research papers

Journal of
Applied
Determination of the degree of preferred
Crystallography orientation within the March–Dollase approach
ISSN 0021-8898

Emil Zolotoyabko
Received 18 August 2008
Accepted 12 April 2009
Department of Materials Engineering, Technion – Israel Institute of Technology, Haifa 32000, Israel.
Correspondence e-mail: [email protected]

An analytic expression has been derived connecting the degree of preferred

orientation in a polycrystalline material to the March parameter. The latter
defines the spread of angular distribution of crystallite inclinations in the
March–Dollase approach [March (1932). Z. Kristallogr. 81, 285–297; Dollase
(1986). J. Appl. Cryst. 19, 267–272]. In turn, the March parameter can be
extracted from experimental data using either the Rietveld refinement of the
entire diffraction pattern or the measurement of two diffraction intensities
# 2009 International Union of Crystallography originating in the selected crystallographic planes. Working examples taken with
Printed in Singapore – all rights reserved two different types of samples are presented.

1. Introduction sical technique is based on the measurement of pole figures

(Bunge, 1982). For this purpose, a conventional /2 diffract-
Practically all polycrystalline materials reveal some degree of ometer is used, equipped with a one-dimensional (slit)
preferred orientation of crystallites, owing to the forces detector and a special texture attachment. The latter provides
evolving between them during crystal growth and material sample rotation by changing the tilt angle  and azimuthal (in
processing. For this reason, the measurement of preferred the plane of the sample) angle ’. Practically, one measures the
orientation (texture) is of great importance because in this diffraction intensity of a fixed reflection (fixed angle 2) as a
way essential information on material characteristics can be function of  and ’ (pole figure). A pole figure (after appro-
obtained. Therefore, the development of fast and reliable priate instrumental corrections) shows the angular (, ’)
methods to characterize quantitatively the preferred orienta- distribution of the reciprocal lattice vector h = (hkl), which is
tion and its modification under processing is highly desirable. normal to the chosen crystallographic plane. In turn, an
However, despite intensive effort there is still a lack of fast inverse pole figure reflects the relative quantities of crystallites
well established techniques that would allow comparison of in which vectors h are nearly parallel to the sample direction z.
the texture of materials using a single parameter or a few of A more general approach, developed by Roe (1965) and
them. Bunge & Haessner (1968), provides the complete ODF, which
Crystallographic texture is characterized by electron and is defined as the relative volume of crystallites having orien-
X-ray diffraction methods. For electron diffraction, a specific tations between g and g + dg in the coordinate system attached
mode of scanning electron microscopy is used, namely elec- to the main directions of the sample. The ODF brings
tron backscatter diffraction (EBSD) (see e.g. Adams et al., comprehensive information on the preferred orientation and,
1993). In this mode, a specific diffraction pattern (Kikuchi in principle, can be extracted from a number of measured pole
lines) is registered from an individual grain by means of a figures. This can be done, for example, by using a generalized
CCD camera, and this is then converted to the grain orien- spherical harmonic description (see e.g. Ahtee et al., 1989;
tation with respect to the sample surface. By taking diffraction Popa, 1992; Järvinen, 1993; Von Dreele, 1997; Sitepu et al.,
patterns from a number of grains, the partial orientation 2005) or by the Williams–Imhof–Matthies–Vinel (WIMV)
distribution function (ODF) (limited by grain quantity) can be method (see e.g. Matthies & Vinel, 1982; Lutterotti et al., 1997;
reconstructed. The EBSD method works well in a thin layer Matthies et al., 1997). However, the ODF approach is rather
beneath the sample surface that is also restricted laterally on a time consuming because of the large amount of data which
scale in the 100 mm range. must be collected and processed.
Texture in deeper layers and averaged over much larger An essential technical step forward was the use of two-
sample areas is traditionally examined by the X-ray diffraction dimensional detectors for texture examination (Smith &
methods, although nowadays, because of progress in Ortega, 1993). A two-dimensional detector allows the capture
synchrotron micro-diffraction techniques, it is also possible to of a considerable part of the diffraction cone (instead of only
follow the orientation of individual micrometre-sized grains the small region available with a one-dimensional detector)
and their transformations using X-rays (Poulsen, 2004). and thus drastically increases the amount of information
There are several approaches for solving the texture obtained during an individual measurement. By using two-
problem with laboratory X-ray diffraction systems. The clas- dimensional detectors it is also possible, in some cases, to

J. Appl. Cryst. (2009). 42, 513–518 doi:10.1107/S0021889809013727 513

research papers
eliminate the sample tilt in texture measurements (Kurtz et al., the sample surface n, this being also the direction of the
2001), which significantly reduces the data acquisition time. preferred orientation H (Dollase, 1986):
However, as has already been mentioned, all these develop-  3=2
ments fail to provide a simple algorithm that allows the 2 2 1 2
WðÞ ¼ r cos  þ sin  : ð1Þ
quantitative description of the degree  of the preferred r
orientation (0   < 100%) of the sample and its change under The angle  in equation (1) is between the vectors h and H in a
processing using a very limited set of parameters. single crystal. The March parameter, 0 < r  1, determines the
In this paper, we attempt to fill this gap in the framework of shape of the function W() and, respectively, the strength of
the March–Dollase approach (Dollase, 1986; Čapková et al., the preferred orientation. At r = 1 (random powder), W() = 1
1993; Howard & Kisi, 2000). The latter uses the analytical and does not depend on . At r ! 0 (perfect uniaxial
weight function – the March function (March, 1932), which preferred orientation), W() transforms to a delta function,
depends on a single parameter (March parameter, 0 < r  1) – (0). At intermediate r values, 0 < r < 1, the W() function has
in order to describe the preferred orientation of the investi- a maximum at  = 0 which grows rapidly as W(0) = r3 when
gated sample. It is worth emphasizing that, having originally the r value diminishes to zero (see Fig. 1). We stress that the
been developed for uniaxial texture (where the weight func- same function in equation (1) describes the distribution of
tion does not depend on the azimuthal angle ’), this correc- crystallites in which the vector of the reciprocal lattice H is
tion can be successfully used in the Rietveld refinement of inclined by an angle  with respect to the normal n to the
different materials (see e.g. Young, 1993, and references surface of the sample (Dollase, 1986). Thus, the function W()
therein) and is now included in many currently available should be normalized,
refinement packages, such as GSAS (Larson & Von Dreele,
1986). Moreover, the March–Dollase approach can be applied R
=2
MðÞ sin  d ¼ 1; ð2Þ
to describe quantitatively a strong preferred orientation, 0
which causes the best part of the diffraction lines to vanish. In
reflecting the finite total number of crystallites with all
this case, Rietveld refinement is useless, but the March para-
possible inclinations of vector H with respect to vector n. This
meter can still be extracted from the measurement of two
important point is directly proved below.
diffraction intensities originating in the selected crystal-
The natural way of defining the degree of preferred orien-
lographic planes (see e.g. Zamir et al., 2000; Zolotoyabko &
tation is to connect it somehow to the fraction of crystallites
Quintana, 2002a). These considerations show the great
P(0) having the inclination angle  in the range 0    0.
potential of the March–Dollase method.
The choice of characteristic angle 0 will be discussed later. So,
At the same time, there are a very limited number of papers
based on equation (1), the P(0) value can be expressed as
focused on preferred orientation itself and its modification in
specific materials (see e.g. Deyu et al., 1990; Leventouri, 1997; Z0  3=2
1
Zamir et al., 2000; Zolotoyabko & Quintana, 2002a, Sitepu et Pð0 Þ ¼ r cos  þ sin2 
2 2
sin  d: ð3Þ
r
al., 2005). In all these papers, the March parameter r is 0
extracted, whose physical meaning and link to the degree of
preferred orientation is not very straightforward. It is believed Using the replacement cos  = t, the integral of equation (3)
that the degree of preferred orientation is  = 100%(1  r), transforms into
although this statement, to the best of our knowledge, has Rt2  3=2
never been justified. Seemingly, it is based on the obvious facts Pð0 Þ ¼ r3=2 dt 1  ð1  r3 Þt2 : ð4Þ
t1
that  = 0 for r = 1 (random powder) and  tends to 100% in
the limit of r = 0 (perfect uniaxial preferred orientation), as
well as on an intuitive assumption that the degree of preferred
orientation is linearly related to the March parameter r in
between.
Below, we develop an analytical procedure that offers a firm
relationship between the degree of preferred orientation 
and the March parameter r extracted from diffraction
measurements. The resulting equation shows that the depen-
dence (r) is not linear, although it provides the correct values
of (0) = 100% and (1) = 0 at the boundaries of the r range.

2. Interrelation between the degree of preferred

orientation and the March parameter
In the March–Dollase approach, the weight function W() is
Figure 1
introduced to describe the fraction of crystallites in which the The March function W(), calculated for r = 0.1 (solid line). The excess
vectors of the reciprocal lattice h are parallel to the normal to above the value for a random powder [W() = 1; dashed line] is hatched.

514 Emil Zolotoyabko  Preferred orientation J. Appl. Cryst. (2009). 42, 513–518
 research papers
A second replacement, t ¼ ð1  r3 Þ1=2 sin x, yields a tabulated 100% ð"=31=2 Þ. At the other extreme, r << 1, the  value tends
integral to 100% as  ’ 100% ½1  ð3=2Þr. Equation (12) provides
 1=2 Zx2 significantly different  values compared with the linear
r3 dx function l = 100% (1  r), also plotted for comparison in
Pð0 Þ ¼
ð1  r3 Þ cos2 x Fig. 2. For example, at r = 0.5, l = 50% while  = 37.8%.
x1
 1=2 Note that, formally, the approach described by equations
3
r (3)–(12) is developed for the pole in the weight function of
¼ ðtan x2  tan x1 Þ: ð5Þ
ð1  r3 Þ equation (1) located at  = 0 [integration between 0 and 0 in
equation (3)]. However, it is easy to show that exactly the
Finally, using the reverse replacement x¼ same final equation (12) remains valid for the pole in the
arcsin½ð1  r3 Þ1=2 cos , one obtains weight function of equation (1) located at  = /2 [integration
 1=2 n h io0
r3 3 1=2
between /2 and /2 + 0 in equation (3)]. As is known, this
Pð0 Þ ¼ tan arcsin 1  r cos  case corresponds to the March parameter ranging between 1
ð1  r3 Þ 0
and 1. This is standard, for example, for neutron diffraction
¼1 ð6Þ
measurements in transmission mode (see e.g. Sitepu et al.,
with 2001). Therefore, the final equation (12) can equally be used in
 1=2 both situations, i.e. r  1 and r  1.
tan2 0
 ¼ 1þ : ð7Þ
r3
Note that if 0 = /2, then  = 0 and P = 1, which justifies the 3. Working examples
normalization condition of equation (2). For a random powder
In practical terms, the degree of preferred orientation  in the
(r = 1), equation (7) transforms into
framework of the March–Dollase approach can be determined
p ¼ cos 0 : ð8Þ in two different ways. If the effect of preferred orientation is
weak, the best solution is given by applying the Rietveld
The degree of preferred orientation  is the excess of the P(0)
refinement procedure to the whole diffraction pattern,
value in a sample with r 6¼ 1 compared with a random powder,
extracting the March parameter r and determining the  value
i.e.
  via equation (12). In the case of strong preferred orientation,
 ¼ 100% ð1  Þ  ð1  p Þ many of the diffraction lines are drastically suppressed and
 application of Rietveld refinement becomes complicated. In
¼ 100% p  
"  1=2 # this situation, the best way is to measure the intensities of two
tan2 0 reflections, H(HKL) and h(hkl), one of them (H) being that of
¼ 100% cos 0  1 þ : ð9Þ
r3 the preferred orientation. By the aid of equation (1), one finds
that the intensity ratio should be
Obviously, the magnitude of  depends on the 0 value. For
example,  = 0 at 0 = 0. Therefore, we must allow the incli-  3=2
IðHÞ 2 sin2 
nation angle  to be within some reasonable angular interval. ¼ ¼ p cos  þ ; ð13Þ
We propose to restrict the latter to the point where the W() IðhÞ r3
function is equal to its value for a random powder (see Fig. 1),
where  is the angle between vectors H and h and p is the
i.e.
 3=2 intensity ratio for a random powder, which can be routinely
1
Wð0 Þ ¼ r2 cos2 0 þ sin2 0 ¼ 1: ð10Þ
r
Solving equation (10) for r 6¼ 1 yields
 1=2
1r
cos 0 ¼ : ð11Þ
1  r3
Finally, substituting equation (11) into equation (9) yields
 1=2
ð1  rÞ3
 ¼ 100% : ð12Þ
1  r3
Equation (12) can be introduced into computer programs
utilizing refinement of the March parameter r within the
Rietveld routine in order to obtain the degree of the preferred
orientation . The magnitude of  as a function of r is plotted Figure 2
The degree of preferred orientation (r) as a function of the March
in Fig. 2. As expected,  tends to zero when r tends to 1. parameter r, according to equation (12) (solid line). The linear
Defining r = 1  ", one obtains the asymptotic  ’ dependence  = 100%(1  r) is given for comparison.

J. Appl. Cryst. (2009). 42, 513–518 Emil Zolotoyabko  Preferred orientation 515
research papers
calculated or taken from the ICDD files. The March para- characterized by measuring the intensity ratios of two
meter r is restored from equation (13) as diffraction lines.
X-ray diffraction measurements were carried out on the
" #1=3
sin2  5BMD beamline of the Advanced Photon Source (APS) at
r¼ ð14Þ Argonne National Laboratory at an X-ray energy of 10 keV
ð = p Þ2=3  cos2 
(for details, see Zolotoyabko & Quintana, 2002a). The /2
scans were taken from the inner (shiny) surfaces of nearly flat
and is then used to find the  value by means of equation (12). rectangular shell pieces, a few millimetres in size, which were
According to equation (14), in the case of rather weak cut from a mature shell using a diamond saw. The incident
preferred orientation (  p), the accuracy of the r deter- beam was restricted by slits to 1 mm vertically and 3 mm
mination will be higher when choosing a second atomic plane horizontally. A typical /2 scan is plotted on a logarithmic
h to be inclined by a small angle (cos   1) with respect to the scale in Fig. 3 and shows a very strong preferred orientation of
plane H. the [001] type, which is revealed as a greatly enhanced
diffraction intensity originating from the (00l) reflections.
3.1. Preferred orientation in the nacre of mollusc shells Other diffraction lines are strongly suppressed. For example,
We have applied the measurement of diffraction intensity the diffraction intensity ratio for the (002) and (012) atomic
ratios to map the preferred orientation in the nacre layer of planes (having a small enough angle between them,  = 19.2 )
aragonitic bivalve shells of Acanthocardia tuberculata. In these is only p = 0.042 for random aragonite powder. Owing to the
shells, which grow to about 40 mm in size, the entire nacre high degree of preferred orientation, the intensity ratio =
layer is 200–300 mm thick and is built of aragonite tablets, each I(002)/I(012) measured in the nacre layer is greatly increased
a few micrometres long and 200–600 nm thick (see Zolo- and in some shells exceeds = 100. This is a clear case for
toyabko & Quintana, 2002b). The aragonite tablets reveal well applying the algorithm expressed by equations (13) and (14).
developed preferred orientation of the [001] type clearly Using the intensity ratios = I(002)/I(012) measured in
visible in scanning electron microscopy cross sections. This different samples and applying equations (14) and (12), the 
means that the aragonite crystals in the nacre layer grow with values for each sample were extracted. Since the investigated
the c axis of the orthorhombic unit cell oriented preferentially samples were taken from different locations over the entire
along the normal to the inner surface of the shell (adjacent to shell, putting the obtained  values all together on the map
the mollusc mantle). Aragonite tablets have a disk-like shape yields the spatial distribution of the degree of preferred
which is suitable for application of the March–Dollase orientation across the inner surface of the shell. The resulting
approach. More details of various microstructures in mollusc data are presented in the form of a 35-‘pixel’ map in Fig. 4. We
shells are given by Chateigner et al. (2000) and Pokroy & note the rather large range of the measured  values, confined
Zolotoyabko (2003). In the case of a crossed lamellar micro- between 32 and 80%. The minimum  = 32% was found at x =
structure, the degree of preferred orientation is not very high 25, y = 5, near the shell apex, where the oldest shell material is
(see e.g. Pokroy & Zolotoyabko, 2003) and the texture may be concentrated. At the shell’s rim, y = 25, where the youngest
refined together with structural parameters using the Rietveld nacre layer is deposited, the  values are much higher,
refinement procedure (see e.g. Ouhenia et al., 2008). In a well reaching the maximum  = 80% in a position located at x = 20,
developed nacre layer, the degree of preferred orientation y = 25.
may be much higher (see e.g. Zolotoyabko & Quintana,
2002a) and, as we show below, the texture can be successfully

Figure 4
Figure 3 A map of the degree of preferred orientation , showing the distribution
Diffraction data taken from one of the shells at 10 keV, showing the of  values across the inner surface of the shell. Horizontal and vertical
strong preferred orientation of the [001] type. dimensions on the map are given in millimetres.

516 Emil Zolotoyabko  Preferred orientation J. Appl. Cryst. (2009). 42, 513–518
 research papers
3.2. GaN growth on Si substrates
Our second example is taken from the field of artificial
crystal growth, namely epitaxial growth of GaN films on [111]-
oriented Si substrates by the metal–organic chemical vapour
deposition technique. As shown by Zamir et al. (2000), in this
case a buffer layer of AlN should be used in order to enhance
the nucleation rate of GaN. The critical parameters for the
successful growth of GaN on Si are the temperature T and the
duration t of the growth of the buffer layer. To find the opti-
mized growth conditions, we used laboratory X-ray diffraction
measurements carried out using a Philips X-Pert system.
Specifically, we performed /2 scans in the angular range that
included the (002) (basal plane of the wurtzite-type GaN Figure 5
structure) and (102) GaN reflections. The latter is visible in The degree of preferred orientation  in a GaN film, plotted as a function
diffraction scans with no overlap with the strong (002) of the growth temperature of the AlN buffer layer.
reflection. The angle between the (001) and (102) atomic
planes is  = 43.2 and the intensity ratio for random powder is
p = I(002)/I(102) = 0.83. In the case of epitaxial GaN growth
on [111]-oriented Si, the basal plane of GaN becomes the
plane of preferred orientation and the intensity ratio =
I(002)/I(102) is greatly increased. We have used this fact to
explain qualitatively the main features of the effect of the
buffer layer on GaN growth (Zamir et al., 2000). Here, we
quantitatively describe the degree of preferred orientation in
this system, applying equations (14) and (12) to the previously
obtained X-ray diffraction ratios .
The effect of the buffer layer growth temperature on the
quality of the GaN films is shown in Fig. 5. At low tempera-
tures, T ’ 700 K, the grown films are poor quality, char-
acterized by a rather low degree of preferred orientation of Figure 6
The degree of preferred orientation  in a GaN film, plotted as a function
less than 50%. The  values increase rapidly with temperature of the duration of the AlN buffer layer growth.
up to  = 89% at T ’ 760 K. After that, the degree of
preferred orientation increases very slowly with temperature,
reaching the remarkable maximum value of  = 90% at T ’ that we are able to characterize the investigated systems
1100 K. The change in growth mode at 760 K is explained in a quantitatively, i.e. to determine the proportion (in percent) of
paper by Zamir et al. (2000) in terms of the characteristics of ‘well organized’ crystallites and to follow its modification as a
AlN crystallites, which are governed by the interplay between function of various parameters. This is achieved using the
nucleation and growth processes in the AlN layer. newly developed equation (12), which connects the degree of
Another important parameter is the growth duration t of preferred orientation to a single March parameter. The latter
the buffer layer. The degree of preferred orientation achieved can be extracted from diffraction measurements, using Riet-
in a GaN film at T = 768 K, as a function of growth duration t, veld refinement of a whole diffraction pattern if the preferred
is plotted in Fig. 6. At short durations of less than 3 min, and orientation is rather weak, or by measuring the intensity ratios
rather long durations of more than 6 min, the quality of the of two appropriately selected diffraction lines in cases where
GaN films is very poor, revealing an extremely low degree of the preferred orientation is pronounced.
preferred orientation,  ’ 20%. The maximum preferred
orientation, exceeding 86%, is achieved at t = 5 min. The
existence of an optimum growth duration of the buffer layer is
explained by Zamir et al. (2000) as the result of competition
between crystal growth and self-annealing processes.
Experimental data relating to mollusc shells were collected
at Sector 5 of the APS, which is run by the DuPont–North-
western–Dow Collaborative Access Team (DND-CAT). The
4. Conclusions help of J. P. Quintana and B. Pokroy is gratefully acknowl-
Working examples show that the procedure developed here edged. Raw diffraction data from GaN films were kindly
provides valuable information on the degree of preferred provided by S. Zamir and his contribution is highly appre-
orientation in different systems. It can be used to characterize ciated. This work was partially supported by the Technion
material modifications under growth and processing. We stress Fund for Promotion of Research.

J. Appl. Cryst. (2009). 42, 513–518 Emil Zolotoyabko  Preferred orientation 517
research papers
References Matthies, S., Lutteroti, L. & Wenk, H. R. (1997). J. Appl. Cryst. 30,
31–42.
Adams, B. L., Wright, S. I. & Kunze, K. (1993). Metall. Trans. A, 24, Matthies, S. & Vinel, G. W. (1982). Phys. Status Solidi B, 112, K111–
819–831. 114.
Ahtee, M., Nurmela, M., Suortti, P. & Järvinen, M. (1989). J. Appl. Ouhenia, S., Chateigner, D., Belkhir, M. A. & Guilmeau, E. (2008). J.
Cryst. 22, 261–268. Struct. Biol. 163, 175–184.
Bunge, H. J. (1982). Texture Analysis in Materials Science: Pokroy, B. & Zolotoyabko, E. (2003). J. Mater. Chem. 13, 682–688.
Mathematical Methods. London: Butterworth. Popa, N. C. (1992). J. Appl. Cryst. 25, 611–616.
Bunge, H. J. & Haessner, F. (1968). J. Appl. Phys. 39, 5503–5514. Poulsen, H. F. (2004). Three-Dimensional X-ray Diffraction Micro-
Čapková, P., Peschar, R. & Schenk, H. (1993). J. Appl. Cryst. 26, 449– scopy, Springer Tracts in Modern Physics, Vol. 205. Berlin:
452. Springer-Verlag.
Chateigner, D., Hedegaard, C. & Wenk, H.-R. (2000). J. Struct. Geol. Roe, R. J. (1965). J. Appl. Phys. 36, 2024–2031.
22, 1723–1735. Sitepu, H., O’Connor, B. H. & Li, D. (2005). J. Appl. Cryst. 38, 158–
Deyu, L., O’Connor, B., Roach, G. I. D. & Cornell, J. B. (1990). 167.
Powder Diffr. 5, 79–85. Sitepu, H., Prask, H. J. & Vaudin, M. D. (2001). Adv. X-ray Anal. 44,
Dollase, W. A. (1986). J. Appl. Cryst. 19, 267–272. 241–246.
Howard, C. J. & Kisi, E. H. (2000). J. Appl. Cryst. 33, 1434–1435. Smith, K. L. & Ortega, R. B. (1993). Adv. X-ray Anal. 36, 641–
Järvinen, M. (1993). J. Appl. Cryst. 26, 525–531. 647.
Kurtz, D. S., Kozaczek, K. J. & Morgan, P. R. (2001). US Patent No. Von Dreele, R. B. (1997). J. Appl. Cryst. 30, 517–525.
6301330B1. Young, R. A. (1993). Editor. The Rietveld Method. Oxford University
Larson, A. C. & Von Dreele, R. B. (1986). GSAS. Report LAUR 86- Press.
748. Los Alamos National Laboratory, New Mexico, USA. Zamir, S., Meyler, B., Zolotoyabko, E. & Salzman, J. (2000). J. Cryst.
Leventouri, T. (1997). Physica C, 277, 82–86. Growth, 218, 181–190.
Lutterotti, L., Matthies, S., Wenk, H. R., Schultz, A. S. & Richardson, Zolotoyabko, E. & Quintana, J. P. (2002a). Rev. Sci. Instrum. 73,
J. W. (1997). J. Appl. Phys. 81, 594–600. 1663–1667.
March, A. (1932). Z. Kristallogr. 81, 285–297. Zolotoyabko, E. & Quintana, J. P. (2002b). J. Appl. Cryst. 35, 594–599.

518 Emil Zolotoyabko  Preferred orientation J. Appl. Cryst. (2009). 42, 513–518