You are on page 1of 11

Materials Science and Engineering A 398 (2005) 66–76

The evolution of homogeneity and grain refinement during equal-channel


angular pressing: A model for grain refinement in ECAP
Cheng Xu a , Minoru Furukawa b , Zenji Horita c , Terence G. Langdon a,∗
aDepartments of Aerospace & Mechanical Engineering and Materials Science, University of Southern California,
Los Angeles, CA 90089-1453, USA
b Department of Technology, Fukuoka University of Education, Munakata, Fukuoka 811-4192, Japan
c Department of Materials Science and Engineering, Faculty of Engineering, Kyushu University, Fukuoka 812-8581, Japan

Received 2 November 2004; received in revised form 17 February 2005; accepted 16 March 2005

Abstract

Samples of pure aluminum and an Al-6061 alloy were subjected to equal-channel angular pressing (ECAP) for up to eight passes at room
temperature. The Vickers microhardness was recorded on the polished cross-sections of each as-pressed billet and the results are plotted in
the form of color-coded maps to provide a pictorial depiction of the hardness distributions throughout the cross-sections. The results show
the hardness increases by a factor of ∼2 in the first pass through the die but thereafter there is only a minor increase. The microhardness
distribution is homogeneous in the unpressed condition but becomes inhomogeneous after a single pass and then reasonably homogeneous in
subsequent passes. The results reveal some differences between the behavior of pure Al and the Al-6061 alloy. These results and microscopic
observations are used to develop a model for the process of grain refinement during ECAP.
© 2005 Elsevier B.V. All rights reserved.

Keywords: Equal-channel angular pressing; Grain refinement; Hardness; Homogeneity; Severe plastic deformation

1. Introduction ing (ECAP) in which a metallic rod or bar is pressed through


a die constrained within a channel which is bent through an
The processing of metals through the application of severe abrupt angle that is often equal to, or very close to, 90◦ [2–4].
plastic deformation (SPD) has become attractive in recent Processing by ECAP leads to significant strengthening of
years, because it provides the capability of achieving remark- the material at ambient temperatures [5,6] and, provided the
able grain refinement in polycrystalline materials, typically ultrafine grains have a reasonable thermal stability, to the oc-
to the submicrometer or even the nanometer level [1]. There currence of superplastic ductilities at very rapid strain rates
are two primary advantages of using SPD processing. First, it at elevated temperatures [7,8].
has a potential for producing large bulk samples without the Despite the apparent success in producing both high
introduction of any porosity or the contaminants that are often strength and high strain rate superplasticity in materials
inherent in alternative techniques such as inert gas conden- processed by ECAP, it is important to recognize that the
sation or ball milling. Second, conventional SPD processing tensile specimens used to establish these mechanical prop-
can be applied relatively easily to a wide range of metallic erties are generally cut from the as-pressed billets with their
alloys without making any significant changes in the process- gauge lengths lying parallel to the longitudinal or pressing
ing procedure. axes. This means, in effect, that the measured mechanical
The most attractive, and potentially the most useful, properties represent the characteristics associated with the
SPD technique appears to be equal-channel angular press- as-pressed microstructure existing within the central core
regions of the as-pressed billets and the mechanical testing
∗ Corresponding author. Tel.: +1 213 740 0491; fax: +1 213 740 8071. provides no information on the general degree of microstruc-
E-mail address: langdon@usc.edu (T.G. Langdon). tural homogeneity occurring in the cross-sections of the

0921-5093/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.msea.2005.03.083
C. Xu et al. / Materials Science and Engineering A 398 (2005) 66–76 67

billets perpendicular to the pressing axis. An earlier study The Al-6061 alloy was received in the form of a plate,
demonstrated the feasibility of effectively scaling the ECAP 10 mm in thickness, after a T651 heat treatment involving
process to include billets having diameters from 6 to 40 mm a solution treatment at 823 K, tensile straining to ∼2% at
[9] and in these experiments some tensile specimens, having room temperature and aging at 433 K for 18 h. Samples with
gauge lengths of 5 mm, were cut from the 40 mm billet with diameters of 10 mm and lengths of 70 mm were cut from
their gauge lengths oriented in two mutually orthogonal the plate and annealed at 693 K for 4 h with a heating and
directions perpendicular to the pressing direction. Although cooling rate of 10 K h−1 . The initial grain size of the alloy in
the subsequent curves of stress versus strain were essentially the unpressed condition was ∼50 ␮m.
identical for all of these tensile specimens and for other The ECAP processing for both materials was performed
specimens cut parallel to the pressing direction, thereby at room temperature (298 K) using a hydraulic press of
suggesting a reasonably homogeneous microstructure, the 150 tonnes capacity operating at a ram speed of ∼7 mm s−1 .
results are insufficient to provide detailed information on The pressings were conducted using a solid die having an
either the degree of homogeneity within the cross-sections or L-shaped channel bent through an angle of Φ = 90◦ near the
the changes in any inhomogeneities with increasing numbers center of the die and with an additional angle of Ψ = 20◦ rep-
of passes through the die. This information is necessary resenting the outer arc of curvature where the two parts of
because the microstructure may change in the vicinity of the channel intersect. These values of Φ and Ψ lead to an im-
the die walls if, for example, frictional effects become posed strain of ∼1 on each passage of the sample through the
important. die [17]. An earlier investigation, also conducted using pure
The present investigation was initiated with three objec- Al, showed that the use of a solid die with an arc of curvature
tives. First, to use hardness measurements to evaluate the of Ψ = 20◦ has no significant influence on measurements of
homogeneity of the microstructures on sections cut through homogeneity in the as-pressed samples by comparison with a
the as-pressed billets perpendicular to the pressing direction. conventional split die where the channel is bent through 90◦
Second, to evaluate the significance of any evolution in the and there is no arc of curvature so that Ψ = 0◦ [18]. Samples
microstructural homogeneity as a function of the number of were processed by ECAP through selected numbers of passes
passes in ECAP and hence of the imposed strain. Third, to using route BC in which the samples are rotated by 90◦ in the
make use of these observations, together with earlier reports same sense between each pass [19]. This processing route
of microstructural development in ECAP, in order to construct was selected because earlier experiments on high-purity Al
a model to account for the evolution of an ultrafine-grained showed it leads most expeditiously to an array of equiaxed
structure when processing by ECAP. grains separated by boundaries having high angles of misori-
The experiments were conducted using two different ma- entation [12].
terials: high-purity aluminum and an Al-6061 alloy. These Following ECAP, the billets were sectioned perpendic-
materials were selected because there are several earlier stud- ular to their longitudinal axes and then mounted and care-
ies documenting the basic characteristics of high-purity alu- fully polished to a mirror-like finish. Microhardness mea-
minum [10–15] and the Al-6061 alloy [16] when processing surements were taken using an FM-1e microhardness tester
using ECAP. As will be demonstrated, there are microstruc- equipped with a Vickers indenter. A series of individual mea-
tural inhomogeneities in both of these materials after ECAP surements was recorded on each polished section whereby
through a single pass but the microstructures evolve with ad- the Vickers indenter was moved over the surface and mea-
ditional passes and, except only for very small regions in the surements of the Vickers microhardness, Hv, were recorded
vicinity of the bottom surfaces of the billets, they become in a regular grid pattern with spacings between each sepa-
essentially homogeneous in both materials. rate measurement of ∼0.5 mm. This procedure gave a total
of ∼220 individual values of Hv on each polished section.
These individual values of Hv were then plotted in the form
2. Experimental materials and procedures of color-coded contour maps depicting the variation of the
local microhardness over the cross-sectional plane for each
The experiments were conducted using pure aluminum of sample.
99.99% purity and a commercial Al-6061 alloy containing, The internal microstructures of pure Al were examined
in wt.%, 1.01% Mg, 0.59% Si, 0.37% Fe, 0.29% Cu, 0.23% using an Hitachi H-8100 transmission electron microscope
Cr and 0.20% Zn. (TEM) operating at 200 kV. Specimens were prepared for
The aluminum was prepared by cold-rolling from an in- TEM in the form of disks, 3 mm in diameter, oriented in the
got into a plate with a thickness of 25 mm, cutting into a plane of sectioning perpendicular to the longitudinal axes of
block with cross-sectional dimensions of 25 mm × 25 mm the billets. These disks were ground mechanically to thick-
and swaging into cylindrical rods having diameters of 10 mm. nesses of ∼150 ␮m and then thinned to perforation using
For processing by ECAP, individual billets were cut from a twin-jet electropolishing facility with a solution of 10%
these rods with lengths of ∼70 mm. These billets were an- HClO4 , 20% C3 H8 O3 and 70% C2 H5 OH with the solution
nealed at 773 K for 1 h prior to ECAP to give an initial un- maintained at a temperature of 278 K. Observations were
pressed grain size of ∼1 mm. recorded by TEM both in the central regions of the billets
68 C. Xu et al. / Materials Science and Engineering A 398 (2005) 66–76

second pass to a maximum value of Hv ≈ 44 and then a


slight decrease to a value of Hv ≈ 42 after a total of eight
passes.

3.2. Distribution of microhardness through


cross-sections of the billets

3.2.1. Pure aluminum


All of the microhardness measurements were plotted in
the form of contour maps depicting, within incremental steps,
the locations of individual microhardness values on the cross-
sectional planes. In plotting these maps, the positions of the
microhardness measurements were taken from the rectilin-
ear coordinate system used in the experiments with the Y
Fig. 1. The average Vickers microhardness, Hv, vs. the number of passes in
and the Z axes defined as the directions tangential to the
ECAP for pure aluminum and an Al-6061 alloy processed at room temper- top point and the side point of the circular cross-section of
ature using route BC . the billet at the point of exit from the die, respectively: fol-
lowing the coordinate system introduced earlier for ECAP
and in the vicinity of the lower surface. The grain sizes were [11], this corresponds to directions perpendicular to the flow
measured in the TEM and selected area electron diffraction (Y) and longitudinal (Z) planes, respectively. The distribu-
(SAED) patterns were recorded using an aperture size of tions of the microhardness values are represented pictori-
12.3 ␮m. ally through a series of colors defined within these incre-
mental values of Hv, thereby giving a direct and visual
presentation of the sample homogeneity before and after
3. Experimental results ECAP.
Fig. 2 shows the color-coded hardness maps for pure Al
3.1. The variation of the average microhardness with where the number of passes is indicated above each display
strain and the meanings of the colors are given at the lower right. The
map at upper left in Fig. 2 shows the as-received condition
The value of the Vickers microhardness, Hv, was deter- immediately prior to ECAP and it is apparent that the values
mined for each billet by calculating the average of all mi- of Hv, having an average value of Hv ≈ 19, are essentially
crohardness measurements on each plane of sectioning. The independent of the location on the section thereby confirming
results are plotted in Fig. 1 against the total number of passes the presence of a reasonably homogeneous microstructure
in ECAP corresponding to imposed strains up to a maxi- throughout the billet.
mum of ∼8. It is important to note that the error bars associ- The contour maps obtained after processing by ECAP
ated with these points are extremely small because each point are shown in the remaining displays. Two conclusions may
represents the average of ∼220 individual readings. Table 1 be drawn from the display at upper right for a single pass
records the error bars at the 95% confidence limit and it can in ECAP. First, the microhardness after one pass is sig-
be seen the errors in the values of Hv are consistently <1 and nificantly higher than in the unpressed condition and this
in practice they are smaller than the size of the individual increase has occurred over the entire section of the bil-
datum points plotted in Fig. 1. let so that there is no region where the microhardness re-
It is apparent from Fig. 1 that the microhardness of mains at, or even close to, Hv ≈ 19. Second, the distribu-
both materials increases dramatically after a single pass of tion of the microhardness measurements is less homoge-
ECAP by a factor of approximately 2×. Thereafter, there neous than in the unpressed condition and there is a broad
is a small increase in each additional pass for the Al-6061 region in the vicinity of the lower surface where the val-
alloy up to a maximum value of Hv ≈ 99 after six passes ues of Hv are markedly lower than average with values of
whereas for pure Al there is a very small increase in the Hv ≈ 30–35. This region of lower hardness occurs in the ma-

Table 1
Average values of the microhardness and errors at the 95% confidence level for pure Al and the Al-6061 alloy
Material Pass number

0 1 2 3 4 6 8
Pure Al 18.6 ± 0.1 40.4 ± 0.6 44.1 ± 0.5 44.2 ± 0.5 43.2 ± 0.4 43.3 ± 0.4 41.9 ± 0.5
Al-6061 44.7 ± 0.2 77.9 ± 0.6 85.4 ± 0.7 91.7 ± 0.6 94.8 ± 0.9 99.5 ± 0.7 –
C. Xu et al. / Materials Science and Engineering A 398 (2005) 66–76 69

Fig. 2. Color-coded contour maps showing the Vickers microhardness through the cross-sections of pure aluminum billets in the unpressed condition and after
pressing through one, two, three, four and eight passes of ECAP at room temperature using route BC : the significance of the color coding is shown at the lower
right.
70 C. Xu et al. / Materials Science and Engineering A 398 (2005) 66–76

terial passing through the arc of curvature where the two strate that, by comparison with pure Al, it is more difficult
parts of the channel intersect. The presence of this region to establish an equilibrium microstructure in the Al-6061
of lower hardness near the bottom surface reduces the av- alloy.
erage hardness to Hv ≈ 40 although the remainder of the
sample generally has hardness values within the range of 3.3. Microstructural observations on pure aluminum
Hv ≈ 40–50.
The situation after two passes is shown in the left col- There have been extensive microstructural observations of
umn in Fig. 2. Again, there is a region along the lower pure Al after ECAP [10–15] but no observations have been
surface where the hardness is lower than average but in undertaken to determine whether there are any differences
this sample the low hardness values are in the range of between the central and lower portions of the billet after a
Hv ≈ 35–40 and the region is confined to within a width of single pass. Accordingly, these two regions were examined
∼1 mm from the bottom surface of the billet. For the sample separately using TEM.
pressed through three passes there is an essentially homoge- Figs. 4 and 5 show representative areas and SAED patterns
neous distribution of hardness values with some lower val- after one pass near the center of the billet and very close to
ues confined only to a very narrow peripheral region around the bottom surface, respectively: for both sets of photomicro-
the perimeter of the billet having a maximum thickness of graphs, the regions labeled (a) correspond to areas of elon-
<0.2 mm. The situation is similar after four and eight passes gated grains and the regions labeled (b) correspond to areas
and there is a general homogeneity in the hardness values of reasonably equiaxed grains.
for these two conditions except only in a very small region, In the central region of the billet, it was estimated that
<0.2 mm in width, immediately adjacent to the lower sur- the elongated grains and the equiaxed grains each occupied
face. area fractions of ∼50% of the total area. For the elongated
The improvement in homogeneity with increasing num- grains shown in Fig. 4(a), the average widths and lengths
bers of passes is due both to a reduction in the extent of the were ∼1.3 and ∼5 ␮m, respectively. For the equiaxed grains
area at the lower surface where the hardness values are lower shown in Fig. 4(b), the average grain size was ∼1.3 ␮m.
than average and also to the disappearance of isolated spots of Thus, these measurements are in excellent agreement with
inhomogeneity where the measured microhardness is higher earlier reports, conducted on the same material, document-
or lower than average. These spots are visible after one, two ing average grain sizes of ∼1.3 ␮m [11] and ∼1.2 ␮m [15]
and three passes but they are absent after pressing through after four passes using route BC . Furthermore, the mea-
four and eight passes. surements provide a clear demonstration that the equilib-
rium grain size achieved after four passes is dictated solely
3.2.2. The Al-6061 alloy by the deformation characteristics imposed on the sample
The equivalent color-coded contour maps for the Al-6061 in the first pass of ECAP. Close inspection of Fig. 4(a)
alloy are shown in Fig. 3: it is important to note that, for ease reveals some partially developed but reasonably equiaxed
in displaying the data in small incremental values of Hv, the grains lying within the elongated grains, thereby showing
various colors in Figs. 2 and 3 represent different values of the equiaxed array develops from the elongated structure
Hv. with an average size equal to the width of the elongated
This material again exhibits excellent homogeneity in grains. It is apparent also that these grains contain large
the unpressed condition with Hv ≈ 45 throughout the cross- numbers of intragranular dislocations and it is reasonable
section. There is a marked inhomogeneity after a single pass to assume, based on earlier studies [20–23], that the grain
with an average value of Hv ≈ 78 but with a large area, cov- boundaries are in high-energy non-equilibrium configura-
ering more than one-half of the cross-section, where the in- tions.
dividual values of Hv are in the range of ∼65–75 and a The elongated grains were more prevalent in the vicin-
smaller area, primarily near the upper surface, where the ity of the lower surface in Fig. 5 occupying an estimated
individual values of Hv are in the range of ∼90–100. As area fraction of ∼90% with an area fraction of only ∼10%
with pure Al, the region of low Hv occurs in the vicin- for the equiaxed grains. These elongated grains had aver-
ity of the lower surface of the billet but the region is age widths of ∼1.3 ␮m and lengths of ∼5 ␮m so that they
more extensive than in the pure material. This region of were identical in size to the elongated grains in the central
low Hv is partially removed after two passes and almost part of the specimen. It is apparent also that the diffrac-
completely removed after three passes to give a reason- tion spots in the SAED patterns are less broad in Fig. 5
ably homogeneous microstructure. However, unlike pure near the bottom surface of the billet than in Fig. 4 in the
Al where the microstructure becomes homogeneous and vicinity of the center of the billet so that the boundaries
the hardness saturates at Hv ≈ 42–44 after four passes, the introduced by ECAP have lower angles of misorientation
microstructure is also homogeneous in the Al-6061 alloy near the bottom surface. This observation is consistent with
but the hardness continues to slowly increase with aver- the predictions of finite element analyses showing less de-
age values for Hv of ∼92 after three passes, ∼95 after formation in the vicinity of the lower surface in ECAP
four passes and ∼99 after six passes. These results demon- [24–26].
C. Xu et al. / Materials Science and Engineering A 398 (2005) 66–76 71

Fig. 3. Color-coded contour maps showing the Vickers microhardness through the cross-sections of an Al-6061 alloy in the unpressed condition and after
pressing through one, two, three, four and six passes of ECAP at room temperature using route BC : the significance of the color coding is shown at the lower
right.
72 C. Xu et al. / Materials Science and Engineering A 398 (2005) 66–76

Fig. 4. Representative microstructures in the central part of the pure alu- Fig. 5. Representative microstructures near the bottom surface of the pure
minum billet after ECAP through one pass at room temperature showing aluminum billet after ECAP through one pass at room temperature showing
regions of (a) elongated grains and (b) equiaxed grains. regions of (a) elongated grains and (b) equiaxed grains.

The inhomogeneities observed at the lower surfaces of the


4. Discussion billets are attributed to the development of a shearing zone
during ECAP as illustrated schematically in Fig. 6. In theory,
4.1. The development of homogeneity during ECAP the channel within the ECAP die is bent through 90◦ and the
shear strain is applied to the sample along a sharp and well-
An important conclusion from these results is that, in defined shear plane corresponding to the abrupt transition
terms of the hardness distributions, pure Al and the Al-6061 between the two parts of the channel, as depicted in Fig. 6(a).
alloy both develop homogeneous structures throughout the In practice, however, it has been shown through model ex-
as-pressed billets when pressed through a minimum of four periments [28,29], through finite element analyses [30–37]
passes. However, whereas there is no significant change in and through a theoretical consideration of ECAP processing
the hardness distributions in pure Al between four and eight [38] that the use of an ECAP die with a sharp corner having
passes, there is a gradual increase in the average hardness val- an arc of curvature of Ψ = 0◦ leads to the occurrence of a
ues in the Al-6061 alloy up to at least six passes. This result “dead zone” at this outer corner where the billet is no longer
is consistent with an earlier report on Al–Mg alloys where in contact with the die wall. The development of this dead
it was shown that, due to the lower rates of recovery in the zone is depicted in Fig. 6(b) where shearing occurs through
alloys, the as-pressed grain sizes were smaller than in pure an angle θ defined with respect to the upper point where the
Al and it was necessary to impose a larger number of passes two parts of the channel intersect.
in order to establish a homogeneous microstructure [27]. For Although the problems associated with these dead zones
the present experiments, it was established earlier that the can be alleviated or removed through the use of dies hav-
equilibrium grain size of the Al-6061 alloy is ∼0.28 ␮m af- ing movable die walls [38,39], the construction of these dies
ter ECAP [16]. is not easy and, from a practical point of view at least, it
C. Xu et al. / Materials Science and Engineering A 398 (2005) 66–76 73

Fig. 6. Schematic illustration of the principle of ECAP: (a) the theoretical model with shearing occurring along a sharp and well-defined shear plane; (b) the
situation in practice with shearing occurring through an arc denoted by the angle θ.

is more convenient to use conventional dies where the die observations [10,11,14,15], the ultimate equiaxed grain size
walls are fixed. Calculations show the value of the angle θ in this material is also ∼1.3 ␮m up to 12 passes in ECAP.
subtending the dead zone depends upon both the strain hard- Thus, it is important to recognize that the size of the equi-
ening behavior of the material and the ram speed of the press: librium substructural unit is established in the first pass of
for example, the value of θ has been estimated as ∼51◦ for ECAP.
1100 aluminum when the ram speed is ∼1 mm s−1 [35]. In It has been shown also that the ultimate equiaxed grain
the present investigation with pure aluminum, the ram speed sizes in ECAP are comparable to the grain sizes produced
was ∼7 mm s−1 and the arc of curvature machined into the in conventional cold-working [41]. For example, extensive
solid die was Ψ = 20◦ . It is reasonable to conclude, therefore, experiments on the cold-working in compression of com-
that the presence of this small arc of curvature has no effect mercial purity Al at a temperature of 293 K and at strain rates
on the pressing of aluminum because the billets move away from 0.1 to 223 s−1 showed that the ultimate grain sizes were
from the outer corner to create the dead zone. This conclu- in the range from ∼1.13 to ∼1.00 ␮m [42] which is simi-
sion is supported by the similar results reported earlier when lar to, but slightly smaller than, the grain size of ∼1.3 ␮m
pressing Al with dies having arcs of curvature of 0◦ and 20◦ reported in the present investigation on pure Al after pro-
[18]. cessing by ECAP. It is necessary, therefore, to develop a
model for grain refinement incorporating these new obser-
4.2. A microstructural model for grain refinement in vations.
ECAP It is noted initially that, from the TEM observations in
Figs. 4 and 5, the area fractions of the elongated and equiaxed
It is well known that processing by ECAP leads to re- grains are related to the imposed strain. The observations
markable grain refinement in coarse-grained materials but show the grains tend to become more equiaxed when the
the general principles of this grain refinement have not been imposed strain is sufficiently large in the central region of
established. A microstructural model was developed for the the billet whereas near the bottom surface, where the strain
refinement of grains in Armco iron when using SPD process- is lower and the shearing occurs over an angular increment,
ing with high-pressure torsion [40] but the model incorpo- the grains are predominantly elongated.
rates a gradually increasing refinement of the microstructure Fig. 7 is a schematic illustration of a model for the grain
through the continuous introduction of additional disloca- refinement occurring in the central portion of the billet for
tions during the straining process. However, a model of grad- a typical fcc metal, such as aluminum, when processed by
ually increasing refinement is not consistent with the present ECAP. The top row of Fig. 7 labeled (a) denotes the mi-
observations where both the widths of the elongated grains crostructures visible on the X plane after one and two passes
and the average sizes of the equiaxed grains are ∼1.3 ␮m af- when using a die with Φ = 90◦ and processing route BC , the
ter a single pass in ECAP and, as shown by numerous detailed second row labeled (b) shows the microstructures on the Y
74 C. Xu et al. / Materials Science and Engineering A 398 (2005) 66–76

Fig. 7. A model for grain refinement in the central region of the billet in ECAP: (a) the substructures visible on the X plane after one and two passes of ECAP;
(b) the substructures visible on the Y plane after one and two passes of ECAP and (c) the substructures visible on the X and Y planes after four and eight passes
of ECAP; the ultimate equiaxed grain size is defined by d and the parameter ξ denotes the approximate fraction of high-angle boundaries in each substructure
based on detailed measurements on pure Al using orientation imaging microscopy [15].

plane after one and two passes, and the bottom row labeled Fig. 7 also includes estimates of the fractions of bound-
(c) shows the microstructures on both the X and Y planes after aries having high-angle misorientations, ξ, for each of the
totals of four and eight passes, respectively; the orientations microstructures: these values of ξ are best estimates based
of the upper two rows with respect to the orthogonal X, Y and on a plot of the fraction of high-angle boundaries versus the
Z directions are depicted in the small insets between the two number of ECAP passes in pure aluminum [15]. It is apparent
illustrations. that a higher fraction of high-angle boundaries is recorded on
C. Xu et al. / Materials Science and Engineering A 398 (2005) 66–76 75

the X plane for any selected number of passes and this is con- of planes spread over wide angular increments on each or-
sistent with an analysis of the shearing patterns where it was thogonal plane when processing using route BC , through the
shown that shearing extends through a larger angular incre- re-arrangement of the various intersecting arrays introduced
ment on the X plane by comparison with the Y plane when in the different passes into a low-energy configuration and
using route BC [43]. The values of ξ in Fig. 7(c) for the X and through the absorption of many of the intragranular disloca-
Y planes are the average estimates for these two planes af- tions into the subgrain boundaries. The result is illustrated in
ter four and eight passes, respectively [15]. The illustrations Fig. 7(c) where there is an array of essentially equiaxed grains
in Fig. 7 are consistent with the experimental observations, having a reasonably high fraction of high-angle boundaries
with earlier reports of the microstructures present on various after four passes so that ξ ≈ 0.56 [15]. The general appear-
planes of sectioning when pure Al is subjected to ECAP us- ance of the microstructure remains essentially unchanged af-
ing route BC [10,11] and with the shearing patterns predicted ter eight (or more) passes but the OIM observations suggest
theoretically [43]. a gradual increase in the fraction of high-angle boundaries as
In Fig. 7(a), a banded substructure is formed along the the imposed strain increases [15].
Y-direction in the first pass where, as shown in Fig. 4(a), ap- Contrary to the model developed earlier for Armco iron
proximately 50% of the grains are reasonably equiaxed and subjected to SPD processing [40], the model in Fig. 7 ac-
the remaining 50% are elongated: a substructure of this type counts for the initial formation of a banded structure of
is clearly visible in an earlier report [11] and it is apparent elongated subgrains and the subsequent development of an
from OIM that a large fraction of these boundaries have low equiaxed array of grains without the introduction of any ap-
angles of misorientation so that the structure represents an preciable change with increasing strain in the fundamental
array of elongated and equiaxed subgrains containing large dimensions of the smallest substructural unit. In addition, the
numbers of intragranular dislocations and many high-energy correlation of the measured grain size in ECAP with the grain
non-equilibrium boundaries. The width of this banded struc- size produced through cold-working [41] suggests that it is
ture measured perpendicular to the Y-axis is denoted by d. feasible to obtain preliminary information about the poten-
The introduction of a second pass using route BC leads, tial substructure introduced through ECAP by undertaking a
when viewed on the X plane, to a rotation of the banded careful review of the relevant metal-working literature.
array through an angle of ∼27◦ with respect to the Y-axis,
the development of a new banded array lying parallel to the
Y-direction, a reduction in the fraction of elongated grains 5. Summary and conclusions
and a gradual evolution into an equiaxed microstructure, and
a diminution in the numbers of intragranular dislocations so 1. Samples of pure aluminum and an Al-6061 alloy were
that, as these dislocations become absorbed into the subgrain processed by ECAP for up to eight passes and measure-
walls, there is a corresponding increase in the fraction of high- ments of the Vickers microhardness, Hv, were recorded on
angle boundaries. Based on experimental measurements, the the polished cross-sections following a regular rectilinear
fraction of high-angle boundaries in this condition is ξ ≈ 0.56 grid pattern. The results are presented in the form of color-
[15]. The average dimension of the minimum structural unit coded maps depicting the value of the local microhardness
introduced in this second pass is again equal to d. at each point on the section.
Similar illustrations are presented in Fig. 7(b) for obser- 2. The results show there is a homogeneous hardness dis-
vations on the Y plane where the banded arrays lie at an angle tribution in the unpressed materials but the average hard-
of ∼45◦ to the X-axis after ECAP through one pass and at ness increases by a factor of approximately 2× after a
angles of ∼27◦ for the first pass and ∼45◦ for the second single pass and remains reasonably constant with addi-
pass after ECAP through two passes using route BC [43]. tional passes for pure Al but gradually increases for the
The predictions of the model are consistent with microstruc- alloy.
tural observations on pure Al [11] where the minimum struc- 3. The hardness distributions in both materials are inhomo-
tural dimension, d, defines the widths of the banded arrays geneous after one pass with a lower hardness in the vicinity
after one and two passes. The model assumes the subgrains of the lower surface but the distribution becomes essen-
are formed initially with their longer axes oriented along the tially homogeneous with additional passes. Microstruc-
lines of shearing introduced in ECAP and this is consistent tural observations on Al show approximately equal frac-
both with earlier observations on the ECAP processing of tions of equiaxed and elongated grains in the central region
polycrystalline aluminum [11] and with experiments on alu- of the billet but a higher area fraction of elongated grains
minum single crystals showing subgrain bands develop lying (∼90%) near the lower surface.
parallel to the primary slip planes [44]. 4. The experimental results are used with earlier microstruc-
It is well-established that the appearance of the substruc- tural observations to construct a model for grain refine-
tural arrays in pure aluminum become more equilibrated, and ment in ECAP. An important characteristic of this model
essentially identical on the X, Y and Z planes, after pressing is that the ultimate size of the equilibrium equiaxed grains
through a total of four passes [11]. This equilibration oc- is dictated by the width of the subgrain bands introduced
curs through the development of shearing on multiple sets into the material in the first pass through the die.
76 C. Xu et al. / Materials Science and Engineering A 398 (2005) 66–76

Acknowledgements [17] Y. Iwahashi, J. Wang, Z. Horita, M. Nemoto, T.G. Langdon, Scr.


Mater. 35 (1996) 143.
[18] C. Xu, T.G. Langdon, Scr. Mater. 48 (2003) 1.
This work was supported in part by a Grant-in-Aid for
[19] M. Furukawa, Y. Iwahashi, Z. Horita, M. Nemoto, T.G. Langdon,
Scientific Research from the Ministry of Education, Science, Mater. Sci. Eng. A257 (1998) 328.
Sports and Culture of Japan, in part by the Light Metals Ed- [20] R.Z. Valiev, N.A. Krasilnikov, N.K. Tsenev, Mater. Sci. Eng. A137
ucational Foundation of Japan and in part by the National (1991) 35.
Science Foundation of the United States under Grant no. [21] R.Z. Valiev, A.V. Korznikov, R.R. Mulyukov, Mater. Sci. Eng. A168
(1993) 141.
DMR-0243331.
[22] J. Wang, Z. Horita, M. Furukawa, M. Nemoto, N.K. Tsenev, R.Z.
Valiev, Y. Ma, T.G. Langdon, J. Mater. Res. 8 (1993) 2810.
[23] Z. Horita, D.J. Smith, M. Furukawa, M. Nemoto, R.Z. Valiev, T.G.
References Langdon, J. Mater. Res. 11 (1996) 1880.
[24] H.S. Kim, M.H. Seo, S.I. Hong, J. Mater. Proc. Tech. 113 (2001)
[1] R.Z. Valiev, R.K. Islamgaliev, I.V. Alexandrov, Prog. Mater. Sci. 45 622.
(2000) 103. [25] R. Srinivasan, Scr. Mater. 44 (2001) 91.
[2] V.M. Segal, Mater. Sci. Eng. A197 (1995) 157. [26] J.-Y. Suh, H.-S. Kim, J.-W. Park, J.-Y. Chang, Scr. Mater. 44 (2001)
[3] V.M. Segal, Mater. Sci. Eng. A271 (1999) 322. 677.
[4] M. Furukawa, Z. Horita, M. Nemoto, T.G. Langdon, J. Mater. Sci. [27] Y. Iwahashi, Z. Horita, M. Nemoto, T.G. Langdon, Metall. Mater.
36 (2001) 2835. Trans. 29A (1998) 2503.
[5] V.V. Stolyarov, Y.T. Zhu, I.V. Alexandrov, T.C. Lowe, R.Z. Valiev, [28] Y. Wu, I. Baker, Scr. Mater. 37 (1997) 437.
Mater. Sci. Eng. A199 (2001) 59. [29] A. Shan, I.-G. Moon, H.-S. Ko, J.-W. Park, Scr. Mater. 41 (1999)
[6] R.Z. Valiev, I.V. Alexandrov, Y.T. Zhu, T.C. Lowe, J. Mater. Res. 353.
17 (2002) 5. [30] J.R. Bowen, A. Gholinia, S.M. Roberts, P.B. Prangnell, Mater. Sci.
[7] R.Z. Valiev, D.A. Salimonenko, N.K. Tsenev, P.B. Berbon, T.G. Eng. A287 (2000) 87.
Langdon, Scr. Mater. 37 (1997) 1945. [31] J.-W. Park, J.-Y. Suh, Metall. Mater. Trans. 32A (2001) 3007.
[8] S. Komura, Z. Horita, M. Furukawa, M. Nemoto, T.G. Langdon, [32] H.S. Kim, J. Mater. Res. 17 (2002) 172.
Metall. Mater. Trans. 32A (2001) 707. [33] P.B. Prangnell, C. Harris, S.M. Roberts, Scr. Mater. 37 (1997) 983.
[9] Z. Horita, T. Fujinami, T.G. Langdon, Mater. Sci. Eng. A318 (2001) [34] D.P. DeLo, S.L. Semiatin, Metall. Mater. Trans. 30A (1999) 1391.
34. [35] H.S. Kim, M.H. Seo, S.I. Hong, Mater. Sci. Eng. A291 (2000) 86.
[10] Y. Iwahashi, Z. Horita, M. Nemoto, T.G. Langdon, Acta Mater. 45 [36] S.L. Semiatin, D.P. DeLo, E.B. Shell, Acta Mater. 48 (2000) 1841.
(1997) 4733. [37] H.S. Kim, Mater. Sci. Eng. A315 (2001) 122.
[11] Y. Iwahashi, Z. Horita, M. Nemoto, T.G. Langdon, Acta Mater. 46 [38] V.M. Segal, Mater. Sci. Eng. A345 (2003) 36.
(1998) 3317. [39] A. Shan, I.-G. Moon, J.-W. Park, J. Mater. Proc. Tech. 122 (2002)
[12] K. Oh-ishi, Z. Horita, M. Furukawa, M. Nemoto, T.G. Langdon, 255.
Metall. Mater. Trans. 29A (1998) 2011. [40] R.Z. Valiev, Yu.V. Ivanisenko, E.F. Rauch, B. Baudelet, Acta Mater.
[13] Y. Iwahashi, M. Furukawa, Z. Horita, M. Nemoto, T.G. Langdon, 44 (1996) 4705.
Metall. Mater. Trans. 29A (1998) 2245. [41] S.L. Semiatin, P.B. Berbon, T.G. Langdon, Scr. Mater. 44 (2001)
[14] P.B. Berbon, M. Furukawa, Z. Horita, M. Nemoto, T.G. Langdon, 135.
Metall. Mater. Trans. 30A (1999) 1989. [42] H.J. McQueen, J.E. Hockett, Metall. Trans. 1 (1970) 2997.
[15] S.D. Terhune, D.L. Swisher, K. Oh-ishi, Z. Horita, T.G. Langdon, [43] M. Furukawa, Z. Horita, T.G. Langdon, Mater. Sci. Eng. A332
T.R. McNelley, Metall. Mater. Trans. 33A (2002) 2173. (2002) 97.
[16] Z. Horita, T. Fujinami, M. Nemoto, T.G. Langdon, Metall. Mater. [44] Y. Fukuda, K. Oh-ishi, M. Furukawa, Z. Horita, T.G. Langdon, Acta
Trans. 31A (2000) 691. Mater. 52 (2004) 1387.

You might also like