Scripta Materialia 51 (2004) 777–781
www.actamat-journals.com
Microstructural parameters from X-ray diffraction peak broadening
T. Ung
ar *
Department of General Physics, E€otv€os University, P.O. Box 32, H-1518 Budapest, Hungary
Accepted 4 May 2004
Available online 4 June 2004
Abstract
X-ray diffraction peak profiles provide crystallite size, size distribution and dislocation structure. Theoretical size and strain
profile functions are given by the minimum number of parameters required by physics and are compared with TEM results. Strain
anisotropy gives information about dislocation character and Burgers vector populations. Thermal stability determined by X-rays
and calorimetry is discussed for recovery.
2004 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: X-ray line broadening; Dislocation structure; Thermal stability
1. Introduction cations this means a dislocation density larger than
about 5 · 1012 m2 . A pragmatic scheme for the corre-
X-ray diffraction peak profile analysis (XDPPA) is a lation between diffraction and lattice imperfectness is:
powerful alternative to transmission electron micro- size broadening is diffraction order independent, how-
scopy (TEM) for describing the microstructure of crys- ever, strain broadening increases with diffraction order.
talline materials. It is also complementary to TEM since This rule of thumb has many exceptions depending on
the parameters provided by the two different methods the specific crystallite size, especially crystallite shape,
are never identical, and thus enables a more detailed and on the specific type of strain caused by different
understanding of microstructures. In the case of ultra- types of lattice defects. The correct interpretation of
fine-grained materials, the physical properties are these various exceptions is the basis for detailed mod-
strongly related to the microstructure [1–15]. The eling of microstructures based on X-ray line profile
mechanical properties are enhanced, however the duc- analysis.
tility is often strongly reduced [1–6], compaction by
sintering without binding material is a promising pro-
cedure [7–10], favorable soft magnetic properties can be 2. Strain, lattice defects and dislocations
achieved, especially in ultrafine-grained iron or iron base
alloys [11–13]. The characterization of the microstruc- The most common sources of strain can be given as:
ture by TEM is complicated because of the very small (a) dislocations, (b) stacking faults, (c) twinning, (d)
size of crystallites and the extremely high dislocation microstresses, (e) long-range internal stresses, (f) grain
density [2,3]. boundaries, (g) sub-boundaries, (h) internal stresses, (i)
X-ray diffraction peaks broaden when the crystal coherency strains, (j) chemical heterogeneities, (k) point
lattice becomes imperfect. The microstructure means the defects or (l) precipitates and inclusions. Though dislo-
extent and the quality of lattice imperfectness. Accord- cations are listed alone as one of the sources of strain,
ing to the theory of kinematical scattering, X-ray dif- they are the major components in (b)–(g) and can play
fraction peaks broaden either when crystallites become an important role in (h) and (i).
smaller than about a micrometer or if lattice defects are The ideal powder diffraction pattern consists of nar-
present in large enough abundance. In terms of dislo- row, symmetrical, delta-function like peaks, positioned
according to a well-defined unit cell. The aberrations
*
Tel.: +36-1372-28-01; fax: +36-1372-28-11. from the ideal powder pattern can be conceived as: (i)
E-mail address:
[email protected] (T. Ungar). peak shift, (ii) peak broadening, (iii) peak asymmetries,
1359-6462/$ - see front matter 2004 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.scriptamat.2004.05.007
778 T. Ungar / Scripta Materialia 51 (2004) 777–781
Table 1
The most typical correlations between diffraction peak aberrations, i.e. broadening, shifts or asymmetries, and the different elements of micro-
structure
Sources of strain Peak aberrations
Peak shift Peak broadening Peak asymmetry Anisotropic peak Peak shape
broadening
Dislocations + + + +
Stacking faults + + + + +
Twinning + + + + +
Microstresses +
Long-range internal stresses + +
Grain boundaries + +
Sub-boundaries + +
Internal stresses +
Coherency strains + + +
Chemical heterogeneities + + +
Point defects +
Precipitates and inclusions + +
Crystallite smallness + + +
(iv) anisotropic peak broadening and (v) peak shape. term ‘‘restrictedly random’’ for the case when screening
The main correlation between these peak aberrations is important, however not as strong as to produce
and the different elements of microstructure are sum- infinitesimal dipoles, and worked out the profile shape
marized in Table 1. The table shows that (i) peak shift is for such dislocation arrangements and suggested to use
related to the different types of internal stresses and the dislocation density and the effective outer cutoff ra-
planar faults, especially stacking faults or twin bound- dius, q and Re , respectively [17,19]. In order to avoid
aries. (ii) Peak broadening indicates crystallite smallness confusion between Re and other length scales, it is
and microstresses, however, stress gradients and/or better to use the dislocation arrangement parameter
pffiffiffi
chemical heterogeneities can also cause peak broaden- M ¼ Re q [18]. For more detailed characterization,
ing. (iii) Peak asymmetries can be caused by long-range further parameters can be defined [18], Kroener [20], for
internal stresses, planar faults or chemical heterogene- example, suggested to use a dislocation-density tensor.
ities. (iv) Anisotropic peak broadening can result from
anisotropic crystallite shape or anisotropic strain. (v)
Peak shape refers to the way the intensity decays into 3. Size
the background, mainly below the half maximum.
The correlation between peak shape and specific lat- The model for size broadening is based on the simple
tice defects has been discussed in terms of the hierarchy concept that only those planes participate in a diffrac-
of lattice defects by Krivoglaz [16]. Lattice defects are tion experiment whose normal vectors are parallel to the
sorted into three different classes: point defects, linear diffraction vector and that the extension of these planes
defects and planar defects. The strain fields of linear is immaterial for the scattering mechanism [21]. Accord-
defects, especially dislocations, decay as 1=r, and are of ing to this concept, all fractions of all crystals or crys-
long-range character. Due to the reciprocity between tallites which have the same thickness parallel to the
crystal and reciprocal space, the scattering related to diffraction vector correspond to a single ‘‘column’’ with
these defects decays asymptotically between DK 2 and one particular ‘‘length’’. Proceeding this mechanism
DK 3 (where DK ¼ ð2 cos hB =kÞDh, hB is the Bragg angle, through the whole sample, all materials will be sorted
k is the X-ray wavelength and Dh ¼ h hB ). This scat- into the columns with different lengths. Each column
tering is clustered around the fundamental Bragg length will have a weight proportional to the volume
reflections and is called diffraction peak- or line broad- fraction of the material sorted into a particular column.
ening. Within the complex phenomena of line broad- The diffraction peak corresponding to one particular
ening caused by different lattice defects, dislocations column is the squared Fourier transform of the form
play a central role: either as the dominant lattice defects function pertaining to that column length. The final size-
or as the major components in complex lattice defects. broadened peak is the weighted sum of the individual
Planar faults, especially stacking faults and twin bound- peaks corresponding to the individual columns. The
aries are not discussed in the present account. direct meaning of a size-broadened peak is thus related
Dislocations are usually present either as dislocation to the weighted column–length distribution in the
dipoles or in special configurations where screening specimen. This on the other hand, depends on the size,
plays an important role [16–18]. Wilkens [17] used the the shape and the size distribution of the crystallites. In
T. Ungar / Scripta Materialia 51 (2004) 777–781 779
order to obtain the mean size of crystallites, assumptions systems in the crystal. The C hk0 and the q1 and q2
have to be made about their shape and size distribution. parameters have been numerically evaluated and com-
piled for a large number of hexagonal crystals [27] for
the most common 11 slip systems according to Jones
4. Basic principles of peak profile analysis and Hutchinson in [28].
Assuming that size (S) and lattice distortions (D) are
the two sources of X-ray line broadening, the Fourier 5. Whole profile fitting method
transforms are given by the Warren–Averbach equation
[22]: A procedure has been worked out, in which the dif-
2 2 2 fraction pattern is simulated by using ab initio physi-
ln AðLÞ ffi ln ASL 2p L g he2g;L i; ð1Þ
cally well established functions for both, the size and the
where AðLÞ are the absolute values of the Fourier coef- strain profile [29]. The size profile, I S , is given by
ficients of the physical profiles, ASL are the size Fourier assuming either spherical or ellipsoidal crystallite shape
coefficients, g is the absolute value of the diffraction and log-normal size distribution. For spherical crystal-
vector, he2g;L i is the mean square strain and L is the lites, the function is defined by the two parameters: the
Fourier length or variable. In a dislocated crystal he2g;L i median, m, and the variance, r, of the size distribution
is [17,19]: function. The strain profile, I D , is constructed by using
the f ðgÞ Wilkens function, as defined in Eq. (2). It is
he2g;L i ¼ ðb=2pÞ2 pqCf ðgÞ; ð2Þ
defined by: the dislocation density, q, and the disloca-
where g ¼ L=Re , b is the Burgers vector of dislocations tion arrangement parameter, M. In the simplest case,
and C is the contrast or orientation factor of disloca- together with the q parameter in the contrast factors for
tions [23,24]. Wilkens determined the f ðgÞ function for cubic crystals, the numerical procedure is operating with
parallel straight screw dislocations in the entire L range five independent parameters: m, r, q, M and q. Due to
from zero to infinity, see e.g. in Eqs. A.6 to A.8 in [19]. It the physical nature of the microstructure, this is the
can be shown that the Wilkens function, f ðgÞ, has a minimum number of parameters which must be used.
more general validity for edge and curved dislocations The instrumental profile, I i , has to be measured sepa-
and, it can be extended in a simple way for infinitesimal rately. The numerical procedure (called convolution
dislocation dipoles when M goes to zero [25]. multiple whole profile (CMWP) fitting procedure) is
based on the convolution of the three functions, I S ,
4.1. Strain anisotropy I D and I i [30]:
I SIM ¼ I S I D I i þ I BG ; ð5Þ
Strain anisotropy means that neither the breadths nor
the Fourier coefficients of the diffraction profiles are where I BG is the background determined before the
monotonous functions of the diffraction angle or g [26]. simulated I SIM function is fitted to the measured dif-
If dislocations are the major source of strain in a crystal, fraction pattern.
the values of C depend on the relative orientation of the The microstructural parameters of an electrodepos-
Burgers-, the line and the diffraction vectors, b, l and g, ited, ultrafine-grained nickel specimen as determined by
respectively. If either all possible slip systems in a crystal the CMWP procedure are: (i) q ¼ 71
1014 m2 , M ¼ 3,
are more or less equally populated or, if a polycrystal is q ¼ 2, m ¼ 10 nm, and r ¼ 0:68. The corresponding size
more or less texture-free, the contrast factors corre- distribution function is shown in Fig. 1. The arithmetic,
sponding to one particular hkl can be averaged over the the area and the volume averaged means of the crystallite
permutations of these hkls. For cubic crystals, C is [23]: diameters, hxiarith , hxiarea and hxivol , [31] are also indicated
in the figure. A TEM micrograph of the same sample (by
C ¼ C h00 ð1 qH 2 Þ; ð3Þ courtesy of [32]) is shown in Fig. 2. A good correlation
where C h00 is the average dislocation contrast factor of between the X-ray determined area averaged mean
the h00 type reflections and H 2 ¼ ðh2 k 2 þ h2 l2 þ crystallite size and the TEM grain size can be observed.
2 The q ¼ 2 value indicates that the dislocations in the
k 2 l2 Þ=ðh2 þ k 2 þ l2 Þ . The values of C h00 and the q
parameter can be obtained numerically for different specimen are of mixed edge and screw character [24].
dislocation types as functions of the elastic properties
of a crystal [24]. For hexagonal crystals C is [27]:
6. The stability of different dislocation types during
C hkl ¼ C hk0 ½1 þ q1 x þ q2 x2 ; ð4Þ
annealing in fine grain Ti
2
where x ¼ ð2=3Þðl=gaÞ , a is the lattice constant in the
basal plane and q1 and q2 are numerical constants A commercial purity Ti specimen was deformed by
depending on the elastic constants and the active slip eight ECAP passes in the B mode [33]. The Burgers
780 T. Ungar / Scripta Materialia 51 (2004) 777–781
0.12 100
m Electrodeposited Ni
f(x)
<x> arith 75 <a >
Relative fraction[%]
0.08 TEM
50
<x> vol <c+a>
<x> area
0.04
25
<c >
0
0.00
0 25 50 300 400 500 600 700 800 900
x[nm]
Temperature [K ]
Fig. 1. The size distribution function, f ðxÞ, of an electrodeposited, Fig. 3. Burgers vector populations and slip system activity as a func-
ultrafine-grain nickel specimen determined by the X-ray method. The tion of annealing temperature in an ultrafine-grain Ti specimen
mean values obtained by the different averaging procedures together deformed by eight ECAP passes, by courtesy of [33].
with the average grain size determined from the TEM micrograph
in the next figure, are also indicated.
7. Thermal stability of the microstructure in ultrafine-
grained (UFG) Al–Mg–Sc–Zr and UFG-Ti
An aluminum base Al–5.9Mg–0.3Sc–0.18Zr alloy
(at.%) was deformed by the method of high pressure
torsion (HPT) applying 1/2, 1, 3, 5, 10 and 15 rotations,
respectively [38]. Three specimens deformed by 15
rotations were heated in a differential scanning calo-
rimeter (DSC2, Perkin-Elmer) from 320 to 450, 520 and
800 K at a heating rate of 20 K/min. The two specimens
heated to 450 and 520 K were cooled to room temper-
ature by 200 K/min after reaching these temperatures.
The volume averaged crystallite size, the dislocation
density and the heat flow are shown in Fig. 4 as a
function of temperature. It can be seen that at the
temperature where the heat flow starts to increase, (i) the
Fig. 2. TEM micrograph of the same specimen as in Fig. 1, by cour- major part of the dislocation density is annealed out,
tesy of [32]. however, (ii) the crystallite size is still at a low level. The
vector populations and slip system activities have been
400 0.4
investigated on the basis of the experimentally deter-
mined dislocation contrast factors and the theoretical
ρ [ 1013 m-2 ]
Heat power/mass [W/g]
ðmÞ
contrast factors defined in Eq. (4). The measured q1 300
Xarea
0.3
ðmÞ
and q2 and theoretical q1 and q2 parameter values have ρ
been matched by a numerical algorithm described in 200 0.2
detail in [34,35]. In the ECAP deformed specimen, the
Xarea [nm] ;
relative fractions of the hai, hci and hc þ ai Burgers
100 0.1
vectors are 62%, 0–4% and 32–36%, respectively, in Heat Flow
good correlation with earlier X-ray [36] and TEM
[33,37] results. The evolution of the Burgers vector 0 0.0
population and the relative fraction of the three basic 300 400 500 600
slip systems as a function of the annealing experiments is T [K]
shown in Fig. 3. It can be seen that as the temperature
Fig. 4. The dislocation density, the area averaged mean crystallite size
increases the relative fraction of the hc þ ai-type dislo- and the heat flow as a function of temperature in an Al–Mg–Sc–Zr
cations decreases indicating that these dislocations alloy deformed by 15 rotations of high-pressure torsion, by courtesy
disappear faster than the hai or hci-type ones. of [38].
T. Ungar / Scripta Materialia 51 (2004) 777–781 781
crystallite size starts to increase considerably at the [6] Zehetbauer M, Stuewe HP, Vorhaber A, Schafler E, Kohout J.
temperature where the heat flow has its maximum. This Adv Eng Mater 2003;5:330.
[7] Arato P, Bartha L, Porat R, Berger S, Rosen A. Nanostruct
indicates that in the UFG alloy produced by HPT, (a) Mater 1998;10:245.
the dislocation density is reduced by an order of mag- [8] Szepv€
olgyi J, Mohai I, Gubicza J. J Mater Chem 2001;11:859.
nitude before the crystallite size determined by X-rays [9] Qian J, Pantea C, Voronin GA, Zerda TW. J Appl Phys
starts coarsening and that (b) the major part of heat flow 2001;90:1632.
is related to this crystallite size coarsening. [10] Pantea C, Qian J, Zerda TW. J Appl Phys 2002;91:1957.
[11] Herzer G. Scripta Metall 1995;33:1741.
[12] Hernando A, Navarro I, Gorrıa P. Phys Rev B 1995;51:3281.
[13] Kaptas D, Kiss LF, Balogh J, Gubicza J, Kemeny T, Vincze I.
8. Conclusions Hyperf Interact 2002;141/142:175.
[14] Hughes DA, Hansen N. Phys Rev Lett 2001;87:135503.
Crystallite size or size distribution determined by X- [15] Hansen N. J Mater Sci Technol 2001;17:409.
[16] Krivoglaz MA. Theory of X-ray and thermal neutron scattering
ray line profile analysis can be given by the median and by real crystals. Berlin: Springer-Verlag; 1996.
the variance of the size distribution function. The area [17] Wilkens M. Phys Stat Sol (a) 1970;2:359.
weighted mean crystallite size can directly be compared [18] Groma I. Phys Rev B 1998;57:7535.
with size values obtained by TEM micrographs. The X- [19] Wilkens M. In: Simmons JA, de Wit R, Bullough R, editors.
ray crystallite size is always the average of the smallest Fundamental aspects of dislocation theory. Vol. II. Nat Bur Stand
(US) Spec Publ No 317, Washington, DC, USA, 1970. p. 1195.
undistorted volumes in the crystal. The dislocation [20] Kroener E. In: Balian R, Klemean M, Poirier JP, editors. Physics
structure determined by X-ray method can be given by of defects. Amsterdam: North Holland; 1981. p. 215.
the density, q, and the arrangement parameter, M, of [21] Bertaut EF. Acta Cryst 1950;3:14.
dislocations. The evaluation of strain anisotropy can be [22] Warren BE. Progr Metal Phys 1959;8:147.
done in terms of dislocation character or, in the case of [23] Ungar T, Tichy G. Phys Stat Sol (a) 1999;171:425.
[24] Ungar T, Dragomir I, Revesz A, Borbely A. J Appl Cryst
hexagonal crystals, in terms of the relative fraction 1999;32:992.
of the three fundamental dislocation types, hai, hci or [25] Groma I, Csikor F. J Mech Behav Mater, submitted for
hc þ ai. publication.
[26] Caglioti G, Paoletti A, Ricci FP. Nucl Instrum 1958;3:223.
[27] Dragomir IC, Ungar T. J Appl Cryst 2002;35:556.
[28] Jones IP, Hutchinson WB. Acta Metall 1981;29:951.
Acknowledgements [29] Ribarik G, Ungar T, Gubicza J. J Appl Cryst 2001;34:669.
[30] Ribarik G, Ungar T. Gubicza J. Mater Sci Eng A, in press.
This work was supported by the Hungarian Science [31] Ungar T, Gubicza J, Ribarik G, Borbely A. J Appl Cryst
Research Fund, OTKA, grant nos. T031786, T034666 2001;34:298.
[32] Zhilyaev AP, Gubicza J, Nurislamova G, Revesz A, Suri~
nach S,
and T046990.
Baro MD, et al. Phys Stat Sol (a) 2003;171:425.
[33] Zhu YT, Huang JY, Gubicza J, Ungar T, Wang YM, Ma E, et al.
J Mater Res 2003;18:1908.
References [34] Gubicza J, Nam NH, Balogh L, Hellmig RJ, Stolyarov VV, Estrin
Y, et al. J Alloys Comp, in press.
[1] Chokshi AH, Rosen A, Karch J, Gleiter H. Scripta Metall [35] Mathis K, Nyilas K, Axt A, Dragomir IC, Ungar T, Lukac P.
1989;23:1679. Acta Mater 2004;52:2889.
[2] Weertman JR. Mater Sci Eng A 1993;166:161. [36] Kuzel Jr R, Klimanek P. J Appl Cryst 1989;22:299.
[3] Siegel RW. J Phys Chem Solids 1994;55:1097. [37] Zhilyaev AP, Gubicza J, Suri~ nach S, Bar o MD, Langdon TG.
[4] Gubicza J, Arato P, Weber F, Juhasz A. Mater Sci Eng A Mater Sci Forum 2003;426:4507.
1999;259:65. [38] Fatay D, Bastarash E, Nyilas K, Dobatkin S, Gubicza J, Ungar T.
[5] Gubicza J, Weber F. Mater Sci Eng A 1999;263:101. Z Metallkde 2003;7:842.
