11
ELSEVIER Secondary Processing of Electron Backscatter Data from an Aluminum Alloy Robert Davies and Valerie Randle Department of Materials Engineering, University College of Wales, Swansea, Swansea SA2 8PP, United Kingdom The aim of this paper is to show how secondary data processing of electron backscatter diffrac- tion data has been used to glean additional information such as proportions of selected individ- ual texture variants and distributions of misorientation angles between contiguous grains/sub- grains. For this purpose, experimental work was performed on a continuously recrystallized commercial Al-Fe-Si alloy. Deconvolution of the microtexture showed the major components to be the S texture, Cu texture, and (436]<323> texture. Further analysis revealed that from the S textures (123}<634> and {123)<412>, only two of the possible four were present-that is, (123)[634]/(213)[%4] and (123)[412]/(213)[i%2], respectively. The distribution of misorien- tation angles indicated that the proportion of boundaries at O-15” was higher than that ex- pected in a random distribution. Also, the higher proportion observed at 50-60” was ex- plained by the adjacency in the microstructure of texture variants that were twin related. 0 Elsevier Science Inc., 2996 INTRODUCTION Electron backscatter diffraction (EBSD) is becoming an established technique for in situ measurement of grain orientations-that is, microtexture. Briefly, the method relies on positioning a stationary probe on selected regions of microstructure in a scanning electron microscope (SEM). If the specimen is highly tilted with respect to the incident electron beam, the backscattered diffracted signal is sufficiently strong to be captured by a low-light television camera and integrated with dedicated software to give the crystal orientation. The principles and practice of EBSD are described in detail elsewhere [l] . The primary input for orientation analy- sis consists of Kikuchi diffraction patterns collected from operator-selected regions of microstructure. Three main factors degrade this primary input and hence limit applica- tion of EBSD: 1. Grain/s&grain size. A standard EBSD system can measure the orientation of a region of <0.5pm in diameter and ap- 131 MATERIALS CHARACTERIZATION 37:131-141 (1996) 0 Elsevier Science Inc., 1996 655 Avenue of the Americas, New York, NY 10010 proximately 20-50nm in depth. Smaller grain sizes, or the sampling of more than one grain in the probe, result in corrup- tion of the diffraction pattern. High lattice strain. Plastic strain, caused typically by dislocations or high local sol- ute content, progressively blurs the dif- fraction pattern as the level increases until eventually the pattern is unsolvable [2]. The amount of strain that can be tolerated depends both on the material and on the strain distribution; a cellular dislocation substructure often permits adequate un- distorted sampling within cells but not at cell walls [3]. Specimen preparation. The constraints with regard to specimen preparation arise as a corollary to points (1) and (2): the pat- terns originate from the very top layer of the surface, which in turn needs to be rela- tively unstrained. Any agent that occludes the surface, such as a product of etching, or oxidation, or a coating, will obstruct diffraction pattern formation. Further- more, where specimen preparation has 1044.5803/96/$15.00 PII SlO44-5803(96)00095-2

Secondary processing of electron backscatter data from an aluminum alloy

Embed Size (px)

Citation preview

Page 1: Secondary processing of electron backscatter data from an aluminum alloy

ELSEVIER

Secondary Processing of Electron Backscatter Data from an Aluminum Alloy Robert Davies and Valerie Randle Department of Materials Engineering, University College of Wales, Swansea, Swansea SA2 8PP, United Kingdom

The aim of this paper is to show how secondary data processing of electron backscatter diffrac- tion data has been used to glean additional information such as proportions of selected individ- ual texture variants and distributions of misorientation angles between contiguous grains/sub- grains. For this purpose, experimental work was performed on a continuously recrystallized commercial Al-Fe-Si alloy. Deconvolution of the microtexture showed the major components to be the S texture, Cu texture, and (436]<323> texture. Further analysis revealed that from the S textures (123}<634> and {123)<412>, only two of the possible four were present-that is, (123)[634]/(213)[%4] and (123)[412]/(213)[i%2], respectively. The distribution of misorien- tation angles indicated that the proportion of boundaries at O-15” was higher than that ex- pected in a random distribution. Also, the higher proportion observed at 50-60” was ex- plained by the adjacency in the microstructure of texture variants that were twin related. 0 Elsevier Science Inc., 2996

INTRODUCTION

Electron backscatter diffraction (EBSD) is becoming an established technique for in situ measurement of grain orientations-that is, microtexture. Briefly, the method relies on positioning a stationary probe on selected regions of microstructure in a scanning electron microscope (SEM). If the specimen is highly tilted with respect to the incident electron beam, the backscattered diffracted signal is sufficiently strong to be captured by a low-light television camera and integrated with dedicated software to give the crystal orientation. The principles and practice of EBSD are described in detail elsewhere [l] .

The primary input for orientation analy- sis consists of Kikuchi diffraction patterns collected from operator-selected regions of microstructure. Three main factors degrade this primary input and hence limit applica- tion of EBSD: 1. Grain/s&grain size. A standard EBSD

system can measure the orientation of a region of <0.5pm in diameter and ap-

131 MATERIALS CHARACTERIZATION 37:131-141 (1996) 0 Elsevier Science Inc., 1996 655 Avenue of the Americas, New York, NY 10010

proximately 20-50nm in depth. Smaller grain sizes, or the sampling of more than one grain in the probe, result in corrup- tion of the diffraction pattern. High lattice strain. Plastic strain, caused typically by dislocations or high local sol- ute content, progressively blurs the dif- fraction pattern as the level increases until eventually the pattern is unsolvable [2]. The amount of strain that can be tolerated depends both on the material and on the strain distribution; a cellular dislocation substructure often permits adequate un- distorted sampling within cells but not at cell walls [3]. Specimen preparation. The constraints with regard to specimen preparation arise as a corollary to points (1) and (2): the pat- terns originate from the very top layer of the surface, which in turn needs to be rela- tively unstrained. Any agent that occludes the surface, such as a product of etching, or oxidation, or a coating, will obstruct diffraction pattern formation. Further- more, where specimen preparation has

1044.5803/96/$15.00 PII SlO44-5803(96)00095-2

Page 2: Secondary processing of electron backscatter data from an aluminum alloy

132 R. Davies and V. Randle

deformed the surface-which is usually the case where heavy metallographic dia- mond polishing has been used-patterns will also often be obliterated. For most ma- terials, these problems are overcome by employing a suitable finishing stage such as electropolishing or slurry polishing to produce a suitable surface for EBSD anal- ysis [l].

With these three points in mind, most re- search including EBSD has been carried out on annealed and recrystallized materials that have a moderate grain size and respond readily to standard preparation routes. These materials include a range of nickel [4], copper [5], and iron-based alloys [6], many of which are commercially important and have tex- ture-related properties. Commercial alumi- num alloys, particularly when heavily de- formed or partially recrystallized, have been less widely studied because of the difficul- ties both in their preparation and in the subsequent generation of diffraction patterns. With regard to specimen preparation, there are practical obstacles due to the response of aluminum to most etchants; those which ade- quately prepare the surface for EBSD do not concurrently reveal the microstructure and vice versa. The lower atomic number of alu- minum, the larger lattice spacing, and other effects mean that fewer Kikuchi lines are discernible in the diffraction patterns them- selves; moreover, the Kikuchi lines tend to be indistinct. In addition, it is very difficult to distinguish the Kikuchi lines in diffraction patterns obtained from heavily deformed material, as already stated. Figure 1 illus- trates this latter point by comparing the dif- fraction pattern from aluminum deformed 80% and slightly recovered with that from fully recrystallized aluminum having under- gone the same amount of prior deformation.

Notwithstanding the difficulties relating to the application of EBSD to aluminum, an understanding of texture on a microscale in rolled aluminum sheet is crucial to improv- ing process control on a commercial level, leading to better product consistency. Hence there is motivation to overcome the technical challenges implicated above.

FIG. 1. Kikuchi diffraction patterns from aluminium (a) slightly recovered and (b) fully recrystallized, both of which have been cold rolled to 80% reduction.

When the primary input (i.e., diffraction patterns) has been satisfactorily indexed and stored, it is usually represented in the tradi- tional texture and graphical formats-pole figures or orientation distribution functions (ODFs) in Euler space or both. However, much more information can be gleaned by seconda y data processing to obtain proportions of selected texture variants, misorientation parameters, orientation-microstructure corre- lations, and various crystallographic relations [7]. These further processing steps also allow the extracted data to be quantified. It is worth noting that in general these data are inaccessible from macrotexture methods; for example, neither misorientation between neighboring grains/subgrains nor analysis of texture variants is possible. Most com- monly, the algorithms used to obtain the parameters are not part of an EBSD analysis

Page 3: Secondary processing of electron backscatter data from an aluminum alloy

Secondary Processing of EBSD Data

package but have to be produced separately for individual requirements.

The purpose of this article is to demon- strate how we have evolved a methodology for the application of EBSD to a particular ex- perimental program on continuous recrys- tallization in commercial aluminum alloys. The main focus of the paper is to demonstrate secondary data processing procedures- particularly, “deconvolution” of the pole figure to extract and quantify specific texture components and misorientation distribution determination. Differences between the EBSD approach and the traditional X-ray texture route will be highlighted. In addition, specimen preparation routes for EBSD on an Al-Fe-Si alloy will be described in detail.

SECONDARY DATA PROCESSING

PROPORTIONS OF SPECIFIED MICROTEXTURE COMPONENTS

Many textures generated during commercial material processing are multicomponent. For example the so-called copper rolling tex- ture, which is representative of several fee textures after cold rolling deformation in- cluding aluminum, is composed of mainly (112]<111>, (123}<634>, and (110]<112>, where (hkl]<uvw> refers to the idealized rolling plane and rolling direction, respec- tively. In microtexture research, the usual requirement is to analyze the sampled orien- tation population in terms of

l which texture components are present, l quantification of the component propor-

tions, and l distribution of the components in the mi-

crostructure.

The components present can be identified by inspecting a pole figure and matching the observed peaks to particular textures that are likely to be present-for example, the rolling texture components. Because, unlike an X-ray pole figure, the data are displayed discretely rather than continuously, it is more appropriate to refer to “clusters” than to peaks. Furthermore, in a microtexture,

133

pole figure orientations corresponding to in- dividual grains can be identified.

A cluster can be identified from an orien- tation matrix that corresponds to it. For ex- ample, in Fig. 2 the labeled cluster on the 111 pole figure is described by the orienta- tion matrix of a central grain:

0.99806 0.05993 0.01708 0.05974 -0.99815 0.01148 0.01774 0.01044 0.99979

With use of the convention whereby the first row of the matrix stipulates the rolling direction and the last row stipulates the rolling plane, the lowest Miller indices con- version of these two rows gives (OOl)[lOO].

The most crucial way in which EBSD out- put differs from X-ray (macrotexture) out- put is that X-ray pole figures are generated by measurement of diffracted intensity from selected sets of planes-for example, 111 or 100 [8]. This relates directly to the volume fraction of grains having planes of a specific orientation. Furthermore, the multiplicity of planes and other factors needs to be taken into account. Hence the strength intensity of texture components is usually given in terms of “times random” for the X-ray case. For the microtexture case, it is not convenient to experimentally link grain orientations to their volume frac- tions; therefore, the methodology described

FIG. 2. A <ill> pole figure displaying a cluster of grains where the orientation can be determined from the orientation of the central grain as shown. In this example, the cluster is identified as the (001}<100> texture.

Page 4: Secondary processing of electron backscatter data from an aluminum alloy

134

here, texture strengths are quoted directly as proportions of the total number of orien- tations sampled and each orientation repre- sents a different grain. A computer proce- dure calculates proportions of components in a microtexture, using a program that counts how many orientations fall within a chosen limit-usually 15“-of an ideal ori- entation. The required input is the data file of orientations plus a table of ideal orienta- tions. This list must include all the texture components, determined by inspection of the pole figure in conjunction with inspec- tion of individual orientation matrices.

A second key difference between the tex- ture outputs from X-ray diffraction and EBSD is that, in the former, specific pZanes are not dis- tinguished within plane families. A texture such as the S texture, which is referred to as either {123)<412> or {123]<634>, has four differ- ent variants, which are:

(123)[412] (123)[a2] (213)[142] (213)[a2]

or

(123)[634] (123)[634] (213)[364] (213)[364]

It should be stressed that these variants are difierent orientations; for example, a grain boundary would exist between grains hav- ing orientations (123)[634] and (213)[364]. Whereas X-ray diffraction does not distin- guish directly between these cases, EBSD recognizes different variants of the same family of textures, and the secondary pro- cessing software used here has provision for their statistics to be compiled separately.

Texture variants should not be confused with symmetry-related solutions. These solu- tions refer to the 24 different ways, related to the symmetry of the cubic system, in which an orientation can be expressed. Thus, con- tinuing with the example of the four S-tex- ture variants shown above, we can express each of the four variants in 24 equivalent (symmetry related) ways. These solutions are calculated from the orientation matrix. Ta- ble 1 shows each symmetry-related solution for the four texture variants. The secondary processing software needs to take account of the 24 solutions when computing the proportion of texture components present.

R. Davies and V. Randle

Table 1 The 24 Symmetry-Related Solutions

Calculated for the Four S-Texture

Variants

Sl s2 s3 s4

(123)[634] (123)[634) Ll

‘321436 - -- 123 634

321436

132 643 _- 123 634

132 643

2i3 364 -- 123 634

213 364

231326

312 463

231346

312 463

312 463

231346 _- 231346

312 463

213 364

i32 643

321436

213 364 -- - 132 643

321436

:-- 321436 -- - 123 634

321436

132 z43

123 634 -- 132 643

213 364

123 634

213 364

231346

312 463

231346

312 463 --- 312 463

2x 346

231346

3i? 463

213 364

i32 643

321436 -- 213 364 _- 132 643

321436

(213)[364]

312 463

213 364 321463

231346 --

213 364

231346

123 6x

213 364 _-

i23 634

132 643

321236 --- 132 643

321436 -- 321436

l%! 643

132 643 --

321 436

123 634

231% --

312 463 -_ 123 634 -_ _ 231346

312 463

(213)[3&2]

312 463

213 364

321463

231346 ---

213 364 --

231 346

123 634

zi3 364

i23 634 -- 132 643

3214%

132 643 - -- 321 436

321436

132 643

i??2 6a

321436

123 634

231346

312 463

123 634

231346 -- 321 463

MISORIENTATION DISTRIBUTION DETERMINATION

For the present case, (i.e., research into con- tinuous recrystallization), it is pertinent to know if neighboring grains/subgrains are joined by high-angle or low-angle bound- aries. Because EBSD samples an orientation from individual grains, the misorientation between neighboring grains can be calcu- lated. This process is not applicable to X-ray generated textures, because individual grams are not identified. The misorientation ma- trix is calculated from two orientation ma- trices as:

M = AZ-IA,

The misorientation angle is then obtained from:

cos0 = (a,, + aZ2 + a,,-‘)/2

Page 5: Secondary processing of electron backscatter data from an aluminum alloy

Secondary Processing of EBSD Data

Table 2 Chemical Composition (wt. %)

135

Alloy Fe

AA8079 1.33

Si

0.06

CU

0.001

Mil

0.012

M&T

0.001

Cr

0.001

Zn 2-i

0.016 0.015

Details of these procedures are given else- where [9].

An important objective of most microtex- ture investigations is to correlate the orienta- tion or m&orientation of individual grains with microstructural features. The valuable information gleaned can then be used to pro- vide information for a number of research studies including recrystallization and grain growth behavior. There is no straightforward way of representing data that combines fully spatial and orientation or misorientation data, because there may be as many as six indepen- dent variables-three for the (mis)orientation and three for the spatial coordinates. Fre- quently, the data are compressed into a more manageable form either by reducing the num- ber of variables to be displayed or by extract- ing distribution statistics from the data [7].

EXPERIMENTAL PROCEDURE

MATERIAL PROCESSING

The material used in the present experimen- tal investigation is a commercial aluminum alloy, AA8079, the composition of which is given in Table 2. The initial hot-band material

Short tramwM

transverse section section

FIG. 3. Schematic presentation of a sheet specimen showing planes and directions.

of 29.8mm gauge, was cold rolled to a final reduction of 97% by using a laboratory-scale rolling mill. Specimens machined from the cold rolled sheets were then annealed in a preheated air-circulating furnace at a temper- ature of 400°C for 30 min.

SPECIMEN PREPARATION

Specimen preparation for EBSD analysis can be regarded as a critical factor because the quality of the diffraction pattern obtained is strongly dependent on the condition of the specimen surface. To obtain good diffraction

Electron beam

RD

B ND

FIG. 4. (A) Conventional EBSD specimen setup. (B) EBSD specimen setup used in present experiment work.

Page 6: Secondary processing of electron backscatter data from an aluminum alloy

136 R. Davies and V. Randle

patterns, the surface must be free of plastic deformation, as mentioned in the introduc- tion. With this in mind, the only technique that can be used to prepare aluminum speci- mens for EBSD is electrolytic polishing, be- cause conventional polishing will occlude the surface and result in obliterated diffrac- tion patterns. In the present work, specimens were ground and then polished, using an electrolyte consisting of 5% perchloric acid in ethanol at a temperature of -20°C and volt- age of 40V. Total polishing times ranged from 1 to 2 min, depending on the quality of the surface finish prior to electropolishing. Pre- cautions were taken to ensure that the spec- imen did not heat up during the process. These entailed continuously cooling the elec- trolyte with liquid nitrogen and polishing the specimen at 30-s intervals.

Etchants and anodizing reagents that are commonly used to reveal the grain struc- ture of aluminum when viewed in cross- polarized light, such as Barkers reagent [lo], are not appropriate for EBSD studies

of aluminum in the SEM. This is due to for- mation of an anodized layer that tarnishes the grains surface and so prevents crystal- lographic data from being obtained. There- fore, for the determination of microtextural information, EBSD analysis of a specimen must be performed on an as-electropolished surface free of oxide and anodic layers.

MICROTEXTURE ANALYSIS

The EBSD equipment used in the present in- vestigation is attached to a JEOL 6100 SEM. EBSD measurements were taken from lon- gitudinal transverse sections of sheet speci- mens, as illustrated in Fig. 3. The orientations of 300 contiguous grains were analyzed in a direction parallel to the specimen normal. In brief, a specific region of the microstruc- ture was selected where the Kikuchi pattern of the first grain is indexed and saved. The specimen is then moved by the stage control until the electron beam traverses across a grain boundary of an adjacent grain. Because

Page 7: Secondary processing of electron backscatter data from an aluminum alloy

Secondary ProcessingofEBSD Data

I

the boundary is not visible from the speci- The total number of grains chosen for anal- men image, the location of the boundary is ysis depends on the strength or sharpness of identified by a rotation or complete change of the resulting microtexture. Specimens that Kikuchi pattern, indicating a neighboring exhibit a random or weak microtexture may subgrain or grain, respectively. The pattern is require 400 to 500 grain orientation measure- then indexed and saved in the same manner ments to produce a representative pole fig- as the preceding one, and the process is re- ure; whereas, for strongly textured specimens peated until the desired number of grams exhibiting clusters of grains of similar orien- have been analyzed. tation, 200 to 300 measurements may prove

FIG. 5. Deconvoluted <111> microtexture pole figures of AA8079 97% cold rolled material annealed at 4Oo”C, comprising 300 grains: (A) all grains; (B) (123]<634>; (C) {Zll)<lll>; (D) (436]<323>; (E) (OOl)<lOO>; (F) (lOl}tlll>; (G) (001)<310>; (H) (103]<311>; and (I) random.

Page 8: Secondary processing of electron backscatter data from an aluminum alloy

138

sufficient. In the present case, it was found that 300 orientation measurements were suf- ficient to provide a very good representation of microtexture.

The conventional specimen setup for EBSD work is illustrated in Fig. 4(a), where the specimen normal direction is perpendicular to the rolling plane and pole figures are rep- resented with respect to the normal direction. Because measurements were taken from the longitudinal section in the present work, the orientation of the specimen with respect to the microscope stage coordinates is rotated by 90”, as illustrated in Fig. 4(b). Thus, to dis- play the microtexture in conventional pole figure format, all orientation data were rotated +90” about the specimen rolling direction.

EXPERIMENTAL RESULTS AND

DISCUSSION

The most informative and complete method of representing a microtexture would be to plot the Cartesian coordinates of each orien- tation in a three-dimensional unit such as Eu- ler space or Rodrigues-Frank space. Because a relatively small number of grains have been selected in the present work, compared with the thousands of grains averaged by conven- tional X-ray techniques, a three-dimensional representation would prove difficult to inter- pret and display and indeed is less common for microtexture work. A convenient way of representing microtexture is by the use of pole figures, even though the information contained is reduced because they are es- sentially two-dimensional. The grain orien- tation distributions presented in this paper are <ill > microtexture pole figures that were subsequently analyzed according to the scheme described in the section on sec- ondary data processing.

Figure 5(a) displays the microtexture of AA8079 97% cold rolled material annealed at

R. Davies and V. Randle

400°C. A strong rolling texture is exhibited with a large number of grains having the S and Cu texture. Although the pole figure gives a two-dimensional qualitative repre- sentation of microtexture, the next stage is to process the data to separate the overall mi- crotexture to obtain a quantitative descrip- tion of the individual texture components. The deconvoluted microtextures obtained from secondary data processing are shown in Figs. 5(b)-5(i). The major components are the S texture (123]<634>, comprising 28.0% of total grains sampled, the Cu texture {211} <ill>, comprising 21.0%, and a rotated Cu component (436}<323> that is misoriented 11% about the <421> axis, comprising 19.3% of the total sample. There are also weaker components of cube [001]<100>, 8.7%; ro- tated-cube [001]<310>, 3.3%; (103} <311>, 4.3%; and (lOl]<lll> 4.7%. The number of grains associated with each individual tex- ture component and their proportions (%) are summarized in Table 3. Some of these com- ponents (e.g., (436}<323>) are unusual tex- ture features that would probably have been masked or overlooked in an X-ray texture analysis.

For convenience, in the data reported here, the S texture is described by the {123}<634> component, but approximately 50% of these grains are actually oriented about the alternative S component, 1123) <412>. As described earlier, for an in-depth EBSD characterization of the microtexture from this specimen, it is important to know if the grains contained within the S texture are evenly distributed between the four S variants or if certain variants are more fa- vorable. Secondary data processing was em- ployed to calculate the proportion of grains represented by each individual S variant.

The data in Table 3 place all the variants of a texture together. However, the micro- texture analysis allows these variants to be recorded separately, and it turns out

Table 3 Proportion (%) of Grains Representative of Each Texture Component

{OOZl <lOO> {001/<310> /203}<311> {1021<111> /Zll/<lll> /1231<634> /436/<323> Random

8.7 3.3 4.3 4.7 21.0 28.0 19.3 10.7

Page 9: Secondary processing of electron backscatter data from an aluminum alloy

Secondary Processing of EBSD Data

T

L

C

T

+

+ +

:

+

139

D

FIG. 6. <111> microtexture pole figures representing the S-texture variants: (a) (123)[aZ]; (b) (213)[n2]; (c) (123)[6%]; and (d) (213)[%4].

that only two of the possible four S-tex- ture variants are present. The pole figure representation of each of these variants- that is, (123)@i2], (213)[m], (123)[634], and (213)[364]-is shown in Figs. 6(a)-6(d), re- spectively, and the proportion of each S vari- ant as a percentage of the total S texture is given in Fig. 7. In instances where a grain is represented by more than one ideal variant, it is classified in the majority variant cate-

gory. The misorientations between adjacent

grains in the data sets also have been ana- lyzed. For the present case, the interest in this parameter is focused on the misorienta- tion angk, because it is pertinent to know whether grains are linked by high-angle or low-angle boundaries. Figure 8 shows the frequency distribution of misorientation angles, indicating that the proportion of boundaries at O-15” is higher than that ex-

pected in a totally random distribution. Furthermore, there is a higher proportion of misorientations at 50-60”, which is ex- plained by the adjacency in the microstruc-

40

.tj 30

6 b

.f 20 r B e

5 10

0

26.19

17.86

L

(123)[372] (213)[742] (123)[634] (213)[%4]

S-Texture Variants

FIG. 7. Histogram showing the proportion of each S variant as a percentage of the total S texture.

Page 10: Secondary processing of electron backscatter data from an aluminum alloy

140 R. Davies and V. Randle

FIG. 8. Plot of the distribution of grain misorientation angles in AA8079 97% cold rolled and fully recrystallized material compared with a random misorientation distribution.

ture of texture variants that are twin related, such as (123)[a2] and (213)[n2].

Finally, whereas the data in Fig. 8 show how individual grain pairs are related in the microstructure, they do not provide informa- tion on the distribution of low-angle bound- aries throughout the linearly sampled grains. Examination of the misorientations shows that the low-angle boundaries are not clus- tered in regions; rather they are distributed singly or, at the most, two adjacent low-angle boundaries occur. The implication from this analysis is that the material is fully recrystal- lized, which is indicated by the large propor- tion of high-angle grain boundaries. This is confirmed by qualitative assessment of the Kikuchi patterns that showed no diffuseness, hence indicating a strain-free microstructure. In relation to current research on continuous recrystallization, the use of EBSD to monitor the progress of microtexture during anneal- ing will not only glean information on tex- ture development, but also provide access to grain boundary parameters that may explain its mechanism. For example, previous work has suggested that, if the deformed substruc- ture comprises predominantly high-angle grain boundaries, continuous recrystalliza- tion could be explained by normal grain growth commencing from a very fine struc- ture ‘[ll]. Alternatively, if the deformed substructure consists mainly of low-angle

boundaries that gradually increase in misori- entation during annealing, continuous re- crystallization may be attributed to extended recovery [12]. In summary, the secondary processing of electron backscatter data is es- sential for a quantitative in-depth metallurgi- cal study of continuous recrystallization.

CONCLUDING REMARKS

1. Although EBSD patterns from alumi- num alloys tend to be more diffuse, and specimen preparation for these materials more complex, than for many other com- monly used alloys, these difficulties can be overcome.

2.

3.

1

Secondary data processing is necessary to extract the maximum information from EBSD-for example, deconvolution of the pole figure to give texture variant pro- portions and misorientation distributions. EBSD is currently being applied to study continuous recrystallization in an alumi- num alloy.

References

V. Randle, Microtexture Determination and ifs Appli- cations. Institute of Materials, London (1992).

Page 11: Secondary processing of electron backscatter data from an aluminum alloy

Secondary ProcessingofEBSD Data 141

2. A. J. Wilkinson and D. J. Dingley, The distribution of plastic deformation in a metal matrix composite caused by straining transverse to the fibre direction. Acta Metal/. Mater. 40:3357-3368 (1992).

3. V. Randle and D. J. Dingley, Microtexture determi- nation by electron back-scatter diffraction. J. Muter. Sci. 2F4545-4566 (1992).

4. G. Palumbo and K. T. Aust, Grain Boundary Char- acter Distributions in Nickel, Proc. Int. Conf. on Re- crystallisation ‘90, January 22-26, 1990, Woolon- gong, Australia (1990).

5. B. L. Adams, J. W. Zhao, and D. O’Hara, Analysis of interface damage heterogeneity in polycrystalline materials. Acta Met&. Mater. 38:953-966 (1990).

6. V. Randle, Application of electron back-scatter dif- fraction to steel products. Ironmaking Steelmaking 21:209-214 (1994).

7. V. Randle and M. C. Gaul, Representation of elec-

tron back-scatter diffraction data. Mat. Sci. Tech., in press.

8. H. J. Bunge, Three-dimensional texture analysis. Id. Metall. Rev. 32:265-291 (1987).

9. V. Randle, Grain Boundary Geometry in Polycrystak, Inst. Phys. Pub., Bristol, UK (1993).

10. L. J. Barker, Revealing the grain structure of com- mon aluminum alloy metallographic specimens. Trans. ASM 42:347-356 (1950).

11. A. Oscarsson, H.-E. EkstrGm, and B. Hutchinson, Transition from discontinuous to continuous recrys- tallization in strip-cast aluminum alloys. Muter. Sci. Forum 113-115:177-182 (1993).

12. R. 0rsund and E. Nes, Subgrain growth during annealing of heavily deformed metals. Scripta Metall. 23:1187-1192 (1989).

Received May 1996; accepted August 1996