marked ED.
280 M.Richert et al./Materials Science and Engineering A260(1999)275-283 Fig.6.Thin foil from a longitudinal section of a specimen deformed to a strain of 60(66 cycles).The area shows an equiaxed cell/subgrain structure.The extrusion direction is marked ED compression of commercial purity aluminium [15].In axis,e.g.in rolling a lamella structure parallel to the these studies as in the present one,it has been observed rolling plane is a characteristic structure after large that the microstructure after a substantial accumulated reductions of thickness.This is in contrast to the evolu- strain has a visual appearance typical of a high stacking tion towards an equiaxed cell/subgrain structure ob- fault energy material deformed at a relatively high served in specimens deformed under a changing strain homologous temperature,i.e.the structure looks like a direction.However,as smaller strains are generally warm worked (or a recovered)structure containing applied in monotonic loading,it cannot be ruled out many subgrain boundaries.However,a quantitative that at sufficiently high strains this deformation mode analysis is required to underpin these qualitative obser- can also lead to a transformation of the directional vations.As concerns the flow stress it has been found structure into an equiaxed one.That such a structural [14,15]that it saturates at a relatively low strain and is transformation will take place has however not been constant from this strain up to the quite large strains found experimentally.Both in pure (99.996%)alu- which have been applied.It is also found that the stress at which the strain hardening rate is zero increases with an increase in the amplitude of the plastic strain leading to a refinement of the microstructure [14,15].Quantita- tively,the observations agree with the present ones with respect to the evolution both in microstructure and in flow stress. Most studies on highly deformed aluminium have applied monotonic loading,e.g.rolling or wire drawing [4,16,17].It is therefore of interest to compare briefly the effect of monotonic deformation with deformation under a changing strain direction.Firstly,it is apparent that the deformation microstructure at small strains is quite comparable,as will be discussed in the next 5 um subsection.However,at increasing strain,the mi- crostructure under monotonic loading develops into a Fig.7.Thin foil from a longitudinal section of a specimen deformed structure where the cell/subgrain boundaries have a to a strain of 30 (33 cycles).An area shows an almost equiaxed macroscopic orientation with respect to the specimen cell/subgrain structure.The extrusion direction is marked ED
280 M. Richert et al. / Materials Science and Engineering A260 (1999) 275–283 Fig. 6. Thin foil from a longitudinal section of a specimen deformed to a strain of 60 (66 cycles). The area shows an equiaxed cell/subgrain structure. The extrusion direction is marked ED. compression of commercial purity aluminium [15]. In these studies as in the present one, it has been observed that the microstructure after a substantial accumulated strain has a visual appearance typical of a high stacking fault energy material deformed at a relatively high homologous temperature, i.e. the structure looks like a warm worked (or a recovered) structure containing many subgrain boundaries. However, a quantitative analysis is required to underpin these qualitative observations. As concerns the flow stress it has been found [14,15] that it saturates at a relatively low strain and is constant from this strain up to the quite large strains which have been applied. It is also found that the stress at which the strain hardening rate is zero increases with an increase in the amplitude of the plastic strain leading to a refinement of the microstructure [14,15]. Quantitatively, the observations agree with the present ones with respect to the evolution both in microstructure and in flow stress. Most studies on highly deformed aluminium have applied monotonic loading, e.g. rolling or wire drawing [4,16,17]. It is therefore of interest to compare briefly the effect of monotonic deformation with deformation under a changing strain direction. Firstly, it is apparent that the deformation microstructure at small strains is quite comparable, as will be discussed in the next subsection. However, at increasing strain, the microstructure under monotonic loading develops into a structure where the cell/subgrain boundaries have a macroscopic orientation with respect to the specimen axis, e.g. in rolling a lamella structure parallel to the rolling plane is a characteristic structure after large reductions of thickness. This is in contrast to the evolution towards an equiaxed cell/subgrain structure observed in specimens deformed under a changing strain direction. However, as smaller strains are generally applied in monotonic loading, it cannot be ruled out that at sufficiently high strains this deformation mode can also lead to a transformation of the directional structure into an equiaxed one. That such a structural transformation will take place has however not been found experimentally. Both in pure (99.996%) aluFig. 7. Thin foil from a longitudinal section of a specimen deformed to a strain of 30 (33 cycles). An area shows an almost equiaxed cell/subgrain structure. The extrusion direction is marked ED
M.Richert et al.Materials Science and Engineering A260 (1999)275-283 281 2 significantly by the accumulated strain.In contrast,the misorientation across the boundaries is significantly 1.6 affected by the strain as both the average misorienta- tion angle and the angular spread increase with the 12 strain (Fig.9).What is noteworthy is the gradual decrease in the number of low angle boundaries (015).Such a shift in the distribution 安 0.4 has also been observed in monotonic deformation where the strain is increased to a value in the range 3-4 [19].It can be noticed that although the angular distri- 10 20304050 60 70 bution shifts with increasing strain,the fraction of low TRUE STRAIN and medium angle boundary (<15)is still quite large Fig.8.Spacing between boundaries as a function of the accumulated at a strain of 60.This shows that the structure is strain. significantly different from that of a typically polycrys- talline metal,where the grain structure is obtained by minium [12]and in commercial purity aluminium [4],it deformation and recrystallization. has been found in longitudinal section of specimen cold-rolled to a thickness strain of s=2.3 that a well 50 defined lamellar structure has formed with lamellar boundaries almost parallel to the rolling plane.To do 40 such observation at high strain is however not straight- 30 forward owing to the experimental difficulty of prepar- ing edge-on TEM foils from specimens with a very 20 small cross section. 10 Finally.our brief comparison of monotonic loading with deformation under a changing strain direction 0 shows that a saturation stress is not reached in 0 10 20·30 40506070 6 ANGLE OF MISORIENTATIONS(Deg) monotonic loading even at strains in the range of 6 to 7 [4]and that the flow stress in monotonic loading for a similar total strain is higher than the saturation flow stress observed when deforming under a changing 24 strain direction. 20 16 4.2.Microstructural evolution 12 In general,it is observed at low strain that the 8 microstructure is subdivided by extended DDW/MBs forming cell blocks containing ordinary dislocation 0 hH且里o- cells.This structure has a visual appearance typical for (b) 0 10203040506070 aluminium deformed monotonically,e.g.in rolling ANGLE OF MISORIENTATIONS(Deg) [12,17]or in tension [18].At increasing deformation, this banded structure develops slowly into an equiaxed structure.It has been suggested [9,11]that localized g glide and shear are underlying processes for this trans- 20 formation,e.g.that crossing shear bands lead to frag- 16 mentation of the elongated microstructure characteris- 12 tic for the low and medium strain regime.In this connection it will be of importance to examine more closely,how the tendency to form shear bands is af- fected by the evolving deformation microstructure.This 738 0 is because,as seen in Fig.4,it cannot be ruled out that 0 10 2030 40 506070 the tendency to localized shear may be reduced once (c) ANGLE OF MISORIENTATIONS(Deg) the equiaxed structure becomes predominant. Fig.9.Histogram showing the angle of misorientation across An analysis of the microstructural parameters shows boundaries in a longitudinal section.(a)Strain 0.9:(b)Strain 30:(c) that the spacing between the boundaries is not affected Strain 60
M. Richert et al. / Materials Science and Engineering A260 (1999) 275–283 281 Fig. 8. Spacing between boundaries as a function of the accumulated strain. significantly by the accumulated strain. In contrast, the misorientation across the boundaries is significantly affected by the strain as both the average misorientation angle and the angular spread increase with the strain (Fig. 9). What is noteworthy is the gradual decrease in the number of low angle boundaries (uB 4°) and the increase in the number of high angle boundaries (u\15°). Such a shift in the distribution has also been observed in monotonic deformation where the strain is increased to a value in the range 3–4 [19]. It can be noticed that although the angular distribution shifts with increasing strain, the fraction of low and medium angle boundary (B15o ) is still quite large at a strain of 60. This shows that the structure is significantly different from that of a typically polycrystalline metal, where the grain structure is obtained by minium [12] and in commercial purity aluminium [4], it deformation and recrystallization. has been found in longitudinal section of specimen cold-rolled to a thickness strain of o=2.3 that a well defined lamellar structure has formed with lamellar boundaries almost parallel to the rolling plane. To do such observation at high strain is however not straightforward owing to the experimental difficulty of preparing edge-on TEM foils from specimens with a very small cross section. Finally, our brief comparison of monotonic loading with deformation under a changing strain direction shows that a saturation stress is not reached in monotonic loading even at strains in the range of 6 to 7 [4] and that the flow stress in monotonic loading for a similar total strain is higher than the saturation flow stress observed when deforming under a changing strain direction. 4.2. Microstructural e6olution In general, it is observed at low strain that the microstructure is subdivided by extended DDW/MBs forming cell blocks containing ordinary dislocation cells. This structure has a visual appearance typical for aluminium deformed monotonically, e.g. in rolling [12,17] or in tension [18]. At increasing deformation, this banded structure develops slowly into an equiaxed structure. It has been suggested [9,11] that localized glide and shear are underlying processes for this transformation, e.g. that crossing shear bands lead to fragmentation of the elongated microstructure characteristic for the low and medium strain regime. In this connection it will be of importance to examine more closely, how the tendency to form shear bands is affected by the evolving deformation microstructure. This is because, as seen in Fig. 4, it cannot be ruled out that the tendency to localized shear may be reduced once the equiaxed structure becomes predominant. An analysis of the microstructural parameters shows that the spacing between the boundaries is not affected Fig. 9. Histogram showing the angle of misorientation across boundaries in a longitudinal section. (a) Strain 0.9; (b) Strain 30; (c) Strain 60
282 M.Richert et al.Materials Science and Engineering A260 (1999)275-283 58° 5 54950 40° 2,59 o 350 5g0 414256 370 216 22 Fig.10.Thin foil from a longitudinal section of a specimen deformed to a strain of 30.The misorientation angle()across a number of boundaries is marked. In general,the microstructures observed at low and resistance of a dislocation boundary equals that of a medium strain are significantly affected by deformation high angle grain boundary given by the Hall-Petch mechanism involving ordinary glide and localized de- equation [23].An increase in misorientation angle formation in S-bands and shear bands.At high strains above such a critical angle should therefore not lead to there is no direct relationship between the morphology an increase in the flow stress for a constant spacing. of the structure and the deformation mechanism,which This tentative hypothesis is in good accord with the now may involve such processes as sliding and coales- experimental observation,but further quantitative mi- cence of grain boundaries.However,the microstruc- croscopy is required to underpin the hypothesis. tures at all strains are typical Low Energy Dislocation Structures (LEDS)[17,20].Observations supporting 5.Conclusions this statement are the checkerboard contrast observed in TEM micrographs and the increasing angle of mis- Pure,polycrystalline aluminium has been deformed orientation for a constant spacing leading to a reduc- at room temperature in cyclic extrusion compression tion in the energy per unit length of dislocation line (CEC)to true strains in the range 0.9-60.The follow- [19].At a large accumulated strain,many dislocation ing can be concluded: boundaries have been replaced by high angle The microstructure evolves from a cell block struc- boundaries in an equiaxed structure thereby reducing ture at lower strains into an equiaxed cell/subgrain the total surface energy of these boundaries. structure at large strain.This structural transformation is assisted by localized glide and shear.Dynamic or 4.3.Microstructure and flow stress static recrystallization structures have not been observed. The analysis of the microstructure shows a clear The average misorientation angle across the subdivision by dislocation boundaries and high angle boundaries subdividing the structure increases with the boundaries.The spacing between these boundaries is strain over the whole strain range leading to a distribu- almost unaffected by the strain in the range examined. tion of angles,which shows a relatively high concentra- It is therefore suggested that the contribution of the tion of high angle boundaries (>15)at a strain of deformation microstructure to the flow stress primarily 60.The average spacing between the boundaries is depends on how the boundaries affect the slip length practically unaffected by the strain. [21,22].Here,it may be assumed that the resistance of The flow stress increases with the accumulated strain a boundary depends primarily on the average density of to a saturation stress which is reached in the strain dislocations in the boundary,which increases linearly range 5-8.This evolution of flow stress as a function of with the angle of misorientation across the boundary the accumulated plastic strain is in good accord with [23].A critical angle may therefore exist where the the microstructural observations
282 M. Richert et al. / Materials Science and Engineering A260 (1999) 275–283 Fig. 10. Thin foil from a longitudinal section of a specimen deformed to a strain of 30. The misorientation angle (°) across a number of boundaries is marked. In general, the microstructures observed at low and medium strain are significantly affected by deformation mechanism involving ordinary glide and localized deformation in S-bands and shear bands. At high strains there is no direct relationship between the morphology of the structure and the deformation mechanism, which now may involve such processes as sliding and coalescence of grain boundaries. However, the microstructures at all strains are typical Low Energy Dislocation Structures (LEDS) [17,20]. Observations supporting this statement are the checkerboard contrast observed in TEM micrographs and the increasing angle of misorientation for a constant spacing leading to a reduction in the energy per unit length of dislocation line [19]. At a large accumulated strain, many dislocation boundaries have been replaced by high angle boundaries in an equiaxed structure thereby reducing the total surface energy of these boundaries. 4.3. Microstructure and flow stress The analysis of the microstructure shows a clear subdivision by dislocation boundaries and high angle boundaries. The spacing between these boundaries is almost unaffected by the strain in the range examined. It is therefore suggested that the contribution of the deformation microstructure to the flow stress primarily depends on how the boundaries affect the slip length [21,22]. Here, it may be assumed that the resistance of a boundary depends primarily on the average density of dislocations in the boundary, which increases linearly with the angle of misorientation across the boundary [23]. A critical angle may therefore exist where the resistance of a dislocation boundary equals that of a high angle grain boundary given by the Hall-Petch equation [23]. An increase in misorientation angle above such a critical angle should therefore not lead to an increase in the flow stress for a constant spacing. This tentative hypothesis is in good accord with the experimental observation, but further quantitative microscopy is required to underpin the hypothesis. 5. Conclusions Pure, polycrystalline aluminium has been deformed at room temperature in cyclic extrusion compression (CEC) to true strains in the range 0.9–60. The following can be concluded: The microstructure evolves from a cell block structure at lower strains into an equiaxed cell/subgrain structure at large strain. This structural transformation is assisted by localized glide and shear. Dynamic or static recrystallization structures have not been observed. The average misorientation angle across the boundaries subdividing the structure increases with the strain over the whole strain range leading to a distribution of angles, which shows a relatively high concentration of high angle boundaries (u\15°) at a strain of 60. The average spacing between the boundaries is practically unaffected by the strain. The flow stress increases with the accumulated strain to a saturation stress which is reached in the strain range 5–8. This evolution of flow stress as a function of the accumulated plastic strain is in good accord with the microstructural observations
M.Richert et al.Materials Science and Engineering A260 (1999)275-283 283 Acknowledgements International Symposium on Metallurgy and Materials Science, Riso National Laboratory.Roskilde,1981.p.445. We thank H.Nilsson,J.Lindbo,P.B.Olesen for [9]M.Richert,H.J.McQueen,in:H.J.MeQueen,E.V.Konopleva. N.D.Ryan (Eds.),Proceedings of the International Symposium careful assistance with the experimental work with the on Hot Workability of Steels and Light Alloys Composites microscopic analysis and Eva Sorensen for preparing Montreal,1996,p.15. the manuscript.We are also grateful for fruitful discus- [10]Q.Liu,Ultramicroscopy 60(1995)81. sions with D.A.Hughes,D.Juul Jensen,A.Godfrey, [11]M.Richert,Structural and Mechanical Effects of Strain Local- O.B.Pedersen. ization in Al 99.992 and AlMg 5 in the Range of Large Defor- mations,Dissertation,Krakow,1995. [12]B.Bay,N.Hansen,D.A.Hughes,D.Kuhlmann-Wilsdorf,Acta Metall.Mater.40 (1992)205. References [13]E.E.Underwood,Quantitative Stereology.Addison-Wesley, Reading.MA.1970.p.274. [1]J.D.Embury.A.S.Keh,R.M.Fischer.Trans.AIME 236(1966) [14]A.Giese,A.Styczynski,Y.Estrin,Mater.Sci.Eng.A124(1990) 11. 1252. [2]G.Langford,M.Cohen,Metall.Trans.A 6(1975)901. [15]P.E.Armstrong.J.E.Hockett,O.D.Sherby,J.Mech.Phys. [3]J.Gil Sevillano,P.van Houtte,E.Aernoudt,Proc.Mater.Sci.25 Solids30(1982)37. 1981)69. [16]N.Hansen,D.Juul Jensen,in:D.G.Brandon et al.(Eds.), [4]S.S.Hecker,M.G.Stout,in:G.Krauss (Ed.).Deformation Strengths of Metals and Alloys,Freund.London,1991,p.953. Processing and Structures,American Society for Metals,Ohio, [17]N.Hansen,D.A.Hughes.Phys.Status Solidi B 149 (1995)155. 1984,pp.1-46. [18]X.Huang,N.Hansen,Scripta Mater.37 (1997)1. [5]J.Richert,M.Richert,Aluminium 8(1986)604. [19]D.A.Hughes,N.Hansen,Acta Mater.45 (1997)387 [6]D.J.Lloyd,D.Kenny,Acta Metall.28 (1980)639. [20]D.Kuhlmann-Wilsdorf,Phys.Status Solidi A 149 (1995)225 [7]D.A.Hughes,in:N.Hansen,D.Juul Jensen,Y.L.Liu,B.Ralph [21]N.Hansen,in:S.I.Andersen et al.(Eds.),Numerical Predictions (Eds.).Microstructural and Crystallographic Aspects of Recrys- of Deformation Processes and the Behaviour of Real Materials. tallization,Proceedings of the Sixteenth Riso International Sym- Proceedings of the Fifteenth Rise International Symposium on posium on Materials Science,Rise National Laboratory, Materials Science.Rise National Laboratory,Roskilde,1994.p. Roskilde,1995,p.63. 325. [8]A.Korbel,M.Richert,J.Richert,in:N.Hansen,A.Horsewell, [22]N.Hansen.To be presented the AMT'98 Conference Krakok- T.Leffers,H.Lilholt (Eds.),Deformation of Polycrystals;Mech- Krynica,Poland,17-21 May,1998. anisms and Microstructures,Proceedings of the Second Rise [23]J.C.M.Li,Trans.Metall.Soc.AIME 227 (1963)239
M. Richert et al. / Materials Science and Engineering A260 (1999) 275–283 283 Acknowledgements We thank H. Nilsson, J. Lindbo, P.B. Olesen for careful assistance with the experimental work with the microscopic analysis and Eva Sørensen for preparing the manuscript. We are also grateful for fruitful discussions with D.A. Hughes, D. Juul Jensen, A. Godfrey, O.B. Pedersen. References [1] J.D. Embury, A.S. Keh, R.M. Fischer, Trans. AIME 236 (1966) 1252. [2] G. Langford, M. Cohen, Metall. Trans. A 6 (1975) 901. [3] J. Gil Sevillano, P. van Houtte, E.Aernoudt, Proc. Mater. Sci. 25 (1981) 69. [4] S.S. Hecker, M.G. Stout, in: G. Krauss (Ed.), Deformation Processing and Structures, American Society for Metals, Ohio, 1984, pp. 1–46. [5] J. Richert, M. Richert, Aluminium 8 (1986) 604. [6] D.J. Lloyd, D. Kenny, Acta Metall. 28 (1980) 639. [7] D.A. Hughes, in: N. Hansen, D. Juul Jensen, Y.L. Liu, B. Ralph (Eds.), Microstructural and Crystallographic Aspects of Recrystallization, Proceedings of the Sixteenth Risø International Symposium on Materials Science, Risø National Laboratory, Roskilde, 1995, p. 63. [8] A. Korbel, M. Richert, J. Richert, in: N. Hansen, A. Horsewell, T. Leffers, H. Lilholt (Eds.), Deformation of Polycrystals; Mechanisms and Microstructures, Proceedings of the Second Risø International Symposium on Metallurgy and Materials Science, Risø National Laboratory, Roskilde, 1981, p. 445. [9] M. Richert, H.J. McQueen, in: H.J. McQueen, E.V. Konopleva, N.D. Ryan (Eds.), Proceedings of the International Symposium on Hot Workability of Steels and Light Alloys Composites, Montreal, 1996, p. 15. [10] Q. Liu, Ultramicroscopy 60 (1995) 81. [11] M. Richert, Structural and Mechanical Effects of Strain Localization in Al 99.992 and AlMg 5 in the Range of Large Deformations, Dissertation, Krakow, 1995. [12] B. Bay, N. Hansen, D.A. Hughes, D. Kuhlmann-Wilsdorf, Acta Metall. Mater. 40 (1992) 205. [13] E.E. Underwood, Quantitative Stereology, Addison-Wesley, Reading, MA, 1970, p. 274. [14] A. Giese, A. Styczynski, Y. Estrin, Mater. Sci. Eng. A124 (1990) 11. [15] P.E. Armstrong, J.E. Hockett, O.D. Sherby, J. Mech. Phys. Solids 30 (1982) 37. [16] N. Hansen, D. Juul Jensen, in: D.G. Brandon et al. (Eds.), Strengths of Metals and Alloys, Freund, London, 1991, p. 953. [17] N. Hansen, D.A. Hughes, Phys. Status Solidi B 149 (1995) 155. [18] X. Huang, N. Hansen, Scripta Mater. 37 (1997) 1. [19] D.A. Hughes, N. Hansen, Acta Mater. 45 (1997) 387. [20] D. Kuhlmann-Wilsdorf, Phys. Status Solidi A 149 (1995) 225. [21] N. Hansen, in: S.I. Andersen et al. (Eds.), Numerical Predictions of Deformation Processes and the Behaviour of Real Materials, Proceedings of the Fifteenth Risø International Symposium on Materials Science, Risø National Laboratory, Roskilde, 1994, p. 325. [22] N. Hansen. To be presented the AMT’98 Conference KrakokKrynica, Poland, 17–21 May, 1998. [23] J.C.M. Li, Trans. Metall. Soc. AIME 227 (1963) 239.